HIGH-THROUGHPUT SCREENING METHODS IN …digilib.bppt.go.id/sampul/9781118538203.pdf · Frederik...

HIGH-THROUGHPUTSCREENING METHODSIN TOXICITY TESTING

HIGH-THROUGHPUTSCREENING METHODSIN TOXICITY TESTING

Edited by

PABLO STEINBERGInstitute for Food Toxicology and Analytical ChemistryUniversity of Veterinary Medicine HannoverHannover, Germany

Copyright C© 2013 by John Wiley & Sons, Inc. All rights reserved.

Published by John Wiley & Sons, Inc., Hoboken, New Jersey.Published simultaneously in Canada.

No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form orby any means, electronic, mechanical, photocopying, recording, scanning, or otherwise, except aspermitted under Section 107 or 108 of the 1976 United States Copyright Act, without either the priorwritten permission of the Publisher, or authorization through payment of the appropriate per-copy fee tothe Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, (978) 750-8400,fax (978) 750-4470, or on the web at www.copyright.com. Requests to the Publisher for permissionshould be addressed to the Permissions Department, John Wiley & Sons, Inc., 111 River Street, Hoboken,NJ 07030, (201) 748-6011, fax (201) 748-6008, or online at http://www.wiley.com/go/permission.

Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts inpreparing this book, they make no representations or warranties with respect to the accuracy orcompleteness of the contents of this book and specifically disclaim any implied warranties ofmerchantability or fitness for a particular purpose. No warranty may be created or extended by salesrepresentatives or written sales materials. The advice and strategies contained herein may not be suitablefor your situation. You should consult with a professional where appropriate. Neither the publisher norauthor shall be liable for any loss of profit or any other commercial damages, including but not limited tospecial, incidental, consequential, or other damages.

For general information on our other products and services or for technical support, please contact ourCustomer Care Department within the United States at (800) 762-2974, outside the United Statesat (317) 572-3993 or fax (317) 572-4002.

Wiley also publishes its books in a variety of electronic formats. Some content that appears in print maynot be available in electronic formats. For more information about Wiley products, visit our web site atwww.wiley.com.

Library of Congress Cataloging-in-Publication Data:

High-throughput screening methods in toxicity testing / edited by Pablo Steinberg.pages cm

Includes index.ISBN 978-1-118-06563-1 (hardback)1. High throughput screening (Drug development) 2. Toxicity testing.I. Steinberg, Pablo, editor if compilation.RS419.5.H542 2013615.1′9–dc23

2012035755

Printed in the United States of America

ISBN: 9781118065631

10 9 8 7 6 5 4 3 2 1

CONTENTS

PREFACE ix

CONTRIBUTORS xi

PART I GENERAL ASPECTS

1 ToxCast: Predicting Toxicity Potential Through High-ThroughputBioactivity Profiling 3Keith A. Houck, Ann M. Richard, Richard S. Judson, Matthew T. Martin,David M. Reif, and Imran Shah

2 High-Throughput Toxicity Testing in Drug Development: Aims,Strategies, and Novel Trends 33Willem G.E.J. Schoonen, Walter M.A. Westerink, Femke M. van de Water, andHorbach G. Jean

3 Incorporating Human Dosimetry and Exposure Informationwith High-Throughput Screening Data in ChemicalToxicity Assessment 77Barbara A. Wetmore and Russell S. Thomas

4 The Use of Human Embryonic Stem Cells in High-ThroughputToxicity Assays 97Xin Huang, Dan-yan Zhu, and Yi-jia Lou

v

vi CONTENTS

PART II HIGH-THROUGHPUT ASSAYS TO ASSESS DIFFERENTCYTOTOXICITY ENDPOINTS

5 High-Throughput Screening Assays for the Assessment ofCytotoxicity 109Andrew L. Niles, Richard A. Moravec, Tracy J. Worzella, Nathan J. Evans,and Terry L. Riss

6 High-Throughput Flow Cytometry Analysis of Apoptosis 129Francesca de Giorgi and Francois Ichas

7 High Content Imaging-Based Screening for Cellular ToxicityPathways 143Bram Herpers and Bob van de Water

8 The Keratinosens Assay: A High-Throughput Screening Assay toAssess Chemical Skin Sensitization 159Andreas Natsch

9 High-Throughput Screening Assays to Assess ChemicalPhototoxicity 177Satomi Onoue, Yoshiki Seto, and Shizuo Yamada

PART III HIGH-THROUGHPUT ASSAYS TO ASSESS DNA DAMAGEAND CARCINOGENESIS

10 Ames IITM and Ames Liquid Format MutagenicityScreening Assays 193Kamala Pant

11 High-Throughput Bacterial Mutagenicity Testing: VitotoxTM Assay 213Luc Verschaeve

12 Genotoxicity and Carcinogenicity: Regulatory and NovelTest Methods 233Walter M.A. Westerink, Joe C.R. Stevenson, G. Jean Horbach, Femke M. van deWater, Beppy van de Waart, and Willem G.E.J. Schoonen

13 High-Throughput Genotoxicity Testing: The Greenscreen Assay 271Jorg Blumel and Nadine Krause

14 High-Throughput Assays to Quantify the Formation ofDNA Strand Breaks 285Marıa Moreno-Villanueva and Alexander Burkle

CONTENTS vii

15 High-Throughput Versions of the Comet Assay 295Irene Witte and Andre Stang

16 Automated Soft Agar Colony Formation Assay for theHigh-Throughput Screening of Malignant CellTransformation 309Pablo Steinberg

17 High-Throughput Quantification of Morphologically TransformedFoci in Bhas 42 Cells (v-Ha-ras Transfected BALB/c 3T3) UsingSpectrophotometry 317Kiyoshi Sasaki, Ayako Sakai, and Noriho Tanaka

PART IV HIGH-THROUGHPUT ASSAYS TO ASSESSREPRODUCTIVE TOXICITY, CARDIOTOXICITY,AND HAEMATOTOXICITY

18 ReProGlo: A New Stem-Cell-Based High-Throughput Assay toPredict the Embryotoxic Potential of Chemicals 343Frederik Uibel and Michael Schwarz

19 Embryonic Stem Cell Test (EST): Molecular Endpoints TowardHigh-Throughput Analysis of Chemical Embryotoxic Potential 357Peter T. Theunissen, Esther de Jong, Joshua F. Robinson,and Aldert H. Piersma

20 Zebrafish Development: High-Throughput Test Systems to AssessDevelopmental Toxicity 371Stephanie Padilla

21 Single Cell Imaging Cytometry-Based High-Throughput Analysisof Drug-Induced Cardiotoxicity 385Min Jung Kim and Joon Myong Song

22 High-Throughput Screening Assays to Evaluate the CardiotoxicPotential of Drugs 403Carl-Fredrik Mandenius and Thomas Meyer

23 High-Throughput Screening Assays to Evaluate the HematotoxicPotential of Drugs 421Caroline Haglund, Rolf Larsson, and Martin Hoglund

viii CONTENTS

PART V HIGH-THROUGHPUT ASSAYS TO ASSESS DRUGMETABOLISM AND RECEPTOR-RELATED TOXICITY

24 High-Throughput Enzyme Biocolloid Systems for DrugMetabolism and Genotoxicity Profiling Using LC–MS/MS 433James F. Rusling and John Schenkman

25 Higher-Throughput Screening Methods to Identify CytochromeP450 Inhibitors and Inducers: Current Applications and Practice 453David M. Stresser and George Zhang

26 High-Throughput Yeast-Based Assays to Study Receptor-MediatedToxicity 479Johanna Rajasarkka and Marko Virta

27 Evaluating the Peroxisomal Phenotype in High ContentToxicity Profiling 501Jonathan Z. Sexton and Kevin P. Williams

28 A Panel of Quantitative Calux R© Reporter Gene Assays forReliable High-Throughput Toxicity Screening of Chemicals andComplex Mixtures 519Bart van der Burg, Sander van der Linden, Hai-yen Man, Roos Winter, LydiaJonker, Barbara van Vugt-Lussenburg, and Abraham Brouwer

29 DR-Calux R©: A High-Throughput Screening Assay for theDetection of Dioxin and Dioxin-Like Compounds in Food and Feed 533Barbara van Vugt-Lussenburg, Harrie T. Besselink, Bart van der Burg, andAbraham Brouwer

INDEX 547

PREFACE

Conventional approaches to toxicity testing of chemicals and drugs are often decadesold, costly, do not allow high-throughput testing, and are of questionable value whenwanting to estimate human risk. The publication of the document entitled “ToxicityTesting in the 21st Century: A Vision and Strategy” by the US National ResearchCouncil and the implementation of the European legislation on the Registration, Eval-uation, Authorisation and Restriction of Chemicals (REACH) have led to a paradigmshift regarding the strategy to be pursued when evaluating the toxic potential of chem-icals and drugs. Namely, toxicity evaluation should be preponderantly performed byusing high-throughput in vitromethods and toxicity testingmethods in animals shouldplay, if at all, a minimal role.The book gives an overview on a variety of high-throughput screening methods

being used in toxicity testing nowadays and should be of help to all scientists workingin the field of toxicity evaluation and risk assessment of chemicals and drugs inchemico-pharmaceutical as well as biotechnology companies, contract laboratories,academia as well as regulatory agencies. The book chapters are written in such a waythat they lend support to those wanting to establish these methods in their laboratoriesas well as those having to evaluate the data generated. Each chapter describes theprinciple of the method and includes detailed information on data generation, dataanalysis, and the use/application(s) in risk assessment. Moreover, the chapters notonly list the advantages of the high-throughput method over the “conventional”methods used up to now in safety evaluation of chemicals and drugs but also pointout limitations and pitfalls.The book is divided into five parts. Part I includes the strategies pursued nowa-

days to predict the toxicity potential of chemicals and drugs through high-throughputbioactivity profiling, the incorporation of human dosimetry and exposure data into

ix

x PREFACE

high-throughput in vitro toxicity screening, and the use of human embryonic stemcells in high-throughput toxicity assays. Part II presents a variety of high-throughputassays to assess different cytotoxicity endpoints; Part III describes high-throughput assays to assess DNA damage and carcinogenesis; Part IV includes high-throughput assays to assess reproductive toxicity, cardiotoxicity, and hematotoxicity;and Part V presents high-throughput assays to assess drug metabolism and receptor-related toxicity. By including all these above-mentioned aspects, the book should beof great value to toxicologists, pharmacologists, analytical chemists, and pharmaceu-tical scientists working in academic institutions, industry, and regulatory agenciesthat are involved in safety evaluation and risk assessment of chemicals and drugsand an excellent complement to the current literature on toxicology in general andsafety evaluation/risk assessment in particular. Because of the test systems and thetoxicity endpoints described, this book could also be extremely interesting for all sci-entists working in the fields of biochemistry, cell biology, molecular biology, systemsbiology, and computational toxicology.I hereby would like to thank all authors for their excellent contributions. Only

because of them it was possible to conceive a book including such a broad spectrumof toxicity testing methods. The development of high-throughput methods to screenthe toxic potential of drugs and chemicals is a rapidly evolving field. If the one orthe other method was missed, then this omission was not on purpose and an incentiveto actualize this version of the book in the future.

Pablo Steinberg

CONTRIBUTORS

Harrie T. Besselink, BioDetection Systems BV, Amsterdam, The Netherlands

Jorg Blumel, MedImmune, Gaithersburg, MD, USA

Abraham Brouwer, BioDetection Systems BV, Amsterdam, The Netherlands,and Department of Animal Ecology, VU University Amsterdam, Amsterdam,The Netherlands

Bart van der Burg, BioDetection Systems BV, Amsterdam, The Netherlands

Alexander Burkle, Molecular Toxicology Group, Department of Biology, Univer-sity of Konstanz, Konstanz, Germany

Nathan J. Evans, Research Department, Promega Corporation, Madison, WI, USA

Francesca de Giorgi, FluoFarma, Pessac, France

Caroline Haglund, Division of Clinical Pharmacology, Department of Medical Sci-ences, Uppsala University, Uppsala, Sweden

Bram Herpers, Division of Toxicology, The Leiden Amsterdam Center for DrugResearch, Leiden University, Leiden, The Netherlands

Martin Hoglund, Division of Hematology, Department of Medical Sciences, Upp-sala University, Uppsala, Sweden

G. Jean Horbach, Department of Toxicology and Drug Disposition, Merck Sharp& Dohme, Oss, The Netherlands

xi

xii CONTRIBUTORS

Keith A. Houck, National Center for Computational Toxicology, Office of Researchand Development, US Environmental Protection Agency, Research Triangle Park,NC, USA

Xin Huang, Cardiovascular Institute, Clinical Research Center, 2nd Affiliated Hos-pital at School of Medicine, Zhejiang University, Hangzhou, People’s Republic ofChina

Francois Ichas, FluoFarma, Pessac, France

Esther de Jong, Laboratory for Health Protection Research, National Institute forPublic Health and the Environment (RIVM), Bilthoven, The Netherlands, and Insti-tute for Risk Assessment Sciences, Faculty of Veterinary Medicine, Utrecht Univer-sity, Utrecht, The Netherlands

Lydia Jonker, BioDetection Systems BV, Amsterdam, The Netherlands

Richard S. Judson, National Center for Computational Toxicology, Office ofResearch and Development, US Environmental Protection Agency, Research Tri-angle Park, NC, USA

Min Jung Kim, College of Pharmacy, SeoulNationalUniversity, Seoul, SouthKorea

Nadine Krause, Nonclinical Safety, Merz Pharmaceuticals GmbH, Frankfurt(Main), Germany

Rolf Larsson, Division of Clinical Pharmacology, Department of Medical Sciences,Uppsala University, Uppsala, Sweden

Sander van der Linden, BioDetection Systems BV, Amsterdam, The Netherlands

Yi-jia Lou, Institute of Pharmacology, Toxicology and Biochemical Pharmaceutics,Zhejiang University, Hangzhou, People’s Republic of China

Hai-yen Man, BioDetection Systems BV, Amsterdam, The Netherlands

Carl-Fredrik Mandenius, Division of Biotechnology, Department of Physics,Chemistry and Biology, Linkoping University, Linkoping, Sweden

Matthew T. Martin, National Center for Computational Toxicology, Office ofResearch and Development, US Environmental Protection Agency, Research Tri-angle Park, NC, USA

Thomas Meyer, Multi Channel Systems GmbH, Reutlingen, Germany

Richard A. Moravec, Research Department, Promega Corporation, Madison, WI,USA

Marıa Moreno-Villanueva, Molecular Toxicology Group, Department of Biology,University of Konstanz, Konstanz, Germany

Andreas Natsch, Givaudan Schweiz AG, Duebendorf, Switzerland

Andrew L. Niles, Research Department, Promega Corporation, Madison, WI, USA

CONTRIBUTORS xiii

Satomi Onoue, Department of Pharmacokinetics and Pharmacodynamics, Schoolof Pharmaceutical Sciences, University of Shizuoka, Shizuoka, Japan

Stephanie Padilla, Integrated Systems ToxicologyDivision, USEnvironmental Pro-tection Agency, Research Triangle Park, NC, USA

Kamala Pant, BioReliance Corporation, Rockville, MD, USA

Aldert H. Piersma, Laboratory for Health Protection Research, National Institutefor Public Health and the Environment (RIVM), Bilthoven, The Netherlands, andInstitute for Risk Assessment Sciences, Faculty of Veterinary Medicine, UtrechtUniversity, Utrecht, The Netherlands

Johanna Rajasarkka, Department of Food and Environmental Sciences, Universityof Helsinki, Helsinki, Finland

David M. Reif, National Center for Computational Toxicology, Office of Researchand Development, US Environmental Protection Agency, Research Triangle Park,NC, USA

Ann M. Richard, National Center for Computational Toxicology,Office ofResearchand Development, US Environmental Protection Agency, Research Triangle Park,NC, USA

Terry L. Riss, Research Department, Promega Corporation, Madison, WI, USA

Joshua F. Robinson, Laboratory for Health Protection Research, NationalInstitute for Public Health and the Environment (RIVM), Bilthoven, TheNetherlands, and Department of Toxicogenomics, Maastricht University, Maastricht,The Netherlands

James F. Rusling, Department of Chemistry, University of Connecticut, Storrs, CT,USA, and Department of Cell Biology, University of Connecticut Health Center,Farmington, CT, USA

Ayako Sakai, Laboratory ofCell Carcinogenesis,Division ofAlternativeToxicologyTest, Hatano Research Institute, Food and Drug Safety Center, Hadano, Kanagawa,Japan

Kiyoshi Sasaki, Laboratory of Cell Carcinogenesis, Division of Alternative Tox-icology Test, Hatano Research Institute, Food and Drug Safety Center, Hadano,Kanagawa, Japan

John Schenkman, Department of Cell Biology, University of Connecticut HealthCenter, Farmington, CT, USA

Willem G.E.J. Schoonen, Department of Toxicology and Drug Disposition, MerckSharp & Dohme, Oss, The Netherlands

Michael Schwarz, Department of Toxicology, Institute of Experimental andClinicalPharmacology and Toxicology, University of Tubingen, Tubingen, Germany

xiv CONTRIBUTORS

Yoshiki Seto, Department of Pharmacokinetics and Pharmacodynamics, School ofPharmaceutical Sciences, University of Shizuoka, Shizuoka, Japan

Jonathan Z. Sexton, Biomanufacturing Research Institute and Technology Enter-prise, North Carolina Central University, Durham, NC, USA

Imran Shah, National Center for Computational Toxicology, Office of Researchand Development, US Environmental Protection Agency, Research Triangle Park,NC, USA

Joon Myong Song, College of Pharmacy, Seoul National University, Seoul, SouthKorea

Andre Stang, Institut fur Biologie und Umweltwissenschaften, Carl von OssietzkyUniversitat Oldenburg, Oldenburg, Germany

Pablo Steinberg, Institute for Food Toxicology and Analytical Chemistry, Univer-sity of Veterinary Medicine Hannover, Hannover, Germany

Joe C.R. Stevenson, Department of Toxicology and Drug Disposition, Merck Sharp& Dohme, The Netherlands

David M. Stresser, Corning GentestSM Contract Research Services, Corning LifeSciences - Discovery Labware, Woburn, MA, USA

Noriho Tanaka, Laboratory of Cell Carcinogenesis, Division of Alternative Tox-icology Test, Hatano Research Institute, Food and Drug Safety Center, Hadano,Kanagawa, Japan

Peter T. Theunissen, Laboratory for Health Protection Research, National Institutefor Public Health and the Environment (RIVM), Bilthoven, The Netherlands, andDepartment of Toxicogenomics, Maastricht University, Maastricht, The Netherlands

Russell S. Thomas, The Hamner Institutes for Health Sciences, Research TrianglePark, NC, USA

Frederik Uibel, Department of Toxicology, Institute of Experimental and ClinicalPharmacology and Toxicology, University of Tubingen, Tubingen, Germany

Luc Verschaeve, Scientific Institute of Public Health, Operational Direction PublicHealth & Surveillance, Laboratory of Toxicology, Brussels, Belgium

Marko Virta, Department of Food and Environmental Sciences, University ofHelsinki, Helsinki, Finland

Barbara van Vugt-Lussenburg, BioDetection Systems BV, Amsterdam, TheNetherlands

Beppy van de Waart, Department of In Vitro & Environmental Toxicology, WILResearch, DD ’s-Hertogenbosch, The Netherlands

Bob van de Water, Division of Toxicology, The Leiden AmsterdamCenter for DrugResearch, Leiden University, The Netherlands

CONTRIBUTORS xv

Femke M. van de Water, Department of Toxicology and Drug Disposition, MerckSharp & Dohme, Oss, The Netherlands

Walter M.A. Westerink, Department of In Vitro and Environmental Toxicology,WIL Research, DD’s-Hertogenbosch, The Netherlands

Barbara A. Wetmore, The Hamner Institutes for Health Sciences, Research Trian-gle Park, NC, USA

Kevin P. Williams, Biomanufacturing Research Institute and Technology Enter-prise, North Carolina Central University, Durham, NC, USA

Roos Winter, BioDetection Systems BV, Amsterdam, The Netherlands

Irene Witte, Institut fur Biologie und Umweltwissenschaften, Carl von OssietzkyUniversitat Oldenburg, Oldenburg, Germany

Tracy J. Worzella, ResearchDepartment, PromegaCorporation,Madison,WI,USA

Shizuo Yamada, Department of Pharmacokinetics and Pharmacodynamics, Schoolof Pharmaceutical Sciences, University of Shizuoka, Shizuoka, Japan

George Zhang, Corning GentestSM Contract Research Services, Corning LifeSciences – Discovery Labware, Woburn, MA, USA

Dan-yan Zhu, Institute of Pharmacology, Toxicology and Biochemical Pharmaceu-tics, Zhejiang University, Hangzhou, People’s Republic of China

PART I

GENERAL ASPECTS

High-Throughput Screening Methods in Toxicity Testing, First Edition. Edited by Pablo Steinberg.© 2013 John Wiley & Sons, Inc. Published 2013 by John Wiley & Sons, Inc.

1ToxCast: PREDICTING TOXICITYPOTENTIAL THROUGHHIGH-THROUGHPUT BIOACTIVITYPROFILING

Keith A. Houck, Ann M. Richard, Richard S. Judson,Matthew T. Martin, David M. Reif, and Imran Shah

1.1 INTRODUCTION

Chemical safety assessment has long relied on exposing a few species of laboratoryanimals to high doses of chemicals and observing adverse effects. These results areextrapolated to humans by applying safety factors (uncertainty factors) to account forspecies differences, susceptible sub-populations, establishing no observed adverseeffect levels (NOAEL) from the lowest observed adverse effect levels, and data gapsyielding theoretically safe exposure limits. This approach is often criticized for lackof relevance to human health effects due to the many demonstrated differences inphysiology, metabolism, and toxicological effects between humans and rodents orother laboratory animals [1]. Such criticism exists mainly due to the lack of knowl-edge of specific mechanisms of toxicity and whether these are relevant to humans.Toxicological modes of action (MOA) have been elucidated for only a limited numberof chemicals; even fewer chemicals have had their specific molecular mechanismsof action determined. Having such detailed knowledge would facilitate higher con-fidence in species extrapolation and setting of exposure limits. However, tens ofthousands of chemicals currently in commerce and with some potential for humanexposure lack even traditional toxicity testing and much less elucidated modes ormechanisms of toxicity [2]. Understanding mechanisms of toxicity usually results


3

4 ToxCast

from decades-long research dedicated to single chemicals of interest, a model unsuit-able for such vast numbers of chemicals. Even with dedicated research, such effortsare not guaranteed to succeed; the extended focus on understanding the mechanism oftoxicity of 2,3,7,8-tetrachlorodibenzodioxin (TCDD) is an example [3]. Traditionalanimal testing, in addition to the criticisms discussed above, is not appropriate or fea-sible for the large numbers of untested chemicals due to the high costs and numberof animals required [1].One major effort to address this dilemma by providing a high-capacity alternative

is underway, facilitated by integration of the fields of computational toxicology andhigh-throughput in vitro testing [4, 5]. The ultimate goals of this approach are themeans to screen and prioritize thousands of chemicals, predict the potential for humanhealth effects, and derive safe exposure levels for themyriad of chemicals to whichweare exposed. This approach relies on a shift in toxicology research away from “black-box” testing on whole animals and toward an understanding of the direct interactionsof chemicals with a broad spectrum of potential toxicity targets comprising specificmolecular entities and cellular phenotypes. This bioactivity profiling of chemicalsgenerated through the use of high-throughput approaches produces characteristicchemical response profiles, or signatures,whichmaydescribe the potential for toxicityof that chemical [6].Computational analysis and modeling of the results are required to provide insight

into complex datasets and support the development of predictive toxicity algorithmsthat ultimately may serve as the foundation of an in vitro toxicity testing approachreplacing most or all animal testings. The groundwork required for a computationaltoxicology approach is the generation of datasets comprising the quantitative effectsof chemicals on biological targets. Two types of data are required. The first are the testresults from in vitro and/or in silico assays that can be run in high-throughput modeand provide bioactivity profiles for hundreds to thousands of chemicals. The secondis a dataset that details the effects of these chemicals on whole organisms, ideally thespecies of interest. These data are used to anchor and build predictive models thatcan then be applied to chemicals that lack in vivo testing. Generation of the in vitrodataset has become feasible andwidely available as high-throughput in vitro screeningtechnology, developed in support of the drug discovery community. The selection anduse of these assays for computational toxicology will be discussed further in Section4. Obtaining the latter dataset of in vivo effects necessary to build the computationalmodels presents unique challenges. Although thousands of chemicals have beentested using in vivo approaches, only a limited amount of this information has beenreadily available. Much of it lies in formats not readily conducive to computationalanalysis, for example, paper records, in the data stores of private corporations, orprotected by confidentiality clauses [7], and generation of extensive new in vivo datato support the approach is cost prohibitive. The access and collation of these datainto a relational database useful for computational toxicology will be discussed inSection 5.Beyond the technical aspects of generating the data, assembling the collection of

required datasets to support computational approaches is a challenging task in itself.Robust, efficient, and accurate knowledge discovery from large datasets require a

INTRODUCTION 5

robust data infrastructure. There are a number of critical steps in the process begin-ning with designing an underlying architecture to manage the data. Appropriate datamust be selected and preprocessed into common formats usable to computer programs(e.g., standardized field names for the types of attributes beingmeasured, standardizedchemical names and links to other data sources). The use of standardized ontologiescan be particularly useful in the sharing of information across organizations [8].Because of the complexities of achieving this on a large scale, these approaches areperhaps best conducted by large organizations with access to computational scientistsin addition to experts in chemistry, toxicology, statistics, and high-throughput screen-ing (HTS). Examples of integration of these diverse areas of expertise include theU.S.Environmental Protection Agency’s (EPA) ToxCast program [4] and the Tox21 col-laboration between the EPA, theNational Toxicology Program, theNational Institutesof Health Center for Translational Therapeutics—NCTT (formally the NIHChemicalGenomics Center [NCGC]), and the U.S. Food and Drug Administration [9, 10]. Inaddition, a number of large pharmaceutical companies have internal programs in thisarea relying on their own, extensive in-house expertise [11, 12].As described, the ultimate goal is to use high-throughput in vitro assays to rapidly

and inexpensively profile the bioactivity of chemicals of unknown toxicity and makepredictions about their potential for causing various adverse endpoints [4]. Achievinga robust, predictive toxicology testing program is a long-range goal that will need toproceed through a number of systematic stages including proof-of-concept, extensionof chemical and bioassay diversity, refinement, and ultimately, supplementation orreplacement of existing methods. The initial stage involves multiple steps including(1) selecting an appropriate chemical test set for which in vivo data are available; (2)selecting high-throughput biological assays for screening the chemicals; (3) generat-ing the screening data on the chemicals; (4) collating the in vivo anchoring data for thechemicals; and (5) building up predictive models. Such models can then be validatedthrough testing of additional chemicals with known toxicity endpoints to determinethe robustness of the models. It is likely that the development of the test systems,as well as the computational models, will be an iterative process. New biologicalassays and statistical approaches are evaluated for potential inclusion in the program,whereas assays and models not producing useful results are dropped.The success of this stage of the process would be models judged useful for

prioritizing chemicals for the potential to cause specific toxicity endpoints. Thisprioritization will be valuable in the short term by allowing focused and limitedin vivo use of testing resources on chemicals most likely to be of concern. Theresults of targeted testing of designated chemicals for specific endpoints shouldensure a reduced use of test animals as only limited endpoints would need to beevaluated. This targeted testing will also provide an additional validation method forthe testing program, that is, do the adverse endpoints predicted by the models occurto a significant extent in the tested chemicals? Ultimately, refinement of the testingand modeling approaches should allow high-confidence prediction of the likelihoodfor toxicity, thereby avoiding animal testing altogether for many chemicals. Theremainder of this chapter will focus more specifically on providing background onthe steps undertaken in developing the initial stages of the ToxCast testing program at

6 ToxCast

EPA, as well as examples of applications of the program in prioritizing environmentalchemicals for multiple toxicity endpoints.

1.2 CHEMICAL LANDSCAPE

A major driver of the development and use of HTS methods in toxicology is thescope of the chemical problem, that is, tens of thousands of chemicals to whichindividuals are potentially exposed, the majority of which have never been tested inany significant way [2] . What chemicals are of interest and the kind of data that islikely to be available depends on the use of the chemical, which in turn is relatedto the regulations to which the chemicals are subjected. To understand the world ofchemicals that are of concern for potential toxicity and candidates for testing, it isuseful to discuss a set of chemical inventories, some of which are overlapping.

1.2.1 Pesticide Active Ingredients

These are typically the active compounds in pesticide formulations, which aredesigned to be toxic against select types of organisms. A related category of com-pounds falling under this general label are antimicrobials, which are also designed tobe toxic to certain organisms, in this case-targeting fungi or bacteria. These groups ofchemicals are further divided into food-use and nonfood-use actives for the purposeof regulation. EPA sets tolerance levels for pesticides that may be used in specificfoods, for particular reasons, and at particular exposure levels. Thus, EPA regulatesthe maximum amount of pesticide residue permitted to remain on a food approvedfor pesticide application. FDA, in contrast, has the authority to monitor and enforcelevels of food-use pesticides and ensure that they comply with EPA regulations.FDA has additional authority regarding the use of antimicrobials in food packaging[13]. Food-use pesticide actives have the highest data requirements and, for these,a company will typically generate data from 2-year chronic/cancer bioassays in ratsand mice, developmental toxicity studies in rats and rabbits, multigenerational repro-ductive toxicity studies in rats, and other specialized in vivo studies [14]. These aresimilar to the complete set of preclinical studies that are required for human phar-maceuticals. Because of this large data requirement, these chemicals are ideal foruse in building up toxicity prediction models, since one will have near-complete invitro and in vivo datasets. It is not surprising that pesticide actives have some of thesame features and chemical properties as pharmaceutical products, given that theyare often designed to interact with a specific molecular target.

1.2.2 Pesticidal Inerts

These are all of the ingredients in a pesticide product or formulation other than theactive ingredients. Although they are labeled as “inert”, there is no requirement thatthey be nontoxic. These can range from solvents (e.g., benzene) to animal attractants,such as peanut butter or rancid milk. As with the actives, inerts are classified as

CHEMICAL LANDSCAPE 7

food-use and nonfood-use. Regulatory data requirements are, in general, limited,thus resulting in the availability of little in vivo data [15].

1.2.3 Industrial Chemicals

This is an extremely broad class of chemicals including solvents, detergents, plas-tic monomers and polymers, fuels, synthesis intermediates, and dyes. As such, theyare typically not designed to be bioactive, although many do have bioactivity, some-times through interaction with enzymes and receptors, or by chemically reacting withbiomolecules or via physical interactions (e.g., by disrupting cell membranes). Manyof these compounds aremanufactured in very large quantities, posing greater potentialrisks. Such chemicals typically have less stringent regulatory oversight and toxicitytesting requirements but are subject to reporting rules under the Toxics SubstancesControl Act (TSCA). Under TSCA, different reporting requirements and regulatoryscrutiny are applied depending on production volume levels (MPV—medium produc-tion volume chemicals,>25 K tons/year; HPV—high-production volume chemicals,>1 M tons/year). On average, these industrial compounds have lower molecularweight than pesticidal actives or pharmaceuticals, and include many more volatileand semivolatile compounds.

1.2.4 Pharmaceuticals

These are the active ingredients in drugs and, hence, are designed to have specificbioactivity. It is well known that many drugs have toxic side effects, often throughunexpected off-target interactions, and that this is a major economic concern forthe pharmaceutical industry driving up the costs of drug development. In addition,there is increasing concern for toxicity, not only for patients directly taking thedrug, but also for ecological species exposed to these compounds in waste water [16].Despite large amounts of toxicity data submitted to the FDA during the drug-approvalprocess, including clinical data on humans if the drug reaches clinical trials, as wellas additional preclinical toxicity data generated within the pharmaceutical industry,little of these data see the light of day due to confidentiality concerns. As a result,public availability of toxicity data on pharmaceuticals is generally limited to what isavailable in the open literature.

1.2.5 Food Additives/Ingredients

This category includes both natural and synthetic small molecules that are inten-tionally added to food, often to enhance nutritional value (e.g., vitamins), to act aspreservatives, such as in food packaging, or to enhance color or texture. FDA regu-lates allowed tolerances for such chemicals and has the authority to require a batteryof in vitro (primarily genotoxicity) and in vivo toxicity testing to support such reviewswithin the Center for Food Safety and Nutrition (CFSAN) [17]. Such data can bemade publicly available, hence providing a potentially rich source of additional invivo data for computational toxicology modeling.

8 ToxCast

1.2.6 Water Contaminants

EPA regulates chemicals in surface and drinking water, and the relevant chemicalsinclude any of the above categories that enter the water system, as well as metabolitesor degradation products. One example of the latter is disinfection byproducts thatcan result from reactions of chlorine with organic molecules in a drinking watersystem to produce polychlorinated organic compounds. The regulatory authority inthis instance is reactive. First, a chemical has to be detected in water at sufficient levelsto cause some concern, and then sufficient scientific justification must be provided towarrant regulatory action. As a result, toxicity data is generally lacking for many ofthese chemicals, similar to the situation for industrial chemicals.Because there are so many chemicals to which humans and ecological species are

potentially exposed, it is necessary to prioritize among them when setting up a large-scale screening program such as ToxCast or Tox21. The potential for exposure is onecritical aspect of this prioritization, and these and further chemical use-categories areimportant indicators of the potential for exposure. For instance, any chemical that isdirectly in food or water (e.g., food additives or pesticides that leave residues on cropsor chemicals found in drinking water) would have extra weight in a prioritizationscheme. More detailed “use-categories” are also available to help refine estimatesof potential exposure routes. For instance, if a chemical is found in products towhich children are exposed (e.g., baby bottles, clothing), that chemical would havea heightened priority for screening. There is no general mapping of chemicals touse-categories that is publicly available, but the ExpoCast project, affiliated withthe ToxCast project within EPA, is currently developing such a mapping based onmerging data from many different sources [18].The lack of data availability on chemicals, whether it is use-category, exposure

potential, or toxicity data, is one of the major drivers of EPA’s HTS computationaltoxicology program [4]. However, the success of this effort also relies upon the abilityto collate as much available data as possible and systematize and format these datainto computable forms to enable modeling efforts to proceed. To provide a centralresource to support this effort, a large-scale database is being created to gather allpublicly available data on chemicals in the environment through the AggregatedComputational Toxicology Resource (ACToR) effort [19]. Thus far, varying amountsand types of data have been compiled on several hundred thousand chemicals col-lected from over 1,000 different sources, consisting of data types that, for example,include information on hazard (i.e., in vitro and in vivo toxicity data), exposure, use,and production.The above discussion focuses on the chemical landscape of concern for testing

from a regulatory and use or exposure perspective, but an equally important consid-eration for our long-range purposes is providing adequate coverage of the chemicalfeature and property landscape spanned by the various use-category inventories ofchemicals. Given the intimate relationship between the chemical structure and its bio-logical activity, building a computational toxicology approach capable of predictingtoxicity from HTS bioactivity profiles must provide for sufficient coverage of bio-logical pathways and toxicity mechanisms across the chemical landscape of interest.

THE CHEMICAL LIBRARIES 9

This means that a chemical testing library must also provide sufficient coverage ofthe diverse chemical property and features space capable of adequately probing thisbiological mechanism diversity.

1.3 THE CHEMICAL LIBRARIES

To generate the in vitro dataset required for the computational toxicology approach,a chemical library was assembled, with initial and later testing candidates largelydrawn from the chemical inventories described above.Meeting the initial objectives ofproviding proof-of-concept of the HTS computational toxicology approach requireda strong anchoring to in vivo animal toxicity studies. Hence, selection of the initialtesting set for ToxCast, which we refer to as the Phase I chemical library, wasprimarily driven by the availability of detailed, in vivo toxicity data. The existence ofhigh-quality regulatory guideline studies required for chemical safety evaluation ofpesticide active ingredients by EPA motivated the selection of these compounds tofulfill these data requirement needs. Thus, the Phase I library consisted of 309 uniquechemical substances, with more than 90% pesticides and the rest a mixture of invivo data-rich industrial chemicals such as bisphenol A (BPA) and perfluorooctanoicacid (PFOA).

In vitro HTS testing procedures additionally have a number of practical require-ments that affect the types of chemicals that can be tested using current technologies.Obvious concerns are the solubility of the chemical in aqueous buffer, which is themedium used to conduct HTS testing, as well as dimethyl sulfoxide (DMSO), whichis the near universal solvent used to solubilize test chemicals for testing. Addition-ally, volatility is a concern, since the chemicals are run in batch mode and attentioncannot be paid to special handling requirements for volatile or semivolatile chem-icals. A few physical–chemical property filters, primarily molecular weight (MW)and octanol/water partition coefficient (logP), were used to choose the Phase I chem-icals, but the structures of pesticides are such that most met the criteria for inclusionand were soluble in DMSO. The ToxCast Phase I chemical solutions that under-went the initial round of HTS testing were also post-analyzed by analytical qualitycontrol (QC) methods that are amenable to high-throughput application (primarilyliquid chromatography–mass spectrometry [LC/MS] with gas chromatography–massspectrometry [GC/MS] follow-up for compounds not suitable for LC/MS analysis).Identity and purity were confirmed for over 80% of the Phase I library, with themajor-ity of the remaining compounds deemed unsuitable for analysis because they weremetal containing or of low MW. One class of pesticides, consisting of 14 sulfurons,was found to significantly dissociate in DMSO over time, motivating the removal ofthese compounds from further ToxCast testing.The ToxCast Phase I chemical library, despite its relatively small size, contained

a significant amount of chemical and functional diversity, spanning over 40 chem-ical functional classes (e.g., pyrazoles, sulfonamides, organochlorines, pyrethroids,carbamates, organophosphates) and 24 known pesticidal functional classes (e.g.,phenylurea herbicides, organophosphate insecticides, dinitroaniline herbicides). The

10 ToxCast

implication is that although the particular compounds included in this Phase I test setmay not be representative of the larger chemical universe of potential interest suchas antimicrobials, food-additives, drugs, and industrial compounds, the constituentfeatures of these chemicals are potentially capable of representing a much broaderset of chemicals from a wide range of use-categories.Clearly, however, in order to meet the larger objectives of the ToxCast program

for modeling in vivo toxicity, it is necessary to test larger chemical inventories thatinclude greater representation of the various use-categories of high interest, as wellas the more varied chemical and biological interactions that must be probed andcharacterized in order to build general models for predicting toxicity. Following thetesting of the Phase I library, a much larger chemical collection was assembled basedon these considerations for the dual purposes of expanding the ToxCast test libraryand constructing the EPA contribution to the Tox21 library. Nominations for thislibrary were broadly drawn from the previously described inventories and initiallyexceeded 9,000 compounds. Given the much larger structural diversity of the chem-icals nominated, a greater number of compounds were excluded from considerationon the basis of calculated physical–chemical properties, such as MW, vapor pres-sure, boiling point, solubility, and logP. Finally, practical considerations pertainingto physical samples, such as cost, availability, actual solubility in DMSO, and con-firmed volatility, determined whether or not a compound was included in the finalEPA Tox21 inventory, consisting of more than 3,700 unique chemical substances.The ToxCast Phase II chemical library, currently undergoing testing, consists of

776 unique chemical substances, including nine Phase I compounds used as testingreplicates, drawn from the expanded EPATox21 chemical inventory, spanning amuchbroader range of use-cases and chemical structures than in Phase I. For the selection ofPhase II compounds, significant weight was given to those substances with extensivein vivo data available, as well as to toxicity reference substances with well-definedactivities and mechanisms of action. Pursuant to this goal, approximately 30% of thePhase II compounds have in vivo data available from theNational ToxicologyProgramor were generated tomeet EPA or FDA’s regulatory requirements for pesticide or foodadditives. However, due to the relative paucity of data for many of the use-categoriesdescribed previously, many of the chemicals in this expanded collection had relativelylittle or no such data available. In addition, higher weight was given to chemicalson high-interest EPA inventories (such as listed above), as well as to chemicalsthat appeared on multiple inventories or use-categories. The Phase II inventory alsobenefitted from an unprecedented collaboration between EPA and the pharmaceuticalindustry, whereby 135 “failed drugs” were donated by six pharmaceutical companies(Pfizer,Merck, GlaxoSmithKline, Sanofi, Roche, andAstellas), alongwith preclinicaland, in some cases, human clinical data reporting adverse effects. The value of thesedata in extending findings made on chemicals tested only in laboratory animals tothose tested in humans may be significant.The ToxCast Phase I and Phase II inventories total 1,060 unique compounds.

These compounds are being run in the full suite of more than 500 ToxCast assays.Both of these chemical inventories are fully contained within the EPA Tox21 chem-ical inventory, which in turn is a subset of the complete Tox21 collection, totaling

THE BIOLOGICAL ASSAYS 11

approximately 8,200 unique chemical structures. In addition to the failed pharmaceu-ticals, the Tox21 library contains an extensive collection of human pharmaceuticals[20]. Although the Tox21 inventory is much larger and spans much greater chemicaldiversity, this library will only be tested in HTS assays being run at the NCTT and,thus, will have more limited bioactivity profiling data available. On the other hand,the smaller ToxCast Phase I and II chemical inventories will be run in the full suite ofToxCast assays, as well as in the additional Tox21 assays, providing a rich chemicaland biological context for the interpretation of these data. Details of the chemicallibraries can be accessed at http://www.epa.gov/ncct/toxcast/chemicals.html.An expanded analytical quality control process to ensure that the tested chemicals

are indeed what they are intended to be is accompanying the full Tox21 effort.Careful review and curation of chemical identifiers, including names and ChemicalAbstracts Service Registry Numbers (CASRN), as well as reported purity wereextracted from Certificates of Analysis at the time of procurement. Further reviewand chemical structure annotation of the full Tox21 inventory and component ToxCastinventories were carried out within EPA’s Distributed Structure-Searchable Toxicity(DSSTox) project (see http://www.epa.gov/ncct/dsstox/ for access to downloadablestructure files). Following solubilization in DMSO, the chemical identity, purity, and,concentration are determined by appropriate analytical techniques, including LC/MSand follow-up GC/MS. This analysis will be repeated over the course of the use ofthe library to assess compound stability during testing. While complex and costly,such efforts ensure that biological activity measured in an assay is associated with theappropriate chemical structure and, conversely, those negative results are associatedwith a chemical structure only if that chemical was indeed present.

1.4 THE BIOLOGICAL ASSAYS

Selection of in vitro assays for toxicity testing would be relatively straightforwardif the molecular targets underlying mechanisms of toxicity were known. Advancesin HTS technologies to support the drug discovery industry have provided the toolsto develop assays for large numbers of biological targets, ranging from receptors toenzymes to ion channels and more. If a protein has a defined function, it is safe tosay that an in vitro assay can be built to measure effects of chemicals on that function.Techniques such as surface plasmon resonance or LC–MS–MS exist that measurechemical–protein interactions even when the function is unknown [21]. Beyondassays focusing on specific molecular targets, many assays are available to probephenotypic changes induced in cells by chemical exposure including effects onorganelles and cellular structures such as mitochondria, nuclei, cytoskeleton, andcell membrane. Again, with advances in automated fluorescent microscopy screen-ing platforms and associated imaging algorithms, the ability to measure altered cel-lular phenotypes is almost unlimited. However, assays targeting specific proteinsor cellular phenotypes suffer from our lack of detailed knowledge with respect tomechanisms of toxicity that would guide high-confidence assay selection. Excep-tions to this, while clear, are relatively few and include molecular targets such as the

12 ToxCast

potassium ion channel hERG [22], acetylcholinesterase [23], cytochrome P450s [24],drug transporters [25], nuclear receptors including the androgen, estrogen, and arylhydrocarbon receptor (AhR) [26], as well as the 5-HT2b G-protein-coupled recep-tor (GPCR) [27]. In addition, cellular phenotypic assays for genotoxicity, oxidativestress, mitochondria energy homeostasis, calcium release from intracellular stores,and necrotic and apoptotic cell death can be used to determine toxicity, although withless specificity with respect to molecular target. Acceptance of these as valid toxic-ity targets usually resulted from many years of research, sometimes combined withserendipitous findings. Continuing with this model to complete our understanding oftoxicology would be a long, expensive, and arduous route.As an alternative approach, a broadly based interrogation of important families

of biological targets and cellular phenotypes can be conducted efficiently usinghigh-throughput in vitro screens, probing them with large chemical libraries withknown animal and human health effects. The reference in vivo toxicity data for thesechemicals are needed to correlate the in vitro findings with in vivo endpoints. Thetools of computational toxicology can then be applied to analyze, interpret, andmodelthe results, ultimately generating predictive signatures of toxicity compatible withcost-efficient, high-throughput assays conducive to screening unknown chemicals.Defined toxicity targets are usually members of large protein families such as

enzymes (e.g., acetylcholinesterase), receptors (e.g., estrogen receptor), and ion chan-nels (e.g., voltage-gated sodium channels). These protein families make up the major-ity of what is called the “druggable genome”, molecular targets thought to providean opportunity for therapeutic intervention and of high interest to the pharmaceuticalindustry [28]. As a result, hundreds of HTS assays have been developed to supportthis drug discovery research. Since the vast majority of these potential drug targetshave been selected based on believed critical roles in various pathological processes,extension of this thinking suggests that such targets could also be involved in toxicitywhen inappropriately perturbed by xenobiotic chemicals. This served as the impetusfor developing a diverse suite of HTS assays to use for profiling the biological activ-ity of chemical libraries by several groups including ourselves through the ToxCastprogram [4, 11, 12].

In vitro HTS assays facilitate the rapid, parallel generation of large numbers ofindividual assay data points through the use of miniaturized assay formats, automatedliquid dispensers, and high-speed plate readers. The miniaturized assay formats areusually in multi-well plates with densities of 96, 384, or 1536 wells per plate ina single, standardized plate footprint, and use total assay volumes ranging from200 �L down to 5 nL. The assay components can be highly varied and depend to alarge degree on the biological target beingmeasured. For instance, an assaymeasuringkinase activity would have a purified kinase, required cofactors, required substrates,appropriate buffer, and chemical to be tested. In addition, a means of measuring theassay endpoint, here the phosphorylation of the substrate, is required. This couldbe a radioactive or fluorescent technique, a means to detect the loss of ATP or theincrease in ADP, or a separation of the phosphopeptide from the nonphosphorylatedone by means of mobility shift microfluidics assay technology. Cellular phenotypicassays use in vitro cultured cells and automated, fluorescence microscopy to image


chemical perturbations of cellular structures, organelles, and functions followed byspecific imaging algorithms to quantitate results [29]. Examples of this are assaysusing fluorescently tagged antibodies to actin microfilaments to monitor chemicalaffects on the stabilization or destabilization of the cytoskeleton [30].The diversity of techniques used to quantitate HTS results underscores a critical

point of understanding of HTS assays; all assays are susceptible to artifacts, anddifferent assay formats are susceptible to different types of artifacts [31]. Assayartifacts are defined as test chemical-induced events that interfere with the ability tomeasure an accurate assay result such as chemical-induced fluorescent quenching,precipitation of the biological target by chemical aggregation, and inherent chemicalfluorescence among others. Thus, an underlying caveat of any HTS assay is that allresults must be interpreted with caution. In addition to artifacts induced by specifictest chemicals, there are also experimental errors and normal assay variabilities thatcan affect the results. While HTS assays should be validated according to industrystandards (http://spotlite.nih.gov/assay/index.php/) inherent in testing large numbersof chemicals is that some results will not be accurate. Inaccurate results can generateboth false-positive and false-negative findings. Each has its own issues.For toxicity testing, it is strongly desirable to avoid false-negative results that

could miss important activities potentially resulting in endangering the health ofexposed populations. Too many false-positives, however, can invalidate the utilityof the screening by requiring extensive measures to follow-up on the large numberof active chemicals. Unfortunately, decreasing the false-negative rate is usually atthe expense of increasing the false-positive rate; thus, finding the right balance witha robust HTS assay is of high importance. Two methods of utility in providinghigh-confidence results for HTS toxicity testing are to use a concentration–responseformat for testing all chemicals and to have multiple assays using different assaytechnologies for important targets. Concentration–response testing allows testingconcentrations high enough to detect the activity of weakly active chemicals, whileminimizing concern for high concentration-induced artifacts such as cytotoxicity,which can mimic inhibition of functional activity in a cell-based assay. In addition,knowledge about the types of response expected for specific biological targets canhelp discriminate between chemicals affecting the target from those active by artifact.Receptor binding assays, for example, should follow the law of mass action andresulting concentration–response curves should display sigmoidal behavior with aslope near one on a semi-log plot [32]. Results with slopes of 10, for example,should flag the response as potentially suspect. Orthogonal assays are particularlyuseful as, for example, the use of a radioligand receptor binding assay and a cellulartransactivation assay for the estrogen receptor. Chemicals active in both would havea high degree of confidence of being truly active at the receptor site and likelyactive in vivo, assuming the chemical reaches its receptor target. The efficiency ofHTS supports both of these approaches by providing inexpensive screening methodswith sufficient capacity to screen both large numbers of chemicals and at multipleconcentrations [33].However, given the sheer numbers of possibilities, testing of all potential toxicity

targets is not feasible even with HTS technologies. Selection of the assays for testing

14 ToxCast

GPCR

Transcription factors

Cytokine/receptor

Nuclear receptor

Cytochrome P450

Other

Cellular phenotype

Protease

Cell adhesion molecule

Other enzyme

Ion channel

Gene expression

Phosphatase

Transporter

Other receptor

Kinase

FIGURE 1.1 Distribution of assays categories in the ToxCast Phase I testing battery.(See insert for color representation of this figure.)

within the ToxCast program followed a strategy of selecting targets with known linksto toxicity, for which assays were available, combined with widely sampling potentialtargets from the large protein superfamilies including GPCRs, kinases, phosphatases,nuclear receptors, chromatin-modifying enzymes, CYP P450s, ion channels, andtransporters [34]. A list of the families and numbers of assays targeting specificmolecular targets is shown in Figure 1.1. Sampling of these families may provide awindow into potential chemical activity, even when the specific target of a chemical-induced toxicity is not included. This occurs through testing in a concentration–response format, which may allow the detection of chemical promiscuity at higherconcentrations. This can be helpful when a specific target of toxicity is not included inthe assay suite. Due to conservation of protein structurewithin families, it is somewhatmore likely that a chemical will affect other closely related family members, but withdifferent affinities. These may serve as assay surrogates for the actual target and maystill be useful in developing signatures of toxicity.The use of cellular assays provides a means to include large numbers of potential

targets concurrently in a more physiologically relevant format. Such assays usuallyrely on coordinated signaling networks to carry out the downstream function beingmeasured, for example, cell proliferation. There are many nodes in the pathways reg-ulating cell proliferation that are potentially susceptible for chemical perturbation.These include growth factor receptors on the plasma membrane, kinase second mes-sengers transmitting the growth signal to the nucleus, transcriptional regulatory andprotein synthesis machinery, mitotic spindle apparatus, cytoskeletal components, andassociated regulatory enzymes. It is because of this complexity that cell proliferation


has been described as one of the most sensitive endpoints for in vitro toxicology[35]. Endpoints may be more narrowly defined, such as mitochondrial function orDNA damage endpoints, but these also have many potential upstream targets. Thus,cellular assays, in general, lack the ability to clearly identify molecular mechanismsof action. They do, however, put the molecular targets in a more physiologically rel-evant context than generally are found with cell-free, biochemical assays. A valuablestrategy is a combination of a biochemical assay, in which chemical specificity can bedefined, as well as a cellular assay, in which functional efficacy can be demonstrated.The assays used in the ToxCast program provide both a broader coverage of toxicitytargets as well as opportunities to define cellular efficacy for chemicals active in thebiochemical screens.Choosing the appropriate cell type for HTS assays supporting predictive toxicol-

ogy approaches is important to the success of the approach. Many factors need tobe considered and these vary, depending on the goal of the assay. In measuring theability of a chemical to perturb a specific molecular target such as a kinase or anuclear receptor, it may be appropriate to use standard, immortalized cell lines thatprovide robust and highly reproducible results. However, in determining the effectsof chemicals on complex signaling pathways, the use of such cells may be of littlevalue if these pathways have been altered during the immortalization process andadaptation to growth under standard cell culture conditions. In this case, the use ofprimary cells may have distinct advantages and providemore physiologically relevantinformation useful to predicting in vivo effects [36, 37]. However, the use of primarycells also has its limitations in terms of limited passage numbers, batch-to-batchvariation, difficulty to engineer with respect to introducing reporter genes, and lackof large signal-to-background ratios for the endpoints being measured.To effectively use data fromHTS assays for computational toxicology approaches,

it is very useful to acquire complete testing datasets, meaning testing all chemicalsagainst all assays in the testing set and to define standard data handling and analysisprocedures. The ToxCast project used defined chemical libraries, described earlier,in testing against suites of in vitro assays in a concentration–response format. Allchemicals were run in all assays as minimizing missing values in the data matrixgreatly enhances the value of the dataset for subsequent analysis. Screening resultswere used to generate AC50 values, the concentration at which an assay is activatedor inhibited by 50% when compared to the control values, for each chemical–assaypair. The AC50 is somewhat arbitrary in that it often has no direct toxicologicalinterpretation. However, it does provide a means of comparing chemicals within anassay, serves as a flag for activity for a chemical in a given assay, and providesinformation as to its general potency range.The concentration–response curves for ToxCast aremodeled by the four-parameter

Hill equation [38] implemented in the R programming language [39]. Heuristics areemployed to accommodate aspects of assay results that cause implementation of theHill equation to fail. The reasons underlying the curve-fitting failures may applyto all assays or be specific to a given assay or platform. For example, results thatshow no concentration-dependent increase in activity but rather maximal activityat all concentrations tested must be flagged with an AC50 less than the lowest

16 ToxCast

concentration tested. Assays susceptible to cytotoxicity at high concentrations oftenneed the responses obtained at these cytotoxic concentrations removed from the curve-fitting routine. Since it would be very difficult to generate a universally accepted bestmethod for doing curve-fitting to a wide variety of biological assays, it is importantto provide transparency to the process used, as well as access to the underlyingunprocessed data in order for others to apply their own techniques.The combination of chemicals and assay AC50 values defines the basic data matrix

required for the computational toxicology input. Value can be added to the matrixthrough additional metadata. One very useful type of metadata is the mapping of theassays to specific gene ontologies, which are tied to biological pathways annotatedby databases such as GO or KEGG [40]. This bioinformatics approach links chemicaleffects to biological pathways that can provide an additional connection to toxicityendpoints. The annotation is relatively straightforward for most biochemical assaystargeting single proteins. However, the ability to do this properly with cellular assaysis more challenging, since often a specific molecular target of the assay is not known.In some cases, specific biological pathways could be used to annotate the assayendpoint. This approach will be illustrated in Sections 6 and 7.Development of a complete, well-annotated data matrix consisting of curated

chemical structures and their activity against well-characterized biological targets isthe core component of a computational toxicology approach. Such a dataset could bethe final product for a predictive toxicology effort, if the biological assays were allhighly validated surrogates for in vivo toxicology. However, as previously discussed,few such validated targets exist. Therefore, one needs to identify which assays orgroups of assays are linked to toxicity endpoints and can serve as signatures for invivo toxicity. We thus focused much of our early ToxCast screening on chemicalswith rich in vivo toxicity information to use as an anchor for our in vitro results. Thedevelopment of the in vivo database to support this effort will be described next.

1.5 IN VIVO TOXICITY DATABASE

ToxRefDB (Toxicity Reference Database) aims to capture traditional animal toxicitystudies across a variety of study types and endpoints, including short-term and long-term systemic toxicity, cancer, reproductive toxicity, and developmental toxicity[7]. The ToxRefDB project initially focused on capturing previously unpublishedhigh-quality regulatory guideline studies required for chemical safety evaluation bythe EPA. The study submissions were reviewed by the EPA’s Office of PesticidePrograms (OPP) and results consolidated into Data Evaluation Record (DER), whichis the primary data source for ToxRefDB. Study results from this DER as well asfrom other high-quality, publically available studies have been manually curatedinto ToxRefDB’s relational database model. The relational database for ToxRefDBensures data integrity by forcing specific vocabulary to be used across all majorToxRefDB fields. The ToxRefDB relational format follows the following logic: Achemical can have many studies performed, each study can have multiple treatmentgroups (male and female, low-, mid-, and high-dose), and each treatment group can

IN VIVO TOXICITY DATABASE 17

TABLE 1.1 Study and Chemical Counts from the ToxRefDB Website

Summary Statistics

Study count 1,978Chemical count 474Combined chronic/cancer rat 324Combined chronic/cancer mouse 324Multigenerational reproductive rat 352Prenatal developmental rat 365Prenatal developmental rabbit 331Subchronic rodent 302

observe many effects. ToxRefDB has subsequently been integrated into the ACToRsystem, primarily through generic chemical linkages (i.e., CASRN) and is availableas a searchable database (http://actor.epa.gov/toxrefdb). ToxRefDB was designedto capture detailed study design, dosing, and treatment-related effect information.In addition to the relational design of the database, controlled and standardizedvocabularies were used for the vast majority of fields to ensure the uniformity ofthe manually curated and entered legacy toxicity information. The current publicallyreleased version of ToxRefDB has study and chemical effect information on 474chemicals, primarily pesticides due to their consistent and large data coverage ofchronic, cancer, reproductive, and developmental studies. The “Basic Info” page onthe ToxRefDB website contains summary information about the database and theassociated manuscripts. Importantly, the manuscripts release supplemental files withaggregated and detailed endpoints across the full ToxRefDB chemical library. These“flattened” endpoint files (i.e., flat tabular listings) have been directly incorporatedinto the ToxCastDB system for predictive modeling exercises. The “Basic Info” pagealso provides information on the current database and chemical coverage counts foreach study type (Table 1.1).The “home” page of ToxRefDB, similar to that of all ACToR system databases,

allows the user to search by generic chemical. As an example, the key word “azole”was used to search all 474 chemicals in ToxRefDB, by both their assigned chemicalname and all synonyms, and resulted in the return of 46 chemicals (Fig. 1.2). Thered boxes indicate whether or not a study is available in ToxRefDB for the particularstudy type. A “Generic Chemical Page” is displayed, as shown in the ACToRwebsite.However, when accessing the ToxRefDB portion of ACToR, only chemicals withtraditional toxicity data captured in ToxRefDB can be viewed. Under the “ToxicologyData” heading, all ToxRefDB data are displayed in a three-tiered structure. The firsttier contains the study design information, including data quality, species and strain,dose administration, study type, and citation information. The second tier containstreatment group and dosing information, while the third tier indicates the treatment-related effects observed at the various dose levels. The study information is availablefor viewing, but, due to the amount of detailed information stored within each tier,the system does not currently allow for detailed filtering of the data. However, a full

NN

N

NNN

N

O

O

NN

N

NN

N

O

O

O

O

CI

BI

CI

O

Det

ails

deta

ils10

6325

-08-

0

1122

81-7

7-3

1143

69-4

3-6

1162

55-4

8-2

Epo

xico

nazo

le

Tetr

acon

azol

e

Fen

buco

nazo

le

Bro

muc

onaz

ole

deta

ils

deta

ils

deta

ils

Imag

eC

AS

RN

Che

mic

al n

ame

Chr

onic

/Can

cer_

Rat

Chr

onic

/Can

cer_

Mou

seR

epro

duct

ive_

Rat

Dev

elop

men

tal_

Rat

Dev

elop

men

tal_

Rab

bitS

ubch

roni

c_R

od

CR

n

CR

n

CR

n

RR

n

RR

n

DR

nD

Rb

DR

nD

Rb

DR

nD

Rb

S S S

RR

n

RR

n

CM

m

CM

m

FIG

UR

E1.

2Screenshotfrom

theToxRefDBwebsiteforthehigh-levelviewofdataavailableforasetofchemicalsacrossthevariousavailable

studytypes.

18

IN VIVO TOXICITY DATABASE 19

FIGURE 1.3 Screen shot from the ToxRefDB website of the endpoint search page with thesearch criteria and additional field information to be included.

download of the ToxRefDB data is available for each chemical as a csv file, enablingfurther analysis and viewing options.The primary search tool currently available within the ToxRefDB system is located

in the “Search by Endpoint” tab. The page allows the user to select from the standard-ized effect vocabulary, the exact search criteria of interest as well as the additionalfield information to be displayed (Fig. 1.3). The results of searching, for example,“Chronic/Cancer Rat Liver Neoplastic Pathology” returns the lowest effect level(LEL) in mg/kg/day dose, which represents the lowest dose at which a treatment-related change in the selected effect or effects was observed (Fig. 1.4). Each rowfrom the returned search represents a unique study in ToxRefDB, with the low andhigh dose tested (LDT and HDT) provided for reference. If multiple effects areselected, a single LEL is returned, which aggregates all selected effects with a pri-mary goal of providing the field of predictive toxicology a tool for rapidly defining

FIGURE 1.4 Screen shot from the ToxRefDB website of the endpoint search page with theresults of the search displayed.

20 ToxCast

endpoints across a large chemical library. The endpoint search tool can also be usedfor researchers interested in delineating a set of reference chemicals with positiveand negative outcomes for a particular effect or endpoint.ToxRefDB has been applied to multiple problem types, including retrospective

and prospective questions. Retrospectively, ToxRefDB has been used to assess theimpact of specific traditional toxicity endpoints and parameters on the safety regula-tion of chemicals. For example, traditional toxicity testing for reproductive toxicitypotential has relied heavily on a two-generation reproductive toxicity study in rats.However, the importance of the second generation has come into question [41]. Anextended one-generation protocol has been proposed that would only produce a sec-ond generation when triggered, would require far fewer animals, and would derivemore toxicological and kinetic information from each animal used. To assess theimpact of the second generation on risk assessment, ToxRefDB was used as a datasource to systematically evaluate the question, relying on the highly standardizedvocabulary and relational format of ToxRefDB. Based on the data in ToxRefDB, theanalysis indicates that the second generation does not greatly impact the interpreta-tion of the reproductive study from a risk assessment perspective. The two-generationretrospective analysis demonstrated the ability of ToxRefDB to provide a systematicreview of traditional toxicity studies. Additional retrospective analyses are underway,including the analysis of the relative impact and importance of running both rat andrabbit in prenatal developmental toxicity studies.ToxRefDB also stores no-observed and lowest-observed adverse effect levels

(NOAEL and LOAEL) for studies reviewed by EPA and used in the chemical registra-tion process. The Threshold of Toxicological Concern (TTC) is an approach that usesNOAEL/LOAEL distributions and chemical structure characteristics to establish safeexposure levels for chemicals with limited to no toxicity information [42]. ToxRefDBis currently being applied to TTC approaches in numerous venues, including assess-ing the applicability of the standard TTC to antimicrobial pesticide products and therefinement of TTC approaches for specific chemical classes. In the example of theantimicrobial TTC study, all available toxicity study information on antimicrobialsis being collected and entered into ToxRefDB. Antimicrobial pesticides typicallyhave less available toxicity data when compared to conventional pesticides, and thisunderscores the need for alternative safety assessment approaches. With the fullfood-use antimicrobial traditional toxicology dataset available in a standardized andrelational format, detailed analysis of the NOAEL/LOAEL distributions across studytype, endpoint categories, and structural classes can be obtained and compared toother TTC analyses. If found to be similar, then all or a portion of antimicrobialscould be evaluated using a TTC approach.

1.6 PREDICTIVE MODELS

An important use of the HTS data we produce is to develop predictive signaturesof particular types of toxicity. A signature is a pattern of in vitro assay hits thatis predictive of a particular toxic endpoint. The basic approach used is as follows.

PREDICTIVE MODELS 21

First, we select a set of chemicals for which we have both in vitro data and in vivophenotype information, based on data from ToxRefDB. The phenotype or endpointcan either be a quantitative value (e.g., the LEL) or a yes/no (either causes or does notcause the phenotype, at any dose). We then look for statistical correlations betweenthe chemicals active in one or more in vitro assays and the endpoint. Concretely, weask if chemicals that are active in a particular assay are significantly more likely thanchance to be positive for the phenotype. A variety of standard statistical methods canbe used, including multivariate machine learning techniques. Whenever doing thesetypes of analyses, one needs to be particularly careful to not overfit a statistical model,which would result in the creation of a model with little or no forward predictivevalue. One method that one almost always uses is called cross-validation, in whichthe dataset is divided into training and test sets. A model is built on the training setand evaluated on the test set. If the performance in the latter is not significantly betterthan chance, the model is rejected. After the statistical model is built, one typicallylooks at the biological meaning of the signature, for example, is there a support in theliterature for a linkage between the target that is probed by the assay and the endpointbeing evaluated? Finally, one may then modify the model to enhance the biologicalcontent while retaining the statistical predictive power. The ultimate test of a modelis always a forward validation against an entirely new dataset.Here we briefly describe several signatures we have developed using the ToxCast

and ToxRefDB data from Phase I of the ToxCast project. The chemicals in this setare largely pesticidal active ingredients, so they have a wealth of high-quality invivo animal toxicology data. Judson et al. [6] show significant correlations betweenseveral pathways and preneoplastic and neoplastic liver lesions in rats. The targetgenes included the peroxisome proliferator-activated receptors (PPARs)—PPAR�and PPAR� . Activation of these receptors is a well-documented cause of liver tumorsin rodents [43], so that this is an external validation that our screening and statisticalapproach can recover correct biological links. Kleinstreuer et al. [44] have demon-strated associations between particular pathways leading to disrupted vasculogenesisand involving inflammatory chemokine signaling, the vascular endothelial growthfactor pathway and the plasminogen-activating system. Activity in these pathwayscan lead to limb malformations during embryonic development, as demonstrated inprenatal developmental studies in rats and rabbits. This analysis is based on the datafrom ToxCast and ToxRefDB. Martin et al. [45] have used these same approachesto develop signatures for predicting reproductive toxicity. They built composite end-points of male and female effects, including fertility and reproductive fitness. Thesignature included assays related to endocrine disruption (estrogen and androgenreceptor), PPAR activity, liver metabolism as evidenced through activity in a preg-nane X receptor (PXR) assay and in CYP450 assays, and generalized activity againstGPCRs. This model produced a balanced accuracy for sensitivity and specificity of>0.70 in both a cross-validation test set and an independent forward validation set[45]. Finally, Sipes et al. [46] have developed a model of cleft palate and urogeni-tal defects in rat and rabbit. The assays statistically associated with these endpointsinclude the retinoic acid receptor (RAR), interleukins 1A and 8, and the transforminggrowth factor � (TGF-�).

22 ToxCast

Using HTS data for predictive modeling may have opened the door for developingan accurate predictive model of reproductive toxicity. Reproductive toxicity is anaggregated multi-modal and multi-effect outcome. No single assay has the abilityto broadly identify reproductive toxicants and, to date, traditional structure-basedand other methods, have not been able to produce an externally validated predictivemodel of reproductive toxicity. Computational modeling of HTS data allows one toexplore the complex relationships between in vivo observations and networks of invitro activity. One of the more simplistic computational modeling approaches is thedevelopment of a classification model, which aims to accurately classify or predict anoutcome based on a training set with known outcomes. The training set for modelingreproductive toxicity was the set of chemicals in the ToxCast library with high-qualityreproductive toxicity data [45]. The initial inputs into the model were the hundredsof ToxCast assays that were collectively mapped to genes and the aggregate activityacross the assays per gene provided the quantitative inputs into the model. Theassay–gene combinations were further filtered based on a feature selection processthat evaluated the statistical association to the training set data. The filtered gene setwas then weighted in a multivariate model using linear discriminate analysis (LDA)and fivefold cross-validation. Many other approaches and methods could have beendeployed, but our observation has been that the use of complex machine learningalgorithms has a tendency to over-fit the data lowering the output model’s abilityto be externally predictive. The resulting internal model performance statistics weregreater than or equal to 75%balanced accuracy, and therewas no significant differencebetween the training and test set accuracies. The final combined model produced abalanced accuracy of 80%.Among the chemicals selected for external validation, themodel provided accurate

predictions for 16 of the 21 chemicals. The five chemicals with inaccurate predictionsprovide valuable insight into potential limitations or gaps of the model. Interestingly,the five chemicals had a common phenotypic profile with respect to reproductivetoxicity causing reduced early offspring survival, particularly the litter size decreaseswith little to no accompanying effects on reproductive performance or reproductivetract pathology. The reproductive LOAEL for all the five chemicals was set at thehigh dose tested based on the early offspring survival effects, and the parental andoffspring LOAEL were set at the lower dose levels. Based on the inclusive definitionused for defining a positive for reproductive toxicity for model development, all thefive were considered positive but lack evidence of specific fertility-related or develop-mentally sensitive reproductive outcomes. Nonetheless, a gap in model predictivitywas identified and could potentially be filled using additional assay technologies,physical–chemical properties, or structural descriptors.The model development process identified biologically plausible features and

pathways from over 500 assays mapped to less than 100 genes or gene sets and span-ning many reproductively relevant MOA. PPAR� activity was clearly associated withreproductive toxicity, with all 10 PPAR� agonists in the training set causing repro-ductive toxicity. Although a mechanistic link between PPAR activity and fertility orother reproductive impairments remains unclear [47], the role of PPAR in steroid

PREDICTIVE MODELS 23

metabolism and its activity in reproductive tissues infers that it is a plausible targetfor the disruption of endocrine signaling and altered gametogenesis. Androgen andestrogen � activities were also associated with reproductive toxicity. The ToxCastreceptor profiling identified, most if not all, the known antiandrogenic and estrogenicchemicals in the current dataset, but the causal relationship between reproductivetoxicity and steroid receptor activity, absolute and relative potency and efficacy needsto be explored further. CYP enzyme inhibition, as compared to gene induction, wassignificantly associated with reproductive toxicity. Alterations in steroid metabolismthrough CYP induction have previously been associated with reproductive impair-ment [48]. However, the nonspecific inhibition of CYPs may be a surrogate forthe capacity of a chemical to disturb steroid metabolism including inhibition ofkey CYPs involved in steroidogenesis (e.g., CYP19 and CYP17). Related to CYPactivity, PXR interestingly displayed a negative correlation/association with repro-ductive toxicity. In general, PXR lowered the false positive rate of the model bylowering the model score of chemicals with nonspecific and low-potency nuclearreceptor activity. Robust PXR activity is an indication of potent xenobiotic sensingand potentially rapid metabolism. A major component of the model not directlyrelated to nuclear receptor biology and xenobiotic/steroid metabolism was GPCRbinding. Numerous GPCR binding assays were significantly associated with repro-ductive toxicity. Those chosen to represent the GPCR family were selected for sta-tistical and not for biological reasons as there is limited literature information ontheir role in reproduction in contrast to their well-characterized role in the nervoussystem function. Platforms measuring epidermal growth factor receptor, TGF�1 andNF-�B activity were also associated with reproductive toxicity. All the three geneproducts have been shown to modulate the relative sensitivity of developmentaltoxicants, especially AhR signaling [49, 50], and may be indicative of altered xeno-biotic metabolism, cellular proliferation, cell–cell signaling, or potential epigeneticeffects [51, 52]. Overall, the key targets in the model identify plausible MOA leadingto reproductive toxicity and covering antiandrogenic, estrogenic, cholesterol/steroidmetabolism, limited coverage of disruption of steroidogenesis, and altered xenobioticmetabolism MOA.With the availability of an externally validated classification model predicting

reproductive toxicity, the bottleneck of uncharacterized chemicals can be evaluatedeither through improvement in the overall statutory authority to request multigener-ational reproductive studies or in the ability to quantitatively identify reproductivetoxicants. If the statutory authority to request these studies were improved, then thecurrent model in concert with other models, alternative methods, and institutionalknowledge could identify with fairly good accuracy and efficiency for all chemicalsfor which a multigenerational reproductive study should be requested. If the latterwere improved to the point of accurate adverse dose predictions, then the model coulddrastically decrease the need for multigenerational studies and be used in the assess-ment of the majority of environmental chemicals. To do this, improvements in HTSassay reproducibility, metabolic capacity, mode-of-action coverage, reverse toxicoki-netics, and overall model accuracy would need to be made. Placing the classification

24 ToxCast

model into a system modeling context will begin to address this next generation ofresearch questions. For now, the predictive model of reproductive toxicity can go along way in improving reproductive chemical testing efficiency and decision making.

1.7 CHEMICAL PRIORITIZATION

Alternative testing methods (in vitro and in silico) will require extensive refinementand validation before they ultimately replace standardized animal testing. Their moreimmediate utility may lie in their use as tools for chemical prioritization of theextensive chemical inventories of exposure concern. Prioritization is the key link-age between rapidly deployed, computational models and downstream applications.These downstream applications may address the question “In what order do we testchemicals?” or “Which sets of chemicals deserve targeted testing first?” Both ofthese questions involve prioritization in terms of ranking, but the second questionalso involves communication of reasons underlying a ranking. Thus, the ideal priori-tization approach yields both an explicit, prioritized order of chemicals as well as atransparent look at the evidence used.There are several prioritization and decision-analysis approaches being developed

to support chemical prioritization needs [53].One example is theToxPi (ToxicologicalPrioritization Index) framework [54]. This approach can be tailored to diverse sets ofchemicals, evidence (data), and prioritization tasks, because it is based on relative—rather than absolute—rankings. It satisfies both the rank and evidence aspects above,as ToxPi provides a visual, weight-of-evidence index that can be used to rank andcompare chemicals. Its initial application was to aid in the prioritization of chemicalsfor putative endocrine activity, in support of the Endocrine Disruptor ScreeningProgram (EDSP). Composite activity scores across sets of endocrine relevant datafrom the ToxCast assay battery for each chemical were calculated to rank all 309ToxCast Phase I chemicals from the highest to lowest priority. Inclusion of well-studied reference chemicals from the domain of the prioritization is particularlyvaluable. In the EDSP example, the Phase I chemicals contained BPA, methoxychlorand its active metabolite 2,2-bis(p- hydroxyphenyl)-1,1,1-trichloroethane (HPTE),all well-studied chemicals with estrogenic activity. Such chemicals serve to putresults for unknown chemicals in a better toxicological perspective. In the face ofpractical temporal and economic limitations, this estimate of potential endocrineactivity provides a formal rationale for prioritizing resources toward further testing.Alternatively, the ToxPi profiles could be used for chemical “read-across”, analogousto QSAR structural alert models. The read-across can be implemented in terms ofoverall ToxPi bins/clusters of chemicals having similar profiles, or subsets of slicescan be interpreted as in vitro “alerts” to support targeted testing decisions.Future prioritization approaches will have an increasing demand for transparency

and interactivity. This demand is driven by a more informed public, as well asthe thoughtful work of many non-governmental organizations (NGOs) and govern-ment entities. Transparency in both data and models is facilitated by web-accessible

TARGETED TESTING 25

databases and software tools. These interactive prioritization interfaces allow formalinput into decisions by stakeholders, regulators, and the public.

1.8 TARGETED TESTING

The objective of targeted testing of chemicals is to reduce the use of test animals bylimiting the endpoints that need to be evaluated. The way we identify the endpointsfor targeted testing on a given chemical is not straightforward. For new chemicalsand for thousands of existing chemicals in commerce, there is little or no informationabout biological effects. In such cases structural similarity between a new chemicaland a well-studied chemical may be used to infer plausible similarity of biologi-cal effects. While this approach works for some endpoints such as genotoxicity, itis currently difficult to accurately classify chemicals by hazard based on structuralinformation alone. The availability of high-throughput bioactivity profiling makesit feasible to rapidly produce a rich overview of molecular and cellular effects ofthousands of chemicals. Bioactivity profiles can be used to guide targeted tests intwo main ways. First, bioactivity profiles can be phenotypically anchored to adverseeffects from animal testing to discover predictive signatures to classify chemicals byhazard. Second, the molecular and cellular activities in the signatures can be usedto relate early events in the pathways to adverse outcomes. Knowledge about suchpathways, which are also known as adverse outcome pathways (AOPs) [55], can becombined with signatures to improve confidence in their predictions. Predictive sig-natures can serve as a practical tool for identifying endpoints for targeted testing andSection 4 describes their development from bioactivity profiles. Briefly, a signatureis developed using computational tools that mine hundreds of thousands of associ-ations between bioactivity patterns and an endpoint. We select signatures that arestatistically significant and that accurately classify known toxicants. This approachhas been used to develop and evaluate signatures for cancer, reproductive toxicity,and developmental toxicity endpoints using ToxCast bioactivity profiles, and animaltoxicology outcomes from ToxRefDB. In principle, the bioactivity profiles of newchemicals can be matched with signatures to select endpoints for targeted testing. Forinstance, if a new chemical activates inflammatory chemokine signaling, the vascularendothelial growth factor pathway and the plasminogen-activating system accordingto the signature proposed by Kleinstreuer et al. [44], it may be a developmental tox-icant. Similarly, if a chemical is a potent activator of multiple nuclear receptors, itcould be a hepatocarcinogen [43] and may warrant testing in a 2-year bioassay.To use a predictive signature to propose an endpoint for targeted testing, it should

be objectively evaluated using known chemicals first. The predictive accuracy ofthe signature is a measure of confidence but does not guarantee future performance.Future performance depends on the reproducibility and biological breadth of theassays (Section 4), the diversity of the chemical landscape (Section 2), and a sufficientnumber of chemicals to build statistically rigorous signatures of adverse effects.Another approach for improving confidence in empirically derived signatures is tohighlight their role in pathophysiological processes that lead to toxicity.

26 ToxCast

A major challenge for using bioactivity profiling for targeted testing will be toprovide confidence in their ability to extrapolate from in vitro to chronic human healthoutcomes. It is extremely difficult at this time to elucidate the detailed sequence ofcause and effect relationships, so translating in vitro assays and in vivo outcomeswill always involve a level of uncertainty. In addition to this uncertainty, a significantdifference between the two approaches further confounds correlation due to thegeneral lack of xenobiotic metabolism with in vitro approaches. This metabolismcould either activate a chemical to a toxic metabolite or inactivate a toxic chemical toa less toxic one, thus leading to incongruent in vitro and in vivo results. However, it isimportant to note that extrapolating adverse effects from test species to humans is alsofraught with considerable uncertainty. For instance, a number of PPAR activators arerodent hepatocarcinogens, but it is difficult to evaluate the relevance of this outcomein humans given the widespread use of pharmaceuticals targeting the same receptorfamily [56].The effectiveness of targeted testing will improve as diverse evidence about the

events that lead to human toxicity is organized, such as AOPs. An AOP describesthe initiation of molecular events by chemicals, followed by a complex sequenceor network of key molecular, cellular, and tissue level changes that culminate inan adverse effect [55]. In the context of an AOP, we can assume that bioactivityassays directly or indirectly measure changes in early molecular events, and that theendpoints are the toxic outcomes. For instance, PPAR� activation may be considereda molecular initiating event that begins a cascade of subsequent changes. Persistentstimulation with PPAR� activators may lead to proliferative lesions that can progressto neoplasms. However, the AOP in this case includes a series of intermediate effects,among others sustained cell proliferation, hepatic hyperplasia, and preneoplasticlesions.This knowledge can be used to assess the events preceding the apical endpoint

such as cell cycle progression. This is valuable because apical outcomes may onlybe measured in animals, whereas the incipient events such as cell cycle changes maybe observed in cell culture models. Targeted testing approaches are feasible nowby using predictive signatures derived from high-throughput bioactivity assays. Webelieve that confidence in these signatures will improve over time as more chemicalsundergo HTS via programs such as ToxCast. In addition, growing knowledge aboutAOPs may enable additional tiers of in vitro tests, thereby further reducing the needfor animal testing. Sophisticated system models [57, 58] that can accurately estimatethe dynamic changes in AOPs in humans at environmentally relevant exposures mayone day help us design a battery of tiered tests that eliminate the need of animals toevaluate chemical safety.

1.9 CONCLUSION

Changing the nature of toxicity testing in the interest of better identifying and char-acterizing the potential for risk to the health of human and other populations is acentral need in the environmental toxicology field. The use of HTS technologies

REFERENCES 27

and computational toxicology approaches is currently in the initial proof-of-conceptphase. Much progress has been achieved to date including the demonstration of thefeasibility of testing large chemical libraries in diverse HTS assays, collating in vivotoxicity information on myriads of chemicals in a relational database, and buildingpredictive models and prioritization schemes for a number of important toxicologicalendpoints. However, much work remains to be performed. Many important toxic-ity endpoints have not yet been modeled. Many chemical classes have not yet beentested and remain as major obstacles to HTS screening. Lack of significant xenobioticmetabolic activity in in vitro assays remains yet another challenge. Finally, modelswill require extensive refinement and validation before they will serve regulatorypurposes. However, the success of these early steps provides hope that continuedresearch efforts in this arena will eventually lead us to our goal of an efficient, robust,in vitro predictive toxicity screening program that will serve the needs of the public byproviding the capacity to routinely screen existing inventories as well as new chemicalentities for the potential for harm to the health of human and other populations.

DISCLAIMERS

The views expressed in this article are those of the authors and do not necessarilyreflect the views or policies of the US Environmental Protection Agency. Mention oftrade names or commercial products does not constitute endorsement or recommen-dation for use.

ACKNOWLEDGMENTS

The authors thank Robert Kavlock and David Dix for their invaluable contributionstoward developing the computational toxicology approach described here. In addition,we thank Andrew Kligerman and Kevin Crofton for their comments and review ofthis chapter.

REFERENCES

1. Hartung, T. (2009). Toxicology for the twenty-first century. Nature, 460, 208–212.

2. Judson, R.A., Dix, D.J., Houck, K., Martin, M., Kavlock, R., Dellarco, V., Henry, T.,Holderman, T., Sayre, P., Tan, S., Carpenter, T., Smith, E. (2009). The toxicity datalandscape for environmental chemicals. Environmental Health Perspectives, 117, 685–695.

3. Wataru,Y., Peterson, R.E., Tohyama,C. (2010).Molecular targets that link dioxin exposureto toxicity phenotypes. Journal of Steroid Biochemistry and Molecular Biology, 127, 96–101.

4. Dix, D.J., Houck, K.A., Martin, M.T., Richard, A.M., Setzer, R.W., Kavlock, R.J. (2007).The ToxCast program for prioritizing toxicity testing of environmental chemicals. Toxi-cological Sciences, 95, 5–12.

28 ToxCast

5. Kavlock, R., Chandler, K., Houck, K., Hunter, S., Judson, R., Kleinstreuer, N., Knudsen,T., Martin, M., Padilla, S., Reif, D., Richard, A., Rotroff, D., Sipes, N., Dix, D. (2012).Update on EPA’s ToxCast program: Providing high throughput decision support tools forchemical risk management. Chemical Research in Toxicology, 25, 1287–1302.

6. Judson, R.S., Houck, K.A., Kavlock, R.J., Knudsen, T.B., Martin, M.T., Mortensen, H.M.,Reif, D.M., Rotroff, D.M., Shah, I., Richard, A.M., Dix, D.J. (2010). In vitro screeningof environmental chemicals for targeted testing prioritization: The ToxCast project. Envi-ronmental Health Perspectives, 118, 485–492.

7. Martin,M.T., Judson,R.S., Reif,D.M.,Kavlock,R.J., Dix,D.J. (2008). Profiling chemicalsbased on chronic toxicity results from the U.S. EPA ToxRef database. EnvironmentalHealth Perspectives, 117, 392–399.

8. Visser, U., Abeyruwan, S., Vempati, U., Smith, R.P., Lemmon, V., Schurer, S.C. (2011).BioAssay Ontology (BAO): A semantic description of bioassays and high-throughputscreening results. BMC Bioinformatics, 12, 257.

9. Collins, F.S., Gray, G.M., Bucher, J.R. (2008). Toxicology: Transforming environmentalhealth protection. Science, 319, 906–907.

10. Kavlock, R.J., Austin, C.P., Tice, R.R. (2009). Toxicity testing in the 21st Century: Impli-cations for human health risk assessment. Risk Analysis, 29, 485–487.

11. Fliri, A.F., Loging, W.T., Thadeio, P.F., Volkmann, R.A. (2005). Biological spectra analy-sis: Linking biological activity profiles to molecular structure. Proceedings of the NationalAcademy of Sciences USA, 102, 261–266.

12. Bender, A., Scheiber, J., Glick, M., Davies, J.W., Azzaoui, K., Hamon, J., Urban, L.,Whitebread, S., Jenkins, J.L. (2007). Analysis of pharmacology data and the prediction ofadverse drug reactions and off-target effects from chemical structure. ChemMedChem, 2,861–873.

13. U.S. Food and Drug Administration. (1999). Guidance for industry: Antimicrobialfood additives. Available at http://www.fda.gov/Food/GuidanceComplianceRegulatoryInformation/GuidanceDocuments/FoodIngredientsandPackaging/ucm077256.htm.Accessed May 4, 2012.

14. U.S. Environmental Protection Agency. (2012). Data requirements for pesticide reg-istration. Available at http://www.epa.gov/pesticides/regulating/data requirements.htm.Accessed May 4, 2012.

15. Cox, C., Surgan, M. (2006). Unidentified inert ingredients in pesticides: Implicationsfor human and environmental health. Environmental Health Perspectives, 114, 1803–1806.

16. Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S.D., Barber, L.B.,Buxton, H.T. (2002). Pharmaceuticals, hormones, and other organic wastewater contami-nants in U.S. streams, 1999–2000: A national reconnaissance. Environmental Science &Technology, 36, 1202–1211.

17. U.S. Food and Drug Administration. CFSAN: Scope of responsibility. Available at http://www.fda.gov/AboutFDA/CentersOffices/OfficeofFoods/CFSAN/WhatWeDo/default.htm. Accessed May 22, 2012.

18. Cohen-Hubal, E.A., Richard,A., Aylward, L., Edwards, S., Gallagher, J., Goldsmith,M.R.,Isukapalli, S., Tornero-Velez, R., Weber, E., Kavlock, R. (2010). Advancing exposurecharacterization for chemical evaluation and risk assessment. Journal of Toxicology andEnvironmental Health Part B: Critical Reviews, 13, 299–313.

REFERENCES 29

19. Judson, R., Richard, A., Dix, D., Houck, K., Elloumi, F., Martin, M., Cathey, T., Transue,T.R., Spencer, R., Wolf, M. (2008). ACToR—Aggregated Computational ToxicologyResource. Toxicology and Applied Pharmacology, 233, 7–13.

20. Huang, R., Southall, N., Wang, Y., Yasgar, A., Shinn, P., Jadhav, A., Nguyen, D.T., Austin,C.P. (2011). The NCGC pharmaceutical collection: A comprehensive resource of clini-cally approved drugs enabling repurposing and chemical genomics. Science TranslationalMedicine, 3, 80ps16.

21. Boozer, C., Kim, G., Cong, S., Guan, H., Londergan, T. (2006). Looking towards label-freebiomolecular interaction analysis in a high-throughput format: A review of new surfaceplasmon resonance technologies. Current Opinion in Biotechnology, 17, 400–405.

22. Roy, M., Dumaine, R., Brown, A.M. (1996). HERG, a primary human ventricular targetof the nonsedating antihistamine terfenadine. Circulation, 94, 817–823.

23. Marrs, T.C. (1993). Organophosphate poisoning. Pharmacology & Therapeutics, 58, 51–66.

24. Guengerich, F.P. (2001). Common and uncommon cytochrome P450 reactions related tometabolism and chemical toxicity. Chemical Research in Toxicology, 14, 611–650.

25. Szakacs, G., Varadi, A., Ozvegy-Laczka, C., Sarkadi, B. (2008). The role of ABC trans-porters in drug absorption, distribution, metabolism, excretion and toxicity (ADME–Tox).Drug Discovery Today, 13, 379–393.

26. Barlow, S., Kavlock, R.J., Moore, J.A., Schantz, S.L., Sheehan, D.M., Shuey, D.L., Lary,J.M. (1999). Teratology Society Public Affairs Committee position paper: Developmentaltoxicity of endocrine disruptors to humans. Teratology, 60, 365–375.

27. Hutcheson, J.D., Setolab, V., Roth, B.L., Merryman, W.D. (2011). Serotonin receptorsand heart valve disease—it was meant 2B. Pharmacology & Therapeutics, 132, 146–157.

28. Hopkins, A.L., Groom, C.R. (2002). The druggable genome. Nature Reviews of DrugDiscovery, 1, 727–730.

29. Taylor, D.L., Giuliano, K.A. (2005). Multiplexed high content screening assays create asystems cell biology approach to drug discovery. Drug Discovery Today, 2, 149–154.

30. Cartera, C.A., Hamm, J.T. (2009). Multiplexed quantitative high content screening revealsthat cigarette smoke condensate induces changes in cell structure and function throughalterations in cell signaling pathways in human bronchial cells. Toxicology, 261, 89–102.

31. Inglese, J., Johnson, R.L, Simeonov, A., Xia, M., Zheng, W., Austin, C.P., Auld, D.S.(2007). High-throughput screening assays for the identification of chemical probes.NatureChemical Biology, 3, 466–479.

32. Hulme, E.C., Trevethick, M.A. (2010). Ligand binding assays at equilibrium: Validationand interpretation. British Journal of Pharmacology, 161, 1219–1237.

33. Inglese, J., Auld, D.S., Jadhav, A., Johnson, R.L., Simeonov, A., Yasgar, A., Zheng, W.,Austin, C.P. (2006). Quantitative high-throughput screening: A titration-based approachthat efficiently identifies biological activities in large chemical libraries. Proceedings ofthe National Academy of Sciences USA, 103, 11473–11478.

34. Knudsen, T.B., Houck, K.A., Sipes, N.S., Singh, A.V., Judson, R.S., Martin, M.T., Weiss-man, A., Kleinstreuer, N.C., Mortensen, H.M., Reif, D.M., Rabinowitz, J.R., Setzer, R.W.,Richard, A.M., Dix, D.J., Kavlock, R.J. (2011). Activity profiles of 309 ToxCastTM chem-icals evaluated across 292 biochemical targets. Toxicology, 282, 1–15.

35. O’Brien, P.J., Irwin, W., Diaz, D., Howard-Cofield, E., Krejsa, C.M., Slaughter, M.R.,Gao, B., Kaludercic, N., Angeline, A., Bernardi, P., Brain, P., Hougham, C. (2006). High

30 ToxCast

concordance of drug-induced human hepatotoxicity with in vitro cytotoxicity measuredin a novel cell-based model using high content screening. Archives of Toxicology, 80,580–604.

36. Houck, K.A., Dix, D.J., Judson, R.S., Kavlock, R.J., Yang, J., Berg, E.L. (2009). Profilingbioactivity of the ToxCast chemical library using BioMAP primary human cell systems.Journal of Biomolecular Screening, 14, 1054–1066.

37. Berg, E.L., Yang, J., Melrose, J., Nguyen, D., Privat, S., Rosler, E., Kunkel, E.J., Ekins,S. (2010). Chemical target and pathway toxicity mechanisms defined in primary humancell systems. Journal of Pharmacological and Toxicological Methods, 61, 3–15.

38. Hill, A.V. (1910). The possible effects of the aggregation of the molecules of haemoglobinon its dissociation curves. Journal of Physiology, 40, 4–7.

39. Ihaka, R., Gentleman, R. (1996). R: A language for data analysis and graphics. Journal ofComputational and Graphical Statistics, 5, 299–314.

40. Bader, G.D., Cary, M.P., Sander, C. (2006). Pathguide: A pathway resource list. NucleicAcids Research, 34, D504–D506.

41. Cooper, R.L., Lamb, J.C., Barlow, S.M., Bentley, K., Brady, A.M., Doerrer, N.G., Eisen-brandt, D.L., Fenner-Crisp, P.A., Hines, R.N., Irvine, L.F., Kimmel, C.A., Koeter, H., Li,A.A., Makris, S.L., Sheets, L.P., Speijers, G., Whitby, K.E. (2006). A tiered approachto life stages testing for agricultural chemical safety assessment. Critical Reviews inToxicology, 36, 69–98.

42. Kroesa, R., Gallib, C., Munroc, I., Schilterd, B., Trane, L.-A., Walkerf, R., Wurtzeng, G.(2000). Threshold of toxicological concern for chemical substances present in the diet: Apractical tool for assessing the need for toxicity testing. Food and Chemical Toxicology,38, 255–312.

43. Klaunig, J.E., Babich,M.A., Baetcke, K.P., Cook, J.C., Corton, J.C., David, R.M., DeLuca,J.G., Lai, D.Y., McKee, R.H., Peters, J.M., Roberts, R.A., Fenner-Crisp, P.A. (2003).PPARalpha agonist-induced rodent tumors:Modes of action and human relevance.CriticalReviews in Toxicology, 33, 655–780.

44. Kleinstreuer, N.C., Judson, R.S., Reif, D.M., Sipes, N.S., Singh, A.V., Chandler, K.J.,Dewoskin, R., Dix, D.J., Kavlock, R.J., Knudsen, T.B. (2011). Environmental impacton vascular development predicted by high-throughput screening. Environmental HealthPerspectives, 119, 1596–1603.

45. Martin, M.T., Knudsen, T.B., Reif, D.M., Houck, K.A., Judson, R.S., Kavlock, R.J., Dix,D.J. (2011). Predictive model of rat reproductive toxicity from ToxCast high-throughputscreening. Biology of Reproduction, 85, 327–339.

46. Sipes, N.S., Martin, M.T., Reif, D.M., Kleinstreuer, N.C., Judson, R.S., Singh, A.V.,Chandler, K.J., Dix, D.J., Kavlock, R.J., Knudsen, T.B. (2011). Predictive models of pre-natal developmental toxicity from ToxCast high-throughput screening data. ToxicologicalSciences, 124, 109–127.

47. Peraza, M.A., Burdick, A.D., Marin, H.E., Gonzalez, F.J., Peters, J.M. (2006). The toxi-cology of ligands for peroxisome proliferator-activated receptors (PPAR). ToxicologicalSciences, 90, 269–295.

48. Goetz, A.K., Ren, H., Schmid, J.E., Blystone, C.R., Thillainadarajah, I., Best, D.S.,Nichols, H.P., Strader, L.F., Wolf, D.C., Narotsky, M.G., Rockett, J.C., Dix, D.J. (2007).Disruption of testosterone homeostasis as a mode of action for the reproductive toxicityof triazole fungicides in the male rat. Toxicological Sciences, 95, 227–239.

REFERENCES 31

49. Abbott, B.D., Buckalew, A.R., DeVito, M.J., Ross, D., Bryant, P.L., Schmid, J.E. (2003).EGF and TGF-alpha expression influence the developmental toxicity of TCDD: Doseresponse and AhR phenotype in EGF, TGF-alpha, and EGF + TGF-alpha knockoutmice. Toxicological Sciences, 71, 84–95.

50. Tian, Y., Ke, S., Denison, M.S., Rabson, A.B., Gallo, M.A. (1999). Ah receptor and NF-kappa B interactions, a potential mechanism for dioxin toxicity. Journal of BiologicalChemistry, 274, 510–515.

51. Tian, Y.A. (2009). Ah receptor and NF-kappa B interplay on the stage of epigenome.Biochemical Pharmacology, 77, 670–680.

52. Tian, Y.N., Rabson, A.B., Gallo, M. A. (2002). Ah receptor and NF-kappa B interactions:Mechanisms and physiological implications. Chemico-Biological Interactions, 141, 97–115.

53. Gabbert, S., Weikard, H.P. (2010). A theory of chemicals testing and regulation. NaturalResources Forum, 34, 155–164.

54. Reif, D.M., Martin, M.T., Tan, S.W., Houck, K.A., Judson, R.S., Richard, A.M., Knudsen,T.B., Dix,D.J., Kavlock, R.J. (2010). Endocrine profiling and prioritization of environmen-tal chemicals using ToxCast data. Environmental Health Perspectives, 118, 1714–1720.

55. Ankley, G.T., Bennett, R.S., Erickson, R.J., Hoff, D.J., Hornung, M.W., Johnson, R.D.,Mount, D.R., Nichols, J.W., Russom, C.L., Schmieder, P.K., Serrrano, J.A., Tietge, J.E.,Villeneuve, D.L. (2010). Adverse outcome pathways: A conceptual framework to supportecotoxicology research and risk assessment. Environmental Toxicology and Chemistry,29, 730–741.

56. Michalik, L., Desvergne, B., Wahli, W. (2004). Peroxisome-proliferator-activated recep-tors and cancers: Complex stories. Nature Reviews Cancer, 4, 61–70.

57. Shah, I., Wambaugh, J. (2010). Virtual tissues in toxicology. Journal of Toxicology andEnvironmental Health Part B: Critical Reviews, 13, 314–328.

58. Wambaugh, J., Imran, S. (2010). Simulating microdosimetry in a virtual hepatic lobule.PLoS Computational Biology, 6(4), e1000756.

2HIGH-THROUGHPUT TOXICITYTESTING IN DRUG DEVELOPMENT:AIMS, STRATEGIES, AND NOVELTRENDS

Willem G.E.J. Schoonen, Walter M.A. Westerink,Femke M. van de Water, and Horbach G. Jean

2.1 INTRODUCTION

The development of new therapeutic drugs within the pharmaceutical industry is acomplex and lengthy process of optimization. This process starts with the selectionof an optimal target for drug development, which can, for example, be a membranereceptor, a nuclear receptor, a cellular enzyme within a particular cell type, or acell–tissue combination. In case of cell types, one should think about different cellsas a tool for cell-specific receptor or enzyme processes, among others lymphoblasts,B- and T-helper cells, hepatocytes, renal proximal tubuli cells, adrenal glomerulosacells, myocytes, cardiomyocytes, as well as alveolar cells. In case of cell–tissuecombinations, one should think about co-cultures and 3D-cultures, for example, thedirect interaction of fibroblasts/leukocytes/Kupffer cells with hepatocytes, and theindirect interaction by means of metabolites in a sandwich culture system, in whichhepatocytes are cultured in a hanging basket and underneath, in the regular well,the pharmacological tissue or cells of interest [1, 2]. Moreover, precision-cut tissueslices, in which the tissue structure remains intact as in the case of liver tissue [3] andin which fibroblasts, hepatocytes, Kupffer, and stellate cells interact with each other,can be used. As an example for co-, 3D-, and sandwich cultures, lipopolysaccharidescan stimulate Kupffer cells to produce tumor necrosis factor-� or interleukins, which


33

34 HIGH-THROUGHPUT TOXICITY TESTING IN DRUG DEVELOPMENT

in turn are excreted and lead to the activation of their cognate receptors in hepatocytes.In this way, a more physiological interactive setting is created.After the selection of the appropriate in vitro tools and the drug target of choice, the

best ligand may be identified, being either an agonistic or antagonistic compound. Ifthe ligand is known, the assay can be validated, but if the drug target has an unknownligand, a search has to be started by screening large libraries of chemical compoundsin the range of 105 up to 108. This number of compounds can only be screenedby using high-throughput screening (HTS), which was developed in the 1990s with96-well plates, but which was soon extended toward 384-, 1536-, 3072- and 6144-well plates [4]. Not only the well plates, but also the equipment to measure theseplates underwent a rapid development towardHTS systems. Nowadays, very sensitivephotomultipliers and digital cameras have improved the detection limits and devicesof spectrophotometers, fluorometers, luminometers, as well as flow cytometers andmicroscopic imaging equipment have been developed toward HTS [4–7] and/orhigh-content screening (HCS) systems [8–10]. This equipment can be used in a semi-automated or fully robotized fashion. From the target and hit identification onward,these HTS tools were also introduced in the second and third screening phases withinthe pharmaceutical industry, leading to new in vitro screening strategies for both hitand lead optimization before the start of animal studies. In the hit-optimization phase,the hit is improved regarding its potency to interact with its target by changing thechemistry of themolecule, whereas during the lead optimization step, next to potency,improvements with respect to absorption, distribution, metabolism, and excretion(ADME), for which also many in vitro assays are available, are made [11, 12].

2.2 DRUG TOXICITY FAILURE IN (PRE)CLINICAL DEVELOPMENT

As shown by Kola and Landis [13], the implementation of ADME assays has tremen-dously reduced drug failure due to pharmacokinetics. As a consequence of thisreduction, the overall failure rate due to toxicological issues increased from 11% to20% and even up to 40% if (pre)clinical development and marketing of the drug isincluded [13, 14]. The failure rate can differ in the kind of toxicity depending onthe company and the specificity of drugs that are under development. For instanceat Legacy Organon, compounds were developed in the area of birth control and hor-monal replacement therapy to prevent muscle weakness, bone loss, and hot flushesin males and females. In this company, the attrition rate due to genotoxicity, carcino-genicity or reproductive toxicity was 20% (Table 2.1). The pharmacological effectsof the progestagens, estrogens, and androgens under investigation at Organon actthrough their cognate nuclear receptors at dose levels in the range of 50 pM up to 10nM, while the toxic effects of these compounds were found at concentrations of morethan 10�Min the case of genotoxicity, that is, especially the formation ofmicronucleiand chromosomal aberrations. This is in contrast to the above-mentioned more phys-iological concentrations leading to reproductive toxicity, as these classes of steroidalcompounds are involved in reproduction and thus in ovarian cycle control, ovula-tion induction, initiation and maintenance of pregnancy, and spermatogenesis. The

STRATEGY FOR IMPLEMENTATION OF IN VITRO TOXICITY TESTING 35

TABLE 2.1 Percentage of Drug Failure for Toxicity Reasons within Drug Development

Merck Dupont and LegacyTox Failure Roche Bristol-Myers Squibb Organon

Genotoxicity/carcinogenicity 6 8 20Reproductive toxicity 2 5 20Hepatotoxicity 20 15 12CV safety 16 27 12Skin toxicity 10 0 2CNS side effects 10 7 4Blood toxicity 6 7 3Renal toxicity 4 2 12Gastrointestinal toxicity 4 3 3

In bold: two highest risk factors per company.

attrition rates due to genotoxicity, carcinogenicity, or reproductive toxicity weremuchlower in the case of pharmaceutical companies like Roche, Merck Dupont (MD), andBristol-Myers Squibb (BMS) than at Organon [14, 15]. On the other hand, Roche,MD, and BMS had to deal much more frequently with hepatotoxicity (15–20%) andcardiotoxicity (16–27%), while at Organon the attrition rates due to hepatotoxicityand cardiotoxicity were only 12%.Moreover, Roche scored 10% for skin toxicity andcentral nervous system side effects, while within Organon renal toxicity was moreprominent with 12%. To reduce these attrition rates for toxicity, early implementa-tion of in vitro toxicity screening, as in the case of ADME studies, should lead toclear improvements. A strategy for the early implementation in the drug developmentprocess of in vitro toxicity screening will be part of the next section.

2.3 STRATEGY FOR IMPLEMENTATION OF IN VITROTOXICITY TESTING

The development of a new therapeutic drug in a pharmaceutical company can bedivided into several phases (Fig. 2.1). The start of this process was already describedabove with the target and hit finding, which is followed by the hit-optimization phase.After hit optimization, the lead-finding process is started, which flows into the lead-optimization phase and finally into the selection of the drug development candidate.From that moment on, the trajectory of drug discovery is finished and preclinicaland clinical development will start, during which amongst others, drug safety tests inanimals will be performed. One of the challenges is to reduce the attrition rate dueto toxicity late in the development process or even by market withdrawal. In orderto reach this goal, better prediction models for human toxicity may be needed. Inretrospect, these models can be built on those drugs that have failed in the clinic dueto all kinds of toxicity.When we consider the drug development process, early toxicity screening is

performed preferentially before the start of the (pre)clinical development phase.


Target finding

Hit finding

Hit optimization

Lead finding

Lead optimization

Drug dev. candidate

Target discovery Lead discovery

Discovery phase

(Pre) Clinical development phase

Phase I Phase II Phase III Registration Marketing Sales

(Pre) Clinical development Drug approval

FIGURE 2.1 Developmental phases for a new therapeutic drug.

The best strategy may differ between companies and may be different for differentcompound classes, but certain requirements need to be fulfilled as otherwise costswould be too high for the successful implementation of early toxicity screening. Onepremise is the need to reduce costs with respect to compound synthesis and animaltesting. This implies that for an optimal strategy, animal toxicity testing is postponeduntil the start of preclinical development and the amounts of compound needed fortesting are minimized by miniaturization of the test models into 96- or 384-wellplates. The toxicity prediction of compounds can be optimized using in vitro toxicityscreening methods by means of a battery of tests for different kinds of toxicity.Such a battery of tests will be based on the identification of a certain set of criticalissues for specific compounds, such as genotoxicity, carcinogenicity, non-genotoxiccarcinogenicity, cytotoxicity, phase I and II metabolism, endocrine disruption,embryotoxicity, and nuclear-receptor-induced enzyme activation. Besides the varietyof tests, the use of these tests preferentially should meet the following criteria:

1. The amounts of compound needed for testing should not exceed 10–20 mg;

2. The purity of the compound should already be high enough for toxicity testing;

3. The solubility of the compound should be checked at all tested concentrations;

4. The turnaround times of the assays should be short, preferentially between 1and 5 days;

5. The throughput should be high, preferentially at least 40–80 compounds perassay;

6. The assays should identify cell, tissue, and species-specific effects.

Implementation of such a complete battery of different tests can be done by usinga combination of sets of assays based on their format similarity. Robotic as well asmanual pipetting in multiple wells of concentration-dependent dilutions in a wide

STRATEGY FOR IMPLEMENTATION OF IN VITRO TOXICITY TESTING 37

Blank

DMSO stock plate/dilution

12 compounds

10x or 100x Dilution in PBS

10x dilution in assay Assay 1

Assay 2

Assay 3

10-1 M

10-4 M

FIGURE 2.2 Procedure of plating compounds in a concentration gradient and stamping ofthe test plates by robotic instruments.

concentration range, for example, 10−3 to 10−8 M, is performed in either full lanesor rows (Fig. 2.2). The first dilution step of the reference and test compounds shouldalways be carried out in an appropriate solvent like DMSO, while the second and/orthird dilution step of the dilution series are performed in either phosphate-bufferedsaline or a solvent like test buffer or test medium. With cellular proliferation andenzyme assays, no toxic effects have been identified at or below 0.1% of DMSO,but at 1% of DMSO detrimental effects in some of the assays can be observed. Afterthe preparation of the final dilution plates, the compounds can be added to the cellcultures and enzyme preparations or vice versa, depending on the test procedure used.The application of robotic equipment in this phase with easy assay-pipetting formatsand especially multiple pipetting by 8-, 16-, 32-, 48-, 96- and/or 384-well pipettingdevices can simplify the dilution steps, as shown in Figure 2.2.In our view, the lead finding and especially themiddle part of the lead-optimization

phase is the best time in drug discovery for early in vitro toxicity screening. Theadvantages of performing toxicity screening at this stage are that in each project, sev-eral compounds are still in the race for selection and a change of direction due to anunwanted toxicological outcome for one, several, or all compounds can still rescue theprogram. An additional advantage of this procedure is that compounds with an equiv-alent pharmacological and ADME profile can be ranked based on their potential tox-icological side effects. Experiences during the last 10 years have yielded several ben-efits in performing early in vitro toxicity screening for genotoxicity, carcinogenicity,non-genotoxic carcinogenicity, cytotoxicity, phase I and II metabolism, endocrinedisruption, embryotoxicity, and nuclear receptor-induced enzyme activation. Thesetools and their merits in early in vitro toxicity screening will be discussed below.


2.4 ASSAYS FOR IN VITRO GENOTOXICITY SCREENING

In the 1970s, the first in vitro toxicity procedures to test for mutagenicity, the bacterialreverse mutation gene tests, were introduced and led to the full Ames test [16, 17].In this full Ames test, different Salmonella typhimurium and Escherichia coli strainsare used for the prediction of a simple nucleotide mutation or a frameshift mutation,which leads to the impairment of one particular enzyme within the histidine or tryp-tophan synthesis pathway. As a consequence of mutagenesis, the miscoded enzymeis repaired by reverting it into its natural active confirmation. The S. typhimuriumstrains often used for the detection of point mutations are TA100 and TA1535 andfor the detection of frameshift mutations TA98 and TA1537 [18]. In addition, theS. typhimurium TA102 or the E. coli strain WP2 uvrA (pKM101) are commonlyapplied. To identify whether a compound is mutagenic, the bacteria are incubatedwith different concentrations of the compound in the absence or presence of an S9-fraction in growth medium without histidine or tryptophan. Only the revertant cellswill survive and the clones can be counted as a measure of the rate of mutagenesis.Thereby, the S9-fraction stands for a 9000 g supernatant of the liver homogenateof rats that were pretreated for 24 h with Aroclor 1254 or a cocktail of compoundsresponsible for the activation of the aryl hydrocarbon receptor (AhR), pregnane Xreceptor (PXR), and constitutive androstane receptor (CAR). This fraction will haveenhanced activity of a variety of cytochrome P450 enzymes. As reference com-pounds 4-nitroquinoline-1-oxides (4-NQO), methyl methanesulfonates (MMS), 9-aminocamptothecins (9-AC), sodium azides (NaN3), and nitropyrenes are commonlyused in the absence of S9-fraction, while in its presence benzo[a]pyrenes, (B[a]P),2-acetylaminofluorenes (AAF), nitrofurantoins (NF), and cyclophosphamide (CPA)are the favorite candidates [19, 20]. This latter set of compounds needs biotrans-formation by the S9-fraction to produce the mutagenic metabolites. Within severalpharmaceutical companies this U.S. Food andDrugAdministration (FDA)- and Euro-pean Medicines Agency (EMA)-approved test has been automated for HTS for allfive strains or either one or two of those in a 96- or 384-well format. Besides this fullAmes test, also the Ames II test (from Xenometrix) is available as a surrogate test[21, 22]. This test is applied on the TA98 strain and a mixture of TA strains (7001,7002, 7003, 7004, 7005, and 7006).A modernized version of the Ames test was introduced with the VitotoxTM strain

(Gentaur, Kampenhout, Belgium). The S. typhimurium TA104 strain is used to assessDNAdamage by activating the SOS repair system [23–25]. Bymaking use of biotech-nological techniques, one of the DNA repair genes has been replaced by the luciferasegene of the beetle Vibrio fischeri. The activation of this SOS repair gene can then bemeasured by luminescence after treating the bacteria with the appropriate concentra-tions of mutagenic compounds. A second TA104 strain constitutively expressing theluciferase gene has been prepared to correct for the cytotoxic effects of the mutageniccompounds. A ratio of more than 1.5-fold in signal between both strains suffices forthe identification of mutagenic and occasionally also clastogenic compounds. Theadvantage of this assay above the full Ames and Ames II test is that it can be carriedout within 3 h instead of 5 days. In the 384-well plates, 16 compounds can be analyzed

ASSAYS FOR IN VITRO GENOTOXICITY SCREENING 39

by making 15 measurements (every 10 to 15 min) in 3 h, leading to a throughput ofat least 160 compounds a week, which is much faster and more in line with the HTSformat.These bacterial mutagenicity tests, however, will underscore mitotic and meiotic

division failures, which can only be recognized in eukaryotic cells. In the 1980s,human peripheral blood lymphocytes, Chinese hamster ovary (CHO), or lung (V79)cells were used on a regular basis for the classification of clastogenicity (chromosomalaberrations, sister chromatid exchange, in vitro micronuclei) [26, 27]. Although thetesting of the compounds for clastogenicity in the absence and presence of S9-fraction is relatively fast with 3-, 24-, or 48-h time points, the scoring and analysisof 200 metaphases for aberrations or 2000 cells for micronuclei per condition is acumbersomeprocedure taking even skilled persons up to 2–3weeks for one compoundand therefore not suited for HTS [28]. Moreover, the prediction of these test systemsis hampered by a high number of false positives. As a consequence, the EuropeanCentre for the Validation of Alternative Methods (ECVAM) has produced a list of 62compounds for the evaluation of new tests in this area, incorporating 20 true positivesbeing both in vitro and in vivo real genotoxicants, 23 true negatives being both in vitroand in vivo non-genotoxicants and non-genotoxic carcinogens, but more importantly19 false positives, being in vitro genotoxic or non-genotoxic carcinogens, whereas invivo they are either non-genotoxic or non-genotoxic and non-carcinogenic [29, 30].Evaluation of these compounds may help to understand the high number of falsepositives in in vitro clastogenicity tests. One cause may be that in vitro screeningis carried out by visual examination at excessive high dose levels, these being evenabove 10 mM or 5 mg/ml. These high dose levels easily lead to cytotoxicity and ATPdepletion, which may also lead to chromosomal aberrations and in vitro micronucleidue to segregation failures. This high number of false positives in the area of in vitroclastogenic profiling can be improved by automation of this scoring process.Consequently, HCS has been incorporated for the determination of chromoso-

mal aberrations as well as for the analysis of in vitro micronuclei. Especially theoptimization of software tools and digital cameras improved the quality of HCS forscoring in vitro micronuclei. With HCS, 50 instead of 2 compounds can be screenedfor micronuclei within 1 week when compared to the manual procedure [31]. ForCHO and HepG2 cells [32, 33], such assays have been developed and validated. Afterblocking cytokinesis with cytochalasin B followed by fixation, the cells can be stainedwith Hoechst 33342 or DAPI for the identification of nuclei and with DRAQ5TM forthe localization of the cytoplasm. In this way, the segregation can be made betweenbinucleated and mononucleated cells and the number of micronuclei in the binucle-ated cells can be quantified with automated algorithms. The sensitivity, specificity,and predictivity scores for these assays are high, being above 80% for CHO cells,with respect to the regular manual testing devices. Moreover, clastogenic and aneu-genic compounds can be segregated by making a size scaling of the mean distributionof micronuclei, in which the large size indicates the aneugenic compounds, and thesmall size the clastogenic compounds [33, 34].WithHCS techniques, further parameters related to genotoxicity and double-strand

breaks can easily be analyzed. It is commonly known that mutations in vitro and


in vivo in one of the several genes like p53, XPC, RAD51C,MDM2, p21 (CDKN1A),BCRA1, and GADD45A in humans, rats, and mice largely increase the incidence ofcarcinogenicity [35–46], while protein phosphorylation of p53 [43, 47] and � -H2AX[48] are other good biomarkers. The activation of the p53 protein leads to the activationof genes involved in DNA repair, such as XPC, the Xeroderma pigmentosum Cgene, which coordinates nucleotide excision repair within the DNA strands, and theRAD51C as well as the � -H2AX gene, which encode proteins that repair homologousDNA double-strand breaks. MDM2 is a protein that interacts with the p53 protein,preventing the activation of the p53 protein. In case of DNA damage, the MDM2dissociates from the p53 protein, which then can be activated, leading to apoptosisor cell cycle arrest to gain additional time for DNA repair. The CDKN1A (p21)gene encodes a cyclin-dependent kinase inhibitor (p21) involved in cell cycle arrest,while both BCRA1 and GADD45A are genes involved in more general DNA repairprocesses. Therefore, these genes are very important in the prevention of genomicdamage. For instance, the activation of such genes and proteins in lymphoblasts in vivocan predict whether animals or humans have been exposed to radiation or genotoxiccompounds. This makes these genes good in vivo biomarkers that can also be ofconvenience for in vitro toxicity screening. Exposure to genotoxic compounds suchas camptothecin, etoposide, and 5-fluorouracil can indeed lead to the activation of thep53 protein by phosphorylation, p53 then dissociates from MDM2, and translocatesinto the nucleus [49]. Treatment with siRNA against p53 can prevent this process andmakes cells more vulnerable to the genotoxins. Moreover, double-strand breaks willlead to the activation of RAD51C and � -H2AX proteins. With antibodies againstthese two proteins, the influence of genotoxic compounds on these in vivo and invitro biomarker genes can easily be assessed [35, 50].Another way to improve this kind of clastogenic screening is by preparing cell

lines with the proper promoter sites of the above-mentioned proteins, or their respon-sive element in case of p53, in combination with a reporter, this being either greenfluorescence protein (GFP), luciferase (LUC) or�-galactosidase (�-gal), as a readout,making such cellular assays appropriately equipped for HTS screening. Such deviceshave, for example, been developed and validated for yeast (Saccharomyces cere-visiae) with the promoter of the RAD54 gene in combination with GFP (Gentronix,Manchester, UK) [15, 24, 51–54] and �-galactosidase (reMYND, Leuven, Belgium)[55–57] or for liver HepG2 cells with the promoter of the RAD51C gene coupledto firefly luciferase [58]. Both these RAD genes are recombination repair genes fordouble-strand breaks [59, 60]. The expression of these genes can be induced by bothclastogenic and mutagenic compounds and can be measured after different time inter-vals depending on the time table of the cell division process. Also in the case of theseassays, a control with respect to cytotoxicity must be incorporated. For HepG2 cells,a promoter assay was developed that makes use of the promoter region of the genecoding for cystatin A, an antiapoptotic protein, as well as the responsive element ofthe p53 protein, both in combination with a luciferase readout. These assays can beperformed in 96-well plates after 48 h of drug treatment [58]. In a similar fashion,Gentronix developed a lymphoblastoid TK6 cell line in which the regulation of thehuman GADD45A gene is linked to the production of GFP as readout [53]. The

ASSAYS FOR IN VITRO CYTOTOXICITY SCREENING 41

Mutagenicity Clastogenicity

FDA and EMA approved GLP in vitro genotoxicity tests

Alternative in vitro tests

Mutagenicity Clastogenicity

Bacterial cell line • Full Ames with histidine deficiency • Mini-Ames • Ames II

Mammalian cell lines • Thymidine kinase deficiency • Hypoxanthine phosphoribosyl transferase def.

Mammalian cell lines • Microscopic chromosomal aberration • Microscopic micronuclei testing

Bacterial cell line • Luciferase reporter assay (Vitotox)

Yeast cell line • GFP and β-galactosidase assays (Gentronix and reMYND) Mammalian cell lines • High content screening micronuclei testing • High content screening antibody biomarkers (p53, -H2AX, RAD51C, GADD45A) γ • Luciferase reporter assays (p53, RAD51C, Cystatin A, GADD45A)

FIGURE 2.3 Testing procedures for genotoxicity testing.

presence of an S9-fraction as a metabolizing system is needed as with most assayswhen endogenous metabolism is inadequate. With respect to HepG2 cells, there isan ongoing debate whether these cells possess sufficient endogenous metabolism.Several scientists believe that incorporation of HepG2 cells into the analysis for clas-togenicity will improve the final prediction scores due to endogenous metabolism[61, 62]. Under our HepG2 cell culture conditions, 11 reference compounds, whichhave to be metabolized to become genotoxic, were analyzed by performing one of thepreviously described promoter assays and all 11 were identified as being genotoxicin the absence of the S9-fraction [33, 58].Thus, in the case of genotoxicity testing, a battery of tests (Fig. 2.3) is available,

in which the sensitivity is consistently high and the specificity has strongly improvedfrom 55% [29] to above 80%. Application of such tests in the lead-optimization phasewill result in a better guidance along the selection procedures for genotoxicity.

2.5 ASSAYS FOR IN VITRO CYTOTOXICITY SCREENING

The founding father of in vitro cytotoxicity screening is the late Prof. Dr. BjornEkwall (1940–2000), being active in this field from 1980 onward [63–66]. In 1989,a new international multicentre project, the so-called MEIC project, was started on


his behalf to evaluate the relevance of the outcomes of in vitro cytotoxicity tests forhuman toxicity. Hereby, 50 reference compounds, including pharmaceutical drugs,regular salts, heavy metals, several alcohols, and organic solvents, were selected[67]. These studies were carried out from 1989 to 1996 in 97 different labs on61 different assays and included measurements of ATP, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) coloration, neutral red uptake, cell numbers,colony formation, pH changes, protein content, lactate dehydrogenase leakage, andmorphological cellular changes. The overall results were presented 10 years later,showing a good toxicity prediction for 22 out of 50 compounds (44%). The toxiceffects increased considerably in sensitivity with an elongated exposure time from3 to 168 h [68, 69]. Moreover, toxicity dose–response curves between oral dosingin humans and mice or rats as well as between acute toxicity and the IC50 valuesobtained in in vitro studies showed a good correlation [70, 71]. In general, the tox-icity for cells of different tissues was identified as being relatively similar. Becausethese data looked very promising, the ACuteTox project was started in the 21st cen-tury to make a new comparison between 97 reference compounds on human HepG2and rat Fa32 hepatocytes as well as on mouse fibroblast 3T3 cells, in which nowalso the in vivo prediction and serum dose levels were taken into account [72, 73].The obtained data confirmed the conclusions from the MEIC study and showedthat the IC50 values in the in vitro assays were very similar among cells of dif-ferent tissues, but differed more between different species. For example, digoxin,a specific human potassium channel inhibitor, was only identified as being toxicin human tissues. Somewhat later in America a similar approach to these MEICand ACuteTox studies was started by the ToxCast program, which is funded by theU.S. Environmental Protection Agency (EPA) [74]. Here, a set of 400 compoundsare being analyzed and more assays have been incorporated, which is made pos-sible by the introduction of HTS programs to handle the number of assays andcompounds within a short time frame. In this program, all available toxicologicalassay techniques are being combined. This means that not only cell-based prolifer-ation assays, cell viability assays, and biochemical assays, but also HCS by imageanalysis, microelectronic monitoring, patch clamp monitoring, model organisms, aswell as new technologies like toxicogenomics, metabolomics, and proteomics arebeing applied. This high-throughput analysis will be more informative in the end andwill teach us which strategy will be the best for early cytotoxicity screening in thefuture. However, today we have already reached a level that is quite discriminatoryand a review on the state-of-the-art technology will be given in the next paragraphsfor cytotoxicity.General aspects of cytotoxicity can be evaluated with simple assays by means of

medium- to high-throughput screening with spectrophoto-, fluoro-, or luminometricmeasurements [5, 75, 76] (Fig. 2.4). All the basic ingredients for regular cellularphysiology like oxygen consumption, energy metabolism, and cellular proliferation,as well as cell survival can become specific cellular toxicity markers. For studyingmitochondrial activity as a marker of cellular toxicity, a variety of assays can beused, for example, Alamar BlueTM (AB) [77–80], several tetrazolium dyes such as


MTT [80–82], nitro blue tetrazolium (NBT), XTT, and WST-1 [83], Cyto-LiteTM

[84], or MitoXpress (Luxcel, Cork, Ireland) [85, 86]. The latter method is special inthis line as it quantifies the oxygen consumption directly by a specific fluorescenceresonance electron transfer,making theClark electrode equipment redundant [87, 88].Moreover, for the analysis of the energy status, ATP-LiteTM [89] and CellTiter-GloTM

[90] are suitable assays. For cellular proliferation, the most commonly used assaysfor DNA coloration are Hoechst 33342 [91–93], DAPI [94, 95], or crystal violet [96]whereas for incorporation of nucleotides into the DNA chains, [3H]-thymidine orbromodeoxyuridine [81, 82, 97] are also often used.Besides these more general cytotoxic assays, more special toxicity measurements

can be carried out. The active uptake of calcein-acetoxymethyl (Calcein-AM) is usedas an indicator of membrane integrity, while oxidative stress can be measured byglutathione depletion and the formation of reactive oxygen species. The cellular glu-tathione concentration can be determined by coupling glutathione to 5,5′-dithiobis(2-nitrobenzoic acid) (DTNB) [98], monochlorobimane (MCB) [99–101], or a luciferinsubstrate bymeans of glutathione-S-transferase (GST) isozymes [102], while reactiveoxygen species can be assessed by dichlorofluorescein (DCF) [103–105] or a Nrf2responsive element reporter assay coupled to GFP or luciferase [58, 106]. More-over, apoptosis can easily be measured by specific luciferin substrates for caspase3/7 and caspase 8/9, these being Z-DEVD-aminoluciferin and LETD-aminoluciferin[107, 108].Hence, a battery of suitable HTS assays exists (Fig. 2.4), and 8 assays in line

with the MEIC studies were evaluated in house by us, those being the AB, Hoechst33342, ATP-LyteTM, Cyto-LyteTM, DCF, MCB, Calcein-AM and Luxcel assays [88,109, 110]. These assays were performed on human and rat liver cells (HepG2,H4IIE), human cervix carcinoma cells (HeLa), human endometrium cells (ECC-1), and CHO cells. This comparison showed that, as in the case of the MEIC andACuteTox studies, differences among species, but not tissue selectivity, give thelargest discrimination in toxic effects. The selectivity of the identification of very toxiccompounds, these being cytostatic, necrotic, and genotoxic compounds, is 70–80%dependent on the end dosages and cell lines used for the prediction. However, whensteatotic, cholestatic, phospholipidotic, aneugenic, and non-genotoxic carcinogeniccompounds are included, the selectivity value for in vivo liver toxicity is decreasedto 46% (52/113). On the other hand, specificity remained very high (96%, 124/130),leading to a predictivity of 73% (182/249). The increase of the HTS incubation timesfrom 3 to 5 or 7 days and activation of phase I enzymes by a cocktail of AhR, PXR,and CAR ligands did not improve these results.The evaluation of cytotoxicity with the HCS procedure [111–113] showed that

both HepG2 cells and primary hepatocytes can be used as good representatives forcytotoxicity scoring in 96-well plates. In the case of HCS (Fig. 2.4), when comparedto HTS, different fluorescent dyes are used for measuring the intracellular ionized cal-cium status (Fluo-4 AM), the mitochondrial membrane potential (TMRM or TMRE),the plasma membrane permeability (TOTO-3 and YOYO-1), the localization of boththe nucleus and cytoplasm (DRAQ5TM), whereas the same dyes can be applied for


DNA content measurements (Hoechst 33342 and DAPI). The particular strength ofHCS analysis lies in the fact that all these parameters can be analyzed simultaneouslywithin the same cell. For 243 drugs with varying degrees of in vivo toxicity, HCSappeared to have a sensitivity of 93% and a specificity of 98% [111, 114]. These HCSdata were supported by Vanparys [115], who studied 10 compounds being hepato-,cardio-, and nephrotoxic in the clinic and being non-toxic in vivo in rats, and whodemonstrated that 8 of these compounds could be identified as cytotoxic in HepG2cells. Moreover Abraham et al. [116] measured several anti-infective, antilipidemic,and antidiabetic drugs, and showed that trovafloxacin, cerivastatin, and troglitazonealready caused cytotoxic effects in HepG2 cells at dose levels being 5- to 20-foldlower than those of closely related drugs, which were less toxic or even non-toxic inthe clinic. These HCS data with HepG2 cells mimicked the clinical situation, whereasHCS data obtained with primary human and rat hepatocytes only reached a sensitiv-ity of 60% and 70%, respectively [112, 113]. This implies that in vitro cytotoxicityanalysis with HCS in HepG2 cells improves the sensitivity up to 93%, whereas withHTS this level is approximately 50%.HCS has another advantage as it can also be used to measure phospholipido-

sis (PLD). PLD can occur in hereditary lysosomal storage disorders such as theNiemann–Pick disease. Furthermore, exposure to some drugs can induce PLD. Espe-cially cationic amphiphilic drugs with a hydrophobic domain and a hydrophilic sidechain may induce excessive accumulation of lipids in the lysosomes. This PLD ischaracterized by the presence of lamellar bodies in the lysosomes, which can be visu-alized by electron microscopy, and is often accompanied by vacuolization of cells andinfiltration of foamy macrophages into the tissue. Nile Red, LipidTox and NBD-PEare commonly used markers for PLD analysis [117–122] (Fig. 2.4). However, NBD-PE has the best signal to noise ratio for flow cytometry and HCS, whereas thesefluorophores are too weak for a proper HTS analysis. Treatment of cells with mapro-tiline induces PLD; thereby, NBD-PE accumulates in a concentration-dependentmanner in the cytosol of CHO-K1 cells and more particularly in lysosomes (Fig.2.5). Analysis of a reference set of 56 compounds revealed very high sensitivity andspecificity levels with scores above 80% for CHO-K1 and HepG2 cells [122]. Thesedata are also in line with those of flow cytometry [119, 120] and other HCS methods[117, 118].For the identification of steatosis,NileRed,OilRedO, andLipidTox are commonly

used [123, 124]. Steatosis is recognized by fatty acid accumulation in the cytoplasmof cells and its detection in the case of HepG2 cells can be improved by using fattyacid mixtures. Hereto, the incubation medium is enriched with a low proportion ofpalmitic acid (oleate/palmitate in a 2:1 ratio) leading to mild toxicity representativefor benign chronic steatosis or a high proportion of palmitic acid (oleate/palmitatein a 0:3 ratio) causing an acute and harmful effect by excessive fat accumulation[123]. However, since Nile Red and LipidTox are also common markers for PLD, aproper discrimination between PLD and steatosis is difficult. The use of BODIPY493/503 for steatosis can resolve these problems, as the latter has been shown to beable to discriminate the lipid staining in the cytoplasm from that of lysosomal storage[121, 125]. Both HCS and HTS can be carried out for the detection of steatosis.


FIGURE 2.4 Testing procedures for cytotoxicity testing.

Various techniques are available for the fast analysis of a variety of cytotoxicitymarkers as cellular proliferation, membrane integrity, mitochondrial activity, oxida-tive stress, apoptosis, PLD, and steatosis (Fig. 2.4). The predictivity and, in particular,the assay sensitivity appears to be largely in favor of HCS as compared to HTS, asindividual cells can simultaneously be examined for more than one parameter andwith higher yields of accuracy.

FIGURE 2.5 Visual accumulation of NBD-PE in CHO-K1 cells after exposure to 0.0316 and10 �Mmaprotiline and by calculation of the NBD-PE intensity, showing cell death (solid line)and % of fluorescence of NBD-PE compared with the effect of the positive control, 3.16 �Mamiodarone, set at 100% (gray bars) ± SEM. ∗, Significant induction of phospholipidosis (p <

0.01). DAPI stains the nuclei (left), and NBD-PE the lysosomes around these nuclei (middle).


2.6 ASSAYS FOR NUCLEAR RECEPTOR-INDUCED ACTIVATIONOF PHASE I AND II ENZYMES

Nuclear receptor activation and the influence on cellular phase I and II metabolismis mainly mediated by the following three receptor types: (1) AhR, (2) PXR, and(3) CAR. Most studied in this respect is the induction of the cytochrome P450(CYP) enzymes involved in phase I metabolism : i.e., (1) CYP1A1 and CYP1A2,which are activated in particular by AhR ligands such as 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD), 3-methylcholantrene (3MC), �-naphthoflavone and indirubin; (2)CYP3A4 (human) and CYP3A11 (rat, mice), which are induced by PXR ligandslike rifampicin, T0901317, dexamethasone, mifepristone (RU486), and phenobar-bital (human-specific); (3) CYP2B1 (rat, mice), CYP2B2 (rat, mice), CYP2B6(human) and 2C9 (human), which are activated by CAR ligands like phenobarbi-tal (rat-specific), TCBOPOB (rat- and mouse-specific) and CITCO (human-specific)[126–131]. These inducing effects may have an impact on the metabolism of a drug,making it more or less toxic. Besides phase I metabolism, phase II metabolism alsoplays an important role, as it can modulate the further degradation or modification ofa non-toxic and/or toxic metabolite into a state, which makes it less or more toxic andmuch easier to be excreted. In this respect, glucuronidation, sulfation, N-acetylation,or conjugationwith glutathione are themain reactions involved. Several of these phaseII enzymes are concomitantly induced with the phase I enzymes. More specifically,in the case of HepG2 cells, AhR activation leads to an increase in the mRNA levelsof CYP1A1, CYP1A2, UDP-glucuronosyltransferase (UGT) 1A1, UGT1A6, sulfo-transferase (SULT) 1A1, SULT1A2, SULT2A1, GST-M1, and microsomal epoxidehydrolase (EPHX) and to an increase in the enzyme levels as measured for CYP1A1and CYP1A2. PXR activation, on the other hand, enhanced the mRNA levels ofCYP3A4, UGT1A1 and SULT1A2, while CAR activation induced the mRNA lev-els of CYP2B6, UGT1A1, SULT1A2, SULT2A1 and N-acetyltransferase 1 (NAT-1)[132, 133].The toxicity of acetaminophen is well known [134]. If given at high dosages,

the intermediate N-acetyl-p-benzoquinone imine (NAPQI) is formed. Althoughacetaminophen itself can be directly metabolized by glucuronidation and sulfa-tion, NAPQI is still formed by the metabolic conversion via CYP2E1, CYP3A4,CYP3A11, or CYP1A2. Enhancement of the activity of one of these enzymes byoxidative stress, alcohol, or PXR inducers increases the formation of this metabo-lite and its liver toxic effects, as shown by higher alanine aminotransferase andaspartate aminotransferase levels in serum [135]. This toxicity can be reduced byspecific inhibitors of these particular cytochrome P450-enzymes or by the supple-mentation of additional sulfhydryl donors such as N-acetylcysteine or methionine,which can compensate for the loss of glutathione. These donors form a substitute forglutathione, but can also be used as scavengers for the inflammatory process, which isactivated by the intake of acetaminophen. This inflammation process normally leadsto the production of nitric oxide and finally nitrotyrosine in the necrotic centrilobularregions in the liver, but N-acetylcysteine and methionine can prevent this kind ofaccumulation [136].

NUCLEAR RECEPTOR-INDUCED ACTIVATION OF PHASE I AND II ENZYMES 47

With this short view on the metabolism of acetaminophen, the importance ofthe activation of the phase I enzymes CYP1A2, CYP3A4/11, and CYP2E1 for theinduction of toxicity as well as the activation of phase II enzymes like UGTs, SULTs,and GSTs for the inhibition of toxicity is pointed out. It is valuable to know whichof the CYP enzymes are involved in the metabolism of a drug and which phaseII enzymes are improving their (de)toxification and excretion of the drug. For theassessment of the CYP and phase II enzyme changes, a set or sequence of events canbe analyzed. After the activation of a particular receptor or a combination thereof,this activation can be monitored at two levels:

1. The mRNA expression levels of the phase I and II enzymes by regular quan-titative polymerase chain reaction (qPCR) with Sybergreen or Taqman both incombination with cell lysis buffer or the cell to CT method [137, 138];

2. The phase I and II enzyme activities by HPLC/MS, fluorescence, or lumines-cence.

Furthermore, primary liver cells or permanent liver cell lines from humans, ratsor mice, supersomes and microsomes from a donor pool of different individuals orfrom specifically prepared insect cells that overexpress one of the CYP enzymes canbe used. In this assessment, it is important to monitor all relevant CYP enzymesfor human metabolism and toxicity, i.e., CYP1A1, 1A2, 2B6, 2C8, 2C9, 2C18,2C19, 2D6, 2E1, 3A4, and 4A10 [139]. The enzyme activities can be measuredwith HPLC/MS by directly analyzing the conversion rate of the drug or a specificreference compound as well as the formation of their metabolites or by the use of[3H] or [14C] labeled drugs. Moreover, enzyme competition assays can be carried outwith these drugs for each CYP enzyme using a specific fluorophore and a specificinhibitor as reference [140–144]. For these fluorometric assays purified human CYPenzymes that were cloned in an insect cell line (BD Biosciences) are used (Fig.2.6). For a fast track luminometric analysis with hepatocytes and supersomes, it isnowadays possible to use luciferin substrates, which are specific for several of theseCYP enzymes andwhich can pass the cellular membrane by diffusion. These luciferinsubstrates are available for CYP1A1, 1A2, 2B1, 2C8, 2C9, 2C19, 2D6, 3A4, 3A7,and 4A10 (P450-GLOTM assays, www.promega.com) (Fig. 2.6). For CYP1A1, 1A2,2C9, and 3A4 the quantification with these luciferin substrates worked very well onhuman hepatocytes and HepG2 cells in our hands [132] (Fig. 2.7), while CYP4A10could be measured in rat H4IIE cells (Fig. 2.7).The bioactivity of phase II enzymes can also be examined with qPCR, cell to

CT and enzyme activity measurements. In the case of UGTs, several isoforms areof high importance for drug detoxification. Regarding detoxification, UGT1A1 andUGT1A6 are strongly induced via the AhR, while regarding steroid conversion,UGT1A1, UGT2B15 and UGT2B17 are the main enzymes involved. In case ofcholestasis in the human liver, UGT1A1 plays a crucial role. In Gilbert’s syndrome,one allele is deficient, while in the case of the Crigler–Najjar disease types I andII, the enzyme activity is completely lost or only remains active at a 10% level


FIGURE 2.6 In vitro fluorometric and luminometric tests for several phase I and II enzymes.

[145, 146]. These deficiencies will lead to an increased bilirubin level and jaundice,to early death shortly after birth or to the need of a liver transplantation at the ageof 6 to 8 years. Moreover, in patients with Gilbert’s syndrome, treatment with pro-tease inhibitors like indinavir and atazanavir can lead to severe hyperbilirubinemia,which is even more pronounced when associated with other genetic variants ofUGT1A3 and UGT1A7 [147]. There also appears to be a potential toxicity risk inthe case of the estrogens 2-hydroxyestrone estradiol, and ethinyl estradiol as wellas in the case of further compounds such as gemfibrozil, simvastatin, fulvestrant,ezetimibe, ibuprofen, ketoprofen and buprenorphine, which have been identified assubstrates for UGT1A1 [148]. Although for UGT2B4, 2B15 and 2B17 different iso-forms with altered metabolic capacities have been observed, no clear drug-relatedtoxic effects have yet been identified. An increased potential risk may be there forUGT2B7 metabolized drugs, like morphine, oxazepam, zidovudine and epirubicinas well as for the UGT2B15- and 2B17-metabolized C19 steroids tamoxifen and 4-hydroxytamoxifen [149–152]. The UGT1A1, 1A3, 1A4, 1A6, 1A7, 1A8, 1A9, 1A10,2B4, 2B7, 2B15, and 2B17 enzymes, also purified from UGT-cloned insect cell lines,can be used for amore old-fashioned radiometric analysis with [14C]-UDP-glucuronicacid as a co-factor, for HPLC/MS and for competitive conversion assays with the flu-orophores 7-methyl-4-methylcoumarin or 7-hydroxy-6-methylcoumarin (scopoletin)as substrate in combination with a specific inhibitor [153–157]. Moreover, two dif-ferent luciferin substrates are now available, which may give these UGT assays a

NUCLEAR RECEPTOR-INDUCED ACTIVATION OF PHASE I AND II ENZYMES 49

100

1000

10000

100,000

0.01 0.1 1 10

Δlu

min

esce

nce

/min

/105

cells

Compound concentration (nM)

(a)

100

1000

10000

100,000

0.01 0.1 1 10

Δlu

min

esce

nce

/min

/105

cells

Compound concentration (nM)

(b)

FIGURE 2.7 Measurement of the concentration-dependent activation of cytochrome P450enzymes by means of specific luciferin substrates in human (HepG2) and rat (H4IIE) livercells: (A) Activation of CYP1A1 (light grey bars) and CYP1A2 (dark grey bars) by 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) in human hepatocytes; (B) Activation of CYP3A4 (lightgrey bars) by the Tularik compound T0901317 in human and CYP4A10 (dark grey bars) byWY14.643 in rat liver cells.

much higher sensitivity for the 96- and 384-well assay formats (www.promega.com,www.BD.com) (Fig. 2.6).Such assays can also be performed for GSTs. GSTA1, A2, A4, M2, M4, P1, T2,

and Z1C are all available in pure form for analysis. The spectrophoto-, fluoro-, andluminometric assays can be set up with specific substrates, such as 1-chloro-2,4-dinitrobenzene (CDNB) [158], monochlorobimane (MCB), or luciferin-NT (GST-GloTM, www.promega.com). Polymorphism of GSTM1 and T1 is common amongCaucasians and a complete loss of enzyme activity, the so-called null genotype vari-ants, has been identified in 50% and 20% of the Caucasian population, respectively.The presence of one of these variants alone does not show an influence on drugs,


whereas a combination of both hampers the treatment of chemotherapy leading to areduced survival rate [139, 149, 151]. Moreover, busulfan is metabolized by GSTM1and A1 and treatment of null-patients does lead to increased health risk [139, 151].On the other hand, a GSTP1 polymorphism can lead to an increased survival ratein cancer patients treated with fluorouracil and oxaliplatin, since in the presenceof the mutated and less active variant the drug is rescued from degradation andexcretion [151].In analogy to the UGTs and GSTs, similar assays can be set up for SULTs.

For SULT1A1, 1A2, 1A3, 1E1, and 2A1 purified enzymes are available and theradioligand method with 3′-phosphoadenosine-5′-phosphosulfate (PAPS) as well asHPLC/MS can be performed (Fig. 2.6). However, fluorometric and luminometricassays are still not available for this enzyme, making high-throughput analysis morecumbersome. As a colorimetric assay the use of nitrophenol or the fluorophores 7-methyl-4-methylcoumarin and 7-hydroxy-6-methylcoumarin (scopoletin) could bean option if the conversion levels are high enough for detection. This issue may besolved in case of luminescence, if an alternative luciferin substrate becomes avail-able. For SULT1A1, a clear polymorphism exists in Caucasian andAfrican-Americanpopulations [159, 160]. Substantial evidence is available for the link between thisSULT1A1 polymorphism and an increased lung cancer risk, which is related to thereduced metabolic capacity to degrade tobacco constituents. Metabolism of phenolicdrugs may also lead to complications, as shown for tamoxifen, 4-hydroxytamoxifen,and endoxifen [139, 149]. Besides SULT1A1, SULT1A2, 1B1, 1C1, and 2A1 poly-morphisms have been observed, but so far there is no clear association between thesepolymorphisms and an enhanced risk to develop certain human diseases or otherhealth indications [139, 149, 151].Thus, with respect to phase I and II enzymes, HTS methods are available for

CYPs, UGTs, and GSTs by making use of fluoro- and luminometry, while for SULTsfurther exploration in this area is needed. Overall, phase I and II drug metabolizingenzyme polymorphisms do have a large impact on drug metabolism.

2.7 ASSAYS FOR NUCLEAR STEROID RECEPTOR ACTIVATIONAND ENDOCRINE DISRUPTION

Reproductive failure can occur due to disturbances in the production, secretion, oractivation of steroidal hormones as, for instance, androgens during the embryonicdevelopment. Androgens and androgen receptors are very important for embryonicgender development in inducing the regression process of the Mullerian duct and themaintenance of theWolffian duct [161, 162]. In case of impairment of steroidogenesisor a non-functional androgen receptor, this process is not induced. The absence of aro-matase activity in females leads to the accumulation of the androgen testosterone anddisturbance of folliculogenesis and sociosexual behavior, while in males this leadsto the disruption of spermatogenesis and impaired sexual behavior [163]. Duringpuberty, the hypothalamus, pituitary, and gonad axis becomes operational and herethe secretion of gonadotropin releasing hormone (GnRH) is crucial for the start of the

NUCLEAR STEROID RECEPTOR ACTIVATION AND ENDOCRINE DISRUPTION 51

secretion of luteinizing hormone (LH) by the pituitary gland. LH secretion leads to theinitiation of steroidogenesis from cholesterol toward progesterone and testosterone intheca cells of the ovaries and Leydig cells of the testis. This is followed by the secre-tion of follicle-stimulating hormone (FSH) by the pituitary gland for the biosynthesisof aromatase and estrogens, particularly 17�-estradiol (E2), by granulosa cells in theovaries and Sertoli cells in the testis. Later during adulthood the ovarian estrus cyclein females depends on a balanced secretion between FSH and LH, followed by a sharpburst of LH secretion at the end of the follicular maturation phase, which induces theoocyte maturation and ovulation of the egg cell from the dominant Graafian follicles.After the fertilization of the egg has taken place, the egg or blastocyst is implantedinto the lining of the uterus. If the blastocyst is taken up in the endometrial layer,embryonic development can progress and a placenta may be formed. Progesterone,which is produced in the corpora lutea, supports the endometrium in this phase, andlater on this hormone is formed in the placenta to sustain pregnancy. An early risein estrogen levels, like E2, estrone (E1), or estriol (E3) can disturb the reproductivecycle and induce embryotoxicity, leading to a reduction in the number of fetuses.Pregnancy interruption can also be induced by a so-called morning-after pill, whichat most times is an antiprogestagen acting as an antihormone on the progesteronereceptor and its activation or maintenance processes. Moreover, glucocorticoids arealso known to induce malformations like cleft palate during embryonic development,and to disturb folliculogenesis, oocyte maturation, and ovulation during adulthood.A lack of corticosteroids and mineralocorticoids, in particular aldosterone, cortisoland/or corticosterone, can cause death within hours to days after birth. In case of adeficiency of 21-hydroxylase activity, which is crucial for the synthesis of glucocor-ticoids and mineralocorticoids, the androgen levels in utero reach very high levels,thereby leading to the formation of Wolffian ducts in the female (and male) offspring.Moreover, the lack of mineralocorticoids hampers the reuptake of salts in the kidney,especially that of sodium, leading to death soon after birth. Supplementation of thefemale fetus with aldosterone, cortisol, dexamethasone or similar drugs in utero isneeded to prevent this embryonic virilization, whereas treatment of both girls andboys is needed in utero and later life for the regulation of their salt, glucose, and stressbalance to reduce their mortality and morbidity rates [164]. For glucocorticoid recep-tor knockout, dimerization-defective glucocorticoid receptor and mineralocorticoidreceptor knockout mice, it has been shown that these receptors and their dimerizationare crucial in the activation of gene-regulated physiological processes, as the absenceof receptor dimerization leads unequivocally to normal prenatal development, butstill to early birth death [165, 166].Cholesterol-derived steroids like estrogens, androgens, progestagens, mineralo-

corticoids, and glucocorticoids act via their cognate nuclear steroid receptors. Theagonistic and antagonistic activities of the drug of interest or one of its metabolitescan be determined for these human estrogen, androgen, progestagen, glucocorticoid,and mineralocorticoid receptors in CHO cells [144], human uterotrophic osteoblasts,that is, U2OS cells [167, 168], and yeast cells [169, 170] (Fig. 2.8), all contain-ing a specific promoter and luciferase–reporter-based readout. Another possibility toanalyze the steroid receptor activity is to determine the rate of proliferation by using


FIGURE 2.8 Testing procedures for endocrine disruption, non-genotoxic carcinogenicityand embryotoxicity.

human receptor containing breast tumor cell lines such as humanMCF-7 [171], T47D[172, 173], ZR-75-1 [174, 175] and MDA-MB-453 cells [176], or prostate tumor celllines like human LNCaP [177] and PC-93 cells [178]. For GnRH [179], FSH andLH receptors [180, 181], such reporter-based promoter assays are also available(Fig. 2.8), as small molecules may compete for these receptors and form anotherpoint of risk for endocrine disruption.

2.8 ASSAYS FOR NUCLEAR RECEPTOR ACTIVATIONAND NON-GENOTOXIC CARCINOGENICITY

When considering nuclear receptor activation and its role in non-genotoxic carcino-genicity, three receptor types play a dominant role, these being AhR [182–184], theperoxisomal proliferation-activating receptor � (PPAR�) [185], and CAR [186, 187].For all three receptors, there is a clear causal relation between receptor activation andcarcinogenicity in rodent studies and the absence of carcinogenicity in knockout micefor these specific receptors. With respect to PPAR� and CAR it has been shown thatthe introduction of the human PPAR� and CAR variants in knockout mice does notresult in the induction of carcinogenesis [188, 189]. The molecular structure of thehuman protein receptor seems to particularly prevent the activation of the cytochrome

ASSAYS FOR NUCLEAR RECEPTOR ACTIVATION AND EMBRYOTOXICITY 53

P450 enzymes, CYP4A10 by PPAR� and CYP2B1/CYP2B2 by CAR as well as a setof other specific genes (Fig. 2.8). This kind of pattern recognition was confirmed witha hierarchical clustering example for in vivo rat liver analysis by toxicogenomics forcompounds interacting with PPAR�, CAR, PXR liver X receptor (LXR), and AhR(Fig. 2.9a), and more in detail for the activation of CYP2B1 by CAR compounds(Fig. 2.9b). Figure 2.9 shows the selectivity of activation and inactivation of gene setsdue to treatments with specific nuclear receptor compounds. Species specificity, atleast for PPAR� and CAR, plays an important role when looking at carcinogenicity inparticular. On the other hand, a specific development candidate was shown in-houseto cause carcinogenesis in rats via AhR by activation of CYP1A1 and CYP1A2. Alsoin vitro, this compound bound to the rat AhR as shown with radiolabeled compet-itive binding experiments and in vitro activation of the CYP1A1 and 1A2 enzymesin rat H4IIE cells. However, this compound was unable to activate CYP1A1 andCYP1A2 in HepG2 cells, both at the mRNA and enzyme activity level, while allother reference compounds tested, such as TCDD, 3-MC, indirubin, and indigo werenormally active in both H4IIE and HepG2 cells [190]. Thus, not all inducers ofrodent carcinogenesis via AhR, PPAR� or CAR are per definition inducers of humancarcinogenicity. For instance, specific fruits and vegetables are very healthy, but stillinduce AhR activity and its pathways. Resveratrol present in wines is known to bean AhR antagonist [191], while many flavonoids are AhR (ant)agonists and are ableto modulate phase I and II enzyme activities [192, 193]. In rats, indigo is a morepotent agonist than indirubin, while in humans it is the other way around. Moreover,in rats, indigo derivatives prevent tumor induction and tumor growth [194, 195],while in Chinese healthcare indirubin is applied for the treatment of chronic myeloidleukemia and prostate tumors. Recently, an even more specific natural ligand of AhR,6-formylindolo[3,2-b]carbazole (FICZ), formed in the skin during exposure to UVlight, has been characterized as being a possible protective (anticarcinogenic) agentafter UV exposure [196]. This natural ligand appears to be an equally potent activatorof AhR as TCDD.Thus, for early risk assessment, comparison of human and rat receptor outcomes

may be beneficial. The analysis for AhR, PPAR�, and CAR has been described aboveand fluorometric as well as luminometric substrates can be used in combination withthe quantification of mRNA levels of the respective cytochrome P450 enzymes byquantitative PCR or cell to CT methods.

2.9 ASSAYS FOR NUCLEAR RECEPTOR ACTIVATIONAND EMBRYOTOXICITY

For every drug intended for use in women with child-bearing potential, costly in vivotests in rats and rabbits are regulatory requirements. The teratogenic effects elicitedby thalidomide in the 1960s formed the basis for these stringent regulatory procedures[197]. However, these pivotal, but time-consuming tests are scheduled quite late indevelopment. So, a lot of money may already have been spent in (pre)clinical studies


(a)

(b)

FIGURE 2.9 Hierarchical clustering of various nuclear receptor agonists for the aryl hydro-carbon receptor (AhR), constitutive androstane receptor (CAR), liver X receptor (LXR), per-oxisomal proliferation activating receptor � (PPAR�) and pregnane X receptor (PXR) (a) anda magnification of the specific CAR- and PXR-mediated gene activation and inactivation ofCYP2B and CYP3A in particular (b). In the white rectangles A, B and C, the specific up-regulated genes mediated by CAR, PPAR�, and LXR, respectively, are shown, while in D andE the specific PXR- and CAR-modulated CYP3A and CYP2B are visible. (See insert for colorrepresentation of this figure.)

ASSAYS FOR NUCLEAR RECEPTOR ACTIVATION AND EMBRYOTOXICITY 55

before the occurrence of embryotoxicity may lead to the final decision to stop thefurther development of a compound.Therefore, a big effort was put into the development of easy and cheap in vitro

screening assays for the prediction of embryotoxicity. However, embryonic develop-ment is rather complex due to the interaction of the mother with the fetus and not allfacets can be identified in one or several in vitro models. On the other hand, for thepurpose of HTS, the predictive value for embryotoxicity does not need to be above95%, as a predictive value of 70–80% for a preselection might be sufficient in in vitromodel systems.Under the coordination of ECVAM, an extensive interlaboratory validation of

the three most commonly used in vitro embryotoxicity tests has been performed[198]. These three evaluated assays are the whole embryo culture, the micromassassay, and the embryonic stem cell test. Detailed reports on this validation have beenreleased and demonstrate [199, 200] a reasonable to good predictive value, sensitivity,specificity, and selectivity. Especially the micromass assay and the embryonic stemcell test have the potency to be transferred into a high- to medium-throughput format.In order to validate these assays, reference compounds like 5-fluorouracil, 5-bromo-2-deoxyuridine, and methotrexate, all inhibitors of thymidine incorporation into theDNA, as well as all-trans-retinoic acid (vitamin A) were used [201]. The first threecompounds are genotoxicants and cytostatic agents, whereas vitamin A is a retinoicacid receptor (RAR�, �, � ) agonist. Moreover, the embryotoxic compounds valproicacid and methoxyacetic acid were used in these studies. Interestingly, these latter twocompounds appear to increase the sensitivity for estrogen, progesterone, androgen,and thyroid hormone receptors from 2- to 8-fold [202, 203], while methoxyaceticacid appears to be a specific inhibitor of sarcosine oxidase/dehydrogenase presentin the choline shunt and responsible for the synthesis of tetrahydrofolate (THF)derivatives, that is, 5-methyl-THF, 10-formyl-THF, and 5,10-methylene-THF thatare crucial intermediates and/or co-factors in thymidine synthesis [204, 205]. Inthe case of methoxyacetic acid, a link between embryotoxicity and the synthesis ofTHF derivatives and thymidine is made. In this respect, not only alcohol, but alsoinhibitors of aldehyde dehydrogenases andmore specifically of the retinol, retinal, andretinaldehyde dehydrogenases may influence embryotoxicity due to the accumulationof (retin)aldehyde intermediates. At least for acetaldehyde, a breakdown productof ethanol, embryotoxicity is unequivocal [206, 207]. The embryotoxic effects ofmethoxyethanol, ethoxyethanol and other alcoholic variants are alsowell documented[208, 209].In the context of embryotoxicity, two lines of research are becoming important. The

teratogenic effects of vitamin A, thyroid hormones, diethylstilbestrol (a non-steroidalestrogen) and androgens are well known. Therefore, their cognate nuclear receptorassays are of interest. A second line of research may be the interaction of drugs withalcohol and aldehyde oxidases/dehydrogenases involved in the degradation of alco-hols and aldehydes into the corresponding acids and the possibility of product inhibi-tion of these oxidases/dehydrogenases aswell as those involved in the choline shunt asprecursors of THF derivatives, i.e., choline, dimethylglycine and sarcosine oxidasesand dehydrogenases, which more or less similar represent chemical reduction steps.


Vitamin A plays an important role during neural development. Exogenous admin-istration of vitamin A can cause an open spinal cord during embryonic development[210]. Moreover, an excess of vitamin A may also lead to other malformations in theface, limbs, heart, and central nervous system [211, 212]. These different dosagesmay lead to RAR and RXR activation, while a clear synergism between these recep-tors may enforce other gene activities at lower dosages and thereby potentiate theteratogenic effects [213]. Moreover, the changes in vitamin A levels can also beinfluenced by inhibitors of alcohol, acetaldehyde, retinol, retinal, and retinaldehydedehydrogenases or deficiencies in these enzymes. As gene polymorphisms leadingto malformations have also been described for these enzymes, they should be viewedas potential targets that, if hit, may lead to teratogenicity [214].For the thyroid hormone receptor (TR), the thyroid hormones 3,5,3′-

triiodothyronine (T3) and 3,5,3′,5′-tetraiodothyronine (thyroxine, T4) are of crucialimportance. Thyroid hormones are involved in the development of extremities orthe disappearance of, for example, the tail in tadpoles and frogs [215, 216]. Verylow levels of thyroid hormone result in stillbirths, abortions, a decrease in the lit-ter size, and congenital malformations [217]. Very high levels of thyroid hormone,on the other hand, lead to additional fingers, cleft lip and palate, an increase inskeletal malformations, and ear deformities [218]. Thiouracil, propylthiouracil, goldthioglucose, and iodoacetate can prevent the formation of T3 and T4 at the iodideoxidation level by inhibiting the enzyme deiodinase [219, 220], while resorcinol,carbimazole, p-aminobenzoic acid, and sulfaguanidine can act as goitrogenic sub-stances [221]. Overexpression of deiodinase type III in amphibia can prevent tailregeneration [222], and this type of deiodinase is largely expressed in the placenta toprevent maternal thyroid hormone to influence the fetus [223].For human RAR and RXR activation, an assay has been developed in HepG2

cells, which clearly shows the synergism between the different RAR (�,�,� ) andRXR (�,�,� ) isoforms, leading to differences in the potency or efficacy of combinedRAR and RXR ligands to activate the CYP26A promoter. The interference withalcohol, acetaldehyde, retinol, and retinaldehyde dehydrogenases may be examinedwith this CYP26A luciferase reporter assay or by direct inhibition of the CYP26A1protein, which enhances the retinoic acid levels [224]. For TR, reporter assays inCHO cells have been developed, and it has been shown that bisphenol A derivativescould be active as TR antagonists and TR binders [225]. These antagonists appearto prevent the T3-induced tail reduction in tadpoles, which confirms the strength ofthe assay format. For AR and ER, research lines were described in the paragraphon endocrine disruption. Moreover, biochemical analyses on the different humanisoforms of alcohol, acetaldehyde, retinol, retinal, and retinaldehyde dehydrogenasesmay be performed by making use of spectrophotometry, while iodination and deiod-ination of T3 and T4 could be studied with radioactive iodine (125I).The increase in knowledge regarding developmental processes in embryos and

pups/babies makes new screening methods become available. In this respect, espe-cially the development of zebrafish eggs into small larvae and fish as a new state-of-the-art technology is worth to be mentioned. First of all, experiments with zebrafisheggs and larvae are not part of the legislation of animal experimentation. Moreover,

MULTIPLE SIZE HTS FORMAT TO STUDY MECHANISMS OF ACTION 57

the egg and larval development can be studied in a 96-well plate, which makesembryonic development studies relatively easy. Failures in neural development canbe followed quite well in the transparent eggs, while, if hatched, the locomotor activ-ity can be studied [226, 227]. Since the eggs and larvae grow in normal fresh water,compound treatment can be easily performed by adding compounds to the water. Thecompounds can be taken up by diffusion through the skin and in later age through thegills. The progress in this field within the past years is huge and fast progress makesthis in vivo assay very worthwhile as an HTS tool.

2.10 TOXICOGENOMICS: A DIFFERENT MULTIPLE SIZE HTSFORMAT TO STUDY MECHANISMS OF ACTION

Toxicogenomics is a kind of HTS format at the gene level. With toxicogenomics, onecan analyze the activation or inactivation of the complete set of 20,000 up to 50,000gene fragments in rats, mice, dogs, monkeys, or humans in different tissues and bya variety of compounds. In this way, it becomes feasible to track down the mecha-nism of action that a compound induces in time with different dosages [228, 229].Although the analysis is more complex than general HTS, the information obtainedis genome wide and not focused on one particular enzyme, protein, or receptor. Thisdelivers the opportunity to obtain a complete fingerprint of gene changes per individ-ual compound. A discrimination between different kinds of compound classes caneasily be made by hierarchical clustering and principal component analysis on allor subsets of genes. With this technique, all kinds of pathological compounds canbe classified in vivo and subdivided into groups. For example, in the case of com-pounds being toxic in the liver, one could differentiate between necrotic, apoptotic,cholestatic, steatotic, genotoxic, carcinogenic, non-genotoxic carcinogenic, embry-otoxic, and immunotoxic effects. Thereby, it will become feasible to prioritize thenewly developed drugs as being at risk or free of risk. The subset of genes per classobtained can then be used for extrapolation to the in vitro setting.Despite the fact that different mechanisms are active in the in vivo situation with

respect to the in vitro situation, the biochemical-activated pathways can be similarand characteristic. For instance, a comparison of in vivo rat liver tissue with in vitrorat precision-cut liver slices showed that the outcome for acetaminophen and carbontetrachloride with respect to the toxicological claim with ToxShieldTM, a programfrom GeneLogic (now Ore Pharmaceuticals), which characterized the in vivo genechanges on chips, was necrosis in both the in vivo and in vitro situations [230]. Forlipopolysaccharide, a correlation with the gene signature of inflammation was alsofound. More in-depth analysis in this respect is ongoing for these and other necroticand cholestatic compounds.In a similar fashion, we also analyzed the effects of genotoxic, necrotic, and

cholestatic compounds in vitro on human HepG2 cells. The data analysis with prin-cipal component analysis and hierarchical clustering immediately showed a clearsegregation of more than 200 up- and down-regulated genes. This again shows thatvery specific fingerprints and segregation into different classes becomes feasible


FIGURE 2.10 Hierarchical clustering of gene fragments in HepG2 cells treated for 24 or72 h with genotoxic (camptothecin [CAMP], ethinyl estradiol [EE], methyltestosterone [MT],cisplatin [Cis-Pt], doxorubicin (DOX), benzo[a]pyrene [B(a)P], all marked in yellow), andnon-genotoxic but cholestatic (�-naphthylisothiocyanate [ANIT], chlorpromazine [Cl-Pro]) ornecrotic compounds (tetracarbon chloride [CCl4], iproniazid [Ipro], acetaminophen [APAP],all marked in blue). (See insert for color representation of this figure.)

by the selection of the appropriate gene sets for discrimination (Fig. 2.10). In thefuture, this kind of segregation analysis will become more generalized with the newtechnological developments in system biology.

2.11 CONCLUSIONS

Implementation of an extensive battery of in vitro toxicological screening tests iscrucial for the pharmaceutical industry to reduce the attrition rate in late-stage

REFERENCES 59

development. These tests can easily lead to the selection and ranking of drugs inthe stage of preclinical development. Good validation of these in vitro methods within vivo data to obtain the best predictive tools will finally lead to a change in mindsetting and legislation regarding their use to reduce the number of animal tests.General toxicity and more specifically, genotoxicity, carcinogenicity, non-genotoxiccarcinogenicity, cytotoxicity, endocrine disruption, embryotoxicity, and metabolicinteractions can be monitored by making use of HTS or HCS, thereby leading toa more prominent role of in vitro toxicology in the risk assessment process in thefuture.

REFERENCES

1. Braun, A., Hammerle, S., Suda, K., Rothen-Rutishauser, B., Gunthert, M., Kramer,S.D., Wunderli-Allenspach, H. (2000). Cell cultures as tools in biopharmacy. EuropeanJournal of Pharmaceutical Sciences, 11(suppl 2), S51–S60.

2. Zucco, F., De Angelis, I., Testai, E., Stammati, A. (2004). Toxicology investigations withcell culture systems: 20 years after. Toxicology In Vitro, 18, 153–163.

3. Lerche-Langrand, C., Toutain, H.J. (2000). Precision-cut liver slices: characteristics anduse for in vitro pharmaco-toxicology. Toxicology, 153, 221–253.

4. Entzeroth, M., Flotow, H., Condron, P. (2003). Overview on high-throughput screening.Current Protocols in Pharmacology, 9.4, 1–23.

5. Fan, F., Wood, K.V. (2007). Bioluminescent assays for high throughput screening. Assayand Drug Development Technologies, 5, 127–136.

6. Hertzberg, R.P., Pope, A.J. (2000). High-throughput screening: new technology for the21st century. Current Opinion in Chemical Biology, 4, 445–451.

7. Sundberg, S.A. (2000). High-throughput and ultra high-throughput screening: solution-and cell-based approaches. Current Opinion in Biotechnology, 11, 47–53.

8. Gough, A.H., Johnston, P.A. (2006). Requirements, features, and performance of highcontent screening platforms. In: Taylor, D.L., Haskins, J.R., Giuliano, K., editors. HighContent Screening: A Powerful Approach to Systems Cell Biology and Drug Discovery.Totowa: Humana Press Inc.; Methods in Molecular Biology, 356, pp. 41–61.

9. Starkuviene, V., Pepperkok, R. (2007). The potential of high-content high-throughputmicroscopy in drug discovery. British Journal of Pharmacology, 152, 62–71.

10. Zanella, F., Lorens, J.B., Link, W. (2010). High content screening: seeing is believing.Trends in Biotechnology, 28, 237–245.

11. Kassel, D.B. (2004). Applications of high-throughput ADME in drug discovery. CurrentOpinion in Chemical Biology, 8, 339–345.

12. Saunders, K.C. (2004). Automation and robotics in ADME screening. Drug DiscoveryToday, 1, 373–380.

13. Kola, I., Landis, J. (2004). Can the pharmaceutical industry reduce attrition rates? NatureReviews Drug Discovery, 3, 711–715.

14. Guengerich, F.P., MacDonald, J.S. (2007). Applying mechanisms of chemical toxicityto predict drug safety. Chemical Research in Toxicology, 20, 344–369.

15. Schoonen, W.G.E.J., Westerink, W.M.A., Horbach, G.J. (2008). In vitro toxicity screen-ing as pre-selection tool. European Pharmacological Review, 3, 40–44.


16. Ames, B.N., Lee, F.D., Durston, W.E. (1973). An improved bacterial test system for thedetection and classification of mutagens and carcinogens. Proceedings of the NationalAcademy of Sciences USA, 70, 782–786.

17. McCann, J., Choi, E., Yamasaki, E., Ames, B.N. (1975). Detection of carcinogens asmutagens in the Salmonella/microsome test: assay of 300 chemicals. Proceedings of theNational Academy of Sciences USA, 72, 5135–5139.

18. Mortelmans, K., Zeiger, E. (2000). The Ames Salmonella/microsome mutagenicityassay. Mutation Research, 455, 29–60.

19. Musatov, S.A., Anisimov, V.N., Andre, V., Vigreux, C., Godard, T., Gauduchon, P.,Sichel, F. (1998). Modulatory effects of melatonin on genotoxic response of referencemutagens in the Ames test and the comet assay.Mutation Research, 417, 75–84.

20. Jemnitz, K., Veres, Z., Torok, G., Toth, E., Vereckey, L. (2004). Comparative study inthe Ames test of benzo[a]pyrene and 2-aminoanthracene activation using rat hepatic S9and hepatocytes following in vivo or in vitro activation. Mutagenesis, 19, 245–250.

21. Fluckiger-Isler, S., Baumeister, M., Braun, K., Gervais, V., Nasler-Nguyen, N., Reimann,R., Van Gompel, J., Wunderlich, H.-G., Engerhardt, G. (2004). Assessment of the perfor-mance of the Ames IITM assay: a collaborative study with 19 coded compounds.MutationResearch, 558, 181–189.

22. Piegorsch, W.W., Simmons, S.J., Margolin, B.H., Zeiger, E., Gidrol, X.M., Gee, P.(2000). Statistical modelling and analyses of a base-specific Salmonella mutagenicityassay. Mutation Research, 467, 11–19.

23. Lelie van der, D., Regniers, L., Borremans, B., Provoost, A., Verschaeve, L. (1997).The VITOTOX R© test, a SOS bioluminescence Salmonella typhimurium test to measuregenotoxicity kinetics. Mutagenesis, 20, 449–454.

24. Van Gompel, J., Woestenborghs, F., Beerens, D., Mackie, C., Cahill, P.A., Knight, A.W.,Billinton, N., Tweats, D.J., Walmsley, R.M. (2005). An assessment of the utility of theyeast GreenScreen assay in pharmaceutical screening.Mutagenesis, 20, 449–454.

25. Verschaeve, L., Van Gompel, J., Thilemans, L., Regniers, L., Vanparys, P., van der Lelie,D. (1999). VITOTOX R© bacterial genotoxicity and toxicity test for the rapid screening ofchemicals. Environmental and Molecular Mutagenesis, 33, 240–248.

26. Miller,V.P., Stresser,D.M.,Blanchard,A.P., Turner, S., Crespi, C.L. (2000). Fluorometrichigh-throughput screening for inhibitors of cytochrome P450. Annals of the New YorkAcademy of Sciences, 919, 26–32.

27. (a) McGregor, J.T., Casciano, D., Muller, L. (2000). Strategies and testing methods foridentifying mutagenic risks.Mutation Research, 455, 3–20. (b) Mortelmans, K., Zeiger,E. (2000). The Ames Salmonella/microsome mutagenicity assay. Mutation Research,455, 29–60.

28. Jacobson-Kram, D., Contrera, J.F. (2007). Genetic toxicity assessment: employing thebest science for human safety evaluation part I: early screening for potential humanmutagens. Toxicological Sciences, 96, 16–20.

29. Kirkland, D., Pfuhler, S., Tweats, D., Aardema, M., Corvi, R., Darroudi, F., Elhajouji,A., Glatt, H., Hastwell, P., Hayashi, M., Kasper, P., Kirchner, S., Lynch, A., Marzin, D.,Maurici, D., Meunier, J.-R., Muller, L., Nohynek, G., Parry, J., Parry, E., Thybaud, V.,Tice, R., van Benthem, J., Vanparys, Ph., White, P. (2007). How to reduce false positiveresults when undertaking in vitro genotoxicity testing and thus avoid unnecessary follow-up animal tests: report of an ECVAMWorkshop. Mutation Research, 628, 31–55.

REFERENCES 61

30. Tweats, D.J., Scott, A.D., Westmoreland, C., Carmichael, P.L. (2007). Determinationof genetic toxicity and potential carcinogenicity in vitro—challenges post the seventhamendment to the European cosmetics directive. Mutagenesis, 22, 5–13.

31. Hoffman, A.F., Garippa, R.J. (2006). A pharmaceutical company user’s perspective onthe potential of high content screening in drug discovery. In: Taylor, D.L., Haskins,J.R., Giuliano, K., editors. High Content Screening: A Powerful Approach to SystemsCell Biology and Drug Discovery .Totowa: Humana Press Inc.; Methods in MolecularBiology, 356, pp. 19–31.

32. Diaz, D., Scott, A., Carmichael, P., Shi, W., Costales, C. (2007). Evaluation of anautomated in vitro micronucleus assay in CHO-K1 cells.Mutation Research, 630, 1–13.

33. Westerink, W.M.A., Schirris, T.J.J., Horbach, G.J., Schoonen, W.G.E.J. (2011). Thedevelopment and validation of a high content screening in vitro micronucleus assay inCHO-K1 and HepG2 cells.Mutation Research, 724, 7–21.

34. Hashimoto,K., Nakajima,Y.,Matsumara, S., Chatani, F. (2010).An in vitromicronucleusassaywith size-classifiedmicronucleus counting to discrimate aneugens from clastogens.Toxicology In Vitro, 24, 208–216.

35. Baumstark-Khan, C., Hellweg, C.E., Reitz, G. (2010). Cytotoxicity and genotoxicityreporter systems based on the use of mammalian cells. Advances in Biochemical Engi-neering/Biotechnology, 118, 113–151.

36. van Delft, J.H., van Agen, E., van Breda, S.G., Herwijnen, M.H., Staal, Y.C., Kleinjans,J.C. (2005). Comparison of supervised clustering methods to discriminate genotoxicfrom non-genotoxic carcinogens by gene expression profiling. Mutation Research, 575,17–33.

37. Ellinger-Ziegelbauer, H., Stuart, B., Wahle, B., Bomann, W., Ahr, H.J. (2004). Charac-teristic expression profiles induced by genotoxic carcinogens in rat liver. ToxicologicalSciences, 77, 19–34.

38. Goode, E.L., Ulrich, C.M., Potter, J.D. (2002). Polymorphisms in DNA repair genesand associations with cancer risk. Cancer Epidemiology, Biomarkers & Prevention, 11,1513–1530.

39. Knight, A.W., Little, S., Houck, K., Dix, D., Judson, R., Richard, A.,McCarroll, N., Aker-man, G., Yang, C., Birrell, L., Walmsley, R.M. (2009). Evaluation of high-throughputgenotoxicity assays used in profiling the US EPA ToxCast chemicals. Regulatory Toxi-cology and Pharmacology, 55, 188–199.

40. Lio, Y.C., Schild, D., Brenneman, M.A., Redpath, J.L., Chen, D.J. (2004). HumanRad51C deficiency destabilizes XRCC3, impairs recombination, and radiosensitizesS/G2-phase cells. Journal of Biological Chemistry, 279, 42313–42320.

41. Malmlof, M., Paajarvi, G., Hogberg, J., Stenius, U. (2008). Mdm2 as a sensitive andmechanistically informative marker for genotoxicity induced by benzo[a]pyrene anddibenzo[a,l]pyrene. Toxicological Sciences, 102, 232–240.

42. Noon, A.P., Vlatkovic, N., Polanski, R., Maguire, M., Shawki, H., Parsons, K., Boyd,M.T. (2010). p53 andMDM2 in renal cell carcinoma: biomarkers for disease progressionand future therapeutic targets? Cancer, 116, 780–790.

43. Park, S.Y., Lee, S.M., Ye, S.K., Yoon, S.H., Chung, M.H., Choi, J. (2006).Benzo[a]pyrene-induced DNA damage and p53 modulation in human hepatoma HepG2cells for the identification of potential biomarkers for PAH monitoring and risk assess-ment. Toxicological Letters, 167, 27–33.


44. Rotterud, R., Berner, A., Holm, R., Skovlund, E., Fossa, S.D. (2001). p53, p21 andmdm2expression vs the response to radiotherapy in transitional cell carcinoma of the bladder.British Journal Urology International, 88, 202–208.

45. Thacker, J., Zdzienicka, M.Z. (2003). The mammalian XRCC genes: their roles in DNArepair and genetic stability. DNA Repair, 2, 655–672.

46. de Vries, A., van Steeg, H., Breit, T.M., Luijten, M. (2009). Finding transcriptomicsbiomarkers for in vivo identification of (non-)genotoxic carcinogens using wild-type andXpa/p53 mutant mouse models. Carcinogenesis, 30, 1805–1812.

47. Xu, G.W., Mawji, I.M., Macrae, C.L., Koch, C.A., Datti, A., Wrana, J.F., Dennis, J.W.,Schimmer, A.D. (2008). A high-content chemical screen identifies ellipticine as a mod-ulator of p53 nuclear localization. Apoptosis, 13, 413–422.

48. Mattsson, A., Jernstrom, B., Cotgreave, I.A., Bajak, E. (2009). H2AX phosphorylationin A459 cells induced by the bulky and stable DNA adducts of benzo[a]pyrene anddibenzo[a,l]pyrene diol epoxides. Chemico-Biological Interactions, 177, 40–47.

49. Giuliano, K.A., Chen, Y.-T., Taylor, D.L. (2004). High content screening with siRNAoptimizes a cell biological approach to drug discovery: defining the role of p53 activationin the cellular response to anticancer drugs. Journal of Biomolecular Screening, 9, 557–568.

50. Barber, P.R., Locke, R.J., Pierce, G.P., Rothkamm, K., Vojnovic, B. (2007). Gamma-H2AX foci counting: image processing and control software for high-content screening.Proceedings SPIE, 6441, 1M1–1M10.

51. Billinton, N., Barker, M.G., Michel, C.E., Knight, A.W., Heyer, W.-D., Goddard, N.J.,Fielden, P.R., Walmsley, R.M. (1998). Development of a green fluorescent proteinreporter for a yeast genotoxicity biosensor. Biosensors & Bioelectronics, 13, 831–838.

52. Cahill, P.A., Knight, A.W., Billington, N., Barker, M.G., Walsh, L., Keenan, P.O.,Williams, C.V., Tweats, D.J., Walmsley, R.M. (2004). The GreenScreen R© genotoxic-ity assays: a screening validation programme. Mutagenesis, 19, 105–119.

53. Hastwell, P.W., Chai, L.-L., Roberts, K.J., Webster, Th.W., Harvey, J.S., Rees, R.W.,Walmsley, R.M. (2006). High-specificity and high-sensitivity genotoxicity assessmentin a human cell line: validation of the GreenScreen HC GADD45a-GFP genotoxicityassay. Mutation Research, 607, 160–175.

54. Knight, A.W., Billinton, N., Cahill, P.A., Scott, A., Harvey, J.S., Roberts, K.J., Tweats,D.J., Keenan, P.O., Walmsley, R.M. (2007). An analysis of results from 305 compoundstested with the yeast RAD54-GFP genotoxicity assay (GreenScreen GC) - includingrelative predictivity of regulatory tests and rodent carcinogenesis and performance withautofluorescent and coloured compounds. Mutagenesis, 22, 409–416.

55. Cole, G.M., Schild, D., Lovett, S.T., Mortimer, R.K. (1987). Regulation of RAD54- andRAD52-lacZ gene fusions in Saccharomyces cerevisiae in response to DNA damage.Molecular and Cellular Biology, 7, 1078–1084.

56. Averbeck, D., Averbeck, S. (1994). Induction of the genes RAD54 and RNR2 by variousdamaging agents in Saccharomyces cerevisiae. Mutation Research, 315, 123–138.

57. Westerink,W.M.A., Stevenson, J.C.R., Lauwens, A., Griffioen, G., Horbach, G.J., Schoo-nen, W.G.E.J. (2009). Evaluation of the VitotoxTM and RadarScreen assays for the rapidassessment of genotoxicity in the early research phase of drug development. MutationResearch, 676, 113–130.

REFERENCES 63

58. Westerink, W.M.A., Stevenson, J.C.R., Horbach, G.J., Schoonen, W.G.E.J. (2010). Thedevelopment of p53, RAD51C, Cystatin A, and Nrf2 luciferase reporter assays inmetabolic competent HepG2 cells for the mechanistic genotoxicity and oxidative stressassessment in the early phase of the research.Mutation Research, 696, 21–40.

59. Clever, B., Interthal, H., Schmuckli-Maurer, J., King, J., Sigrist, M., Heyer, W.-F. (1997).Recombinational repair in yeast: functional interactions between Rad51 and Rad54proteins. EMBO Journal, 16, 2535–2544.

60. Dronkert,M.L.G., Beverloo, H.B., Johnson, R.D., Hoeijmakers, J.H.J., Jasin,M., Kanaar,R. (2000). Mouse RAD54 affects DNA double-strand break repair and sister chromatidexchange. Molecular and Cellular Biology, 20, 3147–3156.

61. Knasmuller, S., Mersch-Sundermann, V., Kevekordes, S., Darroudi, F., Huber, W.W.,Hoelzl, C., Bichler, J., Majer, B.J. (2004). Use of human-derived liver cell lines forthe detection of environmental and dietary genotoxicants; current status of knowledge.Toxicology, 198, 315–328.

62. Mersch-Sundermann, V., Knasmuller, S., Wu, X.-J., Darroudi, F., Kassie, F. (2004). Useof a human-derived liver cell line for the detection of cytoprotective, antigenotoxic andcogenotoxic agents. Toxicology, 198, 329–340.

63. Ekwall, B. (1980). Toxicity to HeLa cells of 205 drugs as determined by the metabolicinhibition test supplemented by microscopy. Toxicology, 17, 273–295.

64. Ekwall, B. (1980). Preliminary studies on the validity of in vitro measurement of drugtoxicity using HeLa cells III. Drug toxicity in the MIT-24 system compared with mouseand human lethal dosage of 52 drugs. Toxicological Letters, 5, 309–317.

65. Ekwall, B. (1980). Preliminary studies on the validity of in vitro measurement of drugtoxicity using HeLa cells III. Lethal action to man of 43 drugs related to the HeLa celltoxicity of the lethal drug concentrations. Toxicological Letters, 5, 319–331.

66. Ekwall, B., Johansson, A. (1980). Preliminary studies on the validity of in vitromeasure-ment of drug toxicity using HeLa cells. I. Comparative in vitro cytotoxicity of 27 drugs.Toxicological Letters, 5, 299–307.

67. Bondenson, I., Ekwall, B., Hellberg, S., Romert, L., Stenberg, K., Walum, E. (1989).MEIC - A new international multicenter project to evaluate the relevance to humantoxicity of in vitro cytotoxicity tests. Cell Biology and Toxicology, 5, 331–347.

68. Clemedson,C.,McFarlane-Abdulla, E.,Andersson,M., Barile, B., Calleja,M.C., Chesne,C., Clothier, R., Cottin, M., Curren, R., Dierickx, P.J., Ferro, M., Fiskejo, G., Garza-Ocanas, L., Gomez-Lechon, M.J., Gulden, M., Isomaa, B., Janus, J., Judge, P., Kahru,A., Kemp, R.B., Kerszman, G., Kristen, U., Kunimoto, M., Karenlampi, S., Lavrijsen,K., Lewan, L., Lilius, H., Malmsten, A., Ohno, T., Persoone, G., Pettersson, R., Roguet,R., Romert, L., Sandberg, M., Sawyer, Th.W., Seibert, H., Shrivastava, R., Sjostrom,M., Stammati, A., Tanaka, N., Torres-Alanis, O., Voss, J-U., Wakuri, Sh., Walum, E.,Wang, X., Zucco, F., Ekwall, B. (1996). MEIC evaluation of acute systemic toxicity. PartII. In vitro results from 68 toxicity assays used to test the first 30 reference chemicalsand a comparative cytotoxicity analysis. Alternatives to Laboratory Animals (ATLA), 24,273–311.

69. Clemedson, C., Ekwall, B. (1999). Overview of the MEIC results: I. The in vitro - invivo evaluation. Toxicology In Vitro, 13, 657–663.

70. Ekwall, B., Bondesson, I., Castell, J.V., Gomez-Lechon, M.J., Hellberg, S., Hogberg,J., Jover, R., Pondosa, X., Romert, L., Stenberg, K., Walum, E. (1989). Cytotoxicity


evaluation of the first ten MEIC chemicals: acute lethal toxicity in man predicted bycytotoxicity in five cellular assays and by oral LD50 tests in rodents. Alternatives toLaboratory Animals (ATLA), 17, 83–100.

71. Ekwall, B. (1999). Overview of the final MEIC results: II. The in vitro—in vivo evalu-ation, including the selection of a practical battery of cell tests for prediction of acutelethal blood concentrations in humans. Toxicology In Vitro, 13, 665–673.

72. Clemedson, C., Blaauboer, B., Castell, J., Prieto, P., Ristyelli, L., Vericat, J.A., Wen-del, A. (2006). ACuteTox—Optimation and pre-validation of an in vitro test strategyfor predicting human acute toxicity. Alternatives to Animal Experimentation (ALTEX),23(suppl 1), 254–258.

73. Clothier, R., Dierickx, P., Lakhanisky, T., Fabre, M., Betanzos, M., Curren, R., Sjostrom,M., Raabe, H., Bourne, N., Hernandez, V., Mainez, J., Owen, M., Watts, S., Anthonissen,R. (2008). A database of IC50 values and principal component analysis of results fromsix basal cytotoxicity assays, for use in the modelling of the in vivo and in vitro dataof the EU ACuteTox project. Alternatives to Laboratory Animals (ATLA), 36, 503–519.


75. Slater, K. (2001). Cytotoxicity tests for high-throughput drug discovery.Current Opinionin Biotechnology, 12, 70–74.

76. Xia, M., Huang, R., Witt, K.L., Southall, N., Fostel, J., Cho, M.-H., Jadhav, A., Smith,C.S., Inglese, J., Portier, C.J., Tice, R.R., Austin, C.P. (2008). Compound cytotoxicityprofiling using quantitative high-throughput screening. Environmental Health Perspec-tives, 116, 284–291.

77. Andrews, M.J., Garle, M.J., Clothier, R.H. (1997). Reduction of the new tetrazolium dye,Alamar BlueTM, in cultured rat hepatocytes and liver fractions.Alternatives to LaboratoryAnimals (ATLA), 25, 641–653.

78. Nakayama, G.R., Caton, M.C., Nova, M.P., Parandoosh, Z. (1997). Assessment of theAlamar Blue assay for cellular growth and viability in vitro. Journal of ImmunologicalMethods, 240, 205–208.

79. Slaughter,M.R., Bugelski, P.J., O’Brien, P.J. (1999). Evaluation ofAlamarBlue reductionfor in vitro assay of hepatocyte toxicity. Toxicology In Vitro, 13, 567–569.

80. Hamid, R., Rotshteyn, Y., Rabadi, L., Parikh, R., Bullock, P. (2004). Comparison ofAlamar Blue and MTT assays for high throughput screening. Toxicology In Vitro, 18,703–710.

81. Gieni, R.S., Li, Y., HayGlass, K.T. (1995). Comparison of 3H-thymidine incorporationwith MTT- and MTS-based bioassays for human and murine IL-2 and IL-4 analysis.Tetrazolium assays provide markedly enhanced sensitivity. Journal of ImmunologicalMethods, 187, 85–93.

82. Wagner, U., Burkhardt, E., Failing, K. (1999). Evaluation of canine lymphocyteproliferation: comparison of three different colorometric methods with the 3H-thymidine incorporation assay. Veterinary Immunology and Immunopathology, 70, 151–159.

83. Berridge, M.V., Tan, A.S., McCoy, K.D., Wang, R. (1996). The biochemical and cellularbasis of cell proliferation assays that use tetrazolium salts. Biochemica, 4, 14–19.

REFERENCES 65

84. Chan, J.H., Harminder, S.D., Powell-Richards, A., Jones, D.R.E., Harris, I.M. (2001).Effect of ABO blood group mismatching on corneal epithelial cells: an in vitro study.British Journal of Ophthalmology, 85, 1104–1109.

85. Hynes, J., Hill, R., Papkovsky, D.B. (2006). The use of a fluorescence-based oxygenuptake assays in the analysis of cytotoxicity. Toxicology In Vitro, 20, 785–792.

86. Hynes, J., Marroquin, L.D., Ogurtsov, V.I., Christiansen, K.N., Stevens, G.J., Pap-kovsky, D.B., Will, Y. (2006). Investigation of drug-induced mitochondrial toxicityusing fluorescence-based oxygen-sensitive probes. Toxicological Sciences, 92, 186–200.

87. Marroquin, L.D., Hynes, J., Dykens, J.A., Jameison, J.D.,Will, Y. (2007). Circumventingthe crabtree effect: replacing media glucose with galactose increases susceptibility ofHepG2 cells to mitochondrial toxicants. Toxicological Sciences, 97, 539–547.

88. Schoonen, W.G., Stevenson, J.C., Westerink, W.M., Horbach, G.J. (2011). Cytotoxiceffects of 110 reference compounds on rat H4IIE and human HepG2 hepatocytes. IIIMechanistic assays on oxygen consumption with MitoXpress and NAD(P)H productionwith Alamar BlueTM. Manuscript in preparation.

89. Germain,M.A., Hatton, A.,Williams, S.,Matthews, J.B., Stone,M.H., Fisher, J., Ingham,E. (2003). Comparison of the cytotoxicity of clinically relevant cobalt-chromium andalumina ceramic wear particles in vitro. Biomaterials, 24, 469–479.

90. Hannah, R., Beck, M., Moravec, R., Riss, T. (2001). CellTiter-GloTM luminescent cellviability assay: a sensitive and rapid method for determining cell viability. Promega CellNotes, 2, 11–13.

91. Lydon, M.J., Keeler, K.D., Thomas, D.B. (1980). Vital DNA staining and cell sorting byflow microfluorometry. Journal of Cellular Physiology, 102, 175–181.

92. Richards, W.L., Song, M.-K., Krutsch, H., Everts, R.P., Marsden, E., Thorgeirsson, S.S.(1985). Measurement of cell proliferation using Hoechst 33342 for the rapid semiauto-mated microfluorometric determination of chromatin DNA. Experimental Cell Research,159, 235–246.

93. Blaheta, R.A., Franz, M., Auth, M.K.H., Wenisch, H.J.C., Markus, B.H. (1991). Arapid non-radioactive fluorescence assay for the measurement of both cell number andproliferation. Journal of Immunological Methods, 142, 199–206.

94. Petit, J.M., Denis-Gay, M., Ratinaud, M.H. (1993). Assessment of fluorochromes forcellular structure and function studies by flow cytometry. Biology of the Cell, 78, 1–13.

95. Lakowicz, J.R., Gryczynski, I., Malak, H., Schrader, M., Engelhardt, P., Kano, H., Hell,S.W. (1997). Time-resolved fluorescence spectroscopy and imaging of DNA labeledwith DAPI and Hoechst 33342 using three photon excitation. Biophysical Journal, 72,567–578.

96. Nakagawa, T., Sawada, M., Gonzalez, F.J., Yokoi, T., Kamataki, T. (1996). Stableexpression of human CYP2E1 in Chinese hamster cells: high sensitivity to N,N-dimethylnitrosamine in cytotoxicity testing.Mutation Research, 360, 181–186.

97. de Fries, R., Mitsuhashi, M. (1995). Quantification of mitogen induced human lympho-cyte proliferation: comparison of Alamar BlueTM assay to 3H-thymidine incorporationassay. Journal of Clinical Laboratory Analysis, 9, 89–95.

98. Rahman, I., Kode, A., Biswas, K.B. (2006). Assay for the quantitative determination ofglutathione and glutathione disulfide levels using enzymatic recycling methods. NatureProtocols, 1, 3159–3165.


99. Fernandez-Checa, J.C., Kaplowitz, N. (1990). The use of monochlorobimane to deter-mine hepatic GSH levels and synthesis. Analytical Biochemistry, 190, 212–219.

100. Stevenson, D., Wokosin, D., Girkin, J., Grant, M.H. (2002). Measurement of the intracel-lular distribution of reduced glutathione in cultured rat hepatocytes using monochloro-bimane and confocal laser scanning microscopy. Toxicology In Vitro, 16, 609–619.

101. Young, P.R., ConnorsWhite, A.L., Dzido, G.A. (1994). Kinetic analysis of the intra-cellular conjugation of monochlorbimane by IC-21 murine macrophage glutathione-S-transferase. Biochimica et Biophysica Acta, 1201, 461–465.

102. Yasgar, A., Shultz, J., Zhou, W., Wang, H., Huang, F., Murphy, N., Abel, E.L., DiGio-vanni, J., Inglese, J., Simeonov, A. (2010). A high throughput 1536-well luminescenceassay for glutathione S-transferase activity. Assay and Drug Development Technologies,8, 200–211.

103. Gabriel, C., Camins, A., Sureda, F.X., Aquirre, L., Escubedo, E., Pallas, M., Camarasa,J. (1997). Determination of nitric oxide generation in mammalian neurons using dichlo-rofluorescin diacetate and flow cytometry. Journal of Pharmacological and ToxicologicalMethods, 38, 93–98.

104. Sanner, B.M., Meder, U., Zidek, W., Tepel, M. (2002). Effects of glucocorticoids ongeneration of reactive oxygen species in platelets. Steroids, 67, 715–719.

105. Soliman, M.K., Mazzio, E., Soliman, K.F.A. (2002). Levodopa modulating effects ofinducible nitric oxide synthase and reactive oxygen species in glioma cells. Life Sciences,72, 185–198.

106. Fredriksson, L., Herpers, B., Benedetti, G., Matadin, Q., Puigvert, J.C., de Bont, H.,Dragovic, S., Vermeulen, N.P.E., Commandeur, J.N.M., Danen, E., de Graauw, M.,van de Water, B. (2011). Diclofenac inhibits tumor necrosis factor �-induced nuclearfactor-�B activation causing synergistic hepatocyte apoptosis. Hepatology, 53, 2027–2041.

107. O’Brien, M.A., Daily, W.J., Hesselberth, E., Moravec, R.A., Scurria, M.A., Klaubert,D.H., Bulleit, R.F., Wood, K.V. (2005). Homogeneous, bioluminescent protease assays:caspase 3 as a model. Journal of Biomolecular Screening, 10, 137–148.

108. Liu, D., Li, C., Chen, Y., Burnett, C., Liu, X.Y., Downs, S., Colllins, R.D., Hawegir, J.(2004). Nuclear transport of proinflammatory transcription factors is required formassiveliver apoptosis induced by bacterial lipopolysaccharide. Journal of Biological Chemistry,279, 48434–48442.

109. Schoonen, W.G., de Roos, J.A., Westerink, W.M., Debiton, E. (2005). Cytotoxic effectsof 110 reference compounds on HepG2 cells and for 60 compounds on HeLa, ECC-1and CHO cells. II. Mechanistic assays on NAD(P)H, ATP and DNA contents. ToxicologyIn Vitro, 19, 491–503.

110. Schoonen, W.G., Westerink, W.M., de Roos, J.A., Debiton, E. (2005). Cytotoxic effectsof 100 reference compounds on Hep G2 and HeLa cells and of 60 compounds on ECC-1and CHO cells. I. Mechanistic assays on ROS, glutathione depletion and calcein uptake.Toxicology In Vitro, 19, 505–516.

111. O’Brien, P.J., Irwin, W., Diaz, D., Howard-Cofield, E., Krejsa, C.M., Slaughter, M.R.,Gao, B., Kaludercic, N., Angeline, A., Bernardi, P., Brain, P., Hougham, C. (2006). Highconcordance of drug-induced human hepatotoxicity with in vitro cytotoxicity measuredin a novel cell-based model using high content screening. Archives of Toxicology, 80,580–604.

REFERENCES 67

112. Xu, J.J. (2007). High content screeningwith primary rat hepatocytes. Lecture Symposiumon high content screening for toxicity (Informa), 28 June 2007, Le Meridian, Vienna,Austria.

113. Xu, J.J., Henstock, P.V., Dunn, M.C., Smith, A.R., Chabot, J.R., de Graaf, D. (2008).Cellular imaging predictions of clinical drug-induced liver injury.Toxicological Sciences,105, 97–105.

114. Lieggi, N.T., Edvardsson, A., O’Brien, P.J. (2010). Translation of novel anti-cancercytotoxicity biomarkers with high content analysis from an in vitro predictive model toan in vivo cell model. Toxicology In Vitro, 24, 2063–2071.

115. Vanparys, P. (2007). New models for toxicity testing. Lecture on 2nd Conference onpredictive human toxicity and ADME toxicity studies, 25–26 January 2007, Brussels,Belgium.

116. Abraham, V.C., Towne, D.L., Waring, J.F., Warrior, U., Burns, D.J. (2008). Applicationof a high-content multiparameter cytotoxicity assay to prioritize compounds based ontoxicity potential in humans. Journal of Biomolecular Screening, 13, 527–537.

117. Nioi, P., Perry, B.K., Wang, E.-J., Gu, Y.-Z., Snyder, R.D. (2007). In vitro detectionof drug-induced phospholipidosis using gene expression and fluorescent phospholipid-based methodologies. Toxicological Sciences, 99, 162–173.

118. Morelli, J.K., Buehrle, M., Pognan, F., Barone, L.R., Fieles, W., Ciacco, P.J. (2006). Val-idation of an in vitro screen for phospholipidosis using a high-content biology platform.Cell Biology and Toxicology, 22, 15–27.

119. Mesens, N., Steemans, M., Hansen, E., Peters, A., Verheyen, G., Vanparys, P. (2009). A96-well flow cytometric screening assay for detecting in vitro phospholipidosis-inductionin the drug discovery phase. Toxicology In Vitro, 23, 217–226.

120. Mesens, N., Steemans, M., Hansen, E., Verheyen, G., van Goethem, F., van Gompel, J.(2010). Screening for phospholipidosis induced by central nervous drugs: comparing thepredictivity of an in vitro assay to high throughput in silico assays. Toxicology In Vitro,24, 1417–1425.

121. Grandl, M., Schmitz, G. (2010). Fluorescent high content imaging allows the discrimi-nation and quantitation of E-LDL-induced lipid droplets and Ox-LDL-generated phos-pholipidosis in human macrophages. Cytometry, 77A, 231–242.

122. van de Water, F., Havinga, J., Ravesloot, W.T., Horbach, G.J., Schoonen, W.G. (2011).High content screening analysis of phospholipidosis: validation of a 96-well assay withCHO-K1 andHepG2 cells for the prediction of in vivo based phospholipidosis.ToxicologyIn Vitro, 25, 1870–1882.

123. Gomez-Lechon, M.J., Donato, M.T., Martınez-Romero, A., Jimenez, N., Castell, J.V.,O’Connor, J.-E. (2007). A human hepatocellular in vitro model to investigate steatosis.Chemico-Biological Interactions, 165, 106–116.

124. Granneman, J.G., Moore, H.H., Mottillo, E.P., Zhu, Z. (2009). Functional interactionsbetween Mldp (LSDP5) and Abhd5 in the control of intracellular lipid accumulation.Journal of Biological Chemistry, 284, 3049–3057.

125. Donato, M.T., Martınez-Romero, A., Jimenez, N., Negro, A., Herrera, G., Castell, J.V.,O’Connor, J.-E., Gomez-Lechon, M.J. (2009). Cytometric analysis for drug-inducedsteatosis in HepG2 cells. Chemico-Biological Interactions, 181, 417–423.

126. Maglich, J.M., Stoltz, C.M., Goodwin, B., Hawkins-Brown, D., Moore, J.T., Kliewer,S.A. (2002). Nuclear pregnane x receptor and constitutive androstane receptor regulate


overlapping but distinct sets of genes involved in xenobiotic detoxification. MolecularPharmacology, 62, 638–646.

127. Mankowski, D.C., Ekins, S. (2003). Prediction of human drug metabolizing enzymeinduction. Current Drug Metabolism, 4, 381–391.

128. Moore, J.T., Moore, L.B., Maglich, J.M., Kliewer, S.A. (2003). Functional and structuralcomparison of PXR and CAR. Biochimica et Biophysica Acta, 1619, 235–238.

129. Xie, W., Yeuh, M.F., Radominska-Pandya, A., Saini, S.P., Negishi, Y., Bottroff, B.S.,Cabrera, G.Y., Tukey, R.H., Evans, R.M. (2003). Control of steroid, heme, and carcino-gen metabolism by nuclear pregnane X receptor and constitutive androstane receptor.Proceedings of the National Academy of Sciences USA, 100, 4150–4155.

130. Waxman, D.J. (1999). P450 gene induction by structurally diverse xenochemicals: centralrole of nuclear receptorsCAR, PXR, andPPAR.Archives of Biochemistry and Biophysics,369, 11–23.

131. Wei, P., Zhang, J., Dowhan, D.H., Han, Y., Moore, D.D. (2002). Specific and overlappingfunctions of the nuclear hormone receptors CAR and PXR in xenobiotic response.Pharmacogenomics Journal, 2, 117–126.

132. Westerink, W.M.A., Schoonen, W.G.E.J. (2007). Phase I enzyme levels in cryopreservedprimary human hepatocytes and HepG2 cells and their induction in HepG2 cells. Toxi-cology In Vitro, 21, 1581–1591.

133. Westerink, W.M.A., Schoonen, W.G.E.J. (2007). Phase II enzyme levels in HepG2 cellsand cryopreserved human primary human hepatocytes and their induction in HepG2cells. Toxicology In Vitro, 21, 1592–1602.

134. Kienhuis, A.S., Bessems, J.G.M., Pennings, L.R.A., Driesen, M., Luijten, M., van Delft,J.H.M., Peijnenburg, A.A.C.M., van der Ven, L.T.M. (2011). Application of toxicoge-nomics in hepatic systems toxicology for risk assessment: acetaminophen as a case study.Toxicology and Applied Pharmacology, 250, 96–107.

135. Guo, G.L., Moffit, J.S., Nicol, C.J., Ward, J.M., Aleksunes, L.A., Slitt, A.L., Kliewer,S.A., Manautou, J.E., Gonzalez, F.J. (2004). Enhanced acetaminophen toxicity by acti-vation of the pregnane X receptor. Toxicological Sciences, 82, 374–380.

136. James, L.P., McCullough, S.S., Lamps, L.W., Hinson, J.A. (2003). Effect of N-acetylcysteine on acetaminophen toxicity in mice: relationship to reactive nitrogen andcytokine formation. Toxicological Sciences, 75, 458–467.

137. Ando, H., Horibata, Y., Yamashita, S., Oyama, T., Sugimoto, H. (2010). Low-densitylipoprotein and oxysterols suppress the transcription of CTP:Phosphoethanolaminecytidylyltransferase in vitro. Biochimica et Biophysica Acta, 1801, 487–495.

138. Chen, C., Tan, R., Wong, L., Fekete, R., Halsey, J. (2011). Quantification of microR-NAs by real-time RT-qPCR. In PCR Protocols (Park D.J., Ed.). Methods in MolecularBiology, 687, 113–134. Humana Press, Springer Sciences + Business Media, LLC:New York

139. Hines, R.N., Koukouritaki, S.B., Poch, M.T., Stephens, M.C. (2008). Regulatory poly-morphisms and their contribution to interindividual differences in the expression ofenzymes influencing drug and toxicant disposition. Drug Metabolism Reviews, 40, 263–301.

140. Crespi, C.L., Miller, V.P. (1999). The use of heterologously expressed drug metabolizingenzymes - state of the art and prospects for the future. Pharmacology & Therapeutics,84, 121–131.

REFERENCES 69

141. Crespi, C.L., Miller, V.P., Penman, B.W. (1997). Microtiter plate assays for inhibition ofhuman, drug-metabolizing cytochromes P450. Analytical Biochemistry, 248, 188–190.

142. Miller, B., Potter-Locher, F., Seelbach, A., Stopper, H., Utesch, D., Madle, S. (1998).Evaluation of the in vitro micronuclei test as an alternative to the in vitro chromosomalaberration assay: position of the GUM working group on the in vitro micronucleus test.Mutation Research, 410, 81–116.

143. Stresser, D.M., Turner, S.D., Blanchard, A.P., Miller, V.P., Crespi, C.L. (2002).Cytochrome P450 fluorometric substrates: identification of isoform-selective probes forrat CYP2D2 and human CYP3A4. Drug Metabolism & Disposition, 30, 845–852.

144. Schoonen, W.G.E.J., Westerink, W.M.A., Horbach, G.J. (2009). High throughput screen-ing for the analysis of in vitro toxicity. In: Molecular, Clinical and EnvironmentalToxicology (Andreas Luch, Ed.). Birkhauser Publishing: Basel, Switzerland; pp. 401–439.

145. Aono, S., Adachi, Y., Uyama, E., Yamada, Y., Keino, H., Nanno, T., Koiwai, O., Sato,H. (1995). Analysis of genes for bilirubin UDP-glucuronosyltransferase in Gilbert’ssyndrome. Lancet, 345, 958–959.

146. Seppen, J., Bosma, P.J., Goldhoorn, B.G., Bakker, C.T.M., Chowdhury, J.R., Chowd-hury, N.R., Jansen, P.L.M., Oude Elferink, R.P.J. (1994). Discrimination betweenCrigler-Najjar type I and II by expression of mutant bilirubin uridine phosphate-glucuronosyltransferase. Journal of Clinical Investigation, 94, 2385–2391.

147. Lankisch, T.O., Behrens, G., Ehmer, U., Mobius, U., Rockstroh, J., Wehmeier, M.,Kalthoff, S., Freiburg, N., Manns, M.P., Schmidt, R.E., Strassburg, C.P. (2009). Gilbert’ssyndrome and hyperbilirubinemia in protease inhibitor therapy - an extended haplotypeof genetic variants increases risk in indinacir treatment. Journal of Hepatology, 50,1010–1018.

148. Strassburg, C.P. (2008). Pharmacogenetics of Gilbert’s syndrome. Pharmacogenomics,9, 703–715.

149. Ekhart, C., Rodenhuis, S., Smits, P.H.M., Beijnen, J.H., Huitema, A.D.R. (2009). Anoverview of the relations between polymorphisms in drug metabolizing enzymes anddrug transporters and survival after cancer drug treatment. Cancer Treatment Review, 35,18–31.

150. Ouzzine, M., Barre, L., Netter, P., Magdalou, J., Fournel-Gigleux, S. (2003). ThehumanUDP-glucuronosyltransferases: structural aspects and drug glucuronidation.DrugMetabolism Reviews, 35, 287–303.

151. Robert, J., Le Morvan, V., Smith, D., Pourquier, P., Bonnet, J. (2005). Predictingdrug response and toxicity based on gene polymorphisms. Critical Reviews in Oncol-ogy/Hematology, 54, 171–196.

152. Wells, P.G., MacKenzie, P.I., Chowdhury, J.R., Guillemette, C., Gregory, P.A., Ishii,Y., Hansen, A.J., Kessler, F.K., Kim, P.M., Chowdhury, N.R., Ritter, J.K. (2004).Glucuronidation and the UDP-glucuronosyltransferases in health and disease. DrugMetabolism & Disposition, 32, 281–290.

153. Broudy, M.I., Crespi, C.L., Patten, C.J. (2001). A sensitive fluorometric high throughputinhibition assay for human UDP glucuronosyl transferase (UGT) 1A1. Poster presenta-tion: BD Biosciences, property of Bacton, Dickinson and Company.

154. Collier, A.C., Tingle, M.D., Keelan, J.A., Paxton, J.W., Mitchell, M.D. (2000). A highlysensitive fluorescent microplate method for the determination of UDP-glucuronosyl


transferase activity in tissues and placental cell lines. Drug Metabolism & Disposition,28, 1184–1186.

155. Larson, B., Kelts, J.L., Banks, P., Cali, J.J. (2011). Automation and miniaturization of thebioluminescent UGT-Glo assay for screening of UDP-glucuronosyltransferase inhibitionby various compounds. Journal of Laboratory Automation, 16, 38–46.

156. Trubetskoy, O.V., Finel, M., Kurkela, M., Fitzgerald, M., Peters, N.R., Hoff-man, F.M., Trubetskoy, V.S. (2007). High throughput screening assay for UDP-glucuronosyltransferase 1A1 glucuronidation profiling. Assay and Drug DevelopmentTechnologies, 5, 343–354.

157. Uchaipichat, V., Mackenzie, P.I., Guo, X.-H., Gardner-Stephen, D., Galetin, A., Houston,J.B., Miners, J.O. (2004). HumanUDP-glucuronosyltransferases: isoform selectivity andkinetics of 4-methylumbelliferone and 1-naphthol glucuronidation, effects of organicsolvents, and inhibition by diclofenac and probenecid. Drug Metabolism & Disposition,32, 413–423.

158. Kotoh, S., Naito, S., Yokomizo, A., Kohno, K., Kuwano, M., Kumazawa, J. (1997).Enhanced expression of � -glutamylcysteine synthetase and glutathione S-transferasegenes in cisplatin-resistant bladder cancer cells with multidrug resistance phenotype.Journal of Urology, 157, 1054–1058.

159. Coughtrie, M.W., Gilissen, R.A.H.J., Shek, B., Strange, R.C., Fryer, A.A., Jones, P.W.,Bamber, D.E. (1999). Phenol sulphotransferase SULT1A1 polymorphism: moleculardiagnosis and allel frequencies in Caucasian and African populations. Biochemical Jour-nal, 337, 45–49.

160. Lindsay, J., Wang, L., Li, Y., Zhou, S. (2008). Structure, function and polymorphism ofhuman cytosolic sulfotransferases. Current Drug Metabolism, 9, 99–105.

161. Wilson, J.D., George, F.W., Griffin, J.E. (1981). The hormonal control of sexual devel-opment. Science, 211, 1278–1284.

162. Bentvelsen, F.M., Brinkman, A.O., van der Schoot, P., van der Linden, J.E.T.M., vander Kwast, T.H., Boersma, W.J.A., Schroder, F.H., Nijman, J.M. (1995). Developmentalpattern and regulation by androgens of androgen receptor expression in the urogenitaltract of the rat. Molecular and Cellular Endocrinology, 113, 245–253.

163. Simpson, E.R., Clyne, C., Rubin, G., Boon, W.C., Robertson, K., Britt, K., Speed, C.,Jones, M. (2002). Aromatase—a brief overview. Annual Review of Physiology, 64, 93–127.

164. Donaldson, M.D.C., Thomas, P.H., Love, J.G., Murray, G.D., McNinch, A.W., Savage,D.C.L. (1994). Presentation, acute illness, and learning difficulties in salt wasting 21-hydroxylase deficiency. Archives of Disease in Childhood, 70, 214–218.

165. Berger, S., Bleich, M., Schmid, W., Cole, T.J., Peters, J., Watanabe, H., Kriz, W., Warth,R., Greger, R., Schutz, G. (1998). Mineralocorticoid receptor knockout mice: pathophys-iology of Na+ metabolism. Proceedings of the National Academy of Sciences USA, 95,9424–9429.

166. Kleinman, A., Tuckermann, J.P. (2007). Glucocorticoid receptor action in beneficial andside effects of steroid therapy: lessons from conditional knockout mice. Molecular andCellular Endocrinology, 275, 98–108.

167. Sonneveld, E., Riteco, J.A.C., Jansen, H.J., Pieterse, B., Brouwer, A., Schoonen, W.G.,van der Burg, B. (2006). Comparison of in vitro and in vivo screening models forandrogenic and estrogenic activities. Toxicological Sciences, 89, 173–187.

REFERENCES 71

168. Sonneveld, E., Pieterse, B., Schoonen, W.G., van der Burg, B. (2011). Validation of invitro screening models for progestagenic activities: Inter-assay comparison and correla-tion with in vivo activity in rabbits. Toxicology In Vitro, 25, 545–554.

169. Bovee, T.F.H., Helsdingen, R.J.R., Rietjens, I.M.C.M., Keijer, J., Hoogenboom, R.L.A.P.(2004). Rapid yeast estrogen bioassays stably expressing human estrogen receptors �and �, and green fluorescent protein: a comparison of different compounds with bothreceptor types. Journal of Steroid Biochemistry & Molecular Biology, 91, 99–109.

170. Bovee, T.F.H., Helsdingen, R.J.R., Hamers, A.R.M., Van Duursen, M.B.M., Nielen,M.W.F., Hoogenboom, R.L.A.P. (2007). A new highly specific and robust yeast andro-gen bioassay for the detection of agonists and antagonists. Analytical & BioanalyticalChemistry, 389, 1549–1558.

171. Balaguer, P., Boussioux, A.-M., Demirpence, E., Nicolas, J.-C. (2001). Reporter celllines are useful tools for monitoring biological activity of nuclear receptor ligands.Luminescence, 16, 153–158.

172. Legler, J., van den Brink, C.E., Brouwer, A., Murk, A.J., van der Saag, P.T., Vethaak,A.D., van der Burg, B. (1999). Development of a stably transfected estrogen receptor-mediated luciferase reporter gene assay in the human T47D breast cancer cell line.Toxicological Sciences, 48, 55–66.

173. Blankvoort, B.M.G., de Groene, E.M., van Meeteren-Kreikamp, A.P., Witkamp, R.F.,Rodenburg, R.J.T., Aarts, J.M.M.J.G. (2001). Development of an androgen reporter geneassay (AR-LUX) utilizing a human cell line with an endogenously regulated androgenreceptor. Analytical Biochemistry, 298, 93–102.

174. Poulin, R., Baker, D., Labrie, F. (1988). Androgens inhibit basal and estrogen-inducedcell proliferation in the ZR-75–1 human breast tumor cell line. Breast Cancer Researchand Treatment, 12, 213–225.

175. Kloosterboer, H.J., Schoonen, W.G.E.J., Verheul, H.A.M. (2008). Proliferation of breastcells by steroid hormones and their metabolites. In: Breast Cancer Prognosis, Treatmentand Prevention (Pasqualini, J.R., Ed.). Academic Press: New York; pp. 343–386.

176. Wilson, V.S., Bobseine, K., Gray, L.E. Jr. (2004). Development and characterization of acell line that stably expresses an estrogen-responsive luciferase reporter for the detectionof estrogen receptor agonist and antagonists. Toxicological Sciences, 81, 69–77.

177. Thalmann, G.N., Sikes, R.A., Wu, T.T., Degeorges, A., Chang, S., Ozen, M., Pathak,S., Chung, L. (2000). LNCaP progression model of human prostate cancer; androgen-independence and osseous metastasis. The Prostate, 44, 91–103.

178. Heisler, L.E., Evangelou, A., Lew, A.M., Trachtenberg, J., Elsholtz, H.P., Brown, T.J.(1997). Androgen-dependent cell cycle arrest and apoptotic death in PC-3 prostatic cellcultures expressing a full-length human androgen recceptor. Molecular and CellularEndocrinology, 126, 59–73.

179. Oosterom, J., van Doornmalen, E.J.P., Lobregt, S., Blomenrohr, M., Zaman, G.J.R.(2005). High-throughput screening using�-lactamase reporter-gene technology for iden-tification of low-molecular-weight antagonists of the human gonadotropin releasing hor-mone receptor. Assay and Drug Development Technologies, 3, 143–154.

180. Hakola, K., Boogaart, P.V.D., Mulders, J., de Leeuw, R., Schoonen, W., Van Heyst,J., Swolfs, A., Van Casteren, J., Huhtaniemi, I., Kloosterboer, H. (1997). Recombinantfollicle-stimulating hormone; production by Chinese hamster ovary cells, purificationand functional characterization. Molecular and Cellular Endocrinology, 127, 59–69.


181. Hakola, K., Boogaart, P.V.D., Mulders, J., de Leeuw, R., Schoonen, W., Van Heyst,J., Swolfs, A., Van Casteren, J., Huhtaniemi, I., Kloosterboer, H. (1997). Recombinantluteinizing hormone; production by Chinese hamster ovary cells, purification and func-tional characterization. Molecular and Cellular Endocrinology, 128, 47–56.

182. Fingerhut,M.A., Halperin,W.E.,Marlow,D.A., Piacitelli, L.A., Honchar, P.A., Sweeney,M.H., Greife, A.L., Dill, P.A., Steenland, K., Suruda, A.J. (1991). Cancer mortalityin workers exposed to 2,3,7,8-tetrachlorodibenzo-p-dioxin. New England Journal ofMedicine, 324, 212–218.

183. Manz, A., Berger, J., Dwyer, J.H., Flesch-Janys, D., Nagel, S., Waltsgott, H. (1991).Cancer mortality among workers in chemical plant contaminated with dioxin. Lancet,338, 959–964.

184. Zober, A., Messerer, P., Huber, P. (1990). Thirty-four-year mortality follow-up of BASFemployees exposed to 2,3,7,8-TCDD after the 1953 accident. International Archives ofOccupational and Environmental Health, 62, 139–157.

185. Van den Heuvel, P.J.V. (1999). Peroxisome proliferator-activated receptors (PPARs) andcarcinogenesis. Toxicological Sciences, 47, 1–8.

186. Holsapple, M.P., Pitot, H.C., Cohen, S.H., Boobis, A.R., Klaunig, J.E., Pastoor, T.,Dellarco, V.L., Dragan, Y.P. (2006). Mode of action in relevance of rodent liver tumorsto human cancer risk. Toxicological Sciences, 89, 51–56.

187. Phillips, J.M., Goodman, J.I. (2009). Multiple genes exhibit phenobarbital-induced con-stitutive active/androstane receptor-mediated DNA methylation changes during livertumorigenesis and in liver tumors. Toxicological Sciences, 108, 273–289.

188. Lin, J.H. (2008). Applications and limitations of genetically modified mouse models indrug discovery and development. Current Drug Metabolism, 9, 419–438.

189. Morimura, K., Cheung, C., Ward, J.M., Reddy, J.K., Gonzalez, F.J. (2006). Differentialsusceptibility of mice humanized for peroxisome proliferator-activated receptor � toWY-14,643-induced liver tumorigenesis. Carcinogenesis, 27, 1074–1080.

190. Westerink, W.M.A., Stevenson, J.C.R., Schoonen, W.G.E.J. (2008). Pharmacologic pro-filing of human and rat cytochrome P450 1A1 and 1A2 induction and competition.Archives of Toxicology, 82, 909–921.

191. Stewart, J.R., Artime, M.C., O’Brian, C.A. (2003). Resveratrol: a candidate nutritionalsubstance for prostate cancer prevention. Journal of Nutrition, 133, 2440S–2443S.

192. Galati, G., O’Brien, P.J. (2004). Potential toxicity of flavonoids and other dietary phe-nolics: significance for their chemopreventive and anticancer properties. Free RadicalBiology & Medicine, 37, 287–303.

193. Moon, Y.J., Wang, X., Morris, M.E. (2006). Dietary flavonoids: effects on xenobioticand carcinogen metabolism. Toxicology In Vitro, 20, 187–210.

194. Chen, I., McDougal, A., Wang, F., Safe, S. (1998). Aryl hydrocarbon receptor-mediatedantiestrogenic and antitumorigenic activity of diindolylmethane. Carcinogenesis, 19,1631–1639.

195. Kim, Y.S., Milner, J.A. (2005). Targets for indole-3-carbinol in cancer prevention. Jour-nal of Nutritional Biochemistry, 16, 65–73.

196. Fritsche, E., Schafer, C., Calles, C., Bernsmann, T., Bernshausen, T., Wurm, M., Huben-thal, U., Cline, J.E., Hajimiragha, H., Schroeder, P., Klotz, L., Rannug, A., Furst, P.,Hanenberg, H., Abel, J., Krutmann, J. (2007). Lightening up the UV response by

REFERENCES 73

identification of the arylhydrocarbon receptor as a cytoplasmic target for ultravioletB radiation. Proceedings of the National Academy of Sciences USA, 104, 8851–8856.

197. Wu, J.J., Huang, D.B., Pang, K.R., Hsu, S., Tyring, S.K. (2005). Thalidomide: dermato-logical indications, mechanisms of action and side-effects. British Journal of Dermatol-ogy, 153, 254–273.

198. Genschow, E., Spielmann, H., Scholz, G., Seiler, A., Brown, N., Piersma, A., Brady, M.,Clemann, N., Huuskonen, H., Paillard, F., Bremer, S., Becker, K. (2002). The ECVAMinternational validation study on in vitro embryotoxicity tests: results of the definitivephase and evaluation of prediction models. Alternatives to Laboratory Animals (ATLA),30, 151–176.

199. Piersma, A.H. (2004). Validation of alternative methods for developmental toxicity test-ing. Toxicological Letters, 149, 147–153.

200. Piersma, A.H. (2006). Alternative methods for developmental toxicity testing. Basic &Clinical Pharmacological & Toxicology, 98, 427–431.

201. Piersma, A.H., Janer, G., Wolterink, G., Bessems, J.G.M., Hakkert, B.C., Slob, W.(2008). Quantitative extrapolation of in vitro whole embryo culture embryotoxocity datato developmental toxicity in vivo using the benchmark dose approach. ToxicologicalSciences, 101, 91–100.

202. Jansen, M.S., Nagel, S.C., Miranda, P.J., Lobenhofer, E.K., Afshari, C.A., McDonnell,D.P. (2004). Short-chain fatty acids enhance nuclear receptor activity through mitogenactivated protein kinase activation and histone deacetylase inhibition. Proceedings of theNational Academy of Sciences USA, 101, 7199–7204.

203. Tabb, M.M., Blumberg, B. (2006). New modes of action of endocrine-disrupting chem-icals. Molecular Endocrinology, 20, 475–482.

204. Blom, H.J., Shaw, G.M., den Heyer, M., Finnell, R.H. (2006). Neural tube defects andfolate: case far from closed. Nature Reviews Neuroscience, 7, 724–731.

205. Cabrera, R.M., Hill, D.S., Etheredge, A.J., Finnell, R.H. (2004). Investigations into theetiology of neural tube defects. Birth Defects Research, 72, 330–344.

206. Kaufman, M.H. (1997). The teratogenic effects of alcohol following exposure duringpregnancy, and its influence on the chromosome constitution of the pre-ovulatory egg.Alcohol & Alcoholism, 32, 113–128.

207. Khalighi, M., Brezezinski, M.R., Chen, H., Juchau, M.R. (1999). Inhibition of humanprenatal biosynthesis of all-trans-retionoic acid by ethanol, ethanol metabolites andproducts of lipid peroxidation reactions. A possible role for CYP2E1. BiochemicalPharmacology, 57, 811–821.

208. Nelson, B.K., Setzer, J.V., Brightwell,W.S.,Mathinos, P.R., Kuczuk,M.H.,Weaver, T.E.,Goad, P.T. (1984). Comparative inhalation teratogenicity of four glycol ether solventsand an amino acid derivative in rats. Environmental Health Perspectives, 57, 261–271.

209. Mebus, C.A., Welsch, F. (1989). The possible role of one carbon moieties in 2-methoxyethanol and 2-methoxyacetic acid-induced developmental toxicity. ToxicologyApplied Pharmacology, 99, 98–109.

210. Campbell, L.R., Dayton,D.H., Sohal, G.S. (1986).Neural tube defects: a reviewof humanand animal studies on the etiology of neural tube defects. Teratology, 34, 171–187.

211. Sulik, K.K., Dehart, D.B., Rogers, J.M., Chernoff, N. (1995). Teratogenicity of low dosesof all-trans retinoic acid in presomite mouse embryos. Teratology, 51, 398–403.


212. Arnhold, T., Elmazar, M.M.A., Nau, H. (2002). Prevention of vitamin A teratogenesisby phytol or phytanic acid results from reduced metabolism of retinol to the teratogenicmetabolite, all-trans retinoic acid. Toxicological Sciences, 66, 274–282.

213. Elmazar, M.M.A., Nau, H. (2004). Potentiation of the teratogenic effects induced bycoadministration of retinoic acid/phytol with synthetic retinoid receptor ligands.Archivesof Toxicology, 78, 660–668.

214. Niederreither, K., Dolle, P. (2008). Retinoic acid in development: towards an integratedview. Nature Review Genetics, 9, 541–553.

215. Brown, D.D., Wang, Z., Furlow, J.D., Kanamori, A., Schwartman, R.A., Remo, B.F.,Pinder, A. (1996). The thyroid hormone-induced tail resorption program during Xenopuslaevis metamorphosis. Proceedings of the National Academy of Sciences USA, 93, 1924–1929.

216. Wang, Z., Brown, D.D. (1993). Thyroid hormone-induced gene expression program foramphibian tail resorption. Journal of Biological Chemistry, 268, 16270–16278.

217. Hollowell, J.G. Jr., Hannon, W.H. (1997). Teratogen update: iodine deficiency, a com-munity teratogen. Teratology, 55, 389–405.

218. Yang, X.F., Xu, J., Hou, X.H., Guo, H.L., Hao, L.P., Yao, P., Liu, L.G., Sun, X.F. (2006).Developmental toxic effects of chronic exposure to high doses of iodine in the mouse.Reproductive Toxicology, 22, 725–730.

219. Kohrle, J. (2000). The deiodinase family: selenoenzymes regulating thyroid hormoneavailability and action. Cellular and Molecular Life Sciences, 57, 1853–1863.

220. Sanders, J.P., van der Geyten, S., Kaptein, E., Darras, V.M., Kuhn, E.R., Leonard, J.L.,Visser, T.J. (1997). Characterization of a propylthiouracil-insensitive type 1 iodothyro-nine deiodinase. Endocrinology, 138, 5153–5160.

221. Barrington, E.J.W. (1975). The endocrine glands of the pharynx. An Introduction toGeneral Comparative Endocrinology, 2nd ed., Oxford: Clarendon Press: pp. 147–184.

222. Huang, H., Marsh-Armstrong, N., Brown, D.D. (1999). Metamorphosis is inhibited intransgenic Xenopus laevis tadpoles that overexpress type III deiodinase. Proceedings ofthe National Academy of Sciences USA, 96, 962–967.

223. Glinoer, D. (1984). Thyroid dysfunction in the pregnant patients. In: DeGroot, L.J.,Larsen, P.R., Refetoff, S., Stanbury, B., editors. The Thyroid and Its Diseases. Wiley:New York; pp. 519–540.

224. Fritzsche, B., Schuchardt, J., Schmidt, A., Nau, H., Schweigert, F.J., Ruhl, R. (2010).CYP26A1-specific antagonist influence on embryonic implantation, gene expression andendogenous retinoid concentration in rats. Reproductive Toxicology, 30, 446–451.

225. Kitamara, S., Kato, T., Lida, M., Jinno, N., Suzuki, T., Ohta, S., Fujimoto, N., Hanada,H., Kashiwagi, K., Kashiwagi, A. (2005). Anti-thyroid hormonal activity of tetrabromo-bisphenol A, a flame retardant, and related compounds: affinity to themammalian thyroidhormone receptor, and effect on tadpole metamorphosis. Life Sciences, 76, 1589–1601.

226. Hill, A.J., Teraoka, H., Heideman, W., Peterson, R.E. (2005). Zebrafish as a modelvertebrate for investigating chemical toxicity. Toxicological Sciences, 86, 6–19.

227. Huuskonen, H. (2005). New models and molecular markers in evaluation of develop-mental toxicity. Toxicology and Applied Pharmacology, 207(2 suppl), S495–S500.

228. Khor, T.I., Ibrahim, S., Kong, A.T. (2006). Toxicogenomics in drug discovery and drugdevelopment: potential applications and future challenges. Pharmacological Research,23, 1659–1663.

REFERENCES 75

229. Orphanides, G. (2003). Toxicogenomics: challenges and opportunities. ToxicologicalLetters, 140–141, 145–148.

230. Elferink, M.G.L., Olinga, P., Draaisma, A.L., Merema, M.T., Bauerschmidt, S., Polman,J., Schoonen, W.G., Groothuis, G.M.M. (2008). Microarray analysis in rat liver slicescorrectly predicts hepatotoxicity as observed in. Toxicology and Applied Pharmacology,229, 300–309.

3INCORPORATING HUMANDOSIMETRY AND EXPOSUREINFORMATION WITHHIGH-THROUGHPUT SCREENINGDATA IN CHEMICAL TOXICITYASSESSMENT

Barbara A. Wetmore and Russell S. Thomas

3.1 INTRODUCTION

Momentum has been growing worldwide in government and industry sectors torefine current strategies used to assess chemical toxicities. In Europe, the regulationsenacted in the Registration, Evaluation, Authorisation, and Restriction of Chemi-cals (REACH) will require chemical manufacturers to provide hazard and exposureinformation on the estimated 30,000 chemicals to be sold in Europe [1, 2]. As this reg-ulation applies to the chemicals themselves, all international companies who exportproducts to the European Union will also be required to comply. In the United States,the Toxic Substances Control Act (TSCA) of 1976 [3] currently requires the Environ-mental Protection Agency (EPA) to demonstrate that a chemical poses risk to humanhealth before requesting the manufacturer to submit toxicity data [4, 5]. Pending leg-islation to reauthorize TSCA, the Safe Chemicals Act of 2011 [6], proposes shiftingthe balance to the chemical manufacturers who would need to demonstrate the safetyof a chemical prior to registration. Regardless of who is ultimately responsible forproviding chemical safety or toxicity data, the complexity and expense involved tocomplete the conventional in vivo toxicity tests is great. A complete in vivo testing


77

78 DOSIMETRY IN HIGH-THROUGHPUT SCREENS

battery for chemicals in the United States typically includes assessment of carcino-genicity and chronic developmental and reproductive toxicities [7, 8]. Further, it hasbeen estimated that the REACH regulations will cost the industry $4.2 billion andrequire the use of over 45 million animals over the next 15 years [9–11].In 2007, the U.S. National Research Council (NRC) published the report “Tox-

icity Testing in the 21st Century: A Vision and a Strategy” [12]. In it, the authorscalled on toxicologists to exploit the advances made in the development of in vitrocell-based systems, high-throughput assays, systems biology, and bioinformatics tomove away from the traditional strategy of administering high doses of chemicals tolaboratory animals to identify potential adverse effects. The traditional toxicity test-ing approaches were established at a time when knowledge of the mode of action andthe role of signaling pathways in biological responses was limited and the high-doseeffects in laboratory species were presumed to be indicative of effects at relevanthuman exposures [13]. In response to the NRC report and related discourse in thetoxicology community, multiple research efforts assessing the use of in vitro data inchemical hazard identification have been initiated, including major efforts at the U.S.EPA and the U.S. National Toxicology Program [14–16].The need to transition away from traditional in vivo toxicity studies is particularly

important for the U.S. EPA, which is under increasing pressure to address concernsregarding the large number of untested chemicals. The ToxCast research project wasestablished in 2007 to evaluate hundreds to thousands of chemicals in a broad panelof in vitro high-throughput screening (HTS) assays at a fraction of the cost and timeof in vivo animal studies [14]. In the first phase of the project, a library of chemicalsfor which in vivo toxicity data were available was screened in concentration–responseformat across hundreds of cell-based and biochemical assays. The potency of eachchemical was captured as a micromolar concentration at which activity was observedin each positive assay [17–23]. The pattern of these nominal potency values amongthe in vitro assays along with other chemical information has been proposed for usein hazard identification and prioritization of chemicals for further testing [24].While nominal in vitro assay potency valuesmay indicate the activity of a chemical

in a given cellular or biochemical assay, these nominal concentrations do not takeinto account the impact of key factors in human pharmacokinetics that will ultimatelydictate the human health risk potential of these chemicals. Lack of consideration ofchemical absorption, bioavailability and clearance from the body, and exposure levelswill lead to an over- or underestimation of this risk [25].We have developed an approach to incorporate chemical-specific human pharma-

cokinetic measurements and exposure information into in vitro toxicity HTS data toprovide a more informative risk characterization of chemicals (Fig. 3.1) [26, 27]. Twocritical determinants of human steady-state pharmacokinetics are hepatic metabolicclearance and plasma protein binding. These parameters are measured using in vitroassays and are then used to estimate the steady-state concentration (Css) that wouldbe achieved for each chemical, given repeated daily exposure at a constant rate—anexposure scenario consistent with daily exposure to chemicals in the environment.Once these values are determined for each chemical, the daily human oral dose (ororal equivalent dose, in mg/kg/day) required to produce in vivo Css values in the

EXPERIMENTAL APPROACH 79

Hepatic clearance

Least sensitiveassay

Most sensitiveassay

Oral equivalent dose(mg/kg/day)

Amount of chemical a humanwould need to consume toachieve blood concentrationsequivalent to in vitro AC50concentrations across all in vitro assays

Plasma proteinbinding (Fub)

Steady-state blood concentration (Css):Upper 95th percentile among healthy

adult men and women, 20–50 years old

In vitro pharmacokineticassays

IVIVE (using Simcyp)

>600 In vitro High-thoughput screens

Cellular assays239 ToxCast phase lchemicals

CI

CI

CI

OH

O

O

N

O

O

Biochemicalassays Concentration at which

bioactivity is observed

AC50 (μm) or LEC

FIGURE 3.1 Schematic representation of the incorporation of dosimetry with high-throughput in vitro screening data to estimate the oral equivalent doses. In vitro pharmacoki-netic data measuring hepatic clearance and plasma protein binding were used to parameterizea population-based IVIVE model that simulated population variability to estimate the steady-state blood concentration (Css). Reverse dosimetry was used to estimate the oral equivalentdose required to achieve the Css at levels that elicit biological activities in the in vitro screeningassays. Oral equivalent doses were derived for each chemical-assay combination. A box andwhisker plot was used to represent the range of oral equivalent doses for a specific chemical.

blood equivalent to the in vitro AC50 or LEC value can be calculated for each HTSassay. The oral equivalent doses for each chemical-assay combination can then becompared to the human oral exposure estimates. This approach provides a robuststrategy to identify those chemicals with an ability to perturb biological pathways atconcentrations relevant to estimated human exposure levels.This chapter will cover the finer details of this approach and other important

considerations as the toxicology community strives to merge traditional risk assess-ment approaches with a quest for the incorporation of high-throughput and cell-basedtechniques into the decision-making process.

3.2 EXPERIMENTAL APPROACH

3.2.1 Approach and Design of the ToxCast Research Project

While a more detailed discussion of the ToxCast research project will be provided in aseparate chapter, some background information on the effort will be helpful in settingthe stage for this chapter. The initial phase of the ToxCast research project measured


activity of 309 compounds against over 600 in vitro assays using nine separate tech-nologies, including receptor-binding and enzyme activity assays, cell-based proteinand RNA expression assays, real-time growth measured by electronic impedance,and fluorescent cellular imaging [17]. Each chemical-assay combination was run inconcentration–response format and values that elicited 50% of the maximal activityfor that assay (AC50) or the lowest effective concentration (LEC) were estimated,depending on the type of dose–response data that were collected for each assay. Thein vitro bioactivity was assumed to be solely the result of the parent compound, asmost of the assays were metabolically inactive. A detailed description of the assaysand associated data has been provided in earlier publications and is beyond the scopeof this chapter [17–23]. All data are publicly available from the ToxCast website(http:/www.epa.gov/ncct/toxcast/data.html).The ToxCast Phase I chemical library consisted of 320 substances, of which 309

were unique chemicals (chemical and associated purity information is available athttp://www.epa.gov/ncct/toxcast/chemicals.html). Most of the chemicals selected foranalysis during Phase I were food-use pesticides for which extensive animal toxicitytesting data were available. Initial strategies sought to use this library as a training set,to match in vitro findings—whether it is specific assay activities or activity patternswithin assay groups—with existing knowledge on in vivo effects. Knowledge of theseinformative assays could then be applied to predict similar effects in the case of data-poor chemicals for which no in vivo data are available. Additional information aboutthe composition of the Phase I chemical library is beyond the scope of this chapter,but readily available in multiple publications [8, 17, 19].

3.2.2 Application of Pharmacokinetic Data in the In Vitro-to-In VivoExtrapolation of Oral Equivalent Doses

In vitro-to-in vivo extrapolation (IVIVE) is a process to utilize data generated withinin vitro assays to estimate in vivo drug or chemical fate. In the past, IVIVE has beenpredominantly developed and applied in the pharmaceutical industry to estimatetherapeutic blood concentrations for specific candidate drugs and identify potentialdrug–drug interactions [28–30]. Due to both legislative mandates and public pressurefor increased toxicity testing, IVIVE is being increasingly used to predict the in vivopharmacokinetic behavior of environmental and industrial chemicals [31].To be truly useful in a high-throughput testing strategy, the generation and incorpo-

ration of in vitro pharmacokinetic data would also need to be completed in a relativelyhigh-throughput manner. To streamline this approach, high-throughput methods tomeasure two key determinants of chemical disposition in the body—plasma proteinbinding and hepatic metabolic clearance—were employed and a series of simplifyingassumptions were made to allow the generation of oral equivalent dose estimates thatwould be suitably protective of human health. Importantly, the methods selected andassumptions made are based on well-supported data and knowledge already gatheredfrom the study of pharmacokinetics and from the validation of various approaches inthe pharmaceutical industry [32, 33]. The single most important factor that limited

EXPERIMENTAL APPROACH 81

the throughput of the approach was the measurement of chemical levels in phar-macokinetic assay samples by analytical chemistry. In the interest of maintaining areasonable throughput, only those chemicals for which methods were readily avail-able and/or for which “generic” HPLC-MS or GC-MS procedures were applicablewere analyzed. In the end, ∼80% or 239 of the 309 ToxCast Phase I chemicals weresuccessfully analyzed.High-throughput equilibrium dialysis [34] was used to measure plasma protein

binding, or the chemical fraction unbound (Fub) in the plasma. Hepatic clearancewas measured in a metabolic stability assay, where loss of the parent chemical wasmonitored over time. Primary cryopreserved human hepatocytes were employedin the metabolic clearance assays due to their ability to retain metabolic functionof both cytochrome P450 (CYP450) and non-CYP450 enzymes [35] and to moreaccurately predict in vivo metabolic activity than other available methods [36].Clearance was determined at two concentrations: (1) 1 �M was selected as it isa standard concentration used in metabolic stability studies in the pharmaceuticalindustry and, typically, a concentration at which metabolic enzyme saturation isnot usually an issue [37, 38]; (2) 10 �M was chosen as it falls in the middle ofthe concentrations tested in the ToxCast assays and ensures detectable levels of thechemicals for the chemical analyses at later time points.Once these pharmacokinetic values were determined, a simple equation [33] was

used to estimate expected steady-state blood concentrations (Css). The equation wasbased on zero-order (i.e., constant) uptake of a daily dose from the gut (assuming100% oral bioavailability) with both renal and hepatic clearance. The steady-stateconcentration in the blood is

Css = k0

(GFR× Fub)+[(Ql × Fub × Clint)(Ql × Fub × Clint)

]

where k0 = input rate (mg/kg/h), which was set to 0.042 (i.e., 1 mg/kg/day) for thepurposes of this work; Fub = unbound fraction of parent compound in the plasma,experimentally determined using equilibrium dialysis; GFR = glomerular filtrationrate, (6.7 L/h) [39]; Ql = liver blood flow (90 L/h) [40]; Clint = intrinsic metabolicclearance, derived by scaling up the experimentally determined Clin vitro to representwhole liver clearance [Clin vitro × HPGL × Vl, where HPGL = hepatocytes pergram liver (137 × 106 cells/g)] [41]; and Vl = volume of the liver (1820 g) [42].Renal clearance is represented by the left side of the denominator, or GFR × Fub.While the Css determination was based on a straightforward and established phar-

macokinetic equation,we incorporatedMonteCarlo analysis to provide an assessmentof the variability of responses one would expect in the human population. Using theSimcyp software (Simcyp Limited, Sheffield, UK; [43]), Monte Carlo analysis wasperformed to simulate variability across a population of 100 healthy individuals ofboth sexes from 20 to 50 years of age. A coefficient of variation of 30% was used forintrinsic and renal clearance. The median, upper, and lower fifth percentiles for theCss were obtained as output. For the 239 chemicals analyzed, the upper fifth percentile


values were 2.2- to 4.7-fold higher than their respective lower fifth percentile value,with a median of 3.9-fold. Availability of Css values that span the whole population,from the most sensitive group, represented by the Css value for the upper 95th per-centile, to the least sensitive group, provides the range of variability one could expectto see in responses elicited by different chemicals.Traditionally, pharmacokinetic models try to relate exposure concentrations or

external dose to the concentration of the chemical in the blood or tissue. This isknown as “forward dosimetry.” Alternately, measured or predicted blood or tissueconcentrations of chemicals can be combined with the known pharmacokinetics ofthese chemicals to estimate the exposure that would be required to achieve these mea-sured biomarker concentrations. This second approach, known as “reverse dosimetry”[44], was employed in our study. Based on the principles of reverse dosimetry, themedian, upper, and lower fifth percentiles for the Css were used as conversion factorsto generate oral equivalent doses according to the following formula:

Oral equivalent dose (mg/kg/d) = ToxCast AC50 or LEC (�M)× 1 mg/kg/d

Css (�M)

In the equation above, the oral equivalent value is linearly related to the in vitroAC50 or LEC and inversely related to Css. This equation is valid only for first-ordermetabolism that is expected at ambient exposure levels. An oral equivalent valuewas generated for each chemical and each AC50 or LEC value across the >600 invitro assays. As Clin vitro was determined at two concentrations, Css values were alsoderived at both concentrations. The Css determined for the chemical concentration(either 1 or 10�M) closest to that of the AC50 or LEC concentration for that particularassay was used in the estimation of the oral equivalent doses.Given the range in potency values across the in vitro assays for a given chemical,

a box and whisker plot provides an easy way to display the oral equivalent doses andcompare against the human exposure estimates. The Tukey approach was used indefining the values for the whiskers, setting them to 1.5 times the interquartile rangebelow or above the 25th or 75th percentiles, respectively. When the minimum oralequivalent value for each chemical was greater than the Tukey-derived value for thelower whisker, that whisker was set to that minimum value. This happened for all butnine of the chemicals in the analysis, so in essence, the most sensitive assay valuewas then used as the comparator to the human exposure estimates. Because of this,and due to the wide range of assay types and end points assessed, each assay wasweighted equally in the analysis.

3.2.3 Estimation of Human Oral Exposures

During the regulatory assessment process, the U.S. EPA Office of Pesticide Programswill estimate anticipated acute and chronic exposures to each pesticide in residen-tial and occupational settings. For the purposes of our analyses, we employed thechronic aggregate exposure estimates that would occur following oral exposure as aresult of pesticide residues on food (including livestock) and through drinking watercontamination. The route of exposure was limited to oral, as this is the primary route

FINDINGS 83

of exposure for many of the Phase I ToxCast chemicals. Occupational and far-fieldexposures were not considered. Exposure estimates were available for 80% of theToxCast Phase I chemicals analyzed and estimates for multiple age- and gender-based subpopulations were tracked. It should be noted that the estimates provided inthese regulatory documents reflect conservative, rather than accurate, values to assureregulators that human health is protected. This fact is reflected in the comparison ofthose chemicals for which concentrations were measured in the National Health andNutrition Examination Survey (NHANES) conducted by the Centers for DiseaseControl [45]. For the two chemicals for which estimates were available from bothsources, the regulatory documents provided more conservative estimates of roughlyan order of magnitude (6.1- and 11.2-fold greater for triclosan and cacodylic acid,respectively [27]).

3.3 FINDINGS

3.3.1 Comparison of the IVIVE Determinations with Existing In Vivo Data

To assess the accuracy of the IVIVE-derived Css predictions, peer-reviewed in vivopharmacokinetic studies were located for 13 of the 239 Phase I chemicals [27]. Thein vivo Css values were determined from this published data and compared against themedian values derived using the Simcyp IVIVE simulation tool, assuming a dose rateof 1 mg/kg/day and the 1 �Mmetabolic clearance rate (Table 3.1). To provide a morecomprehensive assessment, the performance of the IVIVE predictions was testedusing two different assumptions regarding hepatic clearance: (1) restrictive hepaticclearance using Fub determined experimentally in the plasma protein binding assay;and (2) nonrestrictive hepatic clearance where the Fub was set to 0.99. In restrictivehepatic clearance, the amount of chemical bound to plasma proteins restricts theamount available for metabolism, while for nonrestrictive clearance, the on/off kinet-ics of plasma protein binding has a negligible impact on clearance [33]. To replicatethese two processes, the experimentally derived Fub value was used to model restric-tive clearance and the Fub set to 0.99 was used to model nonrestrictive clearance. Inboth cases, the renal clearance was based on the experimentally derived Fub and theGFR. Using the restrictive hepatic clearance assumption, the IVIVE model predic-tions for six chemicals (2,4-dichlorophenoxyacetic acid, bisphenol-A, cacodylic acid,carbaryl, oxytetracycline dihydrate, and triclosan) were comparable (i.e., within oneorder of magnitude) to the Css values derived from published models. Of the sevenremaining chemicals, the IVIVE model significantly (i.e., by greater than an orderof magnitude) overpredicted the Css values for five chemicals (fenitrothion, lindane,parathion, picloram, and thiabendazole) and underpredicted the Css values for twochemicals (PFOS and PFOA). Using the nonrestrictive hepatic clearance assumption,the IVIVE model predictions for 8 of the 13 chemicals (2,4-dichlorophenoxyaceticacid, bisphenol-A, cacodylic acid, carbaryl, fenitrothion, oxytetracycline dihydrate,parathion, and picloram) were comparable to the published studies, with four of theremaining chemicals being underpredicted (lindane, PFOS, PFOA, and triclosan) andone being overpredicted (thiabendazole).


TABLE 3.1 Comparison of In Vitro-to-In Vivo Extrapolation (IVIVE) ModelingResults with Published In Vivo Pharmacokinetic Data or Physiologically BasedPharmacokinetic Models

Cssa (�M) IVIVECss (�M) Cssa (�M) IVIVE (Assuming(Based on (Assuming Restrictive Nonrestrictive Hepatic

Chemical In Vivo Data) Hepatic Clearance) Clearance (Fub = 0.99))2,4-D 9.05–90.05 43.32 43.32Bisphenol A <0.13b 0.37 0.06Cacodylic acid 1.80 3.15 3.15Carbaryl 0.03 0.07 0.03Fenitrothion 0.03 18.26 0.10Lindane 0.46 13.39 0.07Oxytetracycline dihydrate 0.36 1.94 1.94Parathion 0.17 24.64 0.14Perfluorooctane sulfonic acid 19,990c 153.23c 153.23c

Perfluorooctanoic acid 20,120c 53.07c 0.40c

Picloram 0.27 57.38 0.37Thiabendazole 0.45 14.15 14.15Triclosan 2–10 1.56 0.01

aCss, concentration at steady state for 1 mg/kg/day dose. Predicted using the 1 �M metabolic clearancerate.bCss, value in Reference 57 represented total bisphenol A, of which 99% is glucuronidated. The publishedvalue was divided by 100 to estimate the free concentration for this table.cPFOS and PFOA undergo active renal resorption [43, 44] and may explain the discrepancy in the listedvalues.

The assumptions in the IVIVE modeling bias the results toward overprediction(i.e., the assumption of 100% absorption and 0% extrahepatic metabolism both act toincrease Css). For example, the IVIVE-derived Css values for oxytetracycline dihy-drate overpredicted the published model by 7.8-fold, which can be explained by itslow oral bioavailability (<10%) [46, 47]. Although the assumption of nonrestrictivehepatic clearance improved the concordance within this subset of chemicals (8/13vs. 6/13), its use in the broader analysis across all the chemicals would increasethe number of underpredicted Css values. For the two chemicals where the Css val-ues were significantly underpredicted using both assumptions, PFOS and PFOA, thediscrepancy is likely due to active renal resorption, which was not incorporated inthe IVIVE model [48, 49]. Although our assumption of restrictive hepatic clearanceresults in lower concordance due to over-prediction of Css values, the overpredictionwill lead to a lower and more conservative estimate of the oral equivalent doses.

3.3.2 Comparison of Dosimetry-Adjusted Oral Equivalent Dosesand Exposure Estimates

A comparison of the dosimetry-adjusted ToxCast assay data with human exposuresidentified 18—of the 182 with exposure data—with minimum oral equivalent dose

FINDINGS 85

values that were lower than or equal to the most highly exposed subpopulation(Figs. 3.2a and 3.2b). When compared against exposure estimates for the general USpopulation, the number of chemicals with overlapping oral equivalents and exposureswas reduced to 10. Most of these 18 chemicals were herbicides or fungicides, andthe exposure estimates for many of these were based on the presence of residues onfood crops, in livestock, or in drinking water [27]. Two chemicals, 2-phenylphenoland triclosan, have both bactericidal and fungicidal properties and are found in com-mon household items such as soap, toothpaste (triclosan), and disinfectant cleaners(2-phenylphenol).Side-by-side comparison of the distributions of the dosimetry-adjusted oral equiv-

alent values (Figs. 3.2a and 3.2b) against those of the AC50 values (Figs. 3.2c and3.2d) reveals the added value that dosimetry adjustment and exposure informationprovides to the data. Prioritization based on AC50 and LEC values for chemicalswithout consideration for human exposures would lead to unnecessary animal testingfor compounds with no relevant human exposures. Further, prioritization based onin vitro AC50 and LEC values alone could under- or overestimate the risks associatedwith other chemicals because of aspects of biology that are either not captured in vitroor not modeled in silico. Either scenario is untenable from a regulatory, economic, oranimal welfare perspective.To provide a risk assessment context to these findings, activity-to-exposure ratios

(AERs) were determined for each chemical based on the minimum oral equivalentdose estimated divided by either the estimated upper limit of human exposure or theexposure estimate for the general US population. This ratio is, in essence, an in vitroassay-derived margin of exposure (MOE), with an AER <1 indicating that exposure(as estimated by the U.S. EPA) occurs at levels sufficient to cause bioactivity. Acrossall chemicals, 50% of the AERs derived for the general US population had a ratio>123.03 and 75% had a ratio >11.48 (Fig. 3.3a). The median AER derived usingestimates for the most highly exposed subpopulations was 44.67, with 75% of theAERs being>6.03 (Fig. 3.3b). It is important to note that for some chemicals theAERmay be skewed due to both the conservative assumptions in the IVIVE models andthe conservative nature of the human exposure estimate derivation in the registrationdocuments. For chemicals with limited residue or exposure information, EPA uses atiered approachwith the first tier providing themost conservative exposure estimation.As actual exposure or residue levels are determined, the exposure estimates arerefined.The standard toxicological studies performed on the Phase I chemicals in support

of product registration primarily assess end points such as histopathology, clinicalchemistry, and body weight. Few mechanistic studies have been performed on thesechemicals. A comparison of the non-human in vivo effects with the most sensitivehuman in vitro assay end points revealed a lack of concordance between the two.The most common in vivo effects reported for the 18 chemicals with an AER <1 arehepatotoxicity or liver tumors (10/18 chemicals), and themost common in vitro effectsare changes in CYP450 activity or expression (8/18 chemicals) [27]. It is important tonote that chemically induced changes in the in vitro assays do not necessarily indicatea toxic or adverse response, but rather indicate biological perturbation potentially

100000

10000

1000

100

10

1

0.1

0.01

0.001

0.0001

0.00001

Ora

l equ

ival

ent d

ose

or e

stim

ated

exp

osur

e(m

g/kg

/day

)

Tebu

pirim

fos

Trifl

umiz

ole

Tric

losa

nF

osth

iaza

teM

etho

xych

lor

Tria

sulfu

ron

Qui

ntoz

ene

For

amsu

lfuro

nF

luaz

inam

Rim

sulfu

ron

Dife

noco

nazo

leB

enom

ylP

ropo

xur

Met

sulfu

ron-

met

hyl

Mal

athi

onP

FO

ATe

fluth

rinTe

bufe

nozi

deIs

oxaf

luto

leE

PT

CH

exac

onaz

ole

Tepr

alox

ydim

2-P

heny

lphe

nol

Dim

etho

ate

Thi

azop

yrD

ifenz

oqua

t met

hyl s

ulfa

teD

icro

toph

osD

imet

hom

orph

6-D

esis

opro

pyla

traz

ine

Oxy

tetr

acyc

line

dihy

drat

eT

hiam

etho

xam

Imaz

etha

pyr

Clo

maz

one

Dic

amba

Oxy

fluor

fen

Cac

odyl

ic a

cid

Ace

phat

eTr

iadi

mef

onC

ycla

nilid

eT

hiac

lopr

idF

luro

xypy

r-m

epty

lP

heno

xyet

hano

lB

isph

enol

-ATe

tram

ethr

inM

etal

axyl

Die

thyh

exyl

pht

hala

teF

enhe

xam

idTe

buth

iuro

nV

incl

ozol

inT

hiop

hana

te m

ethy

lD

azom

etB

rom

oxyn

ilS

-Bio

alle

thrin

Car

bary

lA

lach

lor

Hex

azin

one

Ace

toch

lor

Dib

utyl

pht

hala

teD

imet

hyl p

htha

late

Dia

zoxo

n

Compound

(b)

100000

Fen

tin h

ydro

xide

Qui

noxy

fen

Spi

roxa

min

eN

iclo

sam

ide

Em

amec

tin b

enzo

ate

Aci

fluor

fen

Par

athi

onE

toxa

zole

Lind

ane

Fen

buco

nazo

leB

ensu

lfuro

n-m

ethy

lIs

oxab

enP

rope

tam

phos

Pyr

ithio

bac-

sodi

umE

sfen

vale

rate

2,4-

DD

ichl

oran

Clo

fent

ezin

eP

FO

SP

ropa

zine

Nitr

apyr

inR

oten

one

Bro

mac

ilF

enar

imol

Dic

losu

lam

Piri

mip

hos-

met

hyl

For

chlo

rfen

uron

Lact

ofen

Dith

iopy

rA

nila

zine

Chl

orpr

opha

mD

iazi

non

Flu

met

ralin

Pyr

aclo

stro

bin

Pyr

idab

enO

xadi

azon

Cou

map

hos

Tetr

acon

azol

eT

hiob

enca

rbF

lum

etsu

lam

Fen

itrot

hion

Triti

cona

zole

Met

hoxy

feno

zide

Fen

thio

nF

ludi

oxon

ilF

luox

astr

obin

Pip

eron

yl b

utox

ide

But

achl

orN

oval

uron

Imaz

aqui

nB

utyl

ate

Pen

dim

etha

linP

erm

ethr

inF

luro

xypy

rB

utra

linR

esm

ethr

inM

ethy

l par

athi

onB

osca

lidTr

iflur

alin

Nor

flura

zon

10000

1000

100

10

1

0.1

0.01

0.001

0.0001

0.00001

(a)O

ral e

quiv

alen

t dos

e or

est

imat

ed e

xpos

ure

(mg/

kg/d

ay)

FIGURE 3.2 Example box and whisker plots depicting (a, b) the oral equivalent dosesand human exposure estimates (both in mg/kg/d) and (c, d) the nominal AC50 and LECconcentrations for each chemical-assay combination. A representative subset of chemicalsanalyzed in the full study [23] are displayed here, ordered from lowest to highest median oralequivalent dose. Human exposure estimates for age- and gender-based populations are shownusing the following symbols: �, general US population; �, <1 year old; �, 1–2 years old;

100000

10000

1000

100

10

1

0.1

0.01

0.001

0.0001

0.00001

Tebu

pirim

fos

Trifl

umiz

ole

Tric

losa

nF

osth

iaza

teD

iuro

nM

etho

xych

lor

Qui

ntoz

ene

Tria

sufu

ron

For

amsu

lfuro

nF

luaz

inam

Rim

sulfu

ron

Dife

noco

nazo

leB

enom

ylP

ropo

xur

Met

sulfu

ron-

met

hyl

Mal

athi

onP

FO

ATe

fluth

rinTe

bufe

nozi

deIs

oxaf

luto

leE

PT

CH

exac

onaz

ole

Tepr

alox

ydim

2-P

heny

lphe

nol

Dim

etho

ate

Thi

azop

yrD

ifenz

oqua

t met

hyl s

ulfa

teD

icro

toph

osD

imet

hom

orph

6-D

esis

opro

pyla

traz

ine

Oxy

tetr

acyc

line

dihy

drat

eT

hiam

etho

xam

Imaz

etha

pyr

Dic

amba

Oxy

fluor

fen

Cac

odyl

ic a

cid

Ace

phat

e

Tria

dim

efon

Clo

maz

one

Cyc

lani

lide

Thi

aclo

prid

Flu

roxy

pyr-

mep

tyl

Phe

noxy

etha

nol

Bis

phen

ol-A

Tetr

amet

hrin

Met

alax

ylD

EH

PF

enhe

xam

idTe

buth

iuro

nV

incl

ozol

inT

hiop

hana

te m

ethy

lD

azom

etB

rom

oxyn

ilS

-Bio

alle

thrin

Car

bary

lA

lach

lor

Hex

azin

one

Ace

toch

lor

DB

PD

imet

hyl p

htha

late

Dia

zoxo

n

Compound

AC

50 o

r LE

C (μM

)

(d)

(c) 100000

10000

1000

100

10

1

0.1

0.01

0.001

0.0001

0.00001

AC

50 o

r LE

C (μM

)

Compound

Fen

tin h

ydro

xide

Qui

noxy

fen

Spi

roxa

min

eN

iclo

sam

ide

Em

amec

tin b

enzo

ate

Aci

fluor

fen

Par

athi

onE

toxa

zole

Lind

ane

Fen

buco

nazo

leB

ensu

lfuro

n-m

ethy

lIs

oxab

enP

rope

tam

phos

Pyr

ithio

bacN

aE

sfen

vale

rate

2,4-

DD

ichl

oran

Clo

fent

ezin

eP

FO

SP

ropa

zine

Nitr

apyr

inR

oten

one

Bro

mac

ilF

enar

imol

Dic

losu

lam

Piri

mip

hos-

met

hyl

For

chlo

rfen

uron

Lact

ofen

Dith

iopy

rA

nila

zine

Chl

orpr

opha

mD

iazi

non

Flu

met

ralin

Pyr

aclo

stro

bin

Pyr

idab

enO

xadi

azon

Cou

map

hos

Tetr

acon

azol

eT

hiob

enca

rbF

lum

etsu

lam

Fen

itrot

hion

Triti

cona

zole

Met

hoxy

feno

zide

Fen

thio

nF

ludi

oxon

ilF

luox

astr

obin

Pip

eron

yl b

utox

ide

But

achl

orN

oval

uron

Imaz

aqui

nB

utyl

ate

Pen

dim

etha

linP

erm

ethr

inF

luro

xypy

rB

utra

linR

esm

ethr

inM

ethy

l par

athi

onB

osca

lidTr

iflur

alin

Nor

flura

zon

FIGURE 3.2 (Continued) �, 3–5 years old; �, 1–6 years old; ◦, 6–12 years old; �, 13–19years old; �, 20–49 years old; inverted �, 50+ years old; and ♦, females, 13–49 years old.Arrows indicate those chemicals with oral equivalent doses that occur at or below the upperbound of the human exposure estimate. In (c, d), the AC50 data are ordered as shown in (a) and(b) to better compare the two datasets.


# of

Val

ues

Cum

ulative %

100

80

60

40

20

010–2 10–1 100 101 102 103 104

Activity: Exposure Ratio (AER) distributiongeneral US population

Distribution Information

Median

(a)

Upper quartileLower quartile

123.031122.02

11.48

105 106 107 1080

20

40

60

80

100

No.

of v

alue

s Cum

ulative %

100

80

60

40

20

010–110–2 100 101 102 103 104

Activity: Exposure Ratio (AER) distributionmost highly exposed population

Distribution Information

Median

(b)

Upper quartileLower quartile

44.67478.63

6.03

105 106 107 1080

20

40

60

80

100

FIGURE 3.3 Distribution of the AERs for the ToxCast Phase I chemicals. To calculate theAER values, the minimum oral equivalent dose for each chemical was divided by exposureestimates derived from US regulatory documents. The histograms and a line displaying thecumulative percent values for the AERs derived (a) for the general US population and (b) themost highly exposed subpopulation are provided. The vertical dashed line indicates AER= 1.The percentile information of the distributions is also provided.

DISCUSSION 89

followed by adaptation and return to homeostasis or, alternatively, toxicity. Whetheror not such perturbations in in vitro assays indicate an adaptive or toxic response is thesource of significant debate in the toxicology community. Researchers in the ToxCastprogram are currently grouping in vitro assays based on the signaling pathwaysto which they belong with the goal of defining predictive signatures to be used indetermining the potential in vivo responses [17].

3.4 DISCUSSION

Multiple research efforts in the United States and Europe aim to use in vitro HTSassays in the toxicological assessment of environmental chemicals [15, 50, 51]. How-ever, meaningful incorporation of these in vitro findings into such an assessment isdependent upon adequate consideration of in vivo pharmacokinetics to determine therelevance of these data to the external and internal doses achieved during humanexposure scenarios [25, 52]. Within the pharmaceutical industry, IVIVE modelingapproaches have been widely used to assess the preclinical pharmacokinetics of can-didate molecules [32, 53]. In the environmental field, traditional pharmacokineticmodels and reverse dosimetry have been used to estimate human doses that resultin chemical concentrations in human biofluids that are equivalent to biomonitoringdata [54, 55]. The doses are then compared with the reference dose (RfD) or otherexposure guidance values to provide context to the biomonitoring data [54, 55]. Bycombining advances in the pharmaceutical and biomonitoring fields, focused in vitropharmacokinetic assays and IVIVE modeling offer a similar context for the interpre-tation of in vitro toxicity HTS data through the estimation of oral equivalent doses thatare expected to produce blood concentrations equivalent to in vitro concentrationsshowing activity in the HTS assays.Given the large number of chemicals analyzed in this study and the interest in

maintaining a reasonable throughput, the IVIVE modeling was limited to the oralroute and a set of simplifying assumptions were employed. Each chemical wasassumed to have 100% oral bioavailability and elimination of the parent compoundwas limited to hepatic metabolism and glomerular filtration. These assumptionsshould generally be conservative from a human health standpoint, because lowerabsorption or additional routes of excretion would result in a lower estimate of theoral equivalent dose required to achieve a specific plasma Css. One exception wouldoccur if there were active renal resorption, which would result in a higher plasmaCss at a given dose. Alternatively, high biliary clearance of parent or enterohepaticrecirculation could also play a factor. Our comparison of the IVIVE results againstpublished in vivo studies confirmed this scenario.Evaluation of this high-throughput reverse dosimetry approach has revealed impor-

tant considerations. First, our analysis is predicated on the assumption that plasmaconcentrations equivalent to in vitro AC50 or LEC values would produce responsesin vivo. The concentration of free chemical in an in vitro assay that elicits a responsemay differ from the assigned AC50 or LEC value due to factors such as protein–lipidcomposition of the media and binding of the chemical to plastics [25]. Second, Css


values based on a chronic daily exposure may not be appropriate to use in the assess-ment of certain toxicities, such as those elicited by reproductive or developmentaltoxicants, in which periods or windows of susceptibility are more important [31].Finally, as this approach is refined, values other than the AC50 and LEC values maybe used as the basis for estimating the oral equivalent doses. Given the goal of usingthese assays in prioritization and risk assessment, other methods such as benchmarkdose analysis [56] may be needed to estimate the minimum concentration requiredto observe a biological effect above that seen in controls.It should be noted that the throughput of this approach is limited by the compara-

tively slow speed at which the analytical chemistry samples derived from the in vitropharmacokinetic assays can be processed. Given this, it is likely that this strategywould be utilized in a second or third tier assessment, where it can be applied toa smaller subset of chemicals after an initial screening has been conducted. Thismodification is acceptable given that most of the chemicals identified as having oralequivalents that overlap with exposure estimates were due to a low AC50 value ina sensitive assay rather than to a high human exposure. The ToxCast project hasalready generated in vitro assay data on the Phase II chemical library, which ispopulated primarily with non-food use chemicals with limited in vivo toxicity andexposure data. For these chemicals, a low AC50 value could be used to prioritizechemicals for additional testing (assuming the assay is relevant) rather than exposurelevels.

3.5 CONCLUSION

Regulatory, economic as well as animal welfare concerns have fueled the debate inthe toxicology community about the future of toxicity testing. All visions expressedin the NRC report [12] and in federal and academic position pieces [51, 57] embracethe promise of high-throughput in vitro assays that can identify potential proximalbiochemical and cellular interactions at a fraction of the time and cost of an in vivostudy. If pursued, the transition to a new in vitro testing paradigm should incor-porate a parallel investment in understanding chemical pharmacokinetics, as thisknowledge will serve to transform in vitro bioactivity data to an in vivo context thatcan then be applied, along with human exposure estimates, to determine chemicaltesting priorities [25, 58]. As in vitro assays are refined to better reflect in vivoadverse effects and IVIVE modeling improves, this suite of tools may eventuallybe used not only in hazard-based prioritization but also in chemical risk assessment[59, 60].

FUNDING

TheAmerican Chemistry Council’s LongRange Research Initiative provided fundingfor the research.

REFERENCES 91

ACKNOWLEDGMENT

The authors thank Simcyp Limited for providing access to Simcyp Population-basedADME Simulator under a not-for-profit license agreement.

REFERENCES

1. U.S. Department of Commerce. (2008). The Latest on REACH. U.S. Department ofCommerce, U.S. Commercial Service, U.S. Mission to the European Union. Available:http://www.buyusa.gov/europeanunion/reach. Accessed 20 February 2012.

2. European Commission, Environment Directorate. (2008). What is REACH? Available athttp://e.c./europa.eu/environment/chemicals/reach/reach intro.htm. Accessed 20 Febru-ary 2012.

3. Toxic Substances Control Act. (1976). Public Law. 94–469.

4. Locke, P.A., Myers, D.B., Jr. (2011). A replacement-first approach to toxicity testing isnecessary to successfully reauthorize TSCA. ALTEX, 28, 266–272.

5. Wilson, M.P., Schwarzman, M.R. (2009). Toward a new U.S. chemicals policy: rebuild-ing the foundation to advance new science, green chemistry, and environmental health.Environmental Health Perspectives, 117, 1202–1209.

6. Safe Chemicals Act of 2011, S. 847.

7. Chhabra, R.S., Bucher, J.R., Wolfe, M., Portier, C. (2003). Toxicity characterization ofenvironmental chemicals by the US National Toxicology Program: an overview. Interna-tional Journal of Hygiene and Environmental Health, 206, 437–445.

8. Martin,M.T., Judson,R.S., Reif,D.M.,Kavlock,R.J., Dix,D.J. (2009). Profiling chemicalsbased on chronic toxicity results from the U.S. EPA ToxRef Database. EnvironmentalHealth Perspectives, 117, 392–399.

9. Breithaupt, H. (2006). The costs of REACH. REACH is largely welcomed, but the require-ment to test existing chemicals for adverse effects is not good news for all. EMBO Reports,7, 968–971.

10. Brown, V.J. (2003). REACHing for chemical safety. Environmental Health Perspectives,111, A766–A769.

11. Hofer, T., Gerner, I., Gundert-Remy, U., Liebsch, M., Schulte, A., Spielmann, H., Vogel,R., Wettig, K. (2004). Animal testing and alternative approaches for the human healthrisk assessment under the proposed new European chemicals regulation. Archives ofToxicology, 78, 549–564.

12. NRC. (2007). Toxicity Testing in the 21st Century: A Vision and a Strategy. NationalResearch Council of the National Academies: Washington, DC.

13. Andersen, M.E., Krewski, D. (2009). Toxicity testing in the 21st century: bringing thevision to life. Toxicological Sciences, 107, 324–330.


15. Kavlock, R.J., Austin, C.P., Tice, R.R. (2009). Toxicity testing in the 21st century:implications for human health risk assessment. Risk Analysis, 29, 485–487; discussion492–497.


16. Shukla, S.J., Huang, R., Austin, C.P., Xia, M. (2010). The future of toxicity testing: afocus on in vitro methods using a quantitative high-throughput screening platform. DrugDiscovery Today, 15, 997–1007.

17. Judson, R.S., Houck, K.A., Kavlock, R.J., Knudsen, T.B., Martin, M.T., Mortensen, H.M.,Reif, D.M., Rotroff, D.M., Shah, I., Richard, A.M., Dix, D.J. (2010). In vitro screening ofenvironmental chemicals for targeted testing prioritization: the ToxCast project. Environ-mental Health Perspectives, 118, 485–492.

18. Huang, R., Xia, M., Cho, M.H., Sakamuru, S., Shinn, P., Houck, K.A., Dix, D.J., Judson,R.S., Witt, K.L., Kavlock, R.J., Tice, R.R., Austin, C.P. (2011). Chemical genomics pro-filing of environmental chemical modulation of human nuclear receptors. EnvironmentalHealth Perspectives, 119, 1142–1148.

19. Knudsen, T.B., Houck, K.A., Sipes, N.S., Singh, A.V., Judson, R.S., Martin, M.T., Weiss-man, A., Kleinstreuer, N.C., Mortensen, H.M., Reif, D.M., Rabinowitz, J.R., Setzer, R.W.,Richard, A.M., Dix, D.J., Kavlock, R.J. (2011). Activity profiles of 309 ToxCast chemicalsevaluated across 292 biochemical targets. Toxicology, 282, 1–15.

20. Martin, M.T., Dix, D.J., Judson, R.S., Kavlock, R.J., Reif, D.M., Richard, A.M.,Rotroff, D.M., Romanov, S., Medvedev, A., Poltoratskaya, N., Gambarian, M., Moeser,M., Makarov, S.S., Houck, K.A. (2010). Impact of environmental chemicals on keytranscription regulators and correlation to toxicity end points within EPA’s ToxCast pro-gram. Chemical Research in Toxicology, 23, 578–590.

21. Houck, K.A., Dix, D.J., Judson, R.S., Kavlock, R.J., Yang, J., Berg, E.L. (2009). Profilingbioactivity of the ToxCast chemical library using BioMAP primary human cell systems.Journal of Biomolecular Screening, 14, 1054–1066.

22. Rotroff, D.M., Beam, A.L., Dix, D.J., Farmer, A., Freeman, K.M., Houck, K.A., Jud-son, R.S., LeCluyse, E.L., Martin, M.T., Reif, D.M., Ferguson, S.S. (2010). Xenobiotic-metabolizing enzyme and transporter gene expression in primary cultures of human hep-atocytes modulated by ToxCast chemicals. Journal of Toxicology and EnvironmentalHealth. Part B, Critical Reviews, 13, 329–346.

23. Knight, A.W., Little, S., Houck, K., Dix, D., Judson, R., Richard, A., McCarroll, N., Aker-man, G., Yang, C., Birrell, L., Walmsley, R.M. (2009). Evaluation of high-throughputgenotoxicity assays used in profiling the US EPA ToxCast chemicals. Regulatory Toxicol-ogy and Pharmacology, 55, 188–199.

24. Reif, D.M., Martin, M.T., Tan, S.W., Houck, K.A., Judson, R.S., Richard, A.M., Knudsen,T.B., Dix,D.J., Kavlock, R.J. (2010). Endocrine profiling and prioritization of environmen-tal chemicals using ToxCast data. Environmental Health Perspectives, 118, 1714–1720.

25. Blaauboer, B.J. (2010). Biokinetic modeling and in vitro-in vivo extrapolations. Journalof Toxicology and Environmental Health. Part B, 13, 242–252.

26. Rotroff, D.M., Wetmore, B.A., Dix, D.J., Ferguson, S.S., Clewell, H.J., Houck, K.A.,Lecluyse, E.L., Andersen,M.E., Judson, R.S., Smith, C.M., Sochaski,M.A., Kavlock, R.J.,Boellmann, F., Martin, M.T., Reif, D.M., Wambaugh, J.F., Thomas, R.S. (2010). Incor-porating human dosimetry and exposure into high-throughput in vitro toxicity screening.Toxicological Sciences, 117, 348–358.

27. Wetmore, B.A.,Wambaugh, J.F., Ferguson, S.S., Sochaski,M.A., Rotroff, D.M., Freeman,K., Clewell, H.J., 3rd, Dix, D.J., Andersen, M.E., Houck, K.A., Allen, B., Judson, R.S.,Singh, R., Kavlock, R.J., Richard, A.M., Thomas, R.S. (2012). Integration of dosimetry,exposure, and high-throughput screening data in chemical toxicity assessment. Toxicolog-ical Sciences, 125, 157–174.

REFERENCES 93

28. Gibson, G.G., Rostami-Hodjegan, A. (2007). Modelling and simulation in prediction ofhuman xenobiotic absorption, distribution, metabolism and excretion (ADME): in vitro-invivo extrapolations (IVIVE). Xenobiotica, 37, 1013–1014.

29. Shaffer, C.L., Scialis, R.J., Rong, H., Obach, R.S. (2011). Using Simcyp to projecthuman oral pharmacokinetic variability in early drug research to mitigate mechanism-based adverse events. Biopharmaceutics and Drug Disposition, 33, 72–84.

30. Chen, Y., Jin, J.Y., Mukadam, S., Malhi, V., Kenny, J.R. (2012). Application of IVIVEand PBPK modeling in prospective prediction of clinical pharmacokinetics: strategy andapproach during the drug discovery phase with four case studies. Biopharmaceutics andDrug Disposition, 33, 85–98.

31. Basketter, D.A., Clewell, H., Kimber, I., Rossi, A., Blaauboer, B., Burrier, R., Daneshian,M., Eskes, C., Goldberg, A., Hasiwa, N., Hoffmann, S., Jaworska, J., Knudsen, T.B.,Landsiedel, R., Leist, M., Locke, P., Maxwell, G., McKim, J., McVey, E.A., Ouedraogo,G., Patlewicz, G., Pelkonen, O., Roggen, E., Rovida, C., Ruhdel, I., Schwarz,M., Schepky,A., Schoeters, G., Skinner, N., Trentz, K., Turner, M., Vanparys, P., Yager, J., Zurlo, J.,Hartung, T. (2012). A roadmap for the development of alternative (non-animal) methodsfor systemic toxicity testing—t4 Report. ALTEX, 29, 3–91.

32. De Buck, S.S., Mackie, C.E. (2007). Physiologically based approaches towards the pre-diction of pharmacokinetics: in vitro–in vivo extrapolation. Expert Opinion on DrugMetabolism & Toxicology, 3, 865–878.

33. Wilkinson, R.G., Shand, D.G. (1975). Commentary: physiological approach to hepaticdrug clearance. Clinical Pharmacology and Therapeutics, 18, 377–390.

34. Waters, N.J., Jones, R., Williams, G., Sohal, B. (2008). Validation of a rapid equilibriumdialysis approach for the measurement of plasma protein binding. Journal of Pharmaceu-tical Sciences, 97, 4586–4595.

35. Li, A.P., Lu, C., Brent, J.A., Pham,C., Fackett, A., Ruegg, C.E., Silber, P.M. (1999). Cryop-reserved human hepatocytes: characterization of drug-metabolizing enzyme activities andapplications in higher throughput screening assays for hepatotoxicity, metabolic stability,and drug–drug interaction potential. Chemico-Biological Interactions, 121, 17–35.

36. Jones, H.M., Houston, J.B. (2004). Substrate depletion approach for determining in vitrometabolic clearance: time dependencies in hepatocyte and microsomal incubations. DrugMetabolism and Disposition, 32, 973–982.

37. Naritomi, Y., Terashita, S., Kimura, S., Suzuki, A., Kagayama, A., Sugiyama, Y. (2001).Prediction of human hepatic clearance from in vivo animal experiments and in vitrometabolic studies with liver microsomes from animals and humans.Drug Metabolism andDisposition, 29, 1316–1324.

38. Obach, R.S. (1999). Prediction of human clearance of twenty-nine drugs from hepaticmicrosomal intrinsic clearance data: an examination of in vitro half-life approach andnonspecific binding to microsomes. Drug Metabolism and Disposition, 27, 1350–1359.

39. Rule, A.D., Gussak, H.M., Pond, G.R., Bergstralh, E.J., Stegall, M.D., Cosio, F.G., Larson,T.S. (2004). Measured and estimated GFR in healthy potential kidney donors. AmericanJournal of Kidney Diseases, 43, 112–119.

40. Davies, B., Morris, T. (1993). Physiological parameters in laboratory animals and humans.Pharmaceutical Research, 10, 1093–1095.

41. Barter, Z.E., Bayliss,M.K., Beaune, P.H., Boobis, A.R., Carlile, D.J., Edwards, R.J., Hous-ton, J.B., Lake, B.G., Lipscomb, J.C., Pelkonen, O.R., Tucker, G.T., Rostami-Hodjegan,


A. (2007). Scaling factors for the extrapolation of in vivo metabolic drug clearance fromin vitro data: reaching a consensus on values of human microsomal protein and hepato-cellularity per gram of liver. Current Drug Metabolism, 8, 33–45.

42. Johnson, T.N., Tucker, G.T., Tanner, M.S., Rostami-Hodjegan, A. (2005). Changes inliver volume from birth to adulthood: a meta-analysis. Liver Transplantation, 11, 1481–1493.

43. Jamei, M., Marciniak, S., Feng, K., Barnett, A., Tucker, G., Rostami-Hodjegan, A. (2009).The Simcyp R© Population-based ADME Simulator. Expert Opinion Drug MetabolismToxicology, 5, 211–223.

44. Liao, K.H., Tan, Y.M., Clewell, H.J., 3rd. (2007). Development of a screening approachto interpret human biomonitoring data on volatile organic compounds: reverse dosimetryon biomonitoring data for trichloroethylene. Risk Analysis, 27, 1223–1236.

45. CDC. (2010). National Health and Nutrition Examination Survey’s Fourth NationalReport on Human Exposure to Environmental Chemicals Atlanta, GA. Available athttp://www.cdc.gov/exposurereport/pdf/FourthReport.pdf. Accessed 20 February 2012.

46. Bjorklund, H.V., Bylund, G. (1991). Comparative pharmacokinetics and bioavailability ofoxolinic acid and oxytetracycline in rainbow trout (Oncorhynchus mykiss). Xenobiotica,21, 1511–1520.

47. Nielsen, P., Gyrd-Hansen, N. (1996). Bioavailability of oxytetracycline, tetracycline andchlortetracycline after oral administration to fed and fasted pigs. Journal of VeterinaryPharmacology and Therapeutics, 19, 305–311.

48. Andersen, M.E., Clewell, H.J., 3rd, Tan, Y.M., Butenhoff, J.L., Olsen, G.W. (2006). Phar-macokinetic modeling of saturable, renal resorption of perfluoroalkylacids in monkeys—probing the determinants of long plasma half-lives. Toxicology, 227, 156–164.

49. Loccisano,A.E., Campbell, J.L., Jr., Andersen,M.E., Clewell, H.J., 3rd. (2011). Evaluationand prediction of pharmacokinetics of PFOA and PFOS in the monkey and human usinga PBPK model. Regulatory Toxicology and Pharmacology, 59, 157–175.

50. Knight, A. (2008). Non-animal methodologies within biomedical research and toxicitytesting. ALTEX, 25, 213–231.

51. Abbott, A. (2009). Toxicity testing gets a makeover. Nature, 461, 158.

52. Hays, S.M., Aylward, L.L. (2009). Using Biomonitoring Equivalents to interpret humanbiomonitoring data in a public health risk context. Journal of Applied Toxicology, 29,275–288.

53. Caldwell, G.W., Yan, Z., Tang, W., Dasgupta, M., Hasting, B. (2009). ADME optimiza-tion and toxicity assessment in early- and late-phase drug discovery. Current Topics inMedicinal Chemistry, 9, 965–980.

54. Hays, S.M., Aylward, L.L., LaKind, J.S., Bartels, M.J., Barton, H.A., Boogaard, P.J.,Brunk, C., DiZio, S., Dourson, M., Goldstein, D.A., Lipscomb, J., Kilpatrick, M.E.,Krewski, D., Krishnan, K., Nordberg, M., Okino, M., Tan, Y.M., Viau, C., Yager, J.W.(2008). Guidelines for the derivation of Biomonitoring Equivalents: report from theBiomonitoring Equivalents Expert Workshop. Regulatory Toxicology and Pharmacology,51, S4–S15.

55. LaKind, J.S., Aylward, L.L., Brunk, C., DiZio, S., Dourson, M., Goldstein, D.A., Kil-patrick, M.E., Krewski, D., Bartels, M.J., Barton, H.A., Boogaard, P.J., Lipscomb, J.,Krishnan, K., Nordberg, M., Okino, M., Tan, Y.M., Viau, C., Yager, J.W., Hays, S.M.(2008). Guidelines for the communication of Biomonitoring Equivalents: report from the

REFERENCES 95

Biomonitoring Equivalents Expert Workshop. Regulatory Toxicology and Pharmacology,51, S16–S26.

56. Crump, K.S. (1995). Calculation of benchmark dose from continuous data. Risk Analysis,15, 79–89.

57. Collins, F.S., Gray, G.M., Bucher, J.R. (2008). Toxicology. Transforming environmentalhealth protection. Science, 319, 906–907.

58. CohenHubal, E.A., Richard, A., Aylward, L., Edwards, S., Gallagher, J., Goldsmith,M.R.,Isukapalli, S., Tornero-Velez, R., Weber, E., Kavlock, R. (2010). Advancing exposurecharacterization for chemical evaluation and risk assessment. Journal of Toxicology andEnvironmental Health. Part B, Critical Reviews, 13, 299–313.

59. Bhattacharya, S., Zhang, Q., Carmichael, P.C., Boekelheide, K., Andersen, M.E. (2011).Toxicity testing in the 21st century: defining new risk assessment approaches based onperturbation of intracellular toxicity pathways using the p53 DNA-damage and repairpathway as a prototype. PLoS One, 6, e20887.

60. Judson, R.S., Kavlock, R.J., Setzer, R.W., Cohen Hubal, E.A., Martin, M.T., Knudsen,T.B., Houck, K.A., Thomas, R.S., Wetmore, B.A., Dix, D.J. (2011). Estimating toxicity-related biological pathway altering doses for high-throughput chemical risk assessment.Chemical Research in Toxicology, 24, 451–462.

4THE USE OF HUMAN EMBRYONICSTEM CELLS IN HIGH-THROUGHPUTTOXICITY ASSAYS

Xin Huang, Dan-yan Zhu, and Yi-jia Lou

4.1 INTRODUCTION

Human embryonic stem cells (hESCs) are isolated from the inner cell mass of thepreimplantation blastocyte at this stage by plating them on a monolayer of mitoticallyinactivated mouse embryonic fibroblasts (MEF), which serve as supporting feedercells [1]. hESCs possess two properties that make them particularly valuable: First,they show self-renewal and can theoretically be grown indefinitely. Therefore, theypotentially provide an unlimited source of identical cells. Second, they are pluripotent,meaning that they have the potential to differentiate into any functional cell type of thehuman body. To induce differentiation in vitro, hESCs are usually removed from thefeeders and cultured in suspension, where they spontaneously form three-dimensionalaggregates called embryoid bodies (EBs) that contain derivatives of the three germlayers [2]. Multiple variations of this protocol are now available in the literature,which allow various degrees of lineage bias and the enrichment of specific cell types.These protocols are increasingly based on the use of chemically defined culture mediasupplemented with hormones and growth factors, in sequences and concentrationsoften deduced from signaling pathways known to control differentiation in earlyembryonic development.ESC differentiation assays are a particularly promising approach for replacing

some of the in vivo tests. Numerous protocols are available that allow the differentia-tion of ESCs into cell lineages that represent the three germ layers of the developing


97

98 HUMAN EMBRYONIC STEM CELLS IN HIGH-THROUGHPUT TOXICITY ASSAYS

embryo (ectoderm, mesoderm, and endoderm). hESCs have been differentiated intoa variety of cell types, including neurons, hepatocytes, and cardiomyocytes. The dif-ferentiating cells display a variety of developmental processes suitable to monitoradverse effects of drug or chemical exposure [3]. The core benefit of hESC technol-ogy in drug discovery and toxicology resides precisely in the fact that this renewablesource of human cell types produces cell-based assays to improve our ability to pre-dict human efficacy, drug toxicity, and identify chemical targets at their specific site ofaction. hESCs additionally have the potential to provide solutions to problems relatedto interspecies’ differences and methods for human gene polymorphism screenings,thus supporting robust human hazard identification and optimized drug discoverystrategies [4]. Hence, it is expected that hESC technology will have a significant andpositive impact on the soaring compound attrition rates and increasing costs of drugdiscovery and development [5].

4.2 POTENTIAL ROLE OF hESCs IN HIGH-THROUGHPUTSCREENING STRATEGIES FOR TOXICITY

Cell-based high-throughput screening (HTS) technologies to identify compounds thatinteract with validated molecular targets are already available and new ones are beingdeveloped. Stem cells clearly provide both opportunities and challenges for use indrug discovery. They may provide an unlimited source of reproducible cells from avery wide range of tissues and therefore more accurately reflect the properties of cellsin vivo, ultimately providing many more clinically relevant phenotypes than thosecurrently available. It is likely that they will also model human cells more accurately[6]. The use of hESCs in drug discovery is now rapidly progressing from a nascentlevel of development to a point where the cells may begin to be widely employed inHTS, lead optimization, and profiling.ESCs are also being used in the toxicological assessment of novel compounds

with several ESC-derived models recently being adopted by European regulatoryagencies. For example, Paquette et al. [7] assessed the so-called Embryonic StemCell Test (EST), designed by the European Centre for the Validation of AlternativeMethods (ECVAM) to identify the developmental toxicity of novel compounds. Inthis case, compounds are screened for effects on ESC differentiation as one measureof toxicity. Paquette et al. [7] reported that results from internal studies at Pfizer werecomparable to those previously obtained by ECVAM, thus suggesting that the useof ESCs could provide a useful primary in vitro assay to identify potential terato-gens. Hambor [8] described combined efforts of major pharmaceutical companiesto develop hepatocytes derived from human stem cells for use in absorption, dis-tribution, metabolism, and excretion (ADME) analyses and reproductive toxicity aswell as cardiotoxicity evaluation. This is in part due to the availability of these cellsfrom human patient cohorts expressing a defined genetic variability in terms of drugmetabolism and toxicity [6].The development of new drugs is costly and time-consuming. In particular, the

initial stages of research and development (R&D) require in vitro models to screen

ENDPOINT-BASED HTS IN TOXICITY ASSAYS 99

the activity and toxicity of a large number of compounds. Thus, a suitable modelthat can be used for both effects and safety assessment is extremely important. Cell-based in vitro assays with high human relevance are urgently needed for preclinicalscreening activities [9]. Previous studies have suggested that ESCs could serve as ascreening platform to identify low-molecular-weight compounds that affect endoge-nous stem cell populations and the repair of damaged tissue; in the meantime, aset of screening protocols is available (i.e., the primary screen and the secondaryassay) [9].The use of hESCs to predict the impact of chemicals on human health has the

following potential advantages: An unlimited source of identical cells (self-renew),differentiation towards various different functional human cell types and the abil-ity to directly predict human health effects, the influence of genetic factors, differentsusceptibilities within the human population, and the influence on early human devel-opmental processes [10].Numerous molecular in vitro assays have thus found their way into the toolbox

for pharmaceutical substance screening, including induced pluripotent stem cells(iPSCs) and ESCs. The US National Institutes of Health integrated high-throughputassay development and target evaluation into its “Molecular Libraries Program” in2005, and in 2007 the US Environmental Protection Agency launched the “ToxCast”program for the prediction of toxicity based on in vitro HTS [11].The applicability of HTS for toxicity has been shown in biochemical cytotoxicity

assays, generating dose response curves for up to 60,000 substances in one singleexperiment. The extended concept includes cell viability assays as well as tran-script analyses, and is a valuable tool for identifying species-independent molecularmarkers [11].

4.3 ENDPOINT-BASED HTS IN TOXICITY ASSAYS

4.3.1 hESC Applications for Cardiotoxicity HTS

Cardiotoxicity testing is a key activity in the pharmaceutical industry when tryingto detect detrimental effects of new drugs. A reliable human in vitro model wouldbe beneficial both regarding the selection of lead compounds and the reduction ofthe number of animal experiments. However, the human heart is a complex organcomposed of many distinct types of cardiomyocytes. In contrast, cardiomyocyteclusters (CMCs) derived from hESCs could be an option as a cell-based model. Dataon functional properties of CMCs demonstrate similarities to their in vivo analoguesin humans [11].As an alternative, hESCs have been differentiated into functional cardiomyocytes,

which express numerous cardiomyocyte proteins and display spontaneous rhythmiccontractions [12, 13]. These contractions can be recorded in vitro by using elec-trophysiological measurements such as conventional patch clamp or multi-electrodearrays. Studies using these technologies have shown promising results in predictingcardiac toxicity and QT prolongation of reference drugs [14, 15].


Some challenges remain, since the current differentiation protocols show a lim-ited yield of functional cardiomyocytes with variable degrees of purity. Moreover,ESC-derived cardiomyocytes display an early and heterogeneous phenotype that dif-fers from adult cardiomyocytes. During mammalian development, the heart growsthrough proliferation of its component cell types, 60% ofwhich are cardiac fibroblastsand 30% force-generating heart muscle cells (cardiomyocytes), while the remainingcells are vascular endothelial and smooth muscle cells. After birth, the proliferationof cardiomyocytes rapidly declines and the heart mainly grows by an increase incardiomyocyte size [16].The EST uses a continuous murine ESC line, which can be induced to differentiate

via the formation of EBs into contracting cardiomyocytes. After 10 days of culture,the extent of differentiation is evaluated microscopically, and compounds inhibitingthis differentiation process are considered potential developmental toxicants. TheEST is an in vitro tool to assess not only the developmental toxic potency but alsothe cardiotoxicity of test compounds in early development.Recently, the automated image recording of contractile cardiomyocyte-like cells

in the EST was described [17]. This method allows an unbiased high-throughputassessment of the embryotoxic potency of test compounds, resulting in an out-come comparable to manual analysis [17]. The image analysis system consists ofan inverted microscope equipped with a motorized (x,y)-stage, focus drive, and aFirewire monochrome CCD camera. The UGR software running on a standard PC isused to control both the stage and the camera. The selection of the EBs occurs manu-ally by moving the stage to the centre of the EBs, followed by an auto-focus step, andsubsequent time-lapse image recording. The area of contractile cells is computed ateach time point, and the automated assessment of contractility in EBs can be used todetermine drug-induced cardiotoxicity. More recently, studies suggest that molecularbiological approaches, such as transcriptomics, may help to improve the predictionaccuracy of this test system.

4.3.2 hESC Applications for Neurotoxicity HTS

In vitro, the differentiation of hESCs into neurons proceeds as a multistep processthat in many ways recapitulates development of embryonic neurons. Recently, astudy investigated the potential of a hESC-based model to detect neuronal develop-ment toxicity during embryogenesis [18]. The first step was the establishment of aprotocol for neuronal differentiation of hESCs. Neuron precursor cell differentiationwas performed according to literature [19], while Stummann et al. [18] developeda new protocol for maturation of the neuron precursor cells into neuron-like cells.The novelties, as compared to the protocol in the literature [19], were the use ofNeural Progenitor Cell Basal MediumTM (NPCB medium) supplemented with glialcell-derived neurotrophic factor (GDNF) and the replating of the neuroepithelial pre-cursors in a monolayer. The advantages were higher replating efficiency, appearanceof an increased number of cells with neuron-like morphology, as well as a smallerstandard error of the means and amore steady increase of the assessed neuronal mark-ers (�-tubulin III and MAP2) indicating more continuous neuronal differentiation. In

ENDPOINT-BASED HTS IN TOXICITY ASSAYS 101

order to assess whether the hESC model can detect chemically induced developmen-tal toxicity, the authors exposed the differentiating cells to methylmercury (MeHg),which is known to cause structural developmental abnormalities in the brain. Twoseparate exposure intervals were used to determine the effects of MeHg on neuronalprecursor formation and their further maturation. The formation of precursors wassensitive to MeHg at noncytotoxic concentrations, as the expression of several neu-ronal mRNA markers was modified. In contrast, noncytotoxic MeHg concentrationsdid not affect the mRNA marker expression in maturated cells, indicating that theneuronal precursor cell formation is more sensitive to MeHg than the later stagesof neuronal differentiation. The experiments demonstrate that the hESC assay canprovide alerts for the adverse effects of MeHg on neuronal development.Neurons derived from hESCs are potentially valuable in drug screening and a

possible source of donor tissue for transplantation in Parkinson’s disease. Recently,a simple and highly reproducible culture protocol that induces expandable dopamine(DA) neuron progenitors from hESCs in attached cultures was described [20]. Theprotocol consists of two simple steps: Coculturing with PA6 feeder cells, a bonemarrow-derived stromal cell line, and culturing on gelatin-coated slides. In contrast,during the expansion phase a simpler and faster cell dissociation method was used.With this latter technique, we found out that the neural progenitors proliferate abouttenfold every 2 weeks (1:3 passage every 5–7 days) and still retain a relativelyhomogeneous cellular composition. Therefore, the protocol can be used to generatehumanDAneuronsmay be suitable for the study of diseasemechanisms, drug toxicitystudies, drug screening, and intracerebral transplantation.

4.3.3 hESC Applications for Hepatoxicity HTS

Differentiation protocols triggering cultured pluripotent human ES cells into func-tional hepatocytes have been found to mimic hepatogenesis when adding solublemedium factors and reconstructing the cell matrix [21–24].hESCs have been differentiated into hepatocyte-like cells that show a number of

liver-specific morphological and functional characteristics, such as the expressionof liver-specific genes, glycogen storage, albumin secretion, and cytochrome P450(CYP) activity [22, 24–26]. ESCs are differentiated according to a three-step dif-ferentiation protocol, in which the definitive endoderm is first formed, followed byhepatic progenitor cells, which are finally differentiated and maturated into func-tional hepatocyte-like cells [27, 28]. After 2 days preincubation, hESCs are ready fora further experiment. The first step takes 3 days, in which a combination of activin,basic fibroblast growth factor (FGF2), bone morphogenetic protein 4 (BMP4), andLY294004 drives differentiation of hESCs into the definitive endoderm. The secondstep takes 5 days, in which differentiation of human endoderm cells into hepaticprogenitors occurs. The third step takes 10 days, in which maturation of hepatic pro-genitors into hepatocyte-like cells takes place. This differentiation procedure is bettersuited for toxicity HTS, since it better represents the in vivo hepatic developmentalprocess and the generated hepatocyte-like cells display improved functional charac-teristics (expression of liver-associated genes and proteins, as well as CYP-dependent


enzyme activities). Furthermore, the index at the endpoint of any step could be usedas a suitable evaluation platform of HTS.Hepatocytes derived from hESCs could provide a defined and renewable source

of human cells relevant for cell replacement therapies and toxicology studies. Twoindependent protocols for deriving hepatocyte-like cells from hESCs and iPSCs havebeen described and the obtained cells were further characterized by immunocyto-chemistry, real-time (RT)-polymerase chain reaction, and in vitro functional assays[29]. The analysis of the hepatocyte-like cell transcriptomes revealed a broad spec-trumofmolecular cascades involving cell surface receptors, transcriptional regulators,CYPs, and associated signaling pathways known to be active during hepatogenesis.Most importantly, hESC- and iPSC-derived hepatocyte-like cells showed vast tran-scriptional similarities as well as potentially relevant differences when compared tofetal liver and adult hepatic progenitors. These findings are vital to the refinementof efficient protocols for the eventual derivation of highly functional patient-specifichepatocyte-like cells that could be suitable for either cell replacement therapies orscreens for drug toxicity.

4.3.4 hESC Applications for Other Toxicity HTS

Immortalized cell lines and live animal models are commonly used for cytotoxicityscreening of biomedical devices and materials. However, these assays poorly reflecthuman physiology and have numerous other disadvantages. An alternative may be toutilize differentiated fibroblastic progenies of hESCs for in vitro toxicity screening.Randomly differentiated hESC-derived fibroblastic progenies can be utilized as analternative to immortalized cell lines in cytotoxicity testing [30]. A colony formingassay was carried out by seeding 100, 300, and 500 cells of either the differentiatedhESC-derived fibroblastic progenies or positive control (hESC-derived mesenchy-mal stem/progenitor cells) in 6-well tissue culture dishes. After 2 weeks of culture,the positive control formed nice colonies which were positively stained by crystalviolet, while the differentiated hESC-derived fibroblastic progenies were incapableof forming any colonies with > 50 cells. Nevertheless, it must be noted that hESCsand their differentiated somatic lineages proliferate slower than the immortalized celllines commonly utilized for cytotoxicity screening. However, this limitation can beovercome by scaling-up in vitro hESC cultures and their differentiated progenies toan industrial scale, i.e., within large-volume bioreactors.

ACKNOWLEDGMENTS

Research was supported by the National Natural Science Fundation of the People’sRepublic of China (grants 91229124, 81173135 and 30973600 to Yi-jia Lou) and theOutstanding Researcher Program in Medicine & Health Foundation of the ZhejiangProvince, People’s Republic of China (grant 2011RCA024 to Xin Huang).

REFERENCES 103

REFERENCES

1. Thomson, J.A., Itskovitz-Eldor, J., Shapiro, S.S., Waknitz, M.A., Swiergiel, J.J., Marshall,V.S., Jones, J.M. (1998). Embryonic stem cell lines derived from human blastocysts.Science, 282, 1145–1147.

2. Bongso, A., Fong, C.Y., Ng, S.C., Ratnam, S. (1994). Isolation and culture of inner cellmass cells from human blastocysts. Human Reproduction, 9, 2110–2117.

3. Seiler, A., Oelgeschlager, M., Liebsch, M., Pirow, R., Riebeling, C., Tralau, T., Luch,A. (2011). Developmental toxicity testing in the 21st century: The sword of Damo-cles shattered by embryonic stem cell assays? Archives of Toxicology, 85, 1361–1372.

4. Stummann, T.C., Bremer, S. (2012). Embryonic stem cells in safety and pharmacology andtoxicology. In: New Technologies for Toxicity Testing (Balls, M., Combes, R.D., Bhogal,N., Eds.). Landes Bioscience and Springer: New York; pp. 14–25.

5. Cezar, G.G. (2007). Can human embryonic stem cells contribute to the discovery of saferand more effective drugs? Current Opinion in Chemical Biology, 11, 405–409.

6. Eglen, R., Reisine, T. (2011). Primary cells and stem cells in drug discovery: emergingtools for high-throughput screening. Assay and Drug Development Technologies, 9, 108–124.

7. Paquette, J.A., Kumpf, S.W., Streck, R.D., Thomson, J.J., Chapin, R.E., Stedman, DB.(2008). Assessment of the embryonic stem cell test and application and use in the pharma-ceutical industry. Birth Defects Research B Developmental and Reproductive Toxicology,83, 104–111.

8. Hambor, J. (2008). Breaking the stem cell technology barrier: Designing renewable,physiologically functional cells for modern drug discovery. World Stem Cell Report,2008, 133–137.

9. Lou, Y.J., Liang, X.G. (2011). Embryonic stem cell application in drug discovery. ActaPharmacologica Sinica, 32, 152–159.

10. Vliet, E.V. (2011). Current standing and future prospects for the technologies proposed totransform toxicity testing in the 21st century. ALTEX, 28, 17–44.

11. Asp, J., Steel, D., Jonsson, M., Ameen, C., Dahlenborg, K., Jeppsson, A., Lindahl, A.,Sartipy, P. (2010). Cardiomyocyte clusters derived from human embryonic stem cells sharesimilarities with human heart tissue. Journal of Molecular Cell Biology, 2, 276–283.

12. Zhang, J., Wilson, G.F., Soerens, A.G., Koonce, C.H., Yu, J., Palecek, S.P., Thomson, J.A.,Kamp, T.J. (2009). Functional cardiomyocytes derived from human induced pluripotentstem cells. Circulation Research, 104, e30–e41.

13. Steel, D., Hyllner, J., Sartipy, P. (2009). Cardiomyocytes derived from human embryonicstemcells characteristics and utility for drug discovery.Current Opinion in Drug Discovery& Development, 12, 133–140.

14. Braam, S.R., Tertoolen, L., van de Stolpe, A., Meyer, T., Passier, R., Mummery, C.L.(2010). Prediction of drug-induced cardiotoxicity using human embryonic stem cell-derived cardiomyocytes. Stem Cell Research, 4, 107–116.

15. Caspi, O., Itzhaki, I., Kehat, I., Gepstein, A., Arbel, G., Huber, I., Satin, J., Gepstein,L. (2009). In vitro electrophysiological drug testing using human embryonic stem cellderived cardiomyocytes. Stem Cells and Development, 18, 161–172.


16. Freund, C., Mummery, C.L. (2009). Prospects for pluripotent stem cell-derived cardiomy-ocytes in cardiac cell therapy and as disease models. Journal of Cellular Biochemistry,107, 592–599.

17. Peters, A.K., Wouwer, G.V., Weyn, B., Verheyen, G.R., Vanparys, P., Gompel, J.V. (2008).Automated analysis of contractility in the embryonic stem cell test, a novel approach toassess embryotoxicity. Toxicology in Vitro, 22, 1948–1956.

18. Stummann, T.C., Hareng, L., Bremer, S. (2009). Hazard assessment of methylmercurytoxicity to neuronal induction in embryogenesis using human embryonic stem cells.Toxicology, 257, 117–126.

19. Zhang, S.C., Wernig, M., Duncan, I.D., Brustle, O., Thomson, J.A. (2001). In vitrodifferentiation of transplantable neural precursors from human embryonic stem cells.Nature Biotechnology, 19, 1129–1133.

20. Morizane, A., Darsalia, V., Guloglu,M.O., Hjalt, T., Carta, M., Li, J.Y., Brundin, P. (2010).A simple method for large-scale generation of dopamine neurons from human embryonicstem cells. Journal of Neuroscience Research, 88, 3467–3478.

21. Cai, J., Zhao, Y., Liu, Y., Ye, F., Song, Z., Qin, H., Meng, S., Chen, Y., Zhou, R., Song, X.,Guo, Y., Ding, M., Deng, H. (2007). Directed differentiation of human embryonic stemcells into functional hepatic cells. Hepatology, 45, 1229–1239.

22. Agarwal, S., Holton, K.L., Lanza, R. (2008). Efficient differentiation of functional hepa-tocytes from human embryonic stem cells. Stem Cells, 26, 1117–1127.

23. Hay, D.C., Fletcher, J., Payne, C., Terrace, J.D., Gallagher, R.C., Snoeys, J., Black, J.R.,Wojtacha, D., Samuel, K., Hannoun, Z., Pryde, A., Filippi, C., Currie, I.S., Forbes, S.J.,Ross, J.A., Newsome, P.N., Iredale, J.P. (2008). Highly efficient differentiation of hESCsto functional hepatic endoderm requires ActivinA and Wnt3a signaling. Proceedings ofthe National Academy of Sciences USA, 105, 12301–12306.

24. Hay, D.C., Zhao, D., Fletcher, J., Hewitt, Z.A., McLean, D., Urruticoechea-Uriguen, A.,Black, J.R., Elcombe, C., Ross, J.A., Wolf, R., Cui, W. (2008). Efficient differentiationof hepatocytes from human embryonic stem cells exhibiting markers recapitulating liverdevelopment in vivo. Stem Cells, 26, 894–902.

25. Basma, H., Soto-Gutierrez, A., Yannam, G.R., Liu, L., Ito, R., Yamamoto, T., Ellis, E.,Carson, S.D., Sato, S., Chen, Y., Muirhead, D., Navarro-Alvarez, N., Wong, R.J., Roy-Chowdhury, J., Platt, J.L., Mercer, D.F., Miller, J.D., Strom, S.C., Kobayashi, N., Fox,I.J. (2009). Differentiation and transplantation of human embryonic stem cell-derivedhepatocytes. Gastroenterology, 136, 990–999.

26. Baharvand, H., Hashemi, S.M., Shahsavani, M. (2008). Differentiation of human embry-onic stem cells into functional hepatocyte-like cells in a serum-free adherent culturecondition. Differentiation, 76, 465–477.

27. Brolen, G., Sivertsson, L., Bjorquist, P., Eriksson, G., Ek, M., Semb, H., Johansson, I.,Andersson, T.B., Ingelman-Sundberg, M., Heins, N. (2010). Hepatocyte-like cells derivedfrom human embryonic stem cells specifically via definitive endoderm and a progenitorstage. Journal of Biotechnology, 145, 284–294.

28. Touboul, T., Hannan, N.R., Corbineau, S., Martinez, A., Martinet, C., Branchereau,S., Mainot, S., Strick-Marchand, H., Pedersen, R., Di Santo, J., Weber, A., Vallier, L.(2010). Generation of functional hepatocytes from human embryonic stem cells underchemically defined conditions that recapitulate liver development. Hepatology, 51, 1754–1765.

REFERENCES 105

29. Jozefczuk, J., Prigione, A., Chavez, L., Adjaye, J. (2011). Comparative analysis of humanembryonic stemcell and induced pluripotent stemcell-derived hepatocyte-like cells revealscurrent drawbacks and possible strategies for improved differentiation. Stem Cells andDevelopment, 20, 1259–1275.

30. Cao, T., Lu, K., Fu, X., Heng, B.C. (2008). Differentiated fibroblastic progenies of humanembryonic stem cells for toxicology screening. Cloning & Stem Cells, 10, 1–9.

PART II

HIGH-THROUGHPUT ASSAYS TOASSESS DIFFERENT CYTOTOXICITYENDPOINTS


5HIGH-THROUGHPUT SCREENINGASSAYS FOR THE ASSESSMENT OFCYTOTOXICITY

Andrew L. Niles, Richard A. Moravec, Tracy J. Worzella,Nathan J. Evans, and Terry L. Riss

5.1 INTRODUCTION

A subtle paradigm shift occurred in the drug discovery industry sometime late in thetwentieth century when cell-based methods quietly supplanted biochemical ones asthe preferred methodology to screen for new bioactive entities. Since then, the use ofcell-based methods has grown steadily.There are two primary reasons for the adoption of cell-based screening. The first is

that cellular systems tend to more faithfully emulate complex biological systems andfacilitate more predictive responses in multicellular organisms. The second equallycompelling reason is that modern laboratory robotics, liquid handlers, and assaychemistries have advanced to the point where it is now technically feasible to routinelyconduct quality experimentation in high-density formats (384-well, 384 low-volume,and 1536-well assay plates).Although cell-based drug discovery screens can be directed at either phenotypic

endpoints or pharmacologically validated targets within cells, cytotoxicity testing isstill a necessary and near compulsory activity for fulfilling the tenants of the scientificmethod. Simply stated, regardless of the motivation for screening a compound librarywith cells, it is incumbent upon the researcher to understand the effects of any newchemical entity on cellular health.


109

110 HIGH-THROUGHPUT SCREENING ASSAYS FOR CYTOTOXICITY ASSESSMENT

5.2 BASIC CONSIDERATIONS FOR MEASUREMENTSOF CELLULAR HEALTH

Choosing the appropriate and relevant cell health endpoint is the first major con-sideration for achieving screening success. Mating technique strengths with screenobjectives, while avoiding inherent method weaknesses, can also greatly mitigatespurious results. For instance, viability assays are distinctly different from cytotoxic-ity assays in that they measure biological activities resident in cells that are necessaryfor themaintenance of cellular survival [1]. Viability assays typically produce loss-of-function readouts in which chemistries report a reduction in signal relative to controlcells after a cytotoxic event (Fig. 5.1). These data are gathered after a predeterminedexposure period as a terminal endpoint. Within the linear range of these chemistries,gathered measurements will be proportional to viable cell number irrespective ofthe length of compound exposure. This feature makes viability assays an attractivemeans to indirectly address cytotoxic effects. Loss-of-function assays, however, canbe fraught with complications by optical interferences, which may improperly reporta false decrease in viability [2].Cytotoxicity assays, however, measure activities consistent with loss of membrane

integrity [3]. These measures are typically achieved by either labeling intracellularcomponents by normally impermeant dyes or measuring enzymatic activities thathave leaked from the cytoplasm into the extracellular milieu. Enzymatic activitymeasurements associated with cytotoxicity are desirable for miniaturized formatsbecause reporting chemistries often amplify the observed response and generaterobust signal windows. These activities are susceptible, however, to normal proteindegradation that impacts enzymatic half-lives and therefore may underestimate actual

FIGURE 5.1 Viability chemistries. U937 cells were dosed for 48 h with actinomycin D.Surrogates of viability—the live cell protease, ATP, and reductase activity—were measuredby applying CellTiter-FluorTM, CellTiter-GloTM, and CellTiter-BlueTM (all from Promega),respectively. A dose-dependent decrease in signal indicates cytotoxicity.

BASIC CONSIDERATIONS FOR MEASUREMENTS OF CELLULAR HEALTH 111

cytotoxic burden in an assaywell depending uponwhen the activity is assayed relativeto the cytotoxic event [3, 4].Cytostatic effects, as defined by interferencewith normal cell division rateswithout

changes in membrane integrity, can be detected by measuring viable cell numbercompared to untreated control. Unfortunately, cytotoxicity chemistries cannot reliablymeasure these cytostatic effects because membrane integrity is maintained.The second major consideration for implementing either a viability or a cytotoxi-

city screen is the appropriateness of the cell model relative to the goals of the study.A quick checklist of questions to address should include:

1. What is the compound exposure period with cells? Shorter (24 h or less) orlonger (up to 72 h)?

2. Will culture conditions permit normal cell division over the chosen exposureperiod?

3. Will transformed cancer cell lines be used or will primary/stem cell/inducedpluripotent-derived cells be used?

4. Will culture conditions or assay chemistries artificially impact cell health inlong-term exposures?

5. Is the objective to identify compounds that may be useful in producing amechanism-of-action (MOA) toxicity or is the objective to screen for potentialcytotoxic liabilities within a library?

Although no concrete rules of guidance exist, cytotoxicity assays should be con-sidered for use in screening environments if the compound exposure period is 24 hor less, and there is a premium in detecting rare or minor degrees of cell death. Asmentioned previously, most cytotoxicity assays leverage the power of repeated enzy-matic cycles of substrate turnover, which tend to boost signals arising from impairedmembrane integrity up to multiple orders of magnitude beyond the basal backgroundof viable cells. Cytotoxicity chemistries should also be considered for implementa-tion in cases of uncharacterized natural product collections, or libraries that containa high percentage of color quenchers that reduce signals independent of changes inmembrane integrity. These reductions in cytotoxicity assay signals after compoundexposure (relative to control) instead of increases enable easy identification of non-conforming data sets and compounds [5]. The compounds may be reexamined usingorthogonal methods or the information recorded during annotation activities.Viability assays, on the other hand, offer the greatest degree of flexibility for defin-

ing the cytotoxic landscape of a compound collection because they can successfullybe employed at any point during a compound exposure. Viability assays typicallyoffer the most utility in high-throughput screening (HTS) situations where cytotoxicor cytostatic compounds cause profound differences in viable cell number after treat-ment relative to a vehicle-treated population. However, compounds that cause modesttoxicity may be hidden in the natural variation of well-to-well cell number introducedby plating technique or positional effects.


The last major consideration for implementing a specific viability or cytotoxicityscreen is the development of a plan to address end user validation. Nearly all com-mercial formulations of chemistries have been rigorously tested and developed tobe sensitive and robust. Particular regard is often given to the rate of false determi-nations that an assay produces under “standardized” conditions. In reality, however,standard conditions are a fallacy as endless combinations of end user cell types andcell densities, compound exposures, compound classes, microtiter plate plastics, liq-uid handlers, and optical readers (fluorescence filters, photomultiplier tube settings)make additional validation activities a necessity. Efforts devoted to assay optimiza-tion prior to the actual screen are well worth the additional time and result in moreconsistent and better quality data. Depending on the complexity of the model, designof experiment (DOE) testing may be of value.Althoughmost screening facilities are adept at addressing the nuances of miniatur-

izing and applying viability or cytotoxicity chemistries to their cell models, perhapsthe most overlooked and underappreciated aspect of assay validation is the choice ofmodel cytotoxins. Any individual cytotoxic phenotype is shaped by dosage, exposureperiod, and cellular susceptibility. Therefore, it is essential to validate and implementcontrol compounds to fully assess the functional robustness of any assay.Model cytotoxins can be categorized into two general classes—compounds

that produce nonspecific cytotoxicity and those that produce specific cytotoxicity(Table 5.1). Mild detergents or ionophores are useful nonspecific cytotoxic agentsbecause they disrupt the cell–membrane structure in all cell types and cause mea-surable changes in membrane integrity. Care should be exercised when applyingdetergents, however, because they may be incompatible with some biomarker activi-ties or assay chemistries.Specific cytotoxins affect different cell types with differential potencies (effective

concentration or EC50). For instance, DNA alkylator-like or microtubule assem-bly poisons are typically more toxic to rapidly dividing cells than to primary or

TABLE 5.1 Model Cytotoxins

Compound Mechanism of Action Exposure Period

Nonspecific agents Digitonin Membrane disruption <15 minSaponin Membrane disruption <15 minTriton-X Membrane disruption <15 minIonomycin Ion channel flux 0–60 min

Specific agents Staurosporine Kinase inhibition 2–6 hActinomycin D Transcription inhibitor 12–36 hNocodazole Microtubule inhibitor 12–48 hCamptothecin Topoisomerase inhibitor 12–48 h17-AAG Heat-shock protein inhibitor 12–48 hEpoxomicin Proteasome inhibitor 12–36 hTrichostatin A Histone deacetylase inhibitor 12–36 hCCCP Mitochondrial complex inhibitor 2–24 hCisplatin Alkylating-like agent 24–48 h

SINGLE-PARAMETER ASSAYS (METABOLISM- AND NONMETABOLISM-BASED) 113

contact-inhibited cancer cells [6]. Similarly, other specific intracellular processes canbe targeted to produce a desired cytotoxic phenotype that can be matched to the goalsof a screen. Last, some compounds are not toxic until metabolized by specializedliver enzymes [7].

5.3 SINGLE-PARAMETER ASSAYS(METABOLISM- AND NONMETABOLISM-BASED)

Single-parameter assays have long been the mainstay of in vitro viability and cyto-toxicity testing because they are simple to employ, cost-effective, and quantitative.Although numerous options exist for lower density formats (96-well or less), few ofthese assay chemistries are sufficiently sensitive and robust enough for HTS activities.Regarding the assay chemistries that have been validated for HTS or miniaturizedapplications, each method carries its own unique set of advantages and detractionsdepending upon the focus of the study.Although available single-parameter assay methods are typically binned into tech-

niques for measuring viability or cytotoxicity, they can be further divided basedupon whether they measure metabolic (e.g., bioreduction or ATP) or nonmetabolic(e.g., proteolytic or other enzymatic) biomarkers. This distinction can be particularlyimportant in the case of metabolic biomarkers, especially if compounds interact andmodulate aspects of bioenergetic pathways without ultimately producing a cytotoxicphenotype [8]. Conversely, nonmetabolic biomarkers tend to be less susceptible tosuch interferences, but may be less sensitive and more limited by other cell modelconditions.

5.3.1 Metabolism-Based Assays

Viable cells are a rich source of reducing capacity that is diminished or completelyablated upon contact with a cytotoxic compound [9, 10]. Therefore, a number ofchemical probes have been developed, which can be reduced in a manner propor-tional to viable cell number. Although the exact mechanism of cellular bioreductionis poorly understood, studies have shown that reducing activities may be attributableto activities contained within, but not restricted to, the endoplasmic reticulum, mito-chondria, and/or associated with enzymes near the cell surface [11]. Tetrazolium salts(MTT,MTS,XTT, andWST) are probes that offer significant utility in low-throughputapplications, but are problematic to employ in miniaturized HTS applications due torequisite dye solubilization steps, insensitivity, and interferences associated with thecolorimetric readout [12, 13].Resazurin-containing formulations, however, have been more widely used for

HTS because of their simplicity, cost-effectiveness, and utility in scaled “add-mix-measure” additions [14, 15]. When viable cells are contacted with resazurinchemistries, the dye is reduced to a highly fluorescent resorufin product that canbe measured at an excitation wavelength of 560 nm and emission of 590 nm [16].


Since the intensity of the fluorescence is directly correlated to viability and cel-lular health, cytotoxic or cytostatic compounds are identified by decreased signalwhen compared with vehicle control-treated populations. Resazurin reagents areadvantageous in that they tend to be stable for extended periods at room tem-perature, allowing for batch processing and increased flexibility with microfluidicdispensers.Resazurin assays, however, are not without limitations in HTS settings. First,

chemical interferences arising from medium adjuncts, such as dithiothreitol, mercap-toethanol, L-cysteine, and reduced glutathione, complicate or contraindicate usage[11]. Second, intrinsic reductase activity varies greatly among common cell types,which may adversely influence required cell numbers or increase the incubationperiod necessary to achieve a suitable signal window. In some circumstances, exten-sion of the incubation period might lead to reagent-related (rather than compound-related) toxicities [17]. Last, standard interferences such as autofluorescence or com-pounds that interfere with intrinsic reductase activities will also falsely score ascytotoxic hits [18].ATP assays are also considered to be metabolism-based viability assays because

ATP is recognized as the basic bioenergetic currency of viable cells. As such, ATPlevels are normally very tightly regulated and proportional to viable cell number[19, 20]. When cells die or become metabolically damaged, they lose the abilityto re-synthesize additional ATP, resulting in precipitously plummeting levels of thebiomarker. This dramatic decrease after a cytotoxic event is due to both endogenousand exogenous ATPase activities resident in the assay well environment. This loss-of-signal format makes it statistically possible to measure changes in cell number orviability due to cytostatic or cytotoxic events.Cellular ATP can be conveniently measured using various add-mix-measure for-

mulations of the “firefly” luciferase enzyme reaction. The general bioluminescentassay scheme is predicated upon supplying nonlimiting concentrations of luciferaseand luciferin to varying concentrations of the analyte ATP liberated from cells viaa lytic buffer. In practice, the most useful formulations of ATP-measuring reagentscontain thermostable, recombinant luciferase that has been selected using a directedevolution approach to impart resilience to chemical interferences [21]. In additionto improved stability of the luciferase enzyme itself, other modifications to thebuffer/lysis component markedly impact signal window and impart a persistent,glow-type luminescent signal that substantially improves flexibility and utility inHTS environments [22].Themeasurement of ATP for the assessment of changes in viability has become the

gold standard for cell-based HTS activities due in part to the method’s unparalleledsensitivity [23]. Although relative ATP levels also greatly vary amongst differentcell types, the assay is useful in a broad linear range and can easily be miniaturizedinto 1536-well formats with exceptional signal windows and low variability, therebyproducing high Z-factors. An additional feature is that bioluminescent measures areexempt from traditional interferences associated with intrinsic compound autofluo-rescence [24]. Although reagent costs for employing the assay are initially somewhat


more expensive than other methods, data quality, robustness, and simplicity typicallyoffset this consideration and make this method the single most attractive choice forscreening.All enzymatic assays are subject to potential interferences. Therefore, nonspecific

luciferase inhibition in an ATP assay is possible, but is statistically rare and greatlydependent upon the robustness of individual commercial formulations [25]. A largerconsideration for employment of ATP assays is the management of thermal gradients[1]. Because luminescence is temperature dependent, caremust be taken to equilibrateboth test culture plates and reagent to room temperature prior to the measurement ofluminescence. Failure to normalize temperature may lead to variation associated withreaction kinetics. Although sufficient mixing is typically achieved during delivery ofthe reagent, three-dimensional cell models like microtissue spheroids may requireadditional time or mechanical dispersion techniques for signal intensity to stabilizerelative to standard monolayer or suspension cultures (unpublished data and personalcommunications).

5.3.2 Nonmetabolism-Based Assays

Measures of membrane integrity are a classic and well-validated means of assessingcell health. Therefore, differential permeability is routinely employed for either theassessment of viability or cytotoxicity in HTS environments. The operating premise isthat either intact cellularmembranes selectively exclude formulations of impermeabledyes or enzymatic activities associated with either viable or nonviable cells can bemeasured in an assay well after a cytotoxic insult.For instance, a recently identified and described fluorogenic probe (Gly-Phe-

AFC) can be applied to viable cells where it can be proteolytically processed aftera brief incubation to produce a fluorescent aminofluorocoumarin (AFC) productthat can be measured at an excitation of 380–400 nm with an emission of 505 nm[26]. Although the absolute identity of the protease(s) responsible for this cleavageevent is currently unknown, the activities have been shown to be constitutive andconserved in all mammalian cells tested and not influenced by any signaling orregulatory event other than those that produce changes in membrane integrity. Thisprobe, ostensibly directed at intrinsic “housekeeping” activities (“live cell protease”),is useful for defining changes in membrane integrity because the protease activitymeasured is extremely unstable in extracellular environments causing only viablecells to contribute to the signal in the assay.A principal advantage of applying this chemistry over other viability assays is

that this reagent is nonlytic and nontoxic within the exposure periods necessaryto produce useable signal windows (30 min to 2 h). Furthermore, the fluorogenicprobe can be concentrated and delivered in reduced volumes to an assay well. Theseattributes are notable for downstream applications, because the chemistry is fullycompatible with other luminescent and spectrally distinct signals that can be con-veniently multiplexed in the same assay well without interference (these featureswill be explored in section 5.4 of this chapter). Despite the numerous advantages,


this method is susceptible to standard interferences associated with fluorescencetechniques and to compounds that inhibit proteolytic activities. Moreover, the probehas also shown limited permeability and utility in three-dimensional tumor spheroidcell models.Biomarkers of cytotoxicity can be measured in the extracellular space and corre-

lated with changes in membrane integrity using fluorescent and luminescent methods.Optimally, these biomarkers should be highly enzymatically active and stable in thesample after the cytotoxic event. These gain-of-signal assays produce signals pro-portional to the degree of cytotoxicity in the assay well and are often amplified bysubstrate turnover cycling in the detection chemistry.A prominent example of this assay technique leverages the measurement of lactate

dehydrogenase (LDH) as the biomarker by coupling its activity with a diaphoraseenzyme in the presence of a reducible resazurin substrate [27]. Nonviable cellsproduce a robust increase in the highly fluorescent resorufin product from LDHactivity leaked into the medium, whereas viable cells contribute a near negligiblesignal. Although this basic reaction chemistry has been used extensively over theyears, recent improvements in commercial formulations have made this techniqueamenable to HTS.The LDH assay has many notable features including that it is well-validated, is

simple, is cost-effective to employ, and produces excellent signal windows. However,like all activity measurements, this method is susceptible to biomarker degradationthat may limit its utility in long-term culture conditions by leading to underestimationof actual cytotoxicity. Furthermore, the presence of LDH in serum may also produceelevated backgrounds and reduce signal to background values. Last, the chemistry isalso susceptible to standard fluorescence interferences and interferences associatedwith nonspecific reduction of the substrate.The newest examples of chemistries for measuring nascent activities associated

with cell death include those directed at a protease(s) normally compartmentalizedwithin viable cells [26]. Similar to the protease(s) described previously for the non-metabolic viability assay, these chemistries can be used to measure activity fromall mammalian cells. However, because the chemical probes have been designedto be cell impermeant, activities are restricted to protease(s) released into the cul-ture medium as a result of cell death. Once in the medium, these protease(s) cancontact and act upon the assay substrate comprised of the tripeptide alanine-alanine-phenylalanine conjugated to rhodamine 110 or aminoluciferin. These fluorogenicand luminogenic assay versions allow for options with regard to preferred detec-tion platform. The fluorogenic option utilizes the improved rhodamine 110 fluo-rophore with an excitation of 485 nm and emission of 520 nm, which obviatesmany optical interferences observed in the “blue spectrum” found in conventionalcoumarin chemistries [28]. The luminescent assay is completely unencumbered byfluorescence artifacts.This “protease-release” assay technology has been successfully applied to HTS

and has been shown to produce acceptable signal windows and low signal varia-tion [29]. Depending upon the observed degree of primary necrosis and the kineticsof apoptosis, the technique may either produce exceptional data or underestimate


profound cytotoxicity in extended cell-compound exposures. Because in vitro cyto-toxicity follows Paracelsus’ tenant of toxicity (“The dose makes the poison”), aquantitative HTS (qHTS) approach of serial dilutions in the same exposure period isoften indicated and can improve the chance for screen success [30–32].The last major nonmetabolic method for ascribing cytotoxicity in vitro is dye

excludability. This technique exploits differential staining of the intracellular compo-nents of cells based onmembrane integrity changes. Althoughmany general excludedvital stains can be employed (trypan blue, etc.) for cytotoxicity applications, dyesspecifically labeling genomic DNA are often preferred for HTS. These dyes tendto be specific for the nucleic acids associated with the major or minor grooves ofDNA from dead cells and increase their fluorescence properties 100–1000 fold uponbinding [33].Excludable DNA dyes therefore produce a gain-of-signal that is proportional to the

dead cell portion of an assay well. Many dyes have been described in the literature,which are useful for this application [34]. These dyes typically vary only in theiraffinity for DNA (relative to other nucleic acids) and in their quantum yield which isa measure of “brightness” and fluorophore efficiency.The principal advantage of using excludable DNA dyes lies in the fact that it

utilizes a nonactivity-based interaction that is not subject to protein degradation.Although DNA does become partially degraded after cytotoxicity as a result ofnuclease activity, it retains sufficient base-pair structure to enable reliable bindingand staining. Therefore, genomic DNA is often a better and longer lived surrogate ofcytotoxicity thanmost activity based enzymatic biomarkers (Fig. 5.2). Conversely, thestoichiometric nature of the DNA/dye complex limits the obtainable signal window.

24 h

24 h

,

,

,

48 h

48 h

72 h

72 h

FIGURE 5.2 DNA dye for cytotoxicity. K562 cells were dosed with staurosporine for72 h in a medium containing an asymmetric cyanine dye (CellToxTM Green from Promega).Fluorescence was measured at 24, 48, and 72 h. A dose-dependent increase in fluorescenceindicates cytotoxicity.


This compressed signal window ultimately limits the practical sensitivity of the dyesin HTS formats.

5.4 MULTIPARAMETRIC METHODS

Much has been written about the unsustainable nature of the current drug discoveryparadigm. Repeated failures to reach clinical efficacy or unexpected cytotoxicitiesgreatly contribute to the burgeoning costs of bringing a compound to market. New,more predictive screening methods are clearly needed to relieve this attrition, andmultiplexed screening offers a tangible and exciting path forward to directly addressthese inefficiencies while improving data content and quality.High content screening (HCS) is an elegant methodology used to address many of

these inefficiencies [35, 36]. Although multiple parameters can be examined with thetechnology, including cellular morphology, it is typically inaccessible to all but thelargest or most capitalized laboratories due to the staggering expense of dedicatedpersonnel and the initial investments in the imaging and data management systems.Furthermore, absolute throughput can be fairly modest, typically relegating its usageto focused libraries or secondary characterization efforts.Unlike HCS, same-well, optically averaged multiplexes (SWOAM) allow for the

collection of multiple signals from an assay well. This methodology uses common,plate-based,multimodal readers that are distinct fromHCSmethods in that they gatherthe average fluorescence or luminescence from a population of cells, rather than fromindividual cells. Practically speaking, their use represents a simpler and more cost-effective entry point into investigating multiple parameters associated with cell healthdue to improved throughput and reduced operator manipulation. In most cases, thechemistries employed are already HTS qualified for single-parameter use with onlysimple modifications with respect to reagent concentration. SWOAM formats includeeither one-step or sequential additions of the assay chemistry, depending upon theparameter and nature of the reagents [5, 37].All multiplexed assays with a cytotoxicity component are implemented for two

primary reasons. The first is to gather as much information as possible about thenature of a cytotoxic event in order to define a potential MOA. The second reasonis to normalize a variety of different types of assays to the number of viable cellsto reveal spurious data points that may impact decision making with regard to hitidentification [38, 39].

5.4.1 Examination of Primary Cytotoxicity (for MOA)

Although intrinsic potency and cellular susceptibility are critical criteria used todefine an individual cytotoxic effect, the kinetics of the cytotoxic response are oftenthe most informative and telling about the ultimate utility of the tested compound.Therefore, the period of time required for producing the cytotoxic phenotype can beof crucial importance. For instance, primary necrosis typically involves nonenergy-dependent cellular mechanisms that produce profound cytoplasmic swelling and

MULTIPARAMETRIC METHODS 119

catastrophic membrane rupture within 5 min to 2 h [40]. The energy-dependent pro-cess of apoptosis, however, proceeds through a series of morphologies that includenuclear condensation, membrane blebbing, and cellular fragmentation [41]. Interest-ingly, apoptosis may occur in as little as 1 h or require as long as 48 h, dependingupon the nature of the stimulus.Another key differentiator between primary necrosis and apoptotic phenotypes

relates to enzyme activation. Cells undergoing apoptosis typically activate a signal-ing cascade that initiates caspase activation [42]. Caspases are cysteine proteases thatassist in the systematic dismantling of the intracellular architecture of a cell. Thisactivity is measurable using either fluorescent or luminescent chemistries, and theappearance of caspase activity positively identifies apoptosis [43, 44]. Althoughnoncaspase-mediated apoptotic mechanisms have been described in the literature,their relative prevalence and significance to homeostasis remains the subject of somedegree of debate [45]. Unfortunately, there are currently no convenient, homoge-neous HTS methods to discriminate between cells that have undergone primary andsecondary necrosis, necessitating the use of fast-acting kinetics and absence of cas-pase activation as an operational definition. Nevertheless, the distinction between anecrosis- or apoptosis-inducing phenotype is pharmacologically relevant and predic-tive of future compound utility.A recently introduced suite of HTS-friendly assays combine viability and caspase

activation measurements to help define individual cytotoxic responses [46]. Bothassays utilize the Gly-Phe-AFC cell-permeable, fluorogenic probe (described in thesingle-parameter, nonmetabolic assays section 5.3.2 of this chapter) as a measureof cell viability and are paired with a luminogenic caspase chemistry. Procedurally,the reagents are applied in a sequential manner. After a defined cell exposure periodwith compounds, the Gly-Phe-AFC fluorogenic substrate is delivered to the assaywell in a physiologically inert buffer. The probe is incubated with cells at 37◦C fora period of at least 30 min (up to 2 h) to produce an accumulation of fluorescentproduct proportional to viable cell number. Fluorescence is then measured using afluorometer. Next, the luminescent caspase chemistry is applied and incubated atroom temperature for 30 min. The steady-state luminescent signal associated withinduction of caspases can then be quantified using a luminometer. The resulting datacan be used to compare viable cell number (fluorescence) to the presence or absenceof caspase activation (luminescence) in the exposure period chosen. It should benoted that caspase activity is typically transient and may not be detected in extendedcell-compound exposures depending upon the kinetics of the response. Therefore,the absence of measurable caspase activity may or may not indicate primary necrosis.For rigorous examination of MOA cytotoxic responses, a time course of 0, 4, 12, 24,48, and 72 h is suggested.Another useful iteration of the viability/caspase activation assay concept includes

a bis-AAF-R110 fluorogenic probe for the measurement of cytotoxicity [38]. Thisbis-AAF-R110 probe is admixed with the viability probe and delivered as describedabove. The probes and resulting readouts are fully compatible due to spectrally dis-tinct fluorescent properties. In other words, the viable and dead cell contributions ofboth assays are fully parsable and do not interfere with each other. In most cases of


primary necrosis, the viability and cytotoxicity measures are ratiometric and inverselyproportional to each other. However, viability and cytotoxicity measures may or maynot be ratiometric in cases of apoptosis where cell-cycle arrest greatly decreasesdivision and reduces the viable cell signal relative to cells that have proceeded tosecondary necrosis and lost membrane integrity. In long-term compound exposures,the activity-based cytotoxicity probe may have limited value due to the aforemen-tioned issues related to dead cell biomarker degradation. Initial studies indicate thatnonactivity probes (DNA dyes) may be of value in this regard and provide a per-sistent and accurate reflection of the actual cytotoxic population (unpublished data).This developing method concept may overcome an important technical limitationassociated with quantifying the dead cell population.Additional multiplexed combinations of assays can be implemented to examine

specific mechanisms leading to cytotoxicity. One particular assay example addressesmitochondrial toxicity. Although global mitochondrial toxicity is multifactorial, pri-mary insults can be assessed by measuring ATP as a surrogate for mitochondrialfunction in a defined culture system. Under normal oxygenated conditions, primarycells generate most of their ATP in the mitochondria via oxidative phosphoryla-tion [47]. Under hypoxic conditions, glycolytic mechanisms are used to meet thecell’s bioenergetic needs. These mechanisms typically work in concert to maintainhomeostatic levels of ATP. However, for ATP levels to be a predictive measure ofmitochondrial function alone, experimental measures must be taken to limit gly-colytic capacity. Marroquin et al. have shown that cells in culture can be forcedto use mitochondrial respiration to generate ATP by changing the source of sugarin the medium from glucose to galactose [47]. Galactose is readily metabolized bymitochondrial mechanisms to produce ATP; however glycolytic processes yield nonet production of ATP. Therefore, in a galactose-only environment, cells are forcedto meet their bioenergetic need for ATP from mitochondrial production rather thanglycolysis. Although ATP generation can be restricted to mitochondria under theseculture conditions, overall reductions in ATP cannot be directly linked to primarymitochondrial insults. For instance, as mentioned in the previous single-parameterATP section above, ATP is a reliable surrogate of viability. If cells lose membraneintegrity by simple mechanical or necrotic insult, ATP levels will decrease. There-fore, there must be a temporal component applied to separate primary reductions inATP from impaired mitochondrial function from treatments or compounds that causenonspecific necrosis. Therefore, a biomarker of membrane integrity is indicated toreveal these nonmitochondrial effects.The assay is conducted by dosing cells with compounds for 1–4 h. The exposure

period should be limited to this time frame to avoid detecting secondarymitochondrialinvolvement, which ultimately occurs as a result of compounds that induce apoptosis.Thirty minutes prior to the end of the exposure period, the fluorogenic cytotoxicityprobe is delivered to the assay wells and incubated at 37◦C to produce the fluorescentproduct that is proportional to the degree of cytotoxicity. After fluorescence datais collected, the ATP assay chemistry is delivered and bioluminescence measured.The two sets of data are then compared to values collected from a vehicle-treatedpopulation of cells. If conducted in a qHTS mode, mitochondrial effects can be

MULTIPARAMETRIC METHODS 121

FIGURE 5.3 Mitochondrial function multiplex. K562 cells were dosed with rotenone for90 min in serum-free, galactose-containing medium. A multiplexed mitochondrial functionassay (Mitochondrial ToxGloTM from Promega) was performed. A dose-dependent decrease inATP without a change in membrane integrity is consistent with specific mitochondrial toxicity.

distinguished from necrosis based on comparison of EC50 values for ATP and themarker for membrane integrity. If the potency is greater against ATP or there is noapparent cytotoxicity, this strongly suggests interference with mitochondrial func-tioning (Fig. 5.3). Single-concentration screens are possible but may be subject tovariation and a greater degree of false determinations.

5.4.2 Secondary Examination of Cytotoxicity (Normalization)

Single-parameter, cell-based screens are implemented to identify either the positiveor negative modulation of a biological activity relative to a control treatment. Assuch, the data are greatly influenced by cell health and cell number in each assaywell. For instance, cytotoxic compounds may produce a decline in apparent activityirrespective of actualmodulation of this activity. Similarly, variation in the cell numberdelivered to assay wells (aggregation, clumping, and settling during dispensing) mayproduce profound increases or decreases independent of the modulation event [48].Normalization to viable cell number remaining at the end of the experiment canreduce variation and mitigate unfounded interpretation of data points.Although dual genetic reporter systems (one constitutive, one responsive) have

been employed for years to normalize signal output, they require either laboriousco-transfection or the generation of stable cell lines expressing both transgenes. A fur-ther detraction from this form of normalization is that differential protein expressionor other bias can occur during compound exposures. Therefore, a simple, nongeneticform of normalization is highly desirable for screening.A new application for an existing set of chemistries is now available to conve-

niently address these screening constraints [49, 50]. The assay uses the Gly-Phe-AFCfluorogenic viability probe, which is delivered to the assay well to measure viablecell number. After the fluorescent signal is gathered, the reporter chemistry is applied


Compound number

FIGURE 5.4 Cell-based normalization by viability. A stably transfected NF�B-responsivereporter cell line (GloResponseTM HEK293 NF�B, Promega) was preincubated with com-pounds from the LOPAC1280 collection for 1 h, then dosed with TNF� agonist for 5 h. Amultiplexed viability and reporter assay chemistry (ONE-Glo + ToxTM Promega) was appliedand fluorescent and luminescent signals were measured. Gray dots represent either positive ornegative genetic reporter responses relative to a vehicle-treated control. Black circles repre-sent cell viability. Because fluorescent viability values demonstrate no appreciable deviancefrom control, increases or reductions in reporter luciferase activity cannot be attributed to cellclumping or cytotoxicity with significantly scored luminescence hits.

to measure the amount of luciferase gene product present after stimulation. A simplecomparison of the values can reveal nonspecific compound effects such as thosewhich produced cytotoxicity or otherwise influence luminescence (Fig. 5.4). Simi-larly, significant variation can be managed by ratiometric treatment of the values.

5.5 SUMMARY

Numerous options exist for implementing a high-throughput cytotoxicity screen(Table 5.2). For instance, single-parameter, metabolic, or nonmetabolic methods withfluorescent or luminescent readouts or multiparametric methods withmixed detectionplatforms capable of measuring multiple biomarkers are available. Although the mul-tiplicity of choices may seem somewhat daunting, simple considerations can be usedto match the proper assay chemistry with individual experimental needs. Ultimately,each assay type or method has its own merits with regard to ease-of-use, time-to-firstresult, assay windows, rate of false determinations, and implementation costs.Measuring the number of viable cells remaining after treatment is a relevant end-

point for cytotoxicity studies because the data can be directly compared to positiveand negative control treatments. These assays tend to be the most versatile optionbecause they can be implemented at any time during cell-compound exposures.More-over, fluorescent and luminescent detection formats are now favored over colorimetricmethods because they are best suited for high-density formats where sensitivity is

SUMMARY 123

TABLE 5.2 Summary of Assays Described

Assay Type SubtypeBiomarkerMeasured

DetectionFormat

Sensitivity(Cells/Well) Advantages

Viability Metabolism Reductase(s) Fluorescence 50 Cost-effective

ATP Luminescence 10 Most sensitive

Nonmetabolism Live cellprotease(s)

Fluorescence 10–50 Amenable tomultiplexing

Cytotoxicity Nonmetabolism LDH Fluorescence ∼200 Well-validated

Dead cellprotease(s)

Fluorescence orluminescence

10–5010

Amenable tomultiplexing

Exposed DNA Fluorescence ∼200 Stablebiomarker

SWOAM Mechanism ofaction

Live cellprotease(s) andcaspase(s) 3/7

Fluorescence andluminescence

10–50∼100

Candifferentiatenecrosis fromapoptosis

Live cellprotease(s),Dead cellprotease(s),andcaspase(s)-3/7

Fluorescence(400 nm/505 nm)Fluorescence(485 nm/520 nm)Luminescence

10–5010–50∼100

Candifferentiatenecrosis fromapoptosis andcytostasis fromcytotoxicity

Dead cellprotease(s) andATP

FluorescenceLuminescence

10–5010

Primarymitochondrialtoxicants

Normalization Live cellprotease(s) andgeneticluciferasereporter

FluorescenceLuminescence

10–50Depends oninductionstrength

Nongeneticnormalizationof responses

required. However, the use of viability chemistries for the determination of cytotoxic-ity does have tangible limitations. For instance, significant well-to-well variability incell number can arise in extended incubations (48–72 h) due to differential seeding orcell growth. This variability may mask modest cytostatic or cytotoxic effects exertedby test compounds. Furthermore, viability assays produce a loss-of-signal measurethat is susceptible to optical quenching or other chemical interferences.Cytotoxicity assay chemistries that measure the number of cells with dam-

aged/leaky membranes offer relevant endpoints when used under properly definedconditions. Unlike viability assays, cytotoxicity assays produce direct proof of acytotoxic event by measuring changes in membrane integrity or temporal enzymeactivation events that ultimately lead to a secondary necrosis phenotype. Fluorescentand luminescent formats are typically best suited for HTS activities and provide again-of-signal measure that tends to provide better signal to background ratios thanviability chemistries.


Cytotoxicity assay formats also have important drawbacks that must be fullyconsidered for successful use. For example, a majority of cytotoxicity chemistriesmeasure previously compartmentalized enzymatic activities that have been releasedinto the culture medium as the result of compromised membrane integrity. Theseenzyme activities are subject to time-dependent degradation that may lead to under-estimation of total cytotoxic burden depending upon when the cytotoxicity measureis taken. The labile nature of the biomarkers coupled with the differential kinetics ofcompound-directed cytotoxicity often dictates that multiple time points be examinedover an exposure period.Same-well, multiparametric methods offer an exciting opportunity to unite the

strengths of viability and cytotoxicity assays while directly addressing their inherentlimitations. Multiplexing assay methods can provide a simple and straightforwardmeans of understanding the principal MOA of toxicity for a compound, whetherthat be organelle specific or pathway dependent. Multiplexing also can be usedto normalize primary responses and mitigate erroneous conclusions due to overtcytotoxicity or disparate cell number.

REFERENCES

1. Niles, A., Moravec, R., Riss, T. (2009). In vitro viability and cytotoxicity testing and same-well multi-parametric combinations for high throughput screening. Current ChemicalGenomics, 3, 33–41.

2. Gakamsky, D., Dennis, R., Smith, D. (2011). Use of fluorescence lifetime technology toprovide efficient protection from false hits in screening applications. Analytical Biochem-istry, 409, 89–97.

3. Riss, T., Moravec, R. (2004). Use of multiple assay endpoints to investigate the effects ofincubation time, dose of toxin, and plating density in cell-based cytotoxicity assays. Assayand Drug Development Technology, 2, 51–62.

4. Niles, A., Moravec, R., Riss, T. (2008). Update on in vitro cytotoxicity assays for drugdevelopment. Expert Opinion in Drug Discovery, 3, 655–669.

5. Stoddart, M.J. (2010). Mammalian cell viability: methods and protocols. Methods inMolecular Biology, 70, 103–122.

6. Lotfi, C., Machado-Santelli, G. (1996). Comparative analysis of colchicines inducedmicronuclei in different cell types in vitro. Mutation Research/Fundamental and Molec-ular Mechanisms of Mutagenesis, 349, 77–83.

7. Chen, Q., Cederbaum, A. (1998). Cytotoxicity and apoptosis produced by cytochromeP450 2E1 in HepG2 cells. Molecular Pharmacology, 53, 638–648.

8. Kemp, D., Brouwer, K. (2004). Viability assessment in sandwich-cultured rat hepatocytesafter xenobiotic exposure. Toxicology In Vitro, 18, 869–877.

9. Gonzalez, R., Tarloff, J. (2001). Evaluation of hepatic subcellular fractions for AlamarBlue and MTT reductase activity. Toxicology In Vitro, 15, 257–259.

10. O’Brien, J., Wilson, I., Orton, T., Pognan, F. (2000). Investigation of the Alamar Blue(resazurin) fluorescent dye for the assessment of mammalian cell cytotoxicity. EuropeanJournal of Biochemistry, 267, 5421–5426.

REFERENCES 125

11. Berridge, M., Tan, A., McCoy, K., Wang, R. (1996). The biochemical and cellular basisof cell proliferation assays that use tetrazolium salts. Biochemica, 4, 14–19.

12. Roehm, N., Rodgers, G., Hatfield, S., Glasebrook, A.L. (1991). An improved colorimetricassay for cell proliferation and viability utilizing the tetrazolium salt XTT. Journal ofImmunological Methods, 142, 257–265.

13. Bigl, K., Schmitt, A., Meiners, I., Munch, G., Arendt, T. (2007). Comparison of results ofthe CellTiter Blue, the tetrazolium and the lactate dehydrogenase assay applied in braincells after exposure to advanced glycation end products. Toxicology In Vitro, 21, 962–971.

14. Ahmed, S., Gogal, R., Walsh, J. (1994). A new rapid and simple nonradioactive assay tomonitor and determine the proliferation of lymphocytes: an alternative to [3H] thymidineincorporation assays. Journal of Immunological Methods, 170, 211–224.

15. Shum, D., Radu, C., Kim, E., Cajuste, M., Shao, Y., Seshan, V., Djaballah, H. (2008). Ahigh density assay format for the detection of novel cytotoxic agents in large chemicallibraries. Journal of Enzymatic Inhibitors Medicinal Chemistry, 23, 931–945.

16. Alamar Biosciences Laboratory. (1996). Antibiotic and cytotoxic drug susceptibilityassays using resazurin and poising agents. US patent 5501959.

17. Squatrito, R., Connor, J., Buller, R. (1995). Comparison of a novel redox dye cell growthassay to the ATP bioluminescence assay. Gynecologic Oncology, 58, 101–105.

18. Hamid, R., Rotshteyn, Y., Rabadi, L., Parikh, R., Bullock, P. (2004). Comparison of alamarblue and MTT assays for high through-put screening. Toxicology In Vitro, 18, 703–710.

19. Crouch, S., Kozlowski, R., Slater, K., Fletcher, J. (1993). The use of ATP bioluminescenceas a measure of cell proliferation and cytotoxicity. Journal of Immunological Methods,160, 81–88.

20. Lundin, A. (1986). Estimation of biomass in growing cell lines by ATP assay.Methods inEnzymology, 133, 27–42.

21. Hall, M., Gruber, M., Hannah, R., Jennens-Cluogh, M.L., Wood, K.V. (1998). Sta-bilization of firefly luciferase using directed evolution. In: Bioluminescence andChemiluminescence—Perspectives for the 21st Century (Roda, A., Pazzagli, M., Kricka,L., Stanley, P.E., Eds). John Wiley & Sons: Chichester, UK; pp. 392–395.

22. Promega Corporation. (2006). Method for detection of ATP. US Patent 7083911.

23. Severson, W., Shindo, N., Sosa, M., Fletcher, 3rd, T., White, E.L., Ananthan, S., Jonsson,C.B. (2007). Development and validation of a high-throughput screen for inhibitors ofSARS CoV and its application in screening of a 100,000 compound library. Journal ofBiomolecular Screening, 12, 33–40.

24. Fan, F., Wood, K. (2007). Bioluminescent assays for high-throughput screening. Assayand Drug Development Technology, 5, 127–136.

25. Kashem, M., Nelson, R., Yingling, J., Pullen, S.S., Prokopowicz, 3rd, A.S., Jones, J.W.,Wolak, J.P., Rogers, G.R., Morelock, M.M., Snow, R.J., Homon, C.A., Jakes, S. (2007).Three mechanistically distinct kinase assays compared: measurement of intrinsic ATPaseactivity identified the most comprehensive set of ITK inhibitors. Journal of BiomolecularScreening, 12, 70–83.

26. Niles, A., Moravec, R., Hesselbeth, P., Scurria, M.A., Daily, W.J., Riss, T.L. (2007).A homogeneous assay to measure live and dead cells in the same sample by detectingdifferent protease markers. Analytical Biochemistry, 366, 197–206.

27. Riss, T., Moravec, R. (2002). Introducing the CytoTox-ONE homogeneous membraneintegrity assay. Cell Notes, 4, 6–9.


28. Grant, S., Sklar, J., Cummings, R. (2002). Development of novel assays for proteolyticenzymes using rhodamine-based fluorogenic substrates. Journal of Biomolecular Screen-ing, 7, 531–540.

29. Cho, M-H., Niles, A., Huang, R., Inglese, J., Austin, C.P., Riss, T., Xia, M. (2008). Abioluminescent cytotoxicity assay for assessment ofmembrane integrity using a proteolyticbiomarker. Toxicology In Vitro, 22, 1099–1106.

30. Rozman, K., Doull, J. (2001). Paracelsus, Haber and Arndt. Toxicology, 160, 191–196.

31. Inglese, J., Auld, D., Jadhav, A., Johnson, R.L., Simeonov, A., Yasgar, A., Zheng, W.,Austin, C.P. (2006). Quantitative high-throughput screening: a titration-based approachthat efficiently identifies biological activities in large chemical libraries. Proceedings ofthe National Academy of Sciences USA, 103, 11473–11478.

32. Xia, M., Huang, R., Witt, K., Southall, N., Fostel, J., Cho, M.H., Jadhav, A., Smith, C.S.,Inglese, J., Portier, C.J., Tice, R.R., Austin, C.P. (2008). Compound cytotoxicity profilingusing quantitative high-throughput screening. Environmental Health Perspectives, 116,284–291.

33. Rye, H.S., Yue, S.,Wemmer, D.E., Quesada,M.A., Haugland, R.P., Mathies, R.A., Glazer,A.N. (1992). Stable fluorescent complexes of double-stranded DNAwith bis-intercalatingasymmetric cyanine dyes: properties and applications. Nucleic Acids Research, 20, 2803–2812.

34. Myers, M. (1998). Direct measurement of cell numbers in microtitre plate cultures usingthe fluorescent dye SYBR green I. Journal of Immunological Methods, 212, 99–103.

35. Giuliano, K., Haskins, J., Taylor, D. (2003). Advances in high content screening for drugdiscovery. Assay Drug Development Technology, 1, 565–577.

36. Giuliano, K.A., DeBiasio, R.L., Dunlay, R.T, Gough, A., Volosky, J.M., Zock, J., Pavlakis,G.N., Taylor, D.L. (1997). High-content screening: a new approach to easing key bottle-necks in the drug discovery process. Journal of Biomolecular Screening, 2, 249–259.

37. Worzella, T., Busch, M., Niles, A. (2008). High throughput automation of multiplexedcell viability assays for viability and cytotoxicity. Cell Notes, 20, 26–29.

38. Niles, A., Worzella, T., Scurria, M., Daily, W., Bernad, L., Guthmiller, P., McNamara, B.,Rashka, K., Lange, D., Riss, T.L. (2006). Multiplexed viability, cytotoxicity and apoptosisassays for cell-based screening. Cell Notes, 16, 12–15.

39. Niles, A., Scurria, M., Bernad, L., McNamara, B., Rashka, K., Lange, D., Guthmiller,P., Riss, T. (2007). Using protease biomarkers to measure viability and cytotoxicity. CellNotes, 19, 16–20.

40. Fink, S., Cookson, B. (2005). Apoptosis, pyroptosis, and necrosis: mechanistic descriptionof dead and dying eukaryotic cells. Infection and Immunity, 73, 1907–1916.

41. Wyllie, A., Kerr, J., Currie, A. (1980). Cell death: the significance of apoptosis. Interna-tional Review of Cytolology, 68, 251–306.

42. Majno, G., Joris, I. (1995). Apoptosis, oncosis, and necrosis: an overview of cell death.American Journal of Pathology, 146, 3–15.

43. Niles, A., Moravec, R., Riss, T. (2008). Caspase activity assays. Methods in MolecularBiology, 414, 137–150.

44. O’Brien, M., Daily, W., Hesselberth, E., Moravec, R.A., Scurria, M.A., Klaubert, D.H.,Bulleit, R.F.,Wood, K.V. (2005). Homogeneous, bioluminescent protease assays: caspase-3 as a model. Journal of Biomolecular Screening, 10, 137–148.

REFERENCES 127

45. Leist, M., Jaattela, M. (2001). Four deaths and a funeral: from caspases to alternativemechanisms. Nature Reviews, 2, 1–10.

46. Schultz, S. (2009). Determining the predictive mechanism of toxicity using a single-wellmultiplexed assay. http://www.promega.com/resources/articles/pubhub/determining-the-predictive-mechanism-of-toxicity-with-apotox-glo/

47. Marroquin, L.D., Hynes, J., Dykens, J.A., Jamieson, J.D., Will, Y. (2007). Circumventingthe Crabtree effect: replacing media glucose with galactose increases susceptibility ofHepG2 Cells to mitochondrial toxicants. Toxicological Sciences, 97, 539–547.

48. Niles, A., Scurria, M., Bernad, L., McNamara, B., Rashka, K., Lange, D., Guthmiller,P., Worzella, T., Riss, T. (2007). Measure relative numbers of live and dead cells andnormalize assay data to cell number. Cell Notes, 18, 15–20.

49. Hooper, K. (2011). Multiplexing cell-based assays: get more biologically relevant data.http://www.promega.com/resources/articles/pubhub/multiplexing-cell-based-assays-get-more-biologically-relevant-data/

50. Schagat, T., Kopish, K. (2010). Optimize transfection of cultured cells. http://www.promega.com/resources/articles/pubhub/optimize-transfection-of-cultured-cells/

6HIGH-THROUGHPUT FLOWCYTOMETRY ANALYSIS OFAPOPTOSIS

Francesca de Giorgi and Francois Ichas

Apoptosis is an active cell death process, which plays an important role in thedevelopment of multicellular organisms and in the regulation and maintenance of thecell populations in tissues, both in physiological and in pathological conditions. Inthe last two decades, the field of apoptosis has become one of themost studied subjectsof biomedical research, and the apoptotic death pathway has been characterized ingreat detail at the structural, biochemical, and genetic level. The understanding of thefundamental mechanisms involved in apoptosis and the development of experimentalapproaches able to specifically quantify the apoptotic markers in dying cells havemade it possible to define its real significance in pharmacology and toxicology [1].For many years, it was assumed that chemically induced injury and death occurredprimarily by necrosis, a not regulated form of death. With the development of alarge panel of innovative apoptosis-specific protocols, it has been possible to showthat apoptosis is the major form of chemically induced cell death and that necrosisis much rarer, occurring only in circumstances of gross cell injury [2]. Therefore,the induction of apoptotic effects when developing new drugs is highly relevant topharmacotoxicological risk assessment.In mammals, two apoptosis pathways, the intrinsic and the extrinsic pathways,

have distinct initiation phases that converge toward a common execution phase. Bothphases involve proteases known as caspases, which are subdivided into two classes—the initiator caspases 8, 9, and 10 acting at the apex of the proteolytic cascade and the


129

130 HIGH-THROUGHPUT FLOW CYTOMETRY ANALYSIS OF APOPTOSIS

effector caspases 3, 6, and 7 eventually degrading the cellular contents into apoptoticbodies, which in turn can easily be engulfed by phagocytes in vivo.In the case of the extrinsic pathway, the activation of the members of the tumor

necrosis factor receptor superfamily leads to the formation of a multiprotein complex,the death-inducing signaling complex (DISC), which activates procaspases 3 and 7via activation of the caspase 8 [3]. The intrinsic pathway is triggered in response tomultiple cellular stress reactions, including growth factor withdrawal, DNA damage,and endoplasmic reticulum stress, and is transduced by the proteins of the Bcl-2family, which mediate the release of mitochondrial proapoptotic proteins such ascytochrome c and Smac/DIABLO. This release allows the formation of anothermultiprotein complex, the apoptosome, which sequentially recruits and activates theinitiator caspase 9 and the effector caspases 3 and 7 [3]. Thus, the apoptosome and theDISC have the common function of activating their respective apical caspases to initi-ate apoptotic pathways that converge into the activation of caspases 3, 6, and 7 [3]. Thedifferent caspases recognize more or less specifically consensus sites represented by atetrapeptide terminated by an aspartate residue. The effector caspases generally targetthe DXXD sequence, which is present in several signaling and structural proteins.Rapidly after the activation of the effector caspases, the cells acquire a series of

phenotypic and molecular features, among which the most spectacular are probablythe condensation of the nucleus caused by fragmentation of DNA by deinhibitedendonucleases [4] and plasma membrane blebbing and cell shrinkage due to the acti-vation of outward potassium leaks [5]. Several less specific events, typical of dyingcells, can also be observed during caspase-dependent cell death, such as the loss of themitochondrial membrane potential and the decrease of the ATP content. Moreover,during in vitro experimentation, i.e., in the absence of phagocytes, a secondary non-specific degeneration that results in rapid cytolysis occurs. It is commonly mistakenfor necrosis, and is often referred to as secondary necrosis. Therefore, apoptotic cellsmay be underestimated, particularly in vitro, unless specific and sensitive parametersare used.Thus, one of the major challenges in the field of toxicology is the ability to

develop protocols that are able to characterize cell death, thereby distinguishingbetween apoptosis and necrosis and discriminating cell death from cell proliferationarrest. Moreover, an in vitro assay to detect the induction of apoptosis applicablefor the screening of drug libraries would require reliability, sensitivity, simplicity,accuracy while using a small number of cells, low time consumption, automation,and reproducibility.Currently, drug-induced cell death can be assessed by different high-throughput

(HT) approaches. Some of these are based on the HT detection of averaged readoutsoriginating from cell populations such as lactate dehydrogenase release, mitochon-drial dye reduction (MTT), or ATP content. These are very quick, are easy to handle,and involve basic microplate readers, but they do not provide any understanding ofthe underlying mechanisms of cell death and rather determine general viability.In contrast, flow cytometry allows a quantitative assessment of a broad range

of apoptotic features at a single-cell level. The development of automated flowcytometry platforms has made possible the use of a small number of cells grown

HIGH-THROUGHPUT FLOW CYTOMETRY ANALYSIS OF APOPTOSIS 131

in multiwell formats, thereby notably contributing to the development of repro-ducible high content assays that are ideally amenable to high-throughput screening(HTS) [6].Similarly, the development of high content imaging platforms coupled to auto-

mated image segmentation has brought sufficient throughput to single-cell imaging[7], but in the specific case of cell death assays this approach should be consideredwith caution, since it is often difficult to accurately segment apoptotic cells that tendto detach from the substrate, floating or remaining stitched to other cells. Thus, flowcytometry still remains the method of choice for the assessment of apoptosis at thesingle-cell level.Measuring apoptosis by flow cytometry comprises “historical methods” based on

the assessment of “downstream” caspase-dependent cell modifications, though theselate events are often not very specific and shared by other pathways of cytotoxicity.For instance, intercalating agents that drastically enhance their fluorescence yieldupon interacting with DNA are used. Depending on the assay setup, they are usedto titrate the cells that underwent DNA fragmentation (cells with fragmented DNAfalling in the well-known “sub G1” subpopulation) or, more simply, to detect thecytolyzed cells, i.e., cells with plasma membrane permeability defects making themfreely permeable for the probe. A variety of cationic fluorescent potentiometric probes(rhodamines, carbocyanines) that are drained by healthy polarized mitochondria areused to determine the mitochondrial transmembrane potential loss associated with themitochondrial failure occurring during cell death. The exposure of phosphatidylserineat the outer leaflet of the plasma membrane that is observed during apoptosis canbe monitored using fluorescently labeled annexin V. However, if caspase-dependentapoptosis does eventually trigger a change of all these readouts, a variety of otherphysiopathological conditions can affect them as well.Thus, a series of new flow cytometry assays have been developed to detect apop-

tosis more specifically and accurately. Several of them involving cell fixation andsubsequent immunofluorescence (IF) processing for the quantification of intracel-lular targets will not be covered here. Though probing very specific steps of theapoptosis signaling cascades, they are not easily amenable to HT flow cytometry.Indeed, the several centrifugation/washing steps that are inherent to IF processing ofcell suspensions cause a cell loss that especially affects the apoptotic cells and that isvery difficult to keep under strict control. This eventually leads to a strong reductionof the Z′ factor [8] that is not compatible with an HTS context.Further assays more amenable to HTS involve either chemical probes or recom-

binant biosensors that can be loaded into or artificially expressed in living cells.Whether chemical probes or recombinant biosensors, these probes are all meant totitrate activated caspases or to detect caspase activity in situ. For this purpose, they areall designed around a specific consensus tetra- or tripeptide sequence that drives theirinteraction with the target active caspase and that can serve as a cleavable substratein the case activity is to be measured.A first “chemical” approach consisted in designing “autoquenched” cell-

permeable caspase substrates endowed with a DEVD consensus peptide for thedetection of effector caspases. These probes, consisting of two rhodamines brought


into close contact by a DEVD linker, increase their total fluorescence uponcleavage of the DEVD sequence, which relieves the autoquenching of the rhodamines(PhiPhiLuxTM) [9]. However, the intracellular persistence of the cleaved productswas often not enough to confer sufficient signal stability to process large experi-mental series. At the present time, the most widespread caspase probes are FLICAs(fluorochrome-labeled inhibitors of caspases) [10]. FLICAs are carboxyfluorescein(FAM)- or fluorescein (FITC)-tagged peptide-fluoromethyl ketone (FMK) inhibitorsof active caspases. Ideally, the tri- or tetrapeptide moiety is meant to drive the inter-action with the active center of the caspase, the reactivity of the FMK then causingirreversible covalent binding. As a matter of fact, exposure of live cells to FLICAsresults in the rapid intracellular diffusion of these probes followed by their covalentbinding to activated caspases in apoptotic cells. However, despite good signal stabilityand selectivity for apoptotic cells, a significant part of the observed FLICAs’ covalentbinding appears to take place at unknown intracellular sites different from caspases[11]. Two further fluorogenic caspase substrate probes reminiscent of PhiPhiLuxTM

that do not need washing steps are MagicRedTM and CellEventTM/NucViewTM. Inboth probes that initially are nonfluorescent, DEVD cleavage restores either thefluorescence of cresyl violet (extinguished by two DEVD substituents) or the DNAintercalating fluorescence of the nonpeptidic moiety of the probe (previously impededby the tetrapeptide).In an HTS context, the approaches based on chemical probes thus exhibit some

distinct limitations: they may exhibit a very good signal-to-noise ratio, but needwashing steps, and are not necessarily selective for activated caspases, or they rely oncleavable fluorogenic substrates and as such do not need washing steps, but exhibit alower dynamic range and in some cases a weak signal stability. It is important to pointout that washing steps may lead to artifactual cell loss during the sample preparationand that this problem may be associated with a probe loading variability especiallyin the late phase of apoptosis when loss of plasma membrane integrity occurs. Lastbut not least, the cost of the probe can be a problem in the context of large campaignsof HT primary screening.Thus, several recombinant biosensors of caspase activity based on fluorescent

proteins have been developed to circumvent the main limitations of the chemicalprobes, but only some of them can be implemented in HT flow cytometry. The firstone was a fluorescence resonance energy transfer (FRET)-based sensor developed tomonitor caspase activity in living cells. It consists of a couple of green fluorescentprotein (GFP) variants fused together with a peptide linker containing one or severalcaspase-cleavage sites. The proximity of the two fluorescent proteins results in intenseFRET, whereas the cleavage of this linker by caspases eliminates FRET because ofphysical separation of the two fluorescent moieties [12, 13]. This FRET probe hasbeen mostly used in “proof-of-concept” imaging experiments and recently in a testscreening campaign using a high content screening platform [14]. Even though it hasbeen shown that caspase activity measurement in living cells with this probe is alsoachievable by flow cytometry [15], the FRET signal produced by fluorescent proteinsis faint and variable, the assay’s Z’ factor collapsing beyond reasonable limits, andthus precluding a routine industrial use.


Another strategy is to use positional probes, which give place to a signal aftermodifying their intracellular localization as a consequence of a proteolytic event.One example is a GFP-coupled substrate bearing a nuclear localization signal (NLS)located at the C-terminus signal and a nuclear export signal (NES) connected bya linker that contains the DEVD motif at the N-terminal end (pCaspase3-sensorvector, Clontech, BD Biosciences, San Jose, CA). The NES sequence dominatesover the C-terminal NLS; thus, the probe is normally cytosolic, but once caspase-3 is activated, the DEVD domain is cleaved, releasing the fluorescent protein stillbearing the C-terminal NLS that in turn is transported into the nucleus. This kind ofprobe needs a simple imaging setup and is easily multiplexable with other probes,because the signal is detectable in a single fluorescence channel. Initially conceivedfor dynamic experiments in living cells, in principle this approach could be adaptedto segmentation-based HT experiments using HCS imaging platforms [16], but is notsuitable for flow cytometry applications.Because of this need, we have developed in our laboratory an alternative approach

to perform apoptosis HT assays with a new genetically encoded probe detecting theactivity of the effector caspases by flow cytometry. This probe was developed bytaking into account the principle of a differential anchorage inside the cell, whichrepresents a generic strategy tomeasure intracellular proteolytic activities: differentialanchorage probes (DAPs) are fluorescent proteins artificially anchored to a subcellularcompartment that can be specifically released into the cytosol by the protease ofinterest [6, 17]. As for the positional biosensors described before, the protease activityinduces a change in the intracellular distribution of the fluorescent protein due to a lossof anchorage that also renders the fluorescent signal sensitive to the permeabilizationof the plasma membrane, allowing the probe to be selectively washed out only inthose cells, in which proteolytic activity occurred. To obtain a DAP monitoring theeffector caspase activity in cells (caspaDAP), we fused the cDNA coding for a GFPto that of a mitochondrial outer membrane (mom) tail-anchored domain [6] througha synthetic linker containing the consensus site for effector caspases 3 and 7, DEVD(Fig. 6.1). This chimeric construct is localized at the mitochondrial surface (Fig. 6.1),with the fluorescent moiety exposed toward the cytosol.Besides the fact that this approach is a good technical option to monitor apoptosis

by imaging (Fig. 6.1), the permeabilization step with digitonin, converting the spatialinformation (anchored/not anchored) into a simple intensity signal (fluorescent/notfluorescent), makes out of caspaDAP a simple tool to evaluate caspase 3/7 activity byflow cytometry. In a population of cells stably expressing the probe, the percentageof nonfluorescent cells after plasma membrane permeabilization provides a directreadout of the percentage of cells in which DEVDase activity has been triggered atsome point (Fig. 6.1).We generated a panel of stably transfected cell clones expressing caspaDAP from

several different human tumor cell lines including adherent and nonadherent cells:HeLa (cervix carcinoma), HCT116 and HT29 (colon carcinoma), DU145, DU145RC0.1, PC-3, and LNCaP (prostate carcinoma), HuH7 (hepatoma), MCF7 (breastcarcinoma), SHSY-5Y (neuroblastoma), HL60 (promyelocytic leukemia), as well asHepG2 (hepatocarcinoma). The experimental protocol for caspaDAP detection is


Cyt c

CaspaDAP Red Overlay

125’ 128’ 131’ 134’

FIGURE 6.1 A differential anchorage probe for effector caspase activity (CaspaDAP). (a)Schematic representation of caspaDAPwith its cleavable DEVDmotif; (b) Time lapse imagingof HeLa cells stably expressing GFP-cytochrome c (Cyt c-GFP) (kind gift of Dr. D.R. Green),transiently transfected with caspaDAP Red (the fluorescent protein is Hcred-tamdem fromEvrogen) and incubated with 1 �M staurosporine (STS); (c) Cytochrome c immunofluores-cence in HeLa cells transiently transfected with Red caspaDAP and treated 6 h with 1�MSTS.Both experiments show the correlation between cytochrome c release and redistribution of cas-paDAP; (d) Flow cytometric profiles of caspaDAP Green (the fluorescent protein is TurboGFPfrom Evrogen) after plasma membrane permeabilization of HeLa cells stably expressing theconstruct, either untreated or treated with 30 nM or 3 �M STS for 24 h. (See insert for colorrepresentation of this figure.)

very similar for all cell lines, essentially differing by the introduction of a detachmentstep in the case of adherent cells. Briefly, cells were simply detached from the plate,resuspended in a saline solution containing digitonin, and immediately measuredby flow cytometry. The optimization of the procedure for each specific cell lineessentially concerned the detachment protocol and the concentration of digitoninnecessary to perform a controlled cell membrane permeabilization.


Due to the very simple protocol for sample preparation, the caspaDAP assay canbe completely managed by an automated platform [6]. This offers several obviousadvantages in toxicological studies aswell as for drug discovery, compared to the othertechniques available for the evaluation of apoptosis. This new approach is quantitativeenough to allow a pharmacological characterization of the tested compounds witha clear determination of the affinity (EC50) and efficacy (maximal response) in thedifferent cell types, thereby specifically detecting DEVDase activity at the single-celllevel.Figure 6.2 shows the dose–response curves characterizing a reference proapop-

totic agent, staurosporine, obtained either by measuring the mitochondrial membranepotential loss with tetramethylrhodamine, methyl ester (TMRM, a cell-permeant,cationic, red-orange fluorescent dye that is readily sequestered by active mitochon-dria) in HeLa cells or by measuring the DEVDase activity in the corresponding stablyexpressing caspaDAP HeLa cell clone. As a control, we exposed the cells to carbonylcyanide m-chlorophenyl hydrazone (CCCP), which is a protonophore that directlydepolarizes the mitochondria without causing caspase activation. The results showthat caspaDAP is specific toward caspase-dependent apoptosis.Beyond specificity issues, the presence of an endogenously expressed probe has

the following advantages: (i) it allows to avoid any probe loading steps, which wouldhave to be dealt with in the HT routine and which could cause artifactual cytotoxicity,and/or decrease the robustness of the assay; (ii) since the probe is present from thevery beginning of the experiment, it allows to “snapshot” the occurrence of caspaseactivity irrespective of the moment at which it emerges, and of whether or not theactivity is maintained in time.The simple permeabilization step causes no signal drift during the measurement.

Furthermore, we found out that after permeabilization it is even possible to postponethe measurements by simply storing the plates at 4◦C, offering a significant advantagefor a simplified scheduling of large screening campaigns. In Figure 6.2, we havesuperposed the dose–response functions obtained in the same experiment measuredjust after the permeabilization step or analyzed after storing the plates for 3 days at4◦C (without medium change or fixation).Drug-induced apoptosis in the hepatocarcinoma cell line HepG2 cells can also

be analyzed by flow cytometry. The liver is a key target organ for drug toxicity, andan important issue in drug discovery deals with the identification of molecules withhepatotoxic potential. Since it is now recognized that apoptosis in the liver plays acentral role in the toxicity of many xenobiotics, the detection of the apoptotic poten-tial is of major interest when developing new pharmaceuticals [2, 18]. Using ourcaspaDAP HepG2 cell line, we assayed the induction of caspase-dependent apop-tosis after challenging the cells with several drugs—hepatotoxic or safe—in doserange experiments multiplexing caspaDAP, cell proliferation (cell count), and mito-chondrial failure (assessment of mitochondrial membrane potential). In Figure 6.3,we show the dose–response curves obtained for compounds known to be hepato-toxic, i.e., (i) amiodarone, an antiarrhythmic agent, (ii) troglitazone, a withdrawn oralantidiabetic agent associated with idiosyncratic hepatotoxicity; (iii) acetaminophen,a compound widely used for the treatment of pain and fever that causes severe

FIG

UR

E6.

2QuantificationofapoptosisbyflowcytometryusingcaspaDAP.Shownarethedose–responsecurvesforstaurosporine(STS,24h)-induced

apoptosismeasuredbyflowcytometryusingFLICA(a),TMRM(b),andcaspaDAPGreen(d)(inastablytransfectedcloneofHeLacells);thetwo

caspaDAPcurvesrepresentthesameexperimentmeasuredeitherattheendoftheSTStreatment(d0)orafter3daysofplatestorageat4◦C(d3).Moreover,

dose–responsecurvesforchemicallyinducedmitochondrialdepolarization(1

�Mcarbonylcyanide

m-chlorophenylhydrazone[CCCP],1h)measuredwith

TMRM(c)orwithcaspaDAP(e)arepresented.FLICA,fluorochrome-labeledinhibitorsofcaspases;TMRM,tetramethylrhodamine,methylester.

136


FIGURE 6.3 Cytotoxic profiles of hepatotoxic drugs in HepG2 cells expressing caspaDAP.Shown are the dose–response curves for DEVDase activity (caspaDAP), proliferation (cellcount), and mitochondrial depolarization (TMRM) in a HepG2 cell clone stably expressingcaspaDAP after 24 h of treatment with amiodarone, acetaminophen, troglitazone, tacrine, orketorolac as negative control. TMRM, tetramethylrhodamine, methyl ester.


liver injury following overdosing; (iv) tacrine, a cholinesterase inhibitor effective inthe treatment of Alzheimer’s disease. As a negative control we used Ketorolac, anonsteroidal anti-inflammatory drug that unlike other members of its family is nothepatotoxic.A comparison of the different dose–response curves reveals that in HepG2 cells

DEVDase activity is triggered by hepatotoxic compounds and very well correlateswith the other parameters, especially in terms of relative EC50 values. Moreover,for each compound, a characteristic and distinguishable dose–response curve wasobtained. There is, however, a notable exception: tacrine at the highest concentra-tions tested is selectively detected with the mitochondrial membrane potential probe.Interestingly, tacrine toxicity was found to be associated with oxidative stress involv-ing the increased formation of reactive oxygen species [19] that are known to be ableto cause upstream mitochondrial depolarization in stress-induced apoptosis [20].CaspaDAP is suited to perform multiplexed assays combining the simultaneous

measurement of DEVDase activity with that of other key cellular parameters suchas cell proliferation, cell cycle progression, and loss of plasma membrane integrity(cytolysis) [6], allowing to distinguish between selective individual “phenotypes” ofcytotoxic effects such as proliferation arrest, apoptosis, or cytolysis/necrosis. Deter-mination of dose–response curves in such multiplexed assays may give a hint as tothe gross mechanism of action of anticancer compounds.We achieved a considerable automation of the multiplexed assay “apoptosis-

proliferation,” based on caspaDAP and the cell proliferation probe DiI [6], andof the “apoptosis-cytolysis” assay, in which the detection of caspaDAP is coupledto ethidium monoazide bromide (EMA), a fluorescent DNA intercalating agent thatonly penetrates into dead cells with a compromised cell membrane integrity andis photochemically fixable to DNA [6]. With such settings and by using a 96-wellformat, the screening throughput is of 200 test conditions per hour integrating 2000single-cell measurements per test condition.In Figure 6.4, we show an example of DAP application (a combination of two

individual multiplexed assays) to characterize the cytotoxic effects of 10 drugs afterexposing cells for 24 or 48 h to the individual compounds. The assay yielded aZ′ factor of 0.837 for the 24-h exposure series and a Z′ factor of 0.909 for a 48-hexposure. The results obtained for the 10 test compounds with this strategy wereintegrated in a graph, in which the EC50 values for caspase activation, proliferationinhibition, and cytolysis after 24 and 48 h of exposure are shown (Fig. 6.4). Asseen in the graph, it is possible to differentiate the cytotoxic profiles of the differentcompounds.Even after a 48-h drug exposure, arabinofuranosyl cytidine (AraC) essentially

induces an antiproliferative effect without apoptosis. In the first 24 h of treatment,5-fluorouracil (5FU) and vincristine also act mainly as antiproliferative agents, butdelayed apoptosis can be observed. In the case of 5FU, apoptosis is induced atsignificantly higher doses than those leading to cell growth arrest, while in the case ofvincristine apoptosis is induced at the same doses as those causing growth inhibition,thus suggesting that vincristine leads to growth inhibition and apoptosis by the samemolecular mechanism of action. In the case of the two topoisomerase inhibitors

FIG

UR

E6.

4Characterizationofthecytotoxiceffectsofdrugsbymultiplexedassays.(a)SchematicrepresentationofEC50values

forcaspase(DEVDase)activity,proliferation,andcytolysiscalculatedfrom

dose–responsecurvesintwomultiplexedassays(apoptosis-

proliferationandapoptosis-cytolysis)forcellstreatedfor24or48hwith10differentcompounds.(b)Dose–responsecurvesregarding

caspaseactivityandproliferationarrestforthreeanticancerdrugs(flavopiridol,paclitaxel,andmiltefosine)assessedafter24or48hof

treatment.(S

eein

sert

for

colo

rre

pres

enta

tion

ofth

isfig

ure.)

139


etoposide and camptothecin, we observed a very similar profile, with dissociateddose-dependent effects on proliferation and apoptosis, corresponding most probablyto different levels of DNA damage.It should be noticed that plasma membrane integrity loss, often referred to as sec-

ondary necrosis, is the less sensitive parameter for detecting cell death, since it takesplace late and can be negative even in the presence of apoptosis after 24 h. In contrast,the multiplexed assay allows discriminating secondary necrosis from apoptosis andis informative regarding the specific kinetics of drug effect development.Besides EC50 values, full dose–response curves in multiplexed assays performed

with cells exposed for two different periods of time to compounds allow to inte-grate more pharmacological features that can help profiling their cytotoxic effect. Forinstance, in Figure 6.4b, we compared the dose–response curves of three drugs,which exhibit different cytotoxicity profiles: (i) flavopiridol, a cyclin-dependentkinase inhibitor, shows no changes in terms of EC50 or maximal efficacy plateauat 24 or 48 h with both readouts, indicating that all its effects occur during thefirst 24 h of exposure; (ii) paclitaxel, which interferes with the normal breakdownof microtubules during cell division, shows constant EC50 values regarding growthinhibition and apoptosis at 24 and 48 h. However, there is a delayed and progressiverecruitment of cells in the apoptotic fraction between 24 and 48 h that increases theheight of the maximal efficacy plateau without change in the EC50; (iii) miltefosine,an antitumor alkylphospholipid, shows a clear distinction between the EC50 valuefor cell growth arrest (which occurs at a low concentration) and the EC50 value forapoptosis (which occurs at a high concentration). In contrast to the two previouscompounds, the latter case is suggestive of two independent mechanisms underlyingthe development of both effects. In addition, there is a strong impact of treatmentduration on the EC50 value for apoptosis, the EC50 value drastically decreasing astime of exposure increases.In conclusion, caspaDAP is a reliable biosensor that can be quantified by HT flow

cytometry, yielding a cost-effective industrial grade assay with high Z’ factor suitedfor large primary screening campaigns, and that can be incorporated into multiplexedassays for the elaboration of quantitative drug signatures.

ACKNOWLEDGMENTS

The authors thank Muriel Petit, Assia Chaibi, Mallory Meulle, Marion Zanese, andLaura Schembri for generating experimental data and Loic Cerf for project manage-ment.

REFERENCES

1. Tait, S.W., Green, D.R. (2010). Mitochondria and cell death: outer membrane permeabi-lization and beyond. Nature Reviews in Molecular Cell Biology, 11, 621–632.

2. Roberts, R.J. (1999). Apoptosis in Toxicology. Taylor & Francis Ltd: New York.

3. Denault, J.B., Salvesen, G.S. (2008). Apoptotic caspase activation and activity. Methodsin Molecular Biology, 414, 191–220.

REFERENCES 141

4. Liu, X., Li, P., Widlak, P., Zou, H., Luo, X., Garrard, W.T., Wang, X. (1998). The 40-kDa subunit of DNA fragmentation factor induces DNA fragmentation and chromatincondensation during apoptosis. Proceedings of the National Academy of Sciences USA,95, 8461–8466.

5. Wei, L., Xiao, A.Y., Jin, C., Yang, A., Lu, Z.Y., Yu, S.P. (2004). Effects of chloride andpotassium channel blockers on apoptotic cell shrinkage and apoptosis in cortical neurons.Pflugers Archiv, 448, 325–334.

6. Schembri, L., Zanese, M., Depierre-Plinet, G., Petit, M., Elkaoukabi-Chaibi, A., Tauzin,L., Florean, C., Lartigue, L., Medina, C., Rey, C., Belloc, F., Reiffers, J., Ichas, F., DeGiorgi, F. (2009). Recombinant differential anchorage probes that tower over the spatialdimension of intracellular signals for high content screening and analysis. AnalyticalChemistry, 81, 9590–9598.

7. Taylor, D.L. (2010). A personal perspective on high-content screening (HCS): from thebeginning. Journal of Biomolecular Screening, 7, 720–725.

8. Zhang, J.H., Chung, T.D., Oldenburg, K.R. (1999). A simple statistical parameter for usein evaluation and validation of high throughput screening assays. Journal of BiomolecularScreening, 4, 67–73.

9. Zapata, J.M., Takahashi, R., Salvesen, G.S., Reed, J.C. (1998). Granzyme release andcaspase activation in activated human T-lymphocytes. Journal of Biological Chemistry,273, 6916–6920.

10. Bedner, E., Smolewski, P., Amstad, P., Darzynkiewicz, Z. (2000). Activation of caspasesmeasured in situ by binding of fluorochrome-labeled inhibitors of caspases (FLICA):correlation with DNA fragmentation. Experimental Cell Research, 259, 308–313.

11. Kuzelova, K., Grebenova, D., Hrkal, Z. (2007). Labeling of apoptotic JURL-MK1 cells byfluorescent caspase-3 inhibitor FAM-DEVD-fmk occurs mainly at site(s) different fromcaspase-3 active site. Cytometry A, 71, 605–611.

12. Xu,X., Gerard,A.L., Huang, B.C., Anderson,D.C., Payan,D.G., Luo,Y. (1998). Detectionof programmed cell death using fluorescence energy transfer. Nucleic Acids Research, 26,2034–2035.

13. Jones, J., Heim, R., Hare, E., Stack, J., Pollok, B.A. (2000). Development and applicationof a GFP-FRET intracellular caspase assay for drug screening. Journal of BiomolecularScreening, 5, 307–318.

14. Joseph, J., Seervi, M., Sobhan, P.K., Retnabai, S.T. (2011). High throughput ratio imagingto profile caspase activity: potential application in multiparameter high content apoptosisanalysis and drug screening. PLoS One, 6, e20114.

15. He, L., Olson, D.P., Wu, X., Karpova, T.S., McNally, J.G., Lipsky, P.E. (2003). A flowcytometric method to detect protein-protein interaction in living cells by directly visu-alizing donor fluorophore quenching during CFP→YFP fluorescence resonance energytransfer (FRET). Cytometry A, 55, 71–85.

16. Giuliano, K.A., Chen, Y.T., Haskins, J.R. (2003). Fluorescent protein biosensors: a newscreening tool moves drug targets out of the test tube and into the cell. Modern DrugDiscovery, 33–37.

17. Schembri, L., Dalibart, R., Tomasello, F., Legembre, P., Ichas, F., De Giorgi, F. (2007).The HA tag is cleaved and loses immunoreactivity during apoptosis. Nature Methods, 4,107–108.

18. O’Brien, P.J., Irwin, W., Diaz, D., Howard-Cofield, E., Krejsa, C.M., Slaughter, M.R.,Gao, B., Kaludercic, N., Angeline, A., Bernardi, P., Brain, P., Hougham, C. (2006). High


concordance of drug-induced human hepatotoxicity with in vitro cytotoxicity measuredin a novel cell-based model using high content screening. Archives of Toxicology, 80,580–604.

19. Osseni, R.A., Debbasch, C., Christen, M.O., Rat, P., Warnet, J.M. (1999). Tacrine-inducedreactive oxygen species in a human liver cell line: the role of anethole dithiolethione as ascavenger. Toxicology In Vitro, 13, 683–688.

20. Tomasello, F., Messina, A., Lartigue, L., Schembri, L., Medina, C., Reina, S., Thoraval,D., Crouzet, M., Ichas, F., De Pinto, V., De Giorgi, F. (2009). Outer membrane VDAC1controls permeability transition of the inner mitochondrial membrane in cellulo duringstress-induced apoptosis. Cell Research, 19, 1363–1376.

7HIGH CONTENT IMAGING-BASEDSCREENING FOR CELLULARTOXICITY PATHWAYS

Bram Herpers and Bob van de Water

7.1 INTRODUCTION

Time-resolved imaging of living cells is essential to understand the complexity of themolecular mechanisms and pathway interactions that underlie the cellular response totoxicants. Classical biochemical cell toxicity readouts test mitochondrial respiration(MTT and MTS assays) [1, 2] and cellular ATP contents (ATPlite) [3] or measurethe leakage of the enzyme lactate dehydrogenase upon cell death [4]. Additionally,counting the number of cells that survive upon treatment, with or without intercalatingDNA-staining dyes (DAPI or Hoechst 33342 for living cells, propidium iodide [PI]for necrotic cells), can be used in high-throughput applications (e.g., fluorescenceactivated cell sorting [FACS]) to verify compound toxicity [5]. These assays areextensively used in high-throughput screening (HTS), with a throughput of tensof thousands of compounds per day. However, the mechanistic information on themolecular programs that underlie the cytotoxic response that is observed in theseassays is limited.Direct depletion of the cellular ATP levels and cell membrane leakage and/or

rupture are hallmarks of necrosis. The programmed cell death response, apopto-sis, depends on the continued availability of energy stores to execute caspase- andnuclease-mediated cleavage of cellular contents. Necroptosis or programmed necrosishas recently been described as a third type of cell death response that can be exe-cuted in the absence of caspases [6, 7]. The activation of these cell death responses


143

144 IMAGING-BASED SCREENING FOR CELLULAR TOXICITY PATHWAYS

can be instantaneous, can develop gradually, or can be triggered only under spe-cial circumstances. The cell’s internal physiology, genetic, and biochemical makeuppredetermine the ultimate response, which is influenced by preexisting conditions,e.g., growth factor depletion, decrease of nutrients, and presence of reactive oxygenspecies (ROS) [8]. The classical cell death measurements capture a single momentin time (mostly 24 h after exposure) and therefore lack information on prior eventsthat promoted the cytotoxic response. To understand these events in a time-resolvedmanner, image-based techniques are extremely illustrative.Fluorescence-based imaging has fundamentally settled itself in a wide variety

of analytical, diagnostic, surgical, and fundamental research-related tools [9, 10].There is a wide range of fluorescent molecular dyes (e.g., Hoechst [nuclei], DRAQ5[nuclei and lipids], acridine-orange [DNA/RNA], monochlorobimane [mBCl,glutathione levels], BODIPY665/676 [lipid peroxidation], PI or SYTOX dyes[nuclei of necrotic cells], TMRM [mitochondrial membrane potential], Fluo-4 AM[cytosolic calcium], CM-H2DCFDA [ROS], cyanine dimers such as TOTO-3 andYOYO-1 [plasma membrane integrity], and CM-DiI [cell membrane]) [11–14],which can be combined with direct or indirect immunofluorescence and a widevariety of fluorescent (cyan fluorescent protein [CFP], green [GFP], yellow [YFP]and red [RFP/mCherry/dsRed])-fusion proteins.This toolbox has opened a windowfor the characterization of almost if not all intracellular structures, nucleic acids,and proteins present within and between cells, or being formed, produced, reshaped,and relocalized upon stimulation. These location changes are often indicative of theintracellular response that is induced, and as such, these can be used to documentthe stress responses by (unknown) toxicants. Here we provide examples of howto use these cellular properties to study the toxicity of (novel) compounds andto functionally characterize genes by RNA interference (RNAi), thereby testingwhether they are involved in toxicant-induced cell stress and cell death responses.Additionally, we describe how to design setups for image-based screening.

7.2 AUTOMATED IMAGING: PREDEFINED UNBIASEDMICROSCOPY FIT FOR HCS

HTS by microscopy assays or high content screening (HCS) is nowadays stronglyintegrated in toxicity screening strategies [13, 15, 16]. In the last decade, multiplemanufacturersmarketedmicroscope systems that vary fromuser-friendly plate-readertype of imagers to sophisticated high-end systems. The systems can be roughlycategorized into four groups of increasing complexity: wide-field imagers for fixedcell samples, (spinning disk) confocal imagers to scale up the detail of the fixedimages, imagers equipped with temperature and CO2 control, and confocal systemsadapted for HCS and equipped with an environment chamber (Table 7.1) [15, 16].The first two types of systems can be fully integrated in an HCS setup in combinationwith robotic plate exchange systems and automated analysis software for large-scaleunattended screening. The latter two systems can be combinedwith robotic systems aswell, but in live cell imaging modus, the same plate resides in the imager for extendedperiods of time or in a repetitive manner, which vastly reduces the throughput.

AUTOMATED IMAGING: PREDEFINED UNBIASED MICROSCOPY FIT FOR HCS 145

TABLE 7.1 Overview of High Content Imagers

High Content Imagers

ImageStream Amnis Wide-Field Flow Imager

Scan∧R Olympus Wide fieldAcumen eX3 TTP LabTech Wide fieldIC200 Kinetic ImageCytometer

Vala Sciences Wide field

CellVoyager CV7000 Yokogawa Electric Corporation Wide fieldINCellAnalyzer 2000 GE Healthcare Wide field (incubator

optional)3i Marianas Intelligent Imaging Innovations Wide field (incubator

optional)ImageXpress Micro (XL) Molecular Devices Wide field (incubator

optional)Hermes WiSCAN IDEA Bio-Medical Wide field + incubatorPathway 435 Becton Dickinson Wide field + confocalCellomics ArrayScan VTI ThermoFisher Wide field + confocal

(incubator optional)Operetta Perkin-Elmer Wide field + confocal

(incubator optional)Pathway 855 Becton Dickinson Wide field + confocal +

incubatorCellVoyager CV6000 Yokogawa Electric Corporation Wide field + confocal +

incubatorImageXpress Ultra Molecular Devices ConfocalTCS SP5 Leica Microsystems ConfocalINCellAnalyzer 6000 GE Healthcare ConfocalOpera PerkinElmer Confocal + incubatorCellVoyager CV1000 Yokogawa Electric Corporation Confocal + incubatorCell Observer Carl Zeiss Confocal + incubator

By combining multiple functional and morphological markers assigned to spe-cific wavelengths, a multiparametric profile of the drug-exposed cells can be made(multiplexing) [11, 14, 17]. For example, in the case of liver toxicity, some relevantexamples are accumulation of lipid droplets, loss of membrane proteins, chromo-somal fragmentation, calcium fluxes, apoptosis induction, micronucleus formation,accumulation of ROS, loss of mitochondrial membrane potential, loss of cell polarity,activation of stress kinases, and induction of heat shock and ER stress chaperones. Forcancer and stem cell biology, analysis of the cell cycle, cell proliferation, migrationand differentiation, and cytoskeletal rearrangements are relevant. Receptor activation,intracellular transport, cell signaling, and neurite outgrowth are important in neuro-biology. Since the number of stress types already exceeds the number of discerniblewavelengths, a selection of readouts has to be made for multiplex imaging.The design of a good multivalent image-based toxicity assay starts by defining the

mechanisms of interest. HCS and multiplex imaging is followed by image segmen-tation, analysis, statistical evaluation, and hit identification [18–20]. The parameters


Hoechst 33342

Hoechst 33258

Hoechst 33342

GFP-p65

53BP1 DAPI

Annexin-V

Cell division

RTK signalingEGFR phospho-ERK

NF-κB oscillation

DNA damageγ-H2AX

Apoptosis

0 6 12 18 24 30 36 42 48 54

11410810296908478726660

0 30 60 90 120 150 180

0 30 60 90 120 150 180 210

8 12 18 24

210

END-POINT HCS ASSAYS 147

that can be extracted from one image can be hundreds per imaged channel, andthis number increases when considering colocalization, 3D reconstruction, and time-series analysis. Not all parameters will be equally informative and/or discriminative,but equally essential to explore in the assay design phase.To automate image analysis, many HCS imaging systems come with their own

software packages for image analysis, but there are also third-party vendors andopen source efforts that allow further exploration of the image information contents[15, 21]. The software provided by the HCS systems often contains all the necessaryalgorithms for segmentation of images required for specific assays, unsupervisedimage analysis, and statistics. The basic principle of some of these preprogrammedHCS packages is described below.

7.3 END-POINT HCS ASSAYS

One essential and also basic assay is designed to identify the individual cells thatwere imaged. Hoechst (33342 or 33258) or DAPI intercalates with DNA. Since theseare two of the few dyes that are rather uniquely excited at a low wavelength (350nm) and emit at 461 nm, these DNA-staining dyes can be taken along in almostany HCS assay for multiple reasons. First, an assay designed to identify growth-inhibiting and cell-damaging compounds based on Hoechst will provide visual andquantifiable evidence for a reduction in the number of nuclei and therefore cellsupon exposure. Counting the number of cells is possible by creating a segmentationmask (requiring thresholding, background subtraction, and water shedding) of eachindividual nucleus and counting the number of objects. As mentioned above, manyfeatures can be measured simultaneously, for instance the size and intensity of theHoechst-stained nuclei [22]. These two parameters are indicative of the compactnessand organization of the cell’s DNA. Dividing cells show a mitotic profile, markedby condensed, ellipsoid, and paired objects (Fig. 7.1, Cell division panel). On theother hand, necrotic cells show small, condensed, and round nuclei. The intensity,shape, and orientation with respect to neighboring objects are therefore indicative of

�FIGURE 7.1 Readouts for cellular changes by HCS. Cell division panel: the fluorescenceintensity of DNA intercalating dyes increases upon chromosomal condensation during mitosis(time in min); 20× magnification. RTK signaling panel: stimulation of RTK signalingactivates ERK. Here, epidermal growth factor (EGF) stimulation leads to epidermal growthfactor receptor (EGFR) uptake from the plasma membrane and subsequent accumulation ofphosphorylated ERK in the nucleus; 20× magnification. NF-�B oscillation panel: the NF-�Bsubunit RelA/p65 coupled to GFP oscillates in and out of the nucleus upon stimulation withTNF� (time in min). Not all cells respond in a synchronized fashion; 20× magnification.DNA damage panel: 2 Gray radiation leads to double-strand DNA breaks, which are populatedby phosphorylated H2AX (�H2AX) and 53BP1 to recruit the DSB repair machinery; 60×magnification. Apoptosis panel: exogenously added fluorescent Annexin-V binds to apoptoticcells that appear in time upon cell death induction (time in h); 10× magnification.


the state a cell is in. The staining with Hoechst alone can therefore already be veryinformative on the drug-induced effect [14] but has also been used in RNAi-basedhigh content screens to identify genes that modulate cell division [23].The potential of image-based drug screening expands by combining the Hoechst

staining with molecular dyes or indirect immunofluorescence (antibody-based stain-ing of fixed cells followed by a secondary antibody coupled to a fluorophore).For example, the phosphorylation of extracellular signal-regulated kinases [ERK]is indicative of receptor tyrosine kinase [RTK] activity [24, 25]. Active, phosphory-lated ERK localizes to the nucleus, which can be detected with an antibody targetingspecifically phosphorylated ERK (Fig. 7.1, RTK signaling panel). Thus, by usingthe nuclear mask, the fraction of positively ERK-stained nuclei is indicative of thestrength of the response. This same principle applies equally to other proteins andtranscription factors that increase their localization and/or activity in the nucleus uponpathway stimulation.An interesting case of transcription factor activation leading to a localization

change into the nucleus is nuclear factor-�B (NF-�B). Cell stress and inflammatoryresponses activate NF-�B to initiate a protective and proliferative transcript program[26]. NF-�B is visibly present in nonactivated cells in the cell’s cytoplasm due tosequestration by inhibitory proteins (I�Bs). Upon stimulation, the I�Bs are degradedand NF-�B translocates from the cytosol into the nucleus within 30 min (Fig. 7.1,NF-�B oscillation panel). The ratio between nuclear and cytoplasmic NF-�B isindicative of the activation of the response, which follows an oscillating pattern [27].To perform this kind of analysis, a second segmentation mask needs to be produced,which allows tomeasure nuclear and cytoplasm intensity [28]. Three basic techniquesexist to perform the cytoplasmic estimation: ring dilation by a few pixels from thenuclear mask (assuming that no nuclei touch and that each nucleus is surrounded byequally distributed cytoplasm), dilation until the cell boundaries are reached (mostlysuitable for solitary cells), and cell boundary estimation based on the topology ofnuclei and cytoplasmic intensities (i.e., Voronoi lines and varieties thereof or activeshape models) as well as on cytoplasmic intensities.Whereas ERK phosphorylation and NF-�B activation primarily depend on exoge-

nous stimuli by growth factors, cytokines, and other signaling molecules, toxicintracellular stress responses are imposed by biochemical changes. These can beinduced by environmental factors, but more often by drugs and metabolites thereof.Chemotherapeutic drugs, similar to radiation therapies, are designed to induce DNAdamage in cancer tissues [29]. A hallmark of DNA damage is the recruitment of theDNA repair machinery to the lesion sites. Two well-documented markers for DNAdamage are � -H2AX and 53BP1, which identify sites of double-strand breaks (DSBs)and activation of the DNA damage response, respectively [30–32]. The number ofpuncta upon immunolabeling for these two markers is indicative of the amount ofDNA lesions present within a nucleus (Fig. 7.1, DNA damage panel). This can bequantified by using the nucleus as an object mask, in which the � -H2AX and 53BP1dots are individually segmented. The overlap ratio between �H2AX and 53BP1is indicative of the functionality of DSB to repair signaling. Since each nucleus

TIME-LAPSE MICROSCOPY OF APOPTOSIS 149

and each of the puncta are recorded as an individual object, the obtained datasetstrongly expands.The purpose of DNA damage induction is to promote apoptosis in cancer cells,

either through activation of p53 or induction of the intrinsic apoptotic cascade. Similarto ERK and NF-�B, p53 activity can be determined by high-throughput microcopyand by quantifying the levels of (phosphorylated) p53 in the nuclei [33]. However,p53 is less prognostic for apoptosis induction in cells expressingmutant p53. To deter-mine whether the anticancer treatment actually potentiates apoptosis, the integrity ofthe mitochondria is a more reliable readout. Mitochondria are the cell’s ATP gener-ators. ATP production relies on the mitochondrial membrane potential, which is lostupon damage, thereby leading to mitochondrial swelling and loss of the interwovennetwork that mitochondria formwithin the cell. The mitochondrial dyesMitoTracker,Rhodamine 123, TMRM, TMRE, JC-1, and Mito-MPS [14, 34, 35] can be used tomonitor the biochemical and morphological alterations in mitochondria induced bytoxicants.Apoptosis induction through the intrinsic, mitochondria-mediated apoptotic cas-

cade depends on the formation of the apoptosome, a multimeric complex comprisedof cytochrome C, which has leaked from the damaged mitochondria, Apaf1, andcaspase 9 [36]. The activated caspase 9 starts to cleave other caspases that existin the cytoplasm as pro-forms, e.g., caspase 8 and caspase 3. By staining the cellswith antibodies raised against the cleaved form of caspase 3, the apoptotic responseinduction within a cell population can be quantified. In this case, segmentation ofthe individual cells is based on Hoechst-stained nuclei followed by dilation to definethe cytoplasm. The percentage of cells that stain positive for active caspase 3 isindicative of the amount of cells undergoing apoptosis. Note that this antibody-basedstaining requires multiple washing steps, which removes loosely attached and thusfully apoptotic cells.Cell stress is a transient process leading to a circumstance-dependent outcome.

If the stress response is able to eradicate the stressor, the response returns to theresting state. Irresolvable stress can ultimately lead to cell death, possibly througha series of other stress responses. To mechanistically study the process of resolvingstress and cumulative stress activity, the factor time has to be incorporated into theresponse assay.

7.4 TIME-LAPSE MICROSCOPY OF APOPTOSIS

Arrest of cell migration, loss of cell polarity, adoption of a rounded shape, frag-mentation of DNA and mitochondria, as well as loss of substrate attachment aremorphological hallmarks of cells with an active apoptotic response [36]. This pro-grammed cell death response shows marked increases in caspase activity, which canbe used in an HCS setup. The peptide-linked fluorescent molecule NucView488 istaken up by living cells and resides in the cytoplasm, but does not become visi-ble until caspases cleave the peptide sequence, leading to nuclear accumulation of


the cleaved NucView488, which starts to emit fluorescence. This caspase activity-reporting assay elegantly demonstrates the proapoptotic properties of the treatmentin living cell populations [37].As a consequence of caspase activity, a P4-type phospholipid flippase is activated,

resulting in the translocation of phosphatidylserine (PS) from the inner plasma mem-brane leaflet to the outside of the cell. In the tissue, this PS acts as a signalingmoleculefor macrophages to eradicate the apoptotic cell debris. Annexin-V is an endogenouscalcium-dependent PS-binding protein. Since only dead cells present PS at the outermembrane leaflet, external Annexin-Vwill only bind to and label apoptotic cells [38].Fluorophore-labeled Annexin-V (e.g., Annexin-V-FITC) has long been used in

combination with FACS analysis to determine the fraction of apoptotic cells in a cellpopulation. Since cells can also die through necrosis, this type of cell death quantifi-cation has successfully been combined with PI staining. Since PI is not membranepermeable, only cells with a leaky membrane (necrotic or late-stage apoptotic cells)allow PI to enter and stain the DNA. This powerful and insightful technique hasthe drawback of being an end-point method. Once collected, the cells need to beprocessed while still alive (fixation causes all PS to redistribute), and the samplingof a representative population needs to be performed, as it is not possible to measurethe time-dependent response in the same set of cells.Similar to the FACS-based Annexin-V labeling, fluorophore-coupled Annexin-V

can also be applied directly to growing cells in culture medium. In combination withmicroscopy, this allows the direct detection of appearing apoptotic cells (Fig. 7.1,Apoptosis panel).

7.5 APPLICATION OF LIVE APOPTOSIS IMAGING TO FUNCTIONALGENOMICS SCREENING

The strength of the live apoptosis assay is the incorporation of the factor time.Apoptosis is a programmed cell death response and depends on various enzymaticprocesses. The activity is influenced by a growing source of players, in part dedicatedto cell death induction by specific insults. The effect can be mild, causing apoptosisin only a small number of cells within a population, or very strong, leading tomassive cell death. The kinetics of the response (e.g., direct versus delayed onsetof the response) can mostly be influenced by the genetic makeup of the cells andthe activity of various signaling pathways. To study the pathways that influencethe execution of the death response (e.g., induced by drug exposure), RNAi can becombined with live apoptosis imaging [39].Screening siRNAs in combination with live apoptosis measurements is preceded

by three essential quality tests. First, the cell model needs to be tested for sensitivitytoward the drug or otherwise intrusive condition that is being studied. Depending onthe desired results, RNAi screening under complete killing conditions will mainlyresult in identification of death-promoting genes, whereas RNAi screening undernormal culture conditions will basically deliver genes as hits that are essential forsurvival. To study drug-specific pathways that control cell death, a condition that

TIME-LAPSE MICROSCOPY OF CELL STRESS DYNAMICS 151

kills only 20–50% of the cells is preferable. Second, the cell line of choice needsto be tested for the capacity of allowing fluorescent Annexin-V to bind. Not all celllines perform equally well in the assay (i.e., cells with high rates of endocytosistake up the fluorescent recombinant Annexin-V while alive, leading to detectionof the concentrated Annexin-V), and not all cell lines show apoptotic cells withthe same properties, thus demanding to test recombinant Annexin-V from differentsources. Annexin-V can be obtained from different suppliers and even coupled todifferent fluorophores. The third test that is required is to determine the efficiencyof knockdown by RNAi. A wide variety of transfection reagents are available fortesting, and many have been optimized for most cell lines [40].To perform RNAi in combination with live apoptosis, the sequence of the exper-

iment is as follows: first, cells are seeded in microplates suitable for imaging andtransfected with the siRNAs, whereby the cells are either added to the siRNA solution(reverse transfection) or the siRNAs are added after cell seeding. Depending on thecell system, 2 to 3 days after transfection, the assay can be performed. This requiresreplacing the medium with culture medium supplemented with the drug of inter-est and the fluorescent recombinant Annexin-V. Controls that should be consideredinclude z-VAD-fmk to essentially block all caspase activity (30 min pre-exposureand continuous presence) and nontargeting siRNAs to verify the nonintrusive effectof the transfection procedure [39].Because apoptotic cells detach from the plate surface, round up, and eventually

float in the cell culture medium, thereby leaving the focal plane of the living cells,the time-lapse imaging of apoptosis is preferably performed using epifluorescencemicroscopy, due to the larger focal plane. For normalization purposes, the transmittedlight or the differential interference contrast (DIC) channel or Hoechst-stained nucleiare also captured along with the Annexin-V. The imaging is usually performed at aregular interval of 30 or 60 min up to 24 h, but can be extended at will. After imaging,the total area or number of apoptotic cells is divided by the total area or number ofall cells to calculate the fraction of apoptotic cells per time point.This live apoptosis assay in combination with RNAi gives an insight into the genes

controlling the apoptotic process, but it can equally be used to monitor the kinetics ofapoptosis induction by drugs and drug combinations, as long as the used compoundsdo not autofluoresce in the imaging channel used. Further mechanistic insight canalso be obtained by combining the assay with different fluorescent markers, such asPI for necrosis, Hoechst for DNA fragmentation, and specific GFP reporter constructsfor stress signaling.

7.6 TIME-LAPSE MICROSCOPY OF CELL STRESS DYNAMICS

Intracellular (stress) responses can be very dynamic. Induction of stress kinases byextracellular factors can occur within seconds, whereas the downstream responsemaypersist for hours, but also only for a few minutes. End-point measurements thereforerequire delicate decision-making, accurate timing for fixation, and acceptance of thefact that the absence of a signal is either due to true absence of the response, or that


the time point selected is too early or too late to visualize the response. Althoughtime-lapse microscopy vastly reduces the throughput of image-based screening, thefunctional information that is gained is highly significant.One of the most intuitive systems deployable for time-lapse microscopy is acti-

vation of a stress response gene. Expression of a GFP-fusion construct driven by thepromoter sequence of the downstream stress response gene creates a switch-on typeof stress response readout. Similar to the appearance of apoptotic cells within a cellpopulation (see above), the amount of GFP-positive cells at a given time point isindicative of the severity of the response [41]. Time-lapse microscopy allows identi-fication of acute stressors and of delayed response inducers. Once identified, a timepoint can be chosen to be followed up in an end-point HTS system.Optimization of time-lapse microscopy for screening depends on the parameters

needed to be able to observe the response, the amount of cells that needs to bevisualized to acquire significant differences between positive and negative controls,and on the time resolution of the response. In practice, the throughput is a trade-off between these three factors. Detailed imaging of single cells takes more timeand provides data derived from a smaller population sample than low-magnificationimaging. This has implications on the interpretation of the obtained results.Individual cells perceive cell stress differently. The local concentration might

differ, the cellular redox state is different, and the circadian clock and cell cyclephase may strongly vary [42]. These factors create variation within a cell population,possibly leading to a bias in the results. For example, imaging of the actin dynamicsat the substrate interaction site requires short interval, high-resolution imaging. Forsegmentation and illustrative purposes, this is preferably performed on single cellswith a similarly organized phenotype. The major population, however, may not existwithout neighboring cells.Population-based high-throughput imaging is a powerful tool to identify less

prevalent to rare events. An important example is the identification of cell cycle-related genes in a high-resolution live cell imaging screen based on a GFP-taggedreporter cell line. The cell cycle process occurs in only a small subset of cells withinany given population at any given time, and as such, a single fixed view will notallow statements on the average cell doubling time or the exact cell cycle phase acell resides in, due to morphological similarities between the pro-metaphase and thepro-anaphase state. The Mitocheck consortium elegantly demonstrated the necessityof combining endogenous markers and live cell imaging to be able to identify novelfactors involved in cell cycle progression [43, 44].The fact that cell population imaging followed by individual cell segmentation is

essential is also demonstrated by live cell imaging of the NF-�B response. Overall, allcells will respond to the TNF�-induced NF-�B signaling when given at a sufficientlyhigh dose (>1 ng/mL). However, the activity of the response in consecutive timepoints can be very different [27, 39, 45]. This is a process that provides tissueplasticity and challenges for biologists. TNF�-induced NF-�B signaling depends onI�B� phosphorylation and degradation, which can be visualized byWestern blotting.However, whereas it is clear from live cell imaging that the nuclear translocationsignal does not happen once but multiple times within a time span of a few hours, the

HIGH CONTENT ANALYSIS 153

Western blot results are less clear in this respect. This is due to the asynchronicity ofthe response. Different response levels elicited in the initial response will influencethe subsequent activity. This process would be difficult to document without livecell imaging.The success of live cell imaging for cell stress activation lies in the ability to discern

between resting and active states between responses. Therefore, the most potentlydeployed live cell imaging systems depend on reporter GFP (or any other fluorescentprotein fusion) constructs that appear or disappear upon stress induction [41]. Dueto the time factor, further accumulation of the GFP reporter within the reportercell can be used to further discriminate between continuous promoter activationand the presence of processes that prevent further activity and negative feedbackmechanisms. These negative feedbackmechanisms are under the influence of delicateratio-dependent protein interactions, which can be perturbed by the introduction ofreporter constructs driven by a strong promoter. Therefore, approaches that introduceGFP-fusion proteins driven by the endogenous promoter elements (e.g., bacterialartificial chromosome [BAC] recombineering) increase the confidence of live cellstress assays [46–48].

7.7 HIGH CONTENT ANALYSIS

HCS has established itself firmly as a powerful analytical tool for advancementsin pharmacology, toxicology, neurology, and oncology, but it most potently con-tributes to the advancement of biology. While intentionally developed as an efficient,qualitative, and quantitative system for in-depth and complex analyses of cellularchanges to profile drugs and screen siRNAs [16], HCS actually provoked a revolu-tion in data analysis and data mining. The enormous amount of data that is derivedfrom multiparametric analyses of single cells exposed to drugs, toxicants, siRNAs,cytokines, nanoparticles, and combinations thereof [39, 49] demands an automatedway of data storage, phenotype extraction, and pattern recognition to mine the datathat are obtained by HCS [50]. This full battery of automated data acquisition, fea-ture extraction, data storage, management, mining, classification, and reporting canbe summed up with the term high content analysis (HCA).We anticipate that HCA will develop with increasing speed at increasing lev-

els of complexity to provide insight into the action of pharmacologically activecompounds; the function of single genes in the intricately interwoven biological net-works; the effect of neighboring cells on intracellular (re)actions, cell organelles,and cell fate decisions, both in (3D) space and time [51–53]. With the emergenceof more sophisticated tools to study biological responses in a more natural con-text (induced pluripotent stem cells, co-cultures, label-free analyses, micropattern-ing, microfluidics, 3D cell cultures, BAC recombineering, and advanced microscopytechniques) [41, 42, 48, 54–58], we are entering an exciting decade that will visuallyimprove the existing toolbox for drug discovery. The next revolution will lie in theautomated recognition of time-imposed morphological/expression and localizationchanges induced by biologically active compounds and to distinguish these from cell


environment-imposed alterations. Gradually, we are getting closer to understandingthe exact mechanisms that underlie (adverse) drug toxicity by creating systems thatcan monitor live cells in increasingly more human in vivo-like model systems.

REFERENCES

1. Gerlier, D., Thomasset, N. (1986). Use of MTT colorimetric assay to measure cell activa-tion. Journal of Immunological Methods, 94, 57–63.

2. Malich, G., Markovic, B., Winder, C. (1997). The sensitivity and specificity of the MTStetrazolium assay for detecting the in vitro cytotoxicity of 20 chemicals using human celllines. Toxicology, 124, 179–192.

3. Miret, S., De Groene, E.M., Klaffke, W. (2006). Comparison of in vitro assays of cellulartoxicity in the human hepatic cell line HepG2. Journal of Biomolecular Screening, 11,184–193.

4. Fernandez-Cruz, E., Escartin, P., Bootello, A., Kreisler,M., Segovia deArana, J.M. (1978).Hepatocyte damage induced by lymphocytes from patients with chronic liver diseases, asdetected by LDH release. Clinical & Experimental Immunology, 31, 436–442.

5. Darzynkiewicz, Z., Bruno, S., Del Bino, G., Gorczyca, W., Hotz, M.A., Lassota, P.,Traganos, F. (1992). Features of apoptotic cells measured by flow cytometry. Cytometry,13, 795–808.

6. Darding, M., Meier, P. (2012). IAPs: guardians of RIPK1. Cell Death & Differentiation,19, 58–66.

7. Han, J., Zhong, C.Q., Zhang, D.W. (2011). Programmed necrosis: backup to and competi-tor with apoptosis in the immune system. Nature Immunology, 12, 1143–1149.

8. Orrenius, S., Nicotera, P., Zhivotovsky, B. (2011). Cell death mechanisms and their impli-cations in toxicology. Toxicological Sciences, 119, 3–19.

9. Pittet, M.J., Weissleder, R. (2011). Intravital imaging. Cell, 147, 983–991.

10. Haney, S.A., LaPan, P., Pan, J., Zhang, J. (2006). High-content screening moves to thefront of the line. Drug Discovery Today, 11, 889–894.

11. Xu, J.J., Henstock, P.V., Dunn, M.C., Smith, A.R., Chabot, J.R., de Graaf, D. (2008).Cellular imaging predictions of clinical drug-induced liver injury. Toxicological Sciences,105, 97–105.


13. Tolosa, L., Pinto, S., Donato, M.T., Lahoz, A., Castell, J.V., O’Connor, J.E., Gomez-Lechon, M.J. (2012). Development of a multiparametric cell-based protocol to screen andclassify the hepatotoxicity potential of drugs. Toxicological Sciences, 127, 187–198.

14. O’Brien, P.J., Irwin, W., Diaz, D., Howard-Cofield, E., Krejsa, C.M., Slaughter, M.R.,Gao, B., Kaludercic, N., Angeline, A., Bernardi, P., Brain, P., Hougham, C. (2006). Highconcordance of drug-induced human hepatotoxicity with in vitro cytotoxicity measuredin a novel cell-based model using high content screening. Archives of Toxicology, 80,580–604.

15. Zanella, F., Lorens, J.B., Link, W. (2010). High content screening: seeing is believing.Trends in Biotechnology, 28, 237–245.

REFERENCES 155

16. Thomas, N. (2010). High-content screening: a decade of evolution. Journal of Biomolec-ular Screening, 15, 1–9.

17. Perlman, Z.E., Slack, M.D., Feng, Y., Mitchison, T.J., Wu, L.F., Altschuler, S.J. (2004).Multidimensional drug profiling by automated microscopy. Science, 306, 1194–1198.

18. Kummel, A., Gubler, H., Gehin, P., Beibel, M., Gabriel, D., Parker, C.N. (2010). Inte-gration of multiple readouts into the z’ factor for assay quality assessment. Journal ofBiomolecular Screening, 15, 95–101.

19. Kummel, A., Selzer, P., Beibel, M., Gubler, H., Parker, C.N., Gabriel, D. (2011). Com-parison of multivariate data analysis strategies for high-content screening. Journal ofBiomolecular Screening, 16, 338–347.

20. Zhang, J.H., Chung, T.D., Oldenburg, K.R. (1999). A simple statistical parameter for usein evaluation and validation of high throughput screening assays. Journal of BiomolecularScreening, 4, 67–73.

21. Gough, A.H., Johnston, P.A. (2007). Requirements, features, and performance of highcontent screening platforms. Methods in Molecular Biology, 356, 41–61.

22. Downey, M.J., Jeziorska, D.M., Ott, S., Tamai, T.K., Koentges, G., Vance, K.W.,Bretschneider, T. (2011). Extracting fluorescent reporter time courses of cell lineagesfrom high-throughput microscopy at low temporal resolution. PLoS One, 6, e27886.

23. Neumann, B., Walter, T., Heriche, J.K., Bulkescher, J., Erfle, H., Conrad, C., Rogers, P.,Poser, I., Held, M., Liebel, U., Cetin, C., Sieckmann, F., Pau, G., Kabbe, R., Wunsche,A., Satagopam, V., Schmitz, M.H., Chapuis, C., Gerlich, D.W., Schneider, R., Eils, R.,Huber, W., Peters, J.M., Hyman, A.A., Durbin, R., Pepperkok, R., Ellenberg, J. (2010).Phenotypic profiling of the human genome by time-lapse microscopy reveals cell divisiongenes. Nature, 464, 721–727.

24. McKay, M.M., Morrison, D.K. (2007). Integrating signals from RTKs to ERK/MAPK.Oncogene, 26, 3113–3121.

25. Caunt, C.J., Rivers, C.A., Conway-Campbell, B.L., Norman,M.R.,McArdle, C.A. (2008).Epidermal growth factor receptor and protein kinase C signaling to ERK2: spatiotemporalregulation of ERK2 by dual specificity phosphatases. Journal of Biological Chemistry,283, 6241–6252.

26. Wullaert, A., Heyninck, K., Beyaert, R. (2006). Mechanisms of crosstalk between TNF-inducedNF-�B and JNK activation in hepatocytes.Biochemical Pharmacology, 72, 1090–1101.

27. Ashall, L., Horton, C.A., Nelson, D.E., Paszek, P., Harper, C.V., Sillitoe, K., Ryan, S.,Spiller, D.G., Unitt, J.F., Broomhead, D.S., Kell, D.B., Rand, D.A., See, V., White, M.R.(2009). Pulsatile stimulation determines timing and specificity of NF-�B-dependent tran-scription. Science, 324, 242–246.

28. Nelson, G., Paraoan, L., Spiller, D.G., Wilde, G.J., Browne, M.A., Djali, P.K., Unitt, J.F.,Sullivan, E., Floettmann, E., White, M.R. (2002). Multi-parameter analysis of the kineticsof NF-�B signalling and transcription in single living cells. Journal of Cell Science, 115,1137–1148.

29. Link Jr., C.J., Burt, R.K., Bohr, V.A. (1991). Gene-specific repair of DNA damage inducedby UV irradiation and cancer chemotherapeutics. Cancer Cells, 3, 427–436.

30. Markova, E., Schultz, N., Belyaev, I.Y. (2007). Kinetics and dose-response of residual53BP1/gamma-H2AX foci: co-localization, relationship with DSB repair and clonogenicsurvival. International Journal of Radiation Biology, 83, 319–329.


31. Kuo, L.J., Yang, L.X. (2008). Gamma-H2AX - a novel biomarker for DNA double-strandbreaks. In Vivo, 22, 305–309.

32. Anderson, L., Henderson, C., Adachi, Y. (2001). Phosphorylation and rapid relocalizationof 53BP1 to nuclear foci upon DNA damage.Molecular and Cellular Biology, 21, 1719–1729.

33. Giuliano, K.A., Chen, Y.T., Taylor, D.L. (2004). High-content screening with siRNAoptimizes a cell biological approach to drug discovery: defining the role of P53 activationin the cellular response to anticancer drugs. Journal of Biomolecular Screening, 9, 557–568.

34. Chazotte, B. (2011). Labeling mitochondria with MitoTracker dyes. Cold Spring HarborProtocols, 8, 990–992.

35. Sakamuru, S., Li, X., Attene-Ramos, M.S., Huang, R., Lu, J., Shou, L., Shen, M., Tice,R.R., Austin, C.P., Xia, M. (2012). Application of a homogenous membrane potentialassay to assess mitochondrial function. Physiological Genomics, 44, 495–503.

36. Elmore, S. (2007).Apoptosis: a reviewof programmed cell death.Toxicological Pathology,35, 495–516.

37. Antczak, C., Takagi, T., Ramirez, C.N., Radu, C., Djaballah, H. (2009). Live-cell imagingof caspase activation for high-content screening. Journal of Biomolecular Screening, 14,956–969.

38. Puigvert, J.C., de Bont, H., van de Water, B., Danen, E.H. (2010). High-throughput livecell imaging of apoptosis.Current Protocols in Cell Biology, Chapter 18: Unit 18.10.1-13.

39. Fredriksson, L., Herpers, B., Benedetti, G., Matadin, Q., Puigvert, J.C., de Bont, H.,Dragovic, S., Vermeulen, N.P., Commandeur, J.N., Danen, E., de Graauw, M., van deWater, B. (2011). Diclofenac inhibits tumor necrosis factor-�-induced nuclear factor-�Bactivation causing synergistic hepatocyte apoptosis. Hepatology, 53, 2027–2041.

40. Lee, S.K., Siefert, A., Beloor, J., Fahmy, T.M., Kumar, P. (2012). Cell-specific siRNAdelivery by peptides and antibodies. Methods in Enzymology, 502, 91–122.

41. Hendriks, G., Atallah, M., Morolli, B., Calleja, F., Ras-Verloop, N., Huijskens, I., Raams-man, M., van deWater, B., Vrieling, H. (2012). The ToxTracker assay: novel GFP reportersystems that provide mechanistic insight into the genotoxic properties of chemicals. Tox-icological Sciences, 125, 285–298.

42. Bakstad, D., Adamson, A., Spiller, D.G., White, M.R. (2012). Quantitative measurementof single cell dynamics. Current Opinion in Biotechnology, 23, 103–109.

43. Neumann, B., Held, M., Liebel, U., Erfle, H., Rogers, P., Pepperkok, R., Ellenberg, J.(2006). High-throughput RNAi screening by time-lapse imaging of live human cells.Nature Methods, 3, 385–390.

44. Held, M., Schmitz, M.H., Fischer, B., Walter, T., Neumann, B., Olma, M.H., Peter, M.,Ellenberg, J., Gerlich, D.W. (2010). CellCognition: time-resolved phenotype annotationin high-throughput live cell imaging. Nature Methods, 7, 747–754.

45. Turner, D.A., Paszek, P., Woodcock, D.J., Nelson, D.E., Horton, C.A., Wang, Y., Spiller,D.G., Rand, D.A., White, M.R., Harper, C.V. (2010). Physiological levels of TNF� stim-ulation induce stochastic dynamics of NF-�B responses in single living cells. Journal ofCell Science, 123, 2834–2843.

46. Poser, I., Sarov, M., Hutchins, J.R., Heriche, J.K., Toyoda, Y., Pozniakovsky, A., Weigl,D., Nitzsche, A., Hegemann, B., Bird, A.W., Pelletier, L., Kittler, R., Hua, S., Naumann,R., Augsburg, M., Sykora, M.M., Hofemeister, H., Zhang, Y., Nasmyth, K., White, K.P.,

REFERENCES 157

Dietzel, S., Mechtler, K., Durbin, R., Stewart, A.F., Peters, J.M., Buchholz, F., Hyman,A.A. (2008). BAC TransgeneOmics: a high-throughput method for exploration of proteinfunction in mammals. Nature Methods, 5, 409–415.

47. Ciotta, G., Hofemeister, H., Maresca, M., Fu, J., Sarov, M., Anastassiadis, K., Stewart,A.F. (2011). Recombineering BAC transgenes for protein tagging.Methods, 53, 113–119.

48. Bird, A.W., Erler, A., Fu, J., Heriche, J.K., Maresca, M., Zhang, Y., Hyman, A.A., Stewart,A.F. (2012). High-efficiency counterselection recombineering for site-directed mutagen-esis in bacterial artificial chromosomes. Nature Methods, 9, 103–109.

49. Solmesky, L.J., Shuman, M., Goldsmith, M., Weil, M., Peer, D. (2011). Assessing cellulartoxicities in fibroblasts upon exposure to lipid-based nanoparticles: a high content analysisapproach. Nanotechnology, 22, 494016.

50. Stuurman, N., Swedlow, J.R. (2012). Software tools, data structures, and interfaces formicroscope imaging. Cold Spring Harbor Protocols, 2012, 50–61.

51. Murphy, R.F. (2012). CellOrganizer: image-derived models of subcellular organizationand protein distribution.Methods in Cell Biology, 110, 179–193.

52. Stacey, G. (2012). Current developments in cell culture technology. Advances in Experi-mental Medicine and Biology, 745, 1–13.

53. Lieggi, N.T., Edvardsson, A., O’Brien, P.J. (2010). Translation of novel anti-cancer cyto-toxicity biomarkers detected with high content analysis from an in vitro predictive modelto an in vivo cell model. Toxicology In Vitro, 24, 2063–2071.

54. Schwarzenberger, T., Wolf, P., Brischwein, M., Kleinhans, R., Demmel, F., Lechner, A.,Becker, B., Wolf, B. (2011). Impedance sensor technology for cell-based assays in theframework of a high-content screening system. Physiological Measurement, 32, 977–993.

55. Degot, S., Auzan, M., Chapuis, V., Beghin, A., Chadeyras, A., Nelep, C., Calvo-Munoz,M.L., Young, J., Chatelain, F., Fuchs, A. (2010). Improved visualization and quantitativeanalysis of drug effects using micropatterned cells. Journal of Visualized Experiments, 46,e2514.

56. Fiolka, R., Shao, L., Rego, E.H., Davidson, M.W., Gustafsson, M.G. (2012). Time-lapsetwo-color 3D imaging of live cells with doubled resolution using structured illumination.Proceedings of the National Academy of Sciences U.S.A., 109, 5311–5315.

57. Huang, Y., Agrawal, B., Clark, P.A., Williams, J.C., Kuo, J.S. (2011). Evaluation of cancerstem cell migration using compartmentalizing microfluidic devices and live cell imaging.Journal of Visualized Experiments, 58, e3297.

58. Li, L., Lu, Y. (2011). Optimizing a 3D culture system to study the interaction betweenepithelial breast cancer and its surrounding fibroblasts. Journal of Cancer, 2, 458–466.

8THE KERATINOSENS ASSAY: AHIGH-THROUGHPUT SCREENINGASSAY TO ASSESS CHEMICAL SKINSENSITIZATION

Andreas Natsch

8.1 SKIN SENSITIZATION AND THE NEED TO SCREENNOVEL CHEMICALS

Small, protein-reactive molecules coming into contact with the skin may modifyendogenous skin proteins to render them immunogenic [1]. Upon repeated contactwith the chemical, a T-cell-mediated immune reaction against the chemicallymodifiedskin proteins may occur. In the sensitization phase of skin sensitization, the naıveimmune system is primed for the first time by the novel epitopes formed by the reactivechemical. This process, next to formation of the covalent protein adduct, involvesthe activation of dendritic cells by the test chemical to trigger their migration intothe local lymph nodes, where they present the novel epitopes to the T-cell repertoire.In order to initiate migration of dendritic cells, a so-called, still somewhat elusive,“danger signal” is thought to be needed at this stage besides the mere formation ofnovel epitopes [2]. Once the dendritic cells presenting the novel epitopes reach thelymph nodes, clonal expansion of epitope-specific T cells may take place [1, 3, 4].Upon repeated contact with the sensitizing agent (which may occur at a much lowerconcentration), the primed T cells residing in the skin will react as an answer tothe specific epitopes, thus leading to an inflammatory response, which is the actualdisease state known as contact allergy or contact dermatitis. This second phase iscalled the elicitation phase. The elicitation phase had been the endpoint of the older


159

160 THE KERATINOSENS ASSAY

FIGURE 8.1 The different physicochemical, chemical, and biological steps involved in thesensitization phase of the skin sensitization reaction. MHC, major histocompatibility complex.

animal tests evaluating inflamed skin upon repeated application of test chemicals(e.g., the guinea pig maximization test [GPMT] [5] and Buehler test [6]), and it isalso the endpoint evaluated when performing human tests such as the diagnosticpatch tests conducted by dermatologists on sensitized individuals or the predictivehuman repeated insult patch test (HRIPT) occasionally used to assess the sensitizationpotential of individual chemicals in humans [6].Currently, the skin sensitization risk of novel chemicals is predicted by the local

lymph node assay (LLNA) in mice, which measures the proliferative response inthe draining lymph nodes of the ears of mice topically treated three times dailyon consecutive days with the test agent [7]. Hence, this assay only includes thesensitization phase and lacks the elicitation step. Similarly, most in vitro assays inadvanced development only mimic a particular aspect or step of the sensitizationphase.The different cell types and the key steps involved in the sensitization phase are

summarized in Figure 8.1. The key cell types involved are (i) the specific T cells (theeffector cells), (ii) the dendritic or more specifically the Langerhans cells (presentingthe antigens to the T cells), and (iii) the keratinocytes (the primary cells in contactwith the sensitizers and thought to contribute to danger signal formation).

1. In a first step, the molecule applied topically must reach the viable epidermis,and bioavailability in the skin is therefore thought to be a prerequisite and amodifying factor for the skin sensitization reaction [8].

THE MECHANISTIC BACKGROUND OF THE KERATINOSENS ASSAY 161

2. Certain molecules are considered prohaptens, indicating that they themselvesare nonsensitizers and need to be metabolically converted by skin enzymes tobecome the causative agents of a sensitization reaction [1].

3. A key step in the process is the formation of a covalent adduct between theskin sensitizer and endogenous proteins and/or peptides in the skin. Thus, onestraightforward approach to mimic this hallmark of the sensitization reaction isto measure reactivity of chemicals with test peptides [10]. A peptide reactivityassay coined direct peptide reactivity assay (DPRA) has been developed andentered the prevalidation process of the European Centre for the Validation ofAlternative Methods (ECVAM) [9].

4. The modified peptides then must be presented by the Langerhans cells (i.e., thedendritic cells in the skin) on their major histocompatibility complex (MHC)molecules.

5. At the same time, the danger signal needs to be induced. It appears to be partof the innate immune reaction [10] and is formed in the absence of specificityconveyed by specific T cells. Based on the work of Kimber et al. [11, 12], theformation of cytokines appears to be a central element of this danger signal.

6. In a last step, stimulated by the triggers/danger signals coming from the innatereactions, the dendritic cells maturate and migrate from the skin into the locallymph node, where they finally present the modified peptides to the T cells andstimulate the proliferation of specific T-cell clones.

Skin sensitization is an important toxicological endpoint for cosmetics and topi-cally applied drugs. The single ingredients of these preparations are applied at lowdoses in relation to the total body weight, and the risks for systemic toxic effects aretherefore usually very low. On the other hand, application is performed repeatedlyand locally. Hence, local concentrations applied to the skin may be sufficient forsome local effects to occur. As the dose per area is the critical metric parameter forthe initiation of a skin sensitization reaction [13], it is not surprising that skin sensiti-zation is probably the most often reported adverse effect regarding the application ofcosmetics. Within the palette of cosmetic ingredients, preservatives, hair dyes, andsome perfumery materials are the most frequent culprits inducing an allergic reaction,but also a wide range of natural extracts do contain sensitizing components.The Cosmetics Directive in Europe imposes a ban on skin sensitization testing in

animals from 2013 onwards. Although scientific consensus has been reached that itis not possible to provide a validated alternative assay by this deadline [14], thereare ongoing large research initiatives to come up with new testing strategies, notablythe COLIPA research program [15] and the European sixth framework programSens-It-Iv [16].

8.2 THE MECHANISTIC BACKGROUNDOF THE KERATINOSENS ASSAY

As indicated above, the only feature that skin sensitizers have in common is theirintrinsic electrophilicity or their metabolization to electrophilic compounds. Onemay


thus speculate that at the stage of danger signal formation, the cells need to somehow“sense” the electrophilic metabolites. One possible sensor mechanism has recentlybeen described: Skin sensitizers appear to have a general effect on cell-surface thiols,and this appears to activate theMAP kinase p38 [17]. In addition, a significant numberof gene chip, RT-PCR, and reporter gene studies published in the last 3 years pointto the Nrf2-Keap1-ARE regulatory pathway as the probably most common signalingpathway induced by skin sensitizers [18–20]. The evidence regarding the above-mentioned hypothesis has recently been reviewed [21, 22]. As an example, Vandebrielet al. [19] exposed human HaCaT cells to eight contact sensitizers and six irritantsand thereafter performed a gene chip analysis on the exposed cells. The analysisindicated that the pathways, in which Keap1/Nrf2-regulated genes and oxidativestress response genes were involved, were the most significantly affected pathways,the prototypic Nrf2-Keap1-regulated gene HMOX1 being the most discriminativemarker. Reporter gene analysis on 116 chemicals in the AREc32 reporter cell lineindicated that the majority of skin sensitizers induce the luciferase response undercontrol of the antioxidant response element (ARE) [23]. Some indications regardingthe involvement of this signaling pathway in contact allergy and particularly in theTH1 response in vivo have been reported [24].The sensor protein Keap1 (Kelch-like ECH-associated protein 1) contains highly

reactive cysteine residues [25]. In the uninduced state, Keap1 targets the transcriptionfactor Nrf2 (nuclear factor-erythroid 2-related factor 2) for Cul3-mediated ubiquitiny-lation and proteolytic degradation in the proteasome (for review see [26]). Covalentmodification of the reactive cysteine residues of Keap1 by small molecules triggersNrf2 accumulation in the nucleus and activates genes (mainly genes coding for phaseII detoxifying enzymes) having an ARE (also called EpRE, electrophile responseelement) in their promoter sequence [25, 27]. Thus, reactivity of a chemical towardspecific cysteine residues inKeap1 triggers the induction of a battery of genes, and thisprotein serves as a cellular “reactivity sensor.” A simplified view of the Keap1-Nrf2-ARE regulatory pathway is depicted in Figure 8.2. One particular gene containingan ARE in its promotor is the AKR1C2 gene coding for an aldo-keto reductase [28].This particular gene was identified as one of the key target genes upregulated bycontact sensitizers in dendritic cells [29, 30].

8.3 THE CREATION OF THE KERATINOSENS REPORTER CELL LINE

The following 56-base-pair element containing the ARE sequence from theAKR1C2 gene was synthesized: 5′-TGGTCGCAAGGTGTGCAAGCTGCTGAGTCACCCTGACTGCATC AACCC CAGGAGCT-3′. This element was linked to theSV40 promotor and placed upstream of the luciferase gene luc2. The resulting plas-mid was transfected into the human keratinocyte cell line HaCaT [31], and cloneswith a stable insertion were selected in the presence of Geneticin/G418. The cloneswere screened with a set of six reference chemicals (weak to strong sensitizers andmethyl salicylate as nonsensitizer). For each clone, the absolute light output and thedynamic range of luciferase induction by sensitizers were evaluated. A clone wasselected based on the following criteria: (i) best signal-to-noise ratio and (ii) highest

THE STANDARD OPERATING PROCEDURE OF THE KERATINOSENS ASSAY 163

FIGURE 8.2 The Nrf2-Keap1-ARE signaling pathway and its hypothetical activation byskin sensitizing chemicals. (Reproduced with permission from Reference 21).

dynamic range if treated with the weak sensitizers lyral and benzyl salicylate. Thisclone, coined the KeratinoSens cell line [32], was then used to develop the standardoperating procedure (SOP) of the KeratinoSens assay.

8.4 THE STANDARD OPERATING PROCEDUREOF THE KERATINOSENS ASSAY

The SOP includes a full dose–response analysis for each chemical. The cells aregrown for 24 h in 96-well plates. The medium is replaced with a medium containingthe test chemical and a final level of 1% of the solvent dimethyl sulfoxide (DMSO).


Each compound is tested at 12 binary dilutions in the range from 0.98 to 2000 �M.Each test plate contains seven test chemicals, six wells with the solvent control, onewell with no cells for background value, and five wells with the positive controlcinnamic aldehyde in five different concentrations (4–64 �M). Three parallel repli-cate plates are run with this same setup, and a fourth parallel plate is prepared forcytotoxicity determination. Cells are incubated for 48 h with the test agents, andthen luciferase activity and cytotoxicity (with the MTT assay) are determined. Thisfull procedure is repeated three times for each chemical. The validity of each sin-gle assay is determined with fixed criteria regarding the variability of the solventcontrol wells and by evaluating the dose–response analysis of the positive controlcinnamic aldehyde included in all assay plates. Figure 8.3 exemplarily shows dose–response curves obtained for the hair dye ingredient p-phenylenediamine, the naturalfragrance component cinnamic aldehyde (the main component of cinnamon oil), andthe preservative methyldibromoglutaronitrile.

8.5 DATA ANALYSIS AND PREDICTION MODEL

The data are processed automatically with an Excel template, which forms part ofthe SOP. This simplifies data analysis and ensures that data are equally processed byall testing laboratories. The following parameters are calculated automatically:

1. Gene induction relative to DMSO control in each well

2. Wells with statistically significant induction over a given threshold (defaultvalue set to 1.5 = 50% enhanced gene activity)

3. Imax, the maximal induction over the full dose response

4. The EC1.5 value, extrapolated concentration for statistically significant induc-tion above threshold (calculated in a similar way as in the LLNA)

5. IC50 and IC70 values for cytotoxicity (i.e., concentration at which cellularviability comprises 50% and 70%, respectively)

6. Whether the EC1.5 determining concentration falls in the cytotoxic range (via-bility <70%).

Based on the above calculations, chemicals can be rated as positive or negative. Achemical is positive if (i) it leads to a statistically significant induction of luciferaseactivity >1.5-fold in at least two of the three experiments, (ii) the EC1.5 value isbelow 1000 �M, and (iii) the viability at the EC1.5 determining concentration (i.e.,the concentration which yields the significant induction >1.5-fold) is >70%.Accordingly, chemicals are rated negative if (i) no significant induction is recorded,

(ii) the induction is only observed in one of three repetitions, (iii) the induction occursabove 1000 �M, or (iv) the induction only occurs at cytotoxic concentrations. (Note:If gene induction is only observed at cytotoxic concentrations, this is indicative of afalse-positive gene induction generated by a skin irritant.)

9(a

)7.

5

4.5

1.5 06 3

110

Con

cent

ratio

n (μ

m)

Fold luciferase induction

100

10000204060

% viability

80100

120

9(b

)7.

5

4.5

1.5 06 3

110

Con

cent

ratio

n (μ

m)

Fold induction10

010

000204060

% viability

80100

120

9(c

)7.

5

4.5

1.5 06 3

110

Con

cent

ratio

n (μ

m)

Fold induction

100

10000204060

% viability

80100

120

FIG

UR

E8.

3Typicaldose–responsecurvesobtainedintheKeratinoSensassayperformedaccordingtothestandardoperatingprocedure:(a)

thehairdyecomponent

p-phenylenediamine(strongsensitizer);(b)thefragranceandflavorcomponentcinnamicaldehyde,themainingredientof

cinnamonbarkoilandamoderatesensitizer;(c)thepreservativemethyldibromoglutaronitrile,amoderatesensitizer.

165


8.6 THE INTRA- AND INTERLABORATORY REPRODUCIBILITY ANDPREVALIDATION STUDIES

A key component of the (pre)validation process is to assess the robustness of an assay.Thus, it is important to prove both intra- and interlaboratory reproducibility. For thispurpose, the assay was transferred to four partner laboratories, and a two-phase studywas performed. To assess transferability, a set of seven reference chemicals wasfirst tested. Once the performance criteria were fulfilled by the testing laboratories,phase II, in which 21 blind-coded test items were evaluated, was started. Overall,the reproducibility of the assay was found to be very good, and the detailed resultshave recently been published [33]. Not only were congruent results obtained for mostchemicals when evaluating the yes/no rating of the chemicals according to the pre-diction model, but also the dose–response curves were overall highly reproducible.Expressed as a geometric standard deviation, the interlaboratory variability of theEC1.5 and IC50 values was 1.56 and 1.35, respectively. These geometric standarddeviations are close to the square root of two, which indicates that the 95.6% confi-dence interval of these determinations is at one well up and down in the microplatedilution series from the geometric mean. The intralaboratory reproducibility was fur-ther evaluated in a detailed study (A. Natsch, unpublished results), and an even lowervariability for intralaboratory repetitions was found (geometric standard deviation at1.2 for both EC1.5 and IC50 values, thus indicating that the 95.6% confidence intervalof these determinations is at 12 well up and down in dilution series from the mean).All these data were summarized, submitted to ECVAM, and accepted to directly enterthe peer-review process as a prevalidation study in 2011.

8.7 THE PREDICTIVITY FOR STANDARD LISTSOF REFERENCE CHEMICALS

There is no large published standard list of chemicals to validate alternative assaysfor the skin sensitization endpoint. However, several smaller lists have indeed beenpublished: (i) Casati et al. [34] reported a list of 16 chemicals, (ii) the Sens-it-ivconsortium published a largely overlapping list [35], and (iii) the US InteragencyCoordinating Committee on the Validation of Alternative Methods (ICCVAM) pro-posed performance standards for alternative endpoints in the LLNA [36]. We createda larger list coined the “Silver list,” which includes the chemicals of these three listsas well as added chemicals selected according to the following criteria: (i) congruentresults in the LLNA and guinea pig tests or congruent results in the LLNA and humanpredictive tests, (ii) chemicals of all important applicability domains, and (iii) chem-icals to cover a broad range of potencies. The data on this list and the three sublistswere previously published [32] and are summarized in Table 8.1.For all chemicals we routinely perform a peptide-binding assay in parallel [37].

This assay is a modification of the DPRA and includes LC–MS-based analysis ofpeptide adduct formation. The overall result of this assay (i.e., whether a chemicalforms covalent peptide adducts) is included in Table 8.1.We propose to rate chemicalspositive if they are positive in the KeratinoSens assay and/or are forming covalent

THE APPLICABILITY DOMAIN AND LIMITATIONS OF THE ASSAYS 167

TABLE 8.1 Cooper Statistics for the Different Tested Standard Lists

SILVER List, WoE:KeratinoSens

ICCVAM and/or AdductList Sens-it-iv ECVAM SILVER Forming[36] List [34] List [35] List a Positive b

Correctpositives

13 12 11 38 41

False-negatives 2 2 1 5 2Correctnegatives

7 9 4 19 19

False-positives 0 1 0 5 5Number of testchemicals

22 24 16 67 67

Sensitivity 86.7 85.7 91.7 88.4 95.3Specificity 100.0 90.0 100.0 79.2 79.2Accuracy 90.9 87.5 93.8 85.1 89.6aList compiled based on weight-of-evidence in vivo data as described in Reference 32.bChemicals rated positive if either positive in (i) the KeratinoSens assay and/or if (ii) adduct formation isobserved in peptide reactivity assay.

peptide adducts. The results obtained and the rationale of this combined approachare further discussed below.

8.8 THE APPLICABILITY DOMAIN AND LIMITATIONS OF THEASSAYS AS DETERMINED THROUGH THE SCREENING

There are twomain groups of chemicals for which limitations have been encountered:

Acyl-transfer agents—As described in Section 8.2, the Nrf2 pathway in principleinteracts with thiol-reactive chemicals. Skin sensitizers may react with a varietyof different nucleophiles in proteins [38], but since the thiol group in cysteineis the most reactive nucleophile under physiological conditions, most skinsensitizers do at least react with thiols. However, there are notable exceptionsto this rule, especially some chemicals exclusively reactingwith lysine residues.One particular exception is a group of chemicals referred to as acyl-transferagents [39], and examples of these agents are given in Table 8.2. They containan ester bond with a good leaving group (in most cases a phenyl group), andthe selective transfer of the acyl group onto lysine residues can be verified inmost cases in the LC–MS peptide-binding assay [37, 38]. Given the establishedmolecular mechanism of Nrf2 activation by thiol-modifying compounds [27],it is not surprising that the selective reactivity with amine groups renders achemical false-negative in the KeratinoSens assay. Although this group ofchemicals is probably relatively small in the chemical universe, it is significantlyrepresented in the standard lists published by ICCVAM. The result obtained forthis group of chemicals thus underlines the need to perform a peptide-bindingassay with a lysine-containing peptide in parallel to the KeratinoSens assay to


TABLE 8.2 Skin Sensitizing Acyl-Transfer Agents Negative in the KeratinoSensAssay, but Positive in the Direct Peptide-Binding Assay

Structure Name Discussion

OO

3,4-dihydrocoumarin Specific reaction with lysine, adductformation in the LC–MS assay, also inReference 38

O

O

Phenyl benzoate Reaction with lysine, adduct formation inthe LC–MS assay, also in Reference 38

O

O

O

Phthalic anhydride Selective reaction with lysine, adductformation in the LC–MS assay, specificreaction with lysine also reported inDPRA [53]

maximize sensitivity as previously discussed [21]. This group of chemicals ismainly responsible for the improved accuracy of the combined approach (Table8.1).

Prohaptens—A second class of false-negatives includes some chemicals con-sidered as prohaptens. These are mainly aromatic compounds with methoxygroups, and metabolization to catechols or quinone methides appears to be aprerequisite to render them peptide-reactive. Nevertheless, it should be high-lighted that the list of chemicals classified correctly by the KeratinoSens assayalso contains a significant number of chemicals considered to be putative pro-and prehaptens [40], especially aromatic and aliphatic amines, �,�-unsaturatedalcohols, catechols, and quinones.

False-positives—Like in any toxicological test, we also observe a number of false-positive reactions. These have previously been discussed in detail [32]. In thiscontext, one interesting case is that of the detergent Tween 80. This chemicalreproducibly induces the Nrf2 response despite the fact that it does not containany reactive structural elements. Recent data indicate that it induces a specificoxidative response in cells, which apparently can trigger the Nrf2 induction(A. Natsch, unpublished observation). Understanding this mechanism appearsimportant in order to discriminate this rare but relevant false-positive responsefrom real positives.

8.9 CASE STUDIES ON SPECIFIC CHEMICAL CLASSES

Besides the studies involving compound lists derived frompublished databases on ref-erence sensitizers, the KeratinoSens assay has recently been applied in two structure-activity studies with specific classes of chemicals.

CASE STUDIES ON SPECIFIC CHEMICAL CLASSES 169

TABLE 8.3 In Vitro-Based Read-Across Predictions for Two Epoxides

Compound

Predicted EC3 ValueBased on

Concentration forGene Induction (mM)

Predicted EC3 ValueBased on CytotoxicConcentration (mM)

Measured EC3 Value(mM)

OS

59 45 35

O

455 304 140

Different derivatives of phenyl glycidyl ether (PGE) used as diluent in epoxyresinsystems were synthesized by the group of Ann-Therese Karlberg, and detailedstructure-activity relationship (SAR) studies in the LLNA were performed [41, 42].This is perhaps the most detailed recent SAR study in the LLNA, and the samechemicals were therefore tested in the KeratinoSens assay. The quantitative relation-ship between the in vivo LLNA response and the in vitro dose–response data wereanalyzed [43], and both the luciferase-inducing dose and the cytotoxic dose corre-lated with the LLNA result. Most interestingly, the correlation results obtained withthe first series of compounds [41] could be used to make in vitro-based read-acrosspredictions for two new compounds [42], whose LLNA results were not known tothe testing laboratory at the time when the predictions were made. These predictionsare summarized in Table 8.3. At the same time, the study also showed the limitationsof such quantitative extrapolations, when used to predict the sensitization potentialof less closely related chemicals [43].In another initiative, a number of surfactants were recently investigated for their

sensitization potential in the LLNA and in the GPMT [44]. These two in vivo testsgave discordant results for most items of this chemical class (i.e., the surfactantswere negative in the GPMT but were positive in the LLNA). Since the surfactantsodium dodecyl sulfate (SDS) is widely known as a false-positive in the LLNA [45],it was suspected that the negative rating of the surfactants in the GPMT could bethe correct answer, although it is difficult to draw a final conclusion from two assayswith contradictory results. To investigate this issue in more detail, the KeratinoSenstest, the peptide-binding assay, and the human Cell Line Activation Test (h-CLAT)test [46] were performed on the same batches of chemicals previously used for theanimal tests. Indeed, all these in vitro assays gave a negative rating of the surfactants,and a weight-of-evidence analysis was presented for these chemicals based on thisvery detailed dataset combining in vitro and in vivo results [44]. In this study, therewas one conflicting result: a glucoside surfactant was rated clearly sensitizing in theGPMT and borderline in the LLNA. This preparation also yielded a borderline resultin the KeratinoSens assay. Given the absence of structural alerts in the glucoside


FIGURE 8.4 Potentially sensitizing impurity identified in a surfactant preparation bymakinguse of a KeratinoSens activity-based fractionation. Note: Position of the methyl group as inthe isononanol used for synthesis.

surfactant, the clearly positive result in the GPMT came as a surprise. Therefore,the preparation was investigated in greater detail. In an activity-guided fractionationstudy using the KeratinoSens assay, a hexane-soluble fraction, which was clearlypositive in the KeratinoSens assay, was isolated. The water-soluble residue on theother hand was completely negative in the in vitro assay. By chemical analysis anddetailed testing in the KeratinoSens assay, an impurity in the hexane fraction from thesurfactant preparation was identified and synthesized, having a very clear structuralalert (�,�-unsaturated aldehyde, Fig. 8.4). This analysis highlights another potentialuse of this and other high-throughput toxicological assays. Since the tests are fastand multiple parallel tests can be run at the same time, different fractions fromindividual preparations or different qualities can easily be compared and potentialtoxicologically active impuritiesmay be identified, aswe have also reported in anotherrecent case study [47].

8.10 THE USE OF THE ASSAY IN AN INTEGRATED TESTINGSTRATEGY (ITS) FOR A COMPLETE REPLACEMENTOF ANIMAL TESTING

There is a broad consensus in the scientific community that the complex processesinvolved in skin sensitization can only be modeled by a battery of assays (batteryapproach), and that this may particularly be true if we wish to predict skin sensitizerpotency. This thinking has theoretically been formalized in two publications [48, 49].As described in Section 8.8, in aminimal testing batterywepropose to at least combinethe KeratinoSens assay with a peptide reactivity assay eventually incorporating ametabolic component, and the rationale for this is discussed above. In the meantime,we have performed a first practical assessment of the battery approach combiningdata from a previous variant of an Nrf2-based assay and from the DPRA peptide-binding assay [23]. While we had used simple arithmetic combinations of the data,a refined approach for data integration was recently proposed [50], thereby using aBayesian net modeling approach to integrate data from different sources. In anotherstudy [51], glutathione-binding data were combined with RT-PCR-based analysis ofNrf2-dependent genes. All the above-mentioned approaches indicated that combiningpeptide reactivity with Nrf2-dependent gene activation testing is a straightforwardminimal approach for an ITS. Finally, the Laboratory of Experimental Toxicology

REFERENCES 171

and Ecology at BASF [52] has recently performed in parallel the KeratinoSens assay,peptide-binding assay, and two dendritic cell assays (hence, all four assays undergoingthe ECVAM prevalidation process) and shown how data from several different assayscan be combined to improve predictivity. As results from the different proposedassays for the screening of skin sensitizers continue to accumulate, such initiativesmay finally come up with ideal combinations of tests to get optimal predictions atreasonable testing costs. At the same time, a note of caution is warranted at thisstage. As the number of assays is growing, and test batteries combining assays areproposed, it is important to define how weight-of-evidence decisions are to be made.If a positive result from one assay is sufficient to rate chemicals positive (as donein most of the simple integrated strategies discussed above), we will be faced withtest batteries having increased sensitivity but reduced specificity, as each assay willproduce some false-positives.

REFERENCES

1. Karlberg, A.T., Bergstrom, M.A., Borje, A., Luthman, K., Nilsson, J.L. (2008). Allergiccontact dermatitis—formation, structural requirements, and reactivity of skin sensitizers.Chemical Research in Toxicology, 21, 53–69.

2. Martin, S.F., Esser, P.R., Weber, F.C., Jakob, T., Freudenberg, M.A., Schmidt, M., Goe-beler, M. (2011). Mechanisms of chemical-induced innate immunity in allergic contactdermatitis. Allergy, 66, 1152–1163.

3. Smith, C.K., Hotchkiss, S.A.M. (2001). Allergic Contact Dermatitis: Chemical andMetabolic Mechanisms. Taylor and Francis: London and New York; 310 pp.

4. Kimber, I., Basketter, D.A., Gerberick, G.F., Ryan, C.A., Dearman, R.J. (2011). Chem-ical allergy: translating biology into hazard characterization. Toxicological Sciences,120(Suppl 1), S238–S268.

5. Wahlberg, J.E., Boman, A. (1985). Guinea pig maximization test. Current Problems inDermatology, 14, 59–106.

6. McNamee, P.M., Api, A.M., Basketter, D.A., Frank Gerberick, G., Gilpin, D.A., Hall,B.M., Jowsey, I., Robinson, M.K. (2008). A review of critical factors in the conductand interpretation of the human repeat insult patch test. Regulatory and ToxicologicalPharmacology, 52, 24–34.

7. Basketter, D.A., Evans, P., Fielder, R.J., Gerberick, G.F., Dearman, R.J., Kimber, I. (2002).Local lymph node assay - Validation, conduct and use in practice. Food and ChemicalToxicology, 40, 593–598.

8. Basketter, D., Pease, C., Kasting, G., Kimber, I., Casati, S., Cronin, M., Diembeck, W.,Gerberick, F., Hadgraft, J., Hartung, T., Marty, J.P., Nikolaidis, E., Patlewicz, G., Roberts,D., Roggen, E., Rovida, C., Van De Sandt, J. (2007). Skin sensitisation and epidermaldisposition: the relevance of epidermal disposition for sensitisation hazard identificationand risk assessment: the report and recommendations of ECVAM workshop 59. ATLAAlternatives to Laboratory Animals, 35, 137–154.

9. Gerberick, G.F., Vassallo, J.D., Bailey, R.E., Chaney, J.G., Morrall, S.W., Lepoittevin,J.P. (2004). Development of a peptide reactivity assay for screening contact allergens.Toxicological Sciences, 81, 332–343.


10. Freudenberg, M.A., Esser, P.R., Jakob, T., Galanos, C., Martin, S.F. (2009). Innate andadaptive immune responses in contact dermatitis: analogy with infections. Giornale Ital-iano di Dermatologia e Venereologia, 144, 173–185.

11. Dearman, R.J., Cumberbatch, M., Kimber, I. (2003). Cutaneous cytokine expression:induction by chemical allergen and paracrine regulation. Journal of Toxicology—Cutaneous and Ocular Toxicology, 22, 69–86.

12. Kimber, I., Cumberbatch, M., Dearman, R.J., Bhushan, M., Griffiths, C.E.M. (2000).Cytokines and chemokines in the initiation and regulation of epidermal Langerhans cellmobilization. British Journal of Dermatology, 142, 401–412.

13. Kimber, I., Dearman, R.J., Basketter, D.A., Ryan, C.A., Gerberick, G.F., McNamee, P.M.,Lalko, J., Api,A.M. (2008).Dosemetrics in the acquisition of skin sensitization: thresholdsand importance of dose per unit area. Regulatory Toxicology and Pharmacology, 52, 39–45.

14. Adler, S., Basketter, D., Creton, S., Pelkonen, O., van Benthem, J., Zuang, V., Andersen,K.E., Angers-Loustau, A., Aptula, A., Bal-Price, A., Benfenati, E., Bernauer, U., Bessems,J., Bois, F.Y., Boobis, A., Brandon, E., Bremer, S., Broschard, T., Casati, S., Coecke, S.,Corvi, R., Cronin, M., Daston, G., Dekant, W., Felter, S., Grignard, E., Gundert-Remy,U., Heinonen, T., Kimber, I., Kleinjans, J., Komulainen, H., Kreiling, R., Kreysa, J., Leite,S.B., Loizou, G., Maxwell, G., Mazzatorta, P., Munn, S., Pfuhler, S., Phrakonkham, P.,Piersma, A., Poth, A., Prieto, P., Repetto, G., Rogiers, V., Schoeters, G., Schwarz, M.,Serafimova, R., Tahti, H., Testai, E., van Delft, J., van Loveren, H., Vinken, M., Worth,A., Zaldivar, J.M. (2011). Alternative (non-animal) methods for cosmetics testing: currentstatus and future prospects-2010. Archives in Toxicology, 85, 367–485.

15. Aeby, P., Ashikaga, T., Bessou-Touya, S., Schepky, A., Gerberick, F., Kern, P., Marrec-Fairley, M., Maxwell, G., Ovigne, J.M., Sakaguchi, H., Reisinger, K., Tailhardat, M.,Martinozzi-Teissier, S., Winkler, P. (2010). Identifying and characterizing chemical skinsensitizers without animal testing: Colipa’s research and method development program.Toxicology In Vitro, 24, 1465–1473.

16. Rovida, C., Basketter, D., Casati, S., de Silva, O., Hermans, H., Kimber, I., Manou, I.,Weltzien, H.U., Roggen, E. (2007). Management of an integrated project (Sens-it-iv) todevelop in vitro tests to assess sensitisation. ATLA Alternatives to Laboratory Animals,35, 317–322.

17. Hirota, M., Suzuki, M., Hagino, S., Kagatani, S., Sasaki, Y., Aiba, S., Itagaki, H. (2009).Modification of cell-surface thiols elicits activation of human monocytic cell line THP-1:possible involvement in effect of haptens 2,4-dinitrochlorobenzene and nickel sulfate.Journal of Toxicological Sciences, 34, 139–150.

18. Natsch, A., Emter, R. (2008). Skin sensitizers induce antioxidant response element depen-dent genes: application to the in vitro testing of the sensitization potential of chemicals.Toxicological Sciences, 102, 110–119.

19. Vandebriel, R.J., Pennings, J.L., Baken, K.A., Pronk, T.E., Boorsma, A., Gottschalk, R.,Van Loveren, H. (2010). Keratinocyte gene expression profiles discriminate sensitizingand irritating compounds. Toxicological Sciences, 117, 81–89.

20. Ade, N., Leon, F., Pallardy, M., Peiffer, J.L., Kerdine-Romer, S., Tissier, M.H., Bonnet,P.A., Fabre, I., Ourlin, J.C. (2009). HMOX1 and NQO1 genes are upregulated in responseto contact sensitizers in dendritic cells andTHP-1 cell line: role of theKeap1/Nrf2 pathway.Toxicological Sciences, 107, 451–460.

REFERENCES 173

21. Natsch, A. (2010). The Nrf2-Keap1-ARE toxicity pathway as a cellular sensor for skinsensitizers - functional relevance and a hypothesis on innate reactions to skin sensitizers.Toxicological Sciences, 113, 284–292.

22. Neves, B.M., Goncalo, M., Figueiredo, A., Duarte, C.B., Lopes, M.C., Cruz, M.T. (2011).Signal transduction profile of chemical sensitisers in dendritic cells: an endpoint to beincluded in a cell-based in vitro alternative approach to hazard identification? Toxicologyand Applied Pharmacology, 250, 87–95.

23. Natsch, A., Emter, R., Ellis, G. (2009). Filling the concept with data: integrating data fromdifferent in vitro and in silico assays on skin sensitizers to explore the battery approachfor animal-free skin sensitization testing. Toxicological Sciences, 107, 106–121.

24. Kim, H.J., Barajas, B., Wang, M., Nel, A.E. (2008). Nrf2 activation by sulforaphanerestores the age-related decrease of T(H)1 immunity: role of dendritic cells. Journal ofAllergy and Clinical Immunology, 121, 1255–1261.

25. Wakabayashi, N., Dinkova-Kostova, A.T., Holtzclaw, W.D., Kang, M.I., Kobayashi, A.,Yamamoto, M., Kensler, T.W., Talalay, P. (2004). Protection against electrophile andoxidant stress by induction of the phase 2 response: fate of cysteines of the Keap1 sensormodified by inducers. Proceedings of the National Academy of Sciences of the UnitedStates of America, 101, 2040–2045.

26. Motohashi, H., Yamamoto, M. (2004). Nrf2-Keap1 defines a physiologically importantstress response mechanism. Trends in Molecular Medicine, 10, 549–557.

27. Dinkova-Kostova, A.T., Holtzclaw, W.D., Kensler, T.W. (2005). The role of Keap1 incellular protective responses. Chemical Research in Toxicology, 18, 1779–1791.

28. Lou, H., Du, S., Ji, Q., Stolz, A. (2006). Induction of AKR1C2 by phase II inducers:identification of a distal consensus antioxidant response element regulated by NRF2.Molecular Pharmacology, 69, 1662–1672.

29. Gildea, L.A., Ryan, C.A., Foertsch, L.M., Kennedy, J.M., Dearman, R.J., Kimber, I.,Gerberick, G.F. (2006). Identification of gene expression changes induced by chemicalallergens in dendritic cells: opportunities for skin sensitization testing. Journal of Inves-tigative Dermatology, 126, 1813–1822.

30. Ryan, C.A., Gildea, L.A., Hulette, B.C., Dearman, R.J., Kimber, I., Gerberick, G.F. (2004).Gene expression changes in peripheral blood-derived dendritic cells following exposureto a contact allergen. Toxicology Letters, 150, 301–316.

31. Boukamp, P., Petrussevska, R.T., Breitkreutz, D., Hornung, J., Markham, A., Fusenig,N.E. (1988). Normal keratinization in a spontaneously immortalized aneuploid humankeratinocyte cell line. Journal of Cell Biology, 106, 761–771.

32. Emter, R., Ellis, G., Natsch, A. (2010). Performance of a novel keratinocyte-based reportercell line to screen skin sensitizers in vitro. Toxicology and Applied Pharmacology, 245,281–290.

33. Natsch, A., Bauch, C., Foertsch, L., Gerberick, F., Normann, K., Hilberer, A., Inglis, H.,Landsiedel, R., Onken, S., Reuter, H., Schepky, A., Emter, R. (2011). The intra- andinter-laboratory reproducibility and predictivity of the KeratinoSens assay to predict skinsensitizers in vitro: results of a ring-study in five laboratories. Toxicology In Vitro, 25,733–744.

34. Casati, S., Aeby, P., Kimber, I., Maxwell, G., Ovigne, J.M., Roggen, E., Rovida, C., Tosti,L., Basketter, D. (2009). Selection of chemicals for the development and evaluation of in


vitro methods for skin sensitisation testing. ATLA Alternatives to Laboratory Animals, 37,305–12.

35. Sens-it-iv. (2009). Sens-it-iv Newsletter 32, October 12, 2009.

36. ICCVAM. (2008). Revised Draft ICCVAM LLNA Performance Standards.

37. Natsch, A., Gfeller, H. (2008). LC-MS-based characterization of the peptide reactivity ofchemicals to improve the in vitro prediction of the skin sensitization potential. Toxicolog-ical Sciences, 106, 464–478.

38. Aleksic, M., Thain, E., Roger, D., Saib, O., Davies, M., Li, J., Aptula, A., Zazzeroni, R.(2009). Reactivity profiling: covalent modification of single nucleophile peptides for skinsensitization risk assessment. Toxicological Sciences, 108, 401–411.

39. Roberts, D.W., Patlewicz, G., Kern, P.S., Gerberick, F., Kimber, I., Dearman, R.J., Ryan,C.A., Basketter, D.A., Aptula, A.O. (2007). Mechanistic applicability domain classifi-cation of a local lymph node assay dataset for skin sensitization. Chemical Research inToxicology, 20, 1019–1030.

40. Kern, P.S., Gerberick, G.F., Ryan, C.A., Kimber, I., Aptula, A., Basketter, D.A. (2010).Local lymph node data for the evaluation of skin sensitization alternatives: a secondcompilation. Dermatitis, 21, 8–32.

41. Niklasson, I.B., Broo, K., Jonsson, C., Luthman, K., Karlberg, A.T. (2009). Reduced sen-sitizing capacity of epoxy resin systems: a structure-activity relationship study. ChemicalResearch in Toxicology, 22, 1787–1794.

42. Niklasson, I.B., Delaine, T., Luthman, K., Karlberg, A.T. (2011). Impact of a heteroatomin a structure-activity relationship study on analogues of phenyl glycidyl ether (PGE) fromepoxy resin systems. Chemical Research in Toxicology, 24, 542–548.

43. Delaine, T., Niklasson, I.B., Emter, R., Luthman, K., Karlberg, A.T., Natsch, A. (2011).Structure-activity relationship between the in vivo skin sensitizing potency of ana-logues of phenyl glycidyl ether and the induction of Nrf2-dependent Luciferase activ-ity in the KeratinoSens in vitro assay. Chemical Research in Toxicology, 24, 1312–1318.

44. Ball, N., Cagen, S., Carrillo, J.C., Certa, H., Eigler, D., Emter, R., Faulhammer, F., Garcia,C., Graham, C., Haux, C., Kolle, S.N., Kreiling, R., Natsch, A., Mehling, A. (2011).Evaluating the sensitization potential of surfactants: integrating data from the local lymphnode assay, guinea pig maximization test, and in vitro methods in a weight-of-evidenceapproach. Regulatory Toxicology and Pharmacology, 60, 389–400.

45. Montelius, J., Wahlkvist, H., Boman, A., Fernstrom, P., Grabergs, L., Wahlberg, J.E.(1994). Experience with the murine local lymph node assay: inability to discriminatebetween allergens and irritants. Acta Dermato-Venereologica, 74, 22–27.

46. Ashikaga, T., Yoshida, Y., Hirota, M., Yoneyama, K., Itagaki, H., Sakaguchi, H.,Miyazawa, M., Ito, Y., Suzuki, H., Toyoda, H. (2006). Development of an in vitro skinsensitization test using human cell lines: the human Cell Line Activation Test (h-CLAT).I. Optimization of the h-CLAT protocol. Toxicology In Vitro, 20, 767–773.

47. Natsch, A., Gfeller, H., Emter, R., Ellis, G. (2010). Use of in vitro testing to identify anunexpected skin sensitizing impurity in a commercial product: a case study. ToxicologyIn Vitro, 24, 411–416.

48. Basketter, D.A., Kimber, I. (2009). Updating the skin sensitization in vitro data assessmentparadigm in 2009. Journal of Applied Toxicology, 29, 545–550.

REFERENCES 175

49. Jowsey, I.R., Basketter, D.A., Westmoreland, C., Kimber, I. (2006). A future approachto measuring relative skin sensitising potency: a proposal. Journal of Applied Toxicology,26, 341–350.

50. Jaworska, J., Harol, A., Kern, P.S., Gerberick, G.F. (2011). Integrating non-animal testinformation into an adaptive testing strategy—skin sensitization proof of concept case.ALTEX, 28, 211–225.

51. McKim, J.M., Keller, D.J., Gorski, J.R. (2010). A new in vitro method for identifyingchemical sensitizers combining peptide binding with ARE/EpRE-mediated gene expres-sion in human skin cells. Cutaneous and Ocular Toxicology, 29, 170–191.

52. Bauch, C., Kolle, S.N., Fabian, E., Pachel, C., Ramirez, T., Wiench, B., Wruck, C.J., vanRavenzwaay, B., Landsiedel, R. (2011). Intralaboratory validation of four in vitro assaysfor the prediction of the skin sensitizing potential of chemicals. Toxicology In Vitro, 25,1162–1168.

53. Gerberick, G.F., Vassallo, J.D., Foertsch, L.M., Price, B.B., Chaney, J.G., Lepoittevin,J.P. (2007). Quantification of chemical peptide reactivity for screening contact allergens:a classification tree model approach. Toxicological Sciences, 97, 417–427.

9HIGH-THROUGHPUT SCREENINGASSAYS TO ASSESS CHEMICALPHOTOTOXICITY

Satomi Onoue, Yoshiki Seto, and Shizuo Yamada

9.1 INTRODUCTION

Phototoxicity is an undesirable response in the skin and eyes, triggered by exposureto sunlight, especially ultraviolet (UV) A/B radiation (UVA: 320–400 nm and UVB:290–320 nm) and visible (VIS) light (400–700 nm) [1]. Recently, the level of inter-est in drug-induced phototoxicity has markedly increased owing to the awarenessamong the scientific community of the increased level of UV radiation from the sunreaching the earth. There are at least three types of phototoxic skin reactions, includ-ing photoirritant, photogenotoxic, and photoallergic cascades, their mechanisms andpathologic features being quite different [2]. Phototoxic reactions in the skin andeyes can be caused by several classes of pharmaceuticals at clinical doses, includ-ing diuretic agents, nonsteroidal anti-inflammatory drugs (NSAIDs), antipsychoticdrugs, antimicrobials, antimalarials, cardiovascular drugs, anticonvulsants, protonpump inhibitors, psoralens, and anticancer drugs [3]. Perceptible adverse effectswould lead to a reduction in medication compliance, and these phototoxic reactionscan be caused by injected, orally, or topically applied phototoxic drugs. Most pho-toirritant reactions result from the systemic administration of the agent, whereasphotoallergic reactions may be caused by either topical or systemic administration ofthe chemicals.In addition to the clinical issues, drug-induced phototoxicity has also thwarted the

development of new drug entities, and the pharmaceutical industry and regulatory


177

178 HIGH-THROUGHPUT SCREENINGASSAYS TOASSESS CHEMICAL PHOTOTOXICITY

agencies have struggled to predict and/or avoid the phototoxic liability. In particular,at early phases of the drug discovery process, the development of an efficacious pho-totoxicity testing system is essential for the avoidance of side effects. As regulatoryagencies recommend the implementation of the 3Rs principle (refinement, reduction,replacement), the interest in the development of high-throughput in vitro methodsand the use of mechanistic information have been increasing in hazard identificationand characterization steps of the risk assessment process [4]. Therefore, a number ofeffective methodologies for the evaluation of the phototoxic risk of chemicals havebeen developed over the past few years, and guidance on the photosafety testing ofmedicinal products was established by the regulatory agencies in the United Statesand the EuropeanUnion in the early 2000s. These guidelines have described the testedcompounds on the basis of light absorption properties, administration routes, and dis-tribution behaviors, and they have recommended some research tools for photosafetyassessment in pharmaceutical research and development. In the present chapter, wesummarize the research tools and high-throughput screening systems developed forphotosafety evaluation of new drug candidates.

9.2 PHOTOTOXIC PATHWAYS

Penetration and absorption of light in the skin can be an important factor in drug-induced phototoxicity as described by the Grotthus–Draper law of photobiologystates; only light that is absorbed can be active in photochemical and photobiolog-ical processes. When a drug molecule absorbs a photon energy, electrons can beprompted from occupied orbitals (the ground state, S0) to an unoccupied orbital(S1, S2) depending upon the bond type and associated energy level (Fig. 9.1a).

FIGURE 9.1 Possible mechanisms of drug-induced phototoxicity. (a) Jablonski diagram.S, singlet state; T, triplet state; IC, internal conversion; ISC, intersystem crossing. Each linebetween the singlet states indicates the excited vibrational states, and excited rotational statesare not shown. (b) Several phototoxic responses caused by photo-activated drugs.

SCREENING SYSTEMS FOR PHOTOSAFETY ASSESSMENT 179

Furthermore, unpaired singlet state electrons (opposite spin) may be converted to atriplet state (parallel spin) by inversion of the spin via intersystem crossing of theabsorbed energy. To return to the ground state from S1, S2, and T1, energy must bedissipated by internal conversion, fluorescence (from the singlet state), phosphores-cence (from the triplet state), or via chemical reaction, giving rise to photoproductsand/or potential external reactions with biomolecules such as proteins, lipids, andDNA (Fig. 9.1b). In addition, molecular oxygen, a triplet radical in its ground state,appears to be the predominant acceptor of excitation energy as its lowest excitedlevel (singlet state) has a comparatively low value. An energy transfer from theexcited triplet photosensitizer to the oxygen (type II photochemical reaction) couldproduce excited singlet oxygen, which, in turn, might participate in lipid- and protein-membrane oxidation or induce DNA damage. An electron or hydrogen transfer couldlead to the formation of free radical species (type I photochemical reaction), produc-ing a direct attack on the biomolecules or, in the presence of oxygen, evolving towardsecondary free radicals such as peroxyl radicals or the very reactive hydroxyl radical,a known intermediate in the oxidative damage of DNA and other biomolecules. Thistoxic pathway involves a series of reactions, including the appearance of superoxideanion radical, its dismutation to form hydrogen peroxide, followed by the hydro-gen peroxide reduction to form the hydroxyl radical. The generation of the radicalinvolves either the photosensitizer or the target biomolecules [5]. Herein, excitationof the drug by light may give rise to reactive oxygen species (ROS) such as singletoxygen and superoxide, which may be one of the causative agents of drug-inducedphototoxicity [2, 6]. In addition to these phototoxic responses, the possibility existsthat the photochemical reaction of compounds with ROS results in the formation ofsome toxic degradation products. For some drugs, such as amiodarone and chlorpro-mazine, metabolites may play a significant role in phototoxic responses [7, 8].

9.3 SCREENING SYSTEMS FOR PHOTOSAFETY ASSESSMENT

An in vivo photopatch test to evaluate drug-induced phototoxicity on the basis ofboth photoirritant and photoallergic reactions in the skin for clinical inspection hasbeen developed [9]. In the early stage of pharmaceutical development, an in vitro 3T3neutral red uptake phototoxicity test (3T3 NRU PT) has been used as a simplifiedalternative approach to estimate the photosafety of chemicals [10, 11]. However, thein vitro 3T3 NRU PT has some drawbacks, such as too many false-positive resultsand low throughput. Therefore, a higher throughput and different pathogenesis-basedscreening systems are required for precise photosafety evaluation in an early phaseof drug discovery. On the basis of the pathogenesis of drug-induced phototoxicity,various assessment tools for the evaluation of the phototoxic potential of chemicalshave been developed over the past few years (Table 9.1). However, these phototoxicassessment tools cannot always be applied to all types of new drugs due to the physic-ochemical properties of the tested chemicals and the limited specificity and sensitivityof the screening systems, so that researchers require a thorough understanding of thelimitations of each assessment. The photosafety assay systems should be designed


TABLE 9.1 Screening Systems for Photosafety Evaluation

Screening Systems References

In silico predictionsDEREK Barratt et al. [12, 13]HOMO–LUMO gap Betowski et al. [14]

Photochemical evaluationsUV absorption Henry et al. [15]ROS assay Onoue et al. [16, 17]D-ROM assay Onoue et al. [18]

In vitro phototoxic assessments(i) Phototoxicity/photoirritancyPhotopatch test Epstein [9]Photo-basophil-histamine-release test Przybilla et al. [20]Photohemolysis model Selvaag et al. [21]Human reconstituted epidermis model Portes et al. [22]3T3 NRU phototoxicity test Spielmann et al. [10]Oxygen consumption in Bacillus subtilis Beani et al. [23]Yeast growth inhibition assay Sugiyama et al. [24](ii) PhotogenotoxicityPhoto Ames test Brendler-Schwaab et al. [6]Photo comet assay Brendler-Schwaab et al. [6]Electrophoretic DNA-photocleaving assay Onoue et al. [25]DNA-binding assay Onoue et al. [26]IBP assay Seto et al. [27](iii) PhotoallergyPig skin model Sarabia et al. [28]Photo-h-CLAT Hoya et al. [29]

for intended types of phototoxic responses such as photoirritation, photogenotoxicity,and/or photoallergy.

9.3.1 In Silico Prediction

In silico prediction systems such as deductive estimation of risk from existing knowl-edge (DEREK) and highest occupied molecular orbital–lowest unoccupied molecularorbital (HOMO–LUMO) gap can help to predict the phototoxic risk of chemicalsbefore chemicals are synthesized. DEREK allows toxicity prediction of chemicalsbased on structures associatedwith the incidence of toxicity [12, 13], and theHOMO–LUMO gap determines the photoreactive potential of chemicals using energy differ-ences between levels of HOMO and LUMO [14]. These in silico prediction methodscould help to reduce the resources required for the synthesis of new drug candidates.

9.3.2 Photochemical Evaluation

After chemicals are newly synthesized, examination of their photochemical propertieswould be useful for phototoxic evaluation. UV spectral analysis provides informationon the photoexcitability of chemicals on the basis of the first law of photochemistry

SCREENING SYSTEMS FOR PHOTOSAFETY ASSESSMENT 181

[15]. Based on the results of UV measurement, the photoexcitability of chemicalscan be determined; however, the photoreactivity of chemicals, possibly leading to aphototoxic reaction, is unclear because some excited chemicals release the photonenergy via emission of fluorescence, phosphorescence, and heat. UV absorption ofchemicals may not always correlate directly with their phototoxic potential. In thiscontext, some compounds might be falsely predicted to be phototoxic or nonphoto-toxic if the decision is only based on the UV spectral analysis. Therefore, another oran additional screeningmethod should be applied for the evaluation of photoreactivityin order to avoid wrong conclusions.On the basis of the pathogenesis of drug-induced phototoxicity, an ROS assay

and derivatives of reactive oxygen metabolites (D-ROM) assay were proposed toevaluate the phototoxic risk of chemicals. The ROS assay can monitor the gener-ation of ROS, such as singlet oxygen and superoxide, from photoirradiated chem-icals, and the ROS data can be used to evaluate the photoreactivity of chemicals[16, 17]. The D-ROM assay can also determine ROS generation from photoirradi-ated chemicals via monitoring of the generation of ROS-derived metabolites, andthe D-ROM assay can shorten irradiation time when compared with that of the ROSassay [18]. These screening systems can comprehensively evaluate the phototoxicpotentials of chemicals such as photoirritation, photogenotoxicity, and photoallergywith a high throughput, and they may be appropriate for evaluating the phototoxicrisk of chemicals as a first screening in an early stage of the drug discovery process.

9.3.3 In Vitro Phototoxic Assessment

As a photoirritant risk assessment tool, the 3T3 NRU PT is practically used to eval-uate the phototoxicity of chemicals. The in vitro 3T3 NRU PT is a highly sensitivemethodology for evaluating phototoxic risk, especially photoirritant risk, and con-tributes to the development of drug candidates. However, the photosafety assessmentof chemicals by 3T3 NRU PT sometimes provides false-positives because of itshigh detection sensitivity [19]. Additionally, UVB radiation is partially blocked inthe 3T3 NRU PT because cells are killed by UVB radiation; therefore, chemicalsonly excited by UVB radiation exposure produce false-negative results in the assay.Further, assay systems based on the release of inflammatory mediators from leuco-cytes (photo-basophil-histamine-release test) [20], membrane damage of erythrocytes(photohemolysis model) [21], morphological change of a three-dimensional humanskin model (human reconstituted epidermis model) [22], change of oxygen consump-tion in Bacillus subtilis (oxygen consumption in Bacillus subtilis) [23], and growthinhibition of yeast cells (yeast growth inhibition assay) [24] could also be useful forin vitro photoirritation assessments.To evaluate the photogenotoxic potential of chemicals, a photo Ames test which is

based on mutations in genes involved in histidine biosynthesis, has mainly been used.Moreover, a photo comet assay, which is a sensitive test to detect DNA damage at thelevel of the individual eukaryotic cell, is also conducted to evaluate the in vivo/in vitrophotogenotoxic potential of drugs [6]. Our group also proposed three photogenotoxic-ity assessment tools based on DNA damage monitored by electrophoresis and a DNA


intercalator. A capillary gel-electrophoretic DNA-photocleaving assay monitors pho-tocleavage of plasmid DNA by photoirradiated chemicals [25]. ADNA-binding assayevaluates the DNA-binding affinity of chemicals on the basis of results from ethidiumdisplacement or fluorescence titration experiments, and the assay could be used asa second screening following the ROS assay for evaluating the photogenotoxic riskof compounds [26]. An intercalator-based photogenotoxicity (IBP) assay monitorsDNA damage by photoirradiated chemicals on the basis of fluorescence emissionfrom DNA–intercalator complexes [27].A pig skin model and a photo human cell line activation test (photo-h-CLAT) were

developed for evaluating the photoallergic potential of chemicals. The pig skin modelevaluates the formation of drug-protein photoadducts in pig skin [28], and the photo-h-CLAT monitors the overexpression of CD86/54 on dendritic cells in the courseof antigen presentation to T lymphocytes [29]. Although the assays have not beeninternationally accepted, they should be effective for evaluating the photoallergicpotential of compounds.

9.4 3T3 NRU PHOTOXICITY TEST

The 3T3 NRU PT was developed and validated under the auspices of ECVAM from1992 to 1997, to establish a valid in vitro alternative to the various in vivo tests inuse [30]. The test is now accepted by the EU Commission and member states asbeing necessary for all compounds showing the absorbance of UVA and VIS light[31]. The 3T3 NRU PT is conducted using Balb/c 3T3 mouse fibroblasts to assess thephototoxicity of a test compound. The assay quantitatively determines the cytotoxicpotential of a test compound by comparing the reduction of the neutral red dye(3-amino-7-dimethylamino-2-methylphenazine hydrochloride) uptake in Balb/c 3T3cell cultures exposed to serial dilutions of a test article with the neutral red dye uptakein the control (the test article vehicle). The Balb/c 3T3 cells were selected as the targetcell because: (i) they are readily available, grow rapidly in culture and respond readilyto toxic assault; (ii) these cells are used in many cytotoxicity studies; (iii) the cells arelargely refractive to doses of UV radiation used to initiate the phototoxicity reactionin this system [32, 33], and (iv) the assay, accepted by regulatory agencies, wasvalidated using 3T3 cells.An unpublished survey of pharmaceutical companies by the European Federation

of Pharmaceutical Industries and Associations has revealed an extremely high per-centage of compounds (> approximately 85%) that are reported as positive in the3T3 NRU PT, but are negative when tested using animal models. This indicates thatthe analytical detection level is far more sensitive than the biological effect level.Despite issues of false-positives, it should be noted that this assay predicts nega-tives with excellent fidelity. This reflects the efforts made to validate the assay andis probably the strongest point recommending its utility. However, given the highproportion of potential drug candidates that absorb in the UVA region together witha 50% 3T3 NRU PT-positive hit rate for UVA-absorbing compounds, the burden of

ROS ASSAY 183

in vivo photosafety testing relative to the clinical impact of phototoxicity is dispro-portionate. Based on in vivo and clinical experience, the 3T3 NRU PT is substantiallyoverpredicting human hazard. The consequences of excessive in vitro photosafetypositives include additional preclinical studies with increased animal usage, addi-tional clinical assessments, increased cost and time of development programmes, andimplications for patient informed consent and labeling documents. Another difficultylies in the assumption that results from the 3T3 NRU PT correlate with in vivo pho-totoxicity responses. Only UVA wavelengths are used for irradiation in the 3T3 NRUPT, while the fluorescent bulbs present in vivaria do not emit UVA. Therefore, forexample, the inference that retinal toxicity seen in preclinical animal models is pho-totoxicity because of a 3T3 NRU PT-positive compound is unjustified. Only whenphototoxicity is observed at the wavelengths that the animals are actually exposed tocan such an inference bemade. The 3T3NRUPTwas developed to detect the photoir-ritant potential of a drug. The test is not designed to predict other adverse effects thatmay arise from the combined action of a chemical and light, for example, photogeno-toxicity, photoallergy, and photocarcinogenicity, although many chemicals, whichshow these specific properties, will react positively in the in vitro 3T3 NRU PT.

9.5 ROS ASSAY

9.5.1 Theoretical Basis of the ROS Assay and Assay Reliability

Determination of ROS formed as a consequence of irradiating pharmaceutical sub-stances withUVA andUVBwould be of help in recognizing their phototoxic potential[16, 17, 34, 35]. In the ROS assay, generation of singlet oxygen was detected by spec-trophotometricmeasurement of p-nitrosodimethyl aniline (RNO) bleaching, followedby decreased absorbance of RNO at 440 nm [36]. Although singlet oxygen does notchemically react with RNO, the RNO bleaching is a consequence of singlet oxygencapture by the imidazole ring, resulting in the formation of a trans-annular peroxideintermediate capable of inducing the bleaching of RNO as follows:

singlet oxygen+ imidazole → [peroxide intermediate] → oxidized imidazole[peroxide intermediate]+ RNO → RNO+ products

The generation of superoxide can be determined by the reduction of nitrobluetetrazolium (NBT) [37], as indicated below. NBT can be reduced by superoxide anionvia a one-electron transfer reaction, yielding partially reduced (2 e−) monoformazan(NBT+ ) as a stable intermediate. Thus, superoxide can reduce NBT toNBT+ , whoseformation can be monitored spectrophotometrically at 560 nm.

Superoxide+ NBT → O2 + NBT+

Previously, a high-throughput ROS assay system was developed for the determi-nation of singlet oxygen and superoxide [34] by using a multiwell plate and a quartz


FIGURE 9.2 Quartz reaction container designed for the reactive oxygen species (ROS) assaythat consists of a quartz plate, a multiwell plate, a Teflon sheet, and a steel multiwell retainer.Use of the shaded wells (B2–G11) is recommended for a robust ROS assay. (Reprinted withsome modifications from J. Pharm. Biomed. Anal. 46, 187–193 (2008) with permission fromElsevier.)

reaction container (Fig. 9.2). Preliminary research demonstrated that the ROS genera-tion on the outer lane (A1-H1, A12-H12, A2-A11, and H2-11) tended to be 1.2–1.9%less than in the inner zone (B2–G11), possibly due to differences in irradiance amongwells. For a robust ROS assay, it would therefore be better to only use the inner zone.The experimental conditions (irradiance uniformity, UV intensity, exposure time,

temperature, and solvent systems) were found to affect the generation of ROS, and,thus, based on these results the conditions of the ROS assay were optimized [34]. Thereproducibility was evaluated by intra- and interday assays, and the intra- and interdayprecisions (relative standard deviation) for the determination of ROS from irradiatedquinine (200 �M) were found to be less than 3.3% and 4.5%, respectively, therebysuggesting that the precision was sufficient. According to the multiple measurements(20 times) of quinine (positive standard) and sulisobenzone (negative standard), Z′

factors for the determination of singlet oxygen and superoxide were calculated to be0.92 and 0.87, respectively, demonstrating that the assay allows a large separationband between samples and blank signals and thereby confirming its suitability forhigh-throughput screening.

9.5.2 Prediction Capacity

Previously, the photosensitizing properties of 39model compounds, as well as ca. 210drug candidates including 11 chemical series, were evaluated using the ROS assay

ROS ASSAY 185

and the 3T3 NRU PT to establish provisional classification criteria for risk assess-ment of drug-induced phototoxicity [16]. With respect to marketed drugs (Fig. 9.3a),most phototoxic drugs tended to cause type I and/or type II photochemical reac-tions, resulting in the generation of singlet oxygen and superoxide. There seemedto be a clear difference between phototoxic drugs and nonphototoxic compounds intheir ability to induce photochemical reactions. A plot analysis of ROS data on themarketed drugs provided classification criteria (5.0× 10−2 for both singlet oxygenand superoxide) to discriminate the photosensitizers from nonphototoxic substances[16]. Of all tested drugs, only 5-fluorouracil (5-FU) was falsely predicted to benonphototoxic, its phototoxic mechanism of action probably being different fromthat of other photosensitizers. Alternatively, the occurrence of phototoxicity in thecase of 5-FU may require concomitance of some biomolecules including DNA andRNA, into which phosphorylated 5-FU could be incorporated. Further research onthe possible cascades involved in 5-FU phototoxicity is needed to understand thelimitation of this assay, and the study may also enable the improvement of theROS assay.In addition to the model compounds, the phototoxic potential of 210 drug candi-

dates consisting of 11 chemical series (azaindoles, benzimidazoles, cyanobenzenes,dihydropyridines, furopyridines, imidazopyridines, pyrazines, pyrazoles, pyridines,pyrimidines, and quinolones) was also assessedwith the use of the ROS assay and 3T3NRU PT (Fig. 9.3b). According to the results of the 3T3 NRU PT, 60 compounds(28.6% of total) were identified as being phototoxic and 14 compounds (6.7% oftotal) as probably phototoxic. All the phototoxic and probably phototoxic compoundsshowed a significant amount of ROS generation that exceeded the threshold level.In addition, 136 compounds (64.8% of total) were found to be nonphototoxic on thebasis of 3T3NRUPTdata, and 46.3%of nonphototoxic compounds showed no signif-icant ROS generation, lying in the subthreshold region. In theory, the ROS assay cancapture all the photochemically active substances as it can detect type I and/or type IIphotochemical reactions induced by irradiated compounds. These photochemicalreactions were observed in a very early stage of drug-induced phototoxic cascades,so that the ROS assay has been thought to be effective for photosafety evaluation ofpharmaceuticals. However, the possibility exists that, as in the case of phototoxins,some photolabile substances could also be recognized as being phototoxic by the ROSassay, because of significant ROS generation during the photodegradation processes.Namely, the ROS assay sometimes provides false-positive predictions as observed inthe present validation study. This could explain in part the data discrepancy observedbetween the ROS assay and the in vitro phototoxicity assay, and better understandingof the limitations of the assay could be of great help to avoid the overestimation andmisleading conclusions.Thus, the provisional classification criteria based on the ROS assay provided

no false-negatives as compared to the 3T3 NRU PT, and they could be used as afirst screening to identify the phototoxic potential of drug candidates. These resultsconfirm the usefulness of the ROS assay in understanding the phototoxicity riskof pharmaceutical substances, and the ROS assay can be used for screening pur-poses at the drug discovery stage. However, the tentative discrimination criteria were


FIGURE 9.3 Predictive capacity of the ROS assay on the basis of data for (a) 39 marketeddrugs and (b) 210 drug candidates. (a-I and b-I) 2D plot analyses of singlet oxygen dataversus superoxide data. According to 3T3 NRU PT data or clinical observations, each drug orcandidate was defined as ◦, nonphototoxic compounds; �, probably phototoxic compounds;and×, phototoxins. In each case, the ROS assay was carried out using an LTX-01 solarsimulator (from Nagano Science, Osaka, Japan) with an irradiation of 1.8 mW/cm2 for 18 h.Data represent mean of four experiments. (i) ROS data for 5-FU and (ii) ROS generation wasnegligible in the case of 12 drug candidates. The shaded region is indicative of low phototoxicpotential according to the classification criteria defined previously [34]. (a-II and b-II) Positiveand negative predictivity of the ROS assay as compared to the in vitro/in vivo phototoxicity.(Reprinted with some modifications from J. Pharm. Biomed. Anal. 47, 967–972 (2008) withpermission from Elsevier.)

REFERENCES 187

obtained with the ROS assay by using the LTX-01 solar simulator (Nagano Science,Osaka, Japan) equipped with a xenon lamp. In case of wanting to use alternative lightsources, their compatibility should be verified. In this context, it should be mentionedthat the Japan Pharmaceutical Manufacturers Association (JPMA), supervised by theJapanese Center for the Validation of AlternativeMethods (JaCVAM), has carried outa validation study on the ROS assay with other solar simulators to verify relevanceand reliability. Currently, the introduction of the validated ROS assay into the photo-safety guideline (ICH Guideline S10: “Photosafety Evaluation of Pharmaceuticals”)of the International Conference on Harmonization of Technical Requirements forRegistration of Pharmaceuticals for Human Use (ICH) is under consideration.

9.6 CONCLUSION AND FUTURE OUTLOOK

At an early stage of drug discovery, high-throughput photosafety screening systemsare an invaluable tool to identify compounds with undesirable properties, so thata number of in vitro assay systems have been developed and validated to identifycompounds that have a high phototoxic potential in vivo. The in vitro photosafetyassessment tools are also effective in elucidating in detail the mechanism of photo-toxicity of chemicals. However, the results from in vitro photosafety evaluations donot always reflect clinical observations on drug-induced phototoxicity. Phototoxicreactions mainly occur in the skin or eyes after topical and systemic administration.Therefore, a characterization of the pharmacokinetic behavior of the tested chemicals,such as their skin and eye distribution and skin and eye retention properties, wouldbe important additional parameters for photosafety evaluation, and the combined useof in vitro photobiochemical/phototoxic and pharmacokinetic data might enable theevaluation of in vivo phototoxic risk with high clinical relevance. Interest in the pho-tosafety of chemicals has continuously increased in regulatory science as well as inthe area of drug discovery. Moreover, the development of effective high-throughputscreening strategies to evaluate the phototoxic risk of chemicals would provide reli-able photosafety assessments and aid productive research and development of newpharmaceuticals.

REFERENCES

1. Epstein, J.H. (1983). Phototoxicity and photoallergy in man. Journal of the AmericanAcademy of Dermatology, 8, 141–147.

2. Epstein, J.H., Wintroub, B.U. (1985). Photosensitivity due to drugs. Drugs, 30, 42–57.

3. Onoue, S., Seto, Y., Gandy, G., Yamada, S. (2009). Drug-induced phototoxicity; an earlyin vitro identification of phototoxic potential of new drug entities in drug discovery anddevelopment. Current Drug Safety, 4, 123–136.

4. Holmes, A.M., Creton, S., Chapman, K. (2010). Working in partnership to advance the3Rs in toxicity testing. Toxicology, 267, 14–19.


5. Quintero, B., Miranda, M.A. (2000). Mechanisms of photosensitization induced by drugs:a general survey. Ars Pharmaceutica, 41, 27–46.

6. Brendler-Schwaab, S., Czich, A., Epe, B., Gocke, E., Kaina, B., Muller, L., Pollet, D.,Utesch, D. (2004). Photochemical genotoxicity: principles and test methods. Report of aGUM task force. Mutation Research, 566, 65–91.

7. Ferguson, J., Addo, H.A., Jones, S., Johnson, B.E., Frain-Bell, W. (1985). A study ofcutaneous photosensitivity induced by amiodarone. British Journal of Dermatology, 113,537–549.

8. Ljunggren, B., Moller, H. (1977). Phenothiazine phototoxicity: an experimental study onchlorpromazine and its metabolites. Journal of Investigative Dermatology, 68, 313–317.

9. Epstein, S. (1964). The photopatch test; its technique, manifestations, and significance.Annals of Allergy, 22, 1–11.

10. Spielmann, H., Balls, M., Brand, M., Doring, B., Holzhutter, H.G., Kalweit, S., Klecak,G., Eplattenier, H.L., Liebsch, M., Lovell, W.W., Maurer, T., Moldenhauer, F., Moore, L.,Pape, W.J., Pfanenbecker, U., Potthast, J., De Silva, O., Steiling, W., Willshaw, A. (1994).EEC/COLIPA project on in vitro phototoxicity testing: first results obtained with a Balb/c3T3 cell phototoxicity assay. Toxicology In Vitro, 8, 793–796.

11. Spielmann, H., Liebsch, M., Doring, B., Moldenhauer, F. (1994). First results of anEC/COLIPA validation project of in vitro phototoxicity testing methods. ALTEX, 11,22–31.

12. Barratt, M.D. (2004). Structure-activity relationships and prediction of the phototoxicityand phototoxic potential of new drugs. Alternatives to Laboratory Animals, 32, 511–524.

13. Barratt, M.D., Castell, J.V., Miranda, M.A., Langowski, J.J. (2000). Development of anexpert system rulebase for the prospective identification of photoallergens. Journal ofPhotochemistry and Photobiology B, 58, 54–61.

14. Betowski, L.D., Enlow, M., Riddick, L. (2002). The phototoxicity of polycyclic aro-matic hydrocarbons: a theoretical study of excited states and correlation to experiment.Computers and Chemistry, 26, 371–377.

15. Henry, B., Foti, C., Alsante, K. (2009). Can light absorption and photostability data beused to assess the photosafety risks in patients for a new drug molecule? Journal ofPhotochemistry and Photobiology B, 96, 57–62.

16. Onoue, S., Kawamura, K., Igarashi, N., Zhou, Y., Fujikawa, M., Yamada, H., Tsuda, Y.,Seto, Y., Yamada, S. (2008) Reactive oxygen species assay-based risk assessment of drug-induced phototoxicity: classification criteria and application to drug candidates. Journalof Pharmaceutical and Biomedical Analysis, 47, 967–972.

17. Onoue, S., Tsuda, Y. (2006). Analytical studies on the prediction of photosensitive/phototoxic potential of pharmaceutical substances. Pharmaceutical Research, 23, 156–164.

18. Onoue, S., Ochi, M., Gandy, G., Seto, Y., Igarashi, N., Yamauchi, Y. Yamada, S. (2010).High-throughput screening system for identifying phototoxic potential of drug candidatesbased on derivatives of reactive oxygen metabolites. Pharmaceutical Research, 27, 1610–1619.

19. Lynch, A.M., Wilcox, P. (2011). Review of the performance of the 3T3 NRU in vitro pho-totoxicity assay in the pharmaceutical industry. Experimental and Toxicologic Pathology,63, 209–214.

REFERENCES 189

20. Przybilla, B., Schwab-Przybilla, U., Ruzicka, T., Ring, J. (1987). Phototoxicity of non-steroidal anti-inflammatory drugs demonstrated in vitro by a photo-basophil-histamine-release test. Photodermatology, 4, 73–78.

21. Selvaag, E. (1997). Evaluation of phototoxic properties of oral antidiabetics and diuretics.Photohemolysis model as a screening method for recognizing potential photosensitizingdrugs. Arzneimittelforschung, 47, 1031–1034.

22. Portes, P., Pygmalion, M.J., Popovic, E., Cottin, M., Mariani, M. (2002). Use of humanreconstituted epidermis Episkin for assessment of weak phototoxic potential of chemicalcompounds. Photodermatology, Photoimmunology and Photomedicine, 18, 96–102.

23. Beani, J.C., Gautron, R., Amblard, P., Bastrenta, F., Harrouch, L., Jardon, P., Reymond,J.L. (1985). Screening for drug photosensitization activity by measuring the variations inoxygen consumption of Bacillus subtilis. Photodermatology, 2, 101–106.

24. Sugiyama, M., Itagaki, H., Kato, S. (1994). In vitro assays to predict phototoxicity ofchemicals: (II) Yeast growth inhibition assay and battery system with photohemolysisassay. AATEX, 2, 193–202.

25. Onoue, S., Igarashi, N., Kitagawa, F., Otsuka, K., Tsuda, Y. (2008). Capillary elec-trophoretic studies on the photogenotoxic potential of pharmaceutical substances. Journalof Chromatography A, 1188, 50–56.

26. Onoue, S., Seto, Y., Oishi, A., Yamada, S. (2009). Novel methodology for predictingphotogenotoxic risk of pharmaceutical substances based on reactive oxygen species (ROS)and DNA-binding assay. Journal of Pharmaceutical Sciences, 98, 3647–3658.

27. Seto, Y., Ochi, M., Onoue, S., Yamada, S. (2010). High-throughput screening strategyfor photogenotoxic potential of pharmaceutical substances using fluorescent intercalatingdye. Journal of Pharmaceutical and Biomedical Analysis, 52, 781–786.

28. Sarabia, Z., Hernandez,D., Castell, J.V., vanHenegouwen,G.M. (2000). Photoreactivity oftiaprofenic acid and suprofen using pig skin as an ex vivomodel. Journal of Photochemistryand Photobiology B, 58, 32–36.

29. Hoya, M., Hirota, M., Suzuki, M., Hagino, S., Itagaki, H., Aiba, S. (2009). Developmentof an in vitro photosensitization assay using human monocyte-derived cells. ToxicologyIn Vitro, 23, 911–918.

30. Liebsch, M., Spielmann, H. (2002). Currently available in vitro methods used in theregulatory toxicology. Toxicology Letters, 127, 127–134.

31. EMEA Report. (2002). Committee for Propriety Medicinal Products (CPMP). Notesfor guidance on photosafety testing CPMP/SWP/398/01. Available at http://www.emea.europa.eu/pdfs/human/swp/039801en.pdf.

32. Organisation for Economic Co-operation and Development. (2004). OECD Guidelinesfor the Testing of Chemicals/Section 4 Health Effects. Test No. 432: In Vitro 3T3 NRUPhototoxicity Test.

33. Holzhutter, H.G. (1997). A general measure of in vitro phototoxicity derived from pairsof dose response curves and its use for predicting the in vivo phototoxicity of chemicals.ATLA, 25, 445–462.

34. Onoue, S., Igarashi, N., Yamada, S., Tsuda, Y. (2008). High-throughput reactive oxygenspecies (ROS) assay: An enabling technology for screening the phototoxic potential ofpharmaceutical substances. Journal of Pharmaceutical and Biomedical Analysis, 46, 187–193.


35. Onoue, S., Yamauchi, Y., Kojima, T., Igarashi, N., Tsuda, Y. (2008). Analytical studies onphotochemical behavior of phototoxic substances; effect of detergent additives on singletoxygen generation. Pharmaceutical Research, 25, 861–868.

36. Kraljic, I., Mohsni, S.E. (1978). A new method for the detection of singlet oxygen inaqueous solutions. Photochemistry and Photobiology, 28, 577–581.

37. Tarpey, M.M., Fridovich, I. (2001). Methods of detection of vascular reactive species:nitric oxide, superoxide, hydrogen peroxide, and peroxynitrite. Circulation Research, 89,224–236.

PART III

HIGH-THROUGHPUT ASSAYS TOASSESS DNA DAMAGE ANDCARCINOGENESIS


10AMES IITM AND AMES LIQUIDFORMAT MUTAGENICITYSCREENING ASSAYS

Kamala Pant

10.1 INTRODUCTION

In the early 1970s, several researchers affirmed the use of bacterial mutation assaysas a simple and rapid means of detecting mutagens and carcinogens [1–3]. Mutagenscan be detected by either forward or reverse mutation systems. Forward mutationsystems detect mutations as a change in the normal phenotype (appearance) of anorganism, whereas reverse mutation systems detect mutations as a reversion from amutant phenotype to a normal phenotype. The most popular reverse mutation assaysuse bacteria, Salmonella typhimurium and Escherichia coli strains.Named for and developed by Dr. Bruce Ames and others in the early 1970’s,

the standard “Ames assay” is one of the most commonly performed assays forsafety assessment and regulatory submissions. The assay is also called bacterialreverse mutation assay or S. typhimurium assay. The purpose of this bacterial reversemutation assay is to evaluate a chemical’s genotoxicity by measuring its ability toinduce reverse mutations at selected loci in several bacterial strains. It is sensitive to awide range of mutagenic and carcinogenic chemicals [4–6]. This simple, quick, andinexpensive assay is one of several genotoxicity assays required for product safetytesting of a variety of products including drugs, medical devices, food additives,industrial chemicals, and pesticides. This assay is used in regulatory toxicologytesting and is conducted according to the guidelines of various regulatory agenciessuch as FDA, EPA, and OECD.


193

194 AMES IITM AND AMES LIQUID FORMAT MUTAGENICITY SCREENING ASSAYS

The “Ames assay” measures genetic damage at the single base level in DNAby using five or more tester strains of bacteria. Each of the S. typhimurium and E.coli strains used in the assay carries a unique mutation that has turned off histidinebiosynthesis in S. typhimurium or tryptophan biosynthesis in E. coli [4, 7]. Becauseof these original mutations, the bacteria require exogenous histidine or tryptophanto survive and will starve to death if grown without these essential nutrients (aux-otrophy). The key to the assay is that bacteria can undergo a compensating mutationturning the essential gene back on (reverse mutation), permitting the cell to grow inthe absence of either histidine or tryptophan (prototrophy). Each bacterial strain wascreated by a specific type of mutation—either a base-pair substitution or a frameshiftmutation. Because a reverse, compensating mutation usually must occur by the samemutagenic mechanism, mechanistic toxicology information is also available fromAmes assay results based on the pattern of which strain(s) reverted.Chemicals can be direct-acting mutagens or they may require conversion by

metabolism to a mutagenic form. For this reason, in vitro genotoxicity assays, includ-ing the Ames assay, are conducted both in the presence and in the absence of anexogenous metabolic activation system. The metabolic activation system consistsof a rat liver homogenate (S9) and cofactors [8, 9]. Enzyme levels in rat livers areinduced prior to processing using either Aroclor 1254 or a mixture of phenobarbitaland benzoflavone.The standard Ames assay designs include preliminary toxicity tests or combined

toxicity and mutation tests followed by a definitive mutation assay [10]. In bothtoxicity and mutation tests, tester strains are combined with an S9 mix (activationcondition) or buffer (no activation), test or control article, a trace of histidine ortryptophan, and molten agar. The bacteria use the trace histidine or tryptophan toundergo several cell divisions, but will stop growing once histidine or tryptophanis used up, leaving a characteristic “background lawn” that decreases in densitywith increasing toxicity. After 48 h, only those cells that have undergone a reversemutation turning the essential gene back on have survived, producing mutant orrevertant colonies. The evaluation of the background lawn density is followed bythe counting of the number of revertant colonies. Mutation results are reported asrevertants per plate. A test article effect is evaluated by comparing the number ofrevertant colonies in test article-treated plates to the concurrent control.The Organisation for Economic Co-operation and Development (OECD) is an

intergovernmental organization dedicated to assisting governments and providingspecific guidance on economic, social, and governance challenges. The OECD hasestablished guidelines for the testing of chemicals, including pharmaceuticals, toassess their safety. In 1997, the OECD Guideline for Testing of Chemicals, Test471: Bacterial Reverse Mutation Test, was adopted. This test was recommendedas an initial screen for genotoxicity, and, in particular, for point mutation-inducingactivity. Point mutations are the cause of many human genetic diseases, and there issubstantial evidence that point mutations are involved in tumor formation in humansand animals [11].Genotoxicity is an important issue in drug development, and testing the potential

of small molecules to induce genetic alterations is an essential part of preclinical

AMES IITM ASSAY 195

safety evaluation. To minimize genotoxicity-related compound attrition seen duringregulatory studies, many important pharmaceutical companies conduct screeningstudies, usually with a variation of Ames protocols together with in silico programs,early on in drug discovery. At present, the screening Ames assay (100 mm plates)requires at least 250 mg test article obtained from a compound batch. However, (a)positive finding(s) in the genotoxicity screening studiesmost of the time require(s) theprogram to stop further development of the molecule and restart the drug discoveryprocess at an earlier stage. Therefore, the possibility of identifying the genotoxicpotential of test articles prior to making a decision to prepare larger amounts ofthe test articles for in vivo toxicity assays could further minimize loss of time andresources to a considerable extent.The number of drugs and chemicals being developed that require good laboratory

practice (GLP) genotoxicity testing for regulatory submission is growing every year.There has been a simultaneous increase in the use of non-GLP and high-throughputgenotoxicity testing systems earlier in product development to screen candidate com-pounds for further development. Successful genotoxicity screening assays need tobe able to predict the results of the full GLP study that eventually will be run forregulatory submission, conserve test article, be cheap, and be run quickly. In orderto serve this expanding market, the Ames IITM assay was developed. The same cellgrowth, treatment, plating, and scoring method can be used for the standard Amesstrains. In this chapter, the Ames IITM assay as well as the Ames liquid format orfluctuation assay using the standard regulatory required S. typhimurium and E. colistrains are discussed.

10.2 AMES IITM ASSAY

10.2.1 Principle of the AmesTM II Assay

The Ames IITM assay is a second-generation bacterial reverse mutation assay devel-oped as a predictive screening assay for genotoxicity. The assay, a modification ofthe traditional “Ames assay,” was developed in Dr. Bruce Ames’ laboratory at theUniversity of California-Berkeley. The traditional Ames assay has been modified insuch a way that it can be used as a high-throughput screening assay, thus requiringvery small amounts of the test article. The assay allows one to determine the presenceor absence of mutants in each of many microwells by measuring cell growth.The Ames IITM assay can detect both frameshift mutations and base-pair sub-

stitutions without metabolic activation as well as under exogenous (S9) metabolicactivation conditions. In this assay, the frameshift mutations are detected using thetraditional S. typhimurium strain TA98. The different types of base-pair substitutionsare detected utilizing six different S. typhimurium strains specifically engineered forthe assay. Each strain carries a different missense mutation in the histidine operon thatis designed to revert uniquely to one of the six possible base substitution combina-tions causing transitions or transversions in base sequence. The six strains (TA7001,TA7002, TA7003, TA7004, TA7005, and TA7006) are combined into a single culture


FIGURE 10.1 Ames IITM assay principle.

called TAMix and treated as if it were an individual strain. The TAMix strain hasa lower spontaneous reversion frequency compared to the standard S. typhimuriumtester strains, which makes it easier to utilize in a microplate fluctuation format.These two strains (TA98 and TAMix) are plated in triplicate 384-well plates. Wellsare determined to be revertants if the indicator medium undergoes a change fromdark grey to light grey or a colony is clearly visible in the well (Fig. 10.1).

10.2.2 Strains Used in the Ames IITM Assay

The Ames IITM mixed strains (TAMix) and the S. typhimurium strain TA98 areused. The TAMix contains an equimolar mixture of the Ames IITM S. typhimuriumTA7001–TA7006 strains. Individually, these strains are designed to revert by onlyone specific base-pair substitution out of six possible changes. Thus, when mixed, allsix base substitution mutations can be represented in one culture. These strains arepatent protected and licensed for use. This assay also uses a separate culture of TA98for the detection of frameshift mutations. TA98 is one of the regulatory requiredstrains and it compliments the base-pair mutation strain TAMix.

10.2.3 Strain Genotypes

Please refer to Table 10.1.

AMES IITM ASSAY 197

TABLE 10.1 Strain Genotype for TA98 and Ames IITM

Strain Mutation Type Target Cell Walla Repairb pKM101c

Mixed strains (TAMix)

TA98 hisD3052 Frameshift GC Rfa uvrB YesTA7001 hisG1775 b.p. sub. A:T>G:C Rfa uvrB YesTA7002 hisC9138 b.p. sub. T:A>A:T Rfa uvrB YesTA7003 hisG9074 b.p. sub. T:A>G:C Rfa uvrB YesTA7004 hisG9133 b.p. sub. G:C>A:T Rfa uvrB YesTA7005 hisG9130 b.p. sub. C:G>A:T Rfa uvrB YesTA7006 hisC9070 b.p. sub. C:G>G:C Rfa uvrB Yes

aThese mutations affect the lipopolysaccharide (LPS) component of the cell envelope. These strains showan increased permeability for bulky molecules.bStrains carrying the uvrB mutation are deficient in excision repair of bulky lesions as measured by theirlack of survival following UV254 irradiation.cThe pKM101 plasmid carries the mucA and mucB genes to compensate for the weak mutagenic activitiesof the umu operon in Salmonella.b.p., base-pair substitution.

10.2.4 Validation Studies

Primary and initial validation was performed in 1998 at the National Institute ofEnvironmental Health Science (NIEHS). In this study, 30 compounds were tested,thereby yielding an 87% overall agreement on positive and negative classificationin the standard Ames assay. The high concordance with the traditional Ames testand reproducibility among cultures and replicates demonstrate the effectiveness ofAmes IITM in identifying S. typhimurium mutations [12, 13]. A further validationstudy was reported by Gervais et al. [14] at the European Environmental MutagenSociety Annual Meeting in 2003. They tested 350 compounds and found 79–85%concordance between their results, results in the standard Ames tests performed intheir laboratory, and results reported in the literature [14]. In a multicenter assessmentby Fluckiger-Isler et al. [15], 19 compounds were tested, and an interlaboratoryconsistency of 89.5% as well as a comparable sensitivity between the Ames IITM

and the traditional Ames test was reported. This study concluded that the Ames IITM

mutagenicity assay is an effective screening alternative to the standard Ames test andrequires less material and labor [15].

10.2.5 Advantages of the Ames IITM System

� Six S. typhimurium strains (TA7000 series) are combined (Ames IITM TAMix)and used as a single mixture for the rapid and effective screening of all types ofbase-pair substitutions.

� The Ames IITM assay is reproducible and quality controlled with very lowvariations from experiment to experiment, because kits produced under strictquality controls are used to perform the assays.


� The amount of test article used in the Ames IITM assay is one-twentieth or lessthan what is needed for the standard Ames assay, thus making the Ames IITM

assay optimal for the screening of chemicals and lead candidate selection in thedrug development process.

� Combining the TAMix with TA98 permits effective screening for both base-pairand frameshift mutagens with only two bacterial cultures.

� The assay is designed to be performed without and with S9 as activation systemas used in the traditional Ames test, increasing predictivity to the level of theregulatory required standard Ames test.

� Lower spontaneous reversion frequencies allow detection of mutagens at lowerconcentrations without loss of sensitivity.

� Liquid format in microplate (384-well plate) leads to increased sensitivity, easeof automation, and scoring.

� Since the 24-well and 384-well plates are used for the test article treatmentand the subsequent plating, respectively, it leads to the use of less media andplasticware.

� The amount of generated waste is significantly less than with the standard Amesassay.

10.2.6 Assay Description

Approximately 107 bacteria are exposed to six concentrations of a test agent, as wellas a positive and a negative control, for 90 min in a medium containing sufficienthistidine to support approximately two cell divisions. After 90 min, the exposurecultures are diluted in a pH indicator medium lacking histidine, and each replicateculture is aliquoted into 48 wells of a 384-well plate. Within 2 days, cells that haveundergone the reversion to his + will grow in the liquid media. Metabolism by thebacterial growth reduces the pH of themedium, changing the color of the pH indicatorof that well from dark grey to light grey. The number of wells containing mutants arecounted for each dose and compared to vehicle control wells containing spontaneousmutants. Each dose is tested in triplicate to allow for statistical analysis of the data(if performed). An increase in the number of revertant colonies upon exposure to atest chemical relative to the vehicle (zero-dose) controls indicates that the chemicalis mutagenic in the Ames IITM assay.

10.2.7 Detailed Method

10.2.7.1 Preparation of TAMix and TA98 Strain Aliquots Since these strains arepatent protected, they can only be purchased as part of a kit. The process for preparingthe strain is that the 7000 series strains are grown individually for approximately 12 h.Thereafter, the optical density of the cultures is measured and adjusted to be the samefor all six strains. These strains are mixed in equimolar concentrations to prepare themixed strain called TAMix. The TAMix strain is good for just one assay; thus, it has

AMES IITM ASSAY 199

to be purchased as part of a kit. The other strain used in the Ames IITM assay, TA98, isa strain that is used in the standard Ames assay and can be purchased, cryopreserved,and used whenever needed. The kits for performing the Ames IITM assay are availablefrom different vendors such as Molecular Toxicology, Inc. [16] or Xenometrix [17].

10.2.7.2 Ames IITM Assay Kit The assay kit contains the tester strains, growth,exposure, and indicator media. The plasticware can be purchased as part of the kit orfrom other labware vendors. In order to perform the assay, the laboratory should beequippedwith an orbital shaker incubator, a 37◦Cdry incubator, biological safety cab-inets (hoods), a spectrophotometer, vortex mixers, assorted large and micropipettes,and multichannel repeater micropipettes.

10.2.7.3 Day 1 Overnight Culture Preparation One vial each of TA98 and TAMixis removed from the−80◦C freezer. The vials are allowed to thaw at room temperatureprior to initiating cultures. Ten milliliters of Oxoid Broth containing 25–50 �g/mLampicillin (Molecular Toxicology, Inc.) is added to 50 mL centrifuge tubes. Tenmilliliters of the thawed strain (TAMix or TA98) are added directly to the mediain each appropriately labeled tube. One tube without any bacteria is to be used asnegative control. Starting of the overnight cultures should be done late in the eveningfor the cultures to be used early in the morning. The tube caps are kept loose to allowsufficient aeration in the incubator. The tubes are placed in an environmental shakerincubator set to 37 ± 1◦C and 250 rpm for approximately 12 h.

Note: As the cultures increase in age, the number of spontaneous revertantsincreases. For assays to be performed later than the incubation stop time, the culturesshould be stored at room temperature until ready to use. Cultures that are more than24-h old should not be used in the experiment.

10.2.7.4 Determination of the Optical Density (OD600) Values of OvernightCultures The overnight cultures are diluted 1:10, and then the OD readings aremeasured at 600 nm. The OD600 reading of each sample containing the overnightcultures and negative control is taken. Each OD600 reading is multiplied by 10 toobtain the actual optical density. It should be made sure that the OD600 values forTAMix and TA98 are at least 2.0, and that the OD600 value of the negative control is∼0.0. If the overnight cultures do not meet the acceptable range criteria, the culturescan be returned to the incubator for additional time. Total incubation time shouldnot exceed 24 h. If the negative control OD600 value is significantly higher than 0.0,then the possibility exists that contamination has occurred. In this case, the overnightcultures cannot be used and must be discarded. New cultures for TA98 and TAMixshould be started.

10.2.7.5 Test Article Dosing Solution Preparation Test article can be measuredby weight on an analytical balance. Liquid test articles may be measured by volume.Unless there are solubility limitations, the maximum exposure concentration can goup to 5 mg/mL (5 �L/mL for liquid test articles measured by volume). However, in


TABLE 10.2 Ames IITM Assay Plate Setup

Vehicle 100 Vehicle 100 Vehicle 100

3.3 333 3.3 333 3.3 33310 1000 10 1000 10 100033 Positive 33 Positive 33 Positive

most cases, the test article concentration of 1.0 mg/mL is used as the highest concen-tration tested. Test article solutions are prepared in solvent at 25× final concentrationtested. The dilution scheme is carried out by first adding the required volume of sol-vent to the second through last dilution tubes, and then adding the required volumeof solvent to the dilution tube containing the weighed or measured sample of the testarticle. The serial dilution is carried out by transferring the required aliquot to thenext dilution tube. The contents of the dilution tube are vortex mixed immediatelyafter addition of the dilution sample, and a different pipette is used to transfer eachsample. The condition of the sample in each dilution tube, that is, clear solution,workable suspension, or insoluble, should also be recorded. This information can behelpful in interpreting the assay results.

10.2.7.6 Assay Format The assay is performed at six test article concentrations(approximately half log apart) with three replicates each and with as well as withoutexogenous metabolic activation. Three replicates of the vehicle (used to solubilize thetest article) and positive controls without metabolic activation are also included. Forexample, if the highest test article concentration tested is 1000 �g/mL, then the lowerconcentrations can be 333, 100, 33, 10, and 3.3 �g/mL. Alternative concentrationssuch as 1000, 500, 250, 125, 62.5, and 31.3 �g/mL can also be used. An example ofthe 24-well plate setup is shown in Table 10.2.The same setup is usedwith andwithoutmetabolic activation. The positive controls

used are: 4-nitroquinoline (4-NQO) at 1.0 �g/mL + 2-nitrofluorene (2-NF) at 2.0�g/mL and 2-aminoanthracene (2-AA) at 5.0 �g/mL with metabolic activation.

10.2.7.7 Preparation of 30% S9 Mix The Ames IITM assay is performed withand without metabolic activation. For the exogenous metabolic activation, Aroclor1254-induced rat liver S9 is used. The S9 homogenate is prepared frommale Sprague-Dawley rats induced with a single intraperitoneal injection of Aroclor 1254 or a mix-ture of phenobarbital and benzoflavone 5 days prior to sacrifice. The S9 homogenatecan be purchased (Molecular Toxicology, Inc. or other vendors) and stored frozen at≤ –60◦C until use. Each batch of S9 homogenate is assayed by the vendor for itsability to metabolize 2-AA and 7,12-dimethylbenzanthracene to forms mutagenic toS. typhimurium strain TA100.Immediately prior to use, the S9 is thawed and mixed with a cofactor pool to

contain 30% S9 homogenate. The S9 mix is prepared on the day of use by combiningthe volumes of reagents listed below in a sterile tube. The S9 mix should be kept onice until used in the assay (Table 10.3).

AMES IITM ASSAY 201

TABLE 10.3 Instructions for Preparing the S9Mixture for the Ames II Assay

Reagent Volume

1.00 M KCl 0.166 mL0.25 M MgCl2·6H2O 0.160 mL0.20 M glucose-6-phosphate 0.126 mL0.04 M NADP+ 0.500 mL0.20 M NaH2PO4 buffer, pH 7.4 2.500 mLSterile water 0.050 mLS9 rat homogenate 1.500 mL

10.2.7.8 For Cultures without S9 Activation (Enough Volume for One TestArticle) To a 50 mL conical tube, 7.2 mL of the exposure medium (MolecularToxicology, Inc. or other vendors) and 800 �L of the appropriate overnight cul-ture are added and mixed well. Immediately after mixing, 240 �L of this exposuremedium–culture mix is dispensed in each well of the labeled 24-well plate. Theappropriate test article dosing solution of 10 �L is added to each well of the 24-wellplate. The 24-well plate is secured to the base of an environmental shaker incubator.The plate is incubated for approximately 90 min at 37 ± 1◦C and 250 rpm. The sameprocess is repeated with the second strain (Table 10.3).

10.2.7.9 For Cultures with S9 Activation (Enough Volume for One Test Article)To a 50 mL conical tube, 6.0 mL of the exposure medium, 800 �L of the appropriateovernight culture, and 1.2 mL of the cold 30% S9 mix are added and mixed well.Immediately, 240 �L of the exposure medium–culture–S9 mix are added to all wellsof the labeled 24-well plate. The appropriate test article dosing solution of 10 mLis transferred to each well of the 24-well plate. The 24-well plate is secured to thebase of an environmental shaker incubator. The plate is incubated for approximately90 min at 37 ± 1◦C and 250 rpm. The same procedure should be followed for thesecond strain.

10.2.7.10 Transfer to 384-Well Plates After the 90-min incubation, the 24-wellplates are removed from the environmental shaker incubator. For each 24-well plate,three 384-well plates should be labeled with the assay information. The 384-wellplate orientation should be checked and made sure it is correct.To each well of the 24-well plates, 2.8 mL of the indicator medium (Molecular

Toxicology, Inc. or other vendors) is added. The 24-well plate should be placed nextto the first 384-well plate. The cover of the 24-well plate should be removed to exposethe first column of wells. With a repeating pipette, the contents in the well shouldbe mixed by gently pipetting up and down. Of the contents from the 24-well plate,50 �L aliquots should be added into 48 wells of the 384-well plate. Two columns ofthe 24-well plate will fill the 384-well plate. Columns 1 and 2 from the 24-well plateare aliquoted into 384-well plate replicate number 1. Columns 3 and 4 are aliquoted


FIGURE 10.2 Scheme showing the different steps of the Ames IITM assay.

into 384-well plate replicate number 2. Columns 5 and 6 are aliquoted into 384-wellplate replicate number 3. When the 384-well plates are filled, the cover should bereplaced and carefully placed into a sealable plastic bag. The plastic bags preventevaporation during the incubation period. The –S9 activation and the + S9 activationplates should be placed into separate bags. After filling the bags, these are placed ina 37◦C dry incubator for 2 days.

10.2.7.11 Plate Scoring and Calculations After removing the bags from the incu-bator, the three replicate 384-well plates for each strain are scored by placing themon top of a light box. The number of positive wells in each 48-well section is counted,and the data are recorded. Positive wells are those that have turned yellow or have abacterial colony visible at the bottom of the plate. A diagram that shows the wholeassay scheme is presented in Figure 10.2.Once all the plates have been scored, the mean number of positive wells per

concentration (i.e., the average of the positive wells for the three replicates for eachdose) as well as the standard deviation of positive wells per concentration (i.e.,the standard deviation values for the mean number of positive wells) is calculated.Thereafter, the “activity by fold induction” (the increase in the number of positivewells in test article-treated wells above the solvent control [spontaneous revertants])is determined. Note: if the mean number of positive wells for the solvent control isless than 1.0, the value is set to 1.0 for this calculation.

Criteria for a valid assay: The positive and vehicle control values should bewithin the ranges shown below. However, the range may be slightly different forthe different laboratories and may change, based on the laboratory historical vehiclecontrol data (Table 10.4).

10.2.7.12 Evaluation Criteria For a test article to be evaluated positive, it mustcause a concentration-dependent significant increase (at least doubling) in the meannumber of positive wells per concentration over a minimum of two increasing

AMES LIQUID FORMAT ASSAY 203

TABLE 10.4 Positive and Negative Control Acceptable Range in Ames II Assay

Strain Positive Controls

Acceptable Range ofpositive Wells per48-Well Section

TAMix No dose ≤54-NQO (1.0 �g/mL) + 2NF (2.0 �g/mL) ≥252-AA (5 �g/mL +S9) ≥15

TA98 No dose ≤84-NQO (1.0 �g/mL) + 2NF (2.0 �g/mL) ≥252-AA (5 �g/mL +S9) ≥25

2-AA, 2-aminoanthracene; 2NF, 2-nitrofluorene; 4-NQO, 4-nitroquinoline.

concentrations of test article, and the increases should also be at least double (≥twofold) the induction compared to negative control. In the experiments where theaverage vehicle control value for the positive wells is less than one, a value of one(1.0) should be substituted for the vehicle control when calculating the fold increasewith the test article and the positive control-treated concentrations. The test articleshould be considered nonmutagenic if the mean number of positive wells for all testerstrains is below the maximum historical negative control range under all experimentalconditions.

10.3 AMES LIQUID FORMAT ASSAY

The Ames, S. typhimurium and E. coli reverse mutation or bacterial reverse mutationassay is the most commonly used assay in genetic toxicology. It is a very well-defined regulatory required assay. The guidelines (FDA, OECD, and ICH guidelines)describe the assaymethod and the types of strains to be used in the plate incorporationor preincubation method of treatment. These strains can also be used in the liquidformat or fluctuation method.There is a need to find out a test article’s mutagenic and genotoxic potential as

early as possible in the drug development cycle. However, at the beginning of drugdevelopment during the discovery phase, the test article is available in a limitedquantity. In order to perform GLP assay predictive screening, there is a need to scaledown the full-scale GLP assay to a level where the amount of compound needed isvery small (5–10 mg) without loss of the assay’s predictive qualities. These screeningassays should be predictive of the standard GLP assays, since most of the time thedecision to develop a drug or not depends on the early screening assays.The Ames IITM assay is one of the screening assays available. However, the Ames

IITM assay only uses one regulatory required strain (TA98), and the other strains usedin the assay are not regulatory required and are patent protected. Thus, the Ames IITM

assay kits have to be purchased to perform the assay.


TABLE 10.5 Genotype for S-typhimurium and E. coli Strains

Histidine Mutation Additional MutationsTryptophan Mutation

hisG46 hisC3076 hisD3052 trpE65 LPS Repair R-factor

TA1535 TA1537 — — rfa �uvrB —TA100 — TA98 — rfa �uvrB +R— — — WP2uvrA — �uvrA —

WP2uvrApkm101 — �uvrA +R

The Ames liquid format assay can be performed with the following strains:S. typhimurium TA98, TA100, TA1535, TA1537 and E. coli strain WP2uvrA orWP2pkm101. All these strains have relatively low number of background revertants,thus making them ideal for the liquid format assay. The reagents and the strains canbe purchased from Molecular Toxicology, Inc. or other vendors. The tester strainscan be propagated and cryopreserved for future use and only the liquid media needto be purchased.The tester strains include the S. typhimurium histidine auxotrophs TA98, TA100,

TA1535, and TA1537 as described by McCann and Ames [4] and the E. coli testerstrainsWP2uvrA andWP2pkm101 as described by Green andMuriel [6]. The detailsregarding the genetic makeup of these strains are summarized in Table 10.5.In addition to a mutation in the histidine operon, each S. typhimurium tester strain

contains additional mutations that enhance sensitivity to some mutagens. The rfamutation results in a cell wall deficiency that increases the permeability of the cell tocertain classes of chemicals such as those containing large ring systems that wouldotherwise be excluded. The deletion in the uvrB gene results in a deficient DNAexcision repair system. Tester strains TA98 and TA100 also contain the pKM101plasmid (carrying the R-factor). It has been suggested that the plasmid increasessensitivity to mutagens by modifying an existing bacterial DNA repair polymerasecomplex involved with the mismatch repair process.TA98 and TA1537 are reverted from histidine dependence (auxotrophy) to histi-

dine independence (prototrophy) by frameshift mutagens. TA100 is reverted by bothframeshift and base substitution mutagens and TA1535 is reverted only by mutagensthat cause base substitutions.The E. coli tester strains have an AT base pair at the critical mutation site within

the trpE gene [18]. Tester strainWP2uvrA has a deletion in the uvrA gene resulting ina deficient DNA excision repair system, whereas the E. coli strain WP2uvrApkm101contains the pKM101 plasmid (carrying the R-factor) similar to TA98 and TA100,thus rendering it more sensitive to the mutagens.Tryptophan revertants can arise due to a base change at the originally mutated site

or by a base change elsewhere in the chromosome, causing the original mutation to besuppressed. Thus, the specificity of the reversion mechanism is sensitive to base-pairsubstitution mutations [7].The S. typhimurium tester strains initially came from Dr. Bruce Ames, University

of California, Berkeley, CA. The E. coli tester strains can be purchased from the


National Collection of Industrial and Marine Bacteria (Aberdeen, Scotland, UK).The tester strains may also be obtained from Molecular Toxicology, Inc. or othervendors using cultures derived from the above sources.

10.3.1 Validation Studies

Primary and initial validation was performed in 1998 at the National Institute ofEnvironmental Health Science (NIEHS). In this study 30 compounds were tested,thereby yielding an 87% overall agreement on positive and negative classification inthe standard Ames assay. The high concordance with the traditional Ames test andreproducibility among cultures and replicates demonstrate the effectiveness of theAmes liquid format assay in identifying Salmonella mutations. In this experiment,TA98 andTA1537, the strains from the standardAmes assay, were also included alongwith the TAMix strain [12]. The concordance was >79% between the liquid formatassay and the standard format Ames assay. Further publications [13, 14] have alsoshown similar results regarding the assay concordance. Over the years, a number ofposters have been presented by different CROs (i.e., Xenometrix andAniara) showingthe concordance between the standard plate-based Salmonella reverse mutation assayand the liquid format assay using the same strains.

10.3.2 Advantages of the Ames Liquid Format Assay

Essentially the advantages are the same as those of the Ames IITM assay, with theexception that here the regulatory required strains are in use:

� The regulatory required strains are used, so that the assay is more predictive ofthe GLP Ames assay.

� The liquid format assay is reproducible and quality controlled with very lowvariations from experiment to experiment, because prepared media with qualitycontrol are used to perform the assays.

� The amount of test article used in the Ames liquid format assay is one-twentiethor less than what is needed for the standard Ames assay, thus making this assayvery desirable for use in the screening of chemicals and lead candidate selec-tion in the drug development process. For screening purposes, sometimes onlytwo strains, TA98 (frameshift mutation) and TA100 (base-pair and frameshiftmutations), are used. Thus, the amount of test article is exactly the same as inthe Ames IITM assay.

� The assay is designed to be performed under the same conditions (without andwith S9 activation) as used in the traditional Ames test, increasing predictivityto the “gold standard” regulatory required standard Ames test.

� Lower spontaneous reversion frequencies allow detection of mutagens at lowerconcentrations without loss of sensitivity.

� Liquid format in microplate (384-well plate) leads to increased sensitivity, easeof automation, and scoring.


� Because the 24-well and 384-well plates are used for the test article treatmentand then for plating, respectively, it leads to use of less media and plasticware.

� The amount of generated waste is a lot less than the standard Ames assay.

10.3.3 Assay Description

In the regular Ames liquid format assay as in the Ames IITM assay, bacteria areexposed to at least six concentrations of a test article as well as to a positive and anegative control for 90 min in a medium containing sufficient histidine to supportapproximately two cell divisions (exposure media) (Molecular Toxicology, Inc. orother vendors). The number of bacteria for some strains such as TA100 and E.coli (WP2uvrA and WP2uvrApkm101) needs to be adjusted. The reason is thatthese strains have a moderately high background revertant count. Thus, if testedwithout diluting, the number of positive wells could be too high to be able to make ameaningful evaluation. After 90min of incubation, the exposure cultures are diluted inpH indicator medium (Molecular Toxicology, Inc. or other vendors) lacking histidine(for Salmonella cultures) or tryptophan (for E. coli cultures), and each replicateculture is aliquoted into 48 wells of a 384-well plate. Within 2 days, cells that haveundergone the reversion to his + or trp + (in the case of S. typhimurium or E.coli strains, respectively) will grow in the liquid media. Due to growth, metabolicprocesses within the bacteria are enhanced, thereby reducing the pH of the mediumand changing the color of the pH indicator (bromocresol purple) of that well fromdark grey to light grey color. The number of wells containing reverse mutants iscounted for each dose and compared to vehicle control wells containing spontaneousmutants. Each dose is tested in triplicate to allow for statistical analysis of the data(if performed). An increase in the number of revertant colonies upon exposure to atest article (chemical) relative to the vehicle (zero-dose) controls indicates that thechemical is mutagenic in the Ames liquid format assay.

10.3.4 Detailed Method

10.3.4.1 Preparation and Culture of Bacterial Strains, Preparation of Test ArticleDilutions Preparation of the tester strains and culture of S. typhimurium TA100,TA1535, TA1537 as well as E. coli WP2uvrA or WP2uvrApkm101 are performedas described for tester strain TA98 (see Ames IITM, Section 10.2.7.3). The growthmedium is the same for all the cultures. The determination of the optical density andthe preparation of the different test article concentrations to be used are the same aspreviously described for the Ames IITM assay.

10.3.4.2 Positive Controls Used The positive controls used are summarized inTable 10.6.

10.3.4.3 Preparation of 30% S9 Mix The Ames liquid format assay is performedwith and without metabolic activation. For the exogenous metabolic activation, Aro-clor 1254- or phenobarbital- and �-naphthoflavone-induced rat liver S9 is used. TheS9 homogenate is prepared from male Sprague-Dawley rats induced with a singleintraperitoneal injection of Aroclor 1254 or phenobarbital and �-naphthoflavone 5


TABLE 10.6 Recommended Control Concentrations

Strain S9 Positive Control Concentration (�g/mL)

Salmonella strains Rat liver 2-aminoanthracene 5.0WP2uvrA 50TA98 and TA100 None Mixture of 2-nitrofluorene

and 4-nitroquinoline2.0 and 1.0

TA1535 N4-aminocytidine 100TA1537 9-aminoacridine 15WP2uvrA 4-nitroquinoline 1.0

days prior to sacrifice. The S9 homogenate can be purchased and stored frozen at≤ —60◦C until use. Each batch of S9 homogenate is assayed by the vendor for itsability to metabolize 2-aminoanthracene and 7,12-dimethylbenzanthracene to formsmutagenic to S. typhimurium TA100.Immediately prior to use, the S9 is thawed and mixed with a cofactor pool to

contain 30% S9 homogenate. The S9 mix is prepared on the day of use by combiningthe volumes of reagents listed below in a sterile tube and should be kept on ice untilused in the assay (Table 10.7)

10.3.4.4 For S. typhimurium Cultures without S9 Activation (Enough Volumefor One Test Article) To a 50 mL centrifuge tube, 7.2 mL of Salmonella exposuremedium and 800 �L (1:10 dilution) of the appropriate overnight culture (TA98,TA1535, and TA1537) are added. For TA100 strain, use 400 �L of diluted culture(1:20 dilution) [17]. The tube contents should be mixed well. Immediately aftermixing, 240 �L of this Salmonella exposure medium–culture mix is added to allwells in the labeled 24-well plate. The appropriate test article dosing solution of10 �L is added to each well of the 24-well plate. The 24-well plate is secured to thebase of an environmental shaker incubator. The plate is incubated for approximately90 min at 37 ± 1◦C and 250 rpm.

Note: The tester strain TA100 needs to be diluted to the appropriate concentrationin order to achieve the background positive well count of ≤12. Normally, a dilutionof 1:20 of the overnight stock in the exposure medium gives the number of positivewells in the appropriate range. The overnight culture can also be diluted 1:4 in the

TABLE 10.7 Instructions for Preparing S9 Mix for the Ames Liquid Format Assay

Reagent Volume for Five Strains (mL)

1.00 M KCl 0.2060.25 M MgCl2·6H2O 0.1980.20 M glucose-6-phosphate 0.1560.04 M NADP 0.6200.20 M NaH2PO4 buffer, pH 7.4 3.162S9 rat homogenate 1.860


growth medium and grown for approximately one more hour to get it in the log phasegrowth prior to using it in the assay.

10.3.4.5 For S. typhimurium Cultures with S9 Activation (Enough Volume forOne Test Article) To a 50 mL centrifuge tube, 6.0 mL of exposure medium and 800�L (1:10 dilution) of the appropriate overnight culture (TA98, TA1535, and TA1537)are added. For TA100, only 400 �L (1:20 dilution) should be added [17]. In the sametube, 1.2 mL of the already prepared cold 30% S9 mix is dispensed. Contents of thetube are mixed well. Immediately after mixing, 240 �L of the exposure medium–culture–S9 mix is dispensed into all wells of the labeled 24-well plate. The 24-wellplates are treated with 10 �L of the appropriate test article dosing solution. The24-well plate is secured to the base of an environmental shaker incubator. The plateis incubated for approximately 90 min at 37 ± 1◦C and 250 rpm.

10.3.4.6 For E. coli Cultures without S9 Activation (Enough Volume for OneTest Article) Into a 50 mL centrifuge tube, 7.2 mL of E. coli exposure medium isdispensed. To the same tube, 240 �L (1:33 dilution) of the WP2uvrApkm101 strainand 560 �L (1:14.3 dilution) of the WP2uvrA strain overnight cultures are added[17]. The tube contents are mixed well. Immediately after mixing, 240 �L of this E.coli exposure medium–Culture mix is dispensed into all wells of the labeled 24-wellplate. The wells are treated with 10 �L of the appropriate test article dosing solutionin each well of the 24-well plate. The 24-well plate is secured to the base of anenvironmental shaker incubator and incubated for approximately 90 min at 37 ±1◦C and 250 rpm.

10.3.4.7 For E. coli Cultures with S9 Activation (Enough Volume for One TestArticle) Into a 50 mL centrifuge tube, 6.0 mL of exposure medium is dispensed. Inthe same tube, 240 �L (1:33 dilution) of the WP2pkm101 strain and 560 �L (1:14.3dilution) of the WP2uvrA strain overnight cultures are added [17]. In the tube, 1.2mL of the already prepared cold 30% S9 mix is also dispensed. The contents in thetube are mixed well. Immediately after mixing, 240 �L of the exposure medium–culture–S9 mix is added to all wells of the labeled 24-well plate. The wells are treatedwith 10 �L of the appropriate test article dosing solution in each well of the 24-wellplate. The 24-well plate is secured to the base of an environmental shaker incubatorand incubated for approximately 90 min at 37 ± 1◦C and 250 rpm.

10.3.4.8 Transfer to 384-Well Plates After the 90-min incubation, the 24-wellplates are removed from the environmental shaker incubator. For each 24-well plate,three 384-well plates should be labeled with the assay information. To each well ofthe 24-well plates, 2.8 mL of appropriate indicator medium (S. typhimurium or E.coli) is added. With a repeater micropipette, the contents are mixed in the well bygently pipetting up and down. From 1 well of the 24-well plate, 50 �L aliquots ofthe contents are dispensed to 48 wells of the 384-well plate. Two columns (8 wells)of the 24-well plate will fill the whole 384-well plate. Columns 1 and 2 from the24-well plate are aliquoted into 384-well plate replicate number 1. Columns 3 and 4


FIGURE 10.3 Scheme showing the different steps of the Ames liquid format assay.

are aliquoted into 384-well plate replicate number 2. Columns 5 and 6 are aliquotedinto 384-well plate replicate number 3. When the 384-well plates are filled, replacethe cover and carefully place the 384-well plates into a sealable plastic bag. Theplastic bags prevent evaporation during the incubation period. The –S9 activation andthe +S9 activation plates should be placed into separate bags. Automated dispenserscan also be used for this transfer process from the 24-well plate to the 384-well plate.A scheme of the whole Ames liquid format assay is shown in Figure 10.3. Afterfilling the bags, these are placed in a 37◦C dry incubator for 2 days.

10.3.4.9 Plate Scoring and Calculations After removing the bags from the incu-bator, the three replicate 384-well plates (with and without S9) are scored for eachstrain/treatment condition by placing them on top of a light box. The number ofpositive wells is counted in each 48-well section, and the data are recorded. Positivewells are those that have turned yellow or have bacterial colony visible at the bottomof the plate.Once all the plates have been scored, the mean number of positive wells per

concentration (i.e., the average of the positive wells for the three replicates for eachdose) as well as the standard deviation of positive wells per concentration (i.e.,the standard deviation values for the mean number of positive wells) is calculated.Thereafter, the “activity by fold induction” (the increase in the number of positivewells in test article-treated wells above the solvent control [spontaneous revertants])is determined. Note: if the mean number of positive wells for the solvent control isless than 1.0, the value is set to 1.0 for this calculation.

Criteria for a valid assay: The positive and vehicle control values should bewithin the range specified in Table 10.8.

10.3.4.10 Evaluation Criteria For a test article to be evaluated positive, it mustcause a concentration-related significant increase (at least doubling) in the mean


TABLE 10.8 Positive and Negative Control Acceptable Range inAmes Liquid Format Assay

Strain Positive ControlsAcceptable Range of Positive

Wells per 48

TA98 Vehicle control ≤82NF (2.0 �g/mL) ≥252-AA (5 �g/mL +S9) ≥25

TA100 Vehicle control ≤124-NQO (1.0 �g/mL) ≥252-AA (5 �g/mL +S9) ≥25

TA1535 Vehicle control ≤8N4-ACT (100 �g/mL) ≥252-AA (5 �g/mL +S9) ≥25

TA1537 Vehicle control ≤89AA (15 �g/mL) ≥252-AA (5 �g/mL +S9) ≥25

E. coliWP2uvrA Vehicle control ≤124-NQO (1.0 �g/mL) ≥252-AA (50 �g/mL +S9) ≥1.5 to twofold over vehicle

controla

E. coliWP2uvrApkm101 Vehicle control ≤124-NQO (1.0 �g/mL) ≥252-AA (50 �g/mL +S9) ≥1.5 to twofold over the vehicle

controla

2-AA, 2-aminoanthracene; 2NF, 2-nitrofluorene; 4-NQO, 4-nitroquinoline; N4-ACT, N4-aminocytidine.aThe mutagenic response in E. coli with 2AA is lower than in S. typhimurium.

number of positive wells per concentration over a minimum of two increasing con-centrations of test article, and the increases should also be at least double (≥ twofold)the induction compared to negative control. In the experiments where the averagevehicle control value for the positive wells is less than one, a value of one (1.0)should be substituted for the vehicle control when calculating the fold increasewith the test article and the positive control-treated concentrations. The test articleshould be considered nonmutagenic if the mean number of positive wells for all testerstrains is below the maximum historical negative control range under all experimentalconditions.

REFERENCES

1. Ames, B.N., McCann, J., Yamasaki, E. (1975). Methods for detecting carcinogensand mutagens with the Salmonella/mammalian microsome mutagenicity test. MutationResearch, 31, 347–364.

2. Slater, E.E., Anderson, M.D., Rosenkranz, H.S. (1971). Rapid detection of mutagens andcarcinogens. Cancer Research, 31, 970–973.

REFERENCES 211

3. Bridges, B.A. (1972). Simple bacterial systems for detectingmutagenic agents.LaboratoryPractice, 21, 413–416.

4. McCann, J., Ames, B.N. (1976). Detection of carcinogens as mutagens in the Salmonella/microsome test: assay of 300 chemicals: discussion. Proceedings of the National Academyof Sciences USA, 73, 950–954.

5. McCann, J., Choi, E., Yamasaki, E., Ames, B.N. (1975). Detection of carcinogens asmutagens in the Salmonella/microsome test: assay of 300 chemicals. Proceedings of theNational Academy of Sciences USA, 72, 5135–5139.

6. Maron, D.M., Ames, B.N. (1983). Revised methods for the Salmonellamutagenicity test.Mutation Research, 113, 173–215.

7. Green, M.H.L., Muriel, W.J. (1976). Mutagen testing using trp+ reversion in Escherichiacoli. Mutation Research, 38, 3–32.

8. Ames, B.N., McCann, J., Yamasaki, E. (1975). Methods for detecting carcinogensand mutagens, with the Salmonella/mammalian-microsome mutagenicity test. MutationResearch, 31, 347–364.

9. Ames, B.N., Durston, W. E., Yamasaki, E., Lee, F.D. (1973). Carcinogens are mutagens: asimple test system combining liver homogenates for activation and bacteria for detection.Proceedings of the National Academy of Sciences USA, 70, 2281–2285.

10. Putman, D.L., Clarke, J.J., Escobar, P., Gudi, R., Krsmanovic, L.S., Pant, K, Wagner,V.O., San, R.H.C, Jacobson-Kram, D. (2006). Genetic toxicity. In: Toxicological TestingHandbook, Principles, Applications and Data Interpretation, 2nd ed. (Jacobson-Kram,D., Keller, K.A., Eds.). Informa Healthcare: New York; pp. 185–248.

11. OECD Guideline for Testing of Chemicals, Test 471: Bacterial reverse mutation test,adopted July 21, 1997.

12. Gee, P., Sommers, C.H., Melick, A.S., Gidrol, X.M., Todd, M.D., Burris, R.B., Nel-son, M.E., Klemm, R.C., Zeiger, E. (1998). Comparison of responses of base-specificSalmonella tester strains with the traditional strains for identifying mutagens: the resultsof a validation study.Mutation Research, 412, 115–130.

13. Gee, P., Maron, D.M., Ames, B.N. (1994). Detection and classification of mutagens: Aset of base-specific Salmonella tester strains. Proceedings of the National Academy ofSciences USA, 91, 11606–11610.

14. Gervais, V., Bijot, D., Claude, N. (2003). Assessment of a screening experience with theAmes IITM test and future prospects. European Environmental Mutagen Society AnnualMeeting: From Hazard to Risk, Aberdeen, p. 120.

15. Fluckiger-Isler, S., Baumeister, M., Braun, K., Gervais, V., Hasler-Nguyen, N., Reimann,R., van Gompel, J., Wunderlich, H.G., Engelhardt, G. (2004). Assessment of the perfor-mance of the Ames IITM assay: a collaborative study with 19 coded compounds.MutationResearch, 558, 181–197.

16. Molecular Toxicology, BioRelianceAmes IITM MutagenicityAssayUserManual, Novem-ber 2008.

17. Ames MPFTM Penta I, Microplate Format Mutagenicity Assay, Xenometrix, Aniara, Ver-sion 3.12_S, November 2009.

18. Wilcox, P., Naidoo, A., Wedd, D.J., Gatehouse, D.G. (1990). Comparison of Salmonellatyphimurium TA102 with Escherichia coliWP2 tester strains.Mutagenesis, 5, 285–291.

11HIGH-THROUGHPUT BACTERIALMUTAGENICITY TESTING:VITOTOXTM ASSAY

Luc Verschaeve

11.1 INTRODUCTION

Assessing genotoxicity is an important aspect of both drug development and chemicalsafety testing (particularly under the REACH program). New pharmaceuticals andother chemicals need to be tested for their toxicological properties, including theirgenotoxic potential. This is done using a test battery as no single test is sufficient todetermine whether the agent is genotoxic or not. This is, amongst others, becausethere are three major endpoints of genetic damage associated with human disease–gene mutation, clastogenicity, and aneuploidy [1]. A single test usually detects onlyone, seldom two of these endpoints. For product registration, tests should be doneaccording to the principles of good laboratory practice (GLP) using a number of vali-dated, recommended, and approved test systems [2]. For other purposes, for example,prescreening [3] or environmental monitoring [4], other tests can be used. As mosttests are time consuming, labor intensive, and therefore expensive, there is a continu-ing search for new short-term and preferentially high-throughput screening tests. Thisis for example of great importance in the “discovery phase” of a new molecule (e.g.,in the case of new medication), where an early selection of candidate compounds forfurther research is desirable. Indeed, despite the efforts by the pharmaceutical indus-try over the last 15 years, some 50% of new chemical entities still fail due to toxicity,which represents a major cost to industry, particularly in the areas of hepatotoxicity,cardiotoxicity, skin toxicity, central nervous system side effects, genotoxicity, and


213

214 HIGH-THROUGHPUT BACTERIAL MUTAGENICITY TESTING

carcinogenicity. The preclinical cost is even more significant, as safety screening isoften the final hurdle in drug discovery before the new chemical entity enters theclinic [3]. High-throughput screening tests that allow a rapid (geno)toxicity evalua-tion are thus of uttermost importance. Such high-throughput screening tests shouldalso allow testing of compounds in milligram amounts because availability of thecompound at this stage is still very limited.Many new developments concern genetically engineered microorganisms such as

bacteria or yeasts [5]. Easy access and simple procedures as well as rapidity and lowcosts are some of the advantages and reasons to use these organisms. Another reasonis of course that genotoxicity data obtained in microorganisms (e.g., using the Amesassay [6, 7]) are also relatively reliable to predict effects in humans. Assays based onmicroorganisms rely either on reversion principles similar to those of the Ames test,or on promoter–reporter fusions that generate a quantifiable dose-dependent signal inthe presence of DNA-damaging compounds and the induction of repair mechanisms.The SOS chromotest [8–10] and the Umu-C test [11] are assays that are based on thelatter. The more recent Vitotox test is another example. It allows the rapid and cost-effective screening for genotoxicity based on the bacterial SOS response mechanism[12, 13]. This test also requires minimal amounts of a test compound, which givesit an advantage over most other test systems. Compared to the Ames test (still the“golden standard” in genetic toxicology), the assay is much faster and correlates toa greater degree with high-throughput screening procedures. Moreover, as for mostother assays, S9 liver fractions from Aroclor-pretreated rats can be used. In this way,potential genotoxic metabolites can be investigated as well.

11.2 PURPOSE OF THE VITOTOX TEST

The Vitotox test was developed with the purpose of obtaining a test that is superiorto the already existing tests in many ways:

1. The test should allow (geno)toxic products to be detected in a simple way.

2. The test should be able to detect genotoxic compounds in a short period of time(for example, a couple of hours at the most).

3. The test should need only minimal amounts of a test compound.

4. The test should correlate well with the Ames assay and show high specificityand sensitivity. It should be able to measure the kinetics of induction of theSOS system as a function of genotoxicity as this can be an important parameterfor the comparison of different genotoxic compounds.

11.3 PRINCIPLE OF THE VITOTOX TEST

The Vitotox test is not much different from the SOS chromotest or Umu-C test inthe sense that it is based on the same principle but utilizes another reporter gene and

CONSTRUCTION OF recN-luxCDABE FUSIONS 215

another SOS promoter. The test is performed with Salmonella typhimurium TA104cells that show DNA damage more directly than for example the Ames test does.The bacteria contain a luciferase gene under transcriptional control of a mutatedrecN promoter (TA 104-recN strain or genox strain). Normally this promoter isstrongly repressed, but in the presence of a DNA-damaging genotoxic compoundthe recA regulator protein recognizes the resultant free ends or mismatches in DNA.This results in a cascade of reactions leading to depression of the strong recN pro-moter and emission of light that can easily be measured using a luminometer. Asit was realized that some compounds may act directly on the luciferase gene (inthe absence of genotoxicity), a second S. typhimurium TA 104 strain (TA104 pr1 orcytox strain) constitutively expressing luciferase is used as a control and to measurecytotoxicity.Genotoxicity is thus expressed as light production and two different recombinant

Salmonella typhimurium TA104 test strains carrying a luciferase operon are used todetermine the genotoxicity and cytotoxicity of the sample:

1. RecN-luxCDABE fusion on a multicopy plasmid allows genotoxicity detection:increased light production is a function of genotoxicity at sublethal concentra-tions.

2. Pr1-luxCDABE fusion on a multicopy plasmid allows cytotoxicity to bedetected and corrected: the decreased light production is a function oftoxicity.

Detailed descriptions of the molecular cloning techniques and construction ofthe recN-luxCDABE fusions are given in reference 12. The TA104 pr1 strain wasintroduced in a subsequent paper [13]. SOS induction, on which the test is based, isschematically presented in Figure 11.1.

11.4 CONSTRUCTION OF recN-luxCDABE FUSIONS

The recN promoter region, which is part of the Escherichia coli recN gene [14],contains twoLexA-binding sites. OneLexA-binding site overlapswith the –35 region,while the second overlaps with the –10 region and the transcription start point of therecN promoter. The E. coli recN promoter was cloned upstream of the luxCDABEoperon of the expression vector pMOL877. This resulted in pMOL1066. Since therecN promoter is under control of the bacterial SOS system, the expression of thelux operon has become SOS regulated. Also, recN promoter derivatives lackingthe LexA2 site or having a promoter up mutation were cloned in pMOL877. Theseconstructs were named pMOL1067 and pMOL1068, respectively. All constructswere introduced in the Ames test strains TA98, TA100, and TA104, which allow thesimultaneous detection of mutagenic effects by the normal Ames assay. Followingdifferent trials, it appeared that the best results were obtained with the two mutated


Inactive RecA proteins

5’

5’

3’5’

Active RecA proteins

DNA

DAMAGED DNASingle-strand-binding(SSB) proteins bind todamaged DNA

Activated RecA proteinsbind around single-strandDNA

Induction of transcription of SOS function

LexA inactivatedby cleavage

Active LexA protein actsas a repressor of theSOS function

Genotoxic compound

lexA recA uvtA uvtB lexA recA uvtA uvtB

FIGURE 11.1 Principle of the Vitotox R© test showing how DNA damage results in inacti-vation of LexA and induction of transcription of SOS function (LexA, sfiA, uvrA, uvrB, andothers, including recN).

recNpromoter constructs. They allowed the detection of genotoxic compoundswithin2–4 h. Usually 5 to 100 times lower concentrations were shown to be genotoxic whencompared to results obtained in the Ames test and SOS chromotest. Since similarresults were obtained with all test strains, further experiments were carried out withstrain TA104 (pMOL1068) that gave somewhat better results in some of the trials [12].The strain containing the recN-luxCDABE fusion was designated as TA104recN2-4(genox strain).

11.5 CONSTRUCTION OF pr1-LUXCDABE FUSION

Besides TA104recN2-4 another construct was made as a “control strain.” Indeed,some compounds seem to act directly on light production (e.g., aldehydes) or enhancethe metabolism of the bacteria creating false-positive results. Therefore, plasmidpMOL 1046 was constructed by random cloning of EcoRI-digested DNA fragmentsfrom Alcaligenes eutrophus CH34 in the luxCDABE expression vector pMOL877.A. eutrophus CH34 is a gram-negative nonpathogenic soil bacterium derived froma site heavily polluted with heavy metals. After transformation into E. coli, cloneswere selected for light production. The best constitutive light emitting clone wasthen selected out of the six different plasmid transformants (= plasmid pMOL1046)and introduced into the S. typhimurium strain TA104. This was named the pr1 strain(cytox strain). It contains lux genes under the control of a constitutive promoter, sothat light production is not influenced by genotoxic compounds. The pr1 strain is usedin parallel with the recN2-4 strain and is cultivated and treated in exactly the sameway. If light emission increases in the pr1 strain, this is not indicative of a genotoxic

VITOTOX R© TEST PROCEDURE 217

event. On the other hand, light emission in pr1 can decrease, indicating a toxic event.The Vitotox test thus simultaneously detects genotoxicity and cytotoxicity [13].

11.6 VITOTOX R© TEST PROCEDURE

As the Vitotox test is performed with Salmonella typhimurium, bacteria that lackthe necessary oxidative enzyme systems for metabolizing foreign compounds toelectrophilic metabolites capable of reacting with DNA, the bacteria are, as in thecase of most other in vitro assays, treated with the test compound in the presenceand absence of a postmitochondrial supernatant (S9 or microsome fraction). This isprepared from the livers of Aroclor-treated rats.In short, bacteria are incubated overnight, and then a dilution of the bacterial

suspension is incubated for one more hour on a rotative shaker. Multiwell plates areused so as to contain the solvent, different concentrations of the test compound, orthe positive control for genotoxicity testing (e.g., 2-aminofluorene in cultures with S9or 4-nitroquinoline oxide without S9). Genotoxicity and cytotoxicity measurementsare performed at a constant temperature (approximately 30◦C) using a temperature-controlled microplate luminometer that enables online measurements of emitted light(e.g., every 5 min over a period of 4 h). After completion of the measurements, thedata are transferred into an Excel macrosheet and the signal-to-noise (S/N) ratio,that is, the light production of exposed bacteria divided by the light production ofnonexposed bacteria, is calculated for each measurement. S/N is calculated for therecN2-4 and pr1 strain separately, as well as the ratio between the maximum S/Nvalues of the recN2-4 and pr1 strains. All calculations occur automatically and arebased on measurements between 60 and 240 min of incubation.Based on experimental grounds, a compound is considered genotoxic when the

following criteria are met:

1. Max S/N (recN2-4)/max S/N (pr1) (for convenience rec/pr1) is >1.5

2. Max S/N in recN2-4 must show a good dose–effect relationship

3. rec/pr1 should show a good dose–effect relationship.

Further criteria are:

1. If S/N in recN2-4 increases very quickly during the first 20 min, this is not agenotoxic effect (SOS induction cannot already occur)

2. If both strains are strongly induced, there is no genotoxicity even when rec/pr1>1.5

3. Toxicity is assumed when the light emission is substantially decreasing in adose-dependent way; this is the case when S/N in recN2-4 or pr1 decreases farbelow 0.8.

At present, the Vitotox R© test is available as a test kit that is distributed by Gentaur(cat. no. 6400000; [email protected]). Therefore, the detailed procedure describedin the test kit can be slightly different from the above-mentioned description.


Genox strain ICR 191 acridine –S9

0

1

2

3

4

0 60 120 180 240

0.02 ppm

0.04 ppm

0.08 ppm

0.16 ppm

0.31 ppm

0.62 ppm

1.25 ppm

2.5 ppm

0

1

2

3

4

0 60 120 180 240Time (min)

Sign

al/N

oise

Cytox strain ICR 191 acridine –S9

Sign

al/N

oise

0

1

2

3

4

0 60 120 180 240Time (min)

0.02 ppm

0.04 ppm

0.08 ppm

0.16 ppm

0.31 ppm

0.62 ppm

1.25 ppm

2.5 ppm

0

1

2

3

4

0 60 120 180 240

0

1

2

3

4

0 60 120 180 240

FIGURE 11.2 Genotoxicity of ICR 191 acridine as seen in the Vitotox test.

11.7 EXAMPLES OF VITOTOX TEST RESULTS

A few examples of Vitotox test results are shown in Figures 11.2–11.8. Figure 11.2gives the S/N ratio in both the genox and cytox strains as seen for different concentra-tions of the mutagen ICR 191 acridine in the absence of a metabolizing S9 fraction.There is a dose-dependent increase in light production (S/N ratio) in the genox strainreaching values well above 1.5, whereas light production in the cytox strain is notenhanced. This indicates that this chemical is indeed genotoxic as increased lightproduction is not due to a direct effect on the lux operon, which is unrelated toDNA damage.Another example, in this case carbadox, is shown in Figure 11.3. As before, there

is an important dose-dependent increase in light production (S/N ratio) in the genoxstrain reaching values well above 1.5, whereas light production in the cytox strain isnot significantly enhanced.Figure 11.4 shows the data for fluoranthene, which requires metabolic activation.

There is again a dose-dependent increase in light production (S/N ratio) in the genoxstrain, but light production in the cytox strain is not enhanced. Here maximum lightproduction is only reached after approximately 240 min.

Genox strain carbadox –S9

0

2

4

6

8

10

0 60 120 180 240

Time (min)

Sign

al/N

oise

0.04ppm

0.08ppm

0.16ppm

0.32ppm

0.64ppm

1.28ppm

2.56ppm

5.12ppm

0.04 ppm

0.08 ppm

0.16 ppm

0.32 ppm

0.64 ppm

1.28 ppm

2.56 ppm

5.12 ppm

Cytox strain carbadox –S9

Sign

al/N

oise

0

2

4

6

8

10

0 60 120 180 240

TIme (min)

0.04ppm

0.08ppm

0.16ppm

0.32ppm

0.64ppm

1.28ppm

2.56ppm

5.12ppm

0.04 ppm

0.08 ppm

0.16 ppm

0.32 ppm

0.64 ppm

1.28 ppm

2.56 ppm

5.12 ppm

FIGURE 11.3 Genotoxicity of carbadox as seen in the Vitotox test.

EXAMPLES OF VITOTOX TEST RESULTS 219

Genox strain fluoranthene +S9

0

0,5

1

1,5

2

2,5

0 60 120 180 240

Time (min)

3.1ppm

6.2ppm

12.5ppm

25ppm

50ppm

100ppm

200ppm

400ppm

3.1 ppm

6.2 ppm

12.5 ppm

25 ppm

50 ppm

100 ppm

200 ppm

400 ppm

Sign

al/N

oise

Time (min)

0

0,5

1

1,5

2

2,5

0 60 120 180 240

3.1ppm

6.2ppm

12.5ppm

25ppm

50ppm

100ppm

200ppm

400ppm

Cytox strain fluoranthene +S9

3.1 ppm

6.2 ppm

12.5 ppm

25 ppm

50 ppm

100 ppm

200 ppm

400 ppm

Sign

al/N

oise

FIGURE 11.4 Genotoxicity of fluoranthene in the presence of a metabolizing S9 fraction asseen in the Vitotox test.

Potassium dichromate (K2Cr2O7) was also tested in the Vitotox test (Fig. 11.5).It is clear that the higher doses are toxic (dramatic decrease in S/N ratio in the cytoxstrain). This is to a certain extent also reflected in the genox strain, which neverthelessstill clearly shows genotoxicity.Figure 11.6 illustrates that Vitotox test results can be very reproducible. In this

case, benzo[a]pyrene was tested in two independent experiments. The figure onlyshows the results for the genox strain, but it is clear that the results are in both testsalmost identical.Two examples ofVitotox tests on environmental water samples are given in Figures

11.7 and 11.8. A test was performed on a water sample from the (polluted) river“Laak” (Zwijndrecht, Belgium), and the results (in the absence of S9) are presentedin Figure 11.7. Looking at the genox strain alone (TA104 recN), one can find thatthere is an apparent genotoxic effect. However, there is a similar increased S/N ratioin the cytox strain (TA104 pr1), which indicates that the water contains chemicalsthat enhance the metabolism or have a catalytic action on processes involved in lightproduction. Therefore, we cannot conclude that this sample is genotoxic.Figure 11.8 shows the results of a test that was performed on a concentrated

surface water sample from the river Meuse. The cytox strain shows a severe “toxic”

Genox strain K2Cr2O7 –S9

0

1

2

3

4

5

6

0 60 120 180 240

Time (min)

0.5 ppm

1 ppm

2 ppm

4 ppm

8 ppm

16 ppm

32 ppm

64 ppm

Sign

al/N

oise

Cytox strain K2Cr2O7 –S9

Time (min)

0

0,5

1

1,5

0 60 120 180 240

0

0,5

1

1,5

0 60 120 180 240

0.5 ppm

1 ppm

2 ppm

4 ppm

8 ppm

16 ppm

32 ppm

64 ppm

Sign

al/N

oise

FIGURE 11.5 Genotoxicity of potassium dichromate (K2Cr2O7) in the absence of a metab-olizing S9 fraction as seen in the Vitotox test.


Genox strain B[a]P + S9

0123

4

567

0 60 120 180 240

0.025 ppm

0.05 ppm

0.1 ppm

0.2 ppm

0.4 ppm

0.8 ppm

1.6 ppm

3.2 ppm

6.4 ppm

0123

4

567

0 60 120 180 240

0123

4

567

0 60 120 180 240

Time (min)

0123

4

567

0 60 120 180 240

Sign

al/N

oise

Genox strain B[a]P + S9

01

23

45

67

0 60 120 180 240

0.025 ppm

0.05 ppm

0.1 ppm

0.2 ppm

0.4 ppm

0.8 ppm

1.6 ppm

3.2 ppm

6.4 ppmTime (min)

01

23

45

67

0 60 120 180 24060 120 180 240

Sign

al/N

oise

FIGURE 11.6 Results from two independent experiments with benzo[a]pyrene (B[a]P;genox strain only).

Genox strain surface waterRiver Laak –S9

0,81

1,21,41,61,8

22,22,42,6

0 60 120 180 240

2.50%

5.00%

7.50%

10%

0,81

1,21,41,61,8

22,22,42,6

0 60 120 180 240

Time (min)

2.50%

5.00%

7.50%

10%

Cytox strain surface waterRiver Laak –S9

0,8

1

1,21,4

1,6

1,8

22,2

2,4

2,6

0 60 120 180 240

Time (min)

2.50%

5.00%

7.50%

10%

0,8

1

1,21,4

1,6

1,8

22,2

2,4

2,6

0 60 120 180 240

2.50%

5.00%

7.50%

10%

FIGURE 11.7 Vitotox test applied to surface water from the river “Laak.” Four dilutions(%) of the original sample were tested.

0

0,5

1

1,5

2

2,5

3

3,5

4

4,5

0 60 120 180 240

Time (min)

0.125%

0.25%

0.5%

1%Sign

al/N

oise

strain concentrated surfacefrom the river meuse (pH 7) –S9

Genoxwater from (

0

0,5

1

1,5

2

2,5

3

3,5

4

4,5

0 60 120 180 240

0.125%

0.25%

0.5%

1%

Time (min)

0

0,2

0,4

0,6

0,8

1

1,2

1,4

0 60 120 180 240

Sign

al/N

oise 0.125%

0.25%

0.5%

1%

strain concentrated surfacefrom the river meuse (pH 7) –S9

Cytoxwater from (

0

0,2

0,4

0,6

0,8

1

1,2

1,4

0 60 120 180 240

0.125%

0.25%

0.5%

1%

FIGURE 11.8 Example of Vitotox test results with a concentrated (XAD-2 resin) surfacewater sample. Tested concentrations are expressed here as percentage of the sample.

APPLICATIONS OF THE VITOTOX TEST 221

response, whereas the results with the genox strain cannot be adequately interpreted.This kind of “odd” result is sometimes found with complex mixtures. In this case, noconclusions can be drawn from the Vitotox test results.

11.8 APPLICATIONS OF THE VITOTOX TEST

11.8.1 Testing Chemicals with Known Genotoxic Propertiesand New Drug Candidates

The Vitotox test has successfully been used for prescreening purposes, especially totest pure compounds and pharmaceutical preparations. This is exemplarily illustratedin Figures 11.2–11.6 and was already demonstrated in the early publications intro-ducing the test [12, 13]. Muto et al. [15] have also shown that the test is very usefulfor the rapid screening of large numbers of chemicals when only a small quantityof a chemical is available. They found a 94% concordance with the Ames assaywhen testing known Ames test-positive (33) and -negative compounds (26) as wellas 18 drug candidates. The authors also validated the performance of the 384-wellmicroplate-based Vitotox test (Vitotox-384; at that time commercialized by ThermoLabsystems) in an attempt to further improve the throughput and use of the test inpharmaceutical development. Therefore, the test was applied to 61 NTP compounds(= recommended for testing by the National Toxicological Program). They found asensitivity of 87% (26/30, excluding three inconclusive results) and a negative speci-ficity of 100% (25/25). In this study, the concordance between the Vitotox test andthe Ames test was 92% [16].It is important to stress that when the Vitotox test was developed [12, 13], mea-

surements were performed during a 4-h period, which was found the most suitabletime period to allow a correct evaluation of genotoxicity. The validation study ofMuto et al. [16] clearly confirmed this choice. A treatment time of 180 min wasspecified by the (former) supplier of the Vitotox test kit, but it was found out that,although in most cases induction of light emission in the genox strain by genotoxinsstarts at about 60 min after incubation and reaches a maximum within 180 min, somecompounds showed “delayed enhancement” of light emission reaching a maximum(above the genotoxicity threshold level) only after 180 min. These chemicals wouldbe scored as nongenotoxic if measurements were stopped at that time. However, theywould have been correctly evaluated as being genotoxic when measurements wereprolonged up to 240 min. This also illustrates the advantage of kinetic analyses in theVitotox test over endpoint assays like the Umu-C test.Muto et al. [16] demonstrated the usefulness of the Vitotox test as an important

screening tool in the early stage of pharmaceutical development. In their hands, only10 mg of a compound is needed and up to 15 chemicals can be tested per day, evenwith a manual procedure. Easy automation is possible and should further enhancethe number of chemicals that can be tested each day.Compared to the Ames test, which needs at least 3–5 days for an endpoint

measurement, the format of the Vitotox test is much faster and more in line withhigh-throughput procedures. In a large comparative study, the Vitotox, RadarScreen,


and Greenscreen assays were evaluated as early screens for mutagenicity and clas-togenicity [17–19]. The RadarScreen assay was developed in order to obtain a toolcapable of replacing the sister chromatid exchange, chromosomal aberration, and/ormicronucleus tests, which are time consuming and have low compound through-put. This assay is based on the activation of the RAD54 promoter linked to a �-galactosidase reporter gene in yeast. The expression of �-galactosidase can easilybe quantified by using the substrate D-luciferin-O-�-galactopyranoside, which isconverted into galactose and luciferin, and luciferin can then be measured lumi-nometrically. This is possible within 24 h. In the Greenscreen assay, the RAD54gene behind the RAD54 promoter has been replaced by the green fluorescent protein(GFP). A drawback of this system is that S9 liver fractions cannot be used and thatsome of the test compounds give autofluorescence at the wavelength of the inducedGFP. A list of 20 genotoxic and 42 nongenotoxic compounds, as defined by theEuropean Centre for the Validation of Alternative Methods working group [20], aswell as an additional 192 compounds, were investigated.The Vitotox test was found to be very promising as its results were very well

in agreement with those of a full Ames test. Prior to the implementation of theVitotox test, 19% of the compounds tested were positive in the Ames assay. Afterthe implementation of the Vitotox test and de-selection of compounds with a positiveresult in this test, Ames-positive compounds were eliminated [19]. In this respect,the Vitotox test is much more accurate than the GreenScreen or RadarScreen assays.This was also previously observed [16]. It should be noted that in this study, atreatment and measuring time of 180 min was adopted for the Vitotox test. However,as indicated before, the sensitivity of this assay can be improved if the treatmenttime is increased to 240 min. Therefore, one may even expect a somewhat improvedsensitivity, specificity, and predictivity. A maximum of 25–40 mg of a test compoundwas used to conduct the Vitotox test in this study.The Vitotox test is, however, a bacterial test, which means that it uses organisms

whose DNA is not packed into chromatin and that lack a spindle apparatus. It cantherefore be expected that it will underestimate mitotic and meiotic segregationfailures, which may occur in eukaryotic cells. Therefore, in this case the RadarScreenand GreenScreen assays may be more useful. This is indeed shown in Table 11.1 [18].It is expected that the RadarScreen assaywill further reduce the number of clastogeniccompounds and that a combination of the Vitotox and RadarScreen assays will bemost valuable as rapid high-throughput screening tests [19].

11.8.2 Testing Secondary Metabolites and Crude Extracts of BiologicalControl Agents

Kouvelis et al. [21] examined the putative mutagenicity of secondary metabolitesproduced by a number of fungal biological control agents (BCAs) with the Amesassay and Vitotox test. Crude extracts and fungal cell extracts of the BCAs werealso examined. Vitotox results were in full accordance with the results of the Amesassay, as neither pure metabolites nor crude extracts of fungal BCAs were foundto be mutagenic in both test systems. The presence and absence of cytotoxicity asevaluated by the Vitotox test was also consistent with the results of the Ames test. It


TABLE 11.1 Comparison between the Vitotox, GreenScreen, and RadarScreen Assays[17–19]

Predictivity Values for the Vitotox, GreenScreen, and RadarScreen Assays inComparison with Scores for Full Ames tests

Vitotox GreenScreen GC RadarScreen

Sensitivity 0.90 0.39 0.55Specificity 0.90 0.98 0.52Predictivity 0.90 0.74 0.53

Predictivity values for the Vitotox, GreenScreen, and RadarScreen assays in comparisonwith scores for in vitro clastogenicity/aneuploidy

Vitotox GreenScreen GC RadarScreen

Sensitivity 0.29 0.22 0.80Specificity 0.89 0.95 0.77Predictivity 0.51 0.57 0.78

Sensitivity values for the Vitotox and RadarScreen assays separately and combinedaccording to scores for in vitro mutagenicity in the Ames test, for in vitro clastogenicity andfor in vitro carcinogenicity assays

Sensitivity (+ and –)

Mutagenicity Clastogenicity Carcinogenicity

Vitotox 0.90 0.29 0.30RadarScreen 0.55 0.80 0.82Vitotox+RadarScreen 0.96 0.78 0.82

was suggested that both the Vitotox and Ames tests are complementary tests that canbe used together in future toxicological analyses.

11.8.3 Further Comparative Studies: Technotox Workshop

ATechnical Workshop was held inMol (Belgium) onMay 8–12, 2000. Its aim was tobring scientists involved in the construction and use of genotoxicity assays togetherand allow them to practically compare different assays using identical samples, testconditions, and well-defined endpoints. In this workshop, more than 10 genotoxicitytests were run side by side on pure chemical and environmental samples in blindtests, and performance of each test was discussed. A global overview of the results issummarized in Table 11.2.The main conclusions were as follows [22]: five tests were bacterial assays based

on the DNA damage-dependent induction of the bacterial SOS system, namely theSOS chromotest, the umu-test, the combined SOS-lux- and Lac-fluoro test, the fiberoptic RecA-lux test, and the Vitotox test. Very similar results were obtained with allthese bacterial tests. However, some difficulties raised. The SOS chromotest couldnot detect a lower 2-aminoanthracene (2-AA) concentration and an effluent from

TA

BL

E11

.2Su

mm

ary

ofth

eR

esul

tsO

btai

ned

atth

eTe

chno

tox

Wor

ksho

p[2

2]

4-NitroquinolineOxide

a2-Aminoanthraceneb

MNNG

cSurfaceWater

dEffluentWater

e

TestName

0.1

�g/mL

5�g/mL

0.1

�g/mL

4�g/mL

200

�g/mL

4�g/mL

4�g/mL

400

�g/mL

CO

LA

TEXT

DNAinteraction

++

–+

++

––

–+

+Mutatox

––

–+(+/−S9)

+(−S9)

––

+(−S9)

SOSchromotest

–+(+/−S9)

+(+S9)

+(+S9)

+(+/−S9)

+(+/−S9)

–+(+S9)

Vitotox

+(−S9)

+(−S9)

+(−S9)

+(+S9)

+(+S9)

+(+S9)

+(+S9)

+(+/−S9)–

–+(−S9)

SOS-

luxand

Lac

-fluo

rotest

+(−S9)

+(−S9)

+(−S9)

+(+S9)

+(+S9)

+(+S9)

+(+S9)

+(−S9)

––

+(−S9)

Umu-Ctest

+(−S9)

+(−S9)

+(−S9)

+(+S9)

+(+S9)

–+(+S9)

+(−S9)

–+(+S9)

+(+/−S9)

Amestest

f+(−S9)

+(−S9)

+(+S9)

+(+S9)

–+(−S9)

–+(+/−S9)

Fiberoptic

f–

+(−S9)

––

–+(−S9)

+(−S9)

––

––

�-FADU(−S9)

++

––

––

++

–+

+DNAalkaline

unwindingfish

+(−S9)

+(−S9)

––

–+(−S9)

––

–

Mammaliangene

expression

f–

––

–+

RAD54-GFP

+/–

+–

+/–

++/–

++

–+

+Comettest

f+(−S9)

+(+/−S9)–

+(+/−S9)

+(+/−S9)–

–+(+/−S9)–

+(−S9)

+(+/−S9)

ToxAlert

TOX

TOX

TOX

Non-TOX

Non-TOX

TOX

TOX

TOX

Non-TOXTOX

TOX

TOX,toxic;

+designatesdetectionofgenotoxicity;–designatesnogenotoxicitywasfoundinthetestedconcentrations.S9indicatesteststhatwereperformedinthe

presenceofametabolizingenzymefraction.Allsampleswerecodedandcodeswerebrokenattheendoftheworkshopinthepresenceofallparticipants.

a4-Nitroquinolineoxidewasgiveninthreeforms:0.1

�g/mLdissolvedindimethylsulfoxide(DMSO),5

�g/mL(DMSO),and0.1

�g/mLdissolvedinwater.

b2-Aminoanthracenewasgiveninconcentrationsof4and200

�g/mLinDMSO

orwater.

c MNNG(=

N-methyl-

N′ -nitro-

N-nitroguanidine)wasgivenin4and400

�g/mlconcentrations(dissolvedinwater).

dSurfacewaterswerefrom

twodifferentorigins(CO

=acanal,LA

=ariver)andweretestedassuch(unconcentrated).

e Aneffluentwatersamplefrom

atextileindustry(TEXT).

f Someofthedatawereonlymadeavailableaftertheworkshop.

224


textile industry was found to be not genotoxic, whereas it was positive using theother bacterial tests. The fiber optic RecA-lux test was still at an early research stage,and therefore the number of samples that could be tested during the workshop wasrestricted. The immobilization of the sensor needs to be optimized to limit matrixeffects and increase the detection limit. Some other tests were also too much timeconsuming, so that not all samples could be tested in the allotted time. The umu-test, the combined SOS-lux- and Lac-fluoro test, and the Vitotox test had very similaroutcomes and dose–effect relationships. However, the umu-test was not able to detecta surface water spiked with 2-AA, whereas the sample was positive in the two othertests. No false-negatives or false-positives were found with the SOS-lux and Vitotoxassays. Compounds that neededmetabolic activation to be positivewere also correctlydetected. Interestingly, water from one “polluted river” (Laarbeek) was found not tobe genotoxic by most bacterial tests but genotoxic when using tests with higher targetorganisms (yeast, fish embryos, human white blood cells). The Laarbeek River watersample was also found to be toxic with the ToxAlert assay, and GC/MS analysisindicated the presence of a number of nitrophenol, alkylphenol, and chlorophenolsubstances [22].Overall, the most sensitive of all tests appeared to be the Vitotox test. This com-

parative exercise thus illustrated the value of the Vitotox test, without of courseminimizing the merits of other assays. It was for example also concluded that theRAD54-GFP test is one of the very promising tests.

11.8.4 Further Comparative Studies: EILATox-Oregon Workshop

Following the workshop in Mol, it was decided to organize another similar work-shop later on. This became the Eilatox-Oregon biomonitoring workshop thatwas especially devoted to environmental toxicant detection [23]. A number ofblind samples were provided to all participating groups. They consisted of anarray of various different toxicants at toxic concentrations. Concentrations werechosen as a function of animal toxicities listed at the Hazardous SubstancesData Bank (http://toxnet.nlm.nih.gov/cgi-bin/sis/htmlgen?HSDB), whereby the lev-els were reduced as needed until the samples were virtually colorless and odorless.Several tests were conducted, using commercially available kits as well as proto-type assays. In this workshop, the Vitotox test was the only genotoxicity test. Theendpoints of the other tests investigated included, among others, larvae mortality(Thamnotoxkit FTM), neonate mortality (Daphtoxkit FTM), and natural luminescenceof photobacteria (BioToxTM Flash). It is therefore difficult to compare the merits ofthe different tests. The Vitotox test was performed on 17 blind-coded chemicals andthree environmental water samples (3-h exposure and measuring time). This exerciseshowed that using 96-well plates, in which eight dilutions of a sample can be testedwith and without S9 metabolic activation, it is possible to analyze 60 samples aweek with one laboratory technician and one luminometer. If more instruments areavailable, one technician could handle over 180 samples a week. By using 384-wellplates and a single luminometer, the number of samples tested by a single laboratorytechnician can be increased from 60 to about 150 per working week [24]. The authors


mention that the test can possibly be improved for environmental genotoxicity testing,for example by adapting the test to another, more appropriate organism as for exam-ple soil bacteria like Pseudomonas. Different reporter genes, for example, the EGFP(enhanced green fluorescent protein) gene can be used in addition to the luciferaseoperon. This can prevent problems such as those that were found with many envi-ronmental samples or mixtures and that may be due to a direct effect on metabolicpathways for which the luciferase operon is responsible (see next section). In thiswork, a 40 �L sample volume was sufficient to perform the complete test series.

11.8.5 Environmental Samples

Environmental pollution can be investigated by performing chemical measurementsin water, air, or soil samples, but very often the responsible chemicals are not known.Furthermore, effects of mixtures that are involved in pollution are not necessarily thesum of the effects of the individual components. Therefore, a global investigationusing biomarkers is more realistic and recommended. Genotoxicity tests are amongstthe important biomarkers that are utilized in environmental monitoring [4], and theVitotox test was also used in this context.According to Sekler et al. [25], who used the Vitotox test to monitor the toxicity

of the intermediates formed during photochemical degradation of p-nitrophenol,this bioassay was found to be a simple, highly sensitive, and fast method and mayserve as a marker for water quality during water purification. An additional featureof this bioassay is that it can be adapted further to that of a fiber optic biosensorsystem if required for field analysis. The ability to evaluate water contaminationusing bioluminescent bacterial sensors has many industrial applications and could beused as a powerful tool in real-time toxicity monitoring of water contamination andthe quality of water purification systems [25].Tests that were performed at the Technotox and Eilatox-Oregon workshops

included environmental samples, for which the Vitotox test rendered satisfactoryresults. The samples were tested as such, and no concentration of the pollutants inthe sample was performed. In most environmental biomonitoring studies includinggenotoxicity testing, samples need to be pretreated and concentrated before they canbe tested. Pollutants in surface waters are usually present in such low concentrationsthat genotoxicity cannot be directly detected. They need therefore to be concen-trated (for example, using XAD-2 resins). Air samples need to be collected on filters(e.g., polyurethane or glass fiber filters), and soil samples need to be eluted (waterextracted) before they can be presented to bacteria or eukaryotic cell cultures that areused in a test. It appears that Vitotox tests performed on such concentrated samplesand/or complex mixtures often give results that are difficult to interpret, usually as aresult of apparent toxicity [26–29]. This was also in part found to be the case withnonconcentrated, but highly polluted surface water samples of a river in Hyderabad(India) [30]. Brits et al. [31] compared the Vitotox test with the Umu-C test (anotherbacterial test that also uses Salmonella typhimurium bacteria) for assessing the geno-toxicity of effluents and air samples. Pure compounds were also tested with bothassays as well as with the Ames test, and their results were compared. Some environ-mental samples were apparently toxic based on the results with the cytox (pr1) and


sometimes also the genox (recN) strain (drastic decrease in light production). Yet, thecompounds were not acutely toxic as these concentrations did not affect the growthof Umu-C test bacteria (Salmonella typhimurium TA1535). This indicates that theremight be an interaction of the sample or some of its constituents with the metabolicpathways in which the luciferase operon is involved. For example, in the luciferasereaction, aldehydes and dioxide are required. The presence of aldehydes or dioxide ina sample can therefore render a false-positive signal, while compounds that competewith luciferase for the aldehydes or dioxide may act as inhibitors and in this mannerlower the luminescence. In this study, it was (again) demonstrated that the Vitotoxtest is a very suitable and faster alternative to the Ames test with a concordance ofover 90%, but testing complex mixtures may be problematic, for example due tothe above-mentioned problems or complex interactions (additive, synergistic, and/orantagonistic effects) between individual constituents in these samples. In such cases,the Vitotox test can often still be used for a first screening, but other tests may berequired when the results are not conclusive or questionable (as in Fig. 11.8). Thisis the reason why it was suggested to improve the test for such samples by usinganother test organism or reporter system [24].

11.8.6 Smoke-Water and Isolated Smoke-Water Compounds

Smoke, smoke-water, and aerosols have a stimulatory effect on seed germination andgrowth vigor of many seedlings [32], which make possible many potential applica-tions related to seed technology (e.g., in agriculture, horticulture, seed pretreatment,weed control, ecological management, and habitat restoration), provided they do notpose a health risk. Recently, a compound that is highly active in promoting germina-tionwas isolated from plant-derived smoke and fromburned cellulose. The compoundwas characterized as the butenolide 3-methyl-2H-furo[2,3-c]pyran-2-one [33, 34]. Itacts at extremely low concentrations, as low as 10−9 M, and over a wide concen-tration range. Aerosol smoke and smoke solutions are extremely complex chemicalmixtures and contain potentially hazardous substances, such as polycyclic aromaticcompounds (PAHs), which are certainly of health concern. Thus, the isolation of thehighly active germination promoter from plant-derived smoke is an important steptowards a more controlled utilization of the effect of biological activity seen withaerosol smoke and smoke solutions. Further constituents of smoke, for example, avariant of the above-mentioned compound, 3,7-dimethyl-2H-furo[2,3-c]pyran-2-one,was also isolated.It is thus very important to investigate the potential toxicity and health effects of

smoke and smoke-water as well as their isolated active compounds. A number ofgenotoxicity tests have already been performed with this purpose. In a first investiga-tion with the Ames and Vitotox tests, 3-methyl-2H-furo[2,3-c]pyran-2-one was foundto be nonmutagenic in the absence and presence of S9 metabolic activation [35]. Theresults are in agreement with another investigation, in which it was reported that thebutenolide does not affect the integrity of DNA during seed germination, althoughit does play some role during transcription and translation [36]. In a further investi-gation on smoke-water and 3,7-dimethyl-2H-furo[2,3-c]pyran-2-one the Vitotox test


was again applied and no genotoxicity was detected, but the higher concentrations(lower dilutions) of the smoke-water showed a “toxic” response [37].

11.8.7 Testing Medicinal Plant Extracts for Genotoxicand Antigenotoxic Properties

According to the World Health Organisation, approximately 80% of the populationin Asian and African countries rely on traditional medicine as their main source ofhealth care. The interest in herbal products worldwide as a re-emerging health aidis fuelled by the rising costs of drugs. Because these plants are natural and haveoften been used for ages, they are mostly considered to be adequate and safe. Yet,many of them were never tested for their potential adverse health effects in the wayour modern pharmaceutical preparations are. Acute toxicity as a result of the useof such medicinal plant extracts is more common than often assumed, especiallyamong children [38]. It is estimated that between 8000 and 20,000 annual deaths arerecorded in South Africa due to incorrect use of medicinal plants [39]. Moreover,long-term effects in terms of, for example, genotoxicity and/or carcinogenicity needto be carefully investigated.Nowadays extracts from traditional medicinal plants are therefore tested in dif-

ferent laboratories worldwide. Not only are their potential genotoxic but also theirpotential antigenotoxic properties studied. Compounds with antigenotoxic propertiesmay indeed be useful as anticancer agents and may eventually be interesting compo-nents in functional food. Overall, tests that are used to investigate genotoxicity and/orantigenotoxicity of (traditional) medicinal plants or their extracts are the same asthose used for other purposes (Ames assay, comet assay, micronucleus test, etc.). Inthis context, the Vitotox test has also been used onmany occasions [40–44]. It appearsthat some of the tested extracts indeed have genotoxic properties, whereas some arepromising as they were found to be antigenotoxic in different complementary geno-toxicity tests [41]. However, such extracts are complex mixtures and, as indicatedbefore, this sometimes poses a problem when using the Vitotox assay. This has forexample been illustrated by van den Bout-van den Beukel et al. [42], who investigatedtoxicity and genotoxicity of plant extracts with promising antifungal properties. Mostof the extracts showed a relative strong toxicity in one or more assays. Although threeplant extracts were genotoxic according to the Vitotox test, the toxic response of otherextracts might mask genotoxicity. This might have been the case of Acacia nilotica,Clausena anisata, and Pteridium aquilinum, for which genotoxicity has previouslybeen reported. It was therefore concluded that the Vitotox test seems less appropriatefor testing the genotoxicity of crude plant extracts. This test can nevertheless still beused for a preliminary rapid screening, but needs to be complemented by other testsdepending on the outcome.

11.8.8 Investigating Genotoxicity of Nonionizing Radiation

There is quite a lot of anxiety amongst the public and authorities regarding possibleadverse health effects from exposure to electromagnetic fields, in particular the fields(or radiation) generated by power lines and wireless communication devices. In

REFERENCES 229

2002 the International Agency for Research on Cancer concluded that extreme low-frequency (ELF) magnetic fields (power lines) are possibly carcinogenic (Group2B of IARC classification; [45]). The same conclusion was reached in 2011 withrespect to radiofrequency fields from mobile phones and their base station antennas[46, 47]. It is therefore not surprising that a lot of research was and is still devotedto the biological effects of such nonionizing radiation. Because of the establishedlink between genotoxicity and carcinogenicity (in many cases) and the possibilitythat electromagnetic fields are carcinogenic it is not surprising that studies on themutagenicity and genotoxicity of such fields are also an important issue. Recently, theVitotox test was used to investigate the genotoxicity of ELFmagnetic fields (includingpossible cooperative effects with known mutagens). In this study, it was shown thatin the Vitotox test ELF magnetic fields up to 500 �T do not induce genetic alterationsand have no cooperative or synergistic action with different chemical agents [48].These results are in accordance with other genotoxicity tests in bacteria as well aswith the majority of other genotoxicity tests [45, 49].

11.9 CONCLUSIONS

The Vitotox test is used for many different purposes, from product testing to envi-ronmental applications and scientific research related to biomedicine and relateddisciplines. The test proves to be a very valuable, rapid, and easy assay for testinggenotoxicity of pure compounds, for example, for the rapid high-throughput screen-ing of candidate drugs in the pharmaceutical industry. It also proves useful in differentother applications, but it should be realized that the test is a bacterial test, whichmeansthat it will normally not be able to detect aneuploidy-inducing agents. Furthermore,the test also seems less appropriate for testing the genotoxicity of complex mixturesand extracts.

REFERENCES

1. Eastmond, D.A., Hartwig, A., Anderson, D., Anwar, W.A., Cimino, M.C., Dobrev, I.,Douglas, G.R., Nohmi, T., Philips, D.H., Vickers, C. (2009). Mutagenicity testing forchemical risk assessment: update of the WHO/IPCS harmonized scheme. Mutagenesis,24, 341–349.

2. Rao, K.S., Xu, Y., Shaw, E., Parton, J.W. (2004). Mutagenicity testing applied for regula-tion of developing products. Current Separations, 20, 141–144.

3. Allen, J., Jeffrey, P., Williams, R., Ratcliffe, A.J. (2010). Approaches to assessing drugsafety in the discovery phase. Drugs of the Future, 35, 67–75.

4. Bierkens, J., Brits, E., Verschaeve, L. (2005). Environmental monitoring for genotoxiccompounds. In: Environmental Toxicity Testing (Thompson, K.C., Wadhia, K., Loibner,A.P., Eds.). Blackwell Publishing Ltd.: Oxford; pp. 229–256.

5. Biran, A., Pedahzur, R., Buchinger, S., Reifferscheid, G., Belkin, S. (2009). Geneticallyengineered bacteria for genotoxicity assessment. The Handbook of Environmental Chem-istry, 5J, 161–186.


6. Ames, B.N., Lee, F.D., Durston, W.E. (1973). An improved bacterial test system for thedetection and classification of mutagens and carcinogens. Proceedings of the NationalAcademy of Sciences USA, 70, 782–786.

7. Mortelmans, K., Zeiger, E. (2000). The Ames Salmonella/microsome mutagenicity assay.Mutation Research, 455, 29–60.

8. Quillardet, P., Hofnung, M. (1985). The SOS Chromotest, a colorimetric bacterial assayfor genotoxins: procedures. Mutation Research, 147, 65–78.

9. Quillardet, P., Huisman, O., D’Ari, R., Hofnung, M. (1982). SOS chromotest, a directassay of induction of an SOS function in Escherichia coli K-12 to measure genotoxicity.Proceedings of the National Academy of Sciences USA, 79, 5971–5975.

10. Quillardet, P., de Bellecombe, C., Hofnung, M. (1985). The SOS Chromotest, a colori-metric bacterial assay for genotoxins: validation study with 83 compounds. MutationResearch, 147, 79–95.

11. Oda, Y., Nakamura,S., Oki, I., Kato, T., Shinagawa, H. (1985). Evaluation of the newsystem (umu-test) for the detection of environmental mutagens and carcinogens.MutationResearch, 147, 219–229.

12. van der Lelie, D., Regniers, L., Borremans, B., Provoost, A., Verschaeve, L. (1997).The VITOTOX test, a SOS-bioluminescence Salmonella typhimurium test to measuregenotoxicity kinetics. Mutation Research, 389, 279–290.

13. Verschaeve, L., Van Gompel, J., Regniers, L., Van Parijs, Ph., van der Lelie, D. (1999).VITOTOX R© genotoxicity and toxicity test for the rapid screening of chemicals. Environ-mental and Molecular Mutagenesis, 33, 240–248.

14. Rostas, K., Morton, S.J., Picksley, S.M., Lloyd, R.G. (1987). Nucleotide sequence andLexA regulation of the Escherichia coli recN gene. Nucleic Acids Research, 15, 5041–5049.

15. Muto, S., Baba, H., Uno, Y. (2003). Evaluation of the VitotoxTMtest as a high-throughput

genotoxicity assay. Environmental Mutagen Research, 25, 69–75.

16. Muto, S., Miyata, K., Tsutsumibata, C., Daigo, H., Ooro, T., Sugiura, K., Baba, H., Uno,Y. (2006). A practical application of the Vitotox

TMtest for genotoxicity screening of phar-

maceuticals. In: Topical Issues in Applied Microbiology and Biotechnology (Verschaeve,L., Ed.). Trivandrum, India: Research Signpost; pp. 1–9, ISBN 81-308-01 37-X.

17. Benfenati, E., Benigni, R., DeMarini, D.M., Helma, C., Kirkland, D., Martin, T.M.,Mazzatorta, P., Ouedraogo-Arras, G., Richard, A.M., Schilter, B., Schoonen, W.G.E.J.,Snyder, R.D., Yang, C. (2009). Predictive models for carcinogenicity and mutagenicity:frameworks, state-of-the-art, and perspectives. Journal of Environmental Science andHealth, Part C, 27, 57–90.

18. Schoonen,W.G.E.J.,Westerink,W.M.A., Horbach, G.J. (2009). High-Throughput screen-ing for analysis of in vitro toxicity. In: Molecular, Clinical and Environmental Toxicol-ogy, Volume 1: Molecular Toxicology (Luch, A., Ed.). Birkhauser Verlag Switserland;pp. 401–452.

19. Westerink, W.M.A., Stevenson, J.C.R., Lauwers, A., Griffioen, G., Horbach, G.J., Schoo-nen, W.G.E.J. (2009). Evaluation of the Vitotox

TMand RadarScreen assays for the rapid

assessment of genotoxicity in the early research phase of drug development. MutationResearch, 676, 113–130.

20. Kirkland, D., Kasper, P., Muller, L., Corvi, R., Speit, G. (2008). Recommended lists ofgenotoxic and non-genotoxic chemicals for assessment of the performance of new or

REFERENCES 231

improved genotoxicity tests: a follow-up to an ECVAM workshop. Mutation Research,653, 99–108.

21. Kouvelis, V.N., Wang, C., Skrobek, A., Pappas, K.M., Typas, M.A., Butt, T.M. (2011).Assessing the cytotoxic and mutagenic effects of secondary metabolites produced byseveral fungal biological control agents with the Ames assay and the Vitotox C© test.Mutation Research, 722, 1–6.

22. Corbisier, P.,Hansen, P.-D., Barcelo,D., Eds. (2000). Proceedings of theBIOSET technicalworkshop on genotoxicity biosensing. VITO report 2001/MIT/P053.

23. Pancrazio, J.J.,McFadden, P.N., Belkin, S.,Marks, R.S. (2004). EILATox-Oregon biomon-itoringworkshop: summary and observations. Journal of Applied Toxicology, 24, 317–321.

24. Merilainen, J., Lampinen, J. (2004). ELAITox-Oregon workshop: blind study evaluationof Vitotox test with genotoxic and cytotoxic sample library. Journal of Applied Toxicology,24, 327–332.

25. Sekler, M.S., Levi, Y., Polyak, B., Novoa, A., Dunlop, P.S.M., Byrne, J.A., Marks, R.S.(2004). Monitoring genotoxicity during the photocatalytic degradation of p-nitrophenol.Journal of Applied Toxicology, 24, 395–400.

26. Brits, E., Schoeters, G., Verschaeve, L. (2004). Genotoxicity of PM10 and extracted organ-ics collected in an industrial, urban and rural area in Flanders, Belgium. EnvironmentalResearch, 96, 109–118.

27. Du Four, V.A., Van Larebeke, N., Janssen, C.R. (2004). Genotoxic and mutagenic activityof environmental air samples in Flanders, Belgium. Mutation Research, 558, 155–167.

28. Du Four, V.A., Janssen, C.R., Brits, E., Van Larebeke, N. (2005). Genotoxic andmutagenicactivity of environmental air samples from different rural, urban and industrial sites inFlanders, Belgium. Mutation Research, 588, 106–117.

29. Verschaeve, L. (2005). The VITOTOX R© genotoxicity test. In: Recent Research Develop-ments in Applied Microbiology and Biotechnology (Pandali, S.G., Ed.). Research Signpost:Trivandrum, India; pp. 33–49, ISBN 81-7736-293-3.

30. Vijayashree, B., Ahuja, Y.R, Regniers, L., Verschaeve, L. (2005). Genotoxicity of theMusi river (Hyderabad, India) investigated with the Vitotox R© test. Folia Biologica, 51,133–139.

31. Brits, E., Regniers, L., Verschaeve, L. (2006). Bacterial genotoxicity tests; Umu-C andVITOTOX test: a comparison with the Ames assay. In: Topical Issues in Applied Micro-biology and Biotechnology (Verschaeve, L., Ed.). Research Signpost: Trivandrum, India;pp. 11–21, ISBN 81-308-0137-X.

32. Sparg, S.G., Kulkarni,M.G., Light,M.E., van Staden, J. (2005). Improving seedling vigourof indigenous medicinal plants with smoke. Bioresource Technology, 96, 1323–1330.

33. Flematti, G.R., Ghisalberti, E.L., Dixon, K.E., Trengove, R.D. (2004). A compound fromsmoke that promotes seed germination. Science, 305, 977.

34. van Staden, J., Jager, A.K., Light, M.E., Burger, B.V. (2004). Isolation of the majorgermination cue from plant-derived smoke. South African Journal of Botany, 70, 654–659.

35. Verschaeve, L., Maes, J., Light, M.E., van Staden, J. (2006). Genetic toxicity testing of3-methyl-2H-furo[2,3-c]pyran-2-one, an important biologically active compound fromplant-derived smoke. Mutation Research, 611, 89–95.

36. Jain, N., Soos, V., Balazs, E., van Staden, J. (2008). Changes in cellular macromolecules(DNA, RNA and protein) during seed germination in tomato, following the use of


a butenolide, isolated from plant-derived smoke. Plant Growth Regulation, 54, 105–113.

37. Trinh, C., Gevaert, L., Kohout, L., Van Staden, J., Verschaeve, L. (2010). Genotoxicityevaluation of two kinds of smoke water and 3,7-dimethyl-2H-furo[2,3-c]pyran-2-one.Journal of Applied Toxicology, 30, 596–602.

38. Stewart, M.J., Moar, J.J., Steenkamp, V., Kokot, M. (1999). Findings in fatal cases of poi-soning attributed to traditional remedies in South Africa. Forensic Science International,101, 177–183.

39. Thomson, S.A. (2000). South African Government Genocide and Ethnopiracy. The GaiaResearch Institute: South Africa.

40. Arora, S., Brits, E., Kaur, S., Kaur, K., Kumar, S., Verschaeve, L. (2005). Evaluation ofgenotoxicity of medicinal plant extracts by the comet- and VITOTOX test. Journal ofEnvironmental Pathology, Toxicology and Oncology, 24, 193–200.

41. Verschaeve, L., Van Staden, J. (2008). Mutagenic and antimutagenic properties of SouthAfrican medicinal plant extracts. Journal of Ethnopharmacology, 119, 575–587.

42. van den Bout-van den Beukel, C., Hamza, O.J.M., Moshi, M.J., Matee, M.O.N., Mokx, F.,Burger, D.M., Koopmans, P.P., Verweij, P.E., Schoonen, W.G.E.J., van der Ven, A.J.A.M.(2008). Evaluation of cytotoxic, genotoxic and CYP450 enzymatic competition effects ofTanzanian plant extracts traditionally used for treatment of fungal infections. Basic andClinical Pharmacology and Toxicology, 102, 515–526.

43. Edziri, H., Mastouri, M., Mahjoub, A., Anthonissen, R., Mertens, B., Cammaerts, S.,Gevaert, L., Verschaeve, L. (2011). Toxic and mutagenic properties of extracts fromTunisian traditional medicinal plants investigated by the neutral red uptake-, VITOTOX-and alkaline comet assays. South African Journal of Botany, 77, 703–710.

44. Chichioco-Hernandez, C., Wudarski, J., Gevaert, L., Verschaeve, L. (2011). Evaluationof cytotoxicity and genotoxicity of some Phillipine medicinal plants. PharmacognosyMagazine, 7, 171–175.

45. IARC. (2002). IARC Monographs on the Evaluation of Carcinogenic Risks to Humans.Vol. 80; Non Ionizing Radiation, Part 1: Static and Extremely Low-Frequency (ELF)Electric and Magnetic Fields. IARC Press: Lyon, France.

46. Baan, R., Lauby-Secretan, B., El Ghissassi, F. et al. on behalf of the WHO InternationalAgency for Research on Cancer Monograph Working Group. (2011). Carcinogenicity ofradiofrequency electromagnetic fields. Lancet Oncology, 12, 624–626.

47. IARC. (2011). IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Vol.102; Non Ionizing Radiation, Part 2: Radiofrequency Electromagnetic Fields [includesmobile telephones]. IARC Press: Lyon, France; in press.

48. Verschaeve, L., Anthonissen, R., Gevaert, L., Maes, A. (2011). Genotoxicity investiga-tion of ELF-magnetic fields in Salmonella typhimurium with the sensitive SOS-basedVITOTOX test. Bioelectromagnetics, 32, 580–584.

49. Bergqvist, U., Brix, J., de Gruijl, F., de Seze, R., Hietanen, M., Jeffereys, J.G.R., Lagroye,I., Lotz, G.W., Owen, R.D., Repacholi, M.H., Saunders, R., Tenforde, T.S., Verschaeve, L.,Veyret, B. (2003). Review of experimental investigations of EMF biological effects (0–100kHz) – ICNIRP Standing committee II. In: Exposure to Static and Low Frequency Elec-tromagnetic Fields, Biological Effects and Health Consequences (Matthes, R., McKinley,A., Bernhardt, J., Vecchia, P., Veyret, B., Eds.). ICNIRP13/2003, ISBN 3-934994-03-2.

12GENOTOXICITY ANDCARCINOGENICITY: REGULATORYAND NOVEL TEST METHODS

Walter M.A. Westerink, Joe C.R. Stevenson, G. JeanHorbach, Femke M. van de Water, Beppy van de Waart, andWillem G.E.J. Schoonen

12.1 INTRODUCTION

Failure rates of 30–40% for toxicity are common and an important reason for drugattrition within the pharmaceutical industry. These high attrition rates in the caseof drug safety assessment are due to either organ or tissue toxicity and can affect,depending on the chemical and behavioral properties of the drug, multiple organs ortissues. To increase the productivity and develop therapeutics for unexplored targetsnew assay methods, new techniques in liquid handling, robotics, analytical tools, andsoftware were developed. Despite these efforts, the yearly number of approved newdrugs still declines [1]. The average developmental costs of a new drug are high andestimated at approximately 800 million USD [2]. If the costs of the high attrition rateof drugs, which is approximately 90%, are included, these costs even increase up to1.5 billion USD.The process of drug development can be divided into four different phases, that

is, discovery, exploratory development, full development, and the launching of thedrug. This whole process usually takes approximately 12–16 years.

1. In the discovery phase biological targets are validated, followed by high-throughput screening for hit or lead identification. Then, lead optimizationis performed to increase the potency and selectivity of the lead compound. Thisphase of 3–4 years ends with the selection of a development candidate.


233

234 GENOTOXICITY AND CARCINOGENICITY

Hepatotoxicity

Renal toxicity

Neural toxicity

Skin toxicity

Blood toxicity

Reproductive toxicity

Gastrointestinal toxicity

0 5 10 15

% Toxicity failure rates

20 25 30

Genotoxicity/carcinogenicity

Cardiovascular safety

FIGURE 12.1 Toxicity failure rates (%) of drugs that were developed at legacy NVOrganon(�), Merck & BMS (�), and Roche (�) between 1960 and 2000 [4, 5].

2. In the exploratory development phase, the preclinical animal and first into man(phase I) studies are carried out to assess the safety of the drug, including assaysto detect genotoxic potential. If no adverse effects are identified with this firstset of regulatory assays, an investigational new drug (IND) application is filedto the regulatory authorities such as the Food and Drug Administration (FDA)and the EuropeanMedicines Agency (EMA). Then, the first into man studies tostudy the safety and pharmacodynamics of the drug in healthy human volunteersare initiated. Successful completion of this second phase of 3–4 years leads tothe acquirement of a statement of no objection (SNOB).

3. In the full developmental phase, large clinical studies (phase II, III) to provethe efficacy and safety of the drug are performed. Moreover, complex in vivoanimal studies such as carcinogenicity testing are carried out. If in these studiesno serious adverse side effects are identified, this leads to the delivery of a fulldevelopmental candidate, which after positive review by regulatory authoritieswill result in the launching of a new drug. This phase takes 6–8 years and isthe most expensive phase of drug development.

4. Then, the introduction within the clinical market as well as pharmacovigilanceare initiated.

Since 30–40% of the new drug candidates fail in the developmental phase due totoxicological side effects [1, 3], it is very beneficial to recognize these adverse effectsof new chemical entities (NCEs) earlier. As shown in Figure 12.1, the toxicity causes

GENOTOXICITY 235

identified with a high prevalence are hepatotoxicity, nephrotoxicity, cardiovascularsafety, neurotoxicity, skin toxicity, reproductive and developmental toxicity (terato-genicity), and genotoxicity/carcinogenicity. The number of compounds that fail dueto each of these endpoints depends on the portfolio of disease-related drug classes ofpharmaceutical companies. Depending on the portfolio, the detection of drugs withgenotoxic or carcinogenic potential for example can account for 6 up to even 20%of the drug failures due to toxicity. Screening for these adverse effects in the leadoptimization phase of drug development, followed by compound deselection, opti-mization, and final selection, will definitely result in drug development candidateswith an improved success rate.Due to the properties of the regulatory tests used to detect the genotoxic potential

of drugs (i.e., low throughput, requirement for a large amount of compound, and lowspecificity), these tests are difficult to use in the lead optimization phase. Recently,several novel higher throughput test systems have been developed that can moreeasily be incorporated into a lead optimization strategy.In this chapter, a review is given on the standard regulatory tests and newly devel-

oped higher throughput test systems. The sensitivity, specificity, and predictivity ofthe tests are discussed. Moreover, a strategy for the implementation of the novel high-throughput assays in the lead optimization phase of drug development is proposed.

12.2 GENOTOXICITY

Genotoxic compounds (genotoxicants) are capable of inducing mutations in DNA.These mutations are defined as permanent changes within DNA, which can be seg-regated into three different types. First, small changes in the DNA can occur at thelevel of bases, the so-called single-point mutations or gene mutations. These smallmutations can result in base-pair substitutions, deletions, or addition of bases, thelatter two resulting in so-called frameshift mutations. On the other hand, some of thebase-pair mutations can be silent. Second, two types of mutations can be defined atthe level of the chromosomes. Structural chromosomal aberrations, also known asclastogenicity, are defined as structural changes within the chromosome. These majorchanges are due to chromosomal breakages, deletions, exchanges, or rearrangementsof DNA. The compounds that cause these structural chromosomal changes are knownas clastogens. The second type of chromosomal mutations refers to the inappropriatesegregation of the chromosomes between the nuclei, causing a change in the numberof chromosomes. This phenomenon is called aneuploidy and is caused by aneugeniccompounds. A genotoxic compound can cause one or more types of these mutations.Genotoxic compounds can be subclassified into compounds that are DNA reactive

themselves. Examples are compounds that are DNA cross-linkers and directly formDNA adducts. Besides that, there are compounds that do not interfere with the DNA,but react with particular proteins within the cell, thereby indirectly affecting theDNA. An example is the inhibition of the enzyme topoisomerase, which influencesthe DNA replication and causes DNA damage (double-strand breaks) in an indirectway. Several genotoxic compounds need metabolic activation before the genotoxic


mode of action manifests itself. Such compounds are called proximate genotoxicantsor carcinogens.

12.3 CARCINOGENICITY

Carcinogens are involved in the induction of tumor formation and can act eitheras genotoxicants or as nongenotoxicants. The description of tumor development(carcinogenesis) is explained by two different models, that is, the multistage modelof Armitage and Doll [4] and the initiation-promotion model of Berenblum [5]. Bothmodels assume the development of a tumor to be a multistage process in whichseveral subsequent mutations are needed for the transformation of a normal cellinto an autonomous growing neoplastic cell. The initiation phase starts with DNAdamage due to endogenous or exogenous factors, as for example exposure to DNA-reactive or indirect genotoxic compounds. After this initiation phase, the mutationsare irreversibly fixed within the DNA, leading to the start of the promotion phase,in which the neoplastic cell develops and becomes manifested into a clinically andpathologically observable tumor. The promotion phase is followed by the progressionphase, which involves the fast growth of the tumor and coincides with the invasionof tumor cells into the surrounding tissue and metastasis.Nongenotoxic carcinogens induce tumor formation by several mechanisms [6].

These mechanisms may include receptor-mediated induction/stimulation of car-cinogenesis, cytotoxicity, endocrine modification, immunosuppression, inflamma-tion, oxidative stress, hyper-/hypomethylation, and inhibition of gap-junction-mediated intercellular communications. Especially the receptor-mediated induc-tion/stimulation is responsible for several forms of rodent- and/or human-specificcarcinogenicity. Examples are species-specific activation of the aryl hydrocarbonreceptor (AhR), the peroxisome proliferator activated receptor � (PPAR�), and theconstitutive androstane receptor (CAR) [8]. Assays that measure the activation ofthese receptors in rodent and human cells in the lead optimization phase of drugdevelopment may be useful for the early detection of rodent and human nongeno-toxic carcinogens. The detailed description of such assays is not within the scope ofthis chapter.

12.4 REGULATORY TESTS TO DETECT PHARMACEUTICALSWITH GENOTOXIC AND CARCINOGENIC POTENTIAL

In regulatory genotoxicity testing, a stepwise tiered approach is applied [7]. In a firststep, in vitro assays with a high sensitivity are used to identify test compounds havinga high intrinsic genotoxic activity. In a second step, specific in vivo tests to determinethe relevance of the in vitro results for the in vivo situation are performed. These invivo genotoxicity studies are also included, because some of the genotoxicants canonly be detected in vivo after metabolic activation [8]. A more detailed decision tree

REGULATORY TESTS TO DETECT PHARMACEUTICALS 237

for this regulatory genotoxicity testing including a description of necessary follow-uptesting is described later in this chapter.When compared to regulatory carcinogenicity testing, genotoxicity testing is rela-

tively cheap and fast. Compounds without genotoxic liability can proceed to the firstinto man clinical trials. The carcinogenic potential, on the other hand, is assessedlater in the full developmental phase of drug development. The regulatory test strat-egy consists of a battery of core and ancillary tests for the identification of the threeforms of genotoxicity (gene mutations, clastogenicity, and aneugenicity), which can-not be detected in one single test. This required standard test battery for genotoxicitytesting of pharmaceuticals is described in the ICH guideline S2B for the registrationof pharmaceuticals for human use [9]. The core battery consists of:

1. The Ames assay to detect gene mutations in bacteria

2. An in vitro chromosome aberration (CA) and/or mouse lymphoma thymidinekinase (TK) assay in mammalian cells

3. An in vivo chromosome damage assay (CA or micronucleus assay).

Specific technical aspects of these regulatory tests are described in ICH guidelineS2A for the registration of pharmaceuticals for human use [9]. These ICH S2A/S2Bguidelines are currently under revision and will be replaced by the S2(R1) guidelinein 2012/2013. In this revised guideline, there are two options for genotoxicity testing:

Option 1:1. The Ames assay to detect gene mutations in bacteria

2. A cytogenetic test to detect chromosomal damage (the in vitro CA test or invitro micronucleus [IVMN] test) or an in vitro mouse lymphoma TK genemutation assay

3. An in vivo test for genotoxicity, generally a test for chromosomal damageusing rodent hematopoietic cells, either for micronuclei or for chromosomalaberrations in metaphase cells.

Option 2:1. The Ames assay to detect gene mutations in bacteria

2. An in vivo assessment of genotoxicity with two different tissues, usually anassay for micronuclei using rodent hematopoietic cells and a second in vivoassay. Typically this would be a Comet assay in liver.

Option 1 in the S2(R1) guideline is quite similar to the tiered approach in theS2B guideline. The difference is the addition of the IVMN assay. Option 2 describesa new test approach, and the Comet assay is a new core test in this test strategy.A major difference in the new guidelines is the 10-fold decrease of the maximumrequired test concentration in the mammalian assay from 10 mM to 1 mM to increase


TABLE 12.1 Performance Definitions for Genotoxicity Testsa

Term Definition

Sensitivity Percentage of carcinogens, genotoxic carcinogens, or in vivo genotoxinspositive in the test

Specificity Percentage of noncarcinogens, nongenotoxic carcinogens + noncarcinogens,or in vivo nongenotoxins negative in the test

Predictivity Percentage of all tested compounds that was predicted correctly

aDepending on the compound set, the definitions are based on the classification of reference compounds ascarcinogens vs. noncarcinogens, genotoxic carcinogens vs. nongenotoxic carcinogens + noncarcinogens,and in vivo genotoxins vs. in vivo nongenotoxins.

the specificity of the regulatory test battery. There are also minor changes in the newguidelines, but the description of these changes is not within the scope of this chapter.The assays of the standard test battery have a different sensitivity, specificity,

and predictivity for carcinogenicity. The performance definitions are summarizedin Table 12.1. For the calculations of the sensitivity, the results of the genotoxicityassays are compared with the results of the carcinogenicity tests. It is howeverimportant to note that several of the carcinogenic compounds act via a nongenotoxicmode of action. Genotoxicity tests will thus never reach a sensitivity of 100% forcarcinogenicity. The commonly used assays from the regulatory genotoxicity testbattery will be described in more detail in Sections 12.4.1–12.4.5, while the overallperformance scores for each test are summarized in Table 12.2. These scores are stillbased on the 10 mM concentration as maximum test concentration, since the scoresbased on the reduced 1 mM level are not yet available.

12.4.1 The Ames Assay

The Ames assay was developed by Bruce Ames and assesses the mutagenic potentialof chemical compounds [10, 11]. This assay is performed in bacteria with test strainsof Salmonella typhimurium, which carry different point and frameshift mutations ina gene involved in histidine biosynthesis or in four S. typhimurium tester strains incombination with one Escherichia coli strain, which due to a gene mutation cannot

TABLE 12.2 The Sensitivity, Specificity, and Predictivity of the Assays of theStandard Regulatory Test Battery for the Assessment of Carcinogenicity [12, 16]

Assay Sensitivity (%) Specificity (%) Predictivity (%)

Ames test 58.8 73.9 62.5Chromosome aberration test 65.6 44.9 59.8Mouse lymphoma TK test 73.1 39.0 62.9Micronucleus test in vitro 78.7 30.8 67.8Micronucleus test in vivo 40.0 75.0 48.0


synthesize tryptophan. The mutated bacterial cells depend on histidine or tryptophanfor their growth and are so-called histidine or tryptophan auxotrophs. Mutagens cancause a reverse mutation, which results in bacteria that are able to grow again onhistidine- or tryptophan-deficient medium. The total number of bacterial coloniesformed is then used as a measure for the mutagenic potential of a compound withrespect to its vehicle control.The diverse strains are used to be able to detect the different kinds of mutagens,

which act via different mechanisms of mutagenicity. Besides the mutations in thehistidine- or tryptophan-synthesizing gene, the test strains also have additional muta-tions to make the strains more sensitive for the detection of mutagens. A mutation inthe genes needed for lipopolysaccharide synthesis, for example, makes the bacterialcell wall of S. typhimuriummore permeable. Moreover, all the strains have a mutationin their excision repair system [10].To mimic the metabolism in bacterial (and mammalian) mutagenicity assays, a

liver fraction (S9 mixture) containing phase I and II drug-metabolizing enzymes fromAroclor 1254-treated male Sprague-Dawley rats is used. Aroclor 1254 stimulates theAhR, pregnane X receptor (PXR), and CAR and leads to high levels of cytochromeP450 (CYP) 1A1,CYP1A2,CYP2B, andCYP3A,which are involved in the activationof a large number of proximate genotoxicants. Therefore, these mutagenicity assaysare performed in the presence and absence of S9mixture to studywhether compoundsare activated or inactivated by metabolism.The specificity of the Ames assay is relatively high in comparison to the other

in vitro genotoxicity tests (Table 12.2). The sensitivity, specificity, and predictivityof the Ames assay calculated by Kirkland et al. [12] were 58.8, 73.9, and 62.5%,respectively.

12.4.2 The Chromosome Aberration (CA) Assay

The CA assay is performed in vitro in cultured mammalian cells. Structural andnumerical damage is scored by microscopic examination of chromosomes in mitoticcells in metaphase. Tests are carried out in the presence and absence of the S9mixture [13, 14]. This assay is often performed in Chinese hamster ovary k1 cells(CHO-k1 cells), in Chinese hamster lung cells (V79 cells) or in mouse, rat, or humanlymphocytes. Scoring needs specialized training and experience. The sensitivity andpredictivity of this test for carcinogenicity are 65.6 and 59.8%, respectively (Table12.2). The specificity of this test is low (44.9%) [12].

12.4.3 The Mouse Lymphoma TK Assay (MLA)

Thymidine monophosphate (TMP) is one of the four desoxyribonucleotidemonophosphates, which does not undergo significant conversion into othernucleotides. Therefore, under normal growth conditions, this TMP pool, being quitesmall and constant, remains well conserved and serves as an important regulatorof DNA synthesis. TMP is normally synthesized by the enzyme TK by phospho-rylation of thymidine. If thymidine is replaced by a lethal precursor analogue, like


5-bromo-deoxyuridine (BrdU) or trifluorothymidine (TFT), cells will die due to thephosphorylation of this precursor. TK-deficient cells lack this enzyme activity andbecome resistant to the cytotoxic effect of the lethal analogue. In the mouse lym-phoma TK assay, the TK-competent L5178Y (TK+ /+ or TK+ /–) cells are treatedwith the test compounds. After treatment, the cells are shifted to a selective mediumcontaining the lethal thymidine analogue, in which normally most cells will die.However, in the presence of a mutagenic compound TK–/– cells, which are resistantto the formation of the cytotoxic precursor, may have been formed. The number ofcell colonies on the test plates is therefore a measure of genotoxicity. The size ofthe colonies gives information about chromosome damage, as large changes in DNAinhibit growth and result in small colonies, whereas large colonies denote gene muta-tions. The sensitivity and predictivity of the mouse lymphoma TK assay are 73.1 and62.9%, respectively (Table 12.2). As in the case of the CA assay, the specificity ofthis assay is low (39%) [12].

12.4.4 The Micronucleus Test

The fourth regulatory genotoxicity assay is the micronucleus test. Chromosomalfragments or complete chromosomes that are the result of DNA damage or errors inthe separation of chromosomes during the cell cycle can sometimes be found outsidethe nucleus in one of the daughter cells. After the nucleus has been divided, theseDNA fragments will decondensate and form a so-calledmicronucleus. By using DNAstaining techniques, these micronuclei become visible and can be counted under themicroscope. The number of these micronuclei per 1000 (binucleated) cells is usedas a measure for genotoxicity. This assay can be performed in vitro with cell linessuch as CHO-k1 and V-79. Micronuclei can also be measured in red blood cellsand bone marrow obtained from in vivo experiments. By using centromeric probes,it is possible to determine whether micronuclei contain complete chromosomes orfragments of chromosomes. These results can then be used to determine whethercompounds possess a clastogenic or aneugenic mode of action [15]. The sensitivity,specificity, and predictivity of the IVMN assay are 78.7, 30.8, and 67.8%, respectively(Table 12.1). The specificity of the in vivomicronucleus assay in bonemarrow ismuchhigher (75%). The sensitivity of the in vivo test is lower (40%) and the predictivity is48% [12, 16].

12.4.5 The Alkaline Comet Assay (Single-Cell Gel Electrophoresis Assay)

The Comet assay was introduced by Singh et al. [17] as a microgel electrophoretictechnique for the detection of DNA damage in single cells by alkaline pretreatment atpH levels above 13. By alkaline treatment above pH 12, the DNA becomes denaturedand unwinded due to the disruption of the hydrogen bonds between the base pairsof the double strands. This already leads to the identification of DNA single-strandbreaks (SSBs). At pH conditions of 12.6 or higher, alkaline labile sites (ALSs), that is,apurinic sites, are also quickly converted into strand breaks. Thus at pH>13, both theexpression of SSB and ALS becomes maximized, leading to optimal sensitivity. The


Comet assay is performed onmicroscopic slides, which are prepared with compound-pretreated living cells embedded in an agarose layer. Subsequently, these cells arelysed in a 1.0% Triton solution for 1 h to release the DNA from the cells, followedby a treatment with an alkaline solution at pH >13 for 20 to 60 min. Thereafter,electrophoresis is carried out at 4◦C for 30 min at 1 V/cm in an electrophoreticbuffer often supplemented with either dimethyl sulfoxide or 8-hydroxyquinolineas radical scavengers to prevent DNA damage due to the experimental procedure.After electrophoresis, the gel is neutralized and stained for DNA with ethidiumbromide, propidium iodide, 4,6-diamidino-2-phenylindole (DAPI), SYBRGreen I, orYOYO-1 [18].Thereafter, the amount of DNA in the Comet tail, tail length, and tail moment

are measured to assess the amount of DNA damage. The big advantage of the invivo Comet assay is the low amount of tissue needed and therefore the possibility tocombine this assay with the in vivo micronucleus assay. Moreover, this assay can beperformed in several tissues and (i.e., liver, blood, and the bone marrow). Most in vivomicronucleus-negative carcinogens forming DNA adducts are detected in the in vivoComet assay [19]. Currently, more modern versions of the Comet assay have becomeavailable, in which a high content (HC) analysis in 24-well plates is performed.

12.4.6 The Impact of Positive Findings for Genotoxic Potentialand Follow-up Testing Strategies

A decision tree for the tiered approach in regulatory genotoxicity testing is shown inFigure 12.2. In general, a combination of the Ames test + MLA + (or) CA assay isused for regulatory in vitro genotoxicity testing [12]. A combination of the Ames test+ MLA + CA assay has a high sensitivity (84.7%), but low specificity (22.9%) forcarcinogenicity [12]. If these tests are negative, the next step is to perform an in vivomicronucleus test. This in vivo test is performed, because some compounds are poorlydetected in vitro. For example, proximate carcinogens, which are activated by phaseII enzymes, are not identified by the current methodologies [20, 21]. If the in vivomicronucleus assay also shows a negative result, it is very likely that the compoundhas no genotoxic potential and the compound can proceed in the development.More tests have to be performed in the rare situation that the in vivomicronucleus

assay gives a positive result after negative results in vitro. It has been shown thatcompounds that increase or decrease the core body temperature for a sustained period,compounds that increase the erythropoiesis in the bone morrow, and compounds thatinhibit protein synthesis induce the number of micronuclei in bone marrow in vivo.Experiments to show thesemodes of action have been described by an IWGTworkinggroup [22]. Such positive results are mostly irrelevant for humans. Mechanistic datato demonstrate the lack of clinical relevance for humans or a non-DNA-reactivemechanism can lead to continuation of the drug candidate development. For non-DNA-reactive genotoxicants (e.g., topoisomerase inhibitors and spindle poisons),a threshold may even be justified. In case of a DNA-reactive mode of action, thedevelopment is normally terminated.


In vitro test battery+

+

+ + –

+––

–

In vivo micronucleus assay

No genotoxic potentialor threshhold mode of action

CONTINUATION

A for humans relevant DNAreactive mode of action

DISCONTINUTION

Additional in vitro and or in vivotesting to assess the mode of action

In vivo micronuclei assaySecond in vivo test (USD or Comet)

FIGURE 12.2 Schematic overview of the decision tree and necessary follow-up testing forregulatory genotoxicity testing in case of pharmaceuticals for human application.

In the case of a positive result in one of the in vitro genotoxicity assays, atleast two follow-up in vivo genotoxicity tests are required. These are the in vivomicronucleus assay and a further test. In the past, the unscheduled DNA synthesis(UDS) test was often used, but nowadays the Comet assay is preferred when testinghuman pharmaceuticals [23]. This is due to the fact that most in vivo micronucleus-negative carcinogens forming DNA adducts are detected in the Comet assay. Of thesecompounds, less than 20% are detected in the UDS assay [23]. Besides these two invivo assays, it might be useful to perform additional in vitro and in vivo assays toelucidate the mode of action causing the positive result.In most cases, the in vivo genotoxicity assays will show a negative result after a

positive in vitro result, as the in vitro assays give a high number of false-positives. Ina retrospective analysis, Kirkland et al. [12, 24] showed that the genotoxicity batteryused in the tiered approach is highly sensitive. About 80–90% of the carcinogensare detected; however, the specificity of the in vitro mammalian genotoxicity assaysis very low [12, 24]. This is also supported by the retrospective analysis performedby Snyder and Green [25]. They showed that 50% of noncarcinogenic marketeddrugs are positive in the mammalian genotoxicity assays, demonstrating the highfalse-positive rate of these tests.Two negative results in vivo principally overrule a positive result in vitro. However,

in case of the development of pharmaceuticals for human use, additional studies areoften initiated to get a clue about the mechanistic basis of the positive result. Inthe case that the in vivo tests are positive, additional experiments might also beuseful to show whether the positive in vivo result is relevant for humans or whetherthe compound acts by a threshold mode of action. In this way, the compound can


TABLE 12.3 Summary of Human Nonrelevant, Indirect, or Threshold Mechanismof Genotoxicitya

Mode of Action DescriptionIn vitro SystemAffected

Possibility toObtain

ExperimentalEvidence

In vitro specific Rat S9 mixture-specific effects

Feeding effects

All, except primaryhepatocytes

Bacteria

Reasonable

ReasonableDirect DNAeffect butwith athreshold

Azo- and nitro-reductionDNA repair deficiencyInadequate detoxificationMetabolic overload (productionofreactive oxygen species,lipid peroxidation andsulphydryl depletion)

BacteriaAllAllMammalian cells

ReasonableDifficultReasonableReasonable

Indirect effect Inhibition of topoisomerasesInhibition of kinasesInhibition of DNA polymerasesImbalance of DNA precursorsEnergy depletionInhibition of protein synthesisNuclease release from lysosomesProtein denaturationAneuploidyHigh toxicity

Mammalian cellsMammalian cellsMammalian cellsMammalian cellsMammalian cellsMammalian cellsMammalian cells

ReasonableReasonableReasonableReasonableDifficultDifficultDifficultDifficultReasonableReasonable

aThe in vitro systems affected and the probability to obtain experimental evidence to support themechanismare shown [26].

be saved from attrition. A summary of human nonrelevant, indirect, or thresholdmechanisms of genotoxicity is given in Table 12.3.In a paper by Kirkland et al. [26], in vitro approaches to determine whether these

effects occur in or are relevant for humans are described. However, the difficultyresides in predicting what mechanism is affected by a compound giving a positive invitro or in vivo genotoxicity result. Toxicogenomic approachesmight be very valuablein this respect, as they can give a clue about the mechanism of action [27].In general, the following test strategy is used to assess the mode of action of a

compound after a positive in vitro result suspected to be irrelevant for the humansituation or due to a mechanism with a certain threshold [26]. First, in vitro assays areperformed to show the indirect or threshold mode of action. Thereafter, in vivo testsare performed. If these tests are positive, evidence must be obtained that this positiveresult is caused by the same mode of action. In case of a human-relevant non-DNAor threshold mode of action, the “No Observed Adverse Effect Level” (NOAEL)must be determined. If the anticipated human dose is much lower, development ofthe compound might continue. In case of a human-relevant DNA-reactive mode ofaction, the development of the compound is usually terminated.


12.4.7 Carcinogenicity Testing

Testing of carcinogenicity is performed later in the process of drug development(parallel to phase II–III clinical studies) and is required for the final market approvalof a compound [28]. Upon availability of the results from the carcinogenicity studies,the genotoxicity results are used as part of the weight of evidence in cancer riskassessment. The assessment of carcinogenicity is described in guideline ICH S1B forregistration of pharmaceuticals for human use [29]. According to this guideline, theinduction of tumors is monitored in a 2-year, lifetime exposure of mice and rats [30].As an alternative to the mouse bioassay, a medium-term transgenic mouse model canbe used [31]. There are several compounds that are carcinogenic in rodents but actby a mechanism that is irrelevant to humans (e.g., compounds that activate PPAR�and CAR in rodents). Therefore, follow-up testing to show the mechanism of actionof the compound is necessary to assess the relevance to humans.

12.5 SCREENING FOR COMPOUNDS WITH GENOTOXIC ANDCARCINOGENIC POTENTIAL WITHIN THE LEAD OPTIMIZATIONPHASE OF DRUG DEVELOPMENT

To improve the success rate of NCEs, early screening of compounds for geno-toxic/carcinogenic potential may be an important tool. The introduction of fastscreening procedures to detect the genotoxic/carcinogenic potential of drug can-didates for the selection or deselection of compounds in the discovery phase of drugdevelopment may be very beneficial [32]. Optimization of compounds in the earlylead optimization phase regarding pharmacologic properties, pharmacokinetics, aswell as bioavailability has already been shown to be a successful strategy, therebyleading to lower drug attrition rates [1]. Likewise, this strategy may be very useful inthe case of genotoxicity/carcinogenicity, which at least at Legacy Organon accountedfor 20% of toxicity failure.Such an approach to avoid drug attrition due to genotoxicity and carcinogenicity

has already been started within the pharmaceutical industry. Hereto, in silico quanti-tative structure activity relationship analysis is often applied in the lead optimizationphase. Software programs, such as DEREK, Topcat, Multicase, and Mutalert, areroutinely used during lead optimization [33]. In addition, potential drugs are oftentested in non-GLP versions of the micronucleus assay and miniaturized versions ofthe Ames tests at the end of the lead optimization phase just before the selectionof the development candidates. These miniaturized Ames versions are still laboriousand need approximately 300 mg of compound, which makes the application of theassay in the early lead optimization phase difficult [34].Screening in the early lead optimization phase raises several issues. During the

early lead optimization phase the number of compounds per discovery project isrelatively high. Moreover, available amount of test compound is low (typically 10–200 mg). This makes that the regulatory test systems for toxicity, which have alow throughput and require a high amount of compound, are not applicable to early

SCREENING WITHIN LEAD OPTIMIZATION 245

toxicity screening. Thus, early testing for toxicological hazard requires medium- orhigh-throughput assays with a high specificity; if not, too many pharmacologicallyinteresting compounds will be deselected. Hence, the exact number of compoundsto be tested depends on the time point in the lead optimization phase in which theassays will be performed. If assays are performed early in the lead optimizationphase, it is possible to use the assays for the optimization of compounds. If assaysare used just before the selection of a development candidate, the assays are used fordeselection.The regulatory genotoxicity assays, asmentioned above, generally have a relatively

low throughput, need a high amount of compound, and are laborious. Furthermore,the specificity of most of these in vitro genotoxicity assays is too low to makeuse of them in the lead optimization phase (Table 12.2). Therefore, in their currentformat these assays are not applicable for medium- or high-throughput screening inthe lead optimization phase. Nevertheless, a few commercially available screeningassays for genotoxicity have been developed, but in general these assays have onlybeen validated with a limited number of compounds. Thus, for early genotoxicityscreening in vitro assays with a higher throughput, a higher specificity, a reducedamount of compound use, and a better validation with proper reference compoundsare needed.In the next paragraphs a strategy to develop, optimize, and validate in vitro assays

for genotoxic carcinogenicity is briefly described. Higher throughput assays based onbacteria, yeast, and human/rodent cell lines are presented and proposed as useful invitromodels. In case of the human cell lines the main focus is on HepG2 cells, as theproperties of these cells are expected to give a better prediction of in vivo genotoxicitythan primary hepatocytes or CHO-k1 and V79 cells. After a brief description of theassays, the sensitivity, specificity, and predictivity of the assays is discussed andcompared with those of the regulatory tests.For this comparison, a compound list recommended by the European Centre for

the Validation of Alternative Methods (ECVAM) was used, as data were available formost test systems. This ECVAM compound list consists of 20 well-defined genotoxiccarcinogens and 42 nongenotoxic compounds. Of the 20 genotoxic carcinogens, 14compounds are positive in the Ames test. The remaining genotoxins cause chro-mosome damage, but give negative or equivocal results in the Ames assay. The 42nongenotoxic compounds (noncarcinogens and nongenotoxic carcinogens) contain19 compounds that often give false-positive results in in vitromammalian genotoxicityassays. Moreover, an additional set of 192 compounds is used herein for comparisonpurposes. The compounds of this additional set can be classified as nongenotoxinsand genotoxins and consist of both in-house compounds from the legacy NVOrganonandwell-defined reference compounds. The additional compound list contains severalsteroidal compounds that have been reported to be clastogenic or aneugenic. Of these192 compounds, Ames test results are available for 145 compounds, mammalian invitro genotoxicity data for 124 compounds, and in vivo genotoxicity data for 70 com-pounds. The individual scores (sensitivity and specificity) of the higher throughputtest systems based on bacteria, yeast, and mammalian cell lines are summarized inTable 12.4. The scores are also depicted in Figures 12.4 and 12.5.


TABLE 12.4 Sensitivity and Specificity Scores of the Novel Higher Throughput Test

ECVAM Additional

Tests Sensitivity (%) Specificity (%) Sensitivity (%) Specificity (%)

VitotoxTM 79 (11/14) 93 (39/42) 86 (42/49) 94 (90/96)RadarScreen 70 (14/20) 83 (35/42) 68 (25/37) 38 (12/32)HCS CHO-k1 IVMN 80 (16/20) 88 (37/42)HCS HepG2 IVMN 60 (12/20) 88 (27/42)HepG2 p53_luc 85 (17/20) 93 (39/42) 71 (27/38) 81 (26/32)HepG2 cystatin A_luc 70 (14/20) 90 (38/42) 61 (23/38) 94 (30/32)HepG2 RAD51C_luc 60 (12/20) 93 (39/42) 29 (11/38) 94 (30/32)HepG2 Nrf2_luc 75 (15/20) 31 (13/42) 76 (29/38) 66 (21/32)GADD45a-GFPAGreenScreen HC

90 (18/20) 85 (35/41)

The scores are given for the additional compound list from the Legacy Organon and/or for the ECVAMcompound list. In the case of theVitotoxTM assay, the scores are given for the correlationwithAmes data. Incase of the other tests, the scores are given for the correlation with genotoxic carcinogenicity vs. nongeno-toxic carcinogenicity + noncarcinogens (ECVAM) or in vivo genotoxicity vs. in vivo nongenotoxicity(additional compounds Legacy Organon).

12.5.1 The Detection of Genotoxicity with Bacterial Screens

Several miniaturized versions of the Ames test are available, such as the MPF Ames,which show a good prediction rate for Ames test results at the end of the leadoptimization phase. These miniaturized versions of the Ames test still have a relativelow throughput and need a relative high amount of compound (approximately 300mg). Higher throughput alternatives for the Ames assay that are more applicableduring lead optimization are the SOS chromotest and VitotoxTM assay. These assaysare based on the induction of bacterial DNA damage repair genes, the so-called SOSresponse. This gene induction can be used for the measurement of the mutagenicpotential of compounds.The SOS chromotest is a colorimetric assay that measures the activation of the

sfiA gene involved in the SOS response in bacteria [35]. In this assay the promoter ofsfiA is linked to the lacZ gene. Activation of the SOS response therefore results in anincreased production of the lacZ protein, which can be measured colorimetrically orluminometrically.The VitotoxTM assay is a more modern version of this assay. The VitotoxTM assay

is a bioluminescent assay with a medium-throughput test potential that also exploitsthe bacterial SOS response mechanism involved in DNA repair [36, 37]. One ofthe important proteins involved in the SOS response is the recN protein, which isunder nonstress conditions repressed by the LexA protein. In the VitotoxTM assaya genetically modified S. typhimurium strain TA104 that contains the lux operon ofVibrio fischeri under transcriptional control of the promoter of the recN gene is used.After incubation of the bacteria in the presence of a genotoxic compound, the recN


promoter is activated as the repressor LexA protein is inactivated. This results in lightproduction, thereby reflecting genotoxicity (bacterial mutagenicity).In case of the ECVAM compound list the correlation with Ames test results was

79% (11/14). The VitotoxTM assay gave a low number of false-positive results as thespecificity (percentage of nongenotoxic compounds that tested negative) was 93%(39/42). Similar high correlations were observed when the additional compound setof Legacy Organon was tested. For these compounds, the correlation with the Amestest was 91% (sensitivity, 86% [42/49]; specificity, 94% [90/96]) [38]. Based onthe available data, the VitotoxTM assay is a good and rapid prescreen for bacterialmutagenicity that easily can be incorporated into a lead optimization strategy.

12.5.2 The Detection of Genotoxicity with Yeast Screens

Bacterial tests have the disadvantage that clastogenic and aneugenic compounds can-not be detected due to the absence of chromosomes and chromosomal segregation.Eukaryotic yeast-based assays may overcome this mechanistic lack by measuring theactivation of specific DNA repair genes involved in chromosomal damage. In thisrespect, several publications have described the application of a RAD54 promoterlinked to a green fluorescent protein (GFP) protein, the so-called GreenScreen assay[39–43]. Induction of the RAD54 promoter due to DNA damage results in increas-ingly fluorescent cells. A similar approach was chosen for the RadarScreen assay, inwhich the RAD54 promoter is linked to a �-galactosidase reporter and luminometricquantification. The advantage of the luminometric readout is that autofluorescence ofcompounds and the S9 mixture does not hamper the measurements.Validation of the assay with respect to the ECVAM compound list resulted in a

correlation with (in vivo) genotoxicity of 79% (sensitivity, 70% [14/20]; specificity,83% [35/42]). Validation with the additional compound set from Legacy Organongave more false-positive results. The predictivity for this compound set was only54% (sensitivity, 68% [25/37]; specificity, 38% [12/32]). Especially the steroids inthe second list of compounds gave false-positive results for in vivo genotoxicity.The exact reason for the results obtained is not known. The explanation for thelow predictivity could be a lack of (species) differences in metabolizing enzymes. Inaddition, aneugens were not detected with the RadarScreen assay. It can be concludedthat application of this assay may lead to the deselection of too many valuablecompounds in the lead optimization phase [38].

12.5.3 The Detection of Genotoxicity with Rodent and Human CellLine-Based Screens

Besides bacterial and yeast cells, mammalian cells can be used to measure the geno-toxic potential of compounds. Especially human cells may be very valuable, becausethey may better reflect the human situation.For the development of human cell-based genotoxicity assays, the choice of pri-

mary and/or permanent cells is a very important aspect. At a first glance, primary


cells seem to be the best choice. However, for the fixation of mutations, cells need todivide. Most primary cells brought into culture are not yet adapted to the high oxygenconcentrations in the cell incubators, are very stressed, and only divide a few times.Thus, dividing permanent cell lines may be more preferable.However, the problem with permanent cell lines is that many have lost their p53

response as well as their capacities for DNA repair and drug metabolization. Inthis respect the human hepatoma HepG2 cells might be helpful, because these cellsdisplay active p53, DNA repair, and phase I and II metabolism, as explained in thenext section. Therefore, different types of assays can be performed with HepG2cells, among them being high content screening (HCS) with an IVMN assay, DNAdamage- and tumor suppression-related biomarker promoter linkage to luciferase-based reporter assays, which are described in more detail later in this chapter. Anothervaluable human cell line for genotoxicity testing might be the human lymphoblastoidcell line TK6, which is used as host for the GADD45a-GFP GreenScreen HC assay[44–47].

12.5.4 Human HepG2 and TK6 Cells

HepG2 cells are derived from liver hepatoma cells, which have retained their phaseI and II metabolizing enzyme activities, enzyme activities that are normally lost incultured cells [48–50]. To mimic metabolic potential in in vitro systems, the rat S9mixture is normally used for genotoxicity testing. Although this mixture properlyperforms phase I metabolism, the metabolites may be unable to pass the human cellmembranes due to the hydrophilicity of the metabolites. Further disadvantages of therat S9mixture are the absence of phase II metabolism and the fact that human-specificmetabolites will not be generated.Cytotoxicity studies in HepG2 cells showed that several compounds, which need

metabolic activation, for example, benzo[a]pyrene,were toxic in this cell line [51, 52].This implied the presence of phase Imetabolic activity inHepG2 cells. Besides phase Imetabolism, phase II metabolism is crucial for genotoxicity testing. Phase II enzymescan play a role in the detoxification of reactive intermediates, thereby preventinggenotoxicity, or in the bioactivation of proximate carcinogens. Phase II enzymeslike sulfotransferases (SULTs) and N-acetyltransferases (NATs) are important forthe activation of many genotoxicants. Glatt et al. [20, 21] reported that more than100 proximate mutagens are activated by SULTs. SULTs are inactive in the S9mixture due to the lack of the proper cofactor in it. Addition of the particular cofactordoes not directly solve this problem, because the conjugates cannot easily enterthe cells due to their charge and hydrophilic character. Therefore, SULT-activatedproximate genotoxicants with S9 mixtures are missed in the standard in vitro testsystems, whereas they have been reported as being genotoxic in HepG2 cells [53].NAT activity is important for the detection of heterocyclic aromatic amines suchas 2-amino-3-methylimidazo[4,5-f]quinoline (IQ) because such compounds requireacetylation for their activation. The levels of these enzymes are low in CHO-k1 cells.Due to NAT activity in HepG2 cells, IVMN assay scores with heterocyclic aromaticamines correlated much better with in vivo carcinogenic effects than results obtained


with other cell lines like CHO-k1 cells. The micronucleus test scores of HepG2studies correlated even better with carcinogenicity in laboratory animals than in vivomicronucleus tests in bone marrow of mice [49]. Further examples are the applicationof HepG2 cells to elucidate the genotoxic potential of mycotoxins [54, 55]. Some ofthese compounds are carcinogenic in vivo, genotoxic in HepG2 cells, but negative inall in vitro regulatory genotoxicity assays. Furthermore, HepG2 cells give promisingresults as several human-specific carcinogens are detected. A more detailed phaseI and II metabolic characterization of the HepG2 cells has recently been published[56–58].Briefly, transcript levels of CYP1A1, 1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6, 2E1,

and 3A4 were measured with quantitative PCR. However, mRNA levels of mostCYPs were significantly lower than in cryopreserved primary human hepatocytes.These results were confirmed with luminometric assays for CYP1A1, 1A2, 2C9, and3A4. Moreover, the stimulation of CYP1A1, 1A2, 2B6, and 3A4 mRNA levels byAhR, Pregnane X Receptor (PXR), and CAR was similar to the regulation in primaryhuman hepatocytes at the mRNA and/or enzyme level. In contrast, CYP2C8 mRNAlevels are inducible with PXR/CAR activators in cryopreserved human hepatocytes,but not in HepG2 cells. Consistent with other studies, CYP2D6 and 2E1 transcriptlevels were not modified after treatment with AhR, PXR, and CAR activators. Thus,although phase I enzyme levels are significantly lower than in primary cells, the levelsof these enzymes are significantly increased after activation of xenobiotic receptors,which explains the toxic effects of several compounds that need metabolic activationin this cell line [56].The transcript levels of phase II UDP-glucuronosyltransferases (UGTs), sul-

fotransferases (SULTs), glutathione S-transferases (GSTs), N-acetyltransferase-1(NAT1), and epoxide hydrolase (EPHX1) enzymes were also measured with quan-titative PCR in HepG2 cells and cryopreserved primary human hepatocytes. Levelsof SULT1A1, 1A2, 1E1, 1A2, and 2A1, microsomal GST 1and GST �1, NAT1 andEPHX1 in HepG2 cells were very similar to levels in cryopreserved primary humanhepatocytes. In contrast, mRNA levels of UGT1A1 and 1A6 were between 10- and1000-fold higher in cryopreserved primary hepatocytes. AhR, PXR, and CAR activa-tors stimulate similar changes in HepG2 cells as in primary human hepatocytes. Thepresence of these phase II enzymes in HepG2 cells can be a valuable cellular systemto predict toxicity for these compounds [57].Another advantage of HepG2 cells for genotoxicity testing is the presence of a

functionally active p53 protein [59]. The tumor suppressor p53 is a potent transcrip-tion factor inducing the expression of genes involved in growth arrest, apoptosis, andDNA repair. Additionally, p53 suppresses the expression of several genes involved instimulating cell division [60]. Under normal circumstances, p53 is kept under tightcontrol by MDM2, an ubiquitin E3 ligase that mediates the ubiquitylation and degra-dation of p53. In response to a range of different cellular stresses [61], of which DNAdamage is the most important one, p53-MDM2 protein interaction is interrupted withthe result that p53 degradation is blocked [62]. Following DNA damage p53 rapidlyaccumulates. The activation of p53-regulated genes results in either cell cycle arrestallowing DNA repair or in p53-dependent apoptosis or cellular senescence in order to


avoid the propagation of genetically defective cells [60, 62]. One of the p53-activatedgenes is the cyclin-dependent kinase (Cdk) inhibitor protein p21. This protein bindsto G1/S-Cdk and S-Cdk and thereby blocks entry into the S-phase.A functionally active p53 protein is also needed for a proper functional nucleotide

excision repair and double-strand break repair. Wang et al. [63] demonstrated thatp53 plays a prominent role in the damage recognition and assembly of the repairmachinery during global genome repair. The recruitment of the Xeroderma pig-mentosum C (XPC) gene, which plays a prominent role in nucleotide excisionrepair, is p53 dependent. Chang et al. [64] showed in the alkaline Comet assaythat after the introduction of p53 into p53-null cells (cells without a functionallyactive p53 protein) the excision repair activity was restored. Similar results werereported in a microarray experiment [65]. The basal expression of XPC was rela-tively low in lymphoblasts that did not have a functionally active p53 protein. Afterthe onset of DNA damage by radiation, only minor changes in the expression ofXPC occurred. In this study, similar effects were observed for the radiation gene 51homolog C (RAD51C) gene, which plays a prominent role in double-strand breakrepair [65].Another important characteristic of HepG2 cells is the presence of the

nuclear factor (erythroid-derived 2)-like 2 (Nrf2) system [66]. Studies haveshown that activation of Nrf2 induces the transcription of phase II detoxifyingenzymes, antioxidant enzymes, and transporter genes that protect against (geno-)toxic compounds. Examples of Nrf2-induced genes are GSTs, NAD(P)H:quinoneoxidoreductase 1, heme oxygenase-1, SULTs, UGTs, and multidrug resistance-associated proteins. The protective effect is based on the metabolization of reactiveoxygen species and (geno-)toxic compounds, followed by excretion. Under home-ostatic cellular conditions, Nrf2 is retained in the cytoplasm by the Keap1 protein.Due to oxidative stress, Nrf2 is released from Keap1 and can translocate into thenucleus, where it activates genes with an electrophile-responsive element (also calledantioxidant-responsive element) in the promoter region such as those coding for theabove-mentioned phase II metabolizing enzymes, transporters (phase III enzymes),and antioxidant genes.Thus the presence of phase I and II metabolism, active p53 proteins, DNA repair

enzymes, and an Nrf2 system in HepG2 cells might be an advantage in (geno-)toxicity testing over the more commonly used cell lines, such as V79 and CHO-k1cell lines, in which these systems are absent or less functional. Several studies havealready shown that HepG2 cells give good results (high sensitivity and specificity) ingenotoxicity assays such as the IVMN test and the Comet assay [67–73]. However,the used compound sets in these studies were small and mainly consisted of positivecontrols. Therefore, the use of HepG2 cells might prove to be an adequate tool tostudy the role of these systems in genotoxicity and lead to a reduction of the numberof falsely predicted positive results. Several genotoxicity assays that were developedwith HepG2 cells are described in the next sections. Moreover, the GADD45a-GFPreporter assay in the human lymphoblastoid cell line TK6 is described. This TK6 cellline contains, as does the HepG2 cell line, a functionally active p53 protein.


12.5.5 HCS Micronucleus Assays in the Rodent Cell Line CHO-k1and the Human Cell Line HepG2

Normally, the scoring of micronuclei on microscopic slides is performed by well-trained operators. This is laborious and leads to a low throughput. By using the HCSassay technique, higher throughput testing with 96- or 384-well plates is feasible,making this micronucleus test suitable for screening purposes in the lead optimizationphase. Briefly, the HCS micronucleus test combines fluorescent multiple-well platemicroscopy with automated image analysis. A first evaluation of a HCS micronucleiassay was performed by Diaz et al. [74]. Additional advantages of HCS are: (i)multiple parameters, such asmicronuclei formation and cytotoxicity, can bemeasuredsimultaneously in the same cells, thereby improving the interpretation of genotoxicityresults; (ii) the diameter of micronuclei can be scored, thereby leading to a clearsegregation of aneugens with large-sized micronuclei and clastogens with small-sized micronuclei. The throughput of this assay is 21 up to 42 compounds per weekwith 5 mg of compound followed by fixation of the cells in the well plates. Theseplates can be used for quantification for more than a half year, making this applicationvery practical [75].The in vitro mammalian genotoxicity assays in CHO-k1 and V79 cells give

in general too many falsely predicted positive results for genotoxicity. To deter-mine which test system would lead to optimal predictions, the compounds of theECVAM reference list were tested in the HCS micronucleus assay with HepG2 andCHO-k1 cells and the results compared regarding sensitivity and specificity. Thisoutcome should give a stronger insight into the discrepancies of the high num-ber of falsely predicted positive results in the regulatory mammalian genotoxicityassays.The sensitivity (number of genotoxic carcinogens correctly predicted; 16/20, 80%)

and the specificity (percentage of nongenotoxins correctly predicted; 37/42, 88%) ofthe assaywith theCHO-k1 cell linewere high. Proximate genotoxins that are activatedby CYP1A (i.e., benzo[a]pyrene and 7,12-dimethylbenzanthracene) were genotoxicin CHO-k1 cells in the absence of the S9 mixture, indicating that these enzymes areprobably active in CHO-k1 cells. Surprisingly, addition of S9 mixture resulted evenin a decrease in the genotoxic effects of these compounds.The sensitivity of the HepG2 cell line was lower (12/20; 60%), whereas the

specificity was high (37/42, 88%). The metabolic capacity in HepG2 cells made itpossible to detect most of the compounds in the ECVAM list that need metabolicactivity to exert their genotoxic activity. Seven out of the nine proximate geno-toxins showed genotoxic potential without the addition of S9 mixture, that is,cyclophosphamide, benzo[a]pyrene, 7,12-dimethylbenzanthracene, dimethylni-trosamine, 2,4-diaminotoluene, 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine(PhIP), and aflatoxin B1. The tests with the proximate genotoxins 2-acetylaminofluorene and IQ were negative, and the addition of S9 mixture did notresult in activation of these two compounds. It was also shown that based on the sizeof the micronuclei genotoxins could be classified as clastogens or aneugens.


Thus, HCS micronucleus assays with CHO-k1 and HepG2 cells are able to detectthe genotoxic (chromosome damaging) potential of chemicals and allow to differen-tiate between clastogens and aneugens.

12.5.6 The GADD45a-GFP GreenScreen HC Genotoxicity Assayin TK6 Cells

Promoters or responsive elements of human genes involved in cell cycle control orof genes that are downstream targets of p53 can be used for the development ofreporter assays that rapidly assess mammalian (human) genotoxicity. Based on thisprinciple, Gentronix has developed the GADD45a-GFP GreenScreen HC reporterassay [76]. Mutagens, clastogens, and aneugens cause increased expression of thehuman GADD45a gene, which is a downstream target of p53. This principle is thebasis of theGADD45a-GFPGreenScreenHCgenotoxicity assay, inwhichGADD45aexpression is linked to the expression of GFP. The host cell line for the GADD45areporter construct is the human lymphoblastoid cell line TK6. The GADD45a-GFPreporter assays have extensively been validated [44–46]. Several compound sets weretested, including the ECVAM compound list for which the sensitivity, specificity, andpredictivity were 90 (18/20), 85 (35/41), and 87% (53/61), respectively. Addition ofthe S9 mixture is needed in the assay as TK6 cells lack drug-metabolizing enzymes.However, this results in a lower throughput and sensitivity [76]. Validation withfurther compound sets gave similar results (i.e., high sensitivity and high specificity).

12.5.7 Luminescence-Based Reporter Assays in the HepG2 Cell Line

Luminescence-based reporter assays in HepG2 cells are promising as a model systemto detect genotoxicity, and based on this principle several reporter cell lines have beendeveloped. Genotoxicity reporter assays containing the responsive element of p53 andthe promoter region of RAD51C and cystatin A have been developed in the HepG2cell line. Both RAD51C and cystatin A are downstream targets of p53. RAD51Cis involved in homologous recombination (double-strand break repair) and cystatinA is a downstream target of p53 in the apoptotic pathway. Moreover, a luciferase-based reporter assay that measures the activation of the Nrf2 electrophile-responsivepathway was generated. The assay principle of the luciferase-based reporter assaysis shown in Figure 12.3.Testing the ECVAM compound list resulted in a predictivity (total percentage of

correctly predicted genotoxic carcinogens plus nongenotoxic compounds) of 82% forthe HepG2 RAD51C_luc assay (sensitivity, 60% [12/20]; specificity, 93% [39/42]),of 84% for the HepG2 cystatin A_luc assay (sensitivity, 70% [14/20]; specificity,90% [38/42]), and of 90% for the HepG2 p53_luc assay (sensitivity, 85% [17/20];specificity, 93% [39/42]). In addition, the percentage of genotoxic compounds thatactivated the Nrf2 pathway was high (75%, 15/20). In the case of the nongeno-toxic compounds in the ECVAM list, only 31% (13/42) activated the Nrf2 path-way (Fig. 12.4).


Genotoxic stress/Oxidative stress

+Luciferin→

96-well plate1

ABCDEFGH

2 3 4 5 6 7 8 9 10 11 12

Iuciferase

RAD51C promoterCystatin A promoterp53 responsive elementNrf2 responsive element

FIGURE 12.3 Assay principle for the luciferase-based reporter assays with the p53- andNrf2-responsive elements and the cystatin A and RAD51C promoters.

The metabolic capacity in HepG2 cells was sufficient to activate seven out of thenine proximate genotoxins, that is, benzo[a]pyrene, 7,12-dimethylbenzanthracene,2-acetylaminofluorene, 2,4-diaminotoluene, IQ, PhIP, and aflatoxin B1. Of theseseven compounds, 2-acetylaminofluorene and IQ were not detected in the HepG2HCS IVMN assay, which might indicate that the activation of the p53-responsiveelement is a more sensitive endpoint for these compounds in HepG2 cells than theformation of micronuclei (DNA damage). Moreover, the activation of the DNA repairresponse in HepG2 cells may have prevented a significant induction of micronucleiin the HepG2 HCS IVMN assay. The proximate genotoxins cyclophosphamide anddimethylnitrosamine were not detected in the genotoxicity reporter assays. This couldbe due to the fact that cyclophosphamide and dimethylnitrosamine are activated byCYP2B6 and 2E1, respectively, and that these CYPs are expressed at low levels inHepG2 cells.The data from the genotoxicity reporter assays were also compared with the avail-

able data on in vivo genotoxicity for the additional set of 192 compounds from theLegacy Organon. For the individual genotoxicity reporter assays, the correlationswith in vivo genotoxicity were 76% for the HepG2 p53_luc assay (sensitivity, 71%[27/38]; specificity, 81% [26/32]), 59% for the HepG2 RAD51C_luc assay (sen-sitivity, 29% [11/38]; specificity, 94% [30/32]), and 76% for the HepG2 cystatinA_luc assay (sensitivity, 61% [23/38]; specificity, 94% [30/32]). As observed withthe ECVAM compound list, the HepG2 p53_luc assay and cystatin A_luc assayare assays with a relative high sensitivity and specificity. The HepG2 RAD51C_lucassay has a lower sensitivity but the specificity is high, thus indicating that only a fewcompounds induce double DNA strand breaks. Of the 108 compounds in the addi-tional list of 192 compounds that show a positive result for bacterial or mammaliangenotoxicity, 62% (67/108 compounds) activate the Nrf2 pathway. This percentage

Gen

oto

xic

Fals

e po

siti

ves

No

n-g

eno

toxi

c

Test systemAmes Vito

tox RadarscreenIn vivo r

egulatory

In vitro m

ammalian regulatory

HCS IVMN CHO

HCS IVMN-HepG2

Cystatin A_luc

p53_luc Rad51C_lucGADD45a GC HC

Nrf2_luc

Test systemAmesVito


egulatory

In vitro m

ammalian regulatory

HCS IVMN CHO

HCS IVMN-HepG2

Cystatin A_luc


Nrf2_luc

Test system Ames Vitotox Radarscreen

In vivo regulatory

In vitro m

ammalian regulatory

HCS IVMN CHO

HCS IVMN-HepG2

Cystatin A_luc


Nrf2_luc

2-A

AF

o-A

nthr

anili

c ac

idA

mitr

ole

Afla

toxi

n B

1B

enzy

l alc

ohol

Am

pici

llin

trih

ydra

teA

zido

thym

idin

et-

But

ylhy

droq

uino

net-

But

yl a

lcoh

olB

enzo

[a]p

yren

e (B

[a]P

)C

urcu

min

n-B

utyl

chl

orid

e C

adm

ium

chl

orid

e2,

4-D

ichl

orop

heno

l(2

-chl

oroe

thyl

)trim

ethy

l-NH

4Cl

Chl

oram

phen

icol

1,3-

Dih

ydro

xybe

nzen

eC

yclo

hexa

none

Cis

plat

inE

thio

nam

ide

N,N

-Dic

yclo

hexy

l thi

oure

aC

yclo

phos

pham

ide

Eth

yl a

cryl

ate

Die

than

olam

ine

2,4-

Dia

min

otol

uene

2-E

thyl

-1,3

-hex

aned

iol

Di-(

2-et

hylh

exyl

)pht

hala

te7,

12-D

MB

AE

ugen

olE

phid

rine

sulp

hate

Dim

ethy

lnitr

osam

ine

Isob

utyl

alde

hyde

Ery

thro

myc

in s

tear

ate

EN

U

D,L

-Men

thol

Flu

omet

ron

Eto

posi

dep-

Nitr

ophe

nol

Hex

achl

oroe

than

eH

ydro

quin

one

Pro

pyl g

alla

teD

-Lim

onen

eIQ

Pht

halic

anh

ydrid

eD

-Man

nito

lM

MS

Sod

ium

sac

char

inM

elam

ine

p-C

hlor

oani

line

Sod

ium

xyl

ene

sulfo

nate

M

ethy

l car

bam

ate

PhI

P.H

Cl

Sul

fisox

azol

eP

hena

nthr

ene

Sod

ium

ars

enite

Ure

aP

henf

orm

in H

CL

Taxo

lP

roge

ster

one

Pyr

idin

eTr

is(2

-eth

ylhe

xyl)p

hosp

hate

Tris

odiu

m E

DTA

trih

ydra

te

FIG

UR

E12

.4SchematicoverviewoftheresultsoftheECVAMcompoundlistintheAmestest,VitotoxTMtest,RadarScreentest,

invi

voregulatory

genotoxicitytests,

invi

tromammalianregulatorygenotoxicitytests,HCSIVMNCHO-K1assay,HCSIVMNHepG2assay,luciferase-basedreporterassays

withHepG2cellsforcystatinA,p53,RAD51CandNrf2,andtheGADD45a-GFPGreenScreenHCassay.Apositiveresultisindicatedinred,while

equivocalY/Nfindingisindicatedinyellow,noeffectingreen,andnodataavailableinwhite.(

See

inse

rtfo

rco

lor

repr

esen

tati

onof

this

figur

e.)

254

COMPARISON OF THE SENSITIVITY AND SPECIFICITY 255

was only 38% (30/80) in the group of compounds with negative or no genotoxicitydata. These data confirm the results that were found with the ECVAM compoundlist, that is, that a large percentage of the genotoxic compounds activate the Nrf2pathway. Activation of the Nrf2 pathway gives information about the mode of actionof a genotoxic compound. However, this assay should not be used to identify andsubsequently deselect genotoxic compounds, as this pathway is also activated bycytotoxic compounds and by various beneficial compounds that protect cells againstgenotoxic and cytotoxic compounds [77–79]. Activation of this pathway results inupregulation of phase-II detoxifying enzymes and antioxidant stress proteins. Inaddition, cross-talk between the p53 and Nrf2 pathways makes the individual roleof Nrf2 in the genotoxic potential assessment as such very difficult [80] (Figs. 12.5and 12.6).In summary, three luciferase-based reporter assays for the rapid assessment of the

genotoxic potential of drugs have been generated. The p53 and cystatin A reporterassays have a high sensitivity and specificity for predicting genotoxic carcinogenic-ity and in vivo genotoxicity. The RAD51C reporter assay is a more specific assaythat gives information about the formation of double-strand breaks. In addition,information about Nrf2 activation can help to elucidate the mode of action of agenotoxicant.

12.6 COMPARISON OF THE SENSITIVITY AND SPECIFICITY OF THEREGULATORY AND NOVEL HIGHER THROUGHPUT IN VITROGENOTOXICITY ASSAY

Several high-throughput assays for the detection of the genotoxic potential of drugshave been described. In this section, the sensitivity and specificity of these newlydeveloped high-throughput in vitro genotoxicity assays are compared with the reg-ulatory in vitro genotoxicity assays. Since in the current regulatory practice combi-nations of assays are used to detect bacterial mutagenicity (gene mutations) as wellas mammalian genotoxicity (chromosome damage), a combined use of the novel testmethods will be discussed.

12.6.1 Sensitivity and Specificity of Combinations of Regulatory In VitroGenotoxicity Assays

Kirkland et al. [12] evaluated the ability of the single and combined use of theregulatory in vitro genotoxicity tests to discriminate rodent carcinogens and non-carcinogens. Hereby, the authors stated that it is important to differentiate betweencarcinogens that act via a genotoxic and those that act via a nongenotoxic mode ofaction, as carcinogens that act via a nongenotoxic mode cannot be readily detectedin in vitro genotoxicity assays [12].The performance scores of the individual in vitro genotoxicity assays are already

summarized in Table 12.2. These data show that the specificity of the individualmammalian in vitro genotoxicity assays for noncarcinogens is relatively low, thereby


Genotoxic compounds

Test

sys

tem

Ames

Vito

tox

Rad

arsc

reen

In v

ivor

egul

ator

yA

In v

itro

mam

alia

n re

gula

tory

Cys

tatin

A_l

uc

p53_

luc_

luc

Rad

51C

_luc

Nrf

2_lu

c

Test

sys

tem

Ames

Vito

tox

Rad

arsc

reen

In v

ivo r

egul

ator

y

In v

itro

mam

alia

n re

gula

tory

Cys

tatin

A_l

uc

p53_

luc_

luc

Rad

51C

_luc

Nrf

2_lu

c

1-Nitropyrene 16α-Hydroxy-Estrone

2,4-Dinitrophenol 17α-Methyltestosterone

2,7-Dinitrofluorene Acetaminophen

*** see below Allylestrenol

2-Hydroxy-Estradiol Bromobenzene

2-Hydroxy-Estrone Canreonate K+

2-Methoxy-Estradiol Carbon tetrachloride

2-Methoxy-Estrone Chlormadinone

3-Methylcholantrene Chlormadinone acetate

4-Hydroxy-Estradiol Chlorpromazine

4-Hydroxy-Estrone Chlorprothixene citrate

4-Nitroquinoline Oxide Clomiphene

Dacarbazine Colchicine

Dantrolene Cyproterone acetate

Doxorubicin Cytarabine

Ellipticin Dexamethasone

Hydralazine Diclofenac

Hydrogen peroxide Diethylstilbestrol

β-Naphthoflavone Diethylthio carbamic acid

Melphalan Dihydroergotamine

Methampyrone Drospirenone

Nitrofurantoin Dydrogesterone

Org 2249 Equilin

Org 2408 Equilin-7α-Methyl

Org 2508 Estradiol-17αOrg 3240 Estradiol-17βOrg 4122 Estriol (E3)

Org 4330 Estrone (E1)

Org 5694 Ethinylestradiol-17β (EE)

Org 5695 Hexachlorobutadiene

Org 5697 Hydrochlorothiazide

Org 5710 Hydroxychloroquine sulfate

Org 5741 ICI 164.384

Org 5784 Imipramine HCl

Org 5796 Levonorgestrel

Org 5867 Methoxyprogesterone

Org 5907 Megestrol acetate

Org 7797 MENT-Bucyclate

Org 9063 Mestranol

Org 9150 Methotrexate

Org 9250 Moxestrol

Org 9252 Norethisterone

Org 20494 Norethynodrel

Org 32018 Noscapine HCl

Org 42671 Org 10325

Salicylamide Org 30029

Tacrine Org 30251

UK-57400 Org 39735

Uramustine Org 4433

5-Fluorouracil Oxymethalone

Rifampicin

***=2-amino-3-methyl-3H-imidazo-[4,5-f]-quinoline Rotenone

Stanozolol

Sulfinpyranone

Tamoxifen

Testosterone

Tolcapone

Trenbolone


resulting in a high percentage of false-positives. In general, a combination of theAmes test + MLA + CA assays is used in regulatory testing [12] (Table 12.5).This combination has a high sensitivity for carcinogens (84.7%, 171/202), but anextremely low specificity for noncarcinogens of only 22.9% (22/96). The combineduse of two other tests (i.e., Ames + CA, AMES + IVMN, AMES + MLA) givessimilar scores [12]. However, the overlap between the compounds used byKirkland etal. [12] and the compounds used for the validation of the novel test methods is ratherlimited. Therefore, a straightforward comparison is not possible. Moreover, sincemechanistic data were lacking for most carcinogens, it was not possible for Kirklandet al. [12] to calculate the performance scores for genotoxic carcinogenicity.To assess the sensitivity and specificity of the novel in vitromodels, the mentioned

“ECVAM compound list” was used [13]. Application of the regulatory accepted teststrategy in the case of the ECVAM compounds results in a sensitivity for genotoxiccarcinogens of 100% (20/20) and specificity of 55% (23/42) (Table 12.5). The lowspecificity is mainly due to the in vitromammalian genotoxicity assays (MLA + CA)that gave 19 false-positive results. Moreover, an additional list of 192 compoundswas used. Genotoxicity and carcinogenicity data are limited for these compounds,and a clear classification in genotoxic carcinogens and nongenotoxins (nongenotoxiccarcinogens + noncarcinogens) is not available. Consequently, it was decided tocompare the assay data with in vivo genotoxicity data, which are the best to dis-criminate between genotoxic carcinogens and nongenotoxins. With this additionalcompound set, the combination AMES + in vitro mammalian assays (Table 12.5)shows a sensitivity of 97% (37/38) and a specificity of 41% (12/29) for in vivo geno-toxicity. Here again, the low specificity is caused by the results from the in vitromammalian genotoxicity assays (Table 12.5).Overall, the two compound sets give, although the way of comparing them was

different, similar results as those in the Kirkland study [12]. A combination of theAMES + in vitro mammalian genotoxicity assays gives a high sensitivity but lowspecificity. The ECVAM and the additional compound list make a direct compar-ison between results from regulatory in vitro genotoxicity tests (Ames + in vitromammalian assays) and the novel higher throughput systems possible. Different com-binations of novel test systems are discussed below and the scores are compared withthose of the regulatory in vitro genotoxicity assays.

�FIGURE 12.5 Schematic overview of the results for the 108 in-house and referencecompounds of the additional compound list from the Legacy Organon that are genotoxicin at least one of the regulatory genotoxicity assays. Results are shown for the Ames test,VitotoxTM test, RadarScreen test, in vivo regulatory tests, in vitro mammalian regulatorygenotoxicity tests, and luciferase-based reporter assays with HepG2 cells for cystatin A, p53,RAD51C, and Nrf2. A positive result is indicated in red, while equivocal Y/N finding isindicated in yellow, no effect in green, and no data available in white. Chemical nomenclatureof Org compounds is given in Table 12.7. (See insert for color representation of thisfigure.)


Non-genotoxic or no data available

Test

sys

tem

Ames

Vito

tox

Rad

arsc

reen

In v

ivo

regu

lato

ry

In v

itro

mam

alia

n re

gula

tory

Cys

tatin

A_l

uc

p53_

luc_

luc

Rad

51C

_luc

Nrf

2_lu

c

Test

sys

tem

Ames

Vito

tox

Rad

arsc

reen

In v

ivo

regu

lato

ry

In v

itro

mam

alia

n re

gula

tory

Cys

tatin

A_l

uc

p53_

luc_

luc

Rad

51C

_luc

Nrf

2_lu

c

2,5-Hexanedione Org 13011

7α-Methylnorethisterone Org 20091

Acetylsalicylic acid Org 20223

Aminophylline Org 20241

Amiodarone Org 20350

Antazoline mesylate Org 20660

Atamestane Org 30189

Atropine sulfate Org 30535

Bishydroxycoumarin Org 30659

CERM 11884 Org 30701

CERM 13061 Org 31710

Clozapine Org 32608

Corticosterone Org 32782

Cortisol Org 34037

Dehydroepiandrosterone Org 34694

Dimethisterone Org 34850

Dopamine Org 34901

Erythromycin Org 37445

Ethacrynic acid Org 42788

Ethionine Orphenadrine citrate

Ferrous sulphate Papaverine HCl

Fluoxymesterone - 17α-CH3 Perhexiline

Flutamide Perphenazine

Furafylline Phentolamine mesylate

Gentamycin A Propylmesterone

Indomethacin Quinidine

Iodoacetate Quinidine sulfate

Iproniazid R1881

Isoprenaline Raloxifen

Ketoconazole Reserpine

Labetalol RU 58668

L-DOPA Strychnine

Lilopristone +17α-CH3 Sulfamoxole

Methadone Sulfaphenazole

Naphazoline nitrate Tularik 0191317

N-ethylmaleimide Digoxin

Onapristone Org 30002

Org 3362 Org 9217

Org 4060 Org 9935

Org 4428

Org 4874

Org 5168

Org 7258

Org 9340

Org 10490

FIGURE 12.6 Schematic overview of the results for the 84 in-house and reference com-pounds from the additional compound list of the Legacy Organon that are nongenotoxic orhave no genotoxicity data in the regulatory genotoxicity tests. Results are shown for the Amestest, VitotoxTM test, RadarScreen test, in vivo regulatory tests, in vitro mammalian regulatorygenotoxicity tests, and luciferase-based reporter assays with HepG2 cells for cystatin A, p53,RAD51C, and Nrf2. A positive result is indicated in red, while equivocal Y/N finding is indi-cated in yellow, no effect in green, and no data available in white. Chemical nomenclature ofOrg compounds is given in Table 12.7. (See insert for color representation of this figure.)


TABLE 12.5 Sensitivity and Specificity Scores (of combinations) of Regulatory InVitro Genotoxicity Assays for Three Compound SetsCompoundSet

Endpoint

Kirkland et al.[12]

Carcinogenicity

Kirkland et al.[81] (ECVAM)Genotoxic

Carcinogenicity

Additional Compound Set Legacy Organon

In Vivo Genotoxicity

Tests 1. Ames test2. In vitromammaliangenotox.assays

1. Ames test2. In vitromammaliangenotox.assays

1. Ames test 2. In vitromammaliangenotox.assays

1. Ames test2. In vitromammaliangenotox.assays

Sensitivity % 84.7 (171/202) 100 (20/20) 48 (16/33) 94 (34/36) 97 (37/38)Specificity % 22.9 (22/96) 55 (23/42) 79 (22/28) 46 (12/26) 41(12/29)

12.6.2 Sensitivity and Specificity of Combinations of the Novel HigherThroughput Genotoxicity Assays and Comparison with the Regulatory InVitro Screening Battery

Taking as starting point that a screening battery for early genotoxicity detectionshould include a bacterial assay to detect gene mutations, the VitotoxTM assay shouldbe chosen. The high sensitivity (86%) and specificity (90%) of the VitotoxTM assayand its excellent correlation with Ames test results makes this assay a valuableprescreen that can rapidly predict Ames test results in the lead optimization phase ofdrug development.An early screening battery should also contain assays for the detection of com-

pounds causing chromosome damage. Several assays that have this potential havebeen evaluated. Sensitivity and specificity scores for combinations of these assays,that is, (i) the yeast-based RadarScreen assay, (ii) the CHO-k1 HCS IVMN assay,(iii) the HepG2 HCS IVMN assay, (iv) the luciferase-based p53 reporter assays inHepG2, and (iv) the GADD45a-GFP GreenScreen HC combined with the VitotoxTM,are summarized in Table 12.6:

1. VitotoxTM + RadarScreen: In case of the ECVAM compounds, the sensi-tivity (80%) and specificity (81%) is high, whereas with the additional 192compounds the specificity score is low, being only 28%. For early discoveryscreening, this would lead to a high number of false-positives.

TABLE 12.6 Performance Scores of Combined Use of the Novel Higher ThroughputTest Systems Presented in this Chapter

TestCombination

1. VitotoxTM

2. RadarScreen

1. VitotoxTM

2. HCSIVMNCHO-k1

1. VitotoxTM

2. HCSIVMNHepG2

1. VitotoxTM

2. HepG2 p53_luc

1. VitotoxTM

2. GADD45AGreenScreen

HC

Compound setname

ECVAM Additional ECVAM ECVAM ECVAM Additional ECVAM

Sensitivity (%) 80 (17/20) 73 (27/37) 95 (19/20) 90 (18/20) 85 (17/20) 76 (29/38) 95 (19/20)Specificity (%) 81 (34/42) 28 (9/32) 83 (35/42) 83 (35/42) 88 (37/42) 72 (23/32) 85 (35/41)

TA

BL

E12

.7O

verv

iew

ofth

eIn

-Hou

seC

ompo

und

Cod

esfr

omth

eL

egac

yO

rgan

onan

dth

eC

orre

spon

ding

Che

mic

alN

ames

OrgCode

ChemicalName

CERM11884

N-(2,6-Dimethylphenyl)-beta-[(2-methylpropoxy)methyl]-N-(phenylmethyl)-1-pyrrolidineethanaminehydrochloride

CERM13061

trans-D,L-N-(1,3-Benzodioxol-5-yl)-2-[(cyclohexyloxy)methyl]-1-methyl-N-(phenylmethyl)-3-piperidinamine

hydrochloride(1:1)salt

Org10325

N-Hydroxy-5,6-dimethoxy-1H-indene-2-carboximidamidehydrochloride

Org10490

6,7,8,9-Tetrahydro-7-methyl-5H-dibenz[b,i][1,6]oxazecine(Z)-2-butenedioate(1:1)

Org13011

1-[4-[4-[4-(Trifluoromethyl)-2-pyridinyl]-1-piperazinyl]butyl]-2-pyrrolidinone(E)-2-butenedioate(1:1)

Org20091

5-Chloro-4-(1,2,5,6-tetrahydro-1-methyl-3-pyridinyl)-2-thiazolamine(E)-2-butenedioate(2:1)salt

Org20223

2-Amino-N-(2-phenylethyl)-N-propyl-5-thiazoleethanamine(E)-2-butenedioate(1:1)salt

Org20241

N-Hydroxy-4-(3,4-dimethoxyphenyl)-2-thiazolecarboximidamide

Org20350

4-(1-Butyl-1,2,5,6-tetrahydro-3-pyridinyl)-5-chloro-2-thiazolamine(Z)-2-butenedioate(1:1)salt

Org20494

4,5-Dihydro-5-methyl-6-[6-[2-(1-piperidinyl)ethoxy]benzo[b]thien-2-yl]-3(2H)-pyridazinonehydrochloride

Org20660

2-Chloro-11-(1,2,5,6-tetrahydro-1-methyl-3-pyridinyl)dibenzo[b,f][1,4]thiazepine(E)-2-butenedioate(1:1)salt

Org2249

2,3,9,13b-Tetrahydro-2-methyl-1H-dibenz[c,f]imidazo[1,5-a]azepine

Org2408

1,2,3,4,10,14b-hexahydro-2,7-dimethyldibenzo(c,f)pyrazino(1,2-a)azepineˆ(z)-2-butenedioate(1:1)

Org2508

1-(n-hydroxyamidino)-benzocyclobutenehydrochloride

Org30002

5,6,7,8-Tetrahydro-6-methyldibenz[b,h][1,5]oxazoninehydrochloridehemihydrate

Org30029

N-Hydroxy-5,6-dimethoxybenzo[b]thiophene-2-carboximidamidehydrochloride

Org30189

1-[2-Chloro-3-(2-methylpropoxy)propyl]pyrrolidinemonohydrochloride

Org30251

2,3,4,5,6,7-Hexahydro-N-hydroxy-1H-benz[e]indene-2-carboximidamidehydrochloride

Org30535

cis-1,2,3,12b-Tetrahydro-2,7-dimethyl-3aH-dibenz[2,3:6,7]oxepino[4,5-c]pyrrol-3a-ol-Z)-2-butenedioate(1:1)

Org30701

1-[1-[(2-Methylpropoxy)methyl]-2-[[1-(1-propynyl)cyclohexyl]oxy]ethyl]pyrrolidinehydrochloride

Org31710

(6beta,11beta,17beta)-11-[4-(Dimethylamino)phenyl]-4’,5’-dihydro-6-methylspiro[estra-4,9-diene-17,2’(3’H)-furan]-

3-one

Org32018

1-[2-(Phenylmethoxy)ethyl]-4-(3-phenylpropyl)piperazinedihydrochloride

Org3240

n-hydroxy-3,4-dimethoxybicyclo(4.2.0)octa-1,3,5-triene-7-carboximidamideˆhydrochloride

Org32608

2-[2-(4-Chlorophenoxy)phenyl]-4,5-dihydro-1H-imidazole[Z]-2-butenedioate(1:1)

Org32782

2-(3,4-Dichlorophenoxy)benzenemethanaminehydrochloride

Org3362

n-hydroxy-3,5-dimethylbicyclo(4.2.0)octa-1,3,5-triene-7-carboximidamideˆhydrochloride

Org34037

R(-)-6-(4-Chlorophenyl)-2,3,5,6-tetrahydroimidazo[2,1-a]isoquinoline(E)-2-butenenedioate(1:1)salt

Org34694

(7alpha,11E,17alpha)-11-Ethylidene-17-hydroxy-7-methyl-19-norpregn-4-en-20-yn-3-one

260

Org34850

(11beta,17alpha)-11-[4-(Dimethylamino)phenyl]-17-hydroxy-21-[4-(methylsulfonyl)phenyl]-19-norpregna-4,9-dien-

20-yn-3-one

Org34901

(11beta,17alpha)-17-Hydroxy-3-oxo-11-(1-propynyl)-19-norchola-4,20-dien-24-oicaciddeltalactone

Org37445

(3alpha,11beta,17alpha)-11-(2-Propenyl)-19-norpregn-5(10)-en-20-yne-3,17-diol

Org39735

(7alpha,17beta)-7-Ethyl-17-hydroxyestra-4,14-dien-3-one

Org4060

(11beta,17alpha)-11-Ethyl-17-hydroxy-19-norpregn-4-en-20-yn-3-one

Org4122

N-Hydroxy-3,4-dimethoxybenzenepropanamide

Org42671

2,8-dihydroxy-10chloro-11H-[b]-benzofluorene

Org42788

(3alpha,7alpha,11beta,16alpha,17alpha)-7,11,16-Trimethyl-19-norˆpregn-5(10)-en-20-yne-3,17-diol

Org4330

n-hydroxy-3,4-dimethoxybicyclo(4.2.0)octa-1,3,5-triene-7-carboxamide

Org4428

cis-1,3,4,13b-Tetrahydro-2,10-dimethyldibenz[2,3:6,7]oxepino[4,5-c]pyridin-4a(2H)-ol

Org4433

(11beta,17alpha)-11-Ethynyl-17-hydroxy-19-norpregn-4-en-20-yn-3-one

Org4874

n,3,4-trimethoxybenzenepropanamide

Org5168

3a,4,9,9a-Tetrahydro-6,7-dimethoxy-1H-benz[f]isoindole-1,3(2H)-dione

Org5694

N-(Acetyloxy)-3,4-dimethoxybicyclo[4.2.0]octa-1,3,5-triene-7-carboximidamide

Org5695

3,4-Dimethoxy-N-[[(methylamino)carbonyl]oxy]bicyclo[4.2.0]octa-1,3,5-triene-7-carboximidamide

Org5697

3-(3,4-Dimethoxybicyclo[4.2.0]octa-1,3,5-trien-7-yl)-5-(trifluoromethyl)-1,2,4-oxadiazole

Org5710

N,N’-Dihydroxy-3,4-dimethoxybicyclo[4.2.0]octa-1,3,5-triene-7-carboximidamide

Org5741

N-Hydroxy-3,4-dimethoxy-N-methylbicyclo[4.2.0]octa-1,3,5-triene-7-carboximidamidehydrochloride

Org5784

N-Hydroxy-3,4-dimethoxy-N’-methylbicyclo[4.2.0]octa-1,3,5-triene-7-carboximidamidehydrochloride

Org5796

N-Hydroxy-3,4-dimethoxy-7-methylbicyclo[4.2.0]octa-1,3,5-triene-7-carboximidamide(Z)-2-butenedioate

Org5867

N,3,4-Trimethoxybicyclo[4.2.0]octa-1,3,5-triene-7-carboximidamidehydrochloride

Org5907

cis-N-Hydroxy-3,4-dimethoxy-8-methylbicyclo[4.2.0]octa-1,3,5-triene-7-carboximidamidehydrochloride

Org7258

2-(4-Chlorophenyl)-4-phenyl-1H-indene-1,3(2H)-dione

Org7797

(16alpha,17beta)-17-(Methylamino)estra-1,3,5(10)-triene-3,16-diol(Z)-2-butenedioate

Org9063

3,4-Dihydro-N-hydroxy-6,7-dimethoxy-2-naphthalenecarboximidamidehydrochloride(1:1)salt

Org9150

3-(5,6-Dimethoxy-1H-inden-2-yl)-2-propenenitrile

Org9217

(E)-N-Hydroxy-3-(5,6-dimethoxy-2-benzofuranyl)-2-propenimidamidehydrochloride

Org9250

Sodium

5,6-dimethoxy-N-hydroxybenzo[b]thiophene-2-carboxamidemonohydrate

Org9252

4-Chloro-N-hydroxy-5,6-dimethoxybenzo[b]thiophene-2-carboximidamidehydrochloride

Org9340

DL-(1alpha,2beta,4alpha)-4-([1,1’-Biphenyl]-4-ylmethyl)-2-[(1,1-dimethylethyl)amino]

cyclohexanolmethanesulfonate(1:1)salt

Org9935

4,5-Dihydro-6-(5,6-dimethoxybenzo[b]thien-2-yl)-5-methyl-3(2H)-pyridazinone

261


2. VitotoxTM + HCS IVMN CHO-k1: This combination gives for the ECVAMcompounds a sensitivity of 95% (19/20) and a specificity of 83% (35/42),which is much higher than the specificity of 55% (23/42) of the combinationregulatory Ames + in vitro mammalian genotoxicity assays. Causes for thisdifference might be an active p53 protein in CHO-k1 cells, as the p53 inhibitorpifithrin-� was active in our cells, and/or the application of the HCS techniquegiving more consistent and reliable results than manual scoring [82].

3. VitotoxTM + HCS IVMN HepG2: A combination of these assays gives for theECVAM compounds a sensitivity of 90% (18/20) and a specificity of 83%(35/42). Also in this case the specificity is considerably higher than that ofregulatory assays, rendering the combination an advantage as screening tool.

4. VitotoxTM + HepG2 p53-Luc: This combination of assays gives for theECVAM compounds a sensitivity of 85% (17/20) and a specificity of 88%(37/42), being again much higher than the specificity of the regulatory assays.For the set of 192 compounds, a sensitivity of 76% (29/38) and a specificityof 72% (23/32) was found. These sensitivity scores appear to be lower thanthose of regulatory tests (97%, 37/38), but the specificity score is 72%, whichis much higher than the 41% (12/29) of regulatory tests Thus, this combinationis also most suitable for early screening, as shown for this large set of 254compounds with an overall score for sensitivity and specificity of 79% (46/58)and 81% (60/74), respectively.

5. VitotoxTM + GADD45a-GFP GreenScreen HC: This combination gives forthe ECVAM compound list a sensitivity of 95% (19/20) and a specificity of85% (35/41). Due to autofluorescence, it was not possible to test curcumin(false-positive group). The scores obtained with the ECVAM compound listmake this test combination very applicable during lead optimization.

Combinations of the VitotoxTM with the cystatin A and RAD51C reporter assaysare not discussed as the transcription of both these genes is p53-dependent [83] andthe combinations hardly influence the performance scores (Table 12.4). Thus, thep53 reporter assay is the most useful HepG2 reporter assay to detect the genotoxicpotential of drugs, and additional HepG2 reporter assays such as the cystatin A-,RAD51C-, and Nrf2-based test systems may help in elucidating the mode of actionof genotoxic compounds.In summary, the combination VitotoxTM + RadarScreen resulted in a specificity

that is generally too low. The other four assay combinations, namely VitotoxTM withthe CHO-k1 HCS IVMN, HepG2 HCS IVMN, HepG2 p53_luc, or GADD45a-GFPGreenScreen HC assay, seem very useful for genotoxicity screening in the leadoptimization phase of drug development as both sensitivity and specificity are high.The combinations VitotoxTM + HepG2 p53_luc and VitotoxTM + GADD45a-GFPGreenScreen HCwere validated with the largest compound sets. Therefore, one of thetwo above-mentioned combinations should preferably be used for an early screeningstrategy. Further validation of the HCS IVMN assays with a larger compound setshould be performed. Depending on the results, an HCS IVMN assay may be added

REFERENCES 263

to the VitotoxTM + p53_luc or VitotoxTM + GADD45a-GFP GreenScreen HCcombinations or may replace these two mammalian reporter assays. The advantageof the application of an HCS IVMN assay is that it will make possible to directlydiscriminate between aneugens and clastogens.The Committee on Mutagenicity of Chemicals in Food, Consumer Products and

the Environment (COM) has recently published a new guidance for the testing ofchemical substances in the United Kingdom [84], COM being an independent expertadvisory committee. In the new guidance they advise to use genotoxicity screeningassays as a first tier in the case of newly developed chemicals. The screening tests thatwere most promising in their view were the GADD45a and HepG2 reporter assays.The COM guidance has no regulatory status. However, it should be pointed out that,as in the case of earlier published COM guidances, after some time their strategy isoften reflected in guidelines used byUK regulatory authorities. This clearly shows thatnew testing strategies as those described in this chapter most likely will play a moreprominent role in future testing strategies of chemicals as well as pharmaceuticals.

12.7 CONCLUSION

In this chapter, the regulatory test battery and several novel test systems with amedium or high throughput to determine the genotoxic potential of chemicals andpharmaceuticals have been described. The validation data for these higher throughputassays show that an early prediction can be made for bacterial mutagenicity (genemutations) and mammalian genotoxicity (chromosome damage).To develop a strategy for the application of the novel genotoxicity assays in the

lead optimization phase, several combinations of assays were evaluated. Based on theresults available up to now, the combinations VitotoxTM + HepG2 p53 reporter assayand VitotoxTM + GADD45a-GFP GreenScreen HC are the most useful ones whenscreening compounds for their genotoxic potential in the lead optimization phasewithout the risk of obtaining high numbers of false-positives. Further application ofthese assays may prove useful in future development strategies of pharmaceuticalsand other chemicals.

REFERENCES

1. Kola, I., Landis, J. (2004). Can the pharmaceutical industry reduce attrition rates? NatureReview Drug Discovery, 3, 711–715.

2. Tamimi, N.A., Ellis, P. (2009). Drug development: from concept to marketing! NephronClinical Practicum, 113, c125–c131.

3. Benbow, J.W., Aubrecht, J., Banker, M.J., Nettleton, D., Aleo, M.D. (2010). Predictingsafety toleration of pharmaceutical chemical leads: cytotoxicity correlations to exploratorytoxicity studies. Toxicological Letters, 197, 175–182.

4. Armitage, P., Doll, R. (1957). A two-stage theory of carcinogenesis in relation to the agedistribution of human cancer. British Journal of Cancer, 11, 161–169.


5. Berenblum, I., Shubik, P. (1949). An experimental study of the initiating state of car-cinogenesis, and a re-examination of the somatic cell mutation theory of cancer. BritishJournal of Cancer, 3, 109–118.

6. Hernandez, L.G., van Steeg, H., Luijten, M., van Benthem, J. (2009). Mechanisms ofnon-genotoxic carcinogens and importance of a weight of evidence approach. MutationResearch, 682, 94–109.

7. Elliott, B.M. (1994). Genotoxicity testing strategies. Toxicology In Vitro, 8, 871–872.

8. Tweats, D.J., Blakey, D., Heflich, R.H., Jacobs, A., Jacobsen, S.D., Morita, T., Nohmi, T.,O’Donovan, M.R., Sasaki, Y.F., Sofuni, T., Tice, R. (2007). Report of the IWGT workinggroup on strategy/interpretation for regulatory in vivo tests II. Identification of in vivo-only positive compounds in the bone marrow micronucleus test.Mutation Research, 627,92–105.

9. ICH—International Conference on Harmonisation of Technical Requirements for Reg-istration of Pharmaceuticals for Human Use. (2011). Guidelines S2A and B: Guidanceon genotoxicity testing and data interpretation for pharmaceuticals intended for humanuse.

10. Ames, B.N., Lee, F.D., Durston, W.E. (1973). An improved bacterial test system for thedetection and classification of mutagens and carcinogens. Proceedings of the NationalAcademy of Sciences U.S.A., 70, 782–786.

11. Ames, B.N., Durston, W.E., Yamasaki, E., Lee, F.D. (1973). Carcinogens are mutagens: asimple test system combining liver homogenates for activation and bacteria for detection.Proceedings of the National Academy of Sciences U.S.A., 70, 2281–2285.

12. Kirkland, D., Aardema, M., Henderson, L., Muller, L. (2005). Evaluation of the abilityof a battery of three in vitro genotoxicity tests to discriminate rodent carcinogens andnon-carcinogens I. Sensitivity, specificity and relative predictivity. Mutation Research,584, 1–256.

13. Galloway, S.M. (2000). Cytotoxicity and chromosome aberrations in vitro: experience inindustry and the case for an upper limit on toxicity in the aberration assay. Environmentaland Molecular Mutagenesis, 35, 191–201.

14. Galloway, S.M., Armstrong,M.J., Reuben, C., Colman, S., Brown, B., Cannon, C., Bloom,A.D., Nakamura, F., Ahmed, M., Duk, S. (1987). Chromosome aberrations and sisterchromatid exchanges in Chinese hamster ovary cells: evaluations of 108 chemicals. Envi-ronmental and Molecular Mutagenesis, 10(Suppl. 10), 1–175.

15. Kirsch-Volders, M., Vanhauwaert, A., de Boeck, M., Decordier, I. (2002). Importance ofdetecting numerical versus structural chromosome aberrations. Mutation Research, 504,137–148.

16. Benigni, R., Bossa, C., Worth, A. (2010). Structural analysis and predictive value of therodent in vivo micronucleus assay results. Mutagenesis, 25, 335–341.

17. Singh,N.P.,McCoy,M.T., Tice, R.R., Schneider, E.L. (1988).A simple technique for quan-tification of low levels of DNA damage in individual cells. Experimental Cell Research,175, 184–191.

18. Tice, R.R., Agurell, E., Anderson, D., Burlinson, B., Hartmann, A., Kobayashi, H., Miya-mae, Y., Rojas, E., Ryu, J.-C., Sasaki, Y.F. (2000). Single cell gel/Comet assay: guidelinesfor in vitro and in vivo genetic toxicology testing. Environmental and Molecular Mutage-nesis, 35, 206–221.

REFERENCES 265

19. Kirkland, D., Speit, G. (2008). Evaluation of the ability of a battery of three in vitrogenotoxicity tests to discriminate rodent carcinogens and non-carcinogens III. Appropriatefollow-up testing in vivo. Mutation Research, 654, 114–132.

20. Glatt, H., Meinl, W. (2005). Sulfotransferases and acetyltransferases in mutagenicitytesting: technical aspects. Methods in Enzymology, 400, 230–249.

21. Glatt, H.R. (2000). An overview of bioactivation of chemical carcinogens. BiochemicalSociety Transactions, 28, 1–6.

22. Tweats, D.J., Blakey, D., Heflich, R.H., Jacobs, A., Jacobsen, S.D., Morita, T., Nohmi,T., O’Donovan, M.R., Sasaki, Y.F., Sofuni, T., Tice, R. (2007). Report of the IWGTworking group on strategies and interpretation of regulatory in vivo tests I. Increasesin micronucleated bone marrow cells in rodents that do not indicate genotoxic hazards.Mutation Research, 627, 78–91.


24. Kirkland, D., Aardema, M., Muller, L., Makoto, H. (2006). Evaluation of the abilityof a battery of three in vitro genotoxicity tests to discriminate rodent carcinogens andnon-carcinogens II. Further analysis of mammalian cell results, relative predictivity andtumour profiles. Mutation Research, 608, 29–42.

25. Snyder, R.D., Green, J.W. (2001). A review of the genotoxicity of marketed pharmaceu-ticals. Mutation Research, 488, 151–169.

26. Kirkland, D.J., Aardema, M., Banduhn, N., Carmichael, P., Fautz, R., Meunier, J.R.,Pfuhler, S. (2007). In vitro approaches to develop weight of evidence (WoE) and modeof action (MoA) discussions with positive in vitro genotoxicity results. Mutagenesis, 22,161–175.

27. Tweats, D.J., Scott, A.D., Westmoreland, C., Carmichael, P.L. (2007). Determinationof genetic toxicity and potential carcinogenicity in vitro - challenges post the SeventhAmendment to the European Cosmetics Directive. Mutagenesis, 22, 5–13.

28. Jacobs, A., Jacobson-Kram, D. (2004). Human carcinogenic risk evaluation, Part III:assessing cancer hazard and risk in human drug development. Toxicological Sciences, 81,260–262.

29. ICH—International Conference on Harmonisation of Technical Requirements for Regis-tration of Pharmaceuticals for Human Use. (1997). Guideline S1B: testing for carcino-genicity of pharmaceuticals for human use. ICH—International conference on harmoniza-tion of technical requirements for registration of pharmaceuticals for human use.

30. Chhabra, R.S., Huff, J.E., Schwetz, B.S., Selkirk, J. (1990). An overview of prechronicand chronic toxicity/carcinogenicity experimental study designs and criteria usedby the National Toxicology Program. Environmental Health Perspectives, 86, 313–321.

31. Yamamoto, S., Hayashi, Y., Mitsumori, K., Nomura, T. (1997). Rapid carcinogenicitytesting systemwith transgenicmice harboring human prototype c-HRAS gene. LaboratoryAnimal Science, 47, 121–126.

32. Caldwell, G.W., Ritchie, D.M., Masucci, J.A., Hageman, W., Yan, Z. (2001). The newpre-preclinical paradigm: compound optimization in early and late phase drug discovery.Current Topics in Medical Chemistry, 1, 353–366.


33. Snyder, R.D., Pearl, G.S., Mandakas, G., Choy, W.N., Goodsaid, F., Rosenblum, I.Y.(2004). Assessment of the sensitivity of the computational programs DEREK, TOPKAT,and MCASE in the prediction of the genotoxicity of pharmaceutical molecules. Environ-mental and Molecular Mutagenesis, 43, 143–158.

34. Flamand, N., Meunier, J., Meunier, P., Agapakis-Causse, C. (2001). Mini mutagenicitytest: a miniaturized version of the Ames test used in a prescreening assay for pointmutagenesis assessment. Toxicology In Vitro, 15, 105–114.

35. Quillardet, P., Hofnung, M. (1993). The SOS chromotest: a review. Mutation Research,297, 235–279.

36. van der Lelie, D., Regniers, L., Borremans, B., Provoost, A., Verschaeve, L. (1997).The VITOTOX test, an SOS bioluminescence Salmonella typhimurium test to measuregenotoxicity kinetics. Mutation Research, 389, 279–290.

37. Verschaeve, L., van Gompel, J., Thilemans, L., Regniers, L., Vanparys, P., van der Lelie,D. (1999). VITOTOX bacterial genotoxicity and toxicity test for the rapid screening ofchemicals. Environmental and Molecular Mutagenesis, 33, 240–248.

38. Westerink, W.M., Stevenson, J.C., Horbach, G.J., Schoonen, W.G. (2009). Evaluationof the VitotoxTM and Radarscreen assay for the assessment of genotoxicity in the earlyresearch phase of drug development. Mutation Research, 676, 113–130.

39. Knight, A.W., Billinton, N., Cahill, P.A., Scott, A., Harvey, J.S., Roberts, K.J., Tweats,D.J., Keenan, P.O., Walmsley, R.M. (2007). An analysis of results from 305 compoundstested with the yeast RAD54-GFP genotoxicity assay (GreenScreen GC)—including rel-ative predictivity of regulatory tests and rodent carcinogenesis and performance withautofluorescent and coloured compounds. Mutagenesis, 22, 409–416.

40. Afanassiev, V., Sefton, M., Anantachaiyong, T., Barker, G., Walmsley, R., Wolfl, S.(2000). Application of yeast cells transformed with GFP expression constructs containingthe RAD54 or RNR2 promoter as a test for the genotoxic potential of chemical substances.Mutation Research, 464, 297–308.

41. Billinton, N., Barker, M.G., Michel, C.E., Knight, A.W., Heyer, W.D., Goddard, N.J.,Fielden, P.R.,Walmsley, R.M. (1998). Development of a green fluorescent protein reporterfor a yeast genotoxicity biosensor. Biosensors & Bioelectronics, 13, 831–838.

42. Knight, A.W., Goddard, N.J., Billinton, N., Cahill, P.A., Walmsley, R.M. (2002). Fluo-rescence polarization discriminates green fluorescent protein from interfering autofluo-rescence in a microplate assay for genotoxicity. Journal of Biochemical and BiophysicalMethods, 51, 165–177.

43. Walmsley, R.M., Billinton, N., Heyer,W.D. (1997). Green fluorescent protein as a reporterfor the DNA damage-induced gene RAD54 in Saccharomyces cerevisiae. Yeast, 13, 1535–1545.

44. Billington, N., Hastwell, P.W., Beerens, D., Birrell, L., Ellis, P., Maskell, S., Webster, T.,Windebank, S., Woestenborghs, F., Lynch, A.M., Scott, A.D., Tweats, D.J., van Gompel,J., Rees, R.W., Walmsley, R.M. (2008). Interlaboratory assessment of the GreenScreenHC GADD45a-GFP genotoxicity screening assay: an enabling study for independentvalidation as an alternative method. Mutation Research, 653, 23–33.

45. Billington, N., Bruce, S., Hansen, J.R., Hastwell, P.W., Jagger, C., McComb, C., Klug,M.L., Pant, K., Rabinowitz, A., Rees, R., Tate, M., Vinggaard, A.M., Walmsley, R.M.(2010). A pre-validation transferability study of the GreenScreen HC GADD45a-GFPassay with a metabolic activation system (S9).Mutation Research, 700, 44–50.

REFERENCES 267

46. Birrell, L., Cahill, P., Hughes, C., Tate, M., Walmsley, R.M. (2010). GADD45a-GFPGreenScreenHCassay for theECVAMrecommended lists of genotoxic and non-genotoxicchemicals for assessment of new genotoxicity tests. Mutation Research, 695, 87–95.

47. Hastwell, P.W., Webster, T.W., Tate, M., Billington, N., Lynch, A.M., Harvey, J.S.,Rees, R.W., Walmsley, R.M. (2009). Analysis of 75 marketed pharmaceuticals usingthe GADD45a-GFP ‘GreenScreen’ HC genotoxicity assay.Mutagenesis, 24, 455–463.

48. Hewitt, N.J., Hewitt, P. (2004). Phase I and II enzyme characterization of two sources ofHepG2 cell lines. Xenobiotica, 34, 243–256.

49. Knasmuller, S., Schwab, C.E., Land, S.J., Wang, C.Y., Sanyal, R., Kundi, M., Parzefall,W., Darroudi, F. (1999). Genotoxic effects of heterocyclic aromatic amines in humanderived hepatoma (HepG2) cells.Mutagenesis, 14, 533–540.

50. Zeiger, M., Haag, R., Hockel, J., Schrenk, D., Schmitz, H.J. (2001). Inducing effects ofdioxin-like polychlorinated biphenyls on CYP1A in the human hepatoblastoma cell lineHepG2, the rat hepatoma cell line H4IIE, and rat primary hepatocytes: comparison ofrelative potencies. Toxicological Sciences, 63, 65–73.

51. Schoonen, W.G., de Roos, J.A., Westerink, W.M., Debiton, E. (2005). Cytotoxic effectsof 110 reference compounds on HepG2 cells and for 60 compounds on HeLa, ECC-1 andCHO cells. II Mechanistic assays on NAD(P)H, ATP and DNA contents. Toxicology InVitro, 19, 491–503.

52. Schoonen, W.G., Westerink, W.M., de Roos, J.A., Debiton, E. (2005). Cytotoxic effectsof 100 reference compounds on Hep G2 and HeLa cells and of 60 compounds on ECC-1and CHO cells. I Mechanistic assays on ROS, glutathione depletion and calcein uptake.Toxicology In Vitro, 19, 505–516.

53. Kirkland, D., Pfuhler, S., Tweats, D., Aardema, M., Corvi, R., Darroudi, F., Elhajouji,A., Glatt, H., Hastwell, P., Hayashi, M., Kasper, P., Kirchner, S., Lynch, A., Marzin, D.,Maurici, D., Meunier, J.R., Muller, L., Nohynek, G., Parry, J., Parry, E., Thybaud, V., Tice,R., van Benthem, J., Vanparys, P., White, P. (2007). How to reduce false positive resultswhen undertaking in vitro genotoxicity testing and thus avoid unnecessary follow-upanimal tests: Report of an ECVAMWorkshop. Mutation Research, 628, 31–55.

54. Knasmuller, S., Cavin, C., Chakraborty, A., Darroudi, F., Majer, B.J., Huber, W.W.,Ehrlich, V.A. (2004). Structurally related mycotoxins ochratoxin A, ochratoxin B, andcitrinin differ in their genotoxic activities and in their mode of action in human-derivedliver (HepG2) cells: implications for risk assessment. Nutrition Cancer, 50, 190–197.

55. Zhou, S.M., Jiang, L.P., Geng, C.Y., Cao, J., Zhong, L.F. (2010). Patulin-induced oxidativeDNA damage and p53 modulation in HepG2 cells. Toxicon, 55, 390–395.

56. Westerink,W.M., Schoonen,W.G. (2007). CytochromeP450 enzyme levels inHepG2 cellsand cryopreserved primary hepatocytes and their induction in HepG2 cells. Toxicology InVitro, 21, 1581–1591.

57. Westerink, W.M., Schoonen, W.G. (2007). Phase II enzyme levels in HepG2 cells andcryopreserved primary hepatocytes and their induction in HepG2 cells. Toxicology InVitro, 21, 1592–1602.

58. Boehme, K., Dietz, Y., Hewitt, P., Mueller, S.O. (2010). Activation of P53 in HepG2cells as surrogate to detect mutagens and promutagens in vitro. Toxicology Letters, 198,272–281.

59. Vollmer, C.M., Ribas, A., Butterfield, L.H., Dissette, V.B., Andrews, K.J., Eilber, F.C.,Montejo, L.D., Chen, A.Y., Hu, B., Glaspy, J.A., McBride, W.H., Economou, J.S. (1999).


p53 selective and nonselective replication of an E1B-deleted adenovirus in hepatocellularcarcinoma. Cancer Research, 59, 4369–4374.

60. Romer, L., Klein, C., Dehner, A., Kessler, H., Buchner, J. (2006). p53 - a natural cancerkiller: structural insights and therapeutic concepts. Angewandte Chemie (InternationalEdition in English), 45, 6440–6460.

61. Hussain, S.P., Harris, C.C. (2006). p53 biological network: at the crossroads of the cellular-stress response pathway and molecular carcinogenesis. Journal of the Nippon MedicalSchool, 73, 54–64.

62. Helton, E.S., Chen, X. (2007). p53 modulation of the DNA damage response. Journal ofCellular Biochemistry, 100, 883–896.

63. Wang,Q.E., Zhu,Q.,Wani,M.A.,Wani,G., Chen, J.,Wani,A.A. (2003). Tumor suppressorp53 dependent recruitment of nucleotide excision repair factors XPC and TFIIH to DNAdamage. DNA Repair (Amsterdam), 2, 483–499.

64. Chang, Y.C., Jan, K.Y., Cheng, C.A., Liao, C.B., Liu, Y.C. (2008). Direct involvementof the tumor suppressor p53 in nucleotide excision repair. DNA Repair (Amsterdam), 7,751–761.

65. Zschenker, O., Borgmann, K., Streichert, T., Meier, I., Wrona, A., Dikomey, E. (2006).Lymphoblastoid cell lines differing in p53 status show clear differences in basal geneexpression with minor changes after irradiation. Radiotherapeutic Oncology, 80, 236–249.

66. Adachi, T., Nakagawa, H., Chung, I., Hagiya, Y., Hoshijima, K., Noguchi, N., Kuo, M.T.,Ishikawa, T. (2007). Nrf2-dependent and -independent induction of ABC transportersABCC1, ABCC2, and ABCG2 in HepG2 cells under oxidative stress. Journal of Experi-mental and Therapeutic Oncology, 6, 335–348.

67. Beddowes, E.J., Faux, S.P., Chipman, J.K. (2003). Chloroform, carbon tetrachloride andglutathione depletion induce secondary genotoxicity in liver cells via oxidative stress.Toxicology, 187, 101–115.

68. Uhl, M., Helma, C., Knasmuller, S. (1999). Single-cell gel electrophoresis assays withhuman-derived hepatoma (Hep G2) cells.Mutation Research, 441, 215–224.

69. Uhl, M., Ecker, S., Kassie, F., Lhoste, E., Chakraborty, A., Mohn, G., Knasmuller, S.(2003). Effect of chrysin, a flavonoid compound, on the mutagenic activity of 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) and benzo(a)pyrene (B(a)P) in bacterialand human hepatoma (HepG2) cells. Archives Toxicology, 77, 477–484.

70. Uhl, M., Laky, B., Lhoste, E., Kassie, F., Kundi, M., Knasmuller, S. (2003). Effectsof mustard sprouts and allylisothiocyanate on benzo(a)pyrene-induced DNA damage inhuman-derived cells: a model study with the single cell gel electrophoresis/HepG2 assay.Teratogenicity, Carcinogenicity, and Mutagenenicity, (Suppl. 1), 273–282.

71. Majer, B.J., Mersch-Sundermann, V., Darroudi, F., Laky, B., de Wit, K., Knasmuller, S.(2004). Genotoxic effects of dietary and lifestyle related carcinogens in human derivedhepatoma (HepG2, Hep3B) cells.Mutation Research, 551, 153–166.

72. Knasmuller, S., Mersch-Sundermann, V., Kevekordes, S., Darroudi, F., Huber, W.W.,Hoelzl, C., Bichler, J., Majer, B.J. (2004). Use of human-derived liver cell lines forthe detection of environmental and dietary genotoxicants; current state of knowledge.Toxicology, 198, 315–328.

73. Valentin-Severin, I., Le Hegarat, L., Lhuguenot, J.C., Le Bon, A.M., Chagnon, M.C.(2003). Use of HepG2 cell line for direct or indirect mutagens screening: comparativeinvestigation between comet and micronucleus assays.Mutation Research, 536, 79–90.

REFERENCES 269

74. Diaz,D., Scott, A., Carmichael, P., Shi,W., Costales, C. (2007). Evaluation of an automatedin vitro micronucleus assay in CHO-K1 cells. Mutation Research, 630, 1–13.

75. Westerink, W.M., Schirris, T.J., Horbach, G.J., Schoonen, W.G. (2011). Development andvalidation of a high content in vitro micronucleus assay in CHO-k1 and HepG2 cells.Mutation Research, 724, 7–21.

76. Hastwell, P.W., Chai, L.L., Roberts, K.J., Webster, T.W., Harvey, J.S., Rees, R.W., Walms-ley,R.M. (2006). High-specificity and high-sensitivity genotoxicity assessment in a humancell line: validation of the GreenScreen HCGADD45a-GFP genotoxicity assay.MutationResearch, 607, 160–175.

77. Rushmore, T.H., Morton, M.R., Pickett, C.B. (1991). The antioxidant responsive element.Activation by oxidative stress and identification of the DNA consensus sequence requiredfor functional activity. Journal of Biological Chemistry, 266, 11632–11639.

78. Venugopal, R., Jaiswal, A.K. (1996). Nrf1 and Nrf2 positively and c-Fos and Fra1negatively regulate the human antioxidant response element-mediated expression ofNAD(P)H:quinone oxidoreductase1 gene. Proceedings of the National Academy of Sci-ences U.S.A., 93, 14960–14965.

79. Itoh, K., Chiba, T., Takahashi, S., Ishii, T., Igarashi, K., Katoh, Y., Oyake, T., Hayashi,N., Satoh, K., Hatayama, I., Yamamoto, M., Nabeshima, Y. (1997). An Nrf2/small Mafheterodimer mediates the induction of phase II detoxifying enzyme genes through antiox-idant response elements. Biochemical and Biophysical Research Communications, 236,313–322.

80. Chen, W., Sun, Z., Wang, X.J., Jiang, T., Huang, Z., Fang, D., Zhang, D.D. (2009). Directinteraction betweenNrf2 and p21(Cip1/WAF1) upregulates the Nrf2-mediated antioxidantresponse. Molecular Cell, 34, 663–673.

81. Kirkland, D., Kasper, P., Muller, L., Corvi, R., Speit, G. (2008). Recommended lists ofgenotoxic and non-genotoxic chemicals for assessment of the performance of new orimproved genotoxicity tests: a follow-up to an ECVAM workshop. Mutation Research,653, 99–108.

82. Patino-Garcia, B., Hoegel, J., Varga, D., Hoehne, M., Michel, I., Jainta, S., Kreienberg,R., Maier, C., Vogel, W. (2006). Scoring variability of micronuclei in binucleated humanlymphocytes in a case-control study. Mutagenesis, 21, 191–197.

83. Westerink, W.M., Stevenson, J.C., Horbach, G.J., Schoonen, W.G. (2010). The devel-opment of RAD51C, Cystatin A, p53 and Nrf2 luciferase-reporter assays in metaboli-cally competent HepG2 cells for the assessment of mechanism-based genotoxicity and ofoxidative stress in the early research phase of drug development.Mutation Research, 696,21–40.

84. Committee on Mutagenicity of Chemicals in Food, Consumer Products and theEnvironment (COM) (2000). COM Guidance ‘A Strategy for Testing of Chem-icals for Genotoxicity’. Available at http://www.iacom.org.uk/guidstate/documents/COMGuidanceFINAL.pdf.

13HIGH-THROUGHPUT GENOTOXICITYTESTING: THE GREENSCREEN ASSAY

Jorg Blumel and Nadine Krause

The testing of chemical entities for their potential to induce genetic toxicity plays animportant and critical role for the hazard identification and management of chem-icals and for the risk–benefit assessment during preclinical development of phar-maceuticals. The current testing strategy accepted by regulatory agencies relies ona combination of in vitro and short-term in vivo assays to assess the potential toinduce genotoxicity. The International Conference on Harmonization of TechnicalRequirements for Registration of Pharmaceuticals for Human Use (ICH) outlines inthe revised guidance on genotoxicity testing issued in 2011 [1] a standard test battery.The recommended assay panel consists of an in vitro test for gene mutation in bacte-ria, typically the Salmonella typhimurium reverse mutation (so-called Ames) assay,in combination with an in vitromammalian cell assay (chromosome aberration assay,micronucleus test or mouse lymphoma Tk gene mutation assay), and/or an in vivotest for genotoxicity in rodent cells. A very similar testing paradigm is also appliedduring the registration process of industrial-grade chemicals. Although these assayshave widely been used for decades and are generally accepted by regulatory agenciesworldwide, there are still several limitations that illustrate the need for alternativetesting methods and strategies. A frequently referenced thorough retrospective anal-ysis of a large database of more than 700 chemicals revealed that the sensitivity of theregulatory recommended assays is high, but the specificity of the mammalian assaysand even more in combination with the Ames assay is extremely low [2]. Althoughthere is a high likelihood that a true genotoxic substance will be detected using thestandard test battery, there is in turn also a high risk that a nongenotoxic compoundis wrongly assigned a genotoxic liability (false positive). One can argue that this


271

272 HIGH-THROUGHPUT GENOTOXICITY TESTING: THE GREENSCREEN ASSAY

susceptibility for “false positives” is highly desired to ensure human safety, becauseit greatly reduces the probability of any unintended noxious exposure of humans toan undetected genotoxic xenobiotic. That reasoning may be true for a risk assessmentwith drug or chemical registration purposes, where the standard test battery solelyprovides the decisive data set. But on the other hand, the high rate of “false-positive”results constitutes a major problem for the preclinical selection paradigms of poten-tial pharmaceutical drug candidate molecules. Here the high susceptibility of thestandard test battery toward “false-positive” results may lead to unnecessary earlytermination of promising drug candidates. What is noteworthy is that any screen-ing assay is confirmed later during regulatory toxicity testing by the standard testbattery, which applies another level of safety regarding unintended human exposureto a genotoxic substance. In general, toxicity assays used for early screening pur-poses are not intended to support a risk–benefit assessment of a single molecule, butshould provide reasonable selection or ranking criteria for larger libraries of chemicalstructures. Considering the limitations of the standard test battery as outlined above,there is a need for alternative assays to assess genotoxicity not only with reasonablesensitivity but also acceptable specificity. In addition, the experimental protocol ofa genotoxicity screening assay used during the early screening cascade of potentialdrug candidates requires specific methodological characteristics. Early screening sit-uations are usually characterized by a large number of substances to be tested inparallel in a short time with only a limited amount of each compound available (inmost cases only a few milligrams) for conducting the assay. Thus, the assay protocolneeds to be designed to allow a high or at least medium throughput and the assayreadout must be sensitive enough for reliable and reproducible signal detection withlow substance requirement. In contrast to that requirement, the assays outlined in thestandard test battery for genotoxicity testing require several grams of substance andinvolve a time-consuming low-throughput experimental procedure not applicable foran early screening cascade. Altogether, the data as well as procedural limitationsof the standard test battery illustrate the urgent need for a reliable high-throughputassay for the prediction of genotoxicity in mammalian or even better human cells.The in vitro yeast- or human-cell–based GreenScreen assay discussed in this chapterprovides encouraging data to support its use both as an early screening assay forgenotoxicity as well as a potential assay providing further weight of evidence inregulatory safety assessment of compounds during development [3].The GADD45�–GFP GreenScreen HC assay (GreenScreen HC assay) is an

in vitro human cell (HC)–based genotoxicity screening assay using a geneticallymodified p53-proficient human lymphoblastoid TK6 cell line to host a green flu-orescence protein (GFP) reporter system that exploits the proper regulation of thegrowth arrest and DNA damage gene GADD45� [4]. GADD45� plays an impor-tant role in mediating the adaptive response to genotoxic stress by transcriptionalinduction [5–9]. The basic principle of the GreenScreen HC assay is that GFP flu-orescence reflects the level of activity of the GADD45� reporter construct and thedegree of corrected induction of GFP expression reflects the genotoxicity result.Cytotoxicity is measured in parallel to genotoxicity with and without metabolicactivation (S9).

HIGH-THROUGHPUT GENOTOXICITY TESTING: THE GREENSCREEN ASSAY 273

Xho IOri P

Exon4

Intron 3

p53 motifExon3

Asc I

Not I

Bgl II

EGFP

Hygromycin R GADD45αpromoter

GADD45αgene sequence

EBNA-1

pEP-GD53214.8 kb

FIGURE 13.1 Plasmid map showing the GADD45�–GFP reporter system expressed inthe test cell strain. The plasmid backbone contains the Epstein–Barr virus replication origin(Ori P) and the nuclear antigen gene (EBNA-1) as well as a hygromycin B resistance gene(hygromycin R) jointly allowing stable episomal replication of the reporter and positive strainselection. The plasmid contains the GFP sequence (EGFP) with the corresponding upstreampromoter and regulatory gene sequences with exons 3 and 4 and intron 3 comprising a p53binding motif (reproduced from Hastwell et al. [4]).

The GreenScreen HC assay uses two genetically modified TK6 cell lines, a testcell strain expressing the GADD45�–GFP reporter gene (the reporter plasmid isshown in Fig. 13.1), and a control cell strain in which the reporter has a nonexpressedGFP gene [4]. The control cell strain is used to identify compounds that are eitherinherently fluorescent or induce cellular autofluorescence and may otherwise givea false indication of GFP induction in the test cell strain, and allows for effectivecorrection by subtracting such fluorescence from the test cell strain. The GreenScreenHC assay is performed in a 96-well microplate format, which is sufficient for thesimultaneous testing of four compounds regularly up to a concentration of 1 mg/mL(unless further limited by solubility or cytotoxicity) in two series (control and test cellstrains) of nine 2-fold serial dilutions (see Fig. 13.2 for a representative microplatelayout) and requires less than 2 mg of each compound to be tested. Both low and highconcentrations of a standard genotoxicant are included as an intraplate positive controlto provide data acceptance and other controls including vehicle-treated cells as anegative control, assay medium alone to check for contamination, and test compoundwith assay medium to detect any inherent fluorescence or optical absorbance due


1 B L B

2 B L B

3 B H M

4 B H M

1 B L 1

2 B L 2

3 B H 3

4 B H 4

1 2 3 4 5 6 7 8 9 10 11 12

A

B

C

D

E

F

G

H

Control cell strain

Test cell strain

No cell strain

1

L

H

4- Compound number

Serial Dilutions

M

B

Media control

Blank (Diluent)

Low conc. MMS control

High conc. MMS control

FIGURE 13.2 Scheme of the microplate layout of the GreenScreen HC assay (withoutmetabolic activation [-S9]). The layout represents the testing of four compounds each over nineserial dilutions including positive (high and low concentrations of the standard genotoxicantmethyl methanesulfonate [MMS]) and negative controls and both test and control cell strains(modified from refs. 13 and 14).

to color or precipitation [4]. Sample data are usually obtained after 24 and 48 hof incubation (3-h exposure and 45-h recovery in the presence of S9) by usingflow cytometry for S9-treated samples and by either using fluorescence/absorbancespectrometry or flow cytometry for samples not treated with S9, transferred into adata-processing template and converted to a graphical format [4, 10] (see Fig. 13.3for a typical graphical output). The normalized measurement of the induction ofcellular fluorescence is indicative of genotoxicity, while cytotoxicity is assessed bythe normalized reduction of relative suspension growth (i.e., it is not a measure ofcell viability or death). A detailed explanation of normalization/correction calculationand software-based decision thresholds for data interpretation has been published byHastwell et al. [4] and Jagger et al. [10]. Briefly, a positive result for genotoxicityis recorded for data not treated with S9 if induction of GFP fluorescence relativeto a vehicle-treated control is equal to or greater than the genotoxicity threshold of1.5-fold (50% increase in brightness), which is greater than three times the standarddeviation of the mean GFP signal from a larger number of assays with nongenotoxictoxins and nontoxins [4, 11]. A compound is considered to be cytotoxic if the relative


(a)

(b)

0.0

0.5

1.0

1.5

2.0

2.5

4.212.111.050.53Blank 134.7567.3833.6916.848.42

4.212.111.050.53Blank 134.7567.3833.6916.848.42

4-Nitroquinoline 1-oxide [μg/mL]

Fluo

resc

ence

indu

ctio

n

Test cell strain

Genotoxicity threshold

0

20

40

60

80

100

120

4-Nitroquinoline 1-oxide [μg/mL]

Rel

ativ

e ce

ll de

nsity

(%)

Test cell strain

Cytotoxicity threshold

FIGURE 13.3 Example of a graphical output of theGreenScreenHCassaywithoutmetabolicactivation (-S9) showing simultaneous generation of (a) genotoxicity evaluation and (b) indi-cation of cytotoxicity (corrected and normalized signals).

suspension growth is reduced by 20% [4]. Possible influences of decision thresholdsfor defining a compound as genotoxic and cytotoxic in the GreenScreen HC assayare currently being discussed [12].The technical development of the high-throughput genotoxicity testing Green-

Screen assay is based on a reporter-gene construct first published in 1997 [15].The authors used a genetically engineered yeast strain (Saccharomyces cerevisiae)


containing a high copy number of the RAD54 promoter linked to GFP as reportersystem. RAD54 is a yeast gene known to be a sensitive marker for DNA damage andrepair [16] and GFP is widely used to detect gene expression by its characteristicfluorescence. Cells containing the reporter construct responded to treatment withmethyl methanesulfonate (MMS, a well-known DNA alkylating agent) with a dose-dependent increase in GFP-derived fluorescence. These initial experiments, in whichDNA-damage-related gene expression induced by noxious exogenous stimuli wasreliably detectable with the RAD54—GFP reporter construct, already pointed outthe usability of this assay platform as an alternative tool for genotoxicity screening.The main limitation of this assay was the time-consuming preparation of the cellextract prior to analysis of the fluorescence signal. To overcome that problem, somemodifications of the initial experimental setup were introduced, that is, the use ofyeast-enhanced GFP and a yeast host strain with increased RAD45 gene expressionfollowing MMS-induced DNA damage as well as the reduction of the light scatteringeffect. These modifications improved the signal-to-noise ratio and allowed a sim-plified assay procedure using whole cell preparations instead of cell extracts [17].Further work concentrated on the optimization of the assay parameters to increasethe sensitivity and specificity, especially regarding the discrimination of the DNA-damage-induced GFP signal from interfering autofluorescence derived from media,other cell components, or assaymaterials [18–20]. Thefirst results using the optimizedRAD54–GFP reporter construct to detect the genotoxicity of a small set of knownDNA-damaging substanceswith variousmodes of actionwere reported byAfanassievet al. [21]. The authors demonstrated that the reduced reaction volumes required for96-well microplates are sufficient for a reliable quantification of the DNA-damage-induced reporter-gene fluorescence and concluded that the assay is suitable for fullautomation. It was furthermore emphasized that the standardization of the initial cellconcentration is of critical importance to obtain a reproducible sensitivity. The exper-imental data of 20 substances reported by Afanassiev and colleagues [21] suggestedthat differences in substancemetabolism, yeast-specific detoxifyingmechanisms, andpoor substance penetration into the yeast cells must be taken into consideration andmay account for the observed lack of sensitivity of the yeast assay toward some of thetested genotoxic substances. Lichtenberg-Frate et al. [22] showed that a decreasedsensitivity due to a detoxifying efflux bymultidrug resistance transporter proteins canin principle be circumvented by using a genetically engineered yeast strain lackingthe most important efflux transporters. The applicability of the yeast RAD54–GFP-based GreenScreen assay for the preregulatory screening of potential genotoxicitywas assessed in a larger validation study including 102 chemical substances [23]. Thestudy results demonstrated that the GreenScreen yeast assay system detects a broadpattern of genotoxic substances with different modes of action including direct DNAinteracting as well as indirect acting chemicals (i.e., compounds probably leadingto DNA damage by secondary mechanisms). Overall, the authors reported a goodcorrelation between the outcome of this validation study and the published literatureevidence. Some compounds requiring metabolic activation to exert their genotoxicpotential were identified by the yeast assay, providing evidence that the inherent yeastmetabolic capacity is sufficient to metabolically activate pro-mutagens. Nevertheless,


in contrast to the in vitro assay using the standard exogenous S9 metabolic activationsystem, only three out of nine primary aromatic amines/aromatic amides were scoredpositive. This points out the need to increase the sensitivity of the GreenScreen assayeither by adding an exogenous metabolic activation system or by using yeast cellsgenetically engineered to express human drug metabolizing enzymes [24]. Only twocompounds, tritolyl phosphate and cimetidine, were scored uniquely positive (actu-ally “false positive”) in the GreenScreen assay. In line with the earlier, smaller studyof Afanassiev et al. [21], a potential false-negative scoring due to limited physicalpenetration through the high mannoprotein-containing yeast cell wall was identifiedby Cahill and coworkers [23] as a potential issue, although the GreenScreen assayutilizes haploid cells not forming a rigid protective dityrosine layer. But in contrastto regulatory toxicity testing assays, which are usually designed to be overpredictive,a false-negative result in a screening assay is of less concern, since these data arenot used for a final risk–benefit assessment regarding human or animal exposure.In summary, although not all of the 102 tested substances were correctly identified,this validation study provided good evidence that the GreenScreen assay can be usedas a fast and cheap screening assay for genotoxicity. One major advantage of theGreenScreen assay is the low amount of compound required, an issue of criticalimportance in early phases of drug development, in which assay selection is largelydriven by compound availability. Based on the results of these two initial validationstudies, the experimental setup was further modified to overcome the obstacles oflimited assay sensitivity due to poor cell permeability, inherent drug resistance, andrestricted metabolic competence [25]. This included further genetic engineering ofthe yeast host strain, thereby deleting genes coding for cell wall components as wellas for drug resistance inducing pleiotropic efflux transporters and introducing genesof human phase I xenobiotic metabolizing enzymes. Although the results reportedare not straightforward and varied between different mutant strains, the authors con-cluded that this approach may help to increase the assay sensitivity for specificclasses of test substances, that is, large bulky molecules or pro-mutagens requiringcomplex and intense metabolic activation [25]. Another publication summarizing thedata obtained from a set of 305 compounds confirmed the previous good correlationbetween experimental GreenScreen assay data and published literature evidence withan overall specificity of 82% and a low rate of “false-positive” results [26]. This highspecificity is in contrast to the reported low specificity of the ICH-recommendedregulatory mammalian cell genotoxicity test battery, which is characterized by a highsensitivity but a very low specificity [2], that is, true genotoxins are identified with ahigh likelihood but a lot of the positively tested compounds are not genotoxic (“falsepositives”). In a retrospective data analysis Knight et al. [26] found the specificityof the yeast-based GreenScreen assay to be almost identical to the specificity of theAmes reverse mutation assay (82.3% vs. 84.4%). The limited sensitivity of the Green-Screen assay (38.8%) could be significantly increased up to 66.4% when combiningit with the Ames assay. This combination still showed a high specificity (70.1%).Moreover, Van Gompel et al. [27] demonstrated that both the GreenScreen assayand the Ames assay detect a different subset of genotoxic substances. The authorsconcluded that a combination of the high-throughput version of the Ames assay, the


so-called miniaturized Ames II, and the GreenScreen assay provides an effective andsensitive genotoxicity screening strategy, which is of high value for the selection ofdrug candidates.In contrast to the Ames assay using prokaryotic bacteria cells, the yeast-based

GreenScreen assay utilizes cells of eukaryotic origin. Yeast cells share many cellularand molecular features with mammalian cells, especially various features regardingmetabolism and cell architecture. The DNA of yeast cells is packed in a highly orga-nized form in the nucleus and has, as in the case of human DNA, a high proportion ofnoncoding DNA sequences. In addition, yeast and human cells share common DNArepair mechanisms activated following DNA damage. Although the metabolic com-petence of yeast cells is less pronounced and the pattern of genes encoding enzymesinvolved in metabolism is less complex when compared to mammalian cells, theyare in principle susceptible to pro-mutagens that require metabolic activation [23].Taken together, the development and validation of the yeast-based GreenScreen assayclearly ameliorated the predictive value of high-throughput genotoxicity testing. Buteven though the use of eukaryotic cells instead of prokaryotic cells led to a hugetranslational advantage over conventional screening strategies, the ultimate goal is todevelop a genotoxicity screening assay based on mammalian or human cell lines. Thesuccessful introduction of the yeast-based GreenScreen assay as a high-throughputassay and the encouraging published validation of large data sets containing chem-icals as well as proprietary pharmaceutical compounds triggered further research totranslate the underlying molecular and biochemical readout principles into a human-cell-based assay system.Similar to RAD54 in yeast cells, the GADD45� gene was identified as a highly

conserved gene that is upregulated following DNA damage [28] in a wide varietyof mammalian cell types including various rodent cell types, but more importantlyalso in non-neoplastic [29, 30] as well as neoplastic human cell lines [31]. Althoughthe GADD45� biology is rather complex and the encoded protein is involved in anumber of cellular processes, GADD45� upregulation was found to be a suitableand specific response biomarker for genomic stress and DNA damage. Literature evi-dence demonstrated GADD45� gene expression in response to a variety of genotoxicsubstances like cisplatin [8], antimicrotubule agents [32], or benzo[a]pyrene [5]. Inaddition, a study initiated by the Health and Environmental Science Institute (HESI)to analyze the usability of microarray gene expression profiling to guide risk assess-ment of potential genotoxic substances revealed that GADD45� gene expression wasconsistently upregulated following high-dose exposure to a broad pattern of genotoxiccompounds [33]. The modified genotoxicity screening assay using GADD45�–GFPas the reporter-gene system in a human cell line (GreenScreen HC assay) to assessDNA damage was first published in 2006 [4]. The authors used a panel of 75 knowngenotoxic and nongenotoxic substances and reported an accurate prediction of geno-toxicity as compared to published results. Although the concentrations required werehigher compared to the standard test battery, they were still in the range recommendedby regulatory guidances. Importantly, it was shown that the GreenScreen HC assaydid not respond to nongenotoxic cytotoxic stress, although it had previously beenshown that nongenotoxic cytotoxic insults like hypoxia [34] or hyperosmolality [35]


induce GADD45� expression. In addition, the assay correctly segregated in a sub-group of 18 compounds with published negative Ames assay and positive mammaliancell assay results those being negative in a subsequent in vivo carcinogenicity assess-ment (“false positives”) from those with a positive outcome in in vivo carcinogenicitystudies (“true positives”). In this first validation study [4], the GreenScreen HC assayshowed a good sensitivity, which was comparable to that of the standard mammaliancell assays. In contrast to the reported low specificity of the standard mammalianassays [2], the GreenScreen HC assay showed a high specificity, being comparableto that of the Ames assay. The encouraging data of this first study [4] were furtherexpanded by two additional studies using 75 marketed pharmaceuticals [36] and apanel of 61 substances recommended by the European Centre for the Validation ofAlternativeMethods (ECVAM) [11]. Both studies basically confirmed the high speci-ficity and sensitivity of the GreenScreen HC assay already seen in the initial studyby Hastwell et al. [4] using a homogenous set of substances, even when using a morediverse panel of test compounds. The ECVAM panel of test substances included 20substances with a positive readout in an in vitro mammalian cell genotoxicity assay(group I), 23 chemicals with a negative readout in an in vitro mammalian cell geno-toxicity assay (group II), and 17 compounds reported to induce only chromosomalaberrations in the mouse lymphoma assay (group III). The overall concordance of theGreenScreen HC assay was found to be 93% for group I and II and 76% for groupIII [11]. In the study reported by Hastwell and colleagues [36], the specificity of theGreenScreen HC assay was 90–96% as compared to 86, 79 and 54% for the Amestest, the mammalian mutation assay and the in vitro cytogenesis assay, respectively.In addition, no test substance was uniquely positive in the GreenScreen HC assay.Overall, the authors concluded that the specificity of the GreenScreen HC assay ismaintained even when performed in the presence of a metabolic activation systemand that the assay is more predictive of in vivo genotoxicity and carcinogenicity thanany other assay in the ICH-recommended standard test battery.A blinded interlaboratory validation study following the modular assessment cri-

teria as proposed by ECVAMwith 16 genotoxic and nongenotoxic compounds and 4participating test sites revealed a correct prediction in 92% of the conducted assays[37]. These data together with results published by Olaharski et al. [38] demon-strated the robustness and reproducibility of the assay protocol when transferred todifferent laboratories. Olaharski and coworkers evaluated the performance of theGreenScreen HC assay protocol with 91 proprietary and nonproprietary pharmaceu-tical compounds and reported a sensitivity and specificity regarding genotoxicity of30% and 97%, respectively [38]. Furthermore, a protocol for the GreenScreen HCassay was developed and successfully validated, thereby allowing the assessment ofgenotoxic agents that require metabolic activation after incubation with the standardrodent S9 postmitochondrial fraction [10]. Following intralaboratory validation, thisprotocol underwent an interlaboratory validation in consideration of the ECVAMrecommendations including three transfer test laboratories and eight test substances[39]. After further refinement of the assay protocol during the validation procedure,a concordance of 100% between the results of the different test sites was reached.To comply with the specific requirements of high-throughput systems used in early


screening of large pharmaceutical compound libraries, the assay protocol was furtheradapted to allow automated liquid handling in a 96-well microplate format [13].The assay layout described allows the testing of 12 compounds in parallel with 3serial dilutions starting from a fixed top concentration of 100 �M. To assess theassay performance, a commercially available compound library of 1266 pharmaco-logical active compounds from Sigma-Aldrich was used. Although the sensitivity ofthe GreenScreen HC assay decreased when compared to earlier validation studies,the high-throughput protocol maintained the high specificity reported previously [4,11, 36] and showed a better correlation with published data than in silico softwareassessment. The authors have already pointed out the possibility to further increasethe throughput by switching to larger microwell plates. Another study using 320chemical substances of a nonpharmaceutical chemical space derived from the USEnvironmental Protection Agency ToxCastTM project confirmed the usability of theGreenScreen HC assay as a high-throughput assay [40]. As a consequence of theincreasing recognition and use of the GreenScreen HC assay as a screening tool forgenotoxicity, the GADD45�–GFP assay was discussed in a workshop sponsored byHESI on emerging techniques and new strategies in in vitro genetic toxicity [3]. Theworking group considered the GreenScreen HC assay as a maturing assay that can gobeyond the traditional test battery. In detail, the published solid validation data wereacknowledged, and it was emphasized that the capability of the assay to detect a widespectrum of genotoxic mechanisms in an easy assay format that allows medium tohigh throughout with low compound requirement provides an attractive alternative.The working group concluded that the assay may provide additional weight of evi-dence for regulatory safety assessment, especially as a potential follow-up assay forunique in vitro positive substances.In summary, a broad evidence supporting the use of the GreenScreen HC assay as

a screening tool for genotoxicity was published since the initial development of theyeast-based RAD54–GFPGreenScreen assay in 1997 [15] and themodification of theassay principle using a human lymphoblastoid TK6 cell line and the GADD45�–GFPreporter construct (GreenScreen HC) in 2006 [4]. As described in this chapter, thevalidation studies using both pharmaceutical compounds and chemical substanceswith a variety of genotoxic mechanisms of action showed that the GreenScreenassay provides both a good sensitivity and, in contrast to the ICH-compliant stan-dard test battery, a high specificity for genotoxicity and produces robust as well asreproducible results. Published protocol modifications allow the incorporation of anexternal S9 metabolic activation system for the detection of pro-mutagens as wellas high-throughput techniques like robotic liquid handling and multiwell microtiterplates. Beside the successful technical development of a high-throughputmethodwithlow substance consumption, the most striking difference of the GreenScreen assay ascompared to the ICH-recommended test systems is the use of RAD54 or GADD45�coupled to a reporter gene as upstream readout. These molecular readouts providea closer relationship to the actual noxious DNA-damaging insult than the usuallyused downstream effects like gene mutation, chromosomal aberration, or cancer. Thedownstream effects may be reversed by inherent cellular DNA repair mechanismsand therefore prediction based on these parameters may bemisleading. Consequently,

REFERENCES 281

the use of a readout directly involved in maintaining genome integrity and repair canbe considered as a very predictive marker for the potential of a substance to induceDNA damage.However, when applying the GreenScreen assay system one should bear in mind

that the assay setup aims to screen especially larger compound libraries for potentialgenotoxic compounds. Positive data from high-throughput genotoxicity assays aloneare not a suitable replacement for standard regulatory genotoxicity assays. The resultsof the GreenScreen assay can be used as a fast and easy screening tool for genotoxichazard identification. The data can guide the selection and ranking of chemicals orpharmaceutical development candidates out of a larger library by clearly identify-ing compounds with a potential to induce carcinogenicity and thereby decreasingthe attrition rate due to genotoxicity in later stages of development. Although theGreenScreen assay as of today does not qualify to replace the ICH-recommendedstandard test battery, it is a valuable tool to provide further evidence for integratedrisk assessment and can assist to clarify the mechanism of carcinogenicity [40].Both the chemical and the pharmaceutical industry have an urgent demand forhigh-throughput in vitro screening assays to overcome the throughput limitationsof established assays like in vivo studies, to minimize the extensive use of animals,and to avoid unnecessary animal experiments. The GreenScreen assay is beyonddoubt a promising alternative to the standard test battery as outlined in ICH guidelineS2(R1) [1] and a broader application of the GreenScreen assay in industry and thescientific community may further expand its relevance for regulatory genotoxic riskassessment.

REFERENCES

1. International Conference on Harmonization of technical requirements for registration ofpharmaceuticals for human use (2011). ICH harmonized tripartite guideline: guidance ongenotoxicity testing and data interpretation for pharmaceuticals intended for human useS2(R1). Step 4 dated November 09, 2011.

2. Kirkland, D., Aardema, M., Henderson, M., Muller, L. (2005). Evaluation of the abilityof a battery of three in vitro genotoxicity tests to discriminate rodent carcinogens andnon-carcinogens: I. Sensitivity, specificity and relative predictivity. Mutation Research,584, 1–256.

3. Lynch, A.M., Sasaki, J.C., Elespuru, R., Jacobson-Kram, D., Thybaud, V., DeBoeck, M.,Aardema, M.J., Aubrecht, J., Benz, R.D., Dertinger, S.D., Douglas, G.R., White, P.A.,Escobar, P.A., Fornace Jr., A., Honma, M., Naven, R.T., Rusling, J.F., Schiestl, R.H.,Walmsley, R.M., Yamamura, E., van Benthem, J., Kim, J.H. (2011). New and emergingtechnologies for genetic toxicity testing. Environmental and Molecular Mutagenesis, 52,205–223.

4. Hastwell, P.W., Chai, L.L., Roberts, K.J., Webster, T.W., Harvey, J.S., Rees, R.W., Walms-ley, R.M. (2006). High-specificity and high-sensitivity genotoxicity assessment in a humancell line: validation of the GreenScreen HCGADD45a–GFP genotoxicity assay.MutationResearch, 607, 160–175.


5. Akerman, G.S., Rosenzweig, B.A., Domon, O.E., McGarrity, L.J., Blankenship, L.R.,Tsai, C.A., Culp, S.J., MacGregor, J.T., Sistare, F.D., Chen, J.J., Morris, S.M. (2004).Gene expression profiles and genetic damage in benzo(a)pyrene diol epoxide-exposedTK6 cells.Mutation Research, 549, 43–64.

6. Arriola, E.L., Lopez, A.R., Chresta, C.M. (1999). Differential regulation of p21waf-1/cip-1

and Mdm2 by etoposide: etoposide inhibits the p53-Mdm2 autoregulatory feedback loop.Oncogene, 18, 1081–1091.

7. Abbas, T., Olivier, M., Lopez, J., Houser, S., Xiao, G., Kumar, G.S., Tomasz, M., Bar-gonetti, J. (2002). Differential activation of p53 by the various adducts of mitomycin C.Journal of Biological Chemistry, 277, 40513–40519.

8. Smith, M.L., Kontny, H.U., Zhan, Q., Sreenath, A., O’Connor, P.M., Fornace, A.J. Jr.(1996). Antisense GADD45 expression results in decreased DNA repair and sensitizescells to U.V.-irradiation or cisplatin. Oncogene, 13, 2255–2263.

9. Newton, R.K., Aardema, M., Aubrecht, J. (2004). The utility of DNA microarrays forcharacterizing genotoxicity. Environmental Health Perspectives, 112, 420–422.

10. Jagger, C., Tate, M., Cahill, P.A., Hughes, C., Knight, A.W., Billinton, N.,Walmsley, R.M.(2009).Assessment of the genotoxicity of S9-generatedmetabolites using theGreenScreenHC GADD45a–GFP assay.Mutagenesis, 24, 35–50.

11. Birrell, L., Cahill, P., Hughes, C., Tate, M., Walmsley, R.M. (2010). GADD45a–GFPGreenScreen HC assay results for the ECVAM recommended lists of genotoxic and non-genotoxic chemicals for assessment of new genotoxicity tests. Mutation Research, 695,87–95.

12. Olaharski, A.J., Albertini, S., Mueller, L., Zeller, A., Struwe, M., Gocke, E., Kolaja,K. (2011). GADD45� induction in the GreenScreen HC indicator assay does notoccur independently of cytotoxicity. Environmental and Molecular Mutagenesis, 52,28–34.

13. Knight, A.W., Birrell, L., Walmsley, R.M. (2009). Development and validation of a higherthroughput screening approach to genotoxicity testing using the GADD45a–GFP Green-Screen HC Assay. Journal of Biomolecular Screening, 14, 16–30.

14. Walmsley, R.M., Tate, M. (2012). The GADD45a–GFP GreenScreen HC assay.Methodsin Molecular Biology, 817, 231–250.

15. Walmsley, R.M., Billinton,N., Heyer,W.-D. (1997).Green fluorescent protein as a reporterfor the DNA damage-induced gene RAD54 in Saccharomyces cerevisiae. Yeast, 13, 1535–1545.

16. Cole, G.M., Schild, D., Lovett, S.T., Mortimer, R.K. (1987). Regulation of RAD54- andRAD52-lacZ gene fusions in Saccharomyces cerevisiae in response to DNA damage.Molecular and Cellular Biology, 7, 1078–1084.

17. Billinton, N., Barker, M.G., Michel, C.E., Knight, A.W., Heyer, W.-D., Goddard, N.J.,Fielden, P.R.,Walmsley, R.M. (1998). Development of a green fluorescent protein reporterfor a yeast genotoxicity biosensor. Biosensors & Bioelectronics, 13, 831–838.

18. Knight, A.W., Goddard, N.J., Fielden, P.R., Barker, M.G., Billinton, N., Walmsley, R.M.(1999). Fluorescence polarisation of green fluorescent protein (GFP). A strategy forimproved wavelength discrimination for GFP determinations. Analytical Communica-tions, 36, 113–117.

19. Knight, A.W., Goddard, N.J., Fielden, P.R., Gregson, A.L., Billinton, N., Barker, M.G.,Walmsley, R.M. (2000). The application of fluorescence polarisation for the enhanced

REFERENCES 283

detection of green fluorescent protein (GFP) in the presence of cellular auto-fluorescenceand other green fluorescent compounds. Analyst, 125, 499–506.

20. Knight, A.W., Goddard, N.J., Billinton, N., Cahill, P.A., Walmsley, R.M. (2002). Fluo-rescence polarization discriminates green fluorescent protein from interfering autofluo-rescence in a microplate assay for genotoxicity. Journal of Biochemical and BiophysicalMethods, 51, 165–177.

21. Afanassiev, V., Sefton, M., Anantachaiyong, T., Barker, G., Walmsley, R.M., Wolfl, S.(2000). Application of yeast cells transformed with GFP expression constructs containingthe RAD54 or RNR2 promoter as a test for the genotoxic potential of chemical substances.Mutation Research, 464, 297–308.

22. Lichtenberg-Frate, H., Schmitt, M., Gellert, G., Ludwig, J. (2003). A yeast-based methodfor the detection of cyto and genotoxicity. Toxicology in Vitro, 17, 709–716.

23. Cahill, P.A., Knight, A.W., Billinton, N., Barker, M.G., Walsh, L., Keenan, P.O., Williams,C.V., Tweats, D.J., Walmsley, R.M. (2004). The GreenScreen R© genotoxicity assay: ascreening validation programme. Mutagenesis, 19, 105–119.

24. Sengstag, C., Eugster, H.P., Wurgler, F.E. (1994). High promutagen activating capac-ity of yeast microsomes containing human cytochrome P-450 1A and human NADPH-cytochrome P-450 reductase. Carcinogenesis, 15, 837–843.

25. Walsh, L., Hastwell, P.W., Keenan, P.O., Knight, A.W., Billinton, N., Walmsley, R.M.(2005). Genetic modification and variations in solvent increase the sensitivity of the yeastRAD54–GFP genotoxicity assay. Mutagenesis, 20, 317–327.

26. Knight, A.W., Billinton, N., Cahill, P.A., Scott, A., Harvey, J.S., Roberts, K.J., Tweats,D.J., Keenan, P.O., Walmsley, R.M. (2007). An analysis of results from 305 compoundstested with the yeast RAD54–GFP genotoxicity assay (GreenScreen GC)—includingrelative predictivity of regulatory tests and rodent carcinogenesis and performance withautofluorescent and coloured compounds. Mutagenesis, 22, 409–416.

27. Van Gompel, J., Woestenborghs, F., Beerens, D., Mackie, C., Cahill, P.A., Knight,A.W., Billinton, N., Tweats, D.J., Walmsley, M.M. (2005). An assessment of the util-ity of the yeast GreenScreen assay in pharmaceutical screening. Mutagenesis, 20, 449–454.

28. Fornace, A.J. Jr., Alamo, I. Jr., Hollander,M.C. (1988). DNAdamage-inducible transcriptsin mammalian cells. Proceedings of the National Academy of Sciences of the United Statesof America, 85, 8800–8804.

29. Papathanasiou, M.A., Kerr, N.C., Robbins, J.H., McBride, O.W., Alamo, I. Jr., Barrett,S.F., Hickson, I.D., Fornace, A.J. Jr. (1991). Induction by ionizing radiation of the gadd45gene in cultured human cells: lack of mediation by protein kinase C. Molecular andCellular Biology, 11, 1009–1016.

30. Wang, X.W., Zhan, Q., Coursen, J.D., Khan, M.A., Kontny, H.U., Yu, L., Hollander, M.C.,O’Connor, P.M., Fornace, A.J. Jr., Harris, C.C. (1999). GADD45 induction of a G2/M cellcycle checkpoint. Proceedings of the National Academy of Sciences of the United Statesof America, 96, 3706–3711.

31. Carrier, F., Bae, I., Smith, M.L., Ayers, D.M., Fornace, A.J. Jr. (1996). Characterizationof the GADD45 response to ionizing radiation in WI-L2-NS cells, a p53 mutant cell line.Mutation Research, 352, 79–86.

32. Mullan, P.B., Quinn, J.E., Gilmore, P.M., McWilliams, S., Andrews, H., Gervin, C.,McCabe, N., McKenna1, S., White, P. Song, Y.H., Maheswaran, S., Liu, E., Haber, D.A.,


Johnston, P.G., Harkin, D.P. (2001). BRCA1 and GADD45 mediated G2/M cell cyclearrest in response to antimicrotubule agents. Oncogene, 20, 6123–6131.

33. Pennie,W., Pettit, S.D., Lord, P.G. (2004). Toxicogenomics in risk assessment: an overviewof an HESI collaborative research program. Environmental Health Perspectives, 112, 417–419.

34. Price, B.D., Calderwood, S.K. (1992). Gadd45 and Gaddl53 messenger RNA levels areincreased during hypoxia and after exposure of cells to Agents which elevate the levels ofthe glucose-regulated proteins. Cancer Research, 52, 3814–3817.

35. Kultz, D., Madhany, S., Burg, M.B. (1998). Hyperosmolality causes growth arrest ofmurine kidney cells: induction of GADD45 and GADD153 by osmosensing via stress-activated protein kinase 2. Journal of Biological Chemistry, 273, 13645–13651.

36. Hastwell, P.W., Webster, T.W., Tate, M., Billinton, N., Lynch, A.M., Harvey, J.S., Rees,R.W., Walmsley, R.M. (2009). Analysis of 75 marketed pharmaceuticals using theGADD45a–GFP “GreenScreen HC” genotoxicity assay.Mutagenesis, 24, 455–463.

37. Billinton, N., Hastwell, P.W., Beerens, D., Birrell, L., Ellis, P., Maskell, S., Webster, T.W.,Windebank, S., Woestenborghs, F., Lynch, A.M., Scott, A.D., Tweatse, D.J., van Gompel,J., Rees, R.W., Walmsley, R.M. (2008). Interlaboratory assessment of the GreenScreenHC GADD45a–GFP genotoxicity screening assay: an enabling study for independentvalidation as an alternative method. Mutation Research, 653, 23–33.

38. Olaharski, A., Albertini, S., Kirchner, S., Platz, S., Uppala, H., Linc, H., Kolaja, K. (2008).Evaluation of the GreenScreen GADD45�–GFP indicator assay with non-proprietary andproprietary compounds. Mutation Research, 672, 10–16.

39. Billinton, N., Bruce, S., Rytter Hansen, J., Hastwell, P.W., Jagger, C., McComb, C., Klug,M.L., Pant, K., Rabinowitz, A., Rees, R., Tate, M., Vinggaard, A.M., Walmsley, R.M.(2010). A pre-validation transferability study of the GreenScreen HC GADD45a–GFPassay with a metabolic activation system (S9).Mutation Research, 700, 44–50.

40. Knight, A.W., Little, S, Houck, K., Dix, D., Judson, R., Richard, A., McCarroll, N., Aker-man, G., Yang, C., Birrell, L., Walmsley, R.M. (2009). Evaluation of high-throughputgenotoxicity assays used in profiling the US EPA ToxCastTM chemicals. Regulatory Toxi-cology and Pharmacology, 55, 188–199.

14HIGH-THROUGHPUT ASSAYS TOQUANTIFY THE FORMATION OFDNA STRAND BREAKS

Marıa Moreno-Villanueva and Alexander Burkle

14.1 SINGLE CELL GEL ELECTROPHORESIS ASSAY

The single cell gel electrophoresis assay (SCGE) is a method for analyzing DNAstrand break formation in individual cells. It was first described by Ostling andJohanson [1] in 1984 and modified by Singh et al. [2] in 1985. This assay is based onthe capacity of the negatively charged DNA to migrate to the positive electrode in anelectrical field. Due to the highly organized structure of the DNA, large DNA frag-ments from undamaged cells remain compact, while smaller fragments of damagedDNA can move toward the anode. There are basically two versions of this assay: analkaline version for the determination of single-strand breaks and a neutral one fordouble-strand breaks (DSBs; see below). After staining the DNAwith a specific fluo-rescence dye the resulting image under themicroscope is reminiscent of the shape of acomet; the “comet head” represents the high-molecular-weight DNA and the “comettail” contains the migrated fragments arising from strand breaks [3]. Therefore, thistechnique is also known as “comet assay.”

14.1.1 Procedure

The individual steps of the comet assay have been published in detail by Olive andBanath [4]. Basically, cells are embedded in a thin low-melting-point agarose gel andtransferred to amicroscope slide. After the agarose has gelled, the slides are immersedin a lysis solution containing salt at a high concentration and detergent. The lysis


285

286 HIGH-THROUGHPUT ASSAYS TO QUANTIFY DNA STRAND BREAKS

treatment destroys membranes and denatures proteins. The resulting nucleoids [5]are then electrophoresed either at neutral pH (the neutral comet assay) or at pH >13(the alkaline comet assay). In the neutral comet assay, the rate of migrated DNA isrelated to the chromatin loops and is believed to correlate with DSBs, while in thealkaline comet assay the comet tail is formed by single-stranded DNA fragments [6].To visualize the “comets,” DNA is stained with a specific dye such as propidiumiodide, ethidium bromide, or SybrGreen R© and photographed under the microscope.The amount of migrated DNA is a measure for the extent of DNA damage and istypically analyzed using commercially available image analysis software.

14.1.2 Applications

The comet assay is considered the “gold standard” method for measuring DNA strandbreaks. In addition, a large number of assay modifications have been established tocover different aspects of DNA strand break formation, including the investigationof DNA damage and repair at different cell cycle stages [7], discrimination of singleviable, apoptotic, and necrotic cells [8], detection of oxidative DNA damage [9, 10]and DNA crosslinks [11], and DNA damage in buccal mucosal cells [12], amongothers.

14.1.3 Automation and High Throughput

Automated systems are characterized by high throughput. Automated analysis ofslides from the comet assay has been proposed by Frieauff et al. [13], and a modi-fied protocol enables parallel testing of four compounds using 96-well plates, thusdoubling the previous throughput [14]. To screen simultaneously a large number ofsamples for genotoxicity, Stang and Witte described a high-throughput version ofthe comet assay [15], the main modification being the use of a multichamber plate(MCP) suitable for electrophoresis. The cells attach to the agarose at the bottom of theMCP. After treatment, the walls of the MCP can be removed and the comet assay canbe performed. Another very interesting modification is a high-throughput platformfor DNA damage analysis presented by Wood et al. [16]. This method comprisescapturing cells in an agarose microwell array; cells are then trapped in the wells usinggravity. After treatment, samples can be further processed as in the standard cometassay, images are acquired automatically, and a special software identifies the cometsand calculates the comet parameters.Although these approaches provide a higher throughput they so far do not allow

full automation of the comet assay. The development of an automated system for thewhole comet assay procedure might be difficult in view of the many different stepsinvolved.

14.2 FLUORIMETRIC DETECTION OF ALKALINE DNA UNWINDING

The principle of the fluorimetric detection of alkalineDNA unwinding (FADU) assaywas described by Birnboim and Jevcak in 1981 [17]. The detection of DNA strand

FLUORIMETRIC DETECTION OF ALKALINE DNA UNWINDING 287

Control valuesAbsolute DNA

quantity

Control valuesPhysiological

unwinding

Values for various

grades of damage

SybrGreen background

SybrGreen

SybrGreen

Neutr.

Neutr.

SybrGreen

Neutr.

SybrGreen

NaOH

NaOH

Lysis NaOH

NaOH

Neutr.

NoDamage

P0

DamagePx

HighDamage

B

Lysis

Lysis

NoDamage

T Lysis

FIGURE 14.1 Scheme of the principle of the FADU assay. The yellow boxes on the leftside of the figure represent the cells. In the middle the double-stranded DNA with increasinglevels of damage (gray boxes) and increasing extent of unwinding (blue boxes) is presented.The small circles (dark gray boxes) represent the fluorescent dye SybrGreen R© (yellow, nofluorescence signal; green, fluorescence signal). T, P0, and B are controls to be run in parallelwith the experimentally treated cells. In P0 samples alkaline unwinding is allowed and rep-resents the DNA strand breaks under physiological conditions (i.e., without exogenous DNAdamage). In T samples the neutralization immediately follows the lysis and therefore, uponaddition of the alkaline solution, unwinding cannot take place. T samples provide a measure oftotal DNA content and yield a fluorescence signal defined as 100%. In B samples the cellularDNA has completely been unwound as a result of very high damage level, and the resultingfluorescence represents the background. Px samples (P1, P2, P3, P4 . . . Px) are the experi-mental samples to be analyzed and may represent any extent of damage between P0 andB levels. (See insert for color representation of this figure.)

breaks and repair is based on progressiveDNAunwinding (denaturation) under highlycontrolled conditions of alkaline pH, time, and temperature. The starting points for theunwinding process are DNA “open sites” like replication forks or chromosome ends,but also DNA strand breaks induced by reactive oxygen species (ROS), radiation, orgenotoxic chemical compounds. To monitor DNA damage a commercially availablefluorescent dye such as ethidium bromide, SybrGreen R©, or PicoGreen R© is usedas a marker for the amount of DNA that has remained double-stranded during thealkaline incubation, and a decrease in the fluorescence intensity therefore indicates anincrease of DNAunwinding and consequently a greater number of DNA strand breaks(Fig. 14.1).


14.2.1 Procedure

After treatment with DNA-damaging agents, which may be followed by an additionalperiod of time to allow DNA repair, cells are resuspended in a cold isotonic bufferand kept at 0◦C. Thereafter, cells are lysed and proteins are denatured by adding alysis buffer comprising urea at a high concentration. The alkaline buffer is slowlyadded on top of the cell lysate in such a way as to form a second layer, thus avoidingany mechanical mixing with the lysate yet allowing its diffusion into the lysate.After 15 min the temperature is shifted to 30◦C and kept constant for 60 min inorder to allow unwinding of the DNA, starting from DNA strand interruptions. Tostop the unwinding process neutralization buffer is added to the cell lysate and inaddition the temperature is shifted to 22◦C. The amount of double-stranded DNAremaining in the lysate is determined by the fluorescence intensity of the DNA dyeadded at this point. The extent of damage is inversely related to the fluorescenceintensity.

14.2.2 Applications

The FADU assay has been applied to detect mutagen-induced DNA strand breaks[18], to detect repair of UV-induced DNA strand breaks [19], and to investigateenvironmental genotoxic effects in mammalian cells [20], among others.


The fast micromethod DNA single-strand break assay was developed by Batel et al.[21] in 1999. This method is easy to perform in a 96-well microplate. It requiresonly 30 ng DNA per single well and therefore could conveniently be used for thefast analysis of DNA strand breaks [21, 22]. Ullmann et al. [23] improved the fastmicromethod assay by measuring protein content in order to determine the numberof cells in each well and omitting the lysis step. A modified and automated versionof the FADU assay was published by Moreno-Villanueva et al. [24] in 2009. Themain advantages are precise dispensing of solutions (including the gentle addition ofthe alkaline solution) at a controlled rate without any mixing, perfect control of thetemperature of the samples at all times, full protection from light, reduction of thenumber of cells required by more than 100-fold, and finally automatic performanceof all the steps except cell preparation and DNA-damaging treatment. The readout isperformed in a 96-well-plate fluorescence reader. This automation guarantees highreproducibility, sensitivity, and objectivity of the readout, high throughput, easy andquick performance (2 h), and low cost for personnel.The automated FADU assay is a promising tool for a very wide range of appli-

cations, including the early screening of potentially genotoxic compounds [25].Recently, the automated FADUassay has been used in different contexts [26–28]. Twomodified versions have been established: the detection of 8-oxo-dG in mitochondrialDNA [29] and the reverse FADU assay for detection of DNA crosslinks [30].

IMMUNOFLUORESCENCE STAINING OF � -H2AX FOCI 289

14.3 IMMUNOFLUORESCENCE STAINING OF �-H2AX FOCI

Immunofluorescence is a widespread and common technique, which uses specificantibodies to visualize a protein of interest. H2AX is a variant form of histone H2A.H2A is a component of the nucleosome and contributes to the stabilization of thechromatin high-order structure. After induction of DSBs, H2AXbecomes phosphory-lated on serine 139. Phosphorylated H2AX, also termed �H2AX, is typically locatedin foci [31], which can be visualized under the microscope. The detection is based onindirect immunofluorescence using a fluorophore-labeled secondary antibody, whichbinds to a �H2AX-specific primary antibody [32, 33]. It is believed that each �H2AXfocus represents one DSB.

14.3.1 Procedure

After DSB formation and phosphorylation of H2AX, cells first have to be fixed usinga fixative, followed by several washing steps. To reduce unspecific binding, the cellsare incubated in a blocking buffer usually containing bovine serum albumin (BSA),detergent, and phosphate buffered saline (PBS). Thereafter, cells are incubated witha primary antibody targeting �H2AX. The cells are washed again and then incubatedwith a fluorophore-labeled secondary antibody directed against the primary antibody.To visualize the nuclei a DNA-specific fluorescence dye (DAPI, Hoechst) is used.The quantification of DSBs can be assessed by counting the �H2AX foci with thehelp of fluorescencemicroscopy,Western blotting, or flow cytometrymethods. Buffercomponents, antibody dilutions, incubation times, and temperatures depend on thedifferent assay conditions, for example, cell and tissue type, damage induction, andfluorescence readout.

14.3.2 Applications

Phosphorylated H2AX is a sensitive marker of DSBs and has therefore been used todetect genotoxic stress [34, 35], to monitor DNA damage in cancer research [36–38],as an indicator of environmental health risks [39–42], and to analyze DNA damageresponse pathways [43–45], apoptosis [46], and the cell cycle [47].


Immunofluorescence staining of�H2AX foci is recognized as a very sensitivemethodto detect DSBs. The number of foci per nucleus has to be counted by the operatorunder the microscope and represents the number of DSBs, but this is time-consumingand might introduce the same bias. Therefore, the development of high-throughputversions capable of objective assessment of DSB formation has been a goal over thelast years. Ivashkevich et al. [48] proposed a computational approach for automaticfocus counting using standard image processing algorithms; it results in an automaticand user-independent way for counting foci. Avondoglio et al. [49] developed a high-throughput electrochemiluminescent platform to quantify �H2AX levels based on


a sulfo-ester tag-conjugated phospho-H2AX instead of a fluorophore-labeled sec-ondary antibody. A very interesting approach has been developed by Turner et al.[50]: they successfully established a Rapid Automated Biodosimetry Tool (RABIT)for measuring �H2AX from fingerstick-derived samples of blood. This fully auto-mated method uses robotics, lymphocyte handling in 96-well plates, and high-speedimage acquisition, thus providing a high-throughput system for automated detectionof �H2AX in human lymphocytes.

14.4 CONCLUDING REMARKS

Even if these techniques provide useful tools for measuring DNA strand breaks ina high-throughput fashion, there are some concerns. The comet assay is clearly notconsistent enough: the European Comet Assay Validation Group (ECVAG) reportedan interlaboratory coefficient of variation (CV) of 47% [51]. There are large differ-ences in the reported values of DNA damage of leukocytes from healthy humansand it is not clear whether this is due to different comet assay protocols or to realbiological differences between the populations of different countries [52].Many steps of the comet assay protocol (including slide preparation and elec-

trophoresis) affect both intra-assay variability and interassay reproducibility [53]. Onthe other hand, the automated FADU assay is not able to distinguish between double-and single-strand breaks and depends on the availability of a liquid handling device.The fluorescence detection of DSBs assumes that phosphorylation of the histoneH2AX occurs at DSB sites. However, in some cases the presence of �H2AX foci inthe absence of DNA damage has been demonstrated [54]. The limitations and benefitsof this assay have been commented by Lobrich et al. [55].

REFERENCES

1. Ostling, O., Johanson, K.J. (1984). Microelectrophoretic study of radiation-induced DNAdamages in individual mammalian cells. Biochemical and Biophysical Research Commu-nications, 123, 291–298.

2. Singh, N.P., McCoy, M.T., Tice, R.R., Schneider, E.L. (1988). A simple technique forquantitation of low levels of DNA damage in individual cells. Experimental Cell Research,175, 184–191.

3. Olive, P.L., Banath, J.P., Durand, R.E. (1990). Heterogeneity in radiation-induced DNAdamage and repair in tumor and normal cells measured using the “comet” assay. RadiationResearch, 122, 86–94.

4. Olive, P.L., Banath, J.P. (2006). The comet assay: a method to measure DNA damage inindividual cells. Nature Protocols, 1, 23–29.

5. Cook, P.R., Brazell, I.A. (1976). Characterization of nuclear structures containing super-helical DNA. Journal of Cell Science, 22, 303–324.

6. Afanasieva, K., Zazhytska, M., Sivolob, A. (2009). Kinetics of comet formation in single-cell gel electrophoresis: loops and fragments. Electrophoresis, 31, 512–519.

REFERENCES 291

7. McArt, D.G.,McKerr, G., Saetzler, K., Howard, C.V., Downes, C.S.,Wasson, G.R. (2010).Comet sensitivity in assessing DNA damage and repair in different cell cycle stages.Mutagenesis, 25, 299–303.

8. Morley, N., Rapp, A., Dittmar, H., Salter, L., Gould, D., Greulich, K.O., Curnow, A.(2006). UVA-induced apoptosis studied by the new apo/necro-comet-assay which distin-guishes viable, apoptotic and necrotic cells. Mutagenesis, 21, 105–114.

9. Kushwaha, S., Vikram, A., Trivedi, P.P., Jena, G.B. (2011). Alkaline, Endo III and FPGmodified comet assay as biomarkers for the detection of oxidative DNA damage in ratswith experimentally induced diabetes.Mutation Research, 726, 242–250.

10. Gajski, G., Garaj-Vrhovac, V., Orescanin, V. (2008). Cytogenetic status and oxidativeDNA-damage induced by atorvastatin in human peripheral blood lymphocytes: standardand Fpg-modified comet assay. Toxicology and Applied Pharmacology, 231, 85–93.

11. Merk, O., Speit, G. (1999). Detection of crosslinks with the comet assay in relationshipto genotoxicity and cytotoxicity. Environmental and Molecular Mutagenesis, 33, 167–172.

12. Szeto, Y.T., Benzie, I.F., Collins, A.R., Choi, S.W., Cheng, C.Y., Yow, C.M., Tse, M.M.(2005). A buccal cell model comet assay: development and evaluation for human biomon-itoring and nutritional studies.Mutation Research, 578, 371–381.

13. Frieauff, W., Hartmann, A., Suter, W. (2001). Automatic analysis of slides processed inthe Comet assay.Mutagenesis, 16, 133–137.

14. Kiskinis, E., Suter, W., Hartmann, A. (2002). High throughput Comet assay using 96-wellplates. Mutagenesis, 17, 37–43.

15. Stang, A., Witte, I. (2009). Performance of the comet assay in a high-throughput version.Mutation Research, 675, 5–10.

16. Wood, D.K., Weingeist, D.M., Bhatia, S.N., Engelward, B.P. (2010). Single cell trappingand DNA damage analysis using microwell arrays. Proceedings of the National Academyof Sciences USA, 107, 10008–10013.

17. Birnboim, H.C., Jevcak, J.J. (1981). Fluorometric method for rapid detection of DNAstrand breaks in human white blood cells produced by low doses of radiation. CancerResearch, 41, 1889–1892.

18. Dusinska, M., Slamenova, D. (1992). Application of alkaline unwinding assay for detec-tion of mutagen-induced DNA strand breaks. Cell Biology and Toxicology, 8, 207–216.

19. Baumstark-Khan, C., Hentschel, U., Nikandrova, Y., Krug, J., Horneck, G. (2000). Flu-orometric analysis of DNA unwinding (FADU) as a method for detecting repair-inducedDNA strand breaks in UV-irradiated mammalian cells. Photochemistry and Photobiology,72, 477–484.

20. Daniel, F.B., Chang, L.W., Schenck, K.M., DeAngelo, A.B., Skelly, M.F. (1989). Thefurther development of a mammalian DNA alkaline unwinding bioassay with potentialapplication to hazard identification for contaminants from environmental samples. Toxi-cology and Industrial Health, 5, 647–665.

21. Batel, R., Jaksic, Z., Bihari, N., Hamer, B., Fafandel, M., Chauvin, C., Schroder, H.C.,Muller, W.E., Zahn, R.K. (1999). Amicroplate assay for DNA damage determination (fastmicromethod). Analytical Biochemistry, 270, 195–200.

22. Schroder, H.C., Batel, R., Schwertner, H., Boreiko, O., Muller, W.E. (2006). Fastmicromethod DNA single-strand-break assay. Methods in Molecular Biology, 314, 287–305.


23. Ullmann, K., Muller, C., Steinberg, P. (2008). Two essential modifications stronglyimprove the performance of the Fast Micromethod to identify DNA single- and double-strand breaks. Archives of Toxicology, 82, 861–867.

24. Moreno-Villanueva, M., Pfeiffer, R., Sindlinger, T., Leake, A., Muller, M., Kirkwood,T.B., Burkle, A. (2009). A modified and automated version of the “Fluorimetric Detectionof Alkaline DNA Unwinding” method to quantify formation and repair of DNA strandbreaks. BMC Biotechnology, 9, 39.

25. Moreno-Villanueva, M., Eltze, T., Dressler, D., Bernhardt, J., Hirsch, C., Wick, P., vonScheven, G., Lex, K., Burkle, A. (2011). The automated FADU-assay, a potential high-throughput in vitro method for early screening of DNA breakage. ALTEX, 28, 295–303.

26. Kappes, F., Fahrer, J., Khodadoust, M.S., Tabbert, A., Strasser, C., Mor-Vaknin, N.,Moreno-Villanueva, M., Burkle, A., Markovitz, D.M., Ferrando-May, E. (2008). DEKis a poly(ADP-ribose) acceptor in apoptosis and mediates resistance to genotoxic stress.Molecular and Cellular Biology, 28, 3245–3257.

27. Huljic, S., Bruske, E.I., Pfitzenmaier, N., Dietrich, D.R. (2008). Species-specific toxicityof aristolochic acid (AA) in vitro. Toxicology In Vitro, 22, 1213–1221.

28. Mangerich, A., Herbach, N., Hanf, B., Fischbach, A., Popp, O., Moreno-Villanueva, M.,Bruns, O.T., Burkle, A. (2010). Inflammatory and age-related pathologies in mice withectopic expression of human PARP-1. Mechanisms of Ageing and Development, 131,390–404.

29. Wenzel, P., Schuhmacher, S., Kienhofer, J., Muller, J., Hortmann, M., Oelze, M., Schulz,E., Treiber, N., Kawamoto, T., Scharffetter-Kochanek, K., Munzel, T., Burkle, A., Bach-schmid, M.M., Daiber, A. (2008). Manganese superoxide dismutase and aldehyde dehy-drogenase deficiency increase mitochondrial oxidative stress and aggravate age-dependentvascular dysfunction. Cardiovascular Research, 80, 280–289.

30. D↪ebiak,M., Panas, A., Steinritz, D., Kehe, K., Burkle, A. (2010). High-throughput analysisof DNA interstrand crosslinks in human peripheral blood mononuclear cells by automatedreverse FADU assay. Toxicology, 280, 53–60.

31. Rogakou, E.P., Pilch, D.R., Orr, A.H., Ivanova, V.S., Bonner, W.M. (1998). DNA double-stranded breaks induce histone H2AX phosphorylation on serine 139. Journal of Biolog-ical Chemistry, 273, 5858–5868.

32. Nakamura, A., Sedelnikova, O.A., Redon, C., Pilch, D.R., Sinogeeva, N.I., Shroff, R.,Lichten, M., Bonner, W.M. (2006). Techniques for gamma-H2AX detection. Methods inEnzymology, 409, 236–250.

33. Kuo, L.J., Yang, L.X. (2008). Gamma-H2AX—a novel biomarker for DNA double-strandbreaks. In Vivo, 22, 305–309.

34. Paulsen, M.T., Ljungman, M. (2005). The natural toxin juglone causes degradation of p53and induces rapid H2AX phosphorylation and cell death in human fibroblasts. Toxicologyand Applied Pharmacology, 209, 1–9.

35. Watters, G.P., Smart, D.J., Harvey, J.S., Austin, C.A. (2009). H2AX phosphorylation as agenotoxicity endpoint. Mutation Research, 679, 50–58.

36. Ivashkevich, A., Redon, C.E., Nakamura, A.J., Martin, R.F., Martin, O.A. (2012). Use ofthe � -H2AX assay to monitor DNA damage and repair in translational cancer research.Cancer Letters, 327, 123–133.

37. Sak, A., Stuschke, M. (2010). Use of �H2AX and other biomarkers of double-strandbreaks during radiotherapy. Seminars in Radiation Oncology, 20, 223–231.

REFERENCES 293

38. Podhorecka, M., Skladanowski, A., Bozko, P. (2010). H2AX phosphorylation: its role inDNA damage response and cancer therapy. Journal of Nucleic Acids, pii, 920161.

39. Rothkamm, K., Horn, S. (2009). Gamma-H2AX as protein biomarker for radiation expo-sure. Annali dell’Istituto Superiore di Sanita, 45, 265–271.

40. Ibuki, Y., Toyooka, T., Shirahata, J., Ohura, T., Goto, R. (2007). Water soluble fractionof solar-simulated light-exposed crude oil generates phosphorylation of histone H2AX inhuman skin cells under UVA exposure. Environmental and Molecular Mutagenesis, 48,430–439.

41. Kawanishi, M., Watanabe, T., Hagio, S., Ogo, S., Shimohara, C., Jouchi, R.,Takayama, S., Hasei, T., Hirayama, T., Oda, Y., Yagi, T. (2009). Genotoxicity of 3,6-dinitrobenzo[e]pyrene, a novel mutagen in ambient air and surface soil, in mammaliancells in vitro and in vivo. Mutagenesis, 24, 279–284.

42. Mattsson, A., Lundstedt, S., Stenius, U. (2009). Exposure of HepG2 cells to low levels ofPAH-containing extracts from contaminated soils results in unpredictable genotoxic stressresponses. Environmental and Molecular Mutagenesis, 50, 337–348.

43. Jakob, B., Splinter, J., Conrad, S., Voss, K.O., Zink, D., Durante,M., Lobrich,M., Taucher-Scholz, G. (2011). DNA double-strand breaks in heterochromatin elicit fast repair proteinrecruitment, histone H2AX phosphorylation and relocation to euchromatin. Nucleic AcidsResearch, 39, 6489–6499.

44. Van Attikum, H., Gasser, S.M. (2009). Crosstalk between histone modifications duringthe DNA damage response. Trends in Cell Biology, 19, 207–217.

45. Unal, E., Arbel-Eden, A., Sattler, U., Shroff, R., Lichten, M., Haber, J.E., Koshland, D.(2004). DNA damage response pathway uses histone modification to assemble a double-strand break-specific cohesin domain. Molecular Cell, 16, 991–1002.

46. Lu, C., Zhu, F., Cho, Y.Y., Tang, F., Zykova, T., Ma, W.Y., Bode, A.M., Dong, Z. (2006).Cell apoptosis: requirement of H2AX in DNA ladder formation, but not for the activationof caspase-3. Molecular Cell, 23, 121–132.

47. Fernando, R.N., Eleuteri, B., Abdelhady, S., Nussenzweig, A., Andang, M., Ernfors,P. (2011). Cell cycle restriction by histone H2AX limits proliferation of adult neu-ral stem cells. Proceedings of the National Academy of Sciences USA, 108, 5837–5842.

48. Ivashkevich, A.N., Martin, O.A., Smith, A.J., Redon, C.E., Bonner, W.M., Martin, R.F.,Lobachevsky, P.N. (2011). �H2AX foci as a measure of DNA damage: a computationalapproach to automatic analysis.Mutation Research, 711, 49–60.

49. Avondoglio, D., Scott, T., Kil, W.J., Sproull, M., Tofilon, P.J., Camphausen, K. (2009).High throughput evaluation of gamma-H2AX. Radiation Oncology, 4, 31.

50. Turner, H.C., Brenner, D.J., Chen, Y., Bertucci, A., Zhang, J., Wang, H., Lyulko, O.V.,Xu, Y., Shuryak, I., Schaefer, J., Simaan, N., Randers-Pehrson, G., Yao, Y.L., Amundson,S.A., Garty, G. (2011). Adapting the � -H2AX assay for automated processing in humanlymphocytes. 1. Technological aspects. Radiation Research, 175, 282–290.

51. Forchhammer, L., Johansson, C., Loft, S., Moller, L., Godschalk, R.W., Langie, S.A.,Jones, G.D., Kwok, R.W., Collins, A.R., Azqueta, A., Phillips, D.H., Sozeri, O., Stepnik,M., Palus, J., Vogel, U., Wallin, H., Routledge, M.N., Handforth, C., Allione, A., Matullo,G., Teixeira, J.P., Costa, S., Riso, P., Porrini, M., Møller, P. (2010). Variation in themeasurement of DNA damage by comet assay measured by the ECVAG inter-laboratoryvalidation trial.Mutagenesis, 25, 113–123.


52. Møller, P. (2006). Assessment of reference values for DNA damage detected by the cometassay in human blood cells DNA. Mutation Research, 612, 84–104.

53. Møller, P., Moller, L., Godschalk, R.W., Jones, G.D. (2010). Assessment and reductionof comet assay variation in relation to DNA damage: studies from the European CometAssay Validation Group. Mutagenesis, 25, 109–111.

54. Soutoglou, E., Misteli, T. (2008). Activation of the cellular DNA damage response in theabsence of DNA lesions. Science, 320, 1507–1510.

55. Lobrich, M., Shibata, A., Beucher, A., Fisher, A., Ensminger, M., Goodarzi, A.A., Barton,O., Jeggo, P.A. (2010). Gamma-H2AX foci analysis for monitoring DNA double-strandbreak repair: strengths, limitations and optimization. Cell Cycle, 9, 662–669.

15HIGH-THROUGHPUT VERSIONSOF THE COMET ASSAY

Irene Witte and Andre Stang

15.1 INTRODUCTION

The standard procedure of the alkaline comet assay, also named SCGE (single cellgel electrophoresis), was first introduced in 1988 [1]. Today the comet assay is widelyused to test genotoxicity in whole cells. Several kinds of DNA damage, such as DNAsingle-strand breaks, alkali-labile sites, and further damages that can be enzymaticallyconverted into strand breaks, can be detected. The comet assay has been applied infundamental research to analyze different kinds of DNA damage and DNA repair. Inapplied research the comet assay has been used to detect genotoxic compounds orgenotoxicity of environmental samples.In the standard in vitro comet assay, cells are incubated with the toxicant, thereafter

trypsinized, mixed with agarose, and transferred to agarose-coated slides. Subse-quently, the agarose embedded-cells are lysed, if needed incubated with DNA repairenzymes, and the DNA is electrophorised. The DNA migrates in the gel according toits length, which can be microscopically analyzed and the DNA damage quantified.DNA with strand breaks forms a comet, whereby the head of the comet representsundamaged DNA, and the tail represents DNA fragments (Fig. 15.1).The standard comet assay is an excellent method for the sensitive detection of

DNA damage, but is time-consuming and therefore inefficient for screening simulta-neously a large number of samples. Time- and material-saving progress was gainedby performing the comet assay in 96-well plates [2] instead of using Petri dishes.Newly developed fully automated evaluation systems further increased the through-put of samples [3–5]. Another time-intensive step concerns the performance of the


295

296 HIGH-THROUGHPUT VERSIONS OF THE COMET ASSAY

FIGURE 15.1 Comet formation: (a) no damage; (b) low damage; (c) high damage. (Seeinsert for color representation of this figure.)

electrophoresis after incubation of the cells with the genotoxic agents. In the standardin vitro comet assay each sample has to be transferred from the incubation vessel tomicroscope slides precoated with agarose. To shorten this procedure the transfer stepshould be avoided. Chemical treatment and electrophoresis are now performed on thesame plate [6, 7]. In the comet assay as well as in other genotoxicity assays a separatecytotoxicity test is required to exclude false-positive results due to high cytotoxicity[8, 9]. This additional time- and material-consuming step was abolished by integra-tion of the cytotoxicity measurement within the comet assay [6]. The above describedfour advancements over the standard procedure, namely, (1) the performance of thecomet assay in 96-well plates, (2) the omission of the transfer of treated cells to elec-trophoresis slides/plates, (3) the automated evaluation, and (4) the integration of thecytotoxicitymeasurement, are the basis of the high-throughputmode of the standard invitro comet assay. The next section of this chapter describes and compares in detail theexisting high-throughput versions of the comet assay developed by different groups.

15.2 DESCRIPTION OF THE DIFFERENT VERSIONSOF THE HIGH-THROUGHPUT COMET ASSAY

The standard procedure of the alkaline comet assay has been described in detail in theguidelines published byTice et al. [8] (Fig. 15.2, left side). In brief, cells grown in Petridishes or 96-well plates [2] are incubated with the genotoxic agents. After removingthe test substance, the cell samples are washed and one by one removed from the Petridishes or the 96-well plates by trypsinization, or an equivalent technique. Thereafter,the cells of each sample are mixed separately with low-melting (LMP) agarose,followed by a transfer of the cells to the slides precoated with agarose. Trypsinizationand transfer of the cells are time-consuming steps in the standard comet assay.Subsequently, all samples have to undergo individually the same procedure of lysis,alkaline unwinding and electrophoresis, staining, and evaluation of the DNA damage.To avoid the above-mentioned time-consuming steps fourmain developments have

been described over the past few years. These methods can be divided into two groupsof procedures, the methods of Trevigen [10] and Ritter and Knebel [11] on the onehand (Fig. 15.2, second column) and the procedures described by Stang andWitte [6]and Wood et al. [7] on the other hand (Fig. 15.2, third column). All high-throughput

DESCRIPTION OF THE DIFFERENT VERSIONS 297

Standard procedure [8] Procedure by Trevigen [10] &Ritter and Knebel [11]

Procedure by Stang and Witte [6] & Wood et. al. [7]

seeding of cells in Petri dishesor in multiwell plates

treatment of cells withgenotoxic agents

trypsinisation of individualcultures

cytotoxicityassessment

adding LMPagarose one

by one

transfer to individualmicroscope slides

add coverslips

lysis

alkalineelectrophoresis

neutralization, drying,staining

manual localization of nucleion slides; semi-automatic

scoring using imageanalysis system

automatic localization of nucleion slides; automatic scoring

using image analysissystem

automatic localization of nuclei onslides; automatic scoring using

image analysis system





lysis lysis

transfer to one microscope slide


adding LMP agaroseone by one

trypsinisation on individualcultures


seeding of cells in Petri dishesor in multiwell plates

Stang and Witte [6]

seeding of cell in“multi-chamber-plates”

seeding of cells

in modified wellplate

coating withLMP agarose

in Petri dishesor in multiwell

plates


demounting of surroundingplastic walls



integrated cytotoxicityassessment

demounting of surroundingplastic walls

coating with LMPagarose

Wood et al. [7]

FIGURE 15.2 Procedure of the standard and the high-throughput versions of the cometassay.

systems have in common that up to 96 samples can be electrophorised on a multiwellplate.In the procedure of Trevigen [10] and Ritter and Knebel [11] (Fig. 15.2, second

column) the treated cells are trypsinized, mixed with LMP agarose and transferredto special slides/plates. This allows the electrophoresis of up to 96 samples inone electrophoresis chamber. Trevigen utilizes a slide, which consists of a plasticplate with 2, 20, or 96 holes, each surrounded by a hydrophobic ring. The systemof Ritter and Knebel [11] is similar to that of Trevigen. They use a glass plate,which is especially processed by implanting 20 hydrophobic rings into the glass.The hydrophobic effect prevents the LMP agarose, mixed with cells, from runningover the rings. After drying, all steps such as lysis, unwinding, electrophoresis, andneutralization can be performed simultaneously in both systems. As a result, a higherthroughput of samples is reached. However, the procedure of Trevigen [10] andRitter and Knebel [11] still includes a trypsinization step with a subsequent transferof the chemically treated cells.The systems ofWood et al. [7] and Stang andWitte [6] achieve a higher throughput

when compared to themethods of Trevigen [10] or Ritter andKnebel [11] by avoidingthe time-consuming step of trypsinization and transfer of each sample onto an agarose-coated slide/plate (Fig. 15.2, third column). In contrast to the methods of Trevigen[10] and Ritter and Knebel [11], the experiment is conducted on an agarose-coated


96-well plate. All the following steps, including electrophoresis, are then performedwith all samples simultaneously.Electrophoresis has to be performed on a flat surface. Multiwell plates are not

suitable for electrophoresis because of the walls surrounding the wells. Wood etal. [7] and Stang and Witte [6] developed a procedure, in which the walls can beseparated from the agarose-coated plate, thereby allowing electrophoresis to occur ina plane. Thus, the trypsinization step is avoided.The system by Stang and Witte [6] is based on a specially designed 96-well

plate, named multichamber plate (MCP). The bottom plate is coated with an agarosematrix. After seeding, the cells are allowed to attach to the agarose matrix for severalhours before spreading. The cells of all samples remain on the plate during theentire experiment. After treatment the walls are demounted and the bottom platewith the cells is covered with LMP agarose. All the following steps are performedsimultaneously with all 96 samples (Fig. 15.2, third column).The procedure of Wood et al. [7], also described as “cell trapping system,” allows

single cells to fall into a mold surrounded by agarose. These molds of 20 to 50 �min depth and width are produced by a small chop during drying of the agarose gelon the bottom plate, thereby forming microwells. In each microwell one cell can beindividually trapped. After a 15-min sedimentation within the microwells the cellsare covered with agarose. Thereafter, walls and bottom are sealed, and treatment ofthe cells can occur. The cells embedded in agarose allow different treatments of thesingle wells. In contrast to the version of Stang and Witte [6], the system of Woodet al. [7] allows the examination of kinetics. For example, Wood et al. [7] measuredthe repair of DNA damage due to x-rays on one plate at different time intervals [7].False-positive results in the comet assay [12] may be caused by high toxicity of the

chemicals. Therefore, in parallel to the comet assay, cytotoxicity has to be measured[8]. The separate determination of cytotoxicity is a highly time-consuming factor. Themethod of Stang andWitte [6] is the only one that incorporates the cytotoxicity assayinto the comet assay. After chemical treatment the FDA assay is performed by stainingthe cells with fluorescein diacetate [13] in the MCP. Thereafter, the fluorescence ismeasured in a fluorescence reader [6]. Then, the walls of the MCP are demountedand the comet assay is performed.Besides the previously described developments in the high-throughput comet assay

procedure a further time-saving potential lies in the evaluation of the comets. A suc-cessive replacement of the manual evaluation of the comets by automatic scoringsystems has taken place. The interactive (manual) microscopic analysis of individualcomets takes several hours for each single experiment. In addition, interactive anal-ysis compared with automated systems results in a greater number of errors due tovariations in the individual selection of cells and measurement regions during theanalysis [3]. In the past, some automated analyzing systems were developed for theconventional assay [3–5], which reduced the comet scoring time by approximately50% compared with manual evaluation. It made unattended and overnight evaluationpossible because of the use of slide feeders. The automated analysis of 96-well platesshows a higher efficiency than the evaluation of slides, which is due to the layoutof the samples. All 96 samples are located in direct proximity to one another andcan be analyzed on the same plate, while on microscope slides only 1 to 2 samples

VALIDATION OF THE HIGH-THROUGHPUT ASSAYS 299

can be evaluated. The automatic change of the slides is time-consuming. In addition,in the high-throughput systems of Stang and Witte [6] and of Wood et al. [7] cellsare seeded on one level of a plate covered with agarose. In the conventional methodand other high-throughput versions [10, 11] the cells are mixed with agarose andthen plated. Therefore, the cells are found in all planes of the agarose coating. Thefocusing of comets found in different planes of the agarose needs more time than forcomets located in one plane. In conclusion, with the automated system used for the96-well MCP 50 comets per min are determined [14] compared to 5 to 8 comets permin in the case of microscope slides [3, 4].

15.3 VALIDATION OF THE HIGH-THROUGHPUT ASSAYS

This section only describes validation of the high-throughput modes with which thehighest throughput is achieved. These high-throughput versions have only recentlybeen developed [6, 7]. Therefore, only a few validation data have been published upto now in the scientific literature.The high-throughput mode using the MCP evaluated different classes of DNA-

damaging agents. The data were compared with those of the standard assay. Theconcentration-dependent comet formation, as shown in Figure 15.3, showed similar

FIGURE 15.3 Concentration-dependent comet formation induced by methyl methanesul-fonate (�), H2O2 (♦) and tetrachlorohydroquinone (◦); open symbols: conventional assay;filled symbols: high-throughput assay according to Reference 6.


results. Wood et al. [7] obtained comparable results to the standard assay when theyexamined the dose-dependent effect of x-rays, but they did not compare genotoxicchemicals in the standard assay and the high-throughput assay.Several cell lines have been evaluated in the high-throughput comet assay. Accord-

ing to themethod ofWood et al. [7] all adherent and nonadherent cell lines can be used,because the cells are embedded within two layers of agarose. Therefore, treatmentof adherent cells can be done without waiting for their attachment. The procedure ofWood et al. [7] successfully used adherent (ovarian carcinoma cells and primary hep-atocytes) and nonadherent human lymphoblasts. In principle, all cell types can alsobe examined using the method of Stang andWitte [15]. Nonadherent cell lines need tobe centrifuged on the MCP after chemical treatment as shown for lymphocytes [15].In the standard comet assay adherent cells have to be trypsinized before they can be

used. Trypsinization induces DNA damage leading to a doubling of the background[16]. Tominimize this background effect Stang andWitte [6] let the cells recover fromdamage provoked by trypsin. The trypsinized cells were allowed to attach for severalhours after seeding them in the wells of theMCP. Treatment with chemicals was donebefore the cells spread. This lag time after trypsinization reduced the DNA damage inthe untreated control. In the case of nonpermanent human fibroblasts as well as V79,HepG2, and CHO cells a mean of 2%DNA tail in the control was measured [15]. Thisbackground level is 1/4 of that of the high-throughput assay of Wood et al. [7] andthe standard assay performed with blood cells in human monitoring studies [17]. Theenhanced sensitivity of the high-throughput comet assay using the MCP [6] enablesthe detection of low genotoxicity of environmental samples or chemicals [18].Indirectly acting carcinogens have to be converted to ultimate carcinogens, which

are then able to damage DNA. The activation is only possible inmetabolically compe-tent cells such as hepatocytes. For genotoxicity testing eithermetabolically competentcells or nonmetabolically competent cell lines in the presence of the S9 fraction ofliver homogenates are used. Both systems were tested in the high-throughput cometassay with lymphocytes, V79 cells, fibroblasts, and HeLa cells in the presence ofS9, and HepG2 cells in the absence of S9 [6, 7]. In all cell lines genotoxicitywas observed after treatment with the indirectly acting compound pentachlorophe-nol [15]. Moreover, other indirectly acting carcinogens such as benzo[a]pyrene andcyclophosphamide induced comet formation in the high-throughput comet assayeither in nonmetabolizing cells in the presence of S9 fraction or in HepG2 cells [18].

15.4 ADVANTAGES OF THE HIGH-THROUGHPUT COMET ASSAYSCOMPARED TO THE CONVENTIONAL VERSION

The high-throughput versions of the comet assay save time, material, and as a conse-quence drastically reduce the costs compared to the standard assay. These methodsare especially needed when routinely monitoring large numbers of samples. For someapplications only small amounts of samples are available. The use of 96-well platesinstead of Petri dishes drastically reduces the quantity of samples/compounds needed[2]. The method of Stang and Witte [6] requires even less substance because of the

APPLICATION AREAS 301

integrated cytotoxicity measurement. Thereby, cytotoxicity is measured after chem-ical treatment with the same cells used for the determination of comet formation.Using 386-well MCPs may reduce the amount of substance needed even further; firstexperiments in this direction have already been performed (data not shown).Another advantage of the high-throughput over the standard comet assay is seen in

its higher sensitivity and the lower standard errors from sample to sample. This maybe explained by differences in the agarose coating of slides and plates. Zainol et al.[19] demonstrated differences between individual gels, which appreciably contributeto variations in DNA migration. While in the high-throughput systems 96 samplesare electrophorised on the same agarose gel, in the standard assay each sample needsa separately coated slide with individual gels.It is known that comet formation generally does not follow a Gaussian distribution

[20]. This means that in the case of a sample with 50 or more evaluated cometsthe medians (with percentiles) have to be calculated. The mean and the standarddeviation can only be obtained by measuring at least three, but preferably moresamples in parallel. In contrast to the standard assay, the performance of three orfour parallel samples in the high-throughput versions is easy and fast, and allowsdetermination of the mean with standard deviation. The calculation of the standarddeviation revealed a high homogeneity of the samples analyzed in parallel, the valuesbeing similar in the case of n = 3 and n = 6 [14]. Even at low concentrations, fourparallel measurements are sufficient to reach a significance level of p≤.01 or p≤.001for the percentage DNA tail parameter [14].It has also been shown that the time needed for automated evaluation of comets on

theMCPwas reduced by a factor of more than 10 when compared to the time requiredfor the conventional evaluation [14]. In summary, this high-throughput version of thecomet assay combined with the automated evaluating system increased the output bya factor of 180 compared with the standard method.

15.5 APPLICATION AREAS

The high-throughput comet assay facilitates the systematic screening for genotoxicity.Examination of samples in environmental and human monitoring as well as the earlytesting of pharmaceutical candidates demands a high throughput of samples. Weexpect that the high-throughput comet assay can be applied in all areas, in which thestandard assay has been used in the past, as described below.The comet assay is currently the most widely used method to detect DNA lesions

in eco-genotoxicology. Environmental monitoring with respect to genotoxicity ispreferentially done in aquatic systems [21], but also in air [22, 23] and soil [24].Genotoxic contaminations can be detected in cells of sentinel organisms such astobacco, mussels, fishes, and earthworms or in cells often used and easy to handle incell culture labs [25]. In accordance with the standard comet assay [26] examinationof aquatic samples with the high-throughput comet assay using human fibroblastsrevealed genotoxic effects, while no effects were observed in the micronucleus test(MNT; Fig. 15.4) [18]. When comparing the high-throughput comet assay and a


DN

A d

amag

e [ta

il m

omen

t]

0

1

2

3

4

520

30

40

50

60

70

80(a)

Fibroblasts –S9Fibroblasts +S9

Treated sewage waterPCNC

Surface water 1

Surface water 2

Surface water 3

Ground water 1

Ground water 2

Ground water 3

Ground water 4

Mineral water

(b)Micronuclei –S9Micronuclei +S9

0

2

4

% M

NB

N

6

8


Surface water 1

Surface water 2

Surface water 3

Ground water 1

Ground water 2

Ground water 3

Ground water 4

FIGURE 15.4 Mutagenicity/genotoxicity of aquatic samples measured (a) in the high-throughput comet test (b) in the micronucleus test (c) in the high-throughput Ames II testaccording to Reference 18. NC, negative control; PC, positive control; ∗∗ significance level p≤ .01.

APPLICATION AREAS 303

(c) 40

30

20

Am

es II

res

pons

e

10

0


Surface water 1

Surface water 2

Surface water 3

Ground water 1

Ground water 2

Ground water 3

Ground water 4

Mineral water

TA98 +S9TA98 –S9

TAMix –S9TAMix +S9

FIGURE 15.4 (Continued)

high-throughputAmes test (Ames II) it was found out that the two tests complementedeach other, probably because of their different sensitivities toward different classes ofmutagens. The high sensitivity of both tests combined with easy, quick, and economichandling favors the combination of the two test systems for screening environmentalprobes with low levels of genotoxic contaminants [18].Human monitoring data obtained with the alkaline comet assay are accumulating.

Results on environmental and occupational monitoring are summarized in severalreviews (see ref. [27]). The trend nowadays is to examine more occupational thanenvironmental exposure [28]. Confounding factors such as sex, smoking habit, age,and lifestyle are being exhaustively examined [17]. Considering these results as wellas the guidelines for human monitoring as recommended by Dusinska and Collins[27] the high-throughput comet assay will be a valuable tool as a predictor of humanrisk. Human and environmental monitoring will benefit from the low detection limitof the high-throughput comet assay described above.The high-throughput comet assay was also recommended for early drug candidate

selection [29]. Besides the low costs and time saving, the high-throughput cometassay with integrated cytotoxicity measurement only requires small quantities ofthe test substance. This favors the test for screening drug candidates at early stagesof development, that is, at a time when only small amounts of the compounds areusually available. Positively tested compounds may prematurely be discarded, whilenegatively tested candidates may be further developed.


The in vivo comet assay has been gaining attention in regulatory toxicologyinitiatives such as ICH (International Conference on Harmonization of TechnicalRequirements for Registration of Pharmaceuticals for Human Use), the EuropeanCommunity Regulation, and REACH (Registration, Evaluation, Authorisation andRestriction of Chemical Substances) and is currently being regarded as a useful testfor follow-up testing of positive in vitro findings. It is particularly useful as a secondin vivo test (e.g., as an alternative to the in vivo unscheduled DNA synthesis test).Guidelines of the in vivo comet assay were developed by Tice et al. [8] and wereupdated and amended by Hartmann et al. [30] and Burlinson et al. [31, 32]. In thecase of these guidelines 200 to 600 samples have to be analyzed to determine thegenotoxic potential of one substance. In addition, cytotoxicity measurements of theisolated cells are necessary to avoid false-positive results. All these requirementsfavor the use of the high-throughput versions of the in vivo comet assay. Due tothe integrated cytotoxicity measurement the assay described by Stang and Witte [6]may again offer an advantage. Thus, in the future the in vivo comet assay may be anattractive application area of the high-throughput versions of the comet assay.

15.6 COMPARISON OF THE HIGH-THROUGHPUT COMET ASSAYAND THE HIGH-THROUGHPUT MICRONUCLEUS TEST

Great efforts have been made over the past few years to develop high-throughputversions of genotoxicity assays using mammalian cells. The most common are theMNT and the comet assay. Both standard assays were extensively evaluated andwidely used in their conventional mode. About 2700 MNT articles and 4000 cometassay articles were published between 2006 and 2012 according to PubMed. Even so,a high degree of agreement between both assays exists. Positive effects in the MNTversus negative effects in the comet assay may be due to aneugenic effects, whichcan be detected in the MNT but not in the comet assay. Vice versa, negative effectsin the MNT versus positive effects of the comet assay may be caused by the transientnature of DNA damage, which can be seen in the comet assay but not in the MNT.As in the case of the high-throughput comet assay, the MNT is performed in 96-

well plates instead of in Petri dishes [33, 34] to save material. Several fully automatedimage analysis systems have been developed as well [33–36]. Furthermore, systemsto analyze micronuclei on slides [37] as well as in 96-well plates have been described[33, 34]. The time needed for automated evaluation of one sample is similar forthe high-throughput comet assay [14] and for the MNT [33, 38]. In the case ofthe automated micronuclei evaluation systems the predictive negative values werelower as compared to the manual scoring reviewed by Diaz et al. [33]. However, anoptimized image analysis system for a unique application with high sensitivity andsufficient robustness for routine analyses has not been developed so far.In general, the in vitroMNT procedure requires more time than all known in vitro

high-throughput versions of the comet assay. Eighty to ninety hours are necessaryto perform the MNT [39] without evaluation of the micronuclei [38, 40]. The high-throughput comet assay evaluates the same number of genotoxic agents in 12 to

REFERENCES 305

15 h [14]. Therefore, in those cases in which quick results are needed, for example,after accidents with chemicals or for some medical applications, in which individualresponses of the patients have to be determined, the comet assay should be preferred toget a first information on the presence or effects of carcinogenic compounds/mixtures.In addition, in environmental toxicology the comet assay may be superior to theMNT regarding sensitivity and quickness. On the other hand, the MNT is an acceptedgenotoxicity assay in regulatory toxicology for pharmaceuticals, industrial chemicals,agrochemicals, and consumer products, for which OECD guidelines exist [9, 41].Because of the high number of samples needed to test the in vivo genotoxicity of

a single compound, a high-throughput version of the MNT or the comet assay wouldresult in a strong improvement of the examinations. However, up to now there is noexperience with the in vivo high-throughput versions of the MNT or the comet assay.When compared to the comet assay the in vivo MNT shows a lower sensitivity [42],and cannot be applied to nondividing cells of organs. In consequence, only blood/bonemarrow is routinely examined in the in vivoMNT, while the comet assay can easily beused to analyze the most appropriate target tissues such as liver, intestine, or bladder.These limitations of the MNT may favor the use of the high-throughput comet assayin the future.The combination of the in vivo MNT with the in vivo comet assay has repeatedly

been recommended [43–46] to reduce cost and use of animals. The US NationalToxicology Program recently began to routinely incorporate the comet assay and theMNT in their subchronic rodent studies. This may also be an application area of thehigh-throughput versions of both assays in the future.

REFERENCES

1. Singh, N.P., McCoy, M.T., Tice, R.R., Schneider, E.L. (1988). A simple technique forquantitation of low levels of DNA damage in individual cells.Experimental Cell Research,175, 184–191.

2. Kiskinis, E., Suter, W., Hartmann, A. (2002). High throughput comet assay using 96-wellplates. Mutagenesis, 17, 37–43.

3. Bocker, W., Rolf, W., Bauch, T., Muller, W.U., Streffer, C. (1999). Automated cometassay analysis. Cytometry, 35, 134–144.

4. Frieauff, W., Hartmann, A., Suter, W. (2001). Automatic analysis of slides processed inthe comet assay.Mutagenesis, 16, 133–137.

5. Schunck, C., Johannes, T., Varga, D., Lorch, T., Plesch, A. (2004). New developments inautomated cytogenetic imaging: unattended scoring of dicentric chromosomes, micronu-clei, single cell gel electrophoresis, and fluorescence signals. Cytogenetic and GenomeResearch, 104, 383–389.

6. Stang, A., Witte, I. (2009). Performance of the comet assay in a high-throughput version.Mutation Research, 675, 5–10.

7. Wood, D.K., Weingeist, D.M., Bhatiaa, S.N., Engelward, B.P. (2010). Single cell trappingand DNA damage analysis using microwell arrays. Proceedings of the National Academyof Sciences of the United States of America, 107, 10008–10013.


8. Tice, R.R., Agurell, E., Anderson, D., Burlinson, B., Hartmann, A., Kobayashi, H., Miya-mae, Y., Rojas, E., Ryu, J.C., Sasaki, Y.F. (2000). Single cell gel/comet assay: guidelinesfor in vitro and in vivo genetic toxicology testing. Environmental and Molecular Mutage-nesis, 35, 206–221.

9. OECD Guideline 487—In VitroMammalian Cell Micronucleus Test.

10. Trevigen, www.trevigen.com

11. Ritter, D., Knebel, J. (2009). Genotoxicity testing in vitro–development of a higherthroughput analysis method based on the comet assay. Toxicology In Vitro, 23, 1570–1575.

12. Kirkland, D.J., Mueller, L. (2000). Interpretation of the biological relevance of genotoxi-city test results: the importance of thresholds.Mutation Research, 464, 137–147.

13. Rotman, B., Papermaster, B.W. (1966). Membrane properties of living mammalian cellsas studied by enzymatic hydrolysis of fluorogenic esters. Proceedings of the NationalAcademy of Sciences of the United States of America, 55, 134–141.

14. Stang, A., Brendamour, M., Schunck, C., Witte, I. (2010). Automated analysis of DNAdamage in the high-throughput version of the comet assay.Mutation Research, 698, 1–5.

15. Stang, A., Witte, I. (2010). The ability of the high-throughput comet assay to measurethe sensitivity of five cell lines toward methyl methanesulfonate, hydrogen peroxide, andpentachlorophenol. Mutation Research, 701, 103–106.

16. Singh, N.P., Tice, R.R., Stephens, R.E., Schneider, E.L. (1991). Amicrogel electrophoresistechnique for the direct quantitation of DNA damage and repair in individual fibroblastscultured on microscope slides.Mutation Research, 252, 289–296.

17. Møller, P. (2006). The alkaline comet assay: towards validation in biomonitoring of DNAdamaging exposures. Basic and Clinical Pharmacological Toxicology, 98, 336–345.

18. Stang, A. (2009). Eignung der high throughput Version des Comet Assays als Screening-Verfahren. Dissertation, Carl von Ossietzky University, Oldenburg, Germany. Availableat http://oops.uni-oldenburg.de/volltexte/2009/963/pdf/staeig09.pdf

19. Zainol, M., Stoute, J., Almeida, G.M., Rapp, A., Bowman, K.J., Jones, G.D. (2009).Introducing a true internal standard for the comet assay to minimize intra- and inter-experiment variability in measures of DNA damage and repair. Nucleic Acids Research,37, e150.

20. Duez, P., Dehon, G., Kumps, A., Dubois, J. (2003). Statistics of the comet assay: a key todiscriminate between genotoxic effects.Mutagenesis, 18, 159–166.

21. Frenzilli, G., Nigro, M., Lyons, B.P. (2009). The comet assay for the evaluation of geno-toxic impact in aquatic environments. Mutation Research, 681, 80–92.

22. Coronas, M.V., Pereira, T.S., Rocha, J.A.V., Lemos, A.T., Fachel, J.M.G., Salvadori,D.M.F., Vargas, V.M.F. (2009). Genetic biomonitoring of an urban population exposed tomutagenic airborne pollutants. Environment International, 35, 1023–1029.

23. Tovalin, H., Valverde, M., Morandi, M.T., Blanco, S., Whitehead, L., Rojas, E. (2006).DNA damage in outdoor workers occupationally exposed to environmental air pollutants.Occupational and Environmental Medicine, 63, 230–236.

24. Klobucar, G.I., Stambuk, A., Srut, M., Husnjak, I., Merkas, M., Traven, L., Cvetkovic, Z.(2011).Aporrectodea caliginosa, a suitable earthworm species for field based genotoxicityassessment? Environmental Pollution, 159, 841–849.

25. Dhawan, A., Bajpayee, M., Parmar, D. (2009). Comet assay: a reliable tool for the assess-ment of DNA damage in different models. Cell Biology and Toxicology, 25, 5–32.

REFERENCES 307

26. Maffei, F., Carbone, F., Forti, G.C., Buschini, A., Poli, P., Rossi, C., Marabini, L., Radice,S., Chiesara, E., Hrelia, P. (2009). Drinking water quality: an in vitro approach for theassessment of cytotoxic and genotoxic load in water sampled along distribution system.Environment International, 35, 1053–1061.

27. Dusinska, M., Collins, A.R. (2008). The comet assay in human biomonitoring: gene–environment interactions. Mutagenesis, 23, 191–205.

28. Valverde, M., Rojas, E. (2009). Environmental and occupational biomonitoring using thecomet assay. Mutation Research, 681, 93–109.

29. Witte, I., Plappert, U., de Wall, H., Hartmann, A. (2007). Genetic toxicity assessment:employing the best science for human safety evaluation part III: the comet assay as analternative to in vitro clastogenicity tests for early drug candidate selection. ToxicologicalSciences, 97, 21–26.

30. Hartmann, A., Agurell, E., Beevers, C., Brendler-Schwaab, S., Burlinson, B., Clay, P.,Collins, A., Smith, A., Speit, G., Thybaud, V., Tice, R.R. (2003). Recommendations forconducting the in vivo alkaline comet assay. 4th International Comet Assay Workshop.Mutagenesis, 18, 45–51.

31. Burlinson, B., Tice, R.R., Speit, G., Agurell, E., Brendler-Schwaab, S.Y., Collins, A.R.,Escobar, P., Honma, M., Kumaravel, T.S., Nakajima, M., Sasaki, Y.F., Thybaud, V., Uno,Y., Vasquez, M., Hartmann, A. (2007). Fourth International Workgroup on Genotox-icity testing: results of the in vivo comet assay workgroup. Mutation Research, 627,31–35.

32. Burlinson, B. (2012). The in vitro and in vivo comet assays.Methods in Molecular Biology,817, 143–163.

33. Diaz,D., Scott, A., Carmichael, P., Shi,W., Costales, C. (2007). Evaluation of an automatedin vitro micronucleus assay in CHO-K1 cells. Mutation Research, 630, 1–13.

34. Shibai-Ogata, A., Kakinuma, C., Hioki, T., Kasahara, T. (2011). Evaluation of high-throughput screening for in vitromicronucleus test using fluorescence-based cell imaging.Mutagenesis, 26, 709–719.

35. Rossnerova, A., Spatova, M., Schunck, C., Sram, R.J. (2011). Automated scoring oflymphocyte micronuclei by the MetaSystems Metafer image cytometry system and itsapplication in studies of human mutagen sensitivity and biodosimetry of genotoxin expo-sure. Mutagenesis, 26, 169–175.

36. Decordier, I., Papine, A., Vande Loock, K., Plas, G., Soussaline, F., Kirsch-Volders,M. (2011). Automated image analysis of micronuclei by IMSTAR for biomonitoring.Mutagenesis, 26, 163–168.

37. Schunck, C., Johannes, T., Varga, D., Lorch, T., Plesch, A. (2004). New developments inautomated cytogenetic imaging: unattended scoring of dicentric chromosomes, micronu-clei, single cell gel electrophoresis, and fluorescence signals. Cytogenetic and GenomeResearch, 104, 383–389.

38. Westerink, W.M., Schirris, T.J., Horbach, G.J., Schoonen, W.G. (2011). Development andvalidation of a high-content screening in vitromicronucleus assay in CHO-k1 and HepG2cells. Mutation Research, 724, 7–21.

39. Kirsch-Volders, M., Plas, G., Elhajouji, A., Lukamowicz, M., Gonzalez, L., Vande Loock,K., Decordier, I. (2011). The in vitroMN assay in 2011: origin and fate, biological signif-icance, protocols, high throughput methodologies and toxicological relevance. Archivesof Toxicology, 85, 873–99.


40. Decordier, I., Papine, A., Plas, G., Roesems, S., Vande Loock, K., Moreno-Palomo, J.,Cemeli, E., Anderson, D., Fucic, A.,Marcos, R., Soussaline, F., Kirsch-Volders,M. (2009).Automated image analysis of cytokinesis-blocked micronuclei: an adapted protocol and avalidated scoring procedure for biomonitoring. Mutagenesis, 24, 85–93.

41. OECD guideline 474—Mammalian Erythrocyte Micronucleus Test.


43. Rothfuss, A., Honma,M., Czich, A., Aardema,M.J., Burlinson, B., Galloway, S., Hamada,S., Kirkland, D., Heflich, R.H., Howe, J., Nakajima, M., O’Donovan, M., Plappert-Helbig,U., Priestley, C., Recio, L., Schuler, M., Uno, Y., Martus, H.J. (2011). Improvement ofin vivo genotoxicity assessment: combination of acute tests and integration into standardtoxicity testing. Mutation Research, 723, 108–120.

44. Toyoizumi, T., Ohta, R., Kawakami,K.,Nakagawa,Y., Tazura,Y.,Kuwagata,M.,Noguchi,S., Sui, H., Yamakage, K. (2012). Usefulness of combined in vivo skin comet assay andin vivo skin micronucleus test. Mutation Research, 743, 42–51.

45. Bowen, D.E., Whitwell, J.H., Lillford, L., Henderson, D., Kidd, D., Mc Garry, S., Pearce,G., Beevers, C., Kirkland, D.J. (2011). Evaluation of a multi-endpoint assay in rats,combining the bone-marrow micronucleus test, the comet assay and the flow-cytometricperipheral blood micronucleus test. Mutation Research, 722, 7–19.

46. Vasquez, M.Z. (2010). Combining the in vivo comet and micronucleus assays: a practicalapproach to genotoxicity testing and data interpretation. Mutagenesis, 25, 187–199.

16AUTOMATED SOFT AGAR COLONYFORMATION ASSAY FOR THEHIGH-THROUGHPUT SCREENINGOF MALIGNANT CELLTRANSFORMATION

Pablo Steinberg

16.1 INTRODUCTION

Cell transformation has been defined as the induction of certain phenotypic alter-ations in cultured cells characteristic of tumorigenic cells [1] such as altered cellmorphology, uncontrolled cell growth, anchorage-independent cell growth, autocrineproduction of growth factors, and tumor induction in immune-deficient nude mice.The term anchorage-independent growth indicates that cells do not require a solidsubstratum for growth like the glass or plastic surface of a culture dish/flask, butcan grow and form colonies in suspension or soft media. In order to test whethercells have been malignantly transformed, the soft agar colony formation assay, firstdescribed by MacPherson and Montagnier [2] almost 50 years ago, can be per-formed. The results obtained with this assay have consistently been shown to cor-relate with those observed in nude mice [3]. Hence, the extent of colony formationin soft agar is a useful in vitro parameter to predict tumorigenicity in the wholeanimal.Furthermore, the soft agar colony formation assay is being used to (I) test the

chemosensitivity of tumor cells toward established antitumoral agents [4], (II) to iden-tify new anticancer compounds [5], and (III) to establish new therapeutic strategies to


309

310 AUTOMATED SOFT AGAR COLONY FORMATION ASSAY

control cancer cell outgrowth, for example, modification of cancer cell behavior bygene transfection [6, 7]. By doing so the soft agar colony formation assay may helpto replace a high number of animal studies in the field of anticancer drug evaluationin the future [8].The major practical disadvantages of the classic soft agar colony formation assay

are the relatively long time period of 2 to 4 weeks needed to obtain the results, theuse of a considerable amount of plastic labware, and the laborious counting of thecolonies growing in soft agar. In addition, the temperature of the soft agar has to becarefully controlled when adding soft agar to the cells to be tested: if it were too high,it could damage the cells; if it were too low, the agar would solidify.In the last few years many of the above-mentioned problems have been solved.

First, to enable a higher throughput the soft agar colony formation assay has beenminiaturized from the classic 6-well format down to the 96-well microtiter plateformat [7, 11] and a first report on a soft agar assay in a 384-well microtiter plateformat has been published [5]. Second, the length of the experimental procedurehas been shortened down to 1 week [5, 7, 11]. Third, the extremely laborious andinconsistent cell colony counting has been substituted by various very effective andreliable methods such as automated image/volume analysis systems [9, 10] or astaining of the cells with Alamar BlueTM and their subsequent quantitation with aplate reader [5, 7, 11]. Lastly, the soft agar colony formation assay has been automated[11], thereby allowing the high-throughput screening of the ability of cells to growin an anchorage-independent manner. In the present chapter the major achievementsmentioned before will be highlighted.

16.2 96-WELL SOFT AGAR COLONY FORMATION ASSAY

Ke et al. [7] were the first to establish the soft agar colony formation assay in a96-well format coupled to a fluorimetric readout for the quantification of cell pro-liferation, a parameter indicative of cell colony formation. The cell lines HeLaHFand HeLa, the former being nontransformed and the latter being transformed, werecultured in a liquid medium for 1 day as well as in a semisolid soft agar mediumfor 7 days, both in 96-well microplates, to test the anchorage-dependent and theanchorage-independent growth, respectively [7]. At the end of the experimentalperiod an Alamar BlueTM (resazurin) staining was performed to determine the cellnumber in each well. The nonfluorescent resazurin is reduced to the red fluores-cent resorufin in living cells. The amount of resorufin generated in this way ismeasured fluorimetrically. If one assumes that the amount of resorufin generatedis directly proportional to the number of living cells in the individual wells, thefluorescence measurement provides an estimate of the extent of cell growth in softagar [5, 7].In liquid culture the intensity of Alamar BlueTM staining was directly proportional

to the number of HeLaHF and HeLa cells in each well, the HeLa cells proliferatingslightly faster than the HeLaHF cells. In soft agar a linear proportion between the

384-WELL SOFT AGAR COLONY FORMATION ASSAY 311

number of cells per well and the extent of Alamar BlueTM staining was observed inthe case of HeLa cells, while in the case of HeLaHF cells the staining intensity perwell remained at a background level, independently of the number of cells presentin each well. Thus, the soft agar test system developed by Ke et al. [7] allowed todetermine whether a cell line grows anchorage independently within 1 week and in96-well microplates. The 1-week incubation period has not only been shown to beadequate for HeLa, A549, DLD-1, DU145, HCT116, MCF7, and U87 cells [7] butalso for the colon cancer cell lines MIP-101, DLD-2, and HT-29 [11]. Whether thisis also the case of other cell lines must be tested individually.The modification of cancer cell behavior by gene transfection can also be ana-

lyzed by making use of the test system described by Ke et al. [7]. The colon car-cinoma cell line DLD-1 contains an activated K-RAS gene. If a lentiviral vectorincluding siRNA against mutant K-RAS was delivered into DLD-1 cells and thetransduced cells stably expressed mutant K-RAS siRNA, K-RAS mRNA levels weresignificantly reduced, this reduction leading to a concomitant decrease in the abilityof the cells to grow in soft agar [7]. Moreover, the same group showed that the1-week 96-well formatted soft agar colony formation assay can be used to studythe effects of transiently expressed siRNA directed, for example, against K-RAS orPLK, a serine/threonine protein kinase being more strongly expressed in HeLa cellsthan in their nontransformed counterparts. In both cases it could be shown that thereduced expression of the genes led to a strong reduction of HeLa cell growth insoft agar.

16.3 384-WELL SOFT AGAR COLONY FORMATION ASSAY

Anderson et al. [5] were the first to describe a robust 384-well high-throughput softagar colony formation assay. In the above-mentioned study the assay was used toidentify compounds capable of inhibiting the growth of the human lung carcinomacell line HCC827. In order to prevent the attachment of the tumor cells to the bottomof the wells, in a first step 10 �L of a 0.6% agar solution in cell culture mediumwas pipetted into each well. Thereafter, 50 �L of a cell suspension in 0.4% agar andwithin 24 h 10 �L of the compound to be tested diluted in cell culture medium wereadded to each well. After a 7-day incubation period (see below) at 37◦C an AlamarBlueTM staining was performed.In a series of preliminary experiments Anderson et al. [5] concluded that the

optimal conditions to perform the 384-well soft agar colony formation assay were touse 5000 cells per well, to incubate the cells with the test compounds for 7 days, andto expose cells to Alamar BlueTM for 6 or 24 h. Furthermore, they demonstrated thatconcentrations of dimethyl sulfoxide below 0.3% did not affect the assay.In a first validation experiment staurosporine, lapatinib, gefinitib, erlotinib, van-

detanib, and imatinib were tested. Staurosporine is a promiscuous kinase inhibitor[12], lapatinib is an inhibitor of the epidermal growth factor receptor and the ErbB-2tyrosine kinases [13], gefinitib and erlotinib are inhibitors of the epidermal growth


factor receptor tyrosine kinase [14, 15], vandetanib is an inhibitor of the vascularendothelial growth factor receptor 2 tyrosine kinase and to a lesser degree of theepidermal growth factor receptor tyrosine kinase [16], and imitanib is an abl kinaseinhibitor [17]. The IC50 values for staurosporine, gefinitib, erlotinib, and vandetanibregarding their growth inhibiting effect on HCC827 cells were in the range of 0.4 to 5nM, while lapatinib was a factor 100 less potent and imitanib up to a concentration of10�Mdid not affect cell growth. Thus, the 384-well soft agar colony formation assaydeveloped by Anderson et al. [5] excellently mirrored the results previously describedin the scientific literature. This conclusion is supported by the results obtained whentesting a randomly selected library of 9600 compounds [5].

16.4 AUTOMATED 96-WELL SOFT AGAR COLONYFORMATION ASSAY

In a recent study by Thierbach and Steinberg [18] an automated version of the 96-well soft agar colony formation assay, including a detailed step-by-step protocol, wasdescribed. The automation of the assay allows to strongly increase the reproducibilityof the results and to test the effects of various different compounds or differentconcentrations of a single compound on various different cell lines within a veryshort time interval.Thierbach and Steinberg [18] made use of a commercially available pipetting

system (the epMotion R© 5075 LH). It should be noted that a number of alternativepipetting systems are on the market that can be used. The work flow for the automatedsoft agar colony formation assay is schematically shown in Figure 16.1. The onlysteps that have to be performed manually in between are the addition of 2.4% w/vBactoTM agar, 0.85%w/v BactoTM agar, and the cell suspensions to the correspondingreservoirs in the liquid handling system [18]. The final concentration of the BactoTM

agar in the top agar and in the base layer is 1% and 0.35% w/v, respectively, as in thecase of the classic soft agar colony formation assay [2], and the test compound in thetop agar makes 5% v/v.In order to demonstrate the utility of the automated soft agar colony formation

assay, the sensitivity of three human colon cancer cell lines (MIP-101, DLD-2, andHT-29) toward seven different concentrations of the widely used chemotherapeuticagent 5-fluorouracil in a single 96-well microtiter plate was determined. Approxi-mately 40 min after the program started (i.e., after performing the steps delineatedin Fig. 16.1) the 96-well plate was removed from the epMotion R© 5075 LH andincubated at 37◦C, 5% CO2, and 95% humidity. After 6 days the individual cellsformed colonies (Fig. 16.2). The colonies were quantified by adding Alamar BlueTM

(resazurin) to each well. The nonfluorescent resazurin is metabolized to the red flu-orescent resorufin by reductive processes in living cells. After 3 h the amount ofresorufin generated in this way is measured fluorimetrically. If one assumes that theamount of resorufin generated is directly proportional to the number of living cellsin the individual wells, the fluorescence measurement provides an estimate of the

Base layer

Top agar

Bacto agar 2.4 %

Base-premix

Bacto agar 0.85 %

Top-mix

Test compound

Cell culture medium

5× DMEM

Fetal bovine serum

Penicillin/streptomycin

Cells

Top-premix

Water

70°C

50°C

37°C

Roomtemperature

FIGURE 16.1 Schematic representation of the automated soft agar colony formationassay work flow. Dulbecco’s modified Eagle’s medium (DMEM), fetal bovine serum, peni-cillin/streptomycin, complete medium, and distilled water were mixed to yield the base-premixand the top-premix. The base-premix and the BactoTM agar build the base layer, which coatsthe bottom of the plate wells. The top-premix, the cells, the BactoTM agar, and the compound tobe tested constitute the top agar. Reproduced from Reference 18 with permission of Elsevier.

Control 0.8 μM 3.1 μM 50 μM12.5 μM

5-Fluorouracil

FIGURE 16.2 Photomicrographs of representative MIP-101 cell colonies on day 6 afterincubating the cells with increasing concentrations of 5-fluorouracil (phase contrast, 20×objective). Reproduced from Reference 18 with permission of Elsevier.


0.05 0.

2

0.8

3.1

12.5

50.0

5-Fluorouracil [μM]

HT-29

MIP-101

100

75

50

25

0

100

75

50

25

0

0.01

Rel

ativ

e flu

ores

cenc

e [%

]

Con

trol

Rel

ativ

e flu

ores

cenc

e [%

]

100

75

50

25

0

DLD-2

Con

trol

0.01

0.05 0.

2

0.8

12.5

50.0


3.1

FIGURE 16.3 Automated soft agar colony formation assay results obtained after incubatingthe six 96-wellmicrotiter plates for 6 days. The fluorescence obtained is taken as ameasurementfor the ability of the cells to proliferate in soft agar (untreated cells = 100%). A 5-fluorouracilconcentration-dependent inhibition of cell growth is evident for the three cell lines tested. Themean ± standard deviation of six independent experiments per 5-FU concentration is shown.Reproduced from Reference 18 with permission of Elsevier.

extent of cell growth in soft agar [5, 7]. Figure 16.3 clearly shows that the use of5-fluorouracil leads to a concentration-dependent reduction of colony formation inall three cases, each point of the curves representing the mean + standard deviationof six independent experiments. Each experimental condition was tested in quadru-plicate and showed a high reproducibility (i.e., low degree of variation among thequadruplicates) (Fig. 16.4).The automated method considers all the options of classic soft agar colony for-

mation assay protocols, is extremely variable, and substitutes the three-layer by atwo-layer system. Moreover, one part of the culture medium can be replaced by con-ditioned medium and the substances to be tested come directly into contact with thecells. On each 96-well plate three different cell lines or three different cell concen-trations with a substance-diluting series (seven concentrations + solvent control) orup to eight different substances can be tested at the same time on a single 96-wellmicrotiter plate. Taken together, the method allows for the first time to test the cancer-inhibiting potential of a high number of compounds in the soft agar colony formationassay in an automated and highly reproducible way within a short period of time (40min for processing in the epMotion R© 5075 LH, 6 days incubation at 37◦C, 3 h for

REFERENCES 315


Rel

ativ

e fl

uo

resc

ence

un

its

0

5

10

15

20

25

30

35

50.012.53.10.80.20.050.01Control

FIGURE 16.4 Fluorescence units of resorufin formed byHT-29 cells incubated with increas-ing concentrations of 5-fluorouracil. A very low variability among the corresponding quadru-plicates is evident. Reproduced from Reference 18 with permission of Elsevier.

the Alamar BlueTM staining) and does not require special cell culture practice. In thiscontext it should be mentioned that only 70% ethanol is used to disinfect the differentparts of the pipetting system.

REFERENCES

1. Barrett, J.C., Ts’o, P.O. (1978). Evidence for the progressive nature of neoplastic trans-formation in vitro. Proceedings of the National Academy of Sciences USA, 75, 3761–3765.

2. MacPherson, I., Montagnier, L. (1964). Agar suspension culture for the selective assay ofcells transformed by polyoma virus. Virology, 23, 291–294.

3. Freedman, V.H., Shin, S. (1974). Cellular tumorigenicity in nude mice: correlation withcell growth in semi-solid medium. Cell, 3, 355–359.

4. Hamburger, A., Salmon, E. (1977). Primary bioassay of human myeloma stem cells.Journal of Clinical Investigation, 60, 846–854.

5. Anderson, S.N., Towne, D.L., Burns, D.J., Warrior, U. (2007). A high-throughput softagar assay for identification of anticancer compound. Journal of Biomolecular Screening,12, 938–945.

6. Figini, M., Ferri, R., Mezzanzanica, D., Bagnoli, M., Luison, E., Miotti, S., Canevari,S. (2003). Reversion of transformed phenotype in ovarian cancer cells by intracellularexpression of anti folate receptor antibodies. Gene Therapy, 10, 1018–1025.


7. Ke, N., Albers, A., Claassen, G., Yu, D., Chatterton, J.E., Hu, X., Meyhack, B., Wong-Staal, F., Li, Q.-X. (2004). One-week 96-well soft agar growth assay for cancer targetvalidation. BioTechniques, 36, 826–833.

8. Blumenthal, R.D., Goldenberg, D.M. (2007). Methods and goals for the use of in vitroand in vivo chemosensitivity testing.Molecular Biotechnology, 35, 185–197.

9. Wylie, P.G., Bowen,W.P. (2007). Determination of cell colony formation in a high-contentscreening assay. Clinics in Laboratory Medicine, 27, 193–199.

10. Fiebig, H.H., Maier, A., Burger, A.M. (2004). Clonogenic assay with established humantumour xenografts: correlation of in vitro to in vivo activity as a basis for anticancer drugdiscovery. European Journal of Cancer, 40, 802–820.

11. Thierbach, R., Steinberg, P. (2009). Automated soft agar assay for the high-throughputscreening of anticancer compounds. Analytical Biochemistry, 387, 318–320.

12. Karaman, M.W., Herrgard, S., Treiber, D.K., Gallant, P., Atteridge, C.E., Campbell, B.T.,Chan, K.W., Ciceri, P., Davis, M.I., Edeen, P.T., Faraoni, R., Floyd, M., Hunt, J.P., Lock-hart, D.J., Milanov, Z.V., Morrison, M.J., Pallares, G., Patel, H.K., Pritchard, S., Wodicka,L.M., Zarrinkar, P.P. (2008). A quantitative analysis of kinase inhibitor selectivity. NatureBiotechnology, 26, 127–132.

13. Rusnak, D.W., Lackey, K., Affleck, K., Wood, E.R., Alligood, K.J., Rhodes, N., Keith,B.R., Murray, D.M., Knight, W.B., Mullin, R.J., Gilmer, T.M. (2001). The effects ofthe novel, reversible epidermal growth factor receptor/ErbB-2 tyrosine kinase inhibitor,GW2016, on the growth of human normal and tumor-derived cell lines in vitro and in vivo.Molecular Cancer Therapeutics, 1, 85–94.

14. Barker, A.J., Gibson, K.H., Grundy, W., Godfrey, A.A., Barlow, J.J., Healy, M.P., Wood-burn, J.R., Ashton, S.E., Curry, B.J., Scarlett, L., Henthorn, L., Richards, L. (2001).Studies leading to the identification of ZD1839 (IressaTM): an orally active, selective epi-dermal growth factor receptor tyrosine kinase inhibitor targeted to the treatment of cancer.Bioorganic & Medicinal Chemistry Letters, 11, 1911–1914.

15. Pollack, V.A., Savage, D.M., Baker, D.A., Tsaparikos, K.E., Sloan, D.E., Moyer, J.D.,Barbacci, E.G., Pustilnik, L.R., Smolarek, T.A., Davis, J.A., Vaidya, M.P., Arnold, L.D.,Doty, J.L., Iwata, K.K.,Morin,M.J. (1999). Inhibition of epidermal growth factor receptor-associated tyrosine phosphorylation in human carcinomas with CP-358,774: dynamics ofreceptor inhibition in situ and antitumor effects in athymic mice. Journal of Pharmacologyand Experimental Therapeutics, 291, 739–748.

16. Wedge, S.R., Ogilvie, D.J., Dukes, M., Kendrew, J., Chester, R., Jackson, J.A., Boffey,S.J., Valentine, P.J., Curwen, J.O.,Musgrove, H.L., Graham, G.A., Hughes, G.D., Thomas,A.P., Stokes, E.S., Curry, B., Richmond, G.H., Wadsworth, P.F., Bigley, A.L., Hennequin,L.F. (2002). ZD6474 inhibits vascular endothelial growth factor signaling, angiogenesis,and tumor growth following oral administration. Cancer Research, 62, 4645–4655.

17. Buchdunger, E., Zimmermann, J., Mett, H., Meyer, T., Muller, M., Druker, B.J., Lydon,N.B. (1996). Inhibition of the Abl protein-tyrosine kinase in vitro and in vivo by a 2-phenylaminopyrimidine derivative. Cancer Research, 56, 100–104.


17HIGH-THROUGHPUTQUANTIFICATION OFMORPHOLOGICALLYTRANSFORMED FOCI IN Bhas 42CELLS (v-Ha-ras TRANSFECTEDBALB/c 3T3) USINGSPECTROPHOTOMETRY

Kiyoshi Sasaki, Ayako Sakai, and Noriho Tanaka

17.1 INTRODUCTION

Transformation assays using mammalian cells are widely utilized not only to studymechanisms of carcinogenesis at the molecular and cellular level but also to iden-tify chemical carcinogens. Based on the quantification method of transformed cells,the assays fall into three categories: (1) focus formation method [1, 2]; (2) colonyformation method [3, 4]; and (3) soft agar colony formation method [5, 6]. In thefocus formation method, established cell lines (e.g., BALB/c 3T3, NIH/3T3, andC3H10T1/2 cells) are used in almost all experiments, and multilayered and morpho-logically altered foci found amidst monolayered normal cells are scored. Althoughthese cell lines are clones, transformed foci with various different morphological fea-tures appear. Scoring of transformed foci is time-consuming and subjective [1, 2, 7]and therefore problematic.Bhas 42 cells, a clone established by transfection of the v-Ha-ras gene intoBALB/c

3T3 cells, are considered a model of initiated cells, since the cells show cell–cell


317

318 HIGH-THROUGHPUT QUANTIFICATION USING SPECTROPHOTOMETRY

contact inhibition, but are transformed by treatment with tumor promoters [8]. Onthe other hand, Asada et al. [9] found out that tumor initiators can induce cell trans-formation by slightly modifying the protocol; that is, transformed foci are inducedby treatment with tumor promoters for a longer period of time in the stationary phase(promotion assay), and with tumor initiators for a shorter period of time in the growthphase (initiation assay). We have recently been evaluating the Bhas 42 cell trans-formation assay by comparing it with the BALB/c 3T3 transformation assay [10],by analyzing the performance of detection of chemical carcinogenicity [11, 12], ininterlaboratory collaborative studies [13, 14], and in an international validation study[15]. These studies show that the Bhas 42 cell transformation assay is reproducibleand reliable and has the following advantages, when compared to the BALB/c 3T3cell transformation assay: (1) it is a simple assay: treatment with a tumor initiator canbe omitted to detect tumor promoters; (2) it is a short-term assay: the culture period isshortened from 4 to 6 weeks to 2.5 to 3 weeks; (3) it is a sensitive assay: specificity ishigh; and (4) it is an economical assay: six wells of 6-well plates are required for eachdose, instead of 8 to 12 60-mm dishes [16]. However, in spite of these improvementsand access to high-throughput assay screening technologies, the problem regardingthe scoring of transformed foci still remains unresolved, since Bhas 42 transformedfoci show various morphologies, as the parent BALB/c 3T3 cells do (Fig. 17.1).Many studies have concentrated on understanding the characteristics of trans-

formed cells at the molecular and cellular level. Especially, the differences in sen-sitivity to radiation [17, 18], heat [18], proteins [19, 20], and chemicals [21, 22] aswell as differences between normal and transformed cells [23, 24] have extensivelybeen examined. These studies suggested that if a chemical selectively kills normalcells, transformed cells can then be quantified objectively and within a short periodof time using a microplate reader following exposure to a dye such as WST-8 [25]and alamarBlue [26] to monitor the metabolic activity of living cells. We thereforescreened chemicals and found out that hydrogen peroxide specifically kills normalcontact-inhibited Bhas 42 cells (Fig. 17.2). Here, we describe in detail the protocol

FIGURE 17.1 Examples of the various morphologies of transformed Bhas 42 cell foci.Transformed foci in a well (left) and single transformed foci showing different morphologies(the other six pictures).

MATERIALS 319

of the novel method to quantify transformed foci in the Bhas 42 cell transforma-tion assay using 96-well plates by adding WST-8 or alamarBlue after treatment withhydrogen peroxide [27].

(a) (b)

Living transformed cellsDead normal cellsTransformed cellsNormal cells

FIGURE 17.2 Selection of transformed Bhas 42 cells by treatment with hydrogen peroxide.After transforming Bhas 42 cells by treatment with 1 �g/mLMCA, the cells were treated with0.0015% hydrogen peroxide for 24 h. Shown are microphotographs before (a) and after (b)treatment with hydrogen peroxide; both microphotographs show the same position. (See insertfor color representation of this figure.)

17.2 MATERIALS

17.2.1 Cell Line

� Bhas 42 cells (v-Ha-ras-transfected Balb/c 3T3 clone A31-1-1 cells): cat. No.JCRB0149 from JCRB cell bank (Osaka, Japan). They are confirmed to be freeof bacteria, fungi, and mycoplasma.

17.2.2 Reagents

� Eagle’s minimum essential medium (MEM): cat. No. 11095-080 from LifeTechnologies (Grand Island, NY); store at 4◦C.

� Dulbecco’s modified Eagle’s medium/Ham’s F12 (DMEM/F12): cat. No.11330-032 from Life Technologies; store at 4◦C.

� Fetal bovine serum (FBS): from Life Technologies; select a lot which induceslow spontaneous focus formation and high focus formation by the positivecontrol; store at –80◦C.

� Penicillin–streptomycin 100× solution (PS): 10000 units/mL Penicillin Gsodium and 10 mg/mL streptomycin sulfate; cat. No. 15140-122 from LifeTechnologies; store at < –20◦C.

� 0.25% Trypsin: cat. No. 15050-065 from Life Technologies; store at < –20◦C.


� 0.02% Ethylenediaminetetraacetic acid–phosphate-buffered saline (-): cat. No.15040-066 from Life Technologies; store at room temperature.

� 30% Hydrogen peroxide: cat. No. 216763-100ML from Sigma-Aldrich (St.Louis, MO); store at 4◦C.

� WST-8: cat. No. CK04 from Dojindo Laboratories (Kumamoto TechnoResearch Park, Japan); store at 4◦C.

� alamarBlue: cat. No. DAL1100 from Biosource (Camarillo, CA); store at 4◦C.� 25%Glutaraldehyde: cat. No. G6257-100ML fromSigma-Aldrich; store at 4◦C.� Giemsa solution: cat No. 1.09204 from Merck (Darmstadt, Germany); store atroom temperature.

� Dimethyl sulfoxide (DMSO): cat. No. D8418-50ML from Sigma-Aldrich; storeat room temperature.

� 3-Methylcholanthrene (MCA): cat. No. 213942-100MG from Sigma-Aldrich;store at < –20◦C.

� Phorbol 12-myristate 13-acetate (12-O-tetradecanoylphorbol-13-acetate, TPA):cat. No. P-1585-1MG from Sigma-Aldrich; store at < –20◦C.

17.2.3 Equipment

� 96-well microplates: cat. No. 3598 from Corning Incorporated (Corning, NY).� 100-mm plates: cat. No. 430167 from Corning Incorporated.� Electronic 8-channel pipette: Xplorer; cat. No. 4861 000.163 from Eppendorf(Hamburg, Germany).

� Diapers or paper towels: Pampers from The Procter & Gamble Company(Cincinnati, OH).

� Microplate reader: Vmax-S from Molecular Devices (Sunnyvale, CA).

17.3 REAGENT SETUP

� M10F (MEM supplemented with 10% FBS): 500 mL MEM + 56 mL FBS+ 5 mL PS. Use for the expansion of provided cells, cell storage, and the firstculture after thawing. Store at 4◦C.

� DF5F (DMEM/F12 supplemented with 5% FBS): 500 mL DMEM/F12 +26.5 mL FBS + 5 mL PS. Used for routine passages, cell growth assay, andtransformation assay. Store at 4◦C.

� 0.0045% Hydrogen peroxide medium (3× the final concentration): 100 mLDF5F + 15 �L 30% hydrogen peroxide. Needed: 8 to 10 mL for one plate.Prepare before use.

� 10% WST-8 medium (2× the final concentration): 90 mL DF5F + 10 mLWST-8. Used for the cell growth assay. Needed: 13 to 15 mL for one plate.Prepare before use.

PROCEDURES 321

� 20% WST-8 medium (4× the final concentration): 80 mL DF5F + 20 mLWST-8. Used for the transformation assay. Needed: 8 to 10 mL for one plate.Prepare before use.

� 10% alamarBlue medium (2× the final concentration): 90 mL DF5F + 10 mLalamarBlue. Used for the cell growth assay. Needed: 13 to 15 mL for one plate.Prepare before use.

� 20% alamarBlue medium (4× the final concentration): 80 mL DF5F + 20mL alamarBlue. Used for the transformation assay. Needed: 8 to 10 mL for oneplate. Prepare before use.

� 2.5% Glutaraldehyde solution (10× the final concentration): 90 mL distilledwater/ultrapure water + 10 mL 25% glutaraldehyde. Needed: 5 to 7 mL forone plate. Prepare before use.

� 5% Giemsa solution: 95 mL distilled water/ultrapure water + 5 mL Giemsasolution. Needed: 13 to 15 mL for one plate. Store at room temperature.

� Test chemicals: Dissolve or suspend in an appropriate solvent or vehicle (usu-ally distilled water/ultrapure water or DMSO) at a high concentration as a stocksolution and store in aliquots at <–20◦C. Before treatment, dilute to each indi-vidual concentration 20× the final concentration with distilled water/ultrapurewater and 1000× the final concentration with DMSO. Never refreeze thealiquots.

� Negative controls: The solvent/vehicle for test chemicals. The final concentra-tion in the medium: 5% for distilled water/ultrapure water and 0.1% for DMSO(permissible up to 0.5% when test chemicals do not dissolve).

� Positive controls:Aknown tumor initiator,MCA (final concentration: 1�g/mL),for the initiation assay and a tumor promoter, TPA (final concentration: 50ng/mL), for the promotion assay. Prepare 1000× the final concentration withDMSO as stock solutions (MCA: 1 mg/mL and TPA: 50 �g/mL) and store inaliquots at < –20◦C. Never refreeze the aliquots.

17.4 PROCEDURES

17.4.1 Overview of the Bhas 42 Cell Transformation Assay by the HydrogenPeroxide Method

The entire process of the Bhas 42 cell transformation assay consists of the initiationassay and the promotion assay [12, 14, 15]. However, these assays are independent,and either one or the other can be performed, depending on the purpose. In eachassay, in a first step the cell growth assay is performed to set the doses of the testchemicals, and then, in a second step and based on the results of the cell growthassay, the transformation assay and the concurrent cell growth assay are performed.The cell growth assay is performed concurrently to confirm the reproducibility of thecell growth response (Fig. 17.3).


Second step:Detect initiation or promotion potential and

confirm the reproducibility of the cell growth response

First step:Set doses

Assays

Cell growth assayInitiation assay

Cell growth assayPromotion assay

Transformation assayConcurrent cell growth assay

Transformation assayConcurrent cell growth assay

FIGURE 17.3 Flow chart of the Bhas 42 cell transformation assay. The initiation or promo-tion assay is independently repeated as needed.

Figure 17.4 shows the protocol of the transformation assay. Only two points,the number of cells plated and the treatment procedure of the test chemicals, aredifferent between the initiation and promotion transformation assays. In the initiationtransformation assay, 200 cells are plated into the wells, and the cells are treated withthe test chemicals for a shorter period (72 h: from Day 1 to 4) in the growth phase.Conversely, in the promotion transformation assay, 400 cells are plated, and the cells

0 4 7 11 Days 14 21 22

Medium change

Initiation assay (200 cells/well)

Cell growth assay

Transformation assay

Promotion assay (400 cells/well)

Cell growth assay

Transformation assay

FIGURE 17.4 Protocol of the Bhas 42 cell transformation assay by the hydrogen peroxidemethod. Open square: test chemicals (tumor initiators: 72 h, and tumor promoters: 10 days),stippled square: hydrogen peroxide (0.0015%, 24 h), diagonal hatched square: WST-8 oralamarBlue (5%, 3 h).

PROCEDURES 323

are treated with the test chemicals for a longer period (10 days: from Day 4 to 14) inthe stationary phase.In the transformation assay, the transformed cells are grown in fresh medium

after treatment with the test chemicals. On Day 21 after plating, the transformed focireach a size that is sufficient for quantification. In the transformation assay, WST-8or alamarBlue is added following exposure to hydrogen peroxide.In the cell growth assay, the cells are treated with the test chemicals in the same

manner as in the transformation assay, except for the shorter culture period. The cellgrowth assays in the first step and in the second step are performed in the same way.On Day 7 after plating, cell growth is measured using WST-8 or alamarBlue.Whenever experiments are conducted, negative and positive controls are necessary.

When the solvent/vehicle of test chemicals is 5% water, DMSO control is alsoessential as the control for MCA and TPA.M10F medium is used from the moment that frozen cells are thawed and during

the first passage. Thereafter, DF5F medium is used in all other steps of the test. Thisis because M10F medium allows the cells to attach and grow well and the DF5Fmedium reduces spontaneous transformation.In the classic transformation assay, in which the foci are counted microscopically,

all procedures are the same as above, except that the cells are fixed with methanol andstained with Giemsa solution, instead of being treated with WST-8 or alamarBluefollowing exposure to hydrogen peroxide. Then, the transformed foci are visuallyscored under a stereoscopic microscope.

17.4.2 Solubility Test

Before the cells are treated with test chemicals, a solubility test should be performed.The highest final concentration is 5 mg/mL or 10 mM, as in the case of other in vitroshort-term assays. To achieve a final concentration of 5 mg/mL, the stock solutionsare highly concentrated, that is, 100 mg/mL (20× final concentration) in distilledwater/ultrapure water and 5000mg/mL (1000× final concentration) in DMSO. Sincechemicals with such a high solubility are rare, the solubility test is needed to choosebetween distilled water/ultrapure water and DMSO, and to determine the maximumconcentration that can be prepared. Based on the maximum concentration of the stocksolution, the maximum final concentration is calculated.

1. Weigh an appropriate amount of the test chemical into two tubes (e.g., 10mg/tube).

2. Add distilled water/ultrapure water or DMSO into the corresponding tube littleby little until dissolved or equally suspended.

3. Choose the solvent/vehicle with the better solubility and obtain the maximumconcentration of the stock solution.

4. Calculate the maximum final concentration. If 10 mg/mL is the maximum con-centration of the stock solution in distilled water/ultrapure water, the maximumfinal concentration is 0.5 mg/mL, adding it to the medium at 5%. For DMSO,


the maximum final concentration is 0.01 mg/mL, adding it to the medium at0.1%. If the test chemical does not dissolve at 10 mg/mL in DMSO, the finalconcentration can be the same by adding 2 mg/mL to the medium at 0.5%.

17.4.3 Preparation of Cell Stock

Basically, Bhas 42 cells are handled in the same way as other cell lines (BALB/c 3T3,NIH/3T3, and C3H10T1/2 cells) in the transformation assay by the focus formationmethod [1, 2]. To prevent spontaneous transformation, Bhas 42 cells should bepreserved at as low a passage as possible after obtaining the cells. Furthermore, amaster stock and a working stock should be prepared, and then, one frozen tube of theworking stock should be thawed for each transformation assay to ensure reproducibleresults in repeated experiments. M10F medium is used.

1. Quickly thaw the frozen cells obtained in a 37◦C water bath.2. After centrifugation (1200 to 1500 rpm, 3 to 5 min) to remove DMSO, suspendthe cell pellet in 50 mL of M10F medium.

3. Distribute the cell suspension into five 100-mm plates and culture.

4. At 50 to 70% confluence, trypsinize the cells and resuspend at a cell densityof 5× 105 cells/mL in cold M10F medium containing 5% DMSO. If one tube(2.5× 105 cells) is thawed, under these conditions the cultures usually become50 to 70% confluent in five to six days.

5. Freeze in 0.5 mL aliquots (2.5× 105 cells/tube) at –80◦C for more than 2 h andstore in liquid nitrogen as a master stock (30 to 50 tubes).

6. Thaw one tube of the master stock and distribute in five to ten 100-mm platesand prepare a working stock (50 to 100 tubes) in the same manner as the masterstock.

7. Start every transformation assay with a tube from the working stock.

8. After using up all the working stocks, prepare more of them from the masterstock.

17.4.4 Cell Culture and Passage

It is known that a long-term culture at high density and many passages cause spon-taneous transformation of cell lines when performing the focus formation method[1, 2]. Bhas 42 cells should also be passaged before confluency and used at lowpassage for the transformation assay. However, to save frozen cell stocks, the cells athigher passages can be used for the cell growth assay and biochemical assays. DF5Fmedium is used.

1. Culture Bhas 42 cells in 100-mm plates with 10 mL DF5F medium in a humid-ified 5% CO2 incubator at 37◦C.

2. At about 50 to 70% confluence (2 to 2.5× 106 cells/plate), remove the mediumand wash the cells once with about 10 mL of 0.02% EDTA–PBS(-).

PROCEDURES 325

3. Add 0.5 to 1 mL of 0.25% trypsin into the plate and allow cells to detach in theincubator.

4. Suspend the cells in 5 to 10mLDF5Fmediumand centrifuge the cell suspension(1200 to 1500 rpm, 3 to 5 min) to remove trypsin.

5. Resuspend the cells with appropriate volumes of DF5F medium and adjust thecell concentration.

6. Passage the cells once a week at a 500- to 1000-fold dilution without cellcounting.

17.4.5 Preparation of Medium Containing Test Chemicals

The medium containing test chemicals should be prepared before treatment of thecells. All chemical treatment media contain an equal concentration of the sol-vent/vehicle. The final concentration of the solvent/vehicle is 5% with distilledwater/ultrapure water and 0.1% with DMSO (acceptable up to 0.5% when test chem-icals do not dissolve). Table 17.1 shows an example of preparation of medium atthe same final concentration (1 �g/mL) of the test chemical dissolved in distilledwater/ultrapure water or DMSO for the initiation and promotion assays.

1. Place multichannel pipette reservoirs and distribute medium into them.

2. Add chemicals to the medium.

TABLE 17.1 An Example of Preparation of Medium Containing the Test Chemical toTreat the Cells at 1 �g/mL

Test ChemicalSolvent or Vehicle

Assays SubstancesFinal

Conc. (%)

Conc. inSolution(�g/mL)

VolumeAdded(mL)

Conc. inReservoirs(�g/mL)

FinalConc.(�g/mL)

Medium(mL)

Initiation Water 5 20 1 2 1 9DMSO 0.1 1000 0.02 2 1 10

Promotion Water 5 20 1 1 1 19DMSO 0.1 1000 0.02 1 1 20

The test chemical solutions are added to the medium in reservoirs at the concentrations and volumesindicated. Initiation assay: 0.05 mL medium containing the test chemical is added to wells with 0.05 mLmedium. Promotion assay: 0.1 mL medium containing the chemical is replaced with 0.1 mL old medium.Conc., concentration.

17.4.6 Medium Change

We recommend draining off by inverting plates as a method of removing mediumrather than aspiration using Pasteur pipettes. The draining-off method has the follow-ing advantages: (1) quick removal; (2) complete removal; and (3) removal withoutdetaching cells by touching them with Pasteur pipettes.


The medium for the cell growth assay and the transformation assay is transferredfrom the same reservoir. Ninety-six wells for the transformation assay (one plate)and eight wells (one column) for the concurrent cell growth assay are used for eachconcentration (Fig. 17.5).

1. Place diapers or paper towels to catch the medium in a laminar flow cabinet.One can work outside the cabinet if the clean room shows no contamination.

2. Remove the medium by inverting the plates, that is, swing and shake the platesthree times to completely drain off the medium.

3. Dispense the medium gently from tips in contact with the upper part of the wellwalls.

Transformation assay(96 wells/concentration)

Cell growth assay(8 wells/concentration)

Reservoirs

Plates

FIGURE 17.5 Transfer of medium containing chemicals. One plate is used for severalconcentrations in the cell growth assay, and one plate is used for one concentration in thetransformation assay.

17.4.7 Preliminary Cell Growth Assay

We strongly recommend to perform in a first step a preliminary cell growth assaybefore the cell growth assay for dose setting. The cell growth varies a great dealdepending on the test chemicals, that is, not inhibited at the mg/mL level, inhibited atng/mL level, and enhanced at certain concentrations. Additionally, the transformationassay should be performed at optimum concentrations depending on the cell growthresponse. Therefore, the selection of a set of concentrations is more difficult thanthought, especially at a narrower concentration range. The preliminary cell growthassay can cut down on repeated cell growth assays for dose setting.In the preliminary cell growth assay, the cells are observed every day, and the cell

growth is judged under an inverted microscope. For example, many detergents lysethe cells immediately after exposure, and some apoptosis-inducing agents kill thecells associated with morphological changes of apoptosis, such as cellular shrinkage,within a few days. There is no need to treat the cells at such concentrations in the

PROCEDURES 327

growth assay for dose setting. Five to nine concentrations with a wide range (e.g.,0.03, 0.1, 0.3, 1 and 3 mg/mL) are set and one well is used for each concentration.

Day 0

1. Replate 400 cells in the same way as in the initiation assay (17.4.8.1), with theexception that 0.05 mL of a 8000 cells/mL suspension is distributed.

Day 1

2. Add 0.05 mL of DF5F medium containing the test chemical (2× the finalconcentration).

Day 1 to 4

3. Observe the cells every day under an inverted microscope.

17.4.8 Initiation Assay

17.4.8.1 Cell Growth Assay for Dose Setting In a similar way as in genotoxicitytests such as the chromosome aberration assay and themammalian cellmutation assay,genotoxic chemicals have a tendency to induce the highest response at concentrationsshowing strong cytotoxicity in the Bhas 42 cell transformation assay. Therefore, thefinding of a concentration near IC90 is important. Five to nine concentrations are set.

Day –3

1. At about 50 to 70% confluence, plate 0.7 to 1× 105 cells into 100-mm plateswith DF5F medium.

Day 0

2. At about 50 to 70% confluence, distribute 0.05 mL of the cell suspension (4000cells/mL, in DF5F medium) into each well of 96-well plates (200 cells/well).

3. Because of the small volume of medium, lightly tap the plates to spread thecells.

4. After keeping the plates at room temperature for 15 min until the cells haveattached, place the plates in an incubator.

Day 1

5. Add 0.05 mL of DF5F medium containing the test chemical (2× the finalconcentration) into each well.

Day 4

6. Discard the cell culture medium and add 0.1 mL fresh DF5F medium into eachwell.


Day 7

7. Add 0.1 mL of 10% WST-8 medium into each well (final concentration: 5%)and incubate for 3 h.

8. Shake the plates for several seconds (the shaking function of amicroplate readeris recommended) and measure the absorbance at 450 nm.

9. Correct the absorbance of each well by subtracting that of the blank wells(medium only, average of eight wells). Calculate the growth rates relative tothe solvent/vehicle control culture.

17.4.8.2 Transformation Assay and Concurrent Cell Growth Assay Five to nineconcentrations are set based on the results of the cell growth assay. These con-centrations cover a range from little or no cytotoxicity to the highest cytotoxicity(less than 20% survival compared to the control). Ideally, one concentration belowNOEL (no observed effect level, 20% above or below the control), two concentrationsbetween NOEL and IC50, and two concentrations between IC50 and IC90 are tested(Fig. 17.6a).If the cell growth is inhibited within a narrow concentration range, in view of

experimental error, it is good to set one or two additional concentrations above orbelow the determined concentration range in the transformation assay. Dilution factorshould not be greater than the square root 10, for example, 0.1, 0.3, 1, 3 mg/mL areacceptable and 0.003, 0.03, 0.3, 3 mg/mL are not acceptable.

10.80.60.40.200

25

50

75

100

125

150

Concentration (mg/mL)

Rel

ativ

e ce

ll gr

owth

(%

)

(a)

10.80.60.40.200

25

50

75

100

125

150


Rel

ativ

e ce

ll gr

owth

(%

)

(b)

10.80.60.40.200

25

50

75

100

125

150


Rel

ativ

e ce

ll gr

owth

(%

)

(c)

FIGURE 17.6 Examples of dose setting for the transformation assay. From the given data(closed circles) of the cell growth assay, optimum concentrations (triangles) are determined:(a) in the case that inhibition is observed in the initiation assay; (b) in the case that enhancementis observed in the promotion assay; (c) in the case that inhibition is observed in the promotionassay (c).

Day –6 or –7

1. Thaw frozen cells (2.5× 105 cells/tube) of the working stock and distributethem in two to five 100-mm plates with M10F medium.

Day –3

2. Passage the cells in the same way as in the cell growth assay (17.4.8.1).

PROCEDURES 329

Day 0

3. Replate the cells for the transformation assay and the concurrent cell growthassay in the same way as in the cell growth assay.

Day 1

4. Treat the cells with a chemical in the same way as in the cell growth assay.

Day 4

5. Change medium in the same way as in the cell growth assay.

Day 7

6. Transformation assay: Discard cell culture medium and add 0.1 mL freshDF5F medium into each well.

7. Concurrent cell growth assay: Measure the absorbance in the same way as inthe cell growth assay.

Day 10 or 11, and 14

8. Discard cell culture medium and add 0.1 mL fresh DF5F medium into eachwell.

Day 21

9. Add 0.05 mL of 0.0045% hydrogen peroxide medium into each well (finalconcentration: 0.0015%) and incubate for 18 to 24 h.

Day 22

10. Add 0.05 mL of 20%WST-8 medium into each well (final concentration: 5%)and incubate for 3 h.

11. Measure the absorbance in the same way as in the cell growth assay.

12. Correct the absorbance of eachwell by subtracting that of blankwells (mediumonly, average of eight wells).

17.4.9 Promotion Assay

17.4.9.1 Cell Growth Assay for Dose Setting Tumor promoters induce differenteffects on cell growth depending on their chemical properties, that is, they can dose-dependently enhance cell growth (e.g., TPA), dose-dependently inhibit cell growth(e.g., okadaic acid), or enhance at low concentrations but inhibit at high concentrations(e.g., lithocholic acid). Because optimum concentrations for the transformation assayvary depending on the growth curve, accurate ranges of enhancement and/or inhibitionshould be obtained. Five to nine concentrations are set.


Day –3

1. Passage the cells in the same way as in the initiation assay (17.4.8.1).

Day 0

2. Replate the cells in the same way as in the initiation assay except for thedistribution of 0.1 mL of the cell suspension (400 cells/well) and withouttapping the plates.

Day 4

3. Discard cell culture medium and add fresh 0.1 mL DF5F medium containingthe test chemical (1× the final concentration) into each well.

Day 7

4. Measure the absorbance in the same way as in the initiation assay.

17.4.9.2 Transformation Assay and Concurrent Cell Growth Assay Five to nineconcentrations are set based on the results of the cell growth assay. The concentrationsdetermined in the cell growth assay are sometimes not suitable for the transformationassay because of the different treatment periods of time in the cell growth assayfor dose setting (three days) and the transformation assay (10 days). Therefore,concentrations should be set with particular care [28]. For test chemicals that exhibitenhancement and inhibition of cell growth, optimum concentrations are selected insuch away that they cover awide range from little or no effect to inhibition interposingenhancement. Ideally, one concentration below NOEL, three concentrations in therange of growth enhancement, and one concentration in the range of weak growthinhibition are assessed (Fig. 17.6b). For test chemicals that inhibit cell growth dose-dependently, optimum concentrations are selected from little or no cytotoxicity tomoderate cytotoxicity (less than 50% survival compared to the control). Ideally, twoconcentrations below NOEL, two concentrations between NOEL and IC50 and oneconcentration above IC50 are assessed (Fig. 17.6c).Similar to the initiation assay, one or two additional concentrations are added in

the transformation assay if the cell growth is enhanced or inhibited within a narrowconcentration range, and the dilution factor should also not be greater than squareroot 10.

Day –6 or –7

1. Thaw the cells in the same way as in the initiation assay (17.4.8.2).

Day –3

2. Passage the cells in the same way as in the initiation assay.

PROCEDURES 331

Day 0

3. Replate the cells in the same way as in the initiation assay except for distri-bution of 0.1 mL of the cell suspension (400 cells/well) and without tappingthe plates.

Day 4

4. Change medium with 0.1 mL DF5F medium containing the test chemical(1× the final concentration).

Day 7

5. Transformation assay: Changemediumwith 0.1mLDF5Fmedium containingthe test chemical.

6. Concurrent cell growth assay: Measure the absorbance in the same way as inthe initiation assay.

Day 10 or 11

7. Discard cell culture medium and add 0.1 mL fresh DF5F medium containingthe test chemical into each well.

Day 14

8. Discard cell culture medium and add 0.1 mL fresh DF5F medium into eachwell.

Day 21

9. Treat the cells with hydrogen peroxide medium in the same way as in theinitiation assay.

Day 22

10. Treat the cells with WST-8 medium and measure the absorbance in the sameway as in the initiation assay.

17.4.10 Measuring Fluorescence Using AlamarBlue

The hydrogen peroxide method makes use of a dye that interacts with living trans-formed cells. Thus, any cell viability reagent may possibly be used. We recommendalamarBlue, in addition to WST-8, since both dyes were developed as a one-stepreagent. However, alamarBlue can bemonitored by both fluorescence and absorbance,and the fluorescencemeasurement ismore sensitive than the absorbancemeasurement(monitor at 570 nm using 600 nm as a reference).


1. Treat the cells with alamarBlue in the same way as with WST-8.

2. Measure the fluorescence intensity using a fluorescence microplate readerat an excitation wavelength of 530 nm and an emission wavelength of590 nm.

3. Calculate the fluorescence values in the same manner as with WST-8.

17.4.11 Acceptance of Criteria Test

The initiation or promotion assay is repeated independently, as needed, to satisfyacceptance criteria. If the cells are not confluent due to cytotoxicity at the end ofthe transformation assay, the concentration tested is invalid. If contamination ortechnical problems are observed, if the number of damaged wells is seven or morein the transformation assay and five ormore in the cell growth assay, the concentrationtested is invalid.

17.4.12 Data Analysis

After obtaining absorbance or fluorescence values, means and standard deviationsare calculated. Thereafter, the means of the groups are compared by performingappropriate statistical analyses.

17.5 OPTION: CONFIRMATION OF CORRESPONDENCE BETWEENHIGH ABSORBANCE VALUES AND TRANSFORMED FOCI

17.5.1 Procedures

To confirm whether wells with high absorbance or fluorescence values include trans-formed foci, the cells are fixed and stained with Giemsa solution after measuring theabsorbance or fluorescence. Since Bhas 42 cells are easily detached from the wells bytreatment with hydrogen peroxide, the cells must be fixedwith glutaraldehyde, a weakfixative. Transformation frequency is expressed as the number of transformed fociper well when using 6-well plates, while it is expressed as the number of wells withtransformed foci per 96 wells when using 96-well plates (i.e., a well having one trans-formed focus as well as a well having two ormore transformed foci are scored as one).

1. After measuring the absorbance or fluorescence, add 0.02 mL of 2.5% glu-taraldehyde solution into each well (final concentration: about 0.25%).

2. After 30 min or more, remove the cell supernatant.

3. Add 0.1 mL of 5% Giemsa solution into each well and stain the cells for 30min or more.

4. Wash the cells with tap water and let them dry.

5. Score foci according to the following characteristics: (a) more than 100 cells,(b) spindle-shaped cells, (c) deep basophilic staining, (d) random orientation

OPTION: CONFIRMATION OF CORRESPONDENCE 333

of cells at the edge of foci, (e) dense multilayering and (f) invasive growth intothe monolayer [1, 2, 7].

6. Record the presence or absence of transformed foci and the absorbance valuein each well.

7. Express the transformation frequency as the number of wells with foci per 96wells.

17.5.2 Examples of Results Obtained

Figure 17.7 shows a typical example of plates having MCA-induced transformedfoci before and after Giemsa staining. Adding cell viability chromogenic reagentssuch as WST-8 and alamarBlue after treatment with hydrogen peroxide enablesto demonstrate the presence of transformed foci, that is, the color of wells withtransformed foci changes from tinted red to dark orange with WST-8 and from blueto purple with alamarBlue. Even in the case that transformed foci are positionedon the well walls (indicated as arrows in Fig. 17.7) and are easily missed under amicroscope, the hydrogen peroxide method can accurately detect them in the wells.When using alamarBlue and arranging the fluorescence values of all 96 wells in

ascending order, a dose-dependent increase of the total fluorescence values becomesapparent (Fig. 17.8a). In fact, transformation frequency (the number of wells withtransformed foci) increases dose-dependently (MCA 0 �g/mL: 6/96, 0.1 �g/mL:24/96, 0.3 �g/mL: 40/96 and 1 �g/mL: 66/96), and the wells showing high fluores-cence values nicely correspond to wells with transformed foci (Fig. 17.8b). Similarresults have been obtained with WST-8 [27]. Troubleshooting advice is provided inTable 17.2.

0.1% DMSO 1 μg/mL MCA 0.1% DMSO 1 μg/mL MCA

alamarBlue

WST-8

H2O2＋ Dyes Fixation and Giemsa staining

FIGURE 17.7 Concordance between wells colored by dyes and wells with transformed fociin the same plates. Bhas 42 cells subjected to the transformation assay with MCA were treatedwith alamarBlue or WST-8 following exposure to hydrogen peroxide (left four pictures), andthen the cells were fixed and stained with Giemsa solution (right four pictures). Arrows:Although the colors change, transformed foci are not seen because they are on the well sides.(See insert for color representation of this figure.)


968472604836241210

5000

10000

15000

20000

25000 (a)

(b)

Wells

Fluo

resc

ence

0

5000

10000

15000

20000

25000

Fluo

resc

ence

0

5000

10000

15000

20000

25000

Fluo

resc

ence

0

5000

10000

15000

20000

25000

Fluo

resc

ence

0.1% DMSO

96847260483624121Wells

0.1 μg/mL MCA

96847260483624121Wells

0.3 μg/mL MCA

96847260483624121Wells

1 μg/mL MCA

9080706050403020101 1 60

5000

10000

15000

20000

25000

Wells

Fluo

resc

ence

0.1% DMSO

241017260504030201010

5000

10000

15000

20000

25000

Wells

Fluo

resc

ence

0.1 μg/mL MCA

40302010156504030201010

5000

10000

15000

20000

25000

Wells

Fluo

resc

ence

0.3 μg/mL MCA

56504030201014030201010

5000

10000

15000

20000

25000

Wells

Fluo

resc

ence

1 μg/mL MCA

FIGURE 17.8 Dose-dependent increase of alamarBlue fluorescence values correlates withinduction of transformed foci by treatment with MCA in Bhas 42 cells. After measuring thefluorescence, the cells were fixed and stained with Giemsa solution, and then, transformedfoci were scored. The fluorescence values are simply arranged in ascending order (a), and inthe same plates, are arranged with and without transformed foci in ascending order (b). Thered and blue bars indicate fluorescence values of wells with and without transformed foci,respectively. (See insert for color representation of this figure.)

17.6 DISCUSSION

The hydrogen peroxide method has many advantages when compared with the classicobservation method; that is, transformed foci can be objectively quantified and theamount of time as well as labor are significantly reduced. For example, a total of 192plates for 12 test chemicals (for one test chemical: one plate for the negative control,one plate for the positive control, and six plates for the test chemical in the initiation

DISCUSSION 335

and promotion assays, respectively) can objectively be measured by one person inone day. Because the method is simple, just adding two reagents, an automatic systemcan be applied to achieve a high throughput (Table 17.3). Additionally, even whentransformed foci are formed on the well walls, which are often missed when using amicroscope, the hydrogen peroxide method can accurately detect them.We found out that not only morphologically transformed cells but also morpho-

logically normal cells survived treatment with hydrogen peroxide in some wells (Fig.17.8b, the cells of blue bars with high fluorescence values). If these cells were iso-lated and plated in soft agar medium, the colony formation frequencies of these cloneswere intermediate between Bhas 42 cells (nontumorigenic) and a morphologically

TABLE 17.2 Troubleshooting

Problems Possible Causes Solutions

Spontaneous transformationfrequency is high.

Serum is not suitable. Change the lot of serum.Culture period (3 wk) islong.

Shorten the culture period(2.5 wk).

Cells contain transformedcells.

Prepare a working stock again,obtain another lot or obtainthe cells from another source.

Transformation frequencyinduced by positivecontrols is low.

Serum is not suitable. Change the lot of serum.Culture period (3 wk) isshort.

Extend the culture period(3.5–4 wk).

Chemicals are inactivated. Prepare freshly. Do not increaseconcentrations to confirmsensitivity.

v-Ha-ras transfected cellsare lost.

A low possibility, because100% of the Bhas 42 cellscontain v-Ha-ras at highpassages.

Hydrogen peroxide does notkill normal cells or killstransformed cells.

Hydrogen peroxideconcentration is notsuitable.

Adjust the concentrationslightly.

A microplate reader recordsa maximum value.

Incubation period (3 h) islong.

Shorten the incubation period(2–2.5 h).

TABLE 17.3 Advantages of the Hydrogen Peroxide Method

Observation Method Hydrogen Peroxide Method

Equipment Microscope Plate readerQuantification Subjective ObjectiveTime Time consuming observation Short time measuringLabor Fixation, staining, washing, drying Just addingAutomatic system Not easy Easy


transformed Bhas 42 clone (tumorigenic) [27]. These data suggest that the hydrogen-peroxide-resistant morphologically normal cells are pretransformed cells, which turninto morphologically transformed cells with increased malignancy after a furtherculture period. Therefore, the hydrogen peroxide method is a new technique that canbe used to select cells, whose characteristics are altered at the molecular level.The selective killing of contact-inhibited Bhas 42 cells by treatment with hydrogen

peroxide is necrosis. If the cells were stained with fluorescein isothiocyanate–annexinV (to detect apoptosis, green fluorescence), ethidium homodimer III (to detect necro-sis, red fluorescence), and Hoechst 33342 (to detect living cells, blue fluorescence),the nuclei of the transformed foci yielded a blue fluorescence, whereas those of thesurrounding normal cells yielded a red fluorescence [27]. However, when the cellswere treated with hydrogen peroxide for 18 to 24 h in the hydrogen peroxide method,the contact-inhibited cells were killed within 6 h. While most studies report thathydrogen peroxide induces apoptosis [29], the fluorescence data and the rapid celldeath indicate that the cell death of Bhas 42 cells induced by hydrogen peroxide mayoccur through necrosis rather than through apoptosis.Recently, variousmethods that could be used to develop a high-throughputBALB/c

3T3 or Bhas 42 cell transformation assay have been described, including imageanalysis [30, 31], biochemical assays to measure butyrylcholinesterase [32] andalkaline phosphatase activities [33], and an automated version of the soft agar assay[34]. In principle, the hydrogen peroxide method is completely different from allthe above-mentioned methods. We demonstrated that the method could be appliednot only to the Bhas 42 but also to the BALB/c 3T3 cell transformation assay (datanot shown). Moreover, IC50 of hydrogen peroxide in the growth phase of Bhas 42cells was similar to that of transformed Bhas 42 cells, but that of contact-inhibitedBhas 42 cells was lower than that of transformed Bhas 42 cells (data not shown).These data indicate that hydrogen peroxide does not specifically act on the activeHa-ras gene, and the sensitivity of the cells to death induction by hydrogen peroxidechanges, depending on the cellular conditions. However, although further studies onthe molecular mechanisms of action of hydrogen peroxide are necessary, the presentmethod can presumably provide a powerful tool to select and quantify transformedfoci in all other cells in a focus formation method.

ACKNOWLEDGMENTS

This studywas sponsored by theNewEnergy and Industrial TechnologyDevelopmentOrganization in Japan (grant P06040).

REFERENCES

1. Heidelberger, C., Freeman,A.E., Pienta, R.J., Sivak,A., Bertram, J.S., Casto, B.C.,Dunkel,V.C., Francis, M.W., Kakunaga, T., Little, J.B., Schechtman, L.M. (1983). Cell transfor-mation by chemical agents—a review and analysis of the literature. A report of the U.S.Environmental Protection Agency Gene-Tox Program.Mutation Research, 114, 283–385.

REFERENCES 337

2. IARC/NCI/EPAWorking Group (1985). Cellular and molecular mechanisms of cell trans-formation and standardization of transformation assays of established cell lines for theprediction of carcinogenic chemicals: overview and recommended protocols. CancerResearch, 45, 2395–2399.

3. Berwald, Y., Sachs, L. (1963). In vitro cell transformation with chemical carcinogens.Nature, 200, 1182–1184.

4. LeBoeuf, R.A., Kerckaert, K.A., Aardema, M.J., Isfort, R.J. (1999). Use of Syrian hamsterembryo and BALB/c 3T3 cell transformation for assessing the carcinogenic potential ofchemicals. In: The Use of Short- and Medium-term Tests for Carcinogens and Data onGenetic Effects in Carcinogenic Hazard Evaluation (IARC Scientific Publications No.146) (McGregor, D.B., Rice, J.M., Venitt, S., Eds.). International Agency for Research onCancer: Lyon; pp. 409–425.

5. Di Mayorca, G., Greenblatt, M., Trauthen, T., Soller, A., Giordano, R. (1973). Malignanttransformation of BHK21 clone 13 cells in vitro by nitrosamines—a conditional state.Proceedings of the National Academy of Sciences U.S.A., 70, 46–49.

6. O’Hayer, K.M., Counter, C.M. (2006). A genetically defined normal human somaticcell system to study ras oncogenesis in vivo and in vitro. Methods in Enzymology, 407,637–647.

7. Sasaki, K., Bohnenberger, S., Hayashi, K., Kunkelmann, T., Muramatsu, D., Poth, A.,Sakai, A., Salovaara, S., Tanaka, N., Thomas, B.C., Umeda, M. (2012). Photo cataloguefor the classification of foci in the BALB/c 3T3 cell transformation assay. MutationResearch, 744, 42–53.

8. Sasaki, K., Mizusawa, H., Ishidate, M. (1988). Isolation and characterization ofras-transfected BALB/3T3 clone showing morphological transformation by 12-O-tetradecanoyl-phorbol-13-acetate. Japanese Journal of Cancer Research, 79, 921–930.

9. Asada, S., Sasaki, K., Tanaka, N., Takeda, K., Hayashi, M., Umeda, M. (2005). Detectionof initiating as well as promoting activity of chemicals by a novel cell transformationassay using v-Ha-ras-transfected BALB/c 3T3 cells (Bhas 42 cells). Mutation Research,588, 7–21.

10. Muramatsu, D., Sasaki, K., Kuroda, S., Hayashi, K., Tanaka, N., Sakai, A. (2009). Com-parison of sensitivity to arsenic compounds between a Bhas 42 cell transformation assayand a BALB/c 3T3 cell transformation assay.Mutation Research, 675, 66–70.

11. Ohmori, K., Sasaki, K., Asada, S., Tanaka, N., Umeda, M. (2004). An assay method forthe prediction of tumor promoting potential of chemicals by the use of Bhas 42 cells.Mutation Research, 557, 191–202.

12. Sakai, A., Sasaki, K., Muramatsu, D., Arai, S., Endou, N., Kuroda, S., Hayashi, K.,Lim, Y.M., Yamazaki, S., Umeda, M., Tanaka, N. (2010). A Bhas 42 cell transformationassay on 98 chemicals: the characteristics and performance for the prediction of chemicalcarcinogenicity. Mutation Research, 702, 100–122.

13. Ohmori, K., Umeda, M., Tanaka, N., Takagi, H., Yoshimura, I., Sasaki, K., Asasda, S.,Sakai, A., Araki, H., Asakura, M., Baba, H., Fushiwaki, Y., Hamada, S., Kitou, N.,Nakamura, T., Nakamura, Y., Oishi, H., Sasaki, S., Shimada, S., Tsuchiya, T., Uno, Y.,Washizuka, M., Yajima, S., Yamamoto, Y., Yamamura, E., Yatsushiro, T.; Non-GenotoxicCarcinogen StudyGroup in the EnvironmentalMutagen Society of Japan (2005). An inter-laboratory collaborative study by the Non-Genotoxic Carcinogen Study Group in Japan,on a cell transformation assay for tumour promoters using Bhas 42 cells. Alternatives toLaboratory Animals, 33, 619–639.


14. Tanaka, N., Sasaki, K., Hayashi, K., Sakai, A., Asada, S., Muramatsu, D., Kudora, S.,Mizuhashi, F., Nagai, M., Suzuki, H., Imamura, T., Asakura, M., Satoh, H., Sakamoto,A., Nakao, R., Hirose, H., Ishii, N., Umeda, M. (2009). An interlaboratory collaborativestudy on a cell transformation assay using Bhas 42 cells. Alternatives to Animal Testingand Experimentation, 14, 831–848.

15. Sakai, A., Sasaki, K., Hayashi, K., Muramatsu, D., Arai, S., Endou, N., Kuroda, S.,Albrecht, P., Bohnenberger, S., Kunkelmann, T., Asakura, M., Hirose, H., Ishii, N.,Mizuhashi, F., Kasamotoe, S., Nagai, M., Pant, K., Bruce, S.W., Sly, J.E., Yamazaki,S., Umeda, M., Tanaka, N. (2011). An international validation study of a Bhas 42 celltransformation assay for the prediction of chemical carcinogenicity. Mutation Research,725, 57–77.

16. Ohmori, K. (2009). In vitro assays for the prediction of tumorigenic potential of non-genotoxic carcinogens. Journal of Health Science, 55, 20–30.

17. Pirollo, K.F., Tong, Y.A., Villegas, Z., Chen, Y., Chang, E.H. (1993). Oncogene-transformed NIH 3T3 cells display radiation resistance levels indicative of a signal trans-duction pathway leading to the radiation-resistant phenotype. Radiation Research, 135,234–243.

18. Raaphorst, G.P., Vadasz, J.A., Azzam, E.I., Sargent, M.D., Borsa, J., Einspenner, M.(1985). Comparison of heat and/or radiation sensitivity and membrane composition ofseven X-ray-transformed C3H 10T1/2 cell lines and normal C3H 10T1/2 cells. CancerResearch, 45, 5452–5456.

19. Hudziak, R.M., Lewis, G.D., Holmes, W.E., Ullrich, A., Shepard, H.M. (1990). Selectionfor transformation and met protooncogene amplification in NIH 3T3 fibroblasts usingtumor necrosis factor a. Cell Growth & Differentiation, 1, 129–134.

20. Woods, K.M., Fattaey, H., Johnson, T.C., Chapes, S.K. (1994). Selective cytotoxicity oftransformed cells but not normal cells by a sialoglycopeptide growth regulator in the pres-ence of tumor necrosis factor. Biochemical and Biophysical Research Communications,205, 215–220.

21. Yamasaki, H., Katoh, F. (1988). Novel method for selective killing of transformed rodentcells through intercellular communication, with possible therapeutic applications. CancerResearch, 48, 3203–3207.

22. Chow, M., Rubin, H. (1998). Selective killing of preneoplastic and neoplastic cells bymethotrexate with leucovorin. Proceedings of the National Academy of Sciences U.S.A.,95, 4550–4555.

23. Greenberg, A.H., Egan, S.E., Jarolim, L., Wright, J.A. (1987). NK sensitivity of H-rastransfected fibroblasts is transformation- independent. Cellular Immunology, 109, 444–450.

24. Cook, J.L., May, D.L., Wilson, B.A., Walker, T.A. (1989). Differential induction ofcytolytic susceptibility by E1A, myc, and ras oncogenes in immortalized cells. Journal ofVirology, 63, 3408–3415.

25. Ishiyama, M., Miyazono, Y., Sasamoto, K., Ohkura, Y., Ueno, K. (1997). A highly water-soluble disulfonated tetrazolium salt as a chromogenic indicator for NADH as well as cellviability. Talanta, 44, 1299–1305.

26. Fields, R.D., Lancaster, M.V. (1993). Dual-attribute continuous monitoring of cell prolif-eration/cytotoxicity. American Biotechnology Laboratory, 11, 48–50.

27. Sasaki, K., Muramatsu, D., Arai, S., Endou, N., Sakai, A., Umeda, M., Tanaka,N. (2012). Spectrophotometric quantitative transformation assay using a cell viability

REFERENCES 339

chromogenic reagent followed by selective killing of normal contact-inhibited cells withhydrogen peroxide in Bhas 42 cells. Manuscript submitted.

28. Arai, S., Tanaka, N., Sasaki, K., Sakai, A. (2010). A study on the dose setting of testchemicals for the promotion assay in Bhas 42 cell transformation assay. Alternatives toAnimal Testing and Experimentation, 15, 6–13.

29. Wang, J., Yi, J. (2008). Cancer cell killing via ROS: to increase or decrease, that is thequestion. Cancer Biology & Therapy, 7, 1875–1884.

30. Urani, C., Stefanini, F.M., Bussinelli, L., Melchioretto, P., Crosta, G.F. (2009). Imageanalysis and automatic classification of transformed foci. Journal of Microscopy, 234,269–279.

31. Procaccianti, C., Stefanini, F.M., Urani, C. (2011). The cell transformation assay: toward astatistical classification of mixed and intermediate foci images. Alternatives to LaboratoryAnimals, 39, 23–36.

32. Poth, A., Heppenheimer, A., Bohnenberger, S. (2007). Bhas42 cell transformation assayas a predictor of carcinogenicity. Alternatives to Animal Testing and Experimentation, 14,519–521.

33. Creton, S., Aardema, M.J., Carmichael, P.L., Harvey, J.S., Martin, F.L., Newbold, R.F.,O’Donovan, M.R., Pant, K., Poth, A., Sakai, A., Sasaki, K., Scott, A.D., Schechtman,L.M., Shen, R.R., Tanaka, N., Yasaei, H. (2012). Cell transformation assays for predictionof carcinogenic potential: state of the science and future research needs.Mutagenesis, 27,93–101.


PART IV

HIGH-THROUGHPUT ASSAYS TOASSESS REPRODUCTIVE TOXICITY,CARDIOTOXICITY, ANDHAEMATOTOXICITY


18ReProGlo: A NEW STEM-CELL-BASEDHIGH-THROUGHPUT ASSAY TOPREDICT THE EMBRYOTOXICPOTENTIAL OF CHEMICALS

Frederik Uibel and Michael Schwarz

18.1 INTRODUCTION

The European Union Regulation for Registration, Evaluation Authorization andRestriction of Chemicals (REACH) is aimed at collecting and assessing safetydata for all chemicals manufactured or marketed in the EU at a quantity of morethan one ton per year. If such an information is not available, which seems tobe the case for most chemicals, it has to be acquired, usually by animal tests,unless waiving is possible. In 2009, Rovida and Hartung [1] performed an eval-uation of animal numbers and costs required to complete the REACH program,considering an (back then) up-to-date list of already preregistered chemicals issuedby the European Chemicals Agency (ECHA). They concluded that if only invivo tests were applied, completion of REACH requirements for all these chem-icals would demand the use of 54 million animals and cost 9.5 billion euros[1]. Strikingly, testing for reproductive toxicity would make up for as much as90% of all animals used and 70% of the overall costs [1]. Taking these factsinto consideration, it becomes obvious that in vitro alternatives for the assessmentof reproductive toxicity of chemicals are urgently needed. Reproductive toxicityincludes adverse effects on fertility, implantation, and embryonic development.Whendeveloping in vitro assays, given the complexity of the mammalian reproductivecycle, it is obvious that it is necessary to target these aspects separately. Our work


343

344 ReProGlo: A NEW STEM-CELL-BASED HIGH-THROUGHPUT ASSAY

focuses on embryonic development and tries to predict embryotoxic effects of chem-ical agents. Well-established in vitro assays in this area are the rat whole embryoculture (WEC) as well as the rat limb bud micromass test (MM), both of whichuse pregnant animals to obtain primary embryonic cells (MM) or whole embryos(WEC) [2]. The only completely cell-culture-based method available so far, however,is the embryonic stem cell test (EST), developed by Horst Spielmann and colleagues[3]. It uses murine embryonic stem cells (ESCs), which upon leading them througha defined culturing protocol, differentiate into contracting cardiomyocytes (amongother cell types). The number of contracting embryonic bodies is used as an end point,and chemicals that interfere with cardiomyocyte differentiation are classified withthe help of a biostatistical prediction model as not, weakly, or strongly embryotoxic.An additional cytotoxicity assessment is also performed with the ESCs and with anadditional fibroblast cell line, contributing to the prediction model. Like the WECand MM, the EST was formally validated by ECVAM [4]. Disadvantages of the ESTinclude the relatively long duration of the assay (10 days) as well as the bias-sensitiveend point (microscopic analysis of contracting embryonic bodies). Furthermore, itssensitivity to unspecific cytotoxicity and its experimental complexity so far impedeshigh-throughput analysis and automation.Early embryonic development is stringently regulated by several signaling path-

ways, which are evolutionary conserved over a broad range of species [5]. Distur-bances in any of these pathways, for example, mediated by genetic manipulation, areknown to be associated with malformations. The idea was therefore to develop anassay system, which is able to assay pathway alterations as end points for predictionof teratogenicity. These alterations (up- or downregulation) can easily be measuredby using pathway-specific reporter plasmids. We decided to use murine ESCs as amodel system, mainly in order to stay close to the embryonic situation, but in prin-ciple every cell line could be used, given that it shows activity in the pathway(s) ofinterest. We used the ESCs in their undifferentiated state, an approach that differsfrom the EST.We chose the Wnt/�-catenin signaling pathway as a starting point, reasoned by

in-depth experience with the methodology required to measure the activity of thispathway in our laboratory. The relevance of this pathway for embryonic developmenthas been intensively studied. It is evolutionarily highly conserved and involved inprocesses like body axis formation [6] and cardiac differentiation [7]. Knockdownof genes that encode for signaling molecules within the Wnt/�-catenin cascade isassociated with the induction of malformations, as the names of some componentsof the pathway indicate (e.g., “Wingless,” “Dickkopf,” and “Frizzled” [8]).

18.2 ESTABLISHING THE ReProGlo ASSAY PROTOCOL

A Wnt/�-catenin-responsive luciferase reporter vector (SuperTopFlash) was trans-ferred into murine ESCs together with a neomycin/geneticin resistance vectorby lipofection to enable selection of transfected cells. The SuperTopFlash vectorcontains seven repeats of the TCF/LEF binding site, the enhancer element for the

ESTABLISHING THE ReProGlo ASSAY PROTOCOL 345

Seeding of ES cellson 96-well plates

Determination of cell viabilityand reporter activity

Addition ofAlamar Blue

Test chemical

0 24 46 48 [h]

FIGURE 18.1 ReProGlo assay protocol.

corresponding transcription factor downstream of theWnt/ �-catenin pathway, whichcontrol the expression of a firefly luciferase gene. Several clones were isolated andtested for inducibility using lithium chloride as positive control. Lithium inhibitsglycogen synthase kinase-� (GSK-�), which is an established negative regulatorwithin the Wnt/�-catenin pathway. The clone showing the highest dynamic range(#62) was used as the cell line for the establishment of the assay.We developed a very simple test protocol, consisting of the three steps seeding,

treatment, andmeasurement, stretched over 48 h. To allowmedium to high throughputand automation, the assay is performed on 96-well plates. An overview of the standardtest protocol of the ReProGlo assay is given in Figure 18.1. In detail, cells are seededon day 0 at a concentration of 10.000 cells/well and then allowed to attach to theplate for 24 h. On day 1, test chemicals are added to the cells. For this purpose,serial dilutions of the test chemical are prepared, which are then mixed with freshcell culture medium. This ensures equal amounts of solvent for each concentrationof the chemical. Concentration of the solvent, especially when using DMSO orethanol, should be kept as low as possible (see below for details). After preparationof the treatment medium, a complete medium change is performed. The cells areincubated with the test chemical for 24 h. We test two chemicals per plate, each ineight different concentrations. We include a medium control, solvent control, positivecontrol (15mMLiCl) andwells without cells for backgroundmeasurements. For eachconcentration or control we test four technical replicates (wells). On day 2 of theassay, cytotoxicity assessment and immediately afterward determination of pathwayactivity via luciferase measurement is performed. The cytotoxicity assessment isdone by using the noninvasive alamar blue assay, which is based on the reductionof resazurin to resorufin, the latter having fluorescent properties. Cells have to beincubated for 1 to 2 h with the reagent. After measuring the fluorescence, cells arelysed for the subsequent luciferase assay. The detection of both end points in thesame well gives us the possibility to normalize luciferase activity to the alamar bluesignal, which could compensate for possible seeding errors or varying cell growthacross the microtiter plate.For each chemical, the ReProGlo assay gives two readouts: cytotoxicity and path-

way activity, each relative to the solvent control. Regarding pathway activity, one


would then by default calculate EC50 values from the dose–response curves. How-ever, in our experiments we usually do not get sigmoidal dose–response curves thatshow saturation at higher concentration ranges, which would be necessary for thecalculation of the EC50 values. An explanation for this phenomenon could be that theconcentrations tested simply are not high enough, so that instead of the EC50 values,we calculate benchmark concentrations from the respective dose–response curves.These values are defined as a doubling or bisection in signal strength, motivated bythe idea that a variation of pathway activity by the factor 2 should result in a biolog-ically significant effect. The benchmark values are termed BMC2 for upregulationand BMC0.5 for downregulation of the signal. If cytotoxicity occurs in the testedconcentration range, we determine the value for 80% cell viability from the alamarblue dose–response curves (termed BMC0.8). This enables the ReProGlo assay todiscriminate between specific effects on the signaling pathway (BMC2 or BMC0.5 atlower concentrations than BMC0.8), which may hint at specific effects of the chemicalon the signaling pathway, and unspecific effects (BMC0.8 at a lower concentrationthan BMC2 or BMC0.5), which would hint at a more general cytotoxic mode of actionof the test chemical at this concentration.Once the assay procedure was established, we started testing a set of chemicals.

These included the pathway activator and known teratogen lithium chloride, thenegative control sodium chloride, as well as the solvents used in the assay, DMSO andethanol.As substanceswith known teratogenic activity in vivo, we chose hydroxyurea,ochratoxin A, papaverine hydrochloride, all-trans retinoic acid, salicylic acid, andthalidomide. Furthermore, in a second step, we were able to acquire the embryotoxiccompound valproic acid (VPA) together with five different derivatives, each with adistinct embryotoxic activity in vivo [9].Independent experts have classified some of the chemicals we chose before as part

of the ECVAM validation study for the EST [10]. Thereby, the chemicals were sortedin different categories, namely, in those being strongly embryotoxic (hydroxyurea andall-trans retinoic acid) or weakly embryotoxic (lithium chloride, salicylic acid, VPA).In another part of the ReProTect project, coordinated by George Daston (Procter &Gamble), a list of 63 additional chemicals was compiled, including among othersochratoxin A, which was classified as teratogenic in all species, as well as papaverineand thalidomide, which were classified as teratogenic in some species.The results of the first experiments, summarized in Table 18.1, were promising.

We were able to detect inducing (LiCl, all-trans retinoic acid, hydroxyurea) as well asinhibiting effects (papaverine, ochratoxin A) on the Wnt/�-catenin pathway. Sodiumchloride did not show any effect on the pathway, whereas the solvents DMSO andethanol inhibited signaling activity. The latter was only effective at very high con-centrations (BMC0.5 of 147 mM, which translates to 8.8% ethanol in the assay; themaximum concentration of ethanol, when used as a solvent, in the ReProGlo assaywas 0.5%). For DMSO we determined a BMC0.5 of 50.6 mM, which relates to 0.36%DMSO in the assay and which is well above the 0.0625% we had been using at thispoint (and still above the 0.2% we had to use later for some highly lipophilic agents).Lithium chloride showed a clear effect on the pathway with no cytotoxicity in thetested dose range, as did all-trans retinoic acid.Hydroxyurea and ochratoxinA showed

TA

BL

E18

.1R

ePro

Glo

Test

Che

mic

als

wit

hB

ench

mar

kC

once

ntra

tion

s(B

MC

s)

Name

BMC0.8(mM)a

BMC2(mM)b

BMC0.5(mM)c

Lithiumchloride

Noeffect

5.78

±0.803

Noeffect

Ethanol

Noeffect

Noeffect

147.0

±80.7

Dimethylsulfoxide

Noeffect

Noeffect

50.6

±10.4

Sodium

chloride

Noeffect

Noeffect

Noeffect

Hydroxyurea

0.333

±0.124

0.0422

±0.0139

Noeffect

OchratoxinA

0.0518

±0.0212

Noeffect

0.0270

±0.0086

Papaverinehydrochloride

0.00165

±0.00158

Noeffect

0.0702

±0.0462

Retinoicacid,all-trans

Noeffect

5.11

×10

−6±6.28

×10

−6Noeffect

Salicylicacid

Noeffect

Noeffect

Noeffect

Thalidomide

Noeffect

Noeffect

Noeffect

Valproicacid

Noeffect

0.205

±0.0703

Noeffect

(+)-2-Ethyl-4-methyl-pentanoicacid(a)

Noeffect

Noeffect

Noeffect

(R)-2-propyl-4-pentynoicacid(b)

Noeffect

Noeffect

Noeffect

(R)-2-pentyl-4-pentynoicacid(c)

Noeffect

Noeffect

1.05

(S)-2-propyl-4-pentynoicacid(d)

Noeffect

0.277

±0.174

Noeffect

(S)-2-pentyl-4-pentynoicacid(e)

0.828

0.0901

±0.0471

Noeffect

Cyclophosphamide(–activation)

Noeffect

Noeffect

Noeffect

Cyclophosphamide(+

activation)

1.27

0.0122

±0.0033

Noeffect

aEffectiveconcentrationcausinga20%reductionincellviability.

bEffectiveconcentrationcausingatwofoldincreaseinreporteractivity.

c Effectiveconcentrationcausingatwofolddecreaseinreporteractivity.

347


effects on the pathway and additional cytotoxicity, but the specific effects occurredat a much lower concentration than the unspecific ones. Papaverine hydrochlorideshowed unspecific cytotoxic effects well below the effects on the pathway, so we con-sider these to be unspecific as well. Thalidomide and salicylic acid did not show anyeffects in the ReProGlo assay and thus represent false-negatives. We did not get anyfalse-positives. So, at this point, if we leave out the two solvents, the ReProGlo assaydetected four out of the seven known embryotoxicants. One embryotoxic substance(papaverine) was detected, but the effects were rated as unspecific. Two substances,thalidomide and salicylic acid, were false-negatives in the ReProGlo assay. The resultfor thalidomide was not surprising, since it is known that it acts in a species-specificmanner and shows no effects in mice [11]. Salicylic acid is considered to be a weakembryotoxicant; the effective embryotoxic dose being close to the dose produc-ing maternal toxicity [10]. Salicylic acid may act by inhibiting histone deacetylases(HDACs), whereby there is only one study available on the topic. The IC50 for HDACinhibition was reported to be almost 20-fold higher than that of VPA [12]. Therefore,we believe that salicylic acid is too weak a teratogen to be detected in the ReProGloassay.The second set of chemicals, VPA and five of its derivatives, gave us the possibility

to investigate the ability of the ReProGlo assay to rank the activity of structuralanalogues in the assay, which show well-characterized differences in teratogenicpotency in mice [9]. The results for the six chemicals are shown in Figure 18.2. VPAproduced a clear induction of signaling activity, with a BMC2 of around 0.2 mM. Thetwo analogues rated with a higher teratogenic potency than VPA (compounds d and e)also showed a strong specific effect, BMC2 being in approximately the same range asVPA (0.277 mM, compound d) or much lower (0.09 mM, compound e). Compound a,which is nonteratogenic, was also negative in the ReProGlo assay. The two remainingchemicals showed no effect on Wnt/�-catenin signaling in the concentration rangetested. One of them (compound c) showed the same teratogenic potency as VPA inmice, while compound b only showed a very low teratogenic potency. The teratogeniceffect of VPA and its derivatives is probably mediated via inhibition of HDACs [13].The IC50 of HDAC inhibition of the falsely negative compound c is approximately twotimes higher than that of VPA (see Fig. 18.2 and [9]), a fact which was obviously notimportant in the classification of in vivo teratogenic potency of the compound, butmayplay a role for classification of the chemicals in the ReProGlo assay. Unfortunately,no HDAC inhibition data is available for compounds b and d.

18.3 PARTICIPATION IN THE ReProTect FEASIBILITY STUDY

In 2009, the ReProGlo assay was part of the ReProTect feasibility study [14]. Theeffect of a panel of 10 blinded test chemicals on the Wnt pathway activity wasassessed, the panel consisting of seven known embryotoxicants as well as threeknown nonembryotoxic chemicals. Upon completion of the study and subsequentunblinding of the chemicals the study showed that the ReProGlo assay classified allthree nonembryotoxic chemicals correctly, so that again there were no false-positives.

PARTICIPATION IN THE ReProTect FEASIBILITY STUDY 349

35(a)

30 VPA

a

b

c

e

d

VPA

a

b

c

e

(S)-2-propyl-4-pentynoic acid

(S)-2-pentyl-4-pentynoic acid

(R)-2-pentyl-4-pentynoic acid

COOH

COOH

COOH

COOH

COOH

COOH

CH2

CH3

H3C

CH3

H3C

H3C

H3C

H

H

HC

HHC

CH3

CH3

CHH

CH

(R)-2-propyl-4-pentynoic acid

(±)-2-ethyl-4-methyl-pentanoic acid

Valproic acid/-2-propyl-pentanoic

acid

d

25

20

15

10

5

Rep

orte

r ac

tivity

(fo

ld in

duct

ion)

0

0.0

(b)

Structure Compound Name Abbr.Teratogenic

potentialin vivo

0.2 0.4Concentration (mM)

0.6 0.8

3+

0

+

3+

5+

5+

1.0

FIGURE 18.2 Effect of VPA and derivatives thereof on pathway activity in the ReProGloassay. (a) Reporter activity as a function of concentration of VPA and derivatives a–e, relativeto untreated control. (b) Structures and teratogenic potential in vivo of VPA and derivatives(for reference see Reference 9, for comparison see also data given in Table 18.1; reproducedwith permission from Elsevier).

However, it detected only three out of the seven known embryotoxicants. False-negatives included glufosinate ammonium, nitrofen, vinclozolin and bisphenol A.For glufosinate, the false-negative outcome can be explained with the mode of actionof the chemical: glufosinate inhibits glutamine synthase, which leads to the depletionof glutamine and is detrimental during the blastula phase of early development. Inthe ReProGlo assay, the cell culture medium is supplemented with glutamine, whichis necessary for the growth of the ESCs; glutamine in the medium thus masks a


possible effect of glufosinate in the assay. Vinclozolin and bisphenol A both act bybinding to the androgen and the estrogen receptors (AR/ER), respectively, effectsthat were also detected in cell-based reporter assays that took part in the ReProTectfeasibility study [14]. The inability of the ReProGlo assay to detect effects mediatedthrough these receptor interactions delivered important information for the definitionof its applicability domain. Nitrofen, negative in the ReProGlo assay, also showed aresponse in someof theReProTect assays that targetARaffinity; it has been speculatedthat its teratogenicity is mediated via interference with retinoic acid signaling [15].However, since all-trans retinoic acid was positive in the ReProGlo assay at verylow concentrations, the above-mentioned mechanism of action of nitrofen, if valid,is apparently not active in our assay system. Compared with the ReProGlo assay, theEST performed better in the ReProTect feasibility study. It detected six out of theseven embryotoxic chemicals; the only false-negative was glufosinate, very likely forthe same reason as mentioned above for the ReProGlo assay.

18.4 AUTOMATION OF THE ASSAY

Once the functionality of the assay was proven, we wanted to adapt it to an automatedplatform for medium- to high-throughput analyses, which was made possible byparticipation in the European Union project ChemScreen [www.chemscreen.eu]. Itsoon became clear that we would continue to perform the routine culturing as wellas seeding of the cells manually, since automation of these steps would be too cost-intensive. Preparation of the treatment medium and medium change in the 96-wellplates, however, can easily bemanaged by any common liquid handling platformwithan integrated robotic arm and an incubator. On day 2, the automated platform canassist in adding the alamar blue reagent, lysing the cells, and transferring the lysateto a measurement plate. Most manufacturers also offer the possibility to integrate aplate reader into the liquid handling system, so that no user interaction is required totransfer the plates. Again, for cost reasons, the liquid handling system of our choice(Tecan) was not equipped in such a way to offer this possibility. The experiencewith the system as we run it in our laboratory showed that it was neither faster normore accurate than if we would perform the experiment manually. Nevertheless, itsacquisition pays off, since it saves a lot of manual operation time, which can be usedotherwise. In our current setup, our throughput is 16 plates per week, each with 2test chemicals, which gives a total of 32 experiments per week. We usually performat least 4 experiments per chemical, which allows us to complete a maximum of 16chemicals in 2 weeks. This number could be expanded by using a more sophisticatedand technically more advanced liquid handling system.

18.5 INTEGRATION OF A METABOLIC ACTIVATION SYSTEM

Some compounds require metabolic activation to become teratogenic or otherwisetoxic. These so-called pro-teratogens are normally not detected in cell-culture-based

INTEGRATION OF A METABOLIC ACTIVATION SYSTEM 351

FIGURE 18.3 Metabolic activation of CPA by preincubation with primary hepatocytes(method i, see text). (a) STF-luc reporter activity as a function of concentration of CPA plus(squares) or minus (triangles) metabolic activation relative to untreated control. (b) Effectof CPA on cell viability plus (squares) or minus (triangles) metabolic activation relative tountreated control. Reproduced with permission from Elsevier.

test systems due to the inadequate metabolic capacity of permanent cell lines. There-fore, the integration of a metabolic activation system would be a major advantage forthe ReProGlo assay. In principle, such a system could consist of the so-called S9 mix,hepatic microsomes, or primary hepatocytes. We decided to use primary mouse hepa-tocytes in three different technical approaches: (i) for preconditioning of the treatmentmedium, (ii) for coculturing with ESCs, and (iii) in trans-well plates together withESCs. We performed the proof-of-principle experiments with cyclophosphamide(CPA), a well-studied anticancer drug, which is metabolized by cytochrome P450enzymes in hepatocytes preferentially to alkylating intermediates [16]. For the firstapproach, we incubated primary hepatocytes with different concentrations of CPAfor 6 h, then took the supernatant and tested it in the ReProGlo assay. The sameconcentrations of CPA were also tested in our conventional ReProGlo assay withoutmetabolic activation. The results of these experiments are summarized in Figure 18.3.Metabolically activated CPA led to a clear induction of the Wnt signaling pathway,while CPA without activation showed no effect. There was also a slight decrease incell viability, when CPA was activated. The second method, coculturing of primaryhepatocytes with ESCs, was performed in 24-well plates and showed basically thesame results, although the relative induction was lower than with the first method.The third method used trans-well plates with two different compartments for the twocell types, which are connected via a membrane to enable transfer of metabolites.This approach also worked, and the induction of the pathway was roughly in the samerange as that observed when using the coculturing method. The trans-well methodmay be the most elegant one, but also the most cost-intensive and laborious one, sothat it was not further used after the first experiments. After the promising resultswith CPA, we tested more pro-teratogens: valpromide (precursor of VPA), retinol(precursor of all-trans retinoic acid), and methoxyethanol (precursor of methoxy


acetic acid). Unfortunately, none of these chemicals showed any influence on thepathway when incubated with the metabolic activation system by either one of thefirst two methods, while their active forms, when added directly to the ES cells, did.Further studies on this topic confirm the metabolic activation of CPA by hepatocytesin combination with the EST [17] or by hepatic microsomes in combination with therat WEC [18]. Valpromide does not seem to be activated by murine hepatocytes, butby human hepatocytes [17] and possibly by murine hepatic microsomes when addedat high concentrations, while retinol and methoxyethanol are not activated by hepaticmicrosomes [18].As the results with CPA suggest, the addition of a metabolic activation system to

the ReProGlo assay is possible. However, we also found that other pro-teratogenstested were not detected. This observation could be due to a number of causes, but themost likely one is that the reactive intermediate of CPA is rather stable (it is givento humans as a pro-drug and the metabolite, formed mostly in liver, has to reach adistant tumor). Metabolites of the other chemicals may not be formed, only formed ata very low concentration, or be unable to pass the hepatocyte membrane after beingformed. Furthermore, species differences in metabolic activation capacity betweenhumans and mice are well known for some chemicals and may explain in part theobserved negative results.

18.6 CONCLUSIONS

Our main goal was the development of a cell-culture-based in vitro assay for theprediction of teratogenic effects of chemicals. The project was motivated by the needfor new in vitro testing methods and mostly triggered by the REACH regulations. Ourstrategy was to use genetically engineered mouse ESCs stably expressing a reporterfor an essential cellular signaling pathway, the Wnt/�-catenin pathway, which, whenperturbed, is known to cause teratogenic effects in animals and humans. The format ofthe newly developed ReProGlo assaywith its 96-well plate layout allowed automationof the assay for routine medium- to high-throughput analyses. In its present formatit is fast and cost-effective. A major advantage, linked to the comparatively shortincubation period with the test compound (24 h), is that “unspecific” cytotoxicitydoes generally only play a minor role. All-trans retinoic acid, for example, was onlytoxic at very high concentrations in the ReProGlo assay upon 24-h exposure of thecells, but is toxic at a very low concentration in the EST, where it has to be incubatedfor several days [4]. In fact, “unspecific” cytotoxicity and the “specific” effect ondifferentiation show a high correlation in the EST assay [19], which is not the casein the ReProGlo assay.To define the performance of an assay system, its sensitivity and specificity are

critical determinants. By taking into account the so far rather low number of testchemicals tested, the ReProGlo assay showed a high specificity; none of the non-teratogens gave a positive result in the assay. On the other hand, it does not appearto be very sensitive; only strong teratogens were detected in the assay, and one of

REFERENCES 353

the “strong” teratogens, thalidomide, was not detected (please note that there aredifferent definitions of a “strong” teratogen in the literature; one is that the compoundis teratogenic in all species tested (see above), which is not the case for thalidomide).Some of the compounds (vinclozolin and bisphenol A) that were negative in the

ReProGlo assay are thought to be teratogenic by interfering with sex hormone signal-ing. The adverse effect of vinclozolin is assumed to be mediated by its antiandrogenicactivity. Since the AR seems to be expressed in ESCs [20], the antagonizing effectof vinclozolin on AR signaling, if active in the mouse ES cells, does not appear toaffect Wnt signaling, which is the readout of the ReProGlo assay. This is somewhatunexpected, since it is well established that there exists a bilateral crosstalk betweenthe Wnt and the AR signaling pathways [21]. Bisphenol A, active as an estrogeniccompound in vitro, was also negative in the ReProGlo assay. However, in this casethe literature on the teratogenicity of this compound in exposed animals is conflictingand unclear.The end point of the ReProGlo assay is specifically restricted to the Wnt signaling

pathway and as such very specific. One might postulate that, in order to broaden itsapplicability domain, the ReProGlo assay should be extended to a whole test batteryfeaturing several stably transfected cell lines covering other signaling pathways ofinterest. However, there exists a huge degree of crosstalk between the Wnt/�-cateninand other pathways. This has been in part proven by the outcome with some ofthe chemicals in the ReProGlo assay, which are not known to directly interact withcomponents of the Wnt pathway. Complementation with additional readouts may notnecessarily improve the test system, because of the inevitably associated increase inrisk of generating false-positives.Overall, we believe that the ReProGlo assay could develop into a suitable tool

to predict (strong) teratogenic effects of chemicals. It tends to underpredict theembryotoxicity of weakly active compounds, but it appears to be well applicable in atest battery. Its main advantages are the time-saving aspect and its cost-effectiveness.The assay is also very useful for mechanistic studies.

REFERENCES

1. Rovida, C., Hartung, T. (2009). Re-evaluation of animal numbers and costs for in vivotests to accomplish REACH legislation requirements for chemicals—a report by theTransatlantic Think Tank for Toxicology (t4). ALTEX, 26, 187–208.

2. Spielmann,H., Seiler, A., Bremer, S., Hareng, L., Hartung, T., Ahr, H., Faustman, E., Haas,U., Moffat, G.J., Nau, H., Vanparys, P., Piersma, A., Sintes, J.R., Stuart, J. (2006). Thepractical application of three validated in vitro embryotoxicity tests. ATLA, 34, 527–538.

3. Seiler, A.M., Spielmann, H. (2011). The validated embryonic stem cell test to predictembryotoxicity in vitro. Nature Protocols, 6, 961–978.

4. Genschow, E., Spielmann, H., Scholz, G., Pohl, I., Seiler, A., Clemann, N., Bremer, S.,Becker, K. (2004). Validation of the embryonic stem cell test in the International ECVAMValidation Study on three in vitro embryotoxicity tests. ATLA, 32, 209–244.


5. National Research Council (2000). Scientific Frontiers in Developmental Toxicology andRisk Assessment. National Academy Press: Washington, DC.

6. Marikawa,Y. (2006).Wnt/�Catenin signaling and body plan formation inmouse embryos.Seminars in Cell & Developmental Biology, 17, 175–184.

7. Nakamura, T., Sano, M., Songyang, Z., Schneider, M.D. (2003). A Wnt- and �-catenin-dependent pathway for mammalian cardiac myogenesis. Proceedings of the NationalAcademy of Sciences USA, 100, 5834–5839.

8. Klaus, A., Birchmeier, W. (2008). Wnt signalling and its impact on development andcancer. Nature Reviews Cancer, 8, 387–398.

9. Eikel, D., Lampen, A., Nau, H. (2006). Teratogenic effects mediated by inhibition of his-tone deacetylase: evidence from quantitative structure activity relationships of 20 valproicacid derivatives. Chemical Research in Toxicology, 19, 272–278.

10. Brown, N.A. (2002). Selection of test chemicals for the ECVAM international validationstudy on in vitro embryotoxicity tests. ATLA, 30, 177–198.

11. Newman, L.M., Johnson, E.M., Staples, R.E. (1993). Assessment of the effectivenessof animal developmental toxicity testing for human safety. Reproductive Toxicology, 7,359–390.

12. Di Renzo, F., Cappelletti, G., Broccia, M.L., Giavini, E., Menegola, E. (2008). Theinhibition of embryonic histone deacetylases as the possible mechanism accounting foraxial skeletal malformations induced by sodium salicylate. Toxicological Sciences, 104,397–404.

13. Gottlicher, M., Minucci, S., Zhu, P., Kramer, O.H., Schimpf, A., Giavara, S., Sleeman,J.P., Lo Coco, F., Nervi, C., Pelicci, P.G., Heinzel, T. (2001). Valproic acid defines a novelclass of HDAC inhibitors inducing differentiation of transformed cells. EMBO Journal,20, 6969–6978.

14. Schenk, B., Weimer, M., Bremer, S., van der Burg, B., Cortvrindt, R., Freyberger, A.,Lazzari, G., Pellizzer, C., Piersma, A., Schafer, W.R., Seiler, A., Witters, H., Schwarz,M. (2010). The ReProTect Feasibility Study, a novel comprehensive in vitro approach todetect reproductive toxicants. Reproductive Toxicology, 30, 200–218.

15. Noble, B.R., Babiuk, R.P., Clugston, R.D., Underhill, T.M., Sun, H., Kawaguchi, R.,Walfish, P.G., Blomhoff, R., Gundersen, T.E., Greer, J.J. (2007). Mechanisms of actionof the congenital diaphragmatic hernia-inducing teratogen nitrofen. American Journal ofPhysiology, 293, L1079–L1087.

16. Nau, H., Spielmann, H., Lo TurcoMortler, C.M.,Winckler, K., Riedel, L., Obe, G. (1982).Mutagenic, teratogenic and pharmacokinetic properties of cyclophosphamide and someof its deuterated derivatives. Mutation Research, 95, 105–118.

17. Hettwer, M., Reis-Fernandes, M.A., Iken, M., Ott, M., Steinberg, P., Nau, H. (2010).Metabolic activation capacity by primary hepatocytes expands the applicability of theembryonic stem cell test as alternative to experimental animal testing. Reproductive Toxi-cology, 30, 113–120.

18. Luijten, M., Verhoef, A., Westerman, A., Piersma, A.H. (2008). Application of a metab-olizing system as an adjunct to the rat whole embryo culture. Toxicology In Vitro, 22,1332–1336.

19. Marx-Stoelting, P., Adriaens, E., Ahr, H.J., Bremer, S., Garthoff, B., Gelbke, H.P., Piersma,A., Pellizzer, C., Reuter, U., Rogiers, V., Schenk, B., Schwengberg, S., Seiler, A., Spiel-mann, H., Steemans, M., Stedman, D.B., Vanparys, P., Vericat, J.A., Verwei, M., van

REFERENCES 355

der Water, F., Weimer, M., Schwarz, M. (2009). A review of the implementation of theembryonic stem cell test (EST). ATLA, 37, 1–16.

20. Lee, S.H., Heo, J.S., Lee, M.Y., Han, H.J. (2008). Effect of dihydrotestosterone on hydro-gen peroxide-induced apoptosis of mouse embryonic stem cells. Journal of CellularPhysiology, 216, 269–275.

21. Mulholland, D.J., Dedhar, S., Coetzee, G.A., Nelson, C.C. (2005). Interaction of nuclearreceptors with the Wnt/�-catenin /Tcf signaling axis: Wnt you like to know? EndocrineReviews, 26, 898–915.

19EMBRYONIC STEM CELL TEST (EST):MOLECULAR ENDPOINTS TOWARDHIGH-THROUGHPUT ANALYSIS OFCHEMICAL EMBRYOTOXICPOTENTIAL

Peter T. Theunissen, Esther de Jong, Joshua F. Robinson,and Aldert H. Piersma

19.1 INTRODUCTION

Current legislative guidelines for developmental toxicity testing of compounds andpharmaceuticals require many test animals using cumbersome and expensive testingstrategies. The embryonic stem cell test (EST) was designed as an animal-free in vitroscreening assay to determine the ability of a compound to induce developmental tox-icity. First described in 1997 by Spielmann et al. [1], the effect of compounds on thedifferentiation of pluripotent murine embryonic stem cells (mESCs) into cardiomy-ocytes of the mesodermal lineage is taken as a predictor of developmental toxicityin the EST. The original EST paradigm [2] rests on three pillars: (1) testing of cyto-toxicity in 3T3 cells (as a model for maternal toxicity); (2) testing of cytotoxicity inmESC (representing the developing organism); (3) the differentiation assay, measur-ing compound inhibition of mESC differentiation into cardiomyocytes after 10 daysof treatment. As a relatively simple assay to determine developmental toxicity of acompound, the cardiomyocyte differentiation method consists of the following steps(Fig. 19.1):mESCs are first cultured for 3 days using the “hanging dropmethod” in thepresence of a concentration range of the test compound. During this period in culture,


357

358 EMBRYONIC STEM CELL TEST (EST)

Day 0 Day 3

ESC suspensionin hanging drops

Differentiation intocardiomyocyte foci

Aggregate formation in hanging drops

Aggregate development in suspension culture

Aggregate attachmenton tissue culture plastics

Day 10Day 5

FIGURE 19.1 Schematic overview of the EST cardiac differentiation culture method. Hang-ing drops containing ∼750 mESCs each are cultured for 3 days in the presence of a concen-tration range of the test compound. During the hanging drop period, mESCs form aggregates,so-called embryoid bodies (EB). At day 3, EB are transferred to bacterial Petri dishes growingfurther in suspension medium containing the compound. On day 5, each EB is plated into onewell of a 24-well plate, where it attaches to the well surface and develops beating cardiomy-ocyte foci. Each plate is used to test a different compound concentration or control and thenumber of wells containing EB with beating cardiomyocytes is scored.

mESCs form aggregates, also known as embryoid bodies (EBs). On day 3 of culture,EBs are transferred to bacterial Petri dishes growing further in suspension mediumcontaining the test compound. On day 5, each EB is plated into one well of a 24-wellplate, where it attaches to the well surface and develops beating cardiomyocyte foci.Exposure is continued throughout this period. Each plate is used to test a differentcompound concentration or vehicle control and the number of wells containing EBwith beating cardiomyocytes is scored [1]. To validate the EST, the European Centrefor Validation of Alternative Methods (ECVAM) coordinated a blind trial validationstudy [3] similar to past ECVAMmodel validation studies [4], in which four laborato-ries of different international institutes tested 20 compounds. The compounds testeddiffered in their developmental toxic potencies in vivo, required no metabolic acti-vation, and were divided into non-embryotoxic, weakly embryotoxic, and stronglyembryotoxic. In this validation study, the EST obtained an accuracy score of 78%,with a 100% prediction of strong embryotoxic compounds, but a limited prediction ofweak and non-embryotoxic compounds (69% and 73%, respectively). Furthermore,in subsequent studies EST prediction was observed to be even less accurate [5, 6].For example, in one validation study ECVAM compounds were predicted with anaccuracy of 83%, but in the case of receptor-mediated pharmaceuticals only a 53%accuracy was achieved [6]. Recently, ECVAM organized expert workshops to deter-mine the applicability domain of the EST, and to discuss improvements necessary toadvance predictivity of the EST [5, 7]. Recommendations to improve the predictivityof the EST included: (1) implementation of objective molecular endpoints; (2) addi-tion of differentiation into other lineages (e.g., neural and osteoblast lineages); (3)development of a metabolic activation system [5, 7]. In the following years and in aneffort to develop an objective measurement of cell differentiation in the EST, manygroups developed molecular-based assays, ranging from single molecular endpointassessments using real-time polymerase chain reaction (RT-PCR) [8, 9], fluorescence-activated cell sorting (FACS) [10] or built-in reporter genes [11] to global assessment

FOCUSED MOLECULAR-BASED ENDPOINTS IN THE EST 359

strategies using omics technologies, such as transcriptomics [12, 13] and proteomics[14]. In addition, systems similar to the EST, but with different lineages, includingneural [15, 16] and osteoblast lineages [17], have been developed to study the effectsof compounds in a broader spectrum of cell differentiation pathways. Furthermore,the advantages of human embryonic stem cell (hESC) in the EST are being assessed[18, 19] to avoid the necessity of extrapolation between species in view of human riskassessment. In the following sections, wewill discuss the progress in the improvementof the EST by implementation of molecular endpoints and in vivo/in vitro correlationstudies.

19.2 THE IMPLEMENTATION AND ASSESSMENT OF FOCUSEDMOLECULAR-BASED ENDPOINTS IN THE EST

Assessment of compound toxicity in the classic EST is built upon the visual micro-scopic assessment of the presence of beating cardiomyocytes as an endpoint ofdifferentiation. This method has been criticized to be subjective and not completelyaccurate [5] as it does not distinguish between EBs containing few versus manycardiomyocytes. One of the first technologies used to quantify differentiation witha limited number of molecular markers is RT-PCR. Several genes were observedto be suitable markers for cardiomyocyte differentiation in the EST, including theexpression of brachyury (T), a transcription factor regulating mesodermal differenti-ation; myosin heavy chain (MHC), a structural protein expressed in cardiac muscle;and Nkx2.5, a regulator of myocardin expression. Using these markers describingstages of cardiomyocyte differentiation, the embryotoxic effects of multiple com-pounds, including retinoic acid, lithium chloride, 5-fluorouracil, diphenylhydantoin,and valproic acid, could be identified [8, 9, 20]. RT-PCR has also been used to studythe effect of chemicals on additional differentiation route endpoints. Markers usedin these studies included aggrecan for chondrocyte differentiation, osteocalcin forosteoblast differentiation and neurofilament, �III-tubulin and map2, among others,for neural differentiation [21–23].Another method used to quantify differentiation is FACS analysis. Seiler et al.

[10] developed the FACS–EST, based on the intracellular immunostaining of twostructural proteins of the sarcomere apparatus, namely MHC and �-actinin. Thisstudy showed that the FACS–EST has the same sensitivity as the morphologic evalu-ation when testing two strong embryotoxic compounds, 5-fluorouracil and all-transretinoic acid, and a non-embryotoxic compound, penicillin G, while reducing theculture duration from 10 to 7 days [10]. The FACS–EST was further validated bytesting an additional set of 10 chemicals. The FACS–EST analysis, if compared to theclassic morphological evaluation of beating cardiomyocytes, delivered almost identi-cal concentration–response curves, thus indicating that FACS evaluation can reliablyreplace the classical morphologically evaluated differentiation endpoint [24].In the ReProGlo assay mESCs were transfected with the SuperTopFlash luciferase

reporter, thereby allowing to detect drug-induced alterations in the canonical Wnt/�-catenin signaling pathway (WSP), which is involved in the regulation of early


embryonic development [11]. In this assay, cells are cultured in a monolayer andexposed to test chemicals for 22 h. Following exposure, WSP activity was quantifiedby measuring fluorescence signal for luciferase activity. Using this method 13 out ofthe 14 chemicals tested could be correctly classified as being either embryotoxic ornon-embryotoxic [11]. One of the advantages of this assay is that cell viability can bemeasured simultaneously with the classical differentiation endpoint, eliminating theneed for a separate cytotoxicity assay. Furthermore, the length of the assay procedureis substantially reduced compared to the 10-day culture of the classic cardiac EST.The downside of the ReProGlo assay is that only the effect of chemicals on the WSPis evaluated. The WSP is known to play an essential role during embryonic develop-ment and is involved in cell proliferation, migration, cell fate decisions, and patternformations of organs in vertebrates [25]. However, it remains to be seen if an assaythat solely depends on the WSP is sufficient to detect a wide range of embryotoxiccompounds with very different modes of action. In order to be able to make a defini-tive statement regarding the applicability domain of the ReProGlo assay, additionalchemicals will have to be evaluated.Similar to the ReProGlo assay, the adherent cell differentiation and cytotoxic-

ity (ACDC) assay is used to simultaneously measure the effects of compounds ondifferentiation and cytotoxicity [26]. Cells are exposed to compounds as a mono-layer in a 96-well plate format. On day 9 of culture, the effect on differentiationis assessed by the intensity of the immunofluorescent staining for MHC. In thesame well, the intensity values for the DNA/cell stain mix (DRAQ5/Sapphire700) todetect cytotoxic effects of chemicals are measured. DRAQ5 is a stain for living cells,while Sapphire700 is a nuclear and cytoplasmic cell stain for dead cells. With thismethod, the relative potency of acetic acid, 5-fluorouracil, and bromochloroaceticacid were determined, thereby obtaining similar concentration–response curves aswith the classical EST (i.e., via the morphological assessment of beating cardiomy-ocytes [26]. The ACDC assay was included in the ToxCast project, in which 309chemicals were evaluated in approximately 500 cell-free or cell-based assays. Intotal, 56 out of the 309 chemicals disrupted either cell viability or embryonic stemcell (ESC) differentiation [27]. The results of the ACDC assay were correlated withthose of the other 500 assays used in this setup as well as with whole animal tox-icity data. The increased cytotoxicity endpoint mainly correlated with assays thatrevealed disruption of cell–cell signaling and features such as apoptosis or DNAdamage [27]. The cytotoxicity endpoint also associated with rat and rabbit in vivoendpoints, including delayed pubertal development in multigenerational rat studiesand cranial malformations in prenatal rabbit studies. The decreased differentiationendpoint also correlated with a number of general cytotoxicity assays. In addition,there was a strong association with several transcriptional activity assays. Theseincluded mainly genes involved in differentiation, such as Pax6 and Oct1. How-ever, when results were correlated to unique in vivo endpoints the ESC differen-tiation only associated with renal defects in rat and mice developmental studies.One of the greatest limitations of this study is that the chemicals were only testedup to a maximum concentration of 12.5 �M, due to restrictions in dimethyl sul-foxide (vehicle) concentration. Therefore, chemicals biologically active at higher

IMPLEMENTING GLOBAL MOLECULAR-BASED ENDPOINTS IN THE EST 361

concentrations were not detected. The correlation between the ESC differentiationassay and developmental toxicity endpoints in vivo may improve if a wider concen-tration range is used.

19.3 IMPLEMENTING GLOBAL MOLECULAR-BASED ENDPOINTSIN THE EST

A testing strategy with a limited number of molecular endpoints may provide anindication of potential compound developmental toxicity. Nevertheless, predefiningsuch a gene or protein set on the basis of available data will likely miss importantaspects. However, after a global array analysis, it may be possible to identify a smallset of biomarkers covering all or most of the mechanisms behind compound-induceddevelopmental toxicity. Omics technologies, such as transcriptomics and proteomics,allow the study of expression of the whole genome and proteome. These technologiescan be used to identify a small set of biomarkers associated with developmentaltoxicity in the EST and to describe the mechanisms involved in the test system.Omics technologies, especially transcriptomics, have been extensively used to studyboth differentiation of ESC [12, 28] as well as compound effects on differentiationin the EST [13, 29–31].Several studies have characterized the dynamics of the transcriptome underlying

ESC differentiation. Extensive time-series studies investigated early gene expression(from day 3 onward, after 0, 24, 48, 72 and 96 h) in an adapted version of the cardiacEST (ESTc) and in a neural variant of the EST (ESTn) by using transcriptomics[28, 32, 33]. In both models, gene expression clusters transiently changing over timecould be separated, resulting in gene expression profiles describing downregulatedearly pluripotency-related genes, peaking mid early development-related genes (e.g.,homeobox genes) and upregulated late development-related genes (e.g., Myh6 inESTc and Tubb3 in ESTn) over time. The “van Dartel gene set” was introducedto describe 43 genes regulated on days 3 to 4, which are involved in early cardiacdifferentiation in the ESTc [33]. Further comparison of gene expression profilesover time between both ESTc and ESTn demonstrated common and unique geneexpression and pathway enrichment profiles between the two models [Theunissen etal., manuscript submitted].Initial studies examining transcriptomic changes over time in culture in both ESTc

and ESTn have resulted in a standardized protocol for toxicogenomic assessments.In contrast to the original EST protocol, which starts exposure on day 0, compoundexposurewas started at day 3 of the differentiation protocol, to focus on transcriptomiceffects on differentiation rather than proliferation, as was described in an earlier study[34]. Time-course studies early in the EST indicated that the variation between controlsamples increased after 24 h treatment from day 3 onwardwith an increasing variationat 48, 72, and 96 h treatment [12, 28]. In addition, it was observed that changes in geneexpression before 24 h (after 6 and 12 h) were relatively small [33]. Due to thesefindings, the 24-h-treatment time point was deemed optimal in view of statisticalpower and was used for further transcriptomics experiments.


210–1–2

–1–2

01

2

PC1

PC

2

5-FU

VPA

RA

PenG

MAA

CON 0 h

CON 24 h

CON 48 h

FIGURE 19.2 Example of the differentiation track (0–48 h) in the ESTc earlier definedby 26 genes identified as markers for developmental toxicity in the ESTc [13] (modifiedwith permission from D.A. van Dartel). Developmental toxicity is predicted based on thesignificance of deviation of 24 h exposed cultures from the differentiation track. CON 0 h,black circles; CON 24 h, dark grey circles; CON 48 h, light grey circles; black asterisk,developmentally toxic 24 h exposed group; light grey asterisk, non-developmentally toxic 24 hexposed group.

An approach to differentiate between toxic and non-toxic development-relatedcompounds was developed in the ESTc, labeled the “differentiation track” [28] (Fig.19.2). In this case genes significantly regulated across time in unexposed ESC cul-tures were first identified. Next, genes significantly and commonly altered acrosscompound-exposed samples were identified. Finally, the overlap between the twogene sets was used as a classifier. Expression of this gene subset (i.e., the classi-fier) was compared between compound-exposed cultures and time-matched controlsusing principle component analysis (PCA). Significant deviation from the normal“differentiation track” suggests that compounds induce toxicity by altering normaldifferentiation in the ESTc. Using this methodology, 26 genes regulated in the ESTccould successfully discriminate between 18 developmentally toxic and non-toxiccompounds with a prediction rate of 83% [13, 30]. Supplementary studies usingadditional compounds have been completed to validate the use of this methodologyin ESTc as well as in the ESTn [29, 31, 32; Theunissen et al., submitted]. Further-more, combined analysis of these studies in two cross-validation studies has resultedin established classifiers in the ESTc (52 genes with 83% prediction) and ESTn (29genes with 84% prediction) with promising predictive applications for developmentaltoxicity testing [35, 42].

COMPARISON OF EST TO IN VIVO DATA 363

Toxicogenomic signatures furthermore provide complex mechanistic informationregarding the diversity of compound effects. Therefore, toxicogenomics may beused to distinguish between classes of compounds based on common mechanisticeffects. In differentiating ESC [31], strong class signatures were defined between twogroups of toxicants with diverse mechanisms of action (triazoles and phthalates) inassociation with similar effects on the classical endpoint of cardiomyocyte inhibition.Therefore, by using toxicogenomics compounds with similar modes of action can beclassified separately from other compounds, even though the compound classes mayinduce similar morphological effects.In addition, toxicogenomic studies enable increased description of compound

effects across experimental factors such as dose and time. In a concentration–response transcriptomic ESTc study using six concentrations of the triazole flusila-zole, concentration–response-related effects were observed at the transcriptome levelin association with increased inhibition of cardiomyocyte differentiation [29]. Simi-lar findings were observed in a concentration–response study in the ESTn, in whichtwo triazoles and valproic acid were investigated [32]. In addition, a concentration–response ESTc study with four phthalates, mono-2-ethylhexyl phthalate, monobutylphthalate, monobenzyl phthalate, and monomethyl phthalate, showed that transcrip-tomic approaches are able to distinguish between compounds with different potencieswithin a particular compound class and correlate with the potencies of the com-pounds in vivo [43]. In the ESTc as well as the ESTn and at concentrations thatdid not impact cell viability or morphology, genes associated with early develop-ment (e.g., anterior/posterior patterning, embryonic morphogenesis) were regulatedat a lower concentration than intrinsic compound mechanisms of action (e.g., sterolbiosynthesis for triazoles) and cell death. These results suggest that the assessmentof concentration-dependent gene expression may be a more sensitive indicator ofcompound response as compared to classical endpoints of toxicity and provide sup-plementary information on the mechanism of action of a tested compound.

19.4 COMPARISON OF EST TO TRADITIONAL DEVELOPMENTALTOXICITY IN VIVO DATA

In order to evaluate the predictive capacity of the EST, several groups have performedin vitro–in vivo correlation studies to compare in vitro EST results with known invivo developmental toxicity data. In the ECVAM validation study, tested chemicalswere divided into three classes, non- embryotoxic, weakly embryotoxic, and stronglyembryotoxic [3]. This classification did not take into account that chemical potenciesspread over a continuum, is artificial and consequently may reduce the predictivecapacity of the EST. Therefore, a sole distinction between non-embryotoxic andembryotoxic chemicals or among a range of developmental toxicity potencies maybe more accurate to correlate in vitro results with in vivo data.The reliability of the EST in terms of inter- and intra-laboratory variability has

been extensively assessed [3]. However, the applicability domain, that is, the chemicalclasses and biological mechanisms for which it can make a prediction, need to be


further evaluated. A broad range of mechanisms of action can be the primary causeof developmental defects. It is deemed unlikely that the EST is capable of correctlypredicting the embryotoxicity of chemicals with all these different mechanisms ofaction. Therefore, it has been suggested that assessing the predictability of the EST forindividual chemical classes in a category approach may be more useful. This strategyallows the assessment of specific chemical classes and their biological mechanismsfor which the EST can make an accurate prediction. A category approach couldlead to a substantial reduction in animal use for developmental toxicity testing. Ifa chemical class has been correctly predicted in vitro, chemicals belonging to thesame class but with unknown embryotoxic properties could reliably be tested in theEST. In addition, using a read-across approach, a prediction can be made as to theembryotoxic potency of that compound.This category approach was used to assess if the EST could be applied to evaluate

the embryotoxicity of valproic acid analogs. Compounds within this chemical classhave been shown to cover a wide range of embryotoxic potencies, despite sharingsimilar pharmacokinetic properties. Six valproic acid analogs have been tested in theEST and compared to their known in vivo developmental toxicity [8]. This studydemonstrated that the EST provides a good prediction of compound embryotoxicitywithin this chemical class. In a similar study the effects of four glycol ether alkoxyacetic acid metabolites were evaluated [36].While it was found that the EST provideda good prediction for potency ranking of glycol ethers, the relative differences inpotency of the compounds in vitrowas almost an order of magnitude lower comparedto in vivo potency. It was suggested that differences in potency were observed dueto varying kinetic properties of these compounds, mainly consisting of differencesin the elimination rate ranging between 1.5 h and 18.6 h for the tested compounds.The influence of kinetic differences on the in vitro–in vivo correlation were furtherinvestigated by using a physiologically based kinetic (PBK) model to extrapolate invitro results to in vivo dose estimates [37]. The predicted in vivo dose estimates werein accordance with the embryotoxic dose levels found in reported in vivo studies.This clearly illustrates the importance of prior knowledge of kinetic properties ofcompounds when performing in vitro–in vivo comparisons.When testing a compoundin vitro, it is important to keep in mind whether or not the concentration tested isrelevant to the human exposure situation. Using a PBK model could overcome thisissue, elevating the EST from a tool for hazard identification to a method to identifythe risk of chemicals. This would enable the use of the EST in a risk assessmentsetting, since the estimated in vivo doses can be correlated with potential exposurelevels.

19.5 COMPARISONS ACROSS IN VITRO AND IN VIVO MODELS USINGOMIC APPROACHES

Omics assessments (e.g., transcriptomics, proteomics, metabolomics) provide a sen-sitive, robust, and common endpoint that can be compared across in vivo and in vitro

CONCLUSION 365

models, which usually use a variety of model-specific morphological endpoints todetermine developmental toxicity. In addition, cross comparisons across in vitro andin vivomodels may facilitate the identification of relevant biomarkers associated withcompound exposure and developmental toxicity and increase our understanding ofhow systems correlate in terms of response to compounds. Although limited, initialcomparative studies across models have shown the ability to match gene alterationslinked with developmental toxicity in vivo with biomarkers in vitro. In a study byKultima et al. [38], targets of valproic acid associated with effects on neural tubedevelopment in vivo were also shown to be altered in valproic-acid-treated P19 neu-roblastoma cells. Genes commonly altered in the two models represented knownneural tube defect candidate genes and pathways critical for early embryogenesis. Ina parallel transcriptomic comparison study between unexposed whole embryos cul-tures (WEC) and embryos in vivo, high similarities in gene expression changes overtime were identified in the two model systems, including significant gene expressionchanges, direction in regulation (increase or decrease), and corresponding function-ality [39], thus suggesting that WEC fairly well reflect this particular window ofin vivo development at the molecular level. Moreover, very similar gene expressionalterations were observed in all-trans-retinoic-acid-exposed WEC and embryos invivo, despite both common and unique developmental toxicity endpoints [40]. Ina larger comparative study between in vivo and in vitro model systems, includingthe EST, similarities and dissimilarities in methylmercury (MeHg)-induced tran-scriptomic response were analyzed [41]. Greater similarity was observed in terms ofresponse between mouse embryos exposed in utero (two studies), differentiating ESCin the ESTc and ESTn and WEC as compared to adult mouse liver, juvenile mousebrain, and mouse embryonic fibroblast MeHg studies. MeHg was observed to impactdevelopment-related signatures in the early development model systems (in vivo orin vitro) versus the non-traditional developmental model systems. In particular, spe-cific comparisons between mouse embryos in vivo and the ESTc showed overlapregarding alterations of specific genes related to heart development by MeHg, thussuggesting that relevant mechanisms of MeHg toxicity do in fact occur in the EST(Fig. 19.3). Future comparisons between EST and in vivo models as well as otheralternative systems using the emerging toxicogenomic databases should increase themechanistic understanding of the relevance of compound-induced responses in theEST and, therefore, the applicability domain of the EST for developmental toxicitytesting.

19.6 CONCLUSION

Over the years, a number of adaptations and additions of molecular endpoints to theEST with the aim of increasing predictability and, eventually, the throughput of thistest system have been proposed. Furthermore, these studies have provided insight intothe mechanisms behind the EST and the applicability domain of this model systemfor developmental toxicity testing. Incorporation of RT-PCR, immunofluorescent


–1.2

–0.6

0

0.6

1.2

1.20.60–0.6–1.2

MeH

g-in

duce

d F

C (

ES

Tc)

24 h

MeHg-induced FC (Embryo (in vivo)) 8 h

Gata3Tgfb2

Tbx3Nrp1Msx1

Id2Hspg2

FIGURE 19.3 Cross scatter plot comparing methylmercury-induced development-relatedgene expression alterations in mouse embryos in vivo and ESTc (modified from Reference 41).Closed triangles, genes significantly altered in both models (p < .05).

staining, FACS, and reporter genes have provided more objective endpoints whencompared to the classic subjective counting of beating cardiomyocyte foci and haveincreased the throughput of the EST. Addition of omics approaches has greatlyimproved the knowledge on mechanisms underlying cell differentiation as well asdevelopmental toxicity and may lead to further development of classifier gene listsfor developmental toxicity in the EST. Current molecular additions to the EST holdpromise for the development of high-throughput testing through the design of arrayscontaining classifier genes for developmental toxicity. In order to implement theEST including molecular endpoints in developmental toxicity testing, strategies forrisk assessment purposes have to be developed. Additional research to identify theassay’s predictability and applicability domains is needed. Comparisons betweenclassic in vivo studies and the EST, especially in potency ranking studies, will pro-vide information on the applicability domain of the EST. Omics studies comparingacross the EST as well as other in vitro and in vivo model systems will greatlyenhance our knowledge on the similarities and dissimilarities between these mod-els, which will be supportive in further determining the applicability domain of theEST and in designing testing strategies. Current and future improvements of theEST, including addition of molecular endpoints to increase predictability of the assayand definition of the applicability domain, will increase the value of the EST as awell-characterized high-throughput in vitro testing system to predict developmentaltoxicity.

REFERENCES 367

REFERENCES

1. Spielmann, H., Pohl, I., Doring, B., Liebsch, M., Moldenhauer, F. (1997). The embryonicstem cell test (EST), an in vitro embryotoxicity test using two permanent mouse cell lines:3T3 fibroblasts and embryonic stem cells. Toxicology In Vitro, 10, 119–127.

2. Scholz, G., Genschow, E., Pohl, I., Bremer, S., Paparella, M., Raabe, H., Southee, J.,Spielmann, H. (1999). Prevalidation of the embryonic stem cell test (EST)—a new in vitroembryotoxicity test. Toxicology In Vitro, 13, 675–681.

3. Genschow, E., Spielmann, H., Scholz, G., Pohl, I., Seiler, A., Clemann, N., Bremer, S.,Becker, K. (2004). Validation of the embryonic stem cell test in the international ECVAMvalidation study on three in vitro embryotoxicity tests. Alternative Laboratory Animal, 32,209–244.

4. Hartung, T., Bremer, S., Casati, S., Coecke, S., Corvi, R., Fortaner, S., Gribaldo, L.,Halder, M., Hoffmann, S., Roi, A.J., Prieto, P., Sabbioni, E., Scott, L., Worth, A., Zuang,V. (2004). A modular approach to the ECVAM principles on test validity. AlternativeLaboratory Animal, 32, 467–472.

5. Marx-Stoelting, P., Adriaens, E., Ahr, H.J., Bremer, S., Garthoff, B., Gelbke, H.P., Piersma,A., Pellizzer, C., Reuter, U., Rogiers, V., Schenk, B., Schwengberg, S., Seiler, A., Spiel-mann, H., Steemans, M., Stedman, D.B., Vanparys, P., Vericat, J.A., Verwei, M., van derWater, F., Weimer, M., Schwarz, M. (2009). A review of the implementation of the embry-onic stem cell test (EST). The report and recommendations of an ECVAM/ReProTectWorkshop. Alternative Laboratory Animal, 37, 313–328.

6. Paquette, J.A., Kumpf, S.W., Streck, R.D., Thomson, J.J., Chapin, R.E., Stedman, D.B.(2008). Assessment of the embryonic stem cell test and application and use in the phar-maceutical industry. Birth Defects Research Part B Developmental and ReproductiveToxicology, 83, 104–111.

7. Spielmann, H., Seiler, A., Bremer, S., Hareng, L., Hartung, T., Ahr, H., Faustman, E.,Haas, U., Moffat, G.J., Nau, H., Vanparys, P., Piersma, A., Sintes, J.R., Stuart, J. (2006).The practical application of three validated in vitro embryotoxicity tests. The report andrecommendations of an ECVAM/ZEBET workshop (ECVAMWorkshop 57). AlternativeLaboratory Animal, 34, 527–538.

8. de Jong, E., Doedee, A.M., Reis-Fernandes,M.A., Nau, H., Piersma, A.H. (2011). Potencyranking of valproic acid analogues as to inhibition of cardiac differentiation of embryonicstem cells in comparison to their in vivo embryotoxicity. Reproductive Toxicology, 31,375–382.

9. Pellizzer, C.,Adler, S., Corvi, R.,Hartung, T., Bremer, S. (2004).Monitoring of teratogeniceffects in vitro by analysing a selected gene expression pattern. Toxicology In Vitro, 18,325–335.

10. Seiler, A., Visan, A., Buesen, R., Genschow, E., Spielmann, H. (2004). Improvement ofan in vitro stem cell assay for developmental toxicity: the use of molecular endpoints inthe embryonic stem cell test. Reproductive Toxicology, 18, 231–240.

11. Uibel, F., Muhleisen, A., Kohle, C., Weimer, M., Stummann, T.C., Bremer, S., Schwarz,M. (2010). ReProGlo: a new stem cell-based reporter assay aimed to predict embryotoxicpotential of drugs and chemicals. Reproductive Toxicology, 30, 103–112.

12. Theunissen, P.T., Pennings, J.L., Robinson, J.F., Claessen, S.M., Kleinjans, J.C., Piersma,A.H. (2011). Time-response evaluation by transcriptomics of methylmercury effects on


neural differentiation of murine embryonic stem cells. Toxicological Sciences, 122, 437–447.

13. van Dartel, D.A., Pennings, J.L., de la Fonteyne, L.J., van Herwijnen, M.H., van Delft,J.H., van Schooten, F.J., Piersma, A.H. (2010). Monitoring developmental toxicity inthe embryonic stem cell test using differential gene expression of differentiation-relatedgenes. Toxicological Sciences, 116, 130–139.

14. Osman, A.M., van Dartel, D.A., Zwart, E., Blokland, M., Pennings, J.L., Piersma, A.H.(2010). Proteome profiling of mouse embryonic stem cells to define markers for celldifferentiation and embryotoxicity. Reproductive Toxicology, 30, 322–332.

15. Baek, D.H., An, S.Y., Park, J.H., Choi, Y., Park, K.D., Kang, J.W., Choi, K.S., Park,S.H., Whang, M.Y., Han, J., Kim, J.H., Kim, H.S., Geum, D., Yoo, T.M. (2012). Trans-ferability of a modified embryonic stem cell test using a new endpoint for developmentalneurotoxicity. Toxicology Mechanisms and Methods, 22, 118–130.

16. Theunissen, P.T., Schulpen, S.H., van Dartel, D.A., Hermsen, S.A., van Schooten, F.J.,Piersma, A.H. (2010). An abbreviated protocol for multilineage neural differentiationof murine embryonic stem cells and its perturbation by methyl mercury. ReproductiveToxicology, 29, 383–392.

17. zur Nieden, N.I., Davis, L.A., Rancourt, D.E. (2010). Monolayer cultivation of osteo-progenitors shortens duration of the embryonic stem cell test while reliably predictingdevelopmental osteotoxicity. Toxicology, 277, 66–73.

18. Adler, S., Pellizzer, C., Hareng, L., Hartung, T., Bremer, S. (2008). First steps in establish-ing a developmental toxicity testmethod based on human embryonic stem cells.ToxicologyIn Vitro, 22, 200–211.

19. Mehta, A., Konala, V.B., Khanna, A.,Majumdar, A.S. (2008). Assessment of drug induceddevelopmental toxicity using human embryonic stem cells.Cell Biology International, 32,1412–1424.

20. zur Nieden, N.I., Ruf, L.J., Kempka, G., Hildebrand, H., Ahr, H.J. (2001). Molecularmarkers in embryonic stem cells. Toxicology In Vitro, 15, 455–461.

21. Pellizzer, C., Bello, E., Adler, S., Hartung, T., Bremer, S. (2004). Detection of tissue-specific effects by methotrexate on differentiating mouse embryonic stem cells. BirthDefects Research. Part B, Developmental and Reproductive Toxicology, 71, 331–341.

22. Stummann, T.C., Hareng, L., Bremer, S. (2007). Embryotoxicity hazard assessment ofmethylmercury and chromium using embryonic stem cells. Toxicology, 242, 130–143.

23. zur Nieden, N.I., Kempka, G., Ahr, H.J. (2004). Molecular multiple endpoint embryonicstem cell test—a possible approach to test for the teratogenic potential of compounds.Toxicology and Applied Pharmacology, 194, 257–269.

24. Buesen, R., Genschow, E., Slawik, B., Visan, A., Spielmann, H., Luch, A., Seiler, A.(2009). Embryonic stem cell test remastered: comparison between the validated EST andthe new molecular FACS-EST for assessing developmental toxicity in vitro. ToxicologicalSciences, 108, 389–400.

25. Teo, J.L., Kahn, M. (2010). The Wnt signaling pathway in cellular proliferation anddifferentiation: A tale of two coactivators. Advanced Drug Delivery Reviews, 62, 1149–1155.

26. Barrier, M., Jeffay, S., Nichols, H.P., Chandler, K.J., Hoopes, M.R., Slentz-Kesler, K.,Hunter 3rd, E.S. (2011). Mouse embryonic stem cell adherent cell differentiation andcytotoxicity (ACDC) assay. Reproductive Toxicology, 31, 383–391.

REFERENCES 369

27. Chandler, K.J., Barrier, M., Jeffay, S., Nichols, H.P., Kleinstreuer, N.C., Singh, A.V., Reif,D.M., Sipes, N.S., Judson, R.S., Dix, D.J., Kavlock, R., Hunter 3rd, E.S., Knudsen, T.B.(2011). Evaluation of 309 environmental chemicals using a mouse embryonic stem celladherent cell differentiation and cytotoxicity assay. PLoS One, 6, e18540.

28. van Dartel, D.A., Pennings, J.L., van Schooten, F.J., Piersma, A.H. (2010).Transcriptomics-based identification of developmental toxicants through their interfer-ence with cardiomyocyte differentiation of embryonic stem cells. Toxicology and AppliedPharmacology, 243, 420–428.

29. vanDartel, D.A., Pennings, J.L., de la Fonteyne, L.J., Brauers, K.J., Claessen, S., vanDelft,J.H., Kleinjans, J.C., Piersma, A.H. (2011). Concentration-dependent gene expressionresponses to flusilazole in embryonic stem cell differentiation cultures. Toxicology andApplied Pharmacology, 251, 110–118.

30. vanDartel, D.A., Pennings, J.L., de la Fonteyne, L.J., Brauers, K.J., Claessen, S., vanDelft,J.H., Kleinjans, J.C., Piersma, A.H. (2011). Evaluation of developmental toxicant iden-tification using gene expression profiling in embryonic stem cell differentiation cultures.Toxicological Sciences, 119, 126–134.

31. van Dartel, D.A., Pennings, J.L., Robinson, J.F., Kleinjans, J.C., Piersma, A.H. (2011).Discriminating classes of developmental toxicants using gene expression profiling in theembryonic stem cell test. Toxicology Letters, 201, 143–151.

32. Theunissen, P.T., Robinson, J.F., Pennings, J.L., de Jong, E., Claessen, S.M., Kleinjans,J.C., Piersma, A.H. (2012). Transcriptomic concentration-response evaluation of valproicacid, cyproconazole and hexaconazole in the neural embryonic stem cell test (ESTn).Toxicological Sciences, 125, 430–438.

33. van Dartel, D.A., Pennings, J.L., Hendriksen, P.J., van Schooten, F.J., Piersma, A.H.(2009). Early gene expression changes during embryonic stem cell differentiation intocardiomyocytes and their modulation by monobutyl phthalate. Reproductive Toxicology,27, 93–102.

34. van Dartel, D.A., Zeijen, N.J., de la Fonteyne, L.J., van Schooten, F.J., Piersma, A.H.(2009). Disentangling cellular proliferation and differentiation in the embryonic stem celltest, and its impact on the experimental protocol. Reproductive Toxicology, 28, 254–261.

35. Pennings, J.L., van Dartel, D.A., Robinson, J.F., Pronk, T.E., Piersma, A.H. (2011). Geneset assembly for quantitative prediction of developmental toxicity in the embryonic stemcell test. Toxicology, 284, 63–71.

36. de Jong, E., Louisse, J., Verwei, M., Blaauboer, B.J., van de Sandt, J.J., Woutersen, R.A.,Rietjens, I.M., Piersma, A.H. (2009). Relative developmental toxicity of glycol etheralkoxy acid metabolites in the embryonic stem cell test as compared with the in vivopotency of their parent compounds. Toxicological Sciences, 110, 117–124.

37. Louisse, J., de Jong, E., van de Sandt, J.J., Blaauboer, B.J.,Woutersen, R.A., Piersma,A.H.,Rietjens, I.M., Verwei, M. (2010). The use of in vitro toxicity data and physiologicallybased kinetic modeling to predict dose-response curves for in vivo developmental toxicityof glycol ethers in rat and man. Toxicological Sciences, 118, 470–484.

38. Kultima, K., Jergil, M., Salter, H., Gustafson, A.L., Dencker, L., Stigson, M. (2010). Earlytranscriptional responses in mouse embryos as a basis for selection of molecular markerspredictive of valproic acid teratogenicity. Reproductive Toxicology, 30, 457–468.

39. Robinson, J.F., Verhoef, A., Piersma, A. (2012). Transcriptomic analysis of neurulationand early organogenesis in rat embryos: an in vivo and ex vivo comparison. ToxicologicalSciences, 126(1), 255–266.


40. Robinson, J.F., Verhoef, A., Pennings, J.L., Pronk, T.E., Piersma, A.H. (2012). A compari-son of gene expression responses in rat whole embryo culture and in vivo: time-dependentretinoic acid-induced teratogenic response. Toxicological Sciences, 126(1), 242–254.

41. Robinson, J.F., Theunissen, P.T., van Dartel, D.A., Pennings, J.L., Faustman, E.M.,Piersma, A.H. (2011). Comparison of MeHg-induced toxicogenomic responses acrossin vivo and in vitro models used in developmental toxicology. Reproductive Toxicology,32, 180–188.

42. Pennings, J.L., Theunissen, P.T., Piersma, A.H. (2012). An optimized gene set for tran-scriptomics based neurodevelopmental toxicity prediction in the neural embryonic stemcell test. Toxicology. 300(3):158–167.

43. Schulpen, S.H., Robinson, J.F., Pennings, J.L., van Dartel, D.A., Piersma, A.H. (2012).Dose response analysis of monophthalates in the murine embryonic stem cell test assessedby cardiomyocyte differentiation and gene expression. Reproductive Toxicology.

20ZEBRAFISH DEVELOPMENT:HIGH-THROUGHPUT TEST SYSTEMSTO ASSESS DEVELOPMENTALTOXICITY

Stephanie Padilla

20.1 INTRODUCTION

The literature is replete with papers extolling the virtues of using zebrafish (Daniorerio) embryonic development for toxicity or mechanistic screening. There are, how-ever, relatively few studies in which a zebrafish early developmental assaywas used toscreen hundreds to thousands of chemicals. As an illustration of this, a comprehensivereview was recently published cataloging many different types (i.e., morphological,cardiovascular system, nervous system, behavioral, sensory system, digestive system,immune system, pigmentation, and metabolic) of zebrafish developmental toxicityscreening approaches and results [1]. Few of the publications cited screened largenumbers of chemicals. The present chapter will approach the discussion of zebrafishdevelopmental toxicity screens from a different aspect. I begin with a short reviewof zebrafish and their advantages as a small animal alternative model, followed byevidence of developmental concordance between zebrafish and mammals, accom-panied by examples of large-scale screens published to date, and conclude with adiscussion of one factor that may limit the usefulness of a zebrafish developmentalscreen.


371

372 HIGH-THROUGHPUT TEST SYSTEMS TO ASSESS DEVELOPMENTAL TOXICITY

20.2 ZEBRAFISH BACKGROUND

The zebrafish, a small (3 to 5 cm) freshwater teleost fish, is a vertebrate species thatis particularly amenable to large-scale screening of chemical libraries (reviewed inReferences 2–4). These easy to rear and maintain fish produce copious eggs thatcan be reared in microtiter plates for the first week after fertilization. The offspringmature rapidly, with organogenesis complete within 3 days, and by 6 days, the larvaeare free swimming, feeding, and have a large repertoire of behaviors. Clear, detailed,encyclopedic information about zebrafish developmental timing and landmarks isavailable [5, 6]. The zebrafish body plan is much like other vertebrates includingtoxicologically relevant structures such as a liver for metabolic activation with a largecomplement of cytochrome P450s, many that are active very early in development[7–11], a thyroid gland that controls development [12–14], and a blood–brain barrier[15, 16]. Thus, the zebrafish embryo provides an integrated model of development,completewith tissue repairmechanisms and feedback loops, in otherwords, aspects ofdevelopmental and biological complexity that in vitro systems are unable to capture.

20.3 DEVELOPMENTAL CONCORDANCE

Given that the zebrafish represents a rapidly developing vertebrate model amenableto high-throughput screening, questions arise whether the underlying mechanisms ofdevelopment are also similar to mammalian development. Therefore, what evidenceis there that the pathways that control development in the zebrafish are similar tothose that control and pattern development in mammals?Concordance betweenmammalian development and teleost development is emerg-

ing. There are many examples where key developmental signaling pathways and theirregulation are conserved between fish andmammals. A comparison of themammalianand zebrafish genomes revealed long-range conserved synteny, attesting to the similar-ity in important regulatory sequences [17]. Some aspects of the zebrafish and humangenomes are similar [18] and interchangeable, as highly conserved non-coding ele-ments in human genes will regulate gene expression in zebrafish when transfectedduring development [19, 20]. The molecular pathways that control patterns of cellularmovement during gastrulation, a crucial phase in development when the three primarygerm layers and the rudimentary body plan is established, is conserved betweenmam-mals and zebrafish (reviewed in Reference 21). In mice and zebrafish, endothelin-1is an integral signaling molecule involved in craniofacial (jaw) development in both[22], and the same gene (Foxg1) is required for olfactory development in both species[23]. It also appears that some of the molecular pathways regulating arteriogenesisare conserved in mice and fish [24]. In the nervous system, well-known sedativedrugs in humans also elicit sedative effects in larval zebrafish, even to the extent thatenantiomeric differences are mirrored [25]. Additionally, it was found that zebrafishpossess orthologs for 86% of the 1318 human drug targets tested [26].Due to the concordance between zebrafish and mammalian developmental path-

ways, the zebrafish model is often promoted for studying mammalian disease

EXAMPLES OF SCREENS AND VALIDATION APPROACHES 373

(reviewed in References 27, 28). It has been suggested that zebrafish would be anexcellent model for ocular disease [29], ocular motor disorders [30], autism [31] and,in general, for the molecular dissection of developmental pathways [23, 26, 32–37].One area in which many investigators have used zebrafish developmental screen-ing is ototoxicity [38–41]. Zebrafish embryos/larvae were used to screen chemicalsduring the period of lateral line neuromast development; neuromasts are structurallysimilar to mammalian inner ear hair cells. Many chemicals that cause hair cell dam-age in humans also caused neuromast damage in developing zebrafish. A subsequentscreening of a library of thousands of FDA-approved pharmaceuticals identified somechemicals that protected neuromasts from chemically (neomycin)-induced damage[39, 40]. To demonstrate mammalian/teleost concordance, some of the chemicalsidentified as ototoxic or otoprotective in the zebrafish screens were also ototoxicor otoprotective in a mammalian (mature mouse) explant culture of utricle hair cell[38–40].Zebrafish embryonic models combined with morpholino gene expression knock-

down technology are becoming common as translational models by dissecting a toxicmechanism or gene function. This translational pattern is generally as follows: (1) thegene associatedwith a human toxic response or disease condition is identified, usuallythrough epidemiological surveys; (2) the zebrafish ortholog for the gene is identifiedand; (3) using that ortholog, a knockdown is constructed in zebrafish embryos for thecharacterization of the phenotype and rescue parameters. For example, it has beenshown that a transcription factor (IRX3) that had been associated by epidemiologicalstudies with diabetes and obesity in humans is involved in the development of insulin-producing cells in the zebrafish pancreas [20]. Unraveling the toxicity pathways forthalidomide toxicity, Ito and coworkers [34] showed that knocking down variouscomponents of the hypothesized thalidomide toxicity pathway elicited a phenotypein zebrafish similar to that produced by thalidomide treatment in mammals, therebyproviding strong evidence to support their proposed mechanism of toxicity. Subse-quent testing of this mechanism in chicks showed that it was in accordance withthe zebrafish results. Furthermore, in their study of the molecular underpinnings ofschizophrenia,Wood and coworkers [42] delineated the function of two schizophreniacandidate genes first identified in human epidemiological studies. Using knockdowntechnology in developing zebrafish, they found out that both genes were involvedin oligodendrocyte development as well as in the development of certain cerebellarneurons, an outcome that prompted the authors to conclude that there were likelypathway connections between these two, key schizophrenia-associated genes.

20.4 EXAMPLES OF SCREENS AND VALIDATION APPROACHES

Given the congruence between zebrafish and mammalian development discussedabove, many investigators have used embryonic zebrafish in screens to predict whichchemicals are likely to bemammalian toxicants. In these types of studies the zebrafishresults may be evaluated by comparison with mammalian toxicity studies. An exam-ple of this approach [43] compared the zebrafish LC50 (nominal water concentration


causing 50% mortality in the zebrafish embryo/larva) with published mammalianLD50 (the dose causing 50% mortality in rodents) values. Given that this was a com-parison between zebrafish developmental lethality and rodent adult, acute lethality, alinear regression r2 of 0.73 was, nonetheless, encouraging. These authors also showedthat if the test compounds were subdivided into chemical classes (i.e., alcohols, alka-loids, amides, carboxylic acids, glycosides, and “others”), the slopes of the regressionlines for the individual classes ranged from 0.36 (carboxylic acids) to 1.27 (“others”),which was interpreted as meaning that some classes were more toxic to zebrafish thanrodents (slope lower than 1), other classes were equally toxic (slope of 1), and classeswith a slope significantly higher than 1 were less toxic to zebrafish embryos thanrodents. A similar approach was taken by Hermsen and coworkers [44], who com-pared the toxicity of a group of triazole antifungals in a zebrafish developmentalassay (exposure was from day 0 to day 3 post-hatching) to rodent developmentaltoxicity obtained from the toxicity reference database (ToxRefDB; [45]). In this case,the authors did not use death as an endpoint in the zebrafish, but rather comparedthe benchmark nominal concentration calculated to produce a 5% change [46–48]in the general morphology score in the zebrafish larvae at day 3 to the lowest effec-tive developmentally toxic concentration in rodents. This comparison yielded a verystrong relationship with an r2 of 0.88. A subsequent paper [49] assessing the samechemical class in zebrafish embryos found comparable toxicity potencies as Hermsenand coworkers [44], thus demonstrating intralaboratory consistency at least for thetoxicity of this class of chemicals.In a screen for cell cycle inhibitors, Murphey and coworkers [50] used multiple

validation approaches. They screened a library of 16,320 chemicals by assessing thelevel of the serine-10-phosphorylated-histone 3 (pH3), a marker of mitotic cells, inzebrafish embryos exposed to the chemicals. To evaluate the validity of their screen,they also tested 17 chemicals known to disrupt various aspects of the cell cycleand found over 50% (9/17) positive in their zebrafish embryo screening assay. Ofthe eight validation compounds that were negative in the zebrafish embryo assay,seven were active in the in vitro AB9 zebrafish fibroblast culture preparation. Theauthors posited that the compounds were likely negative in the embryos but positivein the in vitro culture because of poor bioavailability to the embryo or becausethey were too embryotoxic at the concentrations used. Therefore, because 16 outof the 17 known positive compounds were active in either the zebrafish embryoassay and/or the zebrafish in vitro fibroblast assay, the authors concluded that “theconservation of drug targets between mammals and zebrafish is high.” Screening ofthe compound library yielded 14 novel compounds that perturbed the cell cycle inzebrafish embryos; only 6 of those also affected the cell cycle in NIH3T3 cells, amouse embryonic fibroblast cell line. On the surface, this seems to yield a less than50% concordance with possible mammalian toxicity using the NIH3T3 cells as asurrogate for mammalian toxicity. The authors further note, however, that 3 more ofthese 14 novel compounds are bound up by serum (a component of the NIH3T3 butnot of the zebrafish assay), which probably accounts for the negative result in theNIH3T3 assay.Relating chemical effects noted in zebrafish to actions in mammals can also be

undertaken using a bioinformatic approach. In their assessment of the rest/wake

EXAMPLES OF SCREENS AND VALIDATION APPROACHES 375

locomotor states in a 4-day larval zebrafish, Rihel and coworkers [51] tested 3968compounds and found that 13.7% (547) altered the behavioral phenotype (degree oflocomotor activity and diurnal variation in activity). Using hierarchical clustering,they found that the larval zebrafish behavior phenotype could group the neuroactivecompounds according to their mammalian mechanism of action. Therefore, a rela-tively simple behavioral assessment in zebrafish larvae was able to classify accuratelyhuman bioactive drugs. In a related study, the acute effects of primarily neuroactivedrugs were assessed [52]. A total of 13,976 chemicals from various libraries weretested using the pattern of locomotor activity in response to a bright light in 30-h-oldzebrafish embryos. They also found that hierarchical clustering of the patterns ofeffects in zebrafish tended to group chemicals with similar functions or mechanismsof action that occur in mammals. Novel molecules were also identified that clusteredwith known molecules. For example, a novel molecule clustered with the monoamineoxidase (MAO) inhibitors, and when tested in vitro was found to be a potent MAOinhibitor, thereby reinforcing the notion that a behavioral test in zebrafish larvae maybe able to screen chemicals for desired neuroactive properties. Kokel and coworkers[52] concluded that “functionally related molecules cause similar phenotypes, [and]behavioral barcodes may be used to sort molecules with different cellular targets intocommon pathways.”Other types of screens in zebrafish do not directly relate toxicity in zebrafish

embryos to mammalian toxicity; these studies are primarily to classify effects ofchemicals, and may not include any direct validation procedure. For example, Kita-mbi and coworkers [53] screened approximately 2000 small molecules for pertur-bations in blood vessel development using transgenic zebrafish. The investigatorswere interested in identifying chemicals that affected zebrafish eye vascularizationas potential drug candidates for mammalian retinal vascular pathology treatment. Inthe initial screen, four compounds were identified that specifically affected retinalvessel morphology in developing zebrafish. Two of these were structurally related,and when the authors tested other (novel) chemicals in the same class, many of thesewere also active in the assay. Thus, the zebrafish screen revealed a class of compoundsthat have the potential to affect retinal vascularization in humans.In summary, as described above, the level and type of validation can take many

forms, mainly depending on the goal(s) of the screen. One of the primary questionsthat must be answered when establishing a screen is to determine the purpose oraim of the screen, as it appears that the more narrow the focus, the more likelythe screen will be predictable. Is the screen designed to predict a certain endpoint,toxicity, or pathway? Related to this is the question of the target species of theprediction, that is, is the screen designed to predict mammalian, avian, or piscinetoxicity? Above are many examples of zebrafish/mammalian correlations. There arealso examples relating zebrafish embryo tests to tests in other species. In a studyrelating zebrafish embryo toxicity to either fish (rainbow trout) or bird (white leghornchicken) embryonic toxicity, it was found that the predictive power was quite good foreither species [54]. The correlations are graphed in Figure 20.1 showing an r2 for thezebrafish/rainbow trout comparison of 0.89, and an r2 for the zebrafish/white leghornchicken comparison of 0.81. Interestingly, the slope of the linear regression linecomparing the 7-day zebrafish embryo developmental assay conducted at 25◦C to the


Zebrafish [log(1/EC50) nmol/Liter]

–7 –6 –5 –4 –3 –2 –1 0

log(

1/EC

50) o

ther

org

anis

m

–7

–6

–5

–4

–3

–2

–1

0

Rainbow Trout White Leghorn Chicken

FIGURE 20.1 Comparison of zebrafish embryonic toxicity with rainbow trout or whiteleghorn chicken embryo toxicity. Data are taken from Table 4 of Reference 54. Thezebrafish/rainbow trout linear regression (solid grey line) has a slope of 1.1 and an r2 =0.89. The zebrafish/white leghorn chicken linear regression (dashed line) has an r2 = 0.81 anda slope of 0.36.

results of the rainbow trout embryo developmental assay conducted at 10◦C showeda value of 1.1, thus meaning that the potencies of the chemicals were very similarin each species. Other investigators have also shown a good correlation betweenzebrafish embryo toxicity and fish toxicity tests in other fish species [55].

20.5 THE QUESTION OF BIOAVAILABILITY

One of the greatest unknowns in embryonic zebrafish chemical screening is the issueof dose and how to determine or estimate bioavailability. Usually, the embryo isexposed by immersion in the rearing solution with the chemical added, which isconvenient and mostly straightforward. The unknown in this immersion scenario,however, is exactly how much and when the chemical reaches the embryo or larva.This is in direct contrast to dosing of laboratory mammals, which usually is done byinjection, gavage, or in the feed; in these scenarios, the investigator can be reasonablysure that the chemical has entered the animal. To compound this situation, for thefirst 2 to 3 days the zebrafish embryo is encased in a chorion, which may serve as anadditional barrier to some chemicals. The zebrafish chorion is approximately 3.5 �mthick, and is composed of three acellular layers with microvillus channels or pores[56] that are approximately 0.17 �m2 in area [57]. Embryos can be dechorionatedeither mechanically or with a pronase solution, but this dechorionation may affectthe integrity and behavior of the embryos [58, 59]. Dechorionation also eliminates

THE QUESTION OF BIOAVAILABILITY 377

the possibility of determining the effect of a chemical on hatching, a commonlyused toxicological endpoint. So given the unknowns regarding immersion chemicalexposure and whether the chorion functions as a barrier to the chemical, the issue ofchemical bioavailability in the embryo/larva is paramount.In the few studies in which investigators have measured the internal dose of a test

compound, it is interesting to note that the dose administered to the embryo/larvawas rarely identical to the nominal concentration in the surrounding water (e.g., seeReferences 59–62). To complicatematters further, the time course of bioaccumulationin the embryo/larva can also greatly vary [60, 63]. For example, when exposedto perfluorooctanesulfonic acid (PFOS) by immersion, the concentration of PFOSnever reached steady state in the embryo, but rather showed a continuous increasein concentration over the 5 days of exposure [61]; this sharply contrasts with thetime course of nicotine bioaccumulation in the embryo, which reaches a steady statewithin a matter of minutes [59]. In fact, there is a general recognition that chemicallipophilicity plays a major role in its toxicity profile in a zebrafish developmentalscreen [49, 60, 64, 65]. For example, Figure 20.2 shows the relationship between thehydrophobicity (i.e., log P, the log of the octanol/water partition coefficient) of eachof the 309 unique chemicals tested in a zebrafish developmental screen comparedto the likelihood that the chemical was toxic to the zebrafish embryo at a nominalconcentration of 80 �M or less [49]. Only 20% of the chemicals with log P below 0were toxic, and as the hydrophobicity of the chemical increased, so did the likelihoodof a chemical’s toxicity to developing zebrafish. This general pattern was also noted

LogP bin

below 0 0–1 1–2 2–3 3–4 4–5 above 5

Perc

ent p

ositi

ve c

hem

ical

s

0

20

40

60

80

100

FIGURE 20.2 Relationship between the octanol/water partition coefficient (log P) of achemical and likelihood that it will perturb development of zebrafish embryos. Chemicalstested were the 309 pesticide active ingredients and metabolites contained in the ToxCastTM

Phase I chemical library (data regraphed from Reference 49). The lower the log P, the lowerthe likelihood that the chemical will be toxic to the developing zebrafish.


by Sachidanandan and coworkers [65] when testing 5760 chemicals: in a zebrafishdevelopmental assay: no chemical with a log P below 0 was positive. Obviously,assessing dose to the embryo or bioavailability of the test chemical is essential forthe interpretation of any chemical screen, especially when the goal is to equatedose or body burden in other organisms. Therefore, to assess dose to the zebrafishembryo or larva, it would be desirable to have a direct measurement of the bodyburden of the chemical. This may be reasonable if only testing a few chemicals, butimpractical when testing hundreds of chemicals, so that a procedure for estimatingbody burden of the chemical in the embryo would be useful, especially in a screeningcontext. The log P seems to be an ideal candidate, as it can be estimated for eachchemical using the chemical structure. There are formulae using the log P to calculatebioaccumulation of a chemical in adult fish and other organisms [66–69], and there isone paper by Petersen and Kristensen [70] that investigated the relationship betweenlog P and bioavailability for selected lipophilic compounds (log Ps ranged from3.37 to 6.5) in zebrafish embryos and larvae, thereby calculating the relationship aslog of the bioconcentration factor = −0.46 + 0.86 log P. Moreover, in order toestimate bioavailability of a given chemical in the zebrafish during the first 7 days ofdevelopment, we reviewed the literature for papers that had made determinations ofinternal dose for chemicals with awide range of log Ps. These data were found in threepapers [59, 71, 72], are plotted in Figure 20.3, and show an excellent correlation with

log P–2 –1 0 1 2 3 4 5 6

Embr

yo/la

rva

conc

entr

atio

nas

% o

f nom

inal

0.01

0.1

1

10

100

1000

10000

100000

% Nominal = antilog[–0.089 + 0.725(Log P)]

FIGURE 20.3 Relationship between log P and body burden of chemical in zebrafishembryos/larvae. (�, = mean of 3 dpf and 7 dpf measures [71]; �, [72]; and gray circle,[59]. Linear regression (–) of these combined data gives an equation relating the concentrationin the embryo to the log P of the chemical: % nominal concentration in the embryo/larva =antilog[−0.089 + 0.725(log P)]. The r2 of this linear regression is 0.81. The dashed line isthe relationship between log P and bioconcentration factor (BCF) calculated by Reference 70,where log BCF= −0.46 + 0.86 (log P) for a group of lipophilic compounds tested in zebrafishembryos and larvae.

REFERENCES 379

an r2 (solid line) equaling 0.81. There is also excellent congruence with the formuladescribed above by Petersen and Kristensen [70] (dashed line, Fig. 20.3). The presentequation or the equation proposed by Petersen and Kristensen [70] seems like a goodplace to start for estimating bioavailability of a chemical in a zebrafish embryonicdevelopmental assay.In summary, a broad scientific community has embraced zebrafish embryos as

a convenient, robust alternative model appropriate for screening the developmentaltoxicity potential of chemicals using a wide range of approaches. As the tools forconducting these types of screens become more common, future studies are antici-pated that will use this model for assessments ranging from general embryo toxicityto the dissection of specific vertebrate developmental pathways.

ACKNOWLEDGMENTS

The author thanks Drs. Robert C. MacPhail and Timothy Shafer for commentingon an earlier version of the chapter. This manuscript has been subjected to reviewby the National Health and Environmental Effects Research Laboratory of the U.S.Environmental Protection Agency and approved for publication. Approval does notsignify that the contents reflect the views of the Agency, nor does mention of tradenames or commercial products constitute endorsement or recommendation for use.

REFERENCES

1. McCollum, C.W., Ducharme, N.A., Bondesson, M., Gustafsson, J.A. (2011). Develop-mental toxicity screening in zebrafish. Birth Defects Research. Part C, Embryo Today, 93,67–114.

2. Grunwald, D.J., Eisen, J.S. (2002). Headwaters of the zebrafish—emergence of a newmodel vertebrate. Nature Reviews Genetics, 3, 717–724.

3. Lieschke, G.J., Currie, P.D. (2007). Animal models of human disease: zebrafish swim intoview. Nature Reviews Genetics, 8, 353–367.

4. Spence, R., Gerlach, G., Lawrence, C., Smith, C. (2008). The behaviour and ecology ofthe zebrafish, Danio rerio. Biological Reviews of the Cambridge Philosophical Society,83, 13–34.

5. Kimmel, C.B., Ballard, W.W., Kimmel, S.R., Ullmann, B., Schilling, T.F. (1995). Stagesof embryonic development of the zebrafish. Developmental Dynamics, 203, 253–310.

6. Parichy, D.M., Elizondo,M.R., Mills, M.G., Gordon, T.N., Engeszer, R.E. (2009). Normaltable of postembryonic zebrafish development: staging by externally visible anatomy ofthe living fish. Developmental Dynamics, 238, 2975–3015.

7. Goldstone, J.V., McArthur, A.G., Kubota, A., Zanette, J., Parente, T., Jonsson, M.E.,Nelson, D.R., Stegeman, J.J. (2010). Identification and developmental expression of thefull complement of cytochrome P450 genes in zebrafish. BMC Genomics, 11, 643.

8. Otte, J.C., Schmidt, A.D., Hollert, H., Braunbeck, T. (2010). Spatio-temporal developmentof CYP1 activity in early life-stages of zebrafish (Danio rerio). Aquatic Toxicology, 100,38–50.


9. Stegeman, J.J., Goldstone, J.V., Hahn, M.E. (2010). Perspectives on zebrafish as a modelin environmental toxicology. In: Fish Physiology, Vol. 29 (Perry, S.F., Ekker, M., Farrell,A.P., Brauner, C.J., Eds.). Academic Press; pp. 367–439.

10. Tao, T., Peng, J. (2009). Liver development in zebrafish (Danio rerio). Journal of Geneticsand Genomics, 36, 325–334.

11. Weigt, S., Huebler, N., Strecker, R., Braunbeck, T., Broschard, T.H. (2011). Zebrafish(Danio rerio) embryos as a model for testing proteratogens. Toxicology, 281, 25–36.

12. Blanton, M.L., Specker, J.L. (2007). The hypothalamic-pituitary-thyroid (HPT) axis infish and its role in fish development and reproduction. Critical Reviews in Toxicology, 37,97–115.

13. Porazzi, P., Calebiro, D., Benato, F., Tiso, N., Persani, L. (2009). Thyroid gland develop-ment and function in the zebrafish model. Molecular and Cellular Endocrinology, 312,14–23.

14. Walpita, C.N., Crawford, A.D., Janssens, E.D., van der Geyten, S., Darras, V.M. (2009).Type 2 iodothyronine deiodinase is essential for thyroid hormone-dependent embryonicdevelopment and pigmentation in zebrafish. Endocrinology, 150, 530–539.

15. Eliceiri, B.P., Gonzalez, A.M., Baird, A. (2011). Zebrafish model of the blood-brainbarrier: morphological and permeability studies. Methods in Molecular Biology, 686,371–378.

16. Jeong, J.Y., Kwon, H.B., Ahn, J.C., Kang, D., Kwon, S.H., Park, J.A., Kim, K.W. (2008).Functional and developmental analysis of the blood-brain barrier in zebrafish. BrainResearch Bulletin, 75, 619–628.

17. Kikuta, H., Laplante, M., Navratilova, P., Komisarczuk, A.Z., Engstrom, P.G., Fredman,D., Akalin, A., Caccamo, M., Sealy, I., Howe, K., Ghislain, J., Pezeron, G., Mourrain,P., Ellingsen, S., Oates, A.C., Thisse, C., Thisse, B., Foucher, I., Adolf, B., Geling,A., Lenhard, B., Becker, T.S. (2007). Genomic regulatory blocks encompass multipleneighboring genes and maintain conserved synteny in vertebrates. Genome Research, 17,545–555.

18. Postlethwait, J.H., Woods, I.G., Ngo-Hazelett, P., Yan, Y.-L., Kelly, P.D., Chu, F., Huang,H., Hill-Force, A., Talbot, W.S. (2009). Zebrafish comparative genomics and the originsof vertebrate chromosomes. Genome Research, 10, 1890–1902.

19. Navratilova, P., Fredman, D., Hawkins, T.A., Turner, K., Lenhard, B., Becker, T.S. (2009).Systematic human/zebrafish comparative identification of cis-regulatory activity aroundvertebrate developmental transcription factor genes. Developmental Biology, 327, 526–540.

20. Ragvin, A., Moro, E., Fredman, D., Navratilova, P., DrivenesØ., Engstrom, P.G., Alonso,M.E., Mustienes Ede, L., Skarmeta, J.L., Tavares, M.J., Casares, F., Manzanares, M., vanHeyningen, V., Molven, A., Njølstad, P.R., Argenton, F., Lenhard, B., Becker, T.S. (2010)Long-range gene regulation links genomic type 2 diabetes and obesity risk regions toHHEX, SOX4, and IRX3. Proceedings of the National Academy of Sciences USA, 107,775–780.

21. Solnica-Krezel, L. (2005). Conserved patterns of cell movements during vertebrate gas-trulation. Current Biology, 15, R213–R228.

22. Clouthier, D.E., Schilling, T.F. (2004). Understanding endothelin-1 function during cran-iofacial development in the mouse and zebrafish. Birth Defects Research. Part C, EmbryoToday, 72, 190–199.

REFERENCES 381

23. Duggan, C.D., DeMaria, S., Baudhuin, A., Stafford, D., Ngai, J. (2008). Foxg1 is requiredfor development of the vertebrate olfactory system. Journal of Neuroscience, 28, 5229–5239.

24. Ren, B., Deng, Y., Mukhopadhyay, A., Lanahan, A.A., Zhuang, Z.W., Moodie, K.L.,Mulligan-Kehoe, M.J., Byzova, T.V., Peterson, R.T., Simons, M. (2010). ERK1/2-Akt1crosstalk regulates arteriogenesis in mice and zebrafish. Journal of Clinical Investigation,120, 1217–1228.

25. Renier, C., Faraco, J.H., Bourgin, P., Motley, T., Bonaventure, P., Rosa, F., Mignot, E.(2007). Genomic and functional conservation of sedative-hypnotic targets in the zebrafish.Pharmacogenetics and Genomics, 17, 237–253.

26. Gunnarsson, L., Jauhiainen, A., Kristiansson, E., Nerman, O., Larsson, D.G. (2008).Evolutionary conservation of human drug targets in organisms used for environmentalrisk assessments. Environmental Science & Technology, 42, 5807–5813.

27. Penberthy, W.T., Shafizadeh, E., Lin, S. (2002). The zebrafish as a model for humandisease. Frontiers in Bioscience, 7, d1439–1453.

28. Shin, J.T., Fishman, M.C. (2002). From Zebrafish to human: modular medical models.Annual Review of Genomics and Human Genetics, 3, 311–340.

29. Gestri, G., Link, B.A., Neuhauss, S.C. (2012). The visual system of zebrafish and its useto model human ocular diseases. Developmental Neurobiology, 72, 302–327.

30. Maurer, C.M., Huang, Y.Y., Neuhauss, S.C. (2011). Application of zebrafish oculomotorbehavior to model human disorders. Reviews in the Neurosciences, 22, 5–16.

31. Tropepe, V., Sive, H.L. (2003). Can zebrafish be used as a model to study the neurodevel-opmental causes of autism? Genes, Brain, and Behavior, 2, 268–281.

32. Dodd, A., Curtis, P.M., Williams, L.C., Love, D.R. (2000). Zebrafish: bridging the gapbetween development and disease. Human Molecular Genetics, 9, 2443–2449.

33. Fetcho, J.R. (2007). The utility of zebrafish for studies of the comparative biology ofmotor systems. Journal of Experimental Zoology. Part B, Molecular and DevelopmentalEvolution, 308, 550–562.

34. Ito, T., Ando, H., Suzuki, T., Ogura, T., Hotta, K., Imamura, Y., Yamaguchi, Y., Handa,H. (2010). Identification of a primary target of thalidomide teratogenicity. Science, 327,1345–1350.

35. Spitsbergen, J.M., Kent,M.L. (2003). The state of the art of the zebrafishmodel for toxicol-ogy and toxicologic pathology research - advantages and current limitations. ToxicologicPathology, 31(suppl), 62–87.

36. Xi, Y., Ryan, J., Noble, S., Yu, M., Yilbas, A.E., Ekker, M. (2010). Impaired dopaminergicneuron development and locomotor function in zebrafish with loss of pink1 function.European Journal of Neuroscience, 31, 623–633.

37. Xu, C., Zon, L.I. (2010). The zebrafish as a model for human disease. In: Fish Physiology,Vol. 29 (Perry, S.F., Ekker, M., Farrell, A.P., Brauner, C.J., Eds.). Academic Press; pp.345–365.

38. Chiu, L.L., Cunningham, L.L., Raible, D.W., Rubel, E.W., Ou, H.C. (2008). Using thezebrafish lateral line to screen for ototoxicity. Journal of the Association for Research inOtolaryngology, 9, 178–190.

39. Coffin, A.B., Ou, H., Owens, K.N., Santos, F., Simon, J.A., Rubel, E.W., Raible, D.W.(2010). Chemical screening for hair cell loss and protection in the zebrafish lateral line.Zebrafish, 7, 3–11.


40. Ou, H.C., Cunningham, L.L., Francis, S.P., Brandon, C.S., Simon, J.A., Raible, D.W.,Rubel, E.W. (2009). Identification of FDA-approved drugs and bioactives that protect haircells in the zebrafish (Danio rerio) lateral line and mouse (Mus musculus) utricle. Journalof the Association for Research in Otolaryngology, 10, 191–203.

41. Ton, C., Parng, C. (2005). The use of zebrafish for assessing ototoxic and otoprotectiveagents. Hearing Research, 208, 79–88.

42. Wood, J.D., Bonath, F., Kumar, S., Ross, C.A., Cunliffe, V.T. (2009). Disrupted-in-schizophrenia 1 and neuregulin 1 are required for the specification of oligodendrocytesand neurones in the zebrafish brain. Human Molecular Genetics, 18, 391–404.

43. Ali, S., van Mil, H.G., Richardson, M.K. (2011). Large-scale assessment of the zebrafishembryo as a possible predictive model in toxicity testing. PLoS One, 6, e21076.

44. Hermsen, S.A., van den Brandhof, E.J., van der Ven, L.T., Piersma, A.H. (2011). Relativeembryotoxicity of two classes of chemicals in a modified zebrafish embryotoxicity testand comparison with their in vivo potencies. Toxicology in Vitro, 25, 745–753.

45. http://actor.epa.gov/toxrefdb/faces/Home.jsp

46. Gaylor, D.W. (1996). Quantalization of continuous data for benchmark dose estimation.Regulatory Toxicology and Pharmacology, 24, 246–250.

47. Gaylor, D.W., Aylward, L.L. (2004). An evaluation of benchmark dose methodology fornon-cancer continuous-data health effects in animals due to exposures to dioxin (TCDD).Regulatory Toxicology and Pharmacology, 40, 9–17.

48. Slob, W. (2002). Dose-response modeling of continuous endpoints. Toxicological Sci-ences, 66, 298–312.

49. Padilla, S., Corum, D., Padnos, B., Hunter, D.L., Beam, A., Houck, K.A., Sipes, N.,Kleinstreuer, N., Knudsen, T., Dix, D.J., Reif, D.M. (2012). Zebrafish developmentalscreening of the ToxCastTM Phase I chemical library. Reproductive Toxicology, 33, 174–187.

50. Murphey, R.D., Stern, H.M., Straub, C.T., Zon, L.I. (2006). A chemical genetic screen forcell cycle inhibitors in zebrafish embryos.Chemical Biology & Drug Design, 68, 213–219.

51. Rihel, J., Prober, D.A., Arvanites, A., Lam, K., Zimmerman, S., Jang, S., Haggarty, S.J.,Kokel, D., Rubin, L.L., Peterson, R.T., Schier, A.F. (2010). Zebrafish behavioral profilinglinks drugs to biological targets and rest/wake regulation. Science, 327, 348–351.

52. Kokel, D., Bryan, J., Laggner, C.,White, R., Cheung, C.Y.,Mateus, R., Healey,D., Kim, S.,Werdich, A.A., Haggarty, S.J., Macrae, C.A., Shoichet, B., Peterson, R.T. (2010). Rapidbehavior-based identification of neuroactive small molecules in the zebrafish. NatureChemical Biology, 6, 231–237.

53. Kitambi, S.S.,McCulloch,K.J., Peterson, R.T.,Malicki, J.J. (2009). Smallmolecule screenfor compounds that affect vascular development in the zebrafish retina. Mechanisms ofDevelopment, 126, 464–477.

54. van Leeuwen, C.J., Grootelaar, E.M., Niebeek, G. (1990). Fish embryos as teratogenic-ity screens: a comparison of embryotoxicity between fish and birds. Ecotoxicology andEnvironmental Safety, 20, 42–52.

55. Lammer, E., Carr, G.J., Wendler, K., Rawlings, J.M., Belanger, S.E., Braunbeck, T.(2009). Is the fish embryo toxicity test (FET) with the zebrafish (Danio rerio) a potentialalternative for the fish acute toxicity test? Comparative Biochemistry and Physiology CToxicology & Pharmacology, 149, 196–209.

REFERENCES 383

56. Bonsignorio, D., Perego, L., Del Giacco, L., Cotelli, F. (1996). Structure and macromolec-ular composition of the zebrafish egg chorion. Zygote, 4, 101–108.

57. Cheng, J., Flahaut, E., Cheng, S.H. (2007). Effect of carbon nanotubes on developingzebrafish (Danio rerio) embryos. Environmental Toxicology and Chemistry, 26, 708–716.

58. Saint-Amant, L., Drapeau, P. (1998). Time course of the development of motor behaviorsin the zebrafish embryo. Journal of Neurobiology, 37, 622–632.

59. Thomas, L.T., Welsh, L., Galvez, F., Svoboda, K.R. (2009). Acute nicotine exposure andmodulation of a spinal motor circuit in embryonic zebrafish. Toxicology and AppliedPharmacology, 239, 1–12.

60. Diekmann, H., Hill, A. (2012). ADMETox in zebrafish. Drug Discovery Today: DiseaseModels, in press. Available at http://dx.doi.org/10.1016/j.ddmod.2012.02.005

61. Huang, H., Huang, C., Wang, L., Ye, X., Bai, C., Simonich, M.T., Tanguay, R.L., Dong, Q.(2010). Toxicity, uptake kinetics and behavior assessment in zebrafish embryos followingexposure to perfluorooctanesulphonicacid (PFOS). Aquatic Toxicology, 98, 139–147.

62. Schreiber, R., Altenburger, R., Paschke, A., Schuurmann, G., Kuster, E. (2009). A novelin vitro system for the determination of bioconcentration factors and the internal dose inzebrafish (Danio rerio) eggs. Chemosphere, 77, 928–933.

63. El-Amrani, S., Pena-Abaurrea, M., Sanz-Landaluze, J., Ramos, L., Guinea, J., Camara,C. (2012). Bioconcentration of pesticides in Zebrafish eleutheroembryos (Danio rerio).Science of the Total Environment, 425, 184–190.

64. Milan, D.J., Peterson, T.A., Ruskin, J.N., Peterson, R.T., MacRae, C.A. (2003). Drugsthat induce repolarization abnormalities cause bradycardia in zebrafish. Circulation, 107,1355–1358.

65. Sachidanandan, C., Yeh, J.R., Peterson, Q.P., Peterson, R.T. (2008). Identification of anovel retinoid by small molecule screening with zebrafish embryos. PLoS One, 3, e1947.

66. Arnot, J.A., Arnot, M., Mackay, D., Couillard, Y., Macdonald, D., Bonnell, M., Doyle,P. (2009). Molecular size cut-off criteria for screening bioaccumulation potential: Fact orfiction? Integrated Environmental Assessment and Management, 6, 210–224.

67. Connell, D.W., Hawker, D.W. (1988). Use of polynomial expressions to describe thebioconcentration of hydrophobic chemicals by fish. Ecotoxicology and EnvironmentalSafety, 16, 242–257.

68. Davies, R.P., Dobbs, A.J. (1984). The prediction of bioconcentration in fish. WaterResearch, 18, 1253–1262.

69. Fu, W., Franco, A., Trapp, S. (2009). Methods for estimating the bioconcentration factorof ionizable organic chemicals. Environmental Toxicology and Chemistry, 28, 1372–1379.

70. Petersen, G.I., Kristensen, P. (1998). Bioaccumulation of lipophilic substances in fishearly life stages. Environmental Toxicology and Chemistry, 17, 1385–1395.

71. Berghmans, S., Butler, P., Goldsmith, P., Waldron, G., Gardner, I., Golder, Z., Richards,F.M., Kimber, G., Roach, A., Alderton, W., Fleming, A. (2008). Zebrafish based assaysfor the assessment of cardiac, visual and gut function - potential safety screens for earlydrug discovery. Journal of Pharmacological and Toxicological Methods, 58, 59–68.

72. Gustafson, A.L., Stedman, D.B., Ball, J., Hillegass, J.M., Flood, A., Zhang, C.X., Panzica-Kelly, J., Cao, J., Coburn, A., Enright, B.P., Tornesi, M.B., Hetheridge, M., Augustine-Rauch, K.A. (2012). Inter-laboratory assessment of a harmonized zebrafish developmentaltoxicology assay - Progress report on phase I. Reproductive Toxicology, 33, 155–164.

21SINGLE CELL IMAGINGCYTOMETRY-BASEDHIGH-THROUGHPUT ANALYSIS OFDRUG-INDUCED CARDIOTOXICITY

Min Jung Kim and Joon Myong Song

21.1 INTRODUCTION

Ion channels exist in the biological membranes of cells and regulate ionic flux acrossthe membranes. They play a pivotal role in maintaining normal cellular homeostasisand cellular signal transduction across the membrane [1]. Ion channels in a cardiaccell are very important because the contraction of cardiac muscle cells in the heartis directly controlled by the action potential generated by the cooperative interac-tions of diverse types of ion channels [2]. The cardiac action potential encompassesventricular depolarization and repolarization. The depolarization of the ventriclesinvolves a decrease in the electrical potential across a membrane and the repolar-ization of the ventricles involves a recovery process to the resting potential. Thecardiac action potential can be measured by performing a surface electrocardio-gram (ECG) [3]. The ECG provides information on the electrical changes, whichinduce ventricular depolarization and repolarization within the heart. Depolarizationof the ventricles is initiated by the fast influx of Na+ through selective sodiumchannels (phase 0). Then, a rapid repolarization occurs (phase 1). During phase 1the fast sodium ion channel is inactivated and a transient outward current occursdue to activating and inactivating outward potassium channels. Subsequently, theplateau phase (phase 2) is sustained by Ca2+ influx through L-type calcium channelsand K+ efflux through the slow delayed rectifier K+ channels. In this phase, the


385

386 ANALYSIS OF DRUG-INDUCED CARDIOTOXICITY

regulatory factors including Na+ –Ca2+ exchanger and Na+ /K+ -pump also con-tribute to restore the original intracellular ion concentrations. During repolarization(phase 3) the negative membrane potential is restored by the closure of calciumchannels and opening of slow and rapid delayed rectifier K+ channels. Inwardlyrectifying potassium channels also contribute to the repolarization. Thereafter, thecells maintain a stable resting membrane potential (phase 4) [4].The action potential (depolarization and repolarization) is a fundamental property

of cardiac cells. Thus, abnormal changes in normal ion channel activities induceundesired biological side effects in cardiac cells, such as prolongation of QT intervalinducing “Torsades de pointes” (TdP) and even sudden cell death [4–7]. The QTinterval, that is, the time between the onset of the Q wave and the end of the T wave,is the period of time in which ventricular depolarization and repolarization occur.Several drugs have adverse effects on the cardiac action potential and cause cardiacdysfunction due to changes in ion channel activities. When such drug-induced sideeffects are discovered, therapeutic drug usage is limited and sometimes a drug hasto be withdrawn from the market [6]. Therefore, such a risk should be detected earlyon during the drug development process. In view of interactions between many ionchannels, the integrated activities of multiple ion channels should be measured at amolecular level. Cell proliferation and death should be investigated at a single celllevel. Consequently, high content screening (HCS) of drug-induced changes in ionchannel activity is anticipated to play a key role in drug discovery and develop-ment. HCS is a drug discovery method that uses images of living cells for moleculediscovery.This chapter has been divided into four sections. The first section describes single

cell imaging cytometry as a useful tool in high-throughput screening (HTS) of drug-induced cardiotoxicity. The second section describes the detection techniques usedto analyze changes in ion concentration/channel permeability that account for drug-induced cardiotoxicity at the molecular level. The ion concentration and channelpermeability are detected by using fluorescent indicators. Each fluorescentmarker canbe visualized separately and simultaneously. The third section describes simultaneousand quantitative detection techniques to determine cell death processes includingapoptosis and necrosis. The fourth section describes how the various techniquespresented in the second and third sections can be applied in a single cell imagingcytometry system to determine cardiotoxicity.

21.2 SINGLE CELL IMAGING CYTOMETRY

21.2.1 Introduction

Several techniques are commonly used to investigate drug-induced cardiotoxicityincluding in vivo the ECG and in vitro the patch-clamp technique [8], the Rb+

efflux assay [9], and the microelectrode assay [10, 11]. Conventional patch clamp-ing in mammalian cells is the most widely used technique to monitor drug-induced

SINGLE CELL IMAGING CYTOMETRY 387

cardiotoxicity [12]. Microelectrodes are patched onto the cell surface and the cur-rent and voltage in the voltage-clamp and current-clamp mode is controlled by usingmicroelectrode amplifiers. Although this technique delivers a lot of information and ishighly accurate, the patch-clamp technique only allows the analysis of drug-inducedeffects on a single ion channel, not on multiple ion channels. Moreover, conven-tional patch-clamping methods show low throughput and demand highly skilledstaff. A HTS technique is an alternative without the above-mentioned disadvan-tages and allows monitoring of drug-induced cardiotoxicity by delivering multi-ple/simultaneous information on ion channel activities.We have developed a generalized platform and quantitative method for cellular

assays based on a hyperspectral imaging system that makes use of a single cellimaging cytometer [13–15]. An assembled system is composed of a C-mount lens, anacousto-optic tunable filter (AOTF; TEAF10-0.45-0.7-S, Brimrose), a charge coupleddevice (CCD) camera, a fluorescence microscope, and a computer system equippedwith image acquisition and analysis software. Multicolor and multispectral imagesindicating multiple molecules in a cell are obtained based on AOTF. In single cellimaging cytometry system, AOTF functions as an electronically tunable emissionfilter. A piezoelectric transducer is united to a crystal of tellurium dioxide (TeO2). Afrequency of light is diffracted and shifted at the Bragg angle using high-frequencyacoustical compressionwaves inAOTF. The diffracted light at a particular wavelengthis transmitted and collected in a CCD camera. Drug-induced changes in diverseion channels and intracellular ion concentrations can be simultaneously monitoredusing fluorescent indicators. In addition, further cellular viability assays (i.e., thosemeasuring cell proliferation, apoptosis, and necrosis) triggered by anticancer drugsare efficiently discriminated using AOTF by observing an entire emission spectrumof fluorophores.The strength of single cell imaging cytometry is that automatic quantification is

possible using an unfocused image. Region selection based on a slightly unfocusedimage followed by background correction allows a uniform intensity distribution toall the objects over the entire image. Minor differences in gray scale values (due touneven cellularmorphology) of a tightly focused image do not allow uniform intensitydistribution over the entire image. On the other hand, a slightly defocused imageminimizes the minor differences in cellular morphology. This helps in achieving auniform intensity distribution over the entire object. Background correction is tosubtract the background image from the defocused image. This allows nullifying anuneven light intensity distribution over the desired image.The features of a single cell imaging cytometer are as follows:

1. The single cell imaging cytometry system is based on uniform threshold inten-sity distribution (TID).

2. The platform consists of an AOTF, C-mount lens, CCD camera, fluorescencemicroscope and commercially available software for image acquisition andanalysis.


3. A method for data acquisition and quantitative analysis, comprising the follow-ing steps:

A. Synchronizing the exposure time of CCD camera with AOTF scan-ning/filtering wavelength to obtain desired image plane (stack images) atdesired wavelength parameters.

B. Acquiring a defocused image, focused image, stack image, and images forbackground correction at an identical X and Y axis position.

C. Applying background correction and threshold adjustment to a defocusedimage.

i. A tightly focused image does not allow even intensity distribution ofthreshold objects all over the image due to cell morphological differ-ences and therefore does not allow a quantitative analysis. A slightlydefocused image along with its background correction allows evenintensity distribution over the entire image.

ii. This allows us to select all the objects larger than threshold in the entireimage.

iii. Quantitative analysis is possible using the region selection around thethreshold objects irrespective of their cellular morphology.

D. Image analysis and measurements of all objects in a defocused image.

E. Eliminating the unwanted contaminants by width deduction using objectclassification.

F. Creating a mask for desired object generation.

G. Making regions around the objects.

H. Saving and loading the regions in stack image file.

I. Automatic region measurements at each plane.

J. Automatic data exported to Microsoft Excel.

K. Allowing quantitative data analysis along with many other data analy-sis parameters like optical density, average gray value, Z position, angle,distance, area, width, image plane, elapsed time, stage label, wavelength,region label, intensity S/N, threshold area, etc.

4. Single cell imaging cytometry as a generalized platform and quantitativemethod can be used to perform a variety of cellular and drug screening assays.

5. Single cell imaging cytometry can be used for simultaneous monitoring ofcellular events, that is, apoptosis and necrosis, and quantitative estimation ofapoptosis and necrosis.

6. Single cell imaging cytometry is used for discriminating the mode of cell death,that is, apoptosis and necrosis, based on a three-dimensional (3D) visualizationeffect.

7. Single cell imaging cytometry can be used to analyze approximately 300 to400 cells, a cell number sufficient to achieve high statistical confidence in assayprotocols.

SINGLE CELL IMAGING CYTOMETRY 389

8. A generalized platform and quantitative method defined in claim one is usedfor quantitative estimation of apoptosis and necrosis.

9. A generalized platform and quantitative method defined in claim one canbe applied for a wide range of fluorochromes having their emission maximabetween 463 to 688 nm approximately.

By extension, single cell imaging cytometry can be coupled to a simple microflu-idic system for quantification of multivariate cellular responses related to ion channelactivities, intracellular ion concentrations, and cell viability assays in drug discoveryand development [13]. Microfluidic systems provide new opportunities for develop-ing efficient single cell imaging cytometry systems. Microfluidic systems facilitateminiaturization of the analysis platform for monitoring and quantification of cellularevents at a single cell level.

21.2.2 Construction of a Single Cell Imaging Cytometry System

The single cell imaging cytometry system can be applied to visualize drug toxicitythrough HTS. The single cell imaging cytometer consists of C-mount lens, AOTF(TEAF10-0.45-0.7-S, Brimrose), CCD camera, fluorescence microscope, and com-mercially available software (Fig. 21.1). Amercury arc lamp is used as an illuminationsource for drug-treated cells in a fluorescence microscope and the laser beam is puri-fied by an interference filter (U-MGFPHQ, Olympus, Tokyo, Japan). The purifiedlaser beam passes through the excitation filter and is reflected by a dichroic filter. Thefiltered laser beam is concentrated through a microscope objective lens (40× /60× )and focused onto the sample platform. Then, the fluorescent probes in the cells areinduced to emit by the filtered laser beam. The fluorescent emission is collected bythe same objective lens, passes through dichroic filter and comes out of the sideport (ø = 25 mm) of the microscope. To reduce the fluorescence beam diameter,the C-mount lens is attached to the side port of the fluorescence microscope. Then,the entire beam passes through the AOTF window of 10× 10 mm. A birefringentcrystal in the AOTF splits the fluorescence beam into two beams, the diffracted lightand undiffracted light. A beam is diffracted at a particular wavelength and the otherbeam is undiffracted. Therefore, the fluorescence image of the cells can be detectedat a particular wavelength. While a CCD camera captures fluorescence images, along-pass filter was placed in front of the CCD camera to remove any laser scattering.The fluorescence images are taken up by the CCD camera in the spectral regionfrom 463 nm to 688 nm. The exposure time of the CCD camera was adjusted to 1 sand the scanning rate of AOTF was set at 1 wavelength/s. The spectral resolutionof the AOTF is modifiable up to 0.1 nm. The AOTF frequency sweeping and theCCD image acquisition are operated coincidently. The detection limit of fluoresceinemission is 1× 10−8 M with the developed single cell imaging cytometer. Data anal-yses of fluorescence cell images are performed automatically using commerciallyavailable software (MetaMorph, Version 7.1.3.0, Molecular Devices).


FIGURE 21.1 (a) Scheme of single cell imaging cytometry. The AOTF alignment consistsof laser (1); exciter (2); beam splitter (2a); Z value (3); objective lens (60×) (4); defocusedthreshold image (5); focused threshold image (6); defocal plane (7); focal plane (8); sample(9); prism (10); focusing lens (11); AOTF (12); undiffracted beam (13); diffracted beam (14);detector (15); Bragg angle (�) (the angle between diffracted and undiffracted beam). (b) Thefocused, focused threshold, defocused image, defocused threshold. In the defocused image,TID is uniform at 20 �m ± 2 away from the focal point. (c) Hyperspectral image of threedifferent fluorophore probes. Apoptosis (Qdot R© 625-labeled indicator), sodium indicator, andpotassium ion channel indicator are measured at 625 nm, 579 nm, and 524 nm by single cellimaging cytometry. Figures (a) and (b) are reproduced fromReference 14with permission fromthe American Chemical Society; Figure (c) is reproduced from Reference 13 with permissionfrom the Royal Society of Chemistry. (See insert for color representation of this figure.)

21.2.3 Data Acquisition of Single Cell Imaging Cytometry

The single cell imaging cytometry system is applied to facilitate not only a rapidand image-based analysis but also a quantitative analysis. Quantitative analysis isaccomplished by using background images, slightly defocused and focused, andimage analysis and quantification is performedwith commercially available software.The background image is captured by using a water-containing platform. Slightlydefocused and focused images of each sample are captured by coordinating the z-axis

QUANTITATIVE HIGH-THROUGHPUT ANALYSIS 391

position of the sample plate at the same (x–y) location. Focused images are used toobtain the fluorescent cell images by subtracting the background image from thefocused image. In a slightly defocused (20 �m ± 2 away from the focal point)image, uniform TID over the entire objective is observed and facilitates automatedregion selection (one region per object). However, in a focused image the intercellularvariation in the Z value interrupts uniform TID over the cells. Based on the cell size(∼10 �m), images are filtered in a range of 1.84211 < N > 10 (N, the number ofobjects). With this process, unwanted contaminants such as serum protein coagu-lants and cell debris are omitted. After the omission step each cell is automaticallymarked as an individual cell by creating the region around the cells. Approximately300 cells can be measured. The average fluorescence intensities of all the selectedregions are measured at different wavelengths, depending on the fluorescent emis-sion wavelength, that is, propidium-iodide-stained cells at 617 nm. Stained cells arediscriminated from unstained cells by comparing with the control based on relativefluorescence intensity values.

21.3 QUANTITATIVE HIGH-THROUGHPUT ANALYSISOF DRUG-INDUCED CARDIOTOXICITY

21.3.1 Potassium Ion Channel

A great diversity of potassium channels is associated with the action potential inthe heart. The transient outward potassium channel is activated during phase 1. Theslow (IKs) and rapid (IKr) components of the delayed rectifier potassium currents,outward potassium currents, are raised in phase 3. The inward rectifier potassiumcurrent (IK1) also increases in phase 3 [4]. In most cases drug-induced or acquiredlong QT syndrome (acLQTS) inhibits the rapid delayed rectifier K+ current (IKr)[16–18]. For example, terfenadine, an antihistamine, prolonged cardiac repolarizationby inhibition of potassium currents in the rapid delayed rectifier potassium channel(IKr) and, thus, triggered TdP as well as sudden death [19]. Especially, the humanether-a-go-go-related gene (HERG) channel is critical to restore the resting state(repolarization, phase 3) of cell membrane. HERG, encoding the �-subunit of IKrchannels, is related to conduct potassium ion efflux in cardiomyocytes. Cardiotoxicdrugs block HERG channel activity. Cisapride is known as a HERG channel blockerby binding to two residues (Tyr 652 and Phe 656) in the S6 domain of IKr channel.Thereby, cisapride prolongs the QT interval and induces TdP and sudden death [20].In general, voltage-sensitive dyes are used to measure the fluorescence signal

changes in membrane potential. The fluorescence resonance energy transfer (FRET)-based assay utilizes different negatively charged membrane-soluble oxonol dyes asvoltage-sensing FRET acceptors and coumarin-tagged phospholipids integrated inthe outer membrane as donors [21]. The FRET-based assay is less sensitive becausethe acceptor and donor interaction is relatively slow and the rate of false positives ishigh. Nevertheless, fluorescence-based methods are suited for HTS/HCS and are lesslabor-intensive.


We have investigated the effect of drugs on potassium channels including theHERG channel by using a fluorescence-based method. This potassium ion channelassay monitors potassium channel and transporter activities by making use of thechemical interaction between thallium ions (Tl+ ) and Tl+ -sensitive dyes. For thepotassium ion channel assay a Tl+ -sensitive dye (FluxORTM dye), loading buffer,assay buffer, and stimulus buffer are used. The FluxORTM dye dissolved in DMSOis a membrane-permeable acetoxymethyl (AM) ester. The FluxORTM dye is dilutedwith FluxORTM assay buffer or Hank’s balanced salt solution (HBSS). PowerLoadTM

concentrate, a formulation of Pluronic R© surfactants, is added to the diluted dyesolution to assist dyes loading into cells. AM ester dyes are stable in PowerLoadTM

concentrate solution. Water-soluble probenecid retains the de-esterified FluxORTM

dye in the cytosol by blocking organic anion pumps in the cellular membrane. Thestimulus buffer consists of the channel opener and thallium concentrate, which haveto be dissolved in chloride-free buffer.The potassium channel is closed in the resting state. With a stimulus solution, the

potassium channel is opened and Tl+ in the extracellular solution passes throughthe potassium channel due to a strong inward driving force. After loading theFluxORTM dye inside the cells, endogenous esterases cleave the non-fluorescent AM

FIGURE 21.2 The potassium channel and intracellular sodium ion assays. The permeabilityof the potassium channel is measured using a Tl+ -sensitive dye and the concentration ofintracellular sodium ion is detected by using a sodium indicator. The Tl+ -sensitive dye andthe sodium indicator penetrate the plasma membrane and bind to Tl+ and intracellular sodiumions, respectively. Non-fluorescent indicators immediately become fluorescent after binding.If the potassium channel is blocked by drugs, the fluorescence intensity of the Tl+ -sensitivedye diminishes. If the concentration of intracellular sodium ion has changed, the fluorescenceintensity of the sodium ion indicator also varies. The Tl+ -sensitive dye and the sodium ionindicator show maximum emission wavelengths at 525 and 579 nm, respectively.


ester, which is a fluorogenic thallium-sensitive indicator (Fig. 21.2). The membrane-permeable Tl+ -sensitive dye then binds to cytosolic Tl+ and a fluorescence emissionis generated inside the cells. The maximum fluorescence intensity of the thallium-sensitive form of the FluxORTM dye is observed at 525 nm. In the case that drugsblock the potassium channel, FluxORTM dye inside the cells cannot bind to Tl+

because Tl+ cannot flow through the closed potassium channel, so that the flu-orescent intensity of FluxORTM dye will diminish when compared to the control(non-drug treatment). This potassium ion channel assay, if compared to the tradi-tional thallium flux, has been improved in the sense that it works under chloride-free conditions. The final concentration of the thallium ion is 2 mM even thoughthe FluxORTM reagent is highly sensitive to thallium. In traditional thallium fluxassays, thallium chloride is used as a thallium ion supplier, even though approxi-mately 4 mM thallium chloride leads to an insoluble precipitate. On the other hand,a high concentration of the thallium ion (50 mM) is achieved by using thalliumsulfate (Tl2SO4).

PROTOCOL 1: QUANTIFICATION OF THE PERMEABILITY OFTHE POTASSIUM ION CHANNEL

Equipment and Reagents

� Cardiomyocytes (e.g., H9c2(2-1) cells)� Quantitative single cell imaging cytometry system� Cardiotoxic chemicals� FluxORTM potassium ion channel assay kit (molecular probes)a

� Loading buffer (pH 7.4)b

Chemicals Volume

PowerLoadTM concentrate, 100× 100 �LFluxORTM reagent, reconstituted in DMSO 10 �LFluxORTM assay buffer, 10× 1 mLProbenecid, reconstituted in deionized water 100 �LDeionized water 8.8 mLTotal volume 10 mL

� Assay buffer

Chemicals Volume

FluxORTM assay buffer, 10× 1 mLProbenecid, reconstituted in deionized water 100 �LDeionized water 8.9 mLTotal volume 10 mL


� Stimulus bufferc

Chemicals +K+ (mL) −K+ (mL)

FluxORTM chloride-free buffer, 5× 1 1K2SO4 concentrate 1Tl2SO4 concentrate 0.5 0.5Deionized water 2.5 3.5Total volume 5 5

Methods

1. Culture the cardiomyocytes in 12-well plates to achieve log phase growthuntil 80% confluence.

2. Incubate the cardiomyocytes with the drugs at 37◦C under 5% CO2d.

3. Completely remove the drug-containing medium and resuspend in loadingbuffer (500 �L). Incubate the cells for 90 min at room temperature in a darkcondition.

4. Prepare the assay buffer and stimulus buffer.

5. After 90 min, wash cells once with dye-free assay buffer.

6. Aspirate the loading buffer and replace with 400 �L of assay buffer for 15min.

7. Add stimulus buffer (100 �L) for 90 s.

8. Wash cells three times with dye-free assay buffer.

9. Detach the cells with accutase (200 �L) for 15 min at 37◦C.10. Collect the cells in a 1.5 mL tube and spin the cells at 230 rcf for 3 min in a

centrifuge, discard supernatant and wash with PBS.

11. Load the cells on the sample platform such as a 384-well plate, 96-wellplate, microfluidic device, etc.

12. Monitor the fluorescence imaging using single cell imaging cytometry.

NotesaThe 100× Probenecid stock solution and the 1000× dye stock solution are stored at −20◦C for upto 6 months and protected from light and moisture. Please avoid repeated freeze–thaw cycles. The restof the materials are stored at 4◦C until use.bLoading buffer is prepared as follows: PowerLoadTM concentrate is premixedwith FluxORTM reagentfor 10 s. Then, deionized water, FluxORTM assay buffer, and probenecid are added to the mixture. Formost applications, the FluxORTM dye is loaded into the cells at room temperature. For best results,all buffers including dye-loading buffer, dye-free assay buffer, and stimulus buffer are prepared justbefore the HCS/HTS assay is performed.cCoadministration of potassium and thallium is carried out in the stimulus buffer for voltage-gatedpotassium channels such as HERG, Kv1.3, Kv2.1, and Kv7.2/7.3. Thallium alone is added to thestimulus buffer for resting and inward rectifier potassium channels such as Kir1.1 and Kir2.1.dThe experimentator decides which drug concentration and incubation time are used.


21.3.2 Concentration of Intracellular Sodium Ion

The concentration of the intracellular sodium ion is controlled by the sodium ionchannel, the Na+ –Ca2+ exchanger, and the Na+ /K+ -pump in the heart. The sodiumion channel plays a critical role in cardiac action potential because the cardiac sodiumchannel is involved in early fast depolarization by opening the channel gate [4]. Thesodium ion is selected and passes through the channel pore located in between S5 andS6 of the �-subunit of the sodium channel [22]. A gene mutation or blocking of thesodium channel by drugs that affect the �-subunit causes cardiac diseases includingacLQTS, atrial fibrillation (AF), progressive cardiac conduction defect (PCCD), sicksinus node syndrome (SSS), and Brugada syndrome (BrS) [23]. The cardiac Na+ –Ca2+ exchanger and the Na+ /K+ -pump are responsible for sodium ion influx intothe cardiomyocyte cytosol during repolarization [24]. The drugs directly bind to andblock these intracellular sodium ion-related channels. The inhibition of the Na+ /K+ -pump causes a decrease in the intracellular potassium ion concentration and anincrease in the intracellular sodium ion concentration. These channels are closelyconnected and affect each other. For example, digoxin is known as a Na+ /K+ -pumpblocker. The blockade of the Na+ /K+ -pump by digoxin triggers an accumulation ofintracellular sodium ions due to a decrease in the sodium ion influx rate by the Na+ –Ca2+ exchanger. The elevated sodium ion level in the cytosol inhibits the extrusionof calcium ions by the Na+ –Ca2+ exchanger.Drug-induced changes in sodium ion levels can be measured by using a sodium

fluorescent indicator. CoroNaTM Red, a cell-permeant fluorescent cationic probe, isone of the widely used sodium indicators that provide information on the sodiumion concentration in the cell. Distinct from an AM ester-based precursor, CoroNaTM

Red is transported into the cells and predominantly localized in the mitochondria.CoroNaTM Red directly binds to sodium ion in the cells; therefore, the fluores-cence intensity of the dye depends on the concentration of intracellular sodiumion. For example, a digoxin-treated cell has a higher fluorescent intensity thandigoxin-untreated cell. The disadvantage of CoroNaTM Red is its quite low sodium-binding affinity (Kd(Na+ ) ∼200 mM), but it is sensitive enough to detect cellularNa+ influxes through voltage-gated ion channels and the pore of ATP-gated ionchannels.

PROTOCOL 2: QUANTIFICATION OF INTRACELLULARSODIUM ION CONCENTRATION


� Cardiomyocytes (e.g., H9c2(2-1) cells)� Quantitative single cell imaging cytometry� Cardiotoxic chemicals� CoroNaTM Red sodium indicator (molecular probes) in DMSO (1 mM)


Methods


2. Incubate cardiomyocytes with the drugs and incubate at 37◦C under 5%CO2a.

3. Completely remove the drug-containingmedium and suspend cells in HBSSor 1× PBS.

4. Dilute stock solution of CoroNaTM Red with 1× PBS and prepare a 1.0 �Mworking solution.

5. Add 500 �L of diluted CoroNaTM Red solution (1.0 �M) to the cells andincubate cells for 15 min at 37◦C.

6. Wash loaded cells twice with 1× PBS and detach cells with accutase for 15min at 37◦C

7. Collect the cells in a 1.5 mL tube and spin the cells at 230 rcf for 3 min in acentrifuge, discard supernatant, and wash with 1× PBS.

8. Resuspend in 10 �L of 1× PBS and analyze, monitor the fluorescenceimaging using single cell imaging cytometryb.

NotesaThe experimentator decides which drug concentration and incubation time is used.bCoroNaTM Red fluorescence indicator has absorption and fluorescence emission maxima at 554 and578 nm, respectively. CoroNaTM Red can be used to monitor mitochondrial sodium transport due tothe accumulation of CoroNaTM Red in mitochondria.

21.3.3 Cell Death: Apoptosis and Necrosis

During the last decade, a major shift in determining cell death dynamics mediated bychemotherapeutic drug toxicity has occurred. Aside from their crucial role in curingdisease, chemotherapeutic drugs can also induce severe side effects including nephro-toxicity, neurotoxicity, immunotoxicity, and sudden death [25]. This may be due topassive necrotic processes. In order to prevent aggravation of cancer, tumor cells maybe induced to undergo programmed cell death (apoptosis). Under physiological condi-tions apoptosis can occur without inflammation [26]. In contrast, pathophysiologicalsettings often induce inflammation due to the onset of secondary necrosis and thestimulation of chemokine and cytokine formation [27]. High concentrations of anti-cancer drugs, which are used to trigger apoptosis, may lead to cellular inflammationresulting in both apoptotic and necrotic cell death. The recent concept of multidrugresistance (MDR) has gained special importance in anticancer drug therapy. MDRresults from the ability of tumor cells to survive even after treatment with any ofa number of anticancer drugs. Patients suffering from MDR therefore experiencesevere damage to normal tissues due to high chemotherapeutic drug concentrations.In addition, the cytotoxic effects are mainly due to unspecific effects in both cancer


and normal cells even with the same stimulus. Recently, a hybrid form of cell death indegenerating neurons that falls in between apoptosis and necrosis has been described[28]. Therefore, if an oncotic (necrotic) process is fallaciously labeled as apoptosis,the emphasis is most likely put on the wrong signalingmechanism leading to a diseasestate, especially in the case of malignant cells.In the early phase of drug discovery, major stress is given to elucidating cell death

pathways to minimize the toxic effects of a candidate drug using in vivo and in vitromodels, but apoptotic detection using in vivo models is often costly due to lengthyexperimental procedures. Moreover, the heterogeneity and the short half-life of apop-totic cells limit quantitative measurements in in vivo model systems. On the otherhand, in vitro cell-based assays to investigate cell death dynamics are complicatedby the late phase of apoptosis, when the cell membrane becomes permeable to vitaldyes in the absence of phagocytes. Therefore, there is presently a need to developrapid diagnosis systems that are compatible with simultaneous discrimination of celldeath pathways to prevent the underestimation of apoptosis, which can result fromthe fast deterioration of early apoptotic cells to secondary necrotic cells. Most ofthe available techniques for discriminating between these cell death pathways arebased on either imaging or spectroscopic analysis. However, each of these techniqueshas its own advantages and disadvantages. Diagnosis systems based on glass fil-ters and broadband excitation sources bear a major problem in the form of a poorspectral resolution. Therefore, discriminating the mode of cell death using multiplefluorophores with these diagnosis systems may lead to misinterpretation of oncoticevents such as apoptosis. Other conventional spectroscopic techniques do not offersimultaneous discrimination of cell death pathways, as they provide signal detectionat every wavelength within a certain spectral range, but only for a single analyte spot.Flow cytometry is now routinely preferred as a rapid tool for elucidating apoptoticand necrotic events by recording the signal detected for each event with differentdetectors. Furthermore, none of the above techniques can provide live imaging ofcellular events.To overcome the limitations inherent to the individual techniquesmentioned above,

hybrid modulation of optical imaging with spectroscopy is utilized. To discriminatebetween apoptotic and necrotic cell death, our study mainly focused on a major,early event in apoptosis, the externalization of phosphatidylserine (PS) onto thecell membrane. By taking advantage of the strong coupling between biotin andstreptavidin, PS moieties were labeled with annexin-V (a calcium-dependent PS-binding 35-kDa protein)-biotin conjugated to the streptavidin-coated Quantum dot(hereafter called Annexin-Qdot). The visualization of the externalized PS moietieson the cellular membrane by photostable Qdot labeling was utilized as evidencefor apoptotic cellular events, whereas propidium iodide (PI) was used to visualizenecrotic cells. The difference in the fluorescence emission spectra of Qdots and PIwas used to discriminate between early apoptotic, late apoptotic, and necrotic cellsthrough AOTF-based real-time spectral imaging. The wavelength scanning of theAOTF hyperspectral imaging system is fast enough to provide a prompt and cleardistinction of the apoptotic cell before its rapid deterioration to secondary necrosis.


Protocol 3 specifically describes how to detect apoptosis and necrosis using PI andQdot. For quantification, the cells are detached from the well, labeled with apoptosisand necrosis indicators, measured by single cell imaging cytometry and analyzedwith MetaMorph.

PROTOCOL 3: CELL DEATH: APOPTOSIS AND NECROSIS


� Cardiomyocytes (e.g., H9c2(2-1) cells)� Quantitative single cell imaging cytometry� Annexin V-biotin� Propidium iodide (PI)� Streptavidin-conjugated Quantum dot (�max = 525 nm, Invitrogen)� Calcium-enriched binding buffer (BD Bioscience, San Jose, CA)

Methods


2. Incubate cardiomyocytes with drugs and incubate at 37◦C under 5% CO2a.

3. Completely remove the drug-containing medium and wash twice with 1×PBS.

4. Detach cells with accutase for 10 min at 37◦C.5. Collect the cells in a 1.5 mL tube and spin the cells at 230 rcf for 3 min in acentrifuge, discard supernatant, and wash with cold 1× PBS.

6. Redisperse the cells in 100 �L of calcium-enriched binding buffer.

7. Add 5 �L of annexin V-biotin for 15 minb.

8. Wash the cells with 1× PBS to remove unbound annexin V-biotin.

9. Resuspend the cells in binding buffer (100 �L).

10. Add 10 nM streptavidin-conjugated Qdotc as well as PI solution (10 �L)and incubate 15 min in the dark.

11. Wash the cells twice with 1× PBS and resuspend in 1× PBS (10 �L).

12. Monitor the fluorescence imaging using single cell imaging cytometryd.

NotesaThe experimentator decides which drug concentration and incubation time is used.bAnnexin V is a human protein (MW= 36 kDa). Annexin V and externalized PS on the cell membraneshow a high affinity for each other during early apoptosis.cTo avoid self-aggregation of Qdots, Qdot is dissolved in binding buffer and vortexed before use.dApoptotic cells are Qdot-positive, whereas necrotic cells are PI-positive.

APPLICATION OF HCS TO ANALYZE DRUG-INDUCED CARDIOTOXICITY 399

21.4 APPLICATION OF HCS TO ANALYZE DRUG-INDUCEDCARDIOTOXICITY BY USING A SINGLE CELL IMAGINGCYTOMETRY SYSTEM AND FLUORESCENT INDICATORS

Diverse ion channels are intimately related to each other. To understand the complexinteractions between ion channels and intracellular ions in the cell, multiple fluores-cent biomarkers indicating different targets should be simultaneously analyzed byHCS. Figure 21.3 describes cisapride- and digoxin-induced cytotoxicity. Cisapride,a commercial gastroprokinetic drug, was withdrawn from the market because it wasfound to bind to and block the potassium ion channel, especially the HERG chan-nel. Digoxin is a Na+ /K+ -dependent ATPase pump blocker. The rat cardiomyocytecell line H9c2(2-1) was treated with the two above-mentioned compounds for 12 or24 h. Subsequently, the permeability of the potassium ion channels and the sodiumion concentrations were determined in combined assays. By using single cell image

FIGURE 21.3 HCS of drug-induced changes in the permeability of the potassium channeland the concentration of intracellular sodium ion in H9c2(2-1) cells. Cisapride (a and c) is apotassium channel blocker, while digoxin (b and d) is a Na+ /K+ -ATPase pump blocker. Thequantitative data (c and d) show that the permeability of the potassium channels diminishedafter cisapride treatment and the concentration of the sodium ion increased after digoxintreatment. Figures are reproduced from Reference 13 with permission from the Royal Societyof Chemistry. (See insert for color representation of this figure.)


cytometry, the cytotoxicity of each drug was quantified in H9c2(2-1) cells. P(%)represents the percentage of fluorescent-tagged single cells, which is calculated usingthe following equation:

P(%) = (P1/P2)× 100

P1, the percentage of normalized fluorescence of indicators in drug-treated singlecells against that in the total cell control (drug-untreated cells).

P2, the percentage of normalized fluorescence of indicators in drug-treated singlecells against that in total cells.

Cisapride inhibited potassium ion channel activity concentration dependently(>25 nM), but did not affect the concentration of intracellular sodium ion withina range of 25 to 200 nM. The IC50 of cisapride obtained using single cell imagingcytometry was similar to that obtained using the patch-clamp technique. Digoxin onlyinduced the increase of sodium ion concentration at concentrations >1 nM. Takentogether, single cell imaging cytometry was able to provide accurate informationon potassium ion channel activities and intracellular sodium ion concentrations incombined assays.Figure 21.4 shows an example for the combined determination of cell death

forms and intracellular sodium ion levels. H9c2(2-1) cells were incubated in

FIGURE 21.4 Combined use of assays to determine cell death and the concentration ofsodium ion in cardiomyocytes. The fluorescence spots of the sodium ion completely overlapwith those of necrotic cells. The figure is reproduced from Reference 13 with permission fromthe Royal Society of Chemistry. (See insert for color representation of this figure.)

REFERENCES 401

camptothecin-containing media for 6 h. Thereafter, the sodium ion levels and celldeath forms were determined. Intracellular sodium ion, apoptosis, and necrosis weredetected at 579, 525, and 619 nm. As shown in Figure 21.4, an increment of thesodium ion levels was associated with necrotic cells.

REFERENCES

1. Griffith,W.H., Jasek,M.C., Bain, S.H.,Murchison,D. (2000).Modification of ion channelsand calcium homeostasis of basal forebrain neurons during aging. Behavioural BrainResearch, 115, 219–233.

2. Bers, D.M. (2002) Cardiac excitation-contraction coupling. Nature, 415, 198–205.

3. Postema, P.G., van Dessel, P.F., Kors, J.A., Linnenbank, A.C., van Herpen, G., Ritsemavan Eck, H.J., van Geloven, N., de Bakker, J.M., Wilde, A.A., Tan, H.L. (2010). Localdepolarization abnormalities are the dominant pathophysiologic mechanism for type 1electro-cardiogram in brugada syndrome a study of electrocardiograms, vectorcardio-grams, and body surface potential maps during ajmaline provocation. Journal of theAmerican College of Cardiology, 55, 789–797.

4. Fermini, B., Fossa, A.A. (2003). The impact of drug-induced QT interval prolongation ondrug discovery and development. Nature Reviews in Drug Discovery, 2, 439–447.

5. Keating, M.T., Sanguinetti, M.C. (2001). Molecular and cellular mechanisms of cardiacarrhythmias. Cell, 104, 569–580.

6. Sanguinetti, M.C., Tristani-Firouzi, M. (2006). hERG potassium channels and cardiacarrhythmia. Nature, 440, 463–469.

7. Nattel, S., Carlsson, L. (2006). Innovative approaches to anti-arrhythmic drug therapy.Nature Reviews in Drug Discovery, 5, 1034–1049.

8. Gussak, I., Litwin, J., Kleiman, R., Grisanti, S., Morganroth, J. (2004). Drug-inducedcardiac toxicity: emphasizing the role of electrocardiography in clinical research and drugdevelopment. Journal of Electrocardiology, 37, 19–24.

9. Roy, M., Dumaine, R., Brown, A.M. (1996). HERG, a primary human ventricular targetof the nonsedating antihistamine terfenadine. Circulation, 94, 817–823.

10. Gintant, G.A., Limberis, J.T., McDermott, J.S., WegnerC.D., CoxB.F. (2001). The caninePurkinje fiber: an in vitro model system for acquired long QT syndrome and drug-inducedarrhythmogenesis. Journal of Cardiovascular Pharmacology, 37, 607–618.

11. Rajamani, S., Studenik, C., Lemmens-Gruber, R., Heistracher, P. (2000). Cardiotoxiceffects of fenfluramine hydrochloride on isolated cardiac preparations and ventricularmyocytes of guinea-pigs. British Journal of Pharmacology, 129, 843–852.

12. Dunlop, J., Bowlby, M., Peri, R., Vasilyev, D., Arias, R. (2008). High-throughput elec-trophysiology: an emerging paradigm for ion-channel screening and physiology. NatureReviews Drug Discovery, 7, 358–368.

13. Kim, M.J., Lee, S.C., Pal, S., Han, E., Song, J.M. (2011). High-content screening of drug-induced cardiotoxicity using quantitative single cell imaging cytometry on microfluidicdevice. Lab on a Chip, 11, 104–114.

14. Naoghare, P.K., Kim, M.J., Song, J.M. (2008). Uniform threshold intensity distribution-based quantitative multivariate imaging cytometry. Analytical Chemistry, 80, 5407–5417.


15. Ki, H.A,Naoghare, P.K., Oh, B.K., Choi, J.W., Song, J.M. (2009). Nondestructive quantumdot-based intracellular serotonin imaging in intact cells. Analytical Biochemistry, 388, 23–27.

16. Dennis, A., Wang, L., Wan, X., Ficker, E. (2007). hERG channel trafficking: novel targetsin drug-induced long QT syndrome. Biochemical Society Transactions, 35, 1060–1063.

17. Ficker, E., Kuryshev, Y.A., Dennis, A.T., Obejero-Paz, C., Wang, L., Hawryluk, P., Wible,B.A., Brown, A.M. (2004). Mechanisms of arsenic-induced prolongation of cardiac repo-larization. Molecular Pharmacology, 66, 33–44.

18. Cordes, J.S., Sun, Z., Lloyd, D.B., Bradley, J.A, Opsahl, A.C., Tengowski, M.W., Chen,X., Zhou, J. (2005). Pentamidine reduces hERG expression to prolong the QT interval.British Journal of Pharmacology, 145, 15–23.

19. Woosley, R.L., Chen, Y., Freiman, J.P., Gillis, R.A. (1993). Mechanism of the cardiotoxicactions of terfenadine. JAMA, 269, 1532–1536.

20. Chen, J., Seebohm, G., Sanguinetti, M.C. (2002). Position of aromatic residues in theS6 domain, not inactivation, dictates cisapride sensitivity of HERG and eag potassiumchannels. Proceedings of The National Academy of Sciences USA, 99, 12461–12466.

21. Gonzalez, J.E., Oades, K., Leychkis, Y., Harootunian, A., Negulescu, P.A. (1999). Cell-based assays and instrumentation for screening ion-channel targets.Drug Discovery Today,4, 431–439.

22. Ruan, Y., Liu, N., Priori, S.G. (2009). Sodium channel mutations and arrhythmias. NatureReviews in Cardiology, 6, 337–348.

23. Amin, A.S., Asghari-Roodsari, A., Tan, H.L. (2010). Cardiac sodium channelopathies.Pflugers Archiv, 460, 223–237.

24. Wang, X.Q., Xiao, A.Y., Sheline, C., Hyrc, K., Yang, A., Goldberg, M.P., Choi, D.W.,Yu, S.P. (2003). Apoptotic insults impair Na+ , K+ -ATPase activity as a mechanism ofneuronal death mediated by concurrent ATP deficiency and oxidant stress. Journal of CellScience, 116, 2099–2110.

25. Efferth, T., Kaina, B. (2010). Toxicity of the antimalarial artemisinin and its derivatives.Critical Reviews in Toxicology, 40, 405–421.

26. Elgrabli, D., Abella-Gallart, S., Robidel, F. Rogerieux, F., Boczkowski, J., Lacroix, G.(2008). Induction of apoptosis and absence of inflammation in rat lung after intratrachealinstillation of multiwalled carbon nanotubes. Toxicology, 253, 131–136.

27. Ziebell, J.M., Morganti-Kossmann, M.C. (2010). Involvement of pro- and anti-inflammatory cytokines and chemokines in the pathophysiology of traumatic brain injury.Neurotherapeutics, 7, 22–30.

28. Northington, F.J., Chavez-Valdez, R., Martin, L.J. (2011). Neuronal cell death in neonatalhypoxia-ischemia. Annals of Neurology, 69, 743–758.

22HIGH-THROUGHPUT SCREENINGASSAYS TO EVALUATE THECARDIOTOXIC POTENTIALOF DRUGS

Carl-Fredrik Mandenius and Thomas Meyer

22.1 INTRODUCTION

Pharmacological agents can cause a diverse range of cardiac toxicities, such as alter-ations in the ability of the heart to contract and/or relax, changes in cardiac rhythm,alteration in blood pressure and ischemia [1]. As a consequence, cardiotoxicity hasbecome one of the main reasons for drug attrition. An example is the withdrawal ofthe nonsteroidal anti-inflammatory drug Vioxx (Merck) due to cardiovascular safetyconcerns. The withdrawal of this late-stage candidate caused substantial loss of rev-enues as well as patient litigations and underscored the need for better and morepredictive safety testing [2]. Thus, cardiotoxicity testing is among the essential safetypharmacology measures in the nonclinical and clinical safety testing of new drugs.This is further sustained by regulatory regulations and guidelines. Cardiac safety

pharmacology evaluation is a core subject of the ICH S7A guideline on “SafetyPharmacology Studies for Human Pharmaceuticals” [3]. The ICH S7A guidelineaims to reveal disrupted organ functions, which are not readily detected by standardtoxicological testing. This includes evaluation of endpoints related to the function ofthe heart.A more cardiac-specific guidance concerns nonclinical evaluation of the potential

of a drug to induce delayed ventricular repolarization. This is addressed in detail inthe ICH S7B guideline, recommending a testing strategy based on electrophysiology


403

404 SCREENING ASSAYS TO EVALUATE THE CARDIOTOXIC POTENTIAL OF DRUGS

TABLE 22.1 Methods for In Vitro Toxicity Testing of Cardiac Cells

In VitroMethodology Principle

ResponseVariable(s)

HTSCapacity Reference

Electrophysiology evaluation:Microelectrodearray(MEA)

Records the QT wave fromelectrodes attached on cellsurface (Multi ChannelSystems)

Field potential(volt)

+++ [17]

Automated patch clamp + [6]

Oxygen uptake rate:Fiber opticalsystem

Quenching by O2 in mediasurrounding cell inmulti-well plates (PreSens)

Dissolved oxygenchange (mgO2/min)

++ [14]

Robotic fibersystem

Quenching by O2 in media surrounding cell inmulti-well plates (Seahorse Bioscience)

+++ [13]

Troponin release:Elecsys R© Immunochemiluminescence

method automated for HTSSpectrometricrecording(luminescence)from conjugatedantibodies

+++ [12]

Surfaceplasmonresonance(SPR)

Target molecule binds to IgGon sensor chips. Massdetected by SPR

Resonance signal +++ [35]

Spectrophotometric assays:Alamar blue +++Sulforhodamine B +++

studies in vitro as well as in vivo [4]. This is further supported by the ICH E14 guide-line on clinical evaluation of QT/QTc interval prolongation by non-antiarrhythmicdrugs [5].Several toxicological mechanisms should be addressed to predict contractile tox-

icity by screening methods. It has been suggested that in silico, in vitro and ex vivobased methods could reveal toxic effects induced by the identified mechanisms byaddressing multiple parameters such as morphological changes in the cytoskeleton,nucleus, and mitochondria of single cells, by assessment of functional (mechanical)parameters such as length of single cell contractions and by addressing biochemicalparameters such as single cell DNA fragmentation and release of LDH, troponin,and annexin [1]. Table 22.1 provides an overview of methods available for test-ing. Of these, electrophysiology testing methods are pivotal for meeting the clinical,toxicological, and regulatory requirements.The severity of the cardiotoxic effects, the regulatory requirements regarding

testing, and the consequences for the drug development process underscore the need

INTRODUCTION 405

Toxicant

Cellsource

PredictionPreparationof cardiac cells

Exposureof toxicant

HTS

1 2 3

HTS

Monitorningof response

Evaluationof response

• Cell type• Cell number• Microstructure• Cell condition

• Exposure time• Amounts• Concentration• Volumes

• Field potential• Respiration• Biomolecular

• Statistical method• Software

FIGURE 22.1 Flow of procedures in high-throughput in vitro screening for cardiotoxicity.The HTS may include the whole flow (outer dotted line) or the measurement instrumentationonly (inner dotted line).

for reliable and efficient screening methods for cardiotoxicity testing. The amountof tests, the substantial evaluation work accompanying the testing, and the urge ofproviding test results early in the drug development process emphasize the value ofhigh-throughput screening (HTS) methods. Figure 22.1 illustrates the process flowof a typical HTS method. The greatest benefit is achieved if the whole flow is undera high-throughput regime. However, in most cases the high-throughput operationsare only applied to the instrumental part of the flow (e.g., an automated analyzer orrobotic unit), while the preparation phase and finishing phase are carried out manuallyand then become bottlenecks of the procedure. Moreover, the preparation and culturephases are normally not as easy to automate as measurements and liquid handling.The advantages of an HTS method can only be used to its full extent, if the wholeassay chain is carried out as an HTS procedure.It could be advocated that the need for HTS for cardiotoxicity testing is not as

high as with other targets. In particular, this could be true when front-loaded safetypharmacology does not aim to screen whole compound libraries but rather kicksin at the level of secondary screening. The aim here is rather to obtain medium-throughput assays with a high level of predictivity and reliability rather than focusingon ultrahigh-throughput.Of high interest is the patch clamp technique. Patch clamp is considered the gold

standard in electrophysiological assessment of cardiac safety pharmacological testingand is used to confirm activity against ion channel targets and in regulatory requestedsafety pharmacological assessments [4]. Voltage- and ligand-gated channels can betested by manual and automated patch clamp with highly specific protocols [6]. Theassay is versatile and selective for a particular ion channel. However, as mentionedbefore, the bottleneck of this classical electrophysiology test is the low throughputand high costs per data point, although the introduction of automated patch clamptechnologies has increased the throughput and helped to reduce the cost per datapoint [7].


Another method of relevance is field potential QT scanning. Modulation of thehuman surface electrocardiogram (ECG), particularly the prolongation of the QTinterval, are considered as important risk factors for ventricular tachycardia of theTorsade de Pointes (TdP) type [8]. The mechanism behind the prolongation of theQT interval is a disturbance in the repolarization of the ventricular action potential.This can be caused by drugs binding to potassium channels and blocking the ion fluxthrough those channels [9].A few QT scanning instruments are commercially available. Some of these are

adapted for HTS and others could be further developed into HTS, so that, conse-quently, they are potentially interesting. In the case of the xCELLigence instrument(RocheApplied Science, Indianapolis, IN, andACEA Inc., SanDiego, CA, USA), thepresence of the cells on top of the electrodes will affect the local ionic environmentat the electrode/solution interface, leading to an increase in the electrode impedance[10]. The more the cells are attached to the electrodes, the larger the increases inelectrode impedance. In addition, the impedance depends on the quality of the cellinteraction with the electrodes. For example, increased cell adhesion or spreading willlead to a larger change in electrode impedance. Thus, electrode impedance, whichis displayed as cell index (CI) values, can be used to monitor cell viability, number,morphology, and adhesion degree in a number of cell-based assays. However, whileimpedance recording allows monitoring cell proliferation and beating rate, there isno information on cardiac action potential shape and duration available.In the case of themicroelectrode array (MEA) system fromMulti Channel Systems

(Reutlingen, Germany) recording is carried out with integrated metal electrodes thatsense the electrical field potentials of spontaneously beating or electrically stimulatedclusters of human embryonic stem cell-derived cardiomyocytes (hESC-CMs) plateddirectly on the MEA dishes (Fig. 22.2). The cells can be cultured for up to 2 weekson the MEA dish and subsequently long-term recordings allow monitoring of thedifferentiation and maturation process [11].Both systems have potential for parallelization and further miniaturization. With

theMEA2100 fromMulti Channel Systems it is possible to record up to 240 channels.By doing this, one can record data from 24 wells in parallel. However, one is notlimited to multi-well solutions. Simply by using different MEAs, the setup can befurther optimized. An up-scaled throughput is available by using multi-well sensorplates (Nipro, Osaka, Japan) with up to 96 wells recorded in parallel.Automated immunoanalytical HTS methods for cardiac markers such as troponin,

myosin, and LDH are available. An example is the Elecsys R© system from RocheDiagnostics (Mannheim, Germany) that analyses troponin T (TnT) in an automatedimmunochemiluminescence assay with applied HTS capacity [12].Another system of interest for testing general cytotoxicity effects in cardiac cells

are instruments for respiration monitoring. The XF Extracellular Flux Analyzer fromSeahorse Bioscience (North Billerica, MA, USA) is a fully integrated instrument thatcan simultaneously measure the cellular respiration of major energy yielding path-ways of cells inmicroplate format [13]. This fast and sensitivemeasurement of cellularbioenergetics is label-free, enabling time-resolved analysis and the reuse of the cells.The instrument allows to perform assays that provide increased throughput for drug

INTRODUCTION 407

(a)

FIGURE 22.2 (a) Microelectrode arrays showing a microelectrode array (MEA) well; (b) arobot for measuring 96-well plates with microelectrodes; (c) a micrograph of microelectrodesattached to cardiomyocytes; (d) recordings of the field potential for a drug at 11 differ-ent concentrations using the MEA device (reproduced with permission from Multi ChannelSystems GmbH).

testing. Its microplates are specifically designed to assay the cellular metabolism ofliving cells, isolatedmitochondria, and pancreatic islets. Themicroplates are designedto maximize the sensitivity of the sensor cartridge measurements. A similar instru-ment is available from PreSens GmbH (Regensburg, Germany), which is based onthe fluorescence detection of oxygen in microtiter plates [14]. These will be furtherdescribed below.A common feature of the in vitro testing methods mentioned above is the design

and development of the test method and its protocol. This involves the selection ofthe cell source, the preparation of a cell culture that is considered relevant for testing,a procedure for the transfer of the cultured cells into a sufficiently stable state fortesting (e.g., a 96-well plate or a microfluidic device), a method for the reproducibleexposure of the cells to the toxicants, a recording method and an evaluation procedurefor interpreting the recorded signals (Fig. 22.1). As indicated in Figure 22.1, severalissues are critical when wanting to establish a sufficiently reliable testing method. In


(b)

(c)

FIGURE 22.2 (Continued)

THE MICROELECTRODE ARRAY SYSTEMS 409

-100 0 100 200 300 400 500

0

CTRL

300 pM

1 nM

3 nM

10 nM

30 nM

100 nM

300 nM

1 μM

3 μM

10 μM

30 μM

μV

t/ms

(d)

FIGURE 22.2 (Continued) (See insert for color representation of Figure 22.2d.)

particular, these issues include cell source, selection of recorded variable, sensitivityof the detection principle as well as quality and capacity of the evaluation method.Access to primary human cardiac cells is always very limited. Animal cells are

an alternative. Their relevance varies with the species and in most cases data areunsatisfactory from a regulatory point of view. Use of human stem cell derivedcells has become a significantly more attractive alternative [15–17]. However, car-diomyocytes derived from human embryos via hESC lines are ethically questionable.Reprogrammed induced pluripotent stem cells (iPSCs) are therefore a much betteralternative, especially when the amount of cell material needed is large as in the caseof HTS protocols [18, 19].At the present time methods to differentiate the human iPSCs into specific cell

types advances considerably. It seems plausible that in a near future iPSC-derived car-diomyocyte sources will be broadly available, thereby facilitating in vitro cardiotox-icity drug testing. Basic iPSC-derived cardiac cells are available already today, forexample, the I-CellTM cardiomyocyte line (Cellular Dynamics International, Madi-son, WI, USA). Further commercially available stem cell-derived cardiomyocytesare the cell lines hES-CMC 002 (Cellartis AB, Goteborg, Sweden) and the ReproCardio 2 (Reprocell Inc., Yokohama, Japan) [20–22].

22.2 THE MICROELECTRODE ARRAY SYSTEMS

As mentioned above, modulation of the human surface ECG, particularly the prolon-gation of the QT interval, are considered to be important risk factors for ventriculartachycardia of the TdP type [8].


Since TdP can be lethal, all new drugs need to be tested both in vitro and in vivoregarding their potential to prolong the QT interval prior to entering clinical trials.The mechanisms underlying QT prolongation reach from the molecular interactionof the drug molecule to the channel protein all the way to the ECG measured basedon electrical phenomena of the whole organ. The assays used to identify arrhythmo-genic lead compounds involve measuring molecular-, cell-, tissue-, and organ-basedparameters in vitro and in vivo.The target cell type is the adult human ventricular cardiomyocyte (CM) of the

working myocardium. Alternatives include animal cells and non-CM cells heterol-ogously expressing the human channel proteins of the relevant ion channels (e.g.,hERG). However, as the expression pattern of the ion channels recruited for repolar-ization widely differs across species, the predictivity of nonhuman cell-based assaysis limited [23, 24].The action potential duration is determined by a multitude of ion channels and the

spatial compartmentalization of functional units within the CMs [25]. The access tohuman CMs with an adult ventricular phenotype could make this analytical mecha-nism a valuable tool for cardiotoxicity testing. CMs created from hESCs have recentlybeen successfully tested in field potential assays using MEAs and multiplexed MEAs(QT well plates) [14, 26, 27].The MEA recording is carried out with integrated metal electrodes that sense the

electrical field potentials of spontaneously beating or electrically stimulated clustersof hESC-CM plated directly on the MEA dishes (Fig. 22.2). The cells can be cul-tured for up to 2 weeks on the MEA dish and subsequently long-term recordingsallow monitoring of the differentiation and maturation process. Once a sufficientlyadult phenotype has been established, a cumulative dose–response experiment withincreasing concentrations of a pharmacologically relevant drug is performed.In order to increase the throughput, it is possible to perform the assay in a multi-

well layout with a limited number of electrodes per well instead of the microelectrodesingle-well approach. To establish an industry standard high-throughput compatibleplatform, a 96-well format that matches the standards defined by the Society forBiomolecular Screening (SBS) can be used. To automate the assay, a liquid handlingrobot can be added to and integrated in the electrophysiological sensing platform. Tofurther enhance the convenience of the assay, a data acquisition and analysis softwaremay assist the user to obtain report sheets with dose–response curves and analysis ofthe proarrhythmic potential of the drug.The method has been evaluated for drugs with specific affinity to various ion

channels involved in the cardiac action potential (lidocaine, nifedipine, E-4031)[27]. In addition, safe control compounds (acetylsalicylic acid), clinically relevantdrug classes such as antiarrhythmics and psychopharmacologically active drugs aswell as drugs with affinity to multiple ion channels have been tested. The latterrepresent a group of drugs that is considered to be a challenge in meeting the safetypharmacological requirements regarding toxicity assays. A selection of compoundsthat have been tested using MEA and QT screen technology is shown in Figure 22.3.Lidocaine, a classical sodium channel blocker, showed a dose-dependent block ofthe rapid depolarizing component of the cardiac field potential (Fig. 22.3a). This is

(a)

(b)

(c)

(d)

(e)

FIGURE 22.3 Field potential recordings with the MEA and the hESC-derived cardiomy-ocyte cell line hES-CMC 002. MEA data for: (a) lidocaine, (b) quinidine, (c) nifedipine, (d)amiloride and (e) acetaminophen. The open dots indicate the relative field potential durationbased on the scale on the right side. The filled dots represent the relative block of depolarizationpeak indicated by the axis on the left side of the diagrams; 100% refers to control conditionsin the absence of the drug. Overlay plots in the left panel illustrate modulation of the signalshape over increasing compound concentrations as indicated by the color code (reproducedfrom Reference 27 with permission from John Wiley & Sons, Ltd.). (See insert for colorrepresentation of this figure.)


in line with data obtained from primary cells and voltage clamp recording in hESC-CM. Quinidine, a class 1A antiarrhythmic drug with known blocking effects on thesodium and repolarizing potassium channels, showed a dose-dependent prolongationof the cardiac field potential recorded by extracellular MEA electrodes (hERG block)and a suppression of the amplitude of the depolarizing component (sodium channelblock; Fig. 22.3b). Nifedipine, a blocker of the cardiac L-type calcium channelknown to be responsible for the plateau phase of the cardiac potential, resulted ina shortening of the field potential (Fig. 22.3c). Amiloride was used to validate theselectivity of the assay, since it is known to be a selective blocker of epithelial sodiumchannels. As shown in Figure 22.3d, the sodium channels expressed in hESC-CM areinsensitive to amiloride. This is a good indicator that the cardiac subtype of sodiumchannels is actually expressed in the hESC-CM. At very high concentrations (over100 �M amiloride), unspecific effects on the repolarization and reduction of beatingfrequency are observed. Acetaminophen (paracetamol) was used as a negative control(Fig. 22.3e). No modulatory effect on the cardiac field potential was observed.However, it should be noted that not all drug testing results yielded clear results

and the results were not always in line with the data obtained from primary CMsand clinical data. This observation may, in part, be explained by a mixture of cardiacphenotypes present in the cultures as well as by hESC-CM being present at differentmaturation stages at the time of analysis.Simply having spontaneous beating as the criteria for inclusion of the hESC-CM

preparation may bemisleading andmore precise parameters are most likely needed toacquire homogenous populations of cardiac phenotypes. Technical issues regardingdrug diffusion and penetration into a 3D cellular aggregate may also compromise theinterpretation of the results. Nevertheless, as the examples above show, the hESC-CM-based MEA assay possesses the potential for automation and up-scaling. However,additional work and standardization are required to establish this model as a validatedsafety pharmacology assay.When using the hESC-derived cardiomyocyte cell line HES03 (University of Lei-

den, The Netherlands), similar results were obtained [26]. Monitoring of extracellularfield potentials using the MEA in this cell line and the generation of derivative fieldpotential duration (FPD) values showed clear dose-dependent responses for 12 cardiacand noncardiac drugs. Figure 22.4 exemplifies this for six of the compounds tested.In these cases, patient serum levels of drugs with known effects on QT intervaloverlapped with prolonged FPD values derived from these hESC-CMs, as predicted.The investigators concluded that FPD prolongation observed in these hESC-CMs isa safety criterion for preclinical evaluation of new drugs in development. The studydemonstrates the utility of MEA for monitoring dose responses toward a wide rangeof compounds in hESC-CMs and shows its utility to predict clinical effects.Induced pluripotent stem cell-derived cardiomyocytes (iPSC-CMs) have also been

used [18, 19]. In a recently published report [21] a high-throughput cardiotoxic-ity assay was described, in which a monolayer of beating human iPSC-CMs wasused to detect drug-induced effects. The assay employed an electrophysiologicaltechnique, whereby assessment of impedance measurements was carried out in 96-well plates for recording the rhythmic, synchronous contractions of the iPSC-CMs.

(a)

(b)

(c)

(d)

(e)

(f)

FIG

UR

E22

.4FieldpotentialrecordingswiththeMEAandthehumanembryonicstem

cell-derivedcardiomyocytecelllineHES03from

the

UniversityofLeiden(Leiden,TheNetherlands).(a)Sodium

peakamplitude–doserelationshipforhESC-CMinthepresenceofincreasingamountsof

lidocaine.(b)Exampleoffieldpotentialrawdata;notethereductionofamplitudeathigherconcentrationsoflidocaine.(c)FPD–doserelationshipfor

hESC-CMinthepresenceofincreasingamountsofnifedipine.(d)Exampleoffieldpotentialrawdata;notethereductionoffieldpotentialduration

athigherconcentrationsofnifedipine.(e)FPD–doserelationshipforhESC-CMinthepresenceofincreasingamountsofE-4031.(f)Exampleof

fieldpotentialrawdata;notetheprolongationoffieldpotentialdurationinresponsetoE-4031(reproducedfrom

Reference26withpermissionfrom

Elsevier).(

See

inse

rtfo

rco

lor

repr

esen

tati

onof

this

figur

e.)

413


Twenty-eight compounds with known cardiac effects were assessed with the assay,thereby yielding data on compound-specific changes in the beat rate and/or the ampli-tude of the impedance. These impedance changes were comparable with the resultsfrom electric field potential assessments obtained in the MEA assay. In the study anindex value for drug-induced arrhythmias was calculated, which enabled the determi-nation of the proarrhythmic potential of each of the tested compounds. In conclusion,the system provides an additional tool for HTS in cardiac safety pharmacology.The planar patch clamp technique (e.g., from Sophion, Nanion, or Cytocentrics)

has recently been applied as an automated assay. Stoelzle and colleagues [28] used thisapproach with mouse ESC-derived cardiomyocytes in suspension for cardiotoxicityprediction. The study used MEAs for recording cardiac ion currents as well asactions in stem cell-derived cardiomyocytes (Cor.At R© cardiomyocytes, Cat. No.XCAC-1010; Lonza, Walkersville, MD, USA). Besides monitoring inhibitory effectsof tested compounds on typical cardiac ion currents, the study also recorded drug-induced modulation of cardiac action potentials in an automated patch clamp system.This combination of an in vitro cardiac cell model with a HTS patch clamp screeningseems to be a useful approach for time and cost-effective cardiotoxicity prediction.

22.3 THE USE OF CELLULAR OXYGEN UPTAKE RATES FORMONITORING THE STATE OF CARDIOMYOCYTES

One of the requirements listed in ICH S7B is the determination of the cellular respi-ratory rate. The oxygen uptake rate (OUR) or respiration provides direct informationon the metabolic activity in cells. Although the techniques used to measure molec-ular oxygen by luminescence have been known for a long time, it is only recentlythat convenient and cost-effective techniques have become available and have pro-vided opportunities for applications in toxicology as well as in other biomedicaland biotechnological fields [29]. Commercially available devices, for example, theOxoplates R© and OxoDishes R© from PreSens, have been reported for respirometricscreening with in vitro toxicity testing purposes [14, 30, 31]. These devices have anexcellent potential to be scaled up for HTS (see Fig. 22.5a).The respiration of the cells is assessed by monitoring the oxygen level in 0.1–

1 mL wells with an oxygen-sensitive reagent embedded in a paste at the bottom ofthe wells. This reagent detects the oxygen level (as well as pH level) by quenching theemission from the spot. This is done noninvasively through the transparent bottom.SensorDishes for oxygen (OxoDishes R© and Oxoplates R©) are available in 24-well and6-well format for screening purposes. The luminescence intensity from the sensorspot in the Oxoplates R© can be measured with any available plate reader; in the case ofthe OxoDishes R© the luminescence lifetimemeasurements are carried out in a 15× 10cm device (Fig. 22.5a) provided by the supplier Presens (SensorDish R© Reader, SDR).The device can easily be placed in a cell culture incubator, thereby allowing controlledatmosphere with respect to temperature, humidity, and CO2 concentration. The setupwith the SensorDish R© micro-plate reader with 24 optical sensors integrated into thewells of a 24-well multi-dish (OxoDish R©) is shown in Fig. 22.5a. The SDR method

THE USE OF CELLULAR OXYGEN UPTAKE RATES FOR MONITORING 415

(a)

(b)

FIGURE 22.5 (a) OxoDish R© placed in the SDR SensorDish R© Reader (PreSens GmbH). (b)Dose–concentration curves for verapamil tested in HL-1 cardiomyocytes. EC50 values werecalculated in the respiration and SRB assay (reproduced from Reference 27 with permissionfrom John Wiley & Sons, Ltd.).

has successfully been used with the cardiomyocyte cell line HL-1 [14, 27]. Sixteencardiotoxic compounds were screened in HL-1 cardiomyocytes [32] using 24-wellOxoDishes R©. The kinetic respiration assay was compared with a sulforhodamine B(SRB) assay as reference (Fig. 22.5b). The SDR and the SRB assays gave comparableresults in terms of EC50 values (median effective concentration of a drug) for themajority of the tested compounds.


An alternative device for oxygen uptake rate monitoring of cardiomyocytes is theExtracellular Flux Analyzer from Seahorse Bioscience [33, 34]. The 24-well micro-plate of the device enables real-time measurement of cells attached to the bottom ofthe wells. The applied protocol of the plate (XF24 Islet CaptureMicroplate) facilitatesa time-efficient coating procedure and allows efficient perfusion of reagents and testsubstances in the 24-well microplate. Typically, the oxygen consumption of 50–70cell islets of cells per well is monitored subsequently in real-time using a similar fiberfluorometric technique as described above.

22.4 SURFACE PLASMON RESONANCE FOR TESTING OF TROPONIN

The expression of TnT, a useful biomarker for studying drug-induced toxicity effectsin cardiac cells, is monitored by using immunoassay protocols. Recently, an assayprocedure suitable for HTS based on surface plasmon resonance (SPR) immunosens-ing was used to monitor release of TnT at-line from active HL-1 cardiomyocytesin vitro when exposed to cardiotoxic substances [35]. Monoclonal human TnTantibodies were used in the SPR sensor for analysis of TnT release. The attaineddetection limit of TnT was 30 ng/mL in a direct assay setup and 10 ng/mL in asandwich format. Exposure of the HL-1 cardiomyocytes to cardiotoxic substancessuch as doxorubicin, troglitazone, quinidine, and cobalt chloride for 6 to 24 h gavesignificant SPR responses, while substances of low cardiotoxicity showed insignifi-cant responses (e.g., ascorbic acid and methotrexate). The SPR results were verifiedwith a certified immunochemiluminescence method (Elecsys R©, see above) exhibit-ing a correlation factor of r2 = 0.790. The troponin SPR assay has been adaptedto a protocol using hESC-CM [36]. The protocol has also been used to monitorthe release of another clinically relevant cardiac biomarker, fatty acid binding pro-tein 3, after exposing hESC-CM to the well-known cardioactive drug compounddoxorubicin. The hESC-CM protocol provides an opportunity to surpass previousmethods, based on rodent cells or cell lines due to its significantly higher tox-icological relevance. These SPR biosensor approaches seem particularly appro-priate for HTS due to their capacity for parallel and high-throughput analysis incomplex media.

22.5 CONCLUSIONS

At present, there are several technological opportunities for implementing HTSmeth-ods in safety pharmacology. The most prominent are accounted for in this chap-ter (Table 22.2). Further sensor technological advancements have been reported[37, 38] although not yet fully validated. As pointed out above, a few criti-cal issues remain to be further developed in order to sustain functional HTSprotocols. In particular this includes (1) the generation and supply of suffi-cient quantities of relevant cells from well characterized cell banks, and (2)the access to patient specific models based on the use of cardiomyocytes from

REFERENCES 417

TABLE 22.2 Drugs Analyzed with Electrophysiology Instrumentation onCardiomyocytes

Compound Method Cell Line Data Reference

Quinidine MEA hES-CMC 002 [11]Lidocaine MEA hES-CMC 002 [11]Nifedipine MEA hES-CMC 002 [11]E-4031 MEA hES-CMC 002 [11]Amiloride MEA hES-CMC 002 [11]Acetaminophen MEA hES-CMC 002 [11]Cisapride MEA HES-3 derived CM [15]Sparfloxacin MEA HES-3 derived CM [15]Terfenadine MEA HES-3 derived CM [15]Domperidone MEA HES-3 derived CM [15]Sertindole MEA HES-3 derived CM [15]Alfuzosin MEA iPSC-CM (i-Cell) [16]Cisapride MEA iPSC-CM (i-Cell) [16]Dofetilide MEA iPSC-CM (i-Cell) [16]Erythromycin MEA iPSC-CM (i-Cell) [16]Flecainide MEA iPSC-CM (i-Cell) [16]Quinidine MEA iPSC-CM (i-Cell) [16]Thioridazine MEA iPSC-CM (i-Cell) [16]Terfenadine MEA iPSC-CM (i-Cell) [16]

selected patient groups, for example, from patients with the long QT syndrometype 2 [39, 40].

ACKNOWLEDGMENTS

The authors wish to thank for contributions by the Invitroheart EU project (LSHB-CT-2007-037636), Dr. Udo Kraushaar, Nipro Inc. and Cellular Dynamics InternationalInc.

REFERENCES

1. Stummann, T., Beilmann, M., Duker, G., Dumotier, B., Fredriksson, J.M., Jones, R.L.,Hasiwa, M., Kang, J., Mandenius, C.F., Meyer, T., Minotti, G., Valentin, J.P., Zunckler,B.J., Bremer, S. (2009). Report and recommendations of the workshop of the EuropeanCentre for the validation of alternative methods for drug-induced cardiotoxicity. Cardio-vascular Toxicology, 9, 107–125.

2. Bresalier, R.S., Sandler, R.S., Quan, H., Bolognese, J.A., Oxenius, B., Horgan, K., Lines,C., Riddell, R., Morton, D., Lanas, A., Konstam,M.A., Baron, J.A. (2005). Cardiovascularevents associated with rofecoxib in a colorectal adenoma chemoprevention trial. NewEngland Journal of Medicine, 352, 1092–1102.


3. ICH (The International Conference on Harmonisation of Technical Requirements forRegistration of Pharmaceuticals for Human Use). (2000). Safety pharmacology studiesfor human pharmaceuticals. ICH Safety Guideline S7A.

4. ICH (The International Conference on Harmonisation of Technical Requirements forRegistration of Pharmaceuticals for Human Use). (2005). The nonclinical evaluation ofthe potential for delayed ventricular repolarization (QT interval prolongation) by humanpharmaceuticals. ICH Safety Guideline S7B.

5. ICH (The International Conference on Harmonisation of Technical Requirements forRegistration of Pharmaceuticals for Human Use). (2005) Note for guidance on the clin-ical evaluation of QT/QTc interval prolongation and proarrhythmic potential for non-antiarrhythmic drugs. ICH Efficacy Guideline E14.

6. Stoelzle, S., Obergrussberger, A., Bruggemann, A., Haarmann, C., George, M., Ketten-hofen, R., Fertig, N. (2011). State-of-the-art automated patch clamp devices: heat activa-tion, action potentials, and high-throughput in ion channel screening.Front Pharmacology,2, 76.

7. Finkel, A., Wittel, A., Yang, N., Handran, S., Hughes, J., Costantin, J. (2006). Populationpatch clamp improves data consistency and success rates in the measurement of ioniccurrents. Journal of Biomolecular Screening, 11, 488–496.

8. Valentin, J.P. (2010). Reducing QT liability and proarrhythmic risk in drug discovery anddevelopment. British Journal of Pharmacology, 159, 5–11.

9. Ponte, M.L., Keller, G.A., Girolamo, G. (2010). Mechanisms of drug induced QT intervalprolongation. Current Drug Safety, 5, 44–53.

10. Roche Diagnostics (2010). Preclinical cardiac safety assessment using mouse ES-derivedCor.At R© cardiomyocytes. xCELLigence System RTCA Cardio Instrument ApplicationNote 12 (September 2010).

11. Meyer, T., Leisgen,C.,Gonser, B.,Gunther, E. (2004).QT-screen: high-throughput cardiacsafety pharmacology by extracellular electrophysiology on primary cardiac myocytes.Assay Drug Development Technology, 2, 507–514.

12. Lotze, U., Lemm, H., Heyer, A., Muller, K. (2011). Combined determination of highlysensitive troponin T and copeptin for early exclusion of acute myocardial infarction:first experience in an emergency department of a general hospital. Vascular Health RiskManagement, 7, 509–515.

13. Tamai, M., Yamashita, A., Tagawa, Y. (2011). Mitochondrial development of the in vitrohepatic organogenesis model with simultaneous cardiac mesoderm differentiation frommurine induced pluripotent stem cells. Journal of Bioscience and Bioengineering, 112,495–500.

14. Beckers, S., Noor, F., Muller-Vieira, U., Mayer, M., Strigun, A., Heinzle, E. (2010).High throughput, non-invasive and dynamic toxicity screening on adherent cells usingrespiratory measurements. Toxicology In Vitro, 24, 686–694.

15. Meyer, T., Sartipy, P., Blind, F., Leisgen, C., Guenther, E. (2007). New cell models andassays in cardiac safety profiling. Expert Opinion Drug Metabolism and Toxicology, 3,507–517.

16. Caspi, O., Itzhaki, I., Kehat, I., Gepstein, A., Gil Arbel, G., Huber, I., Satin, J., Gepstein,L. (2009). in vitro electrophysiological drug testing using human embryonic stem cellderived cardiomyocytes. Stem Cells and Development, 18, 161–172.

REFERENCES 419

17. Su, X., Young, E.W.K., Underkofler, H.A.S., Kamp, T.J., January, C.T., Beebe, D.J.(2011). Microfluidic cell culture and its application in high-throughput drug screening:cardiotoxicity assay for hERG channels. Journal of Biomolecular Screening, 16, 101–111.

18. Wobus, A.M., Loser, P. (2011). Present state and future perspectives of using pluripotentstem cells in toxicology research. Archives of Toxicology, 85, 79–117.

19. Kraushaar, U., Meyer, T., Hess, D., Gepstein, L., Mummery, C.L., Braam, S.R., Guenther,E. (2011). Cardiac safety pharmacology: from human ether-a-gogo related gene channelblock towards induced pluripotent stem cell based disease models. Expert Opinion onDrug Safety, 11, 285–298.

20. Sartipy, P., Bjorquist, P. (2011). Concise review: Human pluripotent stem cell-basedmodels for cardiac and hepatic toxicity assessment. Stem Cells, 29, 744–748.

21. Guo, L., Abrams, R.M.C., Babiarz, J.E., Cohen, J.D., Kameoka, S., Sanders, M.J.,Chiao, E., Kolaja, K.L. (2011). Estimating the risk of drug-induced proarrhythmia usinghuman induced pluripotent stem cell-derived cardiomyocytes. Toxicological Sciences,123, 281–289.

22. Meyer, T., Stuerz, K., Guenther, E., Edamura, M., Kraushaar, U. (2010). Cardiac slices asa tool to predict arrhythmogenic potential of drugs and chemicals. Drug Metabolism andToxicology, 6, 1461–1475.

23. Jonsson, M.K., Duker, G, Tropp, C., Andersson, B., Sartipy, P., Vos, M.A., van Veen, T.A.(2010). Quantified proarrhythmic potential of selected human embryonic stem cell-derivedcardiomyocytes. Stem Cell Research, 4, 189–200.

24. Jonsson, M.K.B., van Veen, T.A.B., Goumans, M.J., Vos, M.A., Duker, G., Sartipy, P.(2009). Improvement of cardiac efficacy and safety models in drug discovery by the useof stem cell-derived cardiomyocytes. Expert Opinion on Drug Discovery, 4, 357–372.

25. Zaza, A. (2010). Control of the cardiac action potential: The role of repolarization dynam-ics. Journal of Molecular and Cellular Cardiology, 48, 106–111.

26. Braam, S.R., Tertoolen, L., van de Stolpe, A., Meyer, T., Passier, R., Mummery, C.L.(2010). Prediction of drug-induced cardiotoxicity using human embryonic stem cell-derived cardiomyocytes. Stem Cell Research, 4, 107–116.

27. Mandenius, C.F., Steel, D., Noor, F., Meyer, T., Heinzle, E., Asp, J., Arain, S., Kraushaar,U., Bremer, S., Class, R., Sartipy, S. (2011). Cardiotoxicity testing using pluripotent stemcell derived human cardiomyocytes and state-of-the-art bioanalytics: a review. Journal ofApplied Toxicology, 31, 191–205.

28. Stoelzle, S., Haythornthwaite, A., Kettenhofen, R., Kolossov, E., Bohlen, H., George, M.,Bruggemann, A., Fertig, N. (2011). Automated patch clamp on mESC-derived cardiomy-ocytes for cardiotoxicity prediction. Journal of Biomolecular Screening, 16, 910–916.

29. Papkovsky, D., Hynes, J., Will, Y. (2006) Respirometric screening technology for ADME-Tox studies. Expert Opinion on Drug Metabolism and Toxicology, 2, 313–323.

30. Deshpande, R., Kirsch, Y., Maas, R., John, G., Krause, C., Heinzle, E. (2005). Microplateswith integrated oxygen sensors for kinetic cell respiration measurement and cytotoxicitytesting in primary and secondary cell lines. Assay Drug Development Technology, 3,299–307.

31. Noor, F., Niklas, J., Muller-Vieira, U., Heinzle, E. (2009). An integrated approach toimproved toxicity prediction for the safety assessment during preclinical drug developmentusing Hep G2 cells. Toxicology and Applied Pharmacology, 237, 221–231.


32. White, S.M., Constantin, P.E., Claycomb, W.C. (2004). Cardiac physiology at the cellularlevel: use of HL-1 cardiomyocytes for studies of cardiacmuscle cell structure and function.American Journal of Physiology and Heart Circulation Physiology, 286, 823–829.

33. Sridharan, V., Guichard, J., Li, C.Y., Muise-Helmericks, R., Beeson, C.C., Wright, G.L.(2008). O2-sensing signal cascade: clamping of O2 respiration, reduced ATP utilization,and inducible fumarate respiration. American Journal of Physiology and Cell Physiology,295, C29–C37.

34. Andersson, H., Kagedal, B., Mandenius, C.F. (2010). Monitoring of troponin release fromcardiomyocytes during exposure of toxic substances using surface plasmon resonancebiosensing. Analytical and Bioanalytical Chemistry, 398, 1395–1402.

35. Andersson, H., Steel, D., Asp, J., Dahlenborg, K., Jonsson, M., Kagedal, B., Jeppsson, A.,Lindahl, A., Sartipy, P., Mandenius, C.F. (2010). Assaying cardiac biomarkers for toxicitytesting using biosensing and cardiomyocytes derived from human embryonic stem cells.Journal of Biotechnology, 150, 175–181.

36. Fritzsche,M.,Mandenius, C.F. (2010). Cell-based sensor systems for toxicity testing usingnovel fluorescence spectroscopic approaches. Analytical and Bioanalytical Chemistry,398, 181–191.

37. Fritzsche, M., Fredriksson, J.M., Carlsson, M., Mandenius, C.F. (2009). A cell-basedsensor system for toxicity testing using multiwavelength fluorescence spectroscopy. Ana-lytical Biochemistry, 387, 271–275.

38. Xiao, L., Hua, Z., Zhang,W.,Wua, C., Yua, H.,Wang, P. (2010). Evaluation of doxorubicintoxicity on cardiomyocytes using a dual functional extracellular biochip. Biosensors andBioelectronics, 26, 1493–1499.

39. Matsa, E., Rajamohan, D., Dick, E. Young, L., Mellor, I., Staniforth, A, Denning, C.(2011). Drug evaluation in cardiomyocytes derived from human induced pluripotent stemcells carrying a longQT syndrome type 2mutation.European Heart Journal, 32, 952–962.

40. Itzhaki, I., Maizels, L., Huber, I., Zwi-Dantsis, L., Caspi, O., Winterstern, A., Feldman,O., Gepstein, A., Arbel, G., Hammerman, H., Boulos, M., Gepstein, L. (2011). Modellingthe long QT syndrome with induced pluripotent stem. Nature, 471, 225–229.

23HIGH-THROUGHPUT SCREENINGASSAYS TO EVALUATE THEHEMATOTOXIC POTENTIALOF DRUGS

Caroline Haglund, Rolf Larsson, and Martin Hoglund

23.1 INTRODUCTION

One of the most common dose-limiting adverse effects in cancer treatment is hema-tological toxicity [1]. The rapid turnover of the hematopoietic cells sensitizes themas targets for anticancer drugs, and this toxicity may lead to serious conditions suchas acute and prolonged neutropenia, thrombocytopenia, and anemia [2]. New anti-cancer drugs are being developed under the paradigm of high specificity with lowtoxicity. Since hematological toxicity often determines a drug’s tolerable dose it canbe advantageous to identify the hematotoxic potential of drug candidates at an earlystage of drug development [3]. The conventional way of predicting hematotoxicity isto perform animal tests, but due to species differences between animals and humans,results can be misleading [4, 5]. Complementing in vivo studies with in vitro assayscan refine the safety margin and may reduce animal testing. The in vitromethods usedso far have been of low throughput, thereby limiting their use. Therefore, efforts havebeen made to develop high-throughput screening assays to evaluate the hematotoxicpotential of drugs.


421

422 EVALUATE THE HEMATOTOXIC POTENTIAL OF DRUGS

23.2 HEMATOPOIESIS

Hematopoiesis is the formation of blood cells, which are derived from the hematopoi-etic stem cell (HSC). The bone marrow is the primary site of survival, growth, anddevelopment of the HSC. The pluripotent HSC is capable of generating virtuallyall of the cell types in the immune system and one HSC is capable of producingone million mature blood cells after 20 cell divisions [6–8] (see Fig. 23.1). HSChas the capability of self-renewal, which means that when the HSC divides, one cellreplaces the stem cell and the other is committed to differentiation. The HSC cangenerate lymphoid and myeloid progenitors, which are also capable of self-renewal[8]. These early committed progenitors express low levels of transcription factorsthat commit them to a specific lineage, and growth factor stimulation determinesthe type of blood cells that is produced. The growth factors, also called cytokines,are glycoprotein hormones that regulate proliferation and differentiation of progen-itor cells, affect the function of mature cells, and prevent apoptosis [6, 9]. Drug-induced hematotoxicity can occur in different parts of the hematopoietic hierarchy,thereby determining the character of toxicity. For example, if a drug acts on theHSC, this effect will be amplified throughout the system. In contrast, if the effectis observed during differentiation, then the effect may be specified for a particularlineage [7].

Primitive lympho-hematopoietic stem cell

Myelopoetic stem cell Lymphopoietic stem cell

T-lymphocyteprogenitor cell

Megakaryopoieticprogenitor cell

Myelomonocyticprogenitor cell

Erythropoieticprogenitor cell

Erythropoieticprecursor

Red blood cell Granulocyte Monocyte

Macrophage

Platelet T-lymphocyte B-lymphocyte

Neutrophilprecursor

Macrophageprecursor

B-lymphocyteprogenitor cell

(Neutrophil, Eosinophil,Basophil)

FIGURE 23.1 The hierarchy and organization of the hematopoietic system.

LIQUID CULTURE ASSAYS 423

23.3 COLONY FORMING UNIT ASSAYS

The most frequently used in vitro method for prediction of drug-induced hematotox-icity is the colony forming unit (CFU) assays. The CFU assay can be performed toreflect different hematopoietic lineages such as the granulocyte-macrophage (CFU-GM), erythroid (CFU-E), and megakaryocyte (CFU-Mk) lineage [10–13]. In thecase of CFU assays with human hematopoietic progenitors, cells are cultured in animmobilizing semisolid medium in the presence of growth factors, stimulating thecells to proliferate, differentiate, and form colonies, which are manually countedafter 14 days of culturing [3, 12, 14, 15]. The CFU assays offer the possibility todistinguish colonies from different lineages, but drawbacks include low throughput,subjective colony count and no quantification of colony, which precludes its use inhigh-throughput screening [16].

23.4 APPLICATIONS OF THE COLONY FORMING UNIT ASSAYS

The CFU assay was established by Bradley and Metcalf in the 1960s and has beenused to identify the hematotoxic drug potential since then [9, 17, 18]. The CFU-GMassay has been applied in different settings. For example, Pessina et al. [4] havesuggested that the CFU-GM assay can be used to predict the maximum tolerateddose (MTD) of a drug. It was demonstrated that an 87% prediction of the MTD couldbe achieved when the 90% inhibitory concentration (IC90) value of a compoundwas determined with the CFU-GM. The CFU-GM assay has also been applied todetect species differences (between mice and humans) in cellular drug sensitivity [4].Hence, there are several reports, in which CFU assays have been used in the courseof drug development, but it has never been implemented as a routine assay, althoughit can provide predictive information regarding compound toxicity [16, 17]. One ofthe main reasons for the limited use of the CFU assay is the low throughput of themethod.

23.5 LIQUID CULTURE ASSAYS

Undifferentiated hematopoietic cells are capable of giving rise to specific committedprogenitors in liquid culture if appropriate cytokine stimulation is provided [2, 19].Thus, liquid culture assays have been used to evaluate hematotoxicity of anticanceragents. In these assays, inhibition of cell growth or cell death have been quantifiedwith dimethylthiazol diphenyltetrazolium bromide (MTT) [20], fluorescein diacetate(FDA) [21], flow cytometry [19], measurement of intracellular ATP [16, 22] or bymanual counting of viable cells after trypan blue staining [2]. In several of the liquidculture assays described the cells were cultured in 96- and 384-well plates, whichenables their use in high-throughput screening as compared to the CFU assays.


23.6 HIGH-THROUGHPUT METHODS FOR THE HEMATOTOXICEVALUATION OF COMPOUNDS

Screening large number of compounds can be very labor intensive. Therefore, someof the desirable qualities of a drug screening assay include simplicity, low costs,and reproducibility. The screening assays must allow a high throughput of testedcompounds as well as provide a reasonably accurate assessment of drug sensitivity[23]. In the early stages of drug development there are usually large numbers ofcandidates that need evaluation. Hence, high throughput is an important quality atthis stage. Although the CFU assay can provide valuable information in preclinicaldrug development, the low throughput has only permitted testing small numbers ofselected candidates, precluding its use in early drug development [16]. Therefore,efforts have been made to develop high-throughput alternatives to the CFU assays topredict drug-induced hematotoxicity.

23.7 THE FMCA-GM SYSTEM

One example of a high-throughput assay is the fluorometric microculture cytotoxi-city assay-granulocyte macrophage (FMCA-GM), which is performed with CD34+

umbilical cord blood cells in a liquid culture, supplemented with cytokines to inducedifferentiation toward the granulocyte-macrophage lineage [21]. Cell viability is ana-lyzed using the highly automated FMCA. This method is based on the ability of cellswith intact cell membranes to convert non-fluorescent FDA to fluorescent fluores-cein, and the fluorescence is proportional to the number of living cells [24]. The greatadvantage of the FMCA-GM is that it is performed in 384-well plates with automatedand objective quantification of viable cells. This makes it possible to investigate thehematotoxic potential of hundreds of compounds in 7 to 14 days with a robust assay.To investigate the possibility of replacing the CFU-GM assay by the FMCA-GM,method validation was performed using drugs with known properties. When the 50%inhibitory concentration (IC50) values of the FMCA-GM were correlated with theIC50 values of the CFU-GM assay obtained from literature, a high correlation wasseen (r= 0.8). It was concluded that the FMCA-GM can be used as a high-throughputalternative to the CFU-GM assay.

23.8 THE HALO R© SYSTEM

In the hematotoxicity Assay via luminescence output (HALO R©) system the CFUassay was redesigned using an instrument-based proliferation readout. The HALO R©

assay detects clonal expansion of cells by measuring luminescence of intracellularATP in methylcellulose-based plates or in liquid culture assays [7, 16, 22]. In theHALO R© assay, the cells are lysed after culturing to release intracellular ATP. Theconcentration of intracellular ATP in the lysate is proportional to the number ofproliferating cells in the cultures and is determined by its reaction with luciferin

ESTIMATE THE HEMATOTOXIC POTENTIAL OF DRUGS 425

in the presence of the enzyme luciferase [22]. In the original HALO R©-96 MeCmethod, mononuclear cells from bone marrow were cultured in semisolid mediumin 96-well plates supplemented with cytokines for lineage-specific differentiation.This assay was further optimized by performing it in liquid suspension and in 96-or 384-well plates with an assay readout after 4–7 days. A good correlation betweenthe CFU assay and the HALO R© assay (r = 0.7) was observed when the assays wereperformed under the same conditions. The number of colonies was proportional tothe ATP luminescence, and it was stated that the HALO R© assay can be used as aninstrument-based high-throughput alternative to the CFU assay [22].

23.9 ALTERNATIVE SOURCES OF HEMATOPOIETIC CELLS

In the classic CFU, FMCA-GM, andHALO R© assays, immature cells such asmononu-clear bone marrow cells or CD34+ umbilical cord blood cells are used for drug char-acterization [3]. Another alternative is using peripheral blood lymphocytes, which areeasy to handle, cheap and in comparison easily accessible [5, 16, 25]. Lymphocytescan be kept in culture for several days under standard culturing conditions, whereasperipheral blood neutrophils are only viable for a few hours [26]. Lymphocytes arewell suited for use in high-throughput screening and have been used in the FMCA.An analysis of drug sensitivity of lymphocytes with FMCA has shown a relativelystrong concordance with results obtained with the CFU assay [5, 21]. The drawbackis that the sensitivity of HSCs and progenitor cells, which are believed to be themost important targets for hematological toxicity, is less likely to be reflected bylymphocytes.

23.10 APPLICATIONS OF HIGH-THROUGHPUT ASSAYS TOESTIMATE THE HEMATOTOXIC POTENTIAL OF DRUGS

The two described high-throughput assays for the estimation of the hematotoxicpotential of drugs, the FMCA-GM and the HALO R© system, have proven to bequalified as alternatives to the CFU assay. Therefore, it is most likely that they canbe used to obtain the same type of information as with the CFU assay. Examples ofhow the FMCA-GM and the HALO R© systems have been applied so far are presentedbelow.The FMCA-GM has been applied together with a panel of tumor cells to obtain

information on the therapeutic index of drugs and was used to rank drugs according tostem cell toxicity. In this respect, the therapeutic index was defined as the differencein sensitivity between the cells in the FMCA-GM and tumor cells. Using the FMCA-GM in this way, it was able to distinguish between conventional cytotoxic drugs andless toxic targeted drugs [25]. The FMCA-GM has also been included in a panel ofnormal cell models from different tissues for drug toxicity profiling.When comparingdrug sensitivity of the normal cell panel (with data obtained with the FMCA), resultsroughly reflected clinically observed toxicity [25].


The HALO R© system has multifunctional capacity and in addition to the stan-dard GM lineage it enables testing of seven cell populations of human origin. Thecell models evaluated are a primitive and a mature stem cell, three myelopoieticprogenitors and two lymphopoietic progenitors. This multilineage panel was set upto determine at what point of the hematopoietic hierarchy the hematotoxic effectoccurs. For example, if the toxicity is seen in the most primitive stem cell model, alllineages will be affected, but if the toxicity is restricted to a committed progenitorthe toxicity may be limited to that particular lineage. In a validation study usingreference drugs with known hematotoxic profile, the results from the HALO R© assayseemed to accurately predict the clinical toxicity profile of a high number of drugstested [16].Using peripheral blood lymphocytes as an alternative source of hematopoietic

cells is methodologically simple and straightforward. It could possibly be used tocomplement the progenitor models and serve as a model for a differentiated and non-proliferating peripheral blood cell. Lymphocytes have also proven to be useful whenpredicting species differences in cellular drug sensitivity [5]. Information on speciesdifferences can help in the transition from in vitro to in vivo models, in selectingappropriate models and in interpreting in vivo data [5, 17, 27].

23.11 LIMITATIONS OF IN VITRO ASSAYS TO ESTIMATE THEHEMATOTOXIC POTENTIAL OF DRUGS

The advantage of in vitro assays like the FMCA-GM and HALO R© assay is that thedrug-induced toxicity is measured in the actual target cell, which is very straight-forward. However, toxicity can depend on a number of factors not included in theseassays, for example drug distribution, cell interactionswith stroma, and activemetabo-lites. Another limitation is that these in vitro methods are likely to be more suitedfor detection of hematotoxicity induced by anticancer drugs compared to non-toxicdrugs with lower toxicity frequency.High-throughput assays should be robust and amenable to standardize. Studies

with promising results have been conducted to assess the reproducibility of theHALO R© system [22]. The FMCA-GM has only been performed at one site so farand, although the FMCA has been routinely used for assessment of cytotoxicity inother cell systems [24], this assay needs further validation.Limiting the testing to one lineage, as in the case of the FMCA-GM, is often

sufficient for a general estimation of the hematotoxic potential of drugs [3], but verylineage-specific effects might be missed. For example, the clinically well-knownthrombocytotoxic effect of bortezomib was not reflected in the FMCA-GM, whichmay be due to the fact that cells in the assay belong to the granulocyte-macrophagelineage [25]. This problem has been approached in the HALO R© system when usingthe multilineage panel including both primitive stem cells as well as committedprogenitor cells. The drawback of this strategy is that utilizing seven cell populationsis much more labor intensive, costly, and lowers the throughput.

REFERENCES 427

23.12 SUMMARY

It is important to reduce the risk of late stage rejection by performing an improvedtoxicity profiling in the preclinical stages of drug development. The gold standardin vitro test system to estimate the hematotoxic potential of a drug, the CFU assay, canbe used in early drug development, but is limited by its low throughput. Replacingthe CFU with high-throughput assays would save time and labor and enable the iden-tification of potentially hematotoxic compounds in large scale assays. Theses assayscould be applied in different settings and stages of preclinical drug development, forexample when choosing among drug candidate analogs with similar activity, selectingthe candidates with most beneficial therapeutic index or during the early characteri-zation of a lead candidate. A further application of high-throughput hematotoxicitymethods is the detection of species differences regarding cellular drug sensitivity. Inconclusion, these in vitro assays for the prediction of hematotoxicity cannot fullymirror the complex interactions in vivo and replace in vivo studies, but they provideimportant predictive information at the preclinical stage of drug development andmay allow to reduce the number of animals used for toxicity testing.

REFERENCES

1. Parent-Massin, D., Hymery, N., Sibiril, Y. (2010). Stem cells in myelotoxicity. Toxicology,267, 112–117.

2. Ferlini, C., Distefano,M., Pierelli, L., Bonanno, G., Fattorossi, A., Battaglia, A., Mancuso,S., Scambia, G. (2001). A new method to evaluate in vitro myelotoxicity of antitumouragents in the first steps of drug development. Pharmacology and Toxicology, 89, 231–236.

3. Volpe, D.A., Warren, M.K. (2003). Myeloid clonogenic assays for comparison of the invitro toxicity of alkylating agents. Toxicology In Vitro, 17, 271–277.

4. Pessina, A., Albella, B., Bayo,M., Bueren, J., Brantom, P., Casati, S., Croera, C., Gagliardi,G., Foti, P., Parchment, R., Parent-Massin, D., Schoeters, G., Sibiril, Y., van den Heuvel,R., Gribaldo, L. (2003). Application of the CFU-GM assay to predict acute drug-inducedneutropenia: an international blind trial to validate a prediction model for the maximumtolerated dose (MTD) of myelosuppressive xenobiotics. Toxicological Sciences, 75, 355–367.

5. Hassan, S.B., Haglund, C., Aleskog, A., Larsson, R., Lindhagen, E. (2007). Primarylymphocytes as predictors for species differences in cytotoxic drug sensitivity. ToxicologyIn Vitro, 21, 1174–1181.

6. Hoffbrand, A.V., Moss, P.A.H., Pettit, J.E. (2007). Essential Haematology. BlackwellPublishing: Oxford; pp. 1–11.

7. Rich, I.N. (2007). High-throughput in vitro hematotoxicity testing and in vitro cross-platform comparative toxicity. Expert Opinion On Drug Metabolism and Toxicology., 3,295–307.

8. Kondo, M., Wagers, A.J., Manz, M.G., Prohaska, S.S., Scherer, D.C., Beilhack, G.F.,Shizuru, J.A., Weissman, I.L. (2003) Biology of hematopoietic stem cells and progenitors:implications for clinical application. Annual Review of Immunology, 21, 759–806.


9. Metcalf, D. (2010). The colony-stimulating factors and cancer. Nature Reviews Cancer,10, 425–434.

10. Pessina, A., Parent-Massin, D., Albella, B., van den Heuvel, R., Casati, S., Croera, C.,Malerba, I., Sibiril, Y., Gomez, S., de Smedt, A., Gribaldo, L. (2009). Application of humanCFU-Mk assay to predict potential thrombocytotoxicity of drugs. Toxicology In Vitro, 23,194–200.

11. Pessina, A., Malerba, I., Gribaldo, L. (2005). Hematotoxicity testing by cell clonogenicassay in drug development and preclinical trials. Current Pharmaceutical Design, 11,1055–1065.

12. Pessina, A., Albella, B., Bueren, J., Brantom, P., Casati, S., Gribaldo, L., Croera, C.,Gagliardi, G., Foti, P., Parchment, R., Parent-Massin, D., Sibiril, Y., van den Heuvel, R.(2001). Prevalidation of a model for predicting acute neutropenia by colony forming unitgranulocyte/macrophage (CFU-GM) assay. Toxicology In Vitro, 15, 729–740.

13. Olaharski, A.J., Uppal, H., Cooper, M., Platz, S., Zabka, T.S., Kolaja, K.L. (2009). In vitroto in vivo concordance of a high throughput assay of bone marrow toxicity across a diverseset of drug candidates. Toxicology Letters, 188, 98–103.

14. Gribaldo, L. (2002). Haematotoxicology: scientific basis and regulatory aspects. Alterna-tives to Laboratory Animals, 30(suppl 2), 111–113.

15. Masubuchi, N., May, R.D., Atsumi, R. (2004). A predictive model of human myelotox-icity using five camptothecin derivatives and the in vitro colony-forming unit granulo-cyte/macrophage assay. Clinical Cancer Research, 10, 6722–6731.

16. Rich, I.N., Hall, K.M. (2005) Validation and development of a predictive paradigm forhematotoxicology using a multifunctional bioluminescence colony-forming proliferationassay. Toxicological Sciences, 87, 427–441.

17. Parchment, R.E., Gordon, M., Grieshaber, C.K., Sessa, C., Volpe, D., Ghielmini, M.(1998). Predicting hematological toxicity (myelosuppression) of cytotoxic drug therapyfrom in vitro tests. Annals of Oncology, 9, 357–364.

18. Bradley, T.R., Metcalf, D. (1966). The growth of mouse bone marrow cells in vitro.Australian Journal of Experimental Biology and Medical Science, 44, 287–299.

19. Dal Negro, G., Vandin, L., Bonato, M., Sciuscio, D. (2006). Toward refinement of thecolony-forming unit-granulocyte/macrophage clonogenic assay: inclusion of a metabolicsystem. Toxicology In Vitro, 20, 743–749.

20. Leglise, M.C., Darodes de Tailly, P., Vignot, J.L., Le Bot, M.A., Le Roux, A.M., Riche,C. (1996). A cellular model for drug interactions on hematopoiesis: the use of humanumbilical cord blood progenitors as amodel for the study of drug-relatedmyelosuppressionof normal hematopoiesis. Cell Biology and Toxicology, 12, 39–53.

21. Haglund, C., Aleskog, A., Hakansson, L.D., Hoglund, M., Jacobsson, S., Larsson, R.,Lindhagen, E. (2010). The FMCA-GM assays, high throughput non-clonogenic alterna-tives to CFU-GM in preclinical hematotoxicity testing. Toxicology Letters, 194, 102–107.

22. Reems, J.A., Hall, K.M., Gebru, L.H., Taber, G., Rich, I.N. (2008). Development of a novelassay to evaluate the functional potential of umbilical cord blood progenitors. Transfusion,48, 620–628.

23. Sharma, S.V., Haber, D.A., Settleman, J. (2010). Cell line-based platforms to evaluate thetherapeutic efficacy of candidate anticancer agents. Nature Reviews Cancer, 10, 241–253.

24. Lindhagen, E., Nygren, P., Larsson, R. (2008). The fluorometric microculture cytotoxicityassay. Nature Protocols, 3, 1364–1369.

REFERENCES 429

25. Haglund, C., Aleskog, A., Nygren, P., Gullbo, J., Hoglund, M., Wickstrom, M., Larsson,R., Lindhagen, E. (2012). In vitro evaluation of clinical activity and toxicity of anticancerdrugs using tumor cells from patients and cells representing normal tissues. CancerChemotherapy and Pharmacology, 69, 697–707.

26. Nagami, K., Kawashima, Y., Kuno, H., Kemi, M., Matsumoto, H. (2002). In vitro cyto-toxicity assay to screen compounds for apoptosis-inducing potential on lymphocytes andneutrophils. Journal of Toxicological Sciences, 27, 191–203.

27. Price, P.S., Keenan, R.E., Swartout, J.C. (2008). Characterizing interspecies uncertaintyusing data from studies of anti-neoplastic agents in animals and humans. Toxicology andApplied Pharmacology, 233, 64–70.

PART V

HIGH-THROUGHPUT ASSAYS TOASSESS DRUG METABOLISM ANDRECEPTOR-RELATED TOXICITY


24HIGH-THROUGHPUT ENZYMEBIOCOLLOID SYSTEMS FOR DRUGMETABOLISM AND GENOTOXICITYPROFILING USING LC–MS/MS

James F. Rusling and John Schenkman

24.1 INTRODUCTION TO METABOLITE-BASED TOXICITY TESTING

A wide range of bioassays and animal studies are used for screening and predictingtoxicity in the pharmaceutical industry [1–3]. Nevertheless, drug candidates that aretoxic are not always identified by these procedures. More importantly, some toxicityissues fail to be uncovered even in clinical trails. If the drug is harmful to only a smallpopulation subset, its toxicity may not be manifested until after the drug is marketed.Consequently, about 30% of drug development failures have been linked to serioustoxicity issues that are not apparent until human clinical trials or beyond. This canlead to potential human tragedy as well as high costs that are ultimately passed on tothe consumer [4]. For these reasons, toxicity needs to be predicted more accuratelyand at the earliest stages of drug development [2, 3].Bioassay methods are typically combined into a panel of tests that can provide a

reasonably good prediction of human in vivo toxicity [1, 2, 4]. Nonetheless, unpre-dicted or so-called “idiosyncratic” drug toxicity can result from inter-individualvariations in human pharmacology. Thus, toxic reactions in specific individuals maybe impossible to predict from batteries of in vitro assays or animal tests.Many toxicity issues are related to drug metabolism [5, 6], for example, in many

cases the metabolites rather than the parent molecules may react with biomolecules.In addition, metabolic issues have been associated with over half of the instances


433

434 HIGH-THROUGHPUT ENZYME BIOCOLLOID SYSTEMS FOR DRUG METABOLISM

of idiosyncratic toxicity involving the top 200 drugs marketed in the USA [5]. Inthis respect, there is a need for simple, cheap, molecular-based reactive metabolitescreening assays in high-throughput formats. Some of these are currently emerging[7–9]. Inexpensive toxicity testing devices suited for use at very early stages of drugdevelopment are needed to contribute molecular information about toxicity of drugcandidates to aid in go/no-go decisions.Reactions of drug molecules or their enzyme-generated metabolites with DNA

can produce covalently linked nucleobase adducts that can initiate cancer [10–12].These adducts most often occur on guanines and adenines in DNA, and serve as goodbiomarkers for cancer risk [13, 14]. Many DNA base adducts are formed when drugmetabolism forms alkylating agents as products [15]. Nitrogen and oxygen atoms onthe nucleobases are preferred sites of nucleophilic attack by these species (Fig. 24.1).Guanine is often the most reactive base, with adenine the second most reactive. AnN7-guanine-styrene oxide adduct is shown in Figure 24.2.Nucleobase adducts are convenient biomarkers for detecting reactive metabolites

and predicting drug toxicity [8, 16]. Reactive metabolites that react with DNA are

NH2

NN

N N

Sugar

Adenine Cytosine

1 5

4

6 78

932

HN

H2N

N

N N

Sugar

Guanine

1 5

4

6 78

932

NH2

N

N

Sugar

O

3 5

6

4

12

OO

HN

N

Sugar

Thymine

3 5

6

4

12

O

FIGURE 24.1 Sites of DNA alkylation subject to attack by reactive drug metabolites Arrowsshow major (bold, solid arrow) and minor (broken arrow) alkylation sites.

INTRODUCTION TO METABOLITE-BASED TOXICITY TESTING 435

(a) (b)

styrene 7,8-oxide

βN 7-styrene oxide-guanine adduct

CH CH2

HN

H2N NN

N

OH

1

4

2

3

6 78

95

OO

α β

FIGURE 24.2 AnN7-guanine adduct formed from reaction of the styrene metabolite styrene7,8-oxide with guanine.

also likely to attack proteins and other biomolecules. However, DNA adducts andtheir structures are readily measured by a number of modern bioanalytical techniquesincluding LC–MS. The term bioactivation is used to define the enzyme-generated for-mation of reactive metabolites by cytochrome P450s (cyt P450) and other metabolicenzymes. DNA damage caused by reactive metabolites falls under the heading geno-toxicity. Many recognizable chemicals are metabolized to DNA-reactive species,including styrene, benzo[a]pyrene, nitrosamines, naphthylamines, tamoxifen, andother chemotherapeutic agents [17–21].Metabolic enzymes important in bioactivation to reactive intermediates include

oxidative catalysts (phase I enzymes) and bioconjugation enzymes (phase IIenzymes). The cyt P450s are the major class of oxidative enzymes (Fig. 24.3a).This large family of iron-heme enzymes (P-Fe) catalyzes oxygen-atom transfer, andis involved in metabolism of about 75% of marketed drugs (Fig. 24.3b) [22–24]. Theyare major players in bioactivating drugs to reactive metabolites, although, of course,not all cyt P450-derived metabolites react with biomolecules.Figure 24.3c depicts the proposed pathway for cyt P450-catalyzed oxidations.

The resting state of the iron-heme in cyt P450s has a labile water molecule boundto FeIII (1) [24]. This form of the enzyme binds substrate RH and eliminates waterto give 2. Substrate sits above the heme iron in a hydrophobic pocket. Next, 2 isreduced to P-FeII by NADPH-dependent cyt P450 reductase to give 3, which thenbinds dioxygen to form a ferrous-dioxygen or ferric-superoxy complex 4 [25]. Thiscomplex is converted to P-Fe(III)-OOH by reduction to 5, and then protonated to 6.P-Fe(III)-OOH (6) can also be generated by the peroxide shunt pathway shown bythe arrow connecting species 2 and 6. This is accomplished experimentally by addinghydrogen peroxide or organic peroxide. Protonation of 6 and dioxygen cleavage formsthe heme-iron(IV)-oxo radical cation 7, which transfers oxygen to bound substrateto make the product (ROH). ROH dissociates and the distal site is again occupiedby water to complete the catalytic cycle. Reaction of ferrous form 3 with carbon


FIGURE 24.3 Cyt P450s and drug metabolism: (a) Structure of cyt P450 1A2; (b) fractionof metabolic enzyme action for established drugs, and (c) pathway for cyt P450-catalyzedmetabolic chemistry.

monoxide produces the P-Fe(II)-CO complex 9 that absorbs light at ∼450 nm andgives the cyt P450s their name.Bioconjugation enzymes such as glutathione transferase, esterases, and acyl trans-

ferases add solubilizing groups to drug reactants (Fig. 24.3b) Bioconjugation enzymescan also bioactivate or deactivate drugs. In some cases, these enzymes act in tandemwith cyt P450s to bioactivate drugs [26, 27].

BIOCOLLOID REACTOR PARTICLES 437

Clearly, it is important in drug development to establish whether metabolites canreact with DNA and, if so, what structures ensue. While methods such as the Amestest and the Comet assay approach this problem using whole cells, our research forthe past decade has been directed at developing approaches to answer the question“can metabolites of a given drug or chemical react with DNA at a significant rate?”Then, for the reactive metabolites we identify, we want to determine the structure ofthe nucleobase adducts and their individual rates of formation.Detection of DNA base adducts from metabolites in a screening format requires

(1) producing the metabolites with a representative set of metabolic enzymes, (2)reacting the metabolites with DNA, and (3) detection of the nucleobase adducts [16].We learned very early on in this research area that doing the metabolic reactions in thepresence of DNA with enzymes and reactants in solution requires very long reactiontimes, often 12 h to several days, to produce measurable quantities of nucleobaseadducts. We thus developed strategies involving ultrathin films combining DNA andenzymes on surfaces. These films provide high concentrations of enzymes and DNAin close proximity to one another, so that when metabolites are formed they mustdiffuse out of the film by passing through the regions of very highDNA concentration.Thus, rates of the metabolite reactions with DNA are enhanced and nucleobaseadducts are formedmuch more efficiently. Our first line of screening involves electro-optical arrays that combine the DNA/enzyme films with a metallopolymer that “lightsup” when undergoing redox cycling in the presence of damaged DNA. We haveestablished that the rate of signal increase in these arrays is proportional to the rateof formation of nucleobase adducts [8, 28]. Drugs or chemicals whose metabolitesreact with DNA are rapidly identified in this assay, and can then be subject to moredetailed analysis by LC–MS/MS. Here again, we use films of DNA and metabolicenzymes to generate the DNA damage, but this time on particles.In the next sections, we describe our approaches to LC–MS screening for reac-

tive metabolites. We start by describing our first biocolloid reactor particle studieswith applications to metabolite profiling and relative metabolite assessment. We thendiscuss the evolution of these biocolloid reactor particles into high-throughput pro-cessing systems for these applications.

24.2 BIOCOLLOID REACTOR PARTICLES

Lvov and Caruso pioneered alternate-charge electrostatic layer-by-layer (LbL) filmassembly to make enzyme-coated nanoparticles for biocatalysis [29]. We used thisversatile LbL film fabrication technique to make films of DNA and enzymes orenzyme-containing microsomes on silica particles [30–33]. Fabrication involvesadsorbing the polyions onto the particle one layer at a time, reversing the chargeof the polyion at each adsorption step, and washing between steps to remove weaklyadsorbed materials. About 20 min is required to reach steady state surface adsorptionfor solutions 0.5–4 mg mL−1 in polyion. As an example, bare silica has a negativecharge in neutral solutions, so that silica particles could first be coated with a solutionof polycations such as poly(dimethyldiallyl amine) (PDDA). Then, a layer of thepolyanion DNA could be adsorbed onto this particle, reversing its change. This can


be followed by a second PDDA layer, then a final layer of negatively charged enzymesor microsomal enzymes. After final washing, these particles can be stored in bufferat 5◦C for a month or more, as the polyion films typically stabilize the enzymes [34].These bioreactor colloids are used to run the enzyme reactions and make metabo-

lites that can react with DNA in the film. The metabolites are formed and then diffusethrough a large concentration of DNA in the films, and react to form DNA adducts.The DNA is then hydrolyzed off the particles to yield individual nucleobase andnucleoside adducts, and the resulting samples are analyzed by LC–MS/MS. Hydrol-ysis can be done thermally or by using enzymes.Our first DNA/enzyme films for LC–MS analysis of metabolic DNA damage were

made on porous carbon cloth. Later, films made on 500 nm diameter silica colloids(Fig. 24.4) provided much larger reaction rates resulting in more sensitive adductdetection in shorter times. These enzyme/DNA biocolloid reactors combined withhydrolysis of the DNA after the reaction, and capillary LC–MS (capLC–MS) withon-line sample preconcentration provides a rapid and sensitive method for measuringand identifying DNA-reactive metabolites [35]. The large active surface area of thebiocolloids facilitates the use of microliter solution volumes to conserve enzyme andobtain more product per unit time for determination of major and minor products ofboth the metabolic and metabolite-DNA reactions.

Silica bead

PDDA

DNA

Enzyme

Enzyme

PSS

FIGURE 24.4 DNA/enzyme film formation on silica microbeads by LbL electrostaticassembly. In this example, cationic poly(diallyldimethylamine) PDDA is first assembled ontothe negatively charged microbeads, followed by sequential adsorption of negatively chargedpoly(styrene sulfonate) (PSS) or DNA, followed by a positively charged enzyme layer. Wash-ing removes weakly adsorbed molecule between each adsorption step. Other arrangements oflayers can also be deposited.


FIGURE 24.5 Bioactivation of NNK (1) by cyt P450 2E1 (CYP). First, �-hydroxylation gives intermediate (2) that loses formaldehyde to form 4-oxo-4-(3-pyridyl)-1-butanediazohydroxide (3). Reaction of 3 with guanines in DNA produces the major adduct7-[4-oxo-4-(3-pyridyl)but-1-yl]Gua) (4). Hydrolysis of 3 produces 4-hydroxy-1-(3-pyridyl)-1-butanone (HPB) (5). The dashed line in 4 represents fragmentation in MS.

To test performance, styrene and 4-(methylnitrosoamino)-1-(3-pyridyl)-butanone(NNK) were oxidized with cyt P450 2E1 biocolloids activated by 1 mM H2O2 [35].Cyt P450 2E1 (CYP 2E1) is a major human enzyme that oxidizes styrene to styreneoxide as the only product (Fig. 24.2) [36]. NNK is an N-nitroso procarcinogen intobacco smoke and a suspected carcinogen [37]. NNK metabolizes to 4-hydroxy-1-(3-pyridyl)-butanone (HPB) as the final product (Fig. 24.5). In these studies, styreneoxide was measured by GC and HPB was measured by LC–MS. The formation rateof both metabolites was initially similar, but over 5 min the rate of styrene oxideproduction exceeded that of HPB (Fig. 24.6).Figure 24.7 shows the results for the DNA adducts formed by reaction of NNK

with DNA/cyt P450 2E1 biocolloids activated by peroxide. Figure 24.7a shows thecapLC/MS-MS chromatogram and Figure 24.7b is the single reaction monitoring(SRM) mass spectrum showing the product peaks after the reaction and neutralthermal hydrolysis [35]. Multiple peaks are probably due to positional isomers. Them/z 299→152 is consistent with pyridyloxobutylation of guanine via attachment atthe N7 location in Figure 24.5. The product species fragments to m/z 152 guanine.The ultimate metabolite HPB is not DNA-reactive in this case; the intermediate 3

reacts with DNA (Fig. 24.5). Figure 24.7c shows that the integrated area of this peakincreases with time similar to HPB in Figure 24.6.The above results illustrate the use of DNA/enzyme biocolloids to measure forma-

tion rates and establish structures of metabolites and DNA adducts. This techniquecan also be used to investigate other metabolic issues such as enzyme inhibition and


FIGURE 24.6 Influence of reaction time on amount of styrene oxide (SO, �) from 1%styrene and HPB (•) from 100 mMNNK found from reaction with PSS/cyt P450 2E1 colloidsactivated by 1 mM H2O2 in pH 5.5 buffer. Adapted with permission from Reference 35(copyright 2008 American Chemical Society).

drug–drug interactions [38]. These applications employed biocolloid reactors madeusing isolated recombinant enzymes that are obtained in limited quantities afterlengthy purification. In addition, the cyt P450s were activated with peroxide, whichis similar but not identical to the natural activation process that utilizes NADPHand a reductase (see Fig. 24.3). To eliminate these drawbacks, we began to userat liver microsomes and human liver microsomes (RLM and HLM) and single-cytP450 (bicistronic) membrane fractions as bioreactor enzyme sources [39]. RLMs andHLMs are commercially available mixtures of lipids, cyt P450s, and bioconjugation

FIGURE 24.7 CapLC/MS-MS results for NNK reaction catalyzed by DNA/cyt P450 2E1biocolloids in pH 5.5 buffer: (a) mass chromatogram after 15 min reaction; (b) mass spectrumanalyzed by SRM (m/z 299 → 152) of same sample as in (a); (c) influence of reaction timeon total integrated area of the nucleobase-adduct peak in (a) at 37 min retention time. Adaptedwith permission from Reference 35 (copyright 2008 American Chemical Society).


enzymes. As a source of single enzymes, large amounts of cyt P450 bicistronic mem-brane fractions can be expressed from E. coli isolated within a few days, as opposed to20–30 days for fully purified human cyt P450s. RLMs, HLMs, and bicistronic mem-brane fractions contain cyt P450 reductase, which allows the activation of enzymesvia the natural pathway, using an NADPH regenerating system. Figure 24.8 illustratesthe use of these enzyme sources in biocolloid reactors and screening arrays. Enzymeturnover rates using these microsomal bioreactors are consistently two- to threefoldlarger when compared to conventional microsomal dispersions, most likely becauseof better accessibility of the enzymes.

FIGURE 24.8 Microsome-DNA films on (a) biocolloid reactors and (b) electrochemilumi-nescence (ECL) array chips. A layer of cationic polymer [e.g., poly(ethyleneimine) or a light-emitting metallopolymer] is initially deposited on a membrane, followed by layers of negativeDNA, polycation, and microsomes including cytochrome P450 (P450) and cytochrome 450reductase (CPR). NADPH is added to reduce CPR, which transfers electrons to cyt P450s. O2and cyt P450s can combine to convert the substrate to reactive metabolites that form DNAadducts in the film. Hydrolysis using heat or enzymes releases labile DNA adducts from thebiocolloid reactors for LC–MS analysis. Abbreviation: PAH, polycyclic aromatic hydrocarbon.


We illustrate the microsomal biocolloid reactions by describing a study on tamox-ifen, a breast cancer prevention drug with species toxicity differences. In short,tamoxifen produces liver tumors in rats but not in humans. LC–MS/MS analysis ofreaction products from microsome-DNA biocolloid nanoreactors on silica nanopar-ticles showed an approximate twofold larger turnover rate using RLM comparedto HLM (Fig. 24.9) [40]. LC–MS/MS directly measured relative formation rates

FIGURE 24.9 Tamoxifen (TAM) results: (a) LC chromatogramwith UV detection (blue) forstandards: a, �-OHTAM; b, 4-OHTAM; c, N-des TAM; d, TAM; e, TAMN-oxide. Microsome-DNA biocolloid reactions are presented in red (RLM) and green (HLM) (×3) for 1 min reac-tion of 25 �M tamoxifen using NADPH regenerating system. (b) Multiple reaction monitoring(MRM) chromatograms for formation of �-(N2-deoxyguanosinyl)tamoxifen using RLM withmass transition from 319 to 261. Chromatogram 1 obtained from control with 25 �M tamox-ifen only. Chromatograms 2, 3, and 4 represent reactions for 1, 2, and 3 min, respectively.Inset is peak area vs. time, normalized as peak area/wt. RLM (red) and peak area/wt. HLM(blue). (c) Daughter ions of 319, with fragmentation pattern illustrated in the inset. (d) MRMchromatogram (m/z 564 protonated �-OHTAM glucuronide to m/z 370 tamoxifen) confirmingformation of �-OHTAM O-glucuronide using HLM/DNA films after 1 min reaction. Adaptedwith permission from Reference 40 (copyright 2009 American Chemical Society). (See insertfor color representation of this figure.)

HIGH-THROUGHPUT ANALYSIS WITH BIOCOLLOID REACTORS 443

for the nucleobase adduct �-(N2-deoxyguanosinyl)tamoxifen. In addition, two- tofivefold more rapid formation rates were found for three major metabolites, that is,�-hydroxytamoxifen, 4-hydroxytamoxifen, and tamoxifen N-oxide, in the case of ratliver enzymes when compared to human liver enzymes. Similar formation rates wereobserved for N-desmethyltamoxifen with rat and human liver microsomes. A betterdetoxifying capacity for human liver microsomes than rat liver microsomes was con-firmed utilizing UDP-glucuronyltransferase in the microsomes in experiments withUDP-glucuronic acid added. Taken together, smaller amounts of DNA damage andhigher detoxication rates presented by HLM biocolloid reactors correlate with thelower risk of tamoxifen for causing liver cancer in humans compared to rats. This canbe attributed to a much more active glucuronidation pathway in humans, as mimickedby the biocolloid reactor studies.

24.3 HIGH-THROUGHPUT ANALYSIS WITHBIOCOLLOID REACTORS

Metabolic Profiling. To increase throughput for metabolic profiling, silica particlebioreactors featuring thin films of enzymes were incorporated into a 96-well plateformat. The plate has filters in the bottom of each well so that the particles canbe coated with the polymer-enzymes LbL films in them and washed. The enzymereactions are run in the plates using the enzyme-coated silica particles. After stop-ping the reaction, products are filtered using vacuum into a 96-well collection plate(Fig. 24.10), leaving the biocolloid particles behind, and then transferred to an LC–MS autosampler for analysis.The utility of this approach was illustrated by investigating the metabolism of the

drugs diclofenac (DCF), troglitazone (TGZ), and raloxifene [41]. An experimentalplan in the 96-well plate (Fig. 24.10) was designed to monitor metabolite structuresand their formation rates using microsomal bioreactors featuring HLM, RLM, andbicistronic cyt P450 3A4. These studies simultaneously compared the effects ofrat and human liver enzymes, and monitored activity of human cyt P450 3A4, amajor metabolic enzyme for these drugs. In general, results were consistent with theknown metabolism of the test compounds. In the same 96-well plate, we monitoredthe influence of enzyme inhibition by ketoconazole and activation by quinidine onthe metabolism of the drug to demonstrate applicability to drug–drug interactionsinvolving enzyme inhibition and acceleration. The experimental plan was used forrapid completion of 96 reactions of a single drug. Some of these results are presentedbelow to illustrate the methodology.When reacted with HLM or RLM dispersions, DFC is primarily hydroxylated to

5′-OH-DCF by cyt P450 3A and 4′-OH-DCF by cyt P450 2C9 (Fig. 24.11a) [42, 43].Figure 24.12 shows results of reacting DFC with RLM biocolloids, in which boththose metabolites were detected. The degradation product of 5′-OH-DCF, diclofenac-2,5-quinone imine (DCF-2,5-QI, tR = 14 min), was also found. When DCF wasincubated with bicistronic cyt P450 3A4-bioreactors, 5′-OH-DCF and DCF-2,5-QIwere detected (Fig. 24.11a). Collision induced dissociation (CID) spectra confirmedthe structure of the metabolites (Fig. 24.12c). In the positive electrospray ionization


FIGURE 24.10 Features of the bioanalytical system for high-throughputmetabolic profiling.(a) Bioreactor assembly: a layer of the cationic polymer polydiallyldimethylammonium chlo-ride (PDDA) was initially deposited on silica nanoparticles, followed by a layer of oppositelycharged microsomes. (b) A reaction/filtration 96-well plate equipped with 10,000 Da-cutoffmass filters showing the liquid chromatography-mass spectrometry (LC–MS)-ready sample-collection plate underneath. (c) Schematic illustration of simultaneous enzyme reaction designfeaturing a 96-well plate. (d) LC–MS/MS analysis with an autosampler. Abbreviations: HLM,human liver microsome; RLM, rat liver microsome. (See insert for color representation of thisfigure.)

(ESI+ ) mode, the product ion profile of the protonated molecule after CID, m/z 312,was derived from loss of H2O (m/z 294) or cleavage of H2O2 (m/z 278). The m/z 253ion is generated from loss of acetic acid. Fragment m/z 266 is generated from loss ofthe COOH group.Measured reaction rates (Fig. 24.13) showed differences in the kinetics engendered

by the different enzymes. For comparison, diclofenac and raloxifene were reactedwith HLM dispersions and with no particles. Metabolite formation rates were two- tothreefold larger using the microsome biocolloid particles compared to the reactionswith the same amount of microsomes (Fig. 24.13a). This example illustrates theability of the system to measure relative enzyme turnover rates for a wide variety ofexperimental conditions.


FIGURE 24.11 Major metabolic pathways of (a) Diclofenac, its major metabolites 5′-hydroxydiclofenac and 4′-hydroxydiclofenac, and minor metabolite diclofenac-2,5-quinoneimine. (b) Troglitazone and its major metabolite troglitazone quinone. (c) Raloxifene, itsglucuronide metabolites and the glutathione adducts following reactive intermediate formationunder cyt P450 metabolism.

Magnetic particles and toxicity screening. The 96-well plate approach for thereaction process was improved by using magnetic particles as the base for biocol-loid reactors in place of the non-magnetic silica particles. This simple substitutionenabled magnetic control of biocolloid reactor preparation. Magnetic manipulationof the particles after reaction greatly simplifies sample handling. This approach


FIGURE 24.12 Mass spectrometry chromatogram in ESI+ ion mode illustrating: (a)metabolites of DCF generated after 5 min. incubation with HLM-bioreactors and NADPH; (b)control incubation in absence of bioreactors; and (c) CID Spectra for 5′-OH-DCF. Adaptedwith permission from Reference 41 (copyright 2008 American Chemical Society).

facilitated extending applications to enzyme/DNA biocolloids to identify reactivemetabolites, which presents challenges using non-magnetic particles. In particular,adding DNA to the colloidal particles, hydrolyzing the DNA after reaction, and sep-arating the nucleobase adducts from the particles and intact DNA were much moreeasily accomplished in the 96-well plates using magnetic particles.Thus, we designed a high-throughput genotoxicity screening method by

using 1 �m diameter magnetic particles (Dynabeads R©, Invitrogen) coated withcytosol/microsome/DNA films as the biocolloid reactors in a 96-well filter plate[44]. Incorporation of microsomal and cytosolic enzymes on these particles provideda broader spectrum of enzymes than we had used before, enabling representationof a wider range of metabolic pathways. The procedures involved are illustrated inFigure 24.14. Briefly, the particles are coated with DNA and enzymes, and reactionsare run with the biocolloids in the 96-well plate with quenching at appropriate times.Reactive metabolites that are generated are trapped by covalent binding to DNAin the films, as described in the above-mentioned examples. The reaction mixtureis filtered off, then DNA is hydrolyzed by thermal or enzyme treatment in the 96-well plate, and nucleobase adducts are collected by filtering the reaction solutionsthrough the porous bottoms of the wells. Placing a flat magnet above the plate before


(a)

(c)

(b)

(d)

FIGURE 24.13 Influence of reaction time expressed as the ratio of the area of the metabolitepeak vs. the internal standard caffeine in the LC chromatogram. for: (a) 5′-OH-DCF, by RLM-(�), HLM-(•), bicis.-3A4-(�)-bioreactors and HLM dispersed in solution (�); (b) TGZQ byRLM-, HLM-, bicis.-3A4-bioreactors; (c) Bioactivation of raloxifene-GSH adduct via HLM-(•)-bioreactors (d) Monitor the formation of 6-R-G by HLM-(•)-bioreactors, 6-R-G by HLMdispersed in solution (�); and 4-R-G by HLM-(�)-bioreactors and R-G by HLM dispersed insolution (�). Adapted with permission from Reference 41 (copyright 2008 American ChemicalSociety).

vacuum filtration brings the magnetic particles to the top of the reaction solutionsso that the filters are not blocked, enabling rapid filtration into another 96-well plate(cf. Figs. 24.10 and 24.14). Then, as with the silica particle biocolloids, the solutionsare transferred to an autosampler and analyzed by LC–MS/MS. Nucleobase adductsof metabolites from ethylene dibromide, N-acetyl-2-aminofluorene, and styrene wereidentified in this way to establish proof-of-concept. The relative formation rates ofthe major DNA adducts generated in these studies correlated well with the rodentgenotoxicity metric TD50 for the three compounds (TD50 is the concentration of testcompound that produces liver tumors in half of a test population of rodents duringtheir normal life span.) This magnetic particle-based method represents our most


(a)

(b)

(c)

(d)

(e)

FIGURE 24.14 Experimental steps shown in counterclockwise sequence for metabolic tox-icity screening using magnetic biocolloid reactor particles in a 96-well plate coupled with LC–MS/MS: (a) enzyme reactions are run in the 96-well filter plate; the center picture illustratesa possible multi-experiment design; (b) while particles are held in the wells by the magneticplate, solution is replaced with a hydrolysis cocktail: (c) hydrolysis is done; (d) the magnet ismoved to the top of the well plate to pull biocolloids away from the filters, and nucleobase/deoxynucleoside adduct samples are filtered into a second 96-well plate; (e) samples in thesecond 96-well plate are transferred to an autosampler and analyzed by LC–MS/MS.

advanced technology for metabolite studies to date, and has the potential for bothhigh-throughput genotoxicity screening and metabolic profiling.

24.4 SUMMARY AND FUTURE PROSPECTS

This chapter describes new approaches utilizing enzyme-DNA biocolloid particlesdrug candidates and other chemicals for metabolic profiling and to screen for reactive

REFERENCES 449

metabolites. DNA-damage end points are measured by LC–MS/MS after only afew minutes reaction time. The 96-well plate biocolloid reactor coupled with LC–MS/MS provides high-throughput reactions for metabolite identification, screeningfor reactive metabolites, and information about chemical structures and formationrates. The bottleneck in this approach is the chromatography (20–50 min. per sample)preceding MS/MS, and it would be advantageous to replace it with a faster method,for example, MALDI or some other alternative. We feel that these technologies arevaluable to incorporate into panels of methods for toxicity screening. Integration ofsuch molecular-information-rich techniques with toxicity bioassays should enablemore informed decision making about drug candidates. In addition, revealing thetoxic-reaction chemistry of a drug may facilitate synthetic plans to design the toxicityout while retaining the desired medicinal effects.

ACKNOWLEDGMENT

This work was supported financially by U.S. PHS grant No. ES03154 from theNational Institute of Environmental Health Sciences (NIEHS), NIH, USA. Theauthors thank co-workers named in joint publication references for their excellentcontributions to this project, without which no progress would have been made.

REFERENCES

1. Kramer, J.A., Sagartz, J.E., Morris, D.L. (2007). The application of discovery toxicologyand pathology towards the design of safer pharmaceutical lead candidates.Nature ReviewsDrug Discovery, 6, 636–649.

2. Mayne, J.T., Ku, W.W., Kennedy, S.P. (2006). Informed toxicity assessment in drugdiscovery: systems-based toxicology.Current Opinion In Drug Discovery & Development,9, 75–83.

3. Nasser, A.E.F., Kamel, A.M., Clarimont, C. (2004). Improving the decision making pro-cess in structural modification of drug candidates: reducing toxicity. Drug DiscoveryToday, 9, 1055–1064.

4. Caldwell, G.W., Yan, L. (2006). Screening for reactive intermediates and toxicityassessment in drug discovery. Current Opinion Drug Discovery & Development, 9,47–50.

5. Stepan, A.F., Walker, D.P., Bauman, J., Price, D.A., Baillie, T.A., Kalgutkar, A.S., Aleo,M.D. (2011). Structural alert/reactivemetabolite concept as applied inmedicinal chemistryto mitigate the risk of idiosyncratic drug toxicity: a perspective based on the criticalexamination of trends in the top 200 drugs marketed in the United States. ChemicalResearch in Toxicology, 24, 1345–1410.

6. Thompson, R.A., Isin, E.M., Li, Y.,Weaver, R.,Weidolf, L.,Wilson, I., Claesson, A., Page,K., Dolgos, H., Kenna, J.G. (2011). Risk assessment and mitigation strategies for reactivemetabolites in drug discovery and development. Chemico-Biological Interactions, 192,65–71.


7. Stoll, D., Bachmann, J., Templin, M.F., Joos, T.O. (2004). Microarray technology: anincreasing variety of screening tools for proteomic research, Drug Discovery Today:Targets, 3, 24–31.

8. Rusling, J.F., Hvastkovs, E.G., Schenkman, J.B. (2007). Toxicity screening using biosen-sors that measure DNA damage. Current Opinion Drug Discovery & Development, 10,67–73.

9. Lynch, A.M., Sasaki, J.C., Elespuru, R., Jacobson-Kram, D., Thybaud, V., De Boeck, M.,Aardema, M.J., Aubrecht, J., Benz, R.D., Dertinger, S.D., Douglas, G.R., White, P.A.,Escobar, P.A., Fornace, A. Jr., Honma, M., Naven, R.T., Rusling, J.F., Schiestl, R.H.,Walmsley, R.M., Yamamura, E., van Benthem, J., Kim, J.H. (2011). New and emergingtechnologies for genetic toxicity testing. Environmental and Molecular Mutagenesis, 52,205–223.

10. Jacoby, W.B. (1980). Enzymatic Basis of Detoxification, Vols. I and II. Academic: NewYork.

11. Friedberg, E.C. (2003). DNA damage and repair. Nature, 421, 436–440.

12. Scharer, O.D. (2003). Chemistry and Biology of DNA Repair. Angewandte Chemie Inter-national Edition, 42, 2946–2974.

13. Phillips, D.H., Farmer, P.B., Beland, F.A., Nath, R.G., Poirier, M.C., Reddy, M.V., Turtle-taub, K.W. (2000). Methods of DNA adduct determination and their application to testingcompounds for genotoxicity. Environmental and Molecular Mutagenesis, 35, 222–233.

14. Warren, A.J., Shields, P.G. (1997).Molecular epidemiology: carcinogen-DNAadducts andgenetic susceptibility. Proceedings of the Society for Experimental Biology and Medicine,216, 172–180.

15. Marnett, L.J., Burcham, P.C. (1993). Endogenous DNA adducts: potential and paradox.Chemical Research in Toxicology, 6, 771–785.

16. Tarun, M., Rusling, J.F. (2005). Measuring DNA nucleobase adducts using neutral hydrol-ysis and liquid chromatography-mass spectrometry. Critical Reviews in Eukaryotic GeneExpression, 15, 295–315.

17. Bond, J.A. (1989). Review of toxicology of styrene. CRC Critical Reviews in Toxicology,19, 227–249.

18. Pauwels, W., Vodiceka, P., Severi, M., Plna, K., Veulemans, H., Hemminki, K. (1996).Comparison of genotoxic potency of styrene 7,8-oxide with � radiation and human cancerrisk estimation of styrene using the rad-equivalence approach. Carcinogenesis, 17, 2673–2680.

19. Umemoto, A., Komaki, K., Monden, Y., Suwa, M., Kanno, Y., Kitagawa, M., Suzuki, M.,Lin, C.-X., Ueyama, Y., Momen, M.A., Ravindernath, A., Shibutani, S. (2001). Identifica-tion and quantification of tamoxifen-DNA adducts in the liver of rats and mice. ChemicalResearch in Toxicology, 14, 1006–1013.

20. Wang, M., McIntee, E.J., Shi, Y., Cheng, G., Upadhyaya, P., Villalta, P.W., Hecht, S.S.(2001). Reactions of �-acetoxy-N-nitrosopyrrolidine with deoxyguanosine and DNA.Chemical Research in Toxicology, 14, 1435–1445.

21. Cavalieri, E.L., Rogan, E.G., Devanesan, P.D., Cremonesi, P., Cerny, R.L., Gross, M.L.,Bodell, W.J. (1990). Binding of benzo[a]pyrene to DNA by cytochrome P-450 cat-alyzed one-electron oxidation in rat liver microsomes and nuclei. Biochemistry, 29, 4820–4827.

22. Schenkman, J.B., Greim, H. (1993). Cytochrome P450. Springer-Verlag: Berlin.

REFERENCES 451

23. Ortiz de Montellano, P.R. (2005). Cytochrome P450, 3rd ed. Kluwer/Plenum: New York.

24. Lippard, S.J., Berg, J.M. (1994). Principles of Bioorganic Chemistry. University ScienceBooks: Mill Valley, CA.

25. Guengerich, F.P. (2001). Common and uncommon cytochrome P450 reactions relatedto metabolism and chemical toxicity. Chemical Research in Toxicology, 14, 611–650.

26. Ioannides, C., Lewis, D.F.V. (2004). Cytochrome P450 in the bioactivation of chemicals.Current Topics in Medicinal Chemistry, 4, 1767–1788.

27. Guengerich, F.P. (2008). Cytochrome P450 and chemical toxicology. Chemical Researchin Toxicology, 21, 70–83.

28. Rusling, J.F., Hvastkovs, E.G., Schenkman, J.B. (2009). Screening for reactivemetabolitesusing genotoxicity arrays and enzyme/DNA biocolloids. In: Drug Metabolism Handbook(Nassar, A., Hollenburg, P.F., Scatina, J.J., Eds.). Wiley: Trenton, NJ; pp. 307–340.

29. Lvov, Y., Caruso, F. (2001). Biocolloids with ordered ureasemultilayer shells as enzymaticreactors. Analytical Chemistry, 73, 4212–4217.

30. Lvov, Y. (2000). Protein Architecture: Interfacing Molecular Assemblies and Immobi-lization Biotechnology (Lvov, Y., Mohwald, H., Eds.). Marcel Dekker: New York; pp.125–167.

31. Lvov, Y. (2001).Handbook of Surfaces and Interfaces of Materials, Vol. 3: NanostructuredMaterials, Micelles and Colloids (Nalwa, R.W., Ed.). Academic Press: San Diego, CA;pp. 170–189.

32. Ariga, K., Hill, J.P., Ji, Q. (2007). Layer-by-layer assembly as a versatile bottom-upnanofabrication technique for exploratory research and realistic application. PhysicalChemistry Chemical Physics, 9, 2319–2340.

33. Zhang, X., Chen, H., Zhang, H. (2007). Layer-by-layer assembly: from conventional tounconventional methods. Chemical Communications, 1395–1405.

34. Rusling, J.F., Hvastkovs, E.G., Hull, D.O., Schenkman, J.B. (2008). Biochemical appli-cations of ultrathin films of enzymes, polyions and DNA (feature article). ChemicalCommunications, 141–154.

35. Bajrami, B., Hvastkovs, E.G., Jensen, G., Schenkman, J.B., Rusling, J.F. (2008). Enzyme-DNA biocolloids for reactive metabolite and DNA adduct detection by chromatography-mass spectrometry. Analytical Chemistry, 80, 922–932.

36. Tanaka, E., Terada, M., Misawa, S. (2000). Cytochrome P450 2E1: its clinical and toxi-cological role. Journal of Clinical Pharmacy and Therapeutics, 3, 165–175.

37. Hecht, S.S. (2003). Tobacco carcinogens, their biomarkers and tobacco-induced cancer.Nature Reviews Cancer, 10, 733–744.

38. Bajrami, B., Krishnan, S., Rusling, J.F. (2008). Microsome biocolloids for rapid drugmetabolism and inhibition assessment by LC-MS. Drug Metabolism Letters, 2, 158–162.

39. Krishnan, S., Bajrami, B., Hvastkovs, E.G., Choudhary, D., Schenkman, J.B., Rusling, J.F.(2008). Synergistic metabolic toxicity screening using microsome/DNA electrochemilu-minescent arrays and nanoreactors. Analytical Chemistry, 80, 5279–5285.

40. Zhao, L., Krishnan, S., Zhang, Y., Schenkman, J.B., Rusling, J.F. (2009). Differencesin metabolite-mediated toxicity of tamoxifen in humans vs. rodents elucidated withDNA/microsome electro-optical arrays and nanoreactors. Chemical Research in Toxi-cology, 22, 341–347.


41. Bajrami, B., Zhao, L., Schenkman, J.B., Rusling, J.F. (2009). Rapid LC-MS drug metabo-lite profiling using microsomal enzyme bioreactors in a parallel processing format. Ana-lytical Chemistry, 81, 9921–9929.

42. Leemann, T., Transon, C., Dayer, P. (1993). Cytochrome P450TB (CYP2C): a majormonooxygenase catalyzing diclofenac 4’-hydroxylation in human liver. Life Sciences, 52,29–34.

43. Tang,W. (2003). Themetabolism of diclofenac - enzymology and toxicology perspectives.Current Drug Metabolism, 4, 319–329.

44. Zhao, L., Schenkman, J.B., Rusling, J.F. (2010). High throughput metabolic toxicityscreening using magnetic biocolloid reactors and LC-MS/MS. Analytical Chemistry, 82,10172–10178.

25HIGHER-THROUGHPUT SCREENINGMETHODS TO IDENTIFYCYTOCHROME P450 INHIBITORS ANDINDUCERS: CURRENT APPLICATIONSAND PRACTICE

David M. Stresser and George Zhang

25.1 INTRODUCTION

High-throughput profiling of chemicals for acceptable toxicological and pharmaco-logic properties continues to advance and evolve. In the past decade, refinementsin testing have been directed at generating data that is not only cost-effective andclinically relevant, but also supportive of an overall improved risk profile, in additionto achieving higher throughput. Two key parameters related to assessing pharmacoki-netic drug–drug interaction (DDI) risk are cytochrome P450 (CYP) inhibition andinduction.

25.2 CYTOCHROME P450 INHIBITION METHODS

Companies developing small molecule drugs typically follow a drug discoveryparadigm of “hit to lead,” where drug candidates undergo testing to screen outundesirable pharmacokinetic and toxicological properties, while retaining propertiesdeemed essential to drug action, such as target receptor affinity and selectivity [1].The need to screen chemical libraries to select those candidates with suitable absorp-tion, distribution, metabolism, and excretion (ADME)/toxicity properties emerged


453

454 CURRENT APPLICATIONS AND PRACTICE

in full force during the early 1990s following reports of cardiac toxicity events insome individuals coadministrated ketoconazole and terfenadine [2]. The mechanismof this toxicity could be traced to the inhibition of CYP3A4-mediated conversion ofterfenadine to its active metabolite by ketoconazole, leading to a toxic accumulationof the cardiotoxic parent molecule. A more recently described example is the use ofselective serotonin reuptake inhibitors that block CYP2D6-mediated conversion ofthe breast cancer agent tamoxifen to the more potent antiestrogen and therapeutic4-hydroxytamoxifen [3]. Among the first in vitro ADME tests developed were thoseaimed at identifying potent inhibitors of cytochrome P450 enzymes. In vitro testinginvolves incubation of test articles together with a nonsaturating concentration ofsubstrate for the CYP enzyme of interest and an enzyme source. A typical experi-mental design to obtain an IC50 value involves testing seven concentrations at 0.5 logspacing with the upper concentration of 100 �M or the limit of aqueous solubility.The percent inhibition is calculated for each concentration of test article and typicallyan IC50 value is estimated using nonlinear regression analysis. Early testing methodsrequired quantification of metabolite using HPLC with absorbance or other means ofdetection, which quickly became a bottleneck in testing throughput due to the longruns required to separate substrate from metabolite in the incubation matrix. Thisissue was effectively addressed by employing a number of novel metabolite detectiontechniques including radiochemical release [4], scintillation proximity [5], and flu-orimetric assays [6]. Additional advances came in the form of luminometric assays[7]. These assays obviated the need of chromatographic separation from the substrateand did not require time-consuming extraction with organic solvents or protein pre-cipitation steps. They rapidly gained popularity, with fluorimetric and luminometricmethods tending to become the more widely adopted. More recently, advances infront-end engineering to allow rapid sampling coupled with the exquisite selectivityof tandem-mass spectrometry have allowed conventional and well-validated probesubstrates to regain in popularity [8, 9].

25.2.1 Cytochrome P450 Inhibition Testing with Fluorimetric andLuminometric Substrates

Fluorimetric analysis of cytochrome P450 activity offers a homogeneous assay withdiscontinuous or real-time detection of metabolite coupled with the advantages ofhigh sensitivity and low cost. These tests typically use a series of resorufins, fluo-resceins, quinolines, and coumarins that when incubated with enzyme and NADPHresult in formation of dealkylated or hydroxylated metabolites. Metabolites can bedistinguished from residual parent molecule in solution because of differing excita-tion and emission wavelength optima [6]. Published the application of fluorimetricsubstrates in assessing inhibition potential of test compounds with the five majorhuman drug-metabolizing CYP enzymes, namely, CYP1A2, 2C9, 2C19, 2D6, and3A4 in a 96-well plate format. Further miniaturization [10] and reducing the numberof concentrations tested to estimate IC50 values [11] can offer even higher throughputat lower cost.

CYTOCHROME P450 INHIBITION METHODS 455

The earliest fluorimetric substrates were either nonselective for an individualCYP enzyme present in liver microsomes or were not characterized as substrates forhuman sources of enzyme. The use of singly expressed recombinant human drug-metabolizing enzymes [12] circumvented the issue of nonselective metabolism aswould occur in human liver microsomes (HLM). Since only one CYP enzyme ispresent in the reaction, all metabolite generated and thus any inhibition observedis attributable to interaction with that enzyme only. By not having to demonstrateselectivity, assays for essentially any CYP isoform could be developed as long asconversion to metabolite was sufficient to allow detection [13]. An ancillary benefit ofusing single cDNA-expressed enzyme is the reduced potential formetabolic depletionof the test compound and elimination of off-target, low-level CYP metabolism ofprobe substrate, both of which can result in artificially elevated IC50 values.Metabolicdepletion can be a particular problem for some potent inhibitors (1) that are substratesfor nontarget (and target) P450s and (2) with IC50 values in the rangewhere significantmetabolic depletion may occur over the course of the incubation, typically 1 �Mor less. Nevertheless, many researchers often prefer the use of HLM over cDNA-expressed enzymes, which offers the potential advantage of using native sources andmolar ratios of CYP enzyme, cytochrome P450 reductase, and cytochrome b5 withinthe native phospholipid environment. Fluorimetric substrates are available that areadequately selective for individual isoforms [14–16]; thus, heterogeneous sources ofenzyme such as HLM may be used. A good understanding of the probe selectivity(when using HLM) and potential substrate-dependent responses for the CYP enzymeof interest is clearly advisable, as this can have significant impact on the in vitro aswell as in vivo inhibition response [17–19].One drawback to the use of fluorimetric substrates is that the test compound

or its metabolites may exhibit fluorescence or fluorescence signal quenching at theexcitation and emission wavelengths used to detect and quantify the metabolite,thereby confounding results. In an analysis of >5000 compounds, the interferencerate was estimated to be 0.8% or less for three different commonly used probes [20].While this rate of interference may be acceptably low for most researchers it may besignificantly higher in certain chemical scaffold series.Luminometry-based methods have also been used to assess CYP inhibition. In

this approach, luciferin-based derivatives can be dealkylated or hydroxylated bycytochrome P450 to ultimately regenerate luciferin [7]. Like fluorimetric meth-ods, luminometry approaches are homogeneous and do not require chromatogra-phy or costly instrumentation. In addition, exquisite analytical sensitivity affordedby luminescence permits the use of less CYP enzyme in the assay. This methodwas recently used to screen a library of >17,000 compounds with five major CYPisoforms, demonstrating its suitability for high-throughput screening [21]. Unlikefluorescence-based assays, interference in analyte detection is eliminated, althoughone should consider the potential for the compound or its metabolites to inhibitluciferase during the signal development phase of the assay. The latter case wouldpresumably be unusual, unless the compounds resemble luciferin in structure. Manyof these probes have also been reported to exhibit high selectivity for target CYPs inHLM [7].


25.2.2 Cytochrome P450 Inhibition Testing Using Drug Probe Substratesand Human Liver Microsomes

In 2006, the FDA issued a draft guidance for drug-drug interaction (DDI) testing thatcontained a table of “preferred” and “acceptable” substrates for use in CYP inhibi-tion assays and their continued use is implied in the 2012 draft guidance [22, 23]. Asubset of these substrates also appears in the 2012 guidance on DDI by the EuropeanMedicines Agency as examples of “well-validated marker reactions [24].” Thesesubstrates were among the very first to be developed as probes and consequentlyhave decades of validation work supporting their use [25]. Moreover, many of theseprobes can be used clinically to assess DDI in vivo, thereby facilitating translationof results obtained in vitro. The aforementioned guidance documents have renewedinterest among investigators in using these conventional “drug” probes, along withmass spectrometry detection, at earlier stages in drug development. However, theseassays lack the convenience of being homogeneous and detection of analyte occursafter separation from other assay components, which unfortunately reduces through-put. For example, conventional LC/MS methods typically achieve cycle times forsingle analytes of 2 to 4 min per sample, compared to 1 min for an entire 96-wellplate of samples for fluorescence- and luminescence-based assays. However, a num-ber of strategies and recent technological advances have greatly improved throughputof mass spectrometry analysis. Peng et al. [26] used a monolithic column to permithigh flow rates and achieve separation and quantitation of metabolites from six CYPsubstrates in 24 s. In this example, incubations were conducted individually andpooled for analysis. Youdim et al. [27] employed a novel technique to concentratesample into a narrow band between two higher volumes of aqueous solvent within theautosampler syringe (up to 5 �L total injection volume). The small sample volumecoupled with this “sandwich injection” and extreme high-pressure liquid chromatog-raphy permitted robust separation from the solvent front and matrix interference.Separation of nine analytes (metabolites and/or their internal standards), representinganalysis of five CYPs, was attained in under 30 s with run times of approximately 1min (Fig. 25.1).This study also deployed a “cocktail” incubation in which five substrates were

incubated concurrently, thus eliminating the need to conduct multiple incubationsand subsequent pooling steps while reducing reagent consumption. The cocktailapproach has been widely viewed as an elegant strategy to decrease overall cycletimes for CYP inhibition screening. A drawback to this approach is the need forrobust and reproducible analyte separation to optimize instrument parameters foreach analyte window. In addition, the various substrates (and their metabolites)must not interfere with catalysis of each other. Finally, detection of 4′-hydroxy(S)-mephenytoin, the metabolite of the most commonly used CYP2C19 probeS-mephenytoin, often presents a challenge in obtaining sufficient analytical sensi-tivity. This is typically addressed by increasing the amount of metabolite formedthrough the addition of more enzyme (e.g., 0.2 mg/mL HLM protein or higher)or extending incubation times. However, under these conditions midazolam, a pre-ferred probe for CYP3A4, may exhibit substrate depletion exceeding 30%, which is


0.10.0

2.0e4

4.0e4

6.0e4

8.0e4

Inte

nsity

, cps

1.0e5

1.2e5

1.4e5

1.6e5

1.8e51.9e5

0.2 0.3 0.4 0.5

F

EB

A

CD

G

H

I

0.6Time, min

0.7 0.8 0.9 1.0

FIGURE 25.1 Analytical trace from cocktail substrate screen analyzed using extreme high-pressure liquid chromatography gradient conditions. A, 1-OH-tacrine; B, D3-dextrorphan;C, dextrorphan; D, 1-OH midazolam; E, D3-4-OH mephenytoin; F, 4-OH mephenytoin; G,flunitrazepam; H, 13C6-4-OH diclofenac; I, 4-OH diclofenac. Reproduced from Reference 27with permission of Elsevier.

generally considered excessive. Despite these challenges, there are published exam-ples that demonstrate that they are surmountable and discovery teams are oftenwilling to accept some degree of method imperfection to permit gains in through-put [28].Recent approaches that combine the sensitivity and selectivity of mass spectrome-

try with a rapid sample purification system via microscale solid-phase extraction andisocratic runs have achieved sample cycle times <10 s. This methodology achievesthroughput approaching that of a plate reader, but also permits significant flexibilityin assay design. This flexibility results from the avoidance of pooling complicationsthat include limitations on the choice of analytes, microsomal protein concentrations,and incubation times. Using such an approach, Lim et al. [8] reduced analysis time toless than 15 min per 96-well plate in support of a six-CYP enzyme testing strategy.The IC50 values generated using conventional LC/MS/MS approaches gave correla-tion coefficients ranging from 0.89 to 1.13, when comparing the IC50 values obtainedfrom the two approaches for each of the CYP enzymes. These results are similar tofindings from our own laboratory (Fig. 25.2).The analytical approach has recently been extended and applied to other ADME

screening assays [9].


0.0010.001 0.01

0.01

0.1

IC50 (μM) by Conventional LC/MS

IC50

(μM

) by

Rap

idF

ireT

M m

ass

spec

trom

etry

Unity

Midazolam

Testosterone

0.1

1

1

10

10

100

100

1000

1000

Global R2 = 0.964

FIGURE 25.2 Correlation of CYP3A4 IC50 values determined on same incubation sam-ples using RapidFireTM mass spectrometry and conventional liquid chromatography massspectrometry. Assays were performed according to Reference [29]. IC50 values are means ofintraplate duplicates calculated based on peak area ratios. Data set includes results (n = 2or 3) for the same enzyme/inhibitor pair, generated in independent experiments. Experimentswere conducted using a testosterone 6�-hydroxylase or midazolam 1′-hydroxylase assay withhuman liver microsomes as the enzyme source.

25.2.3 Time-Dependent Inhibition of Cytochrome P450 and Toxicity

Compounds may be converted to metabolites that are more inhibitory than the parentmolecule. In an in vitroCYP inhibition assay, time-dependent inhibition (TDI) occurswhen the enzyme becomes catalytically compromised as incubation time proceeds.In vivo, this may manifest as larger than expected exposure occurring over timein repeat dosing studies or as DDI. Time-dependent inhibitors generally fall intocategories of reversible inhibition caused by a highly inhibitory metabolite, metabolicintermediate complex formation and mechanism-based inhibition [30]. The latter canbe more problematic, particularly because metabolites are chemically reactive andcan bind covalently to protein and/or heme of the enzyme. Covalent binding bymechanism-based inhibitors has been associated with idiosyncratic toxicity [31].In general, compounds exhibiting TDI properties can be challenging to developinto drugs. In some cases, finding of TDI may be inconsequential, such as in thecase of very low-dose drugs. For example, ethinyl estradiol, a mechanism-based


inactivator of CYP3A4, is given at doses such that the amount of CYP3A4 in liverand intestine greatly exceeds the daily dose of drug, to the point that that evenstoichiometric inactivation (e.g., partition ratio = 1) would have little impact onCYP3A4 metabolism. In other cases, compounds may be “de-risked” if it can beshown that metabolic pathways other than those typically screened are significantin overall elimination of the compound. For example, ezetimibe [32] and raloxifene[33] exhibit significant glucuronidation and this offsets CYP TDI liabilities identifiedin screening assays.

25.2.4 Methods to Detect Time-Dependent Inhibitionof Cytochrome P450 In Vitro

Detection of TDI of cytochrome P450 generally involves measuring the rate or extentof metabolism-dependent inactivation or inhibition of enzyme that occurs over timein an incubation containing test article, NADPH, and the enzyme of interest [30].Following this incubation, the amount of residual, active enzyme is determined bymeasuring enzyme activity with an added probe substrate. Importantly, the decreasein enzyme activity due to the preincubation step must be distinguished from anybackground intrinsic inhibition resulting from the test article (i.e., that which occurswithout the presence of test article metabolites). Multiple preincubation time pointsandmultiple concentrations are required to determine the kinetic end pointsKI, whichmay be defined as the concentration of test article required to reach half-maximalrate of inactivation, and kinact, the maximal rate of inactivation. These parameters areused to determine risk of DDI [23]. Unfortunately, assays to determine KI and kinactare laborious and typically reserved for later-stage compounds. A common approachto detect TDI in an abbreviated assay is the so-called “IC50 shift” assay, where twoIC50 values are generated concurrently—one that assesses primarily intrinsic or directinhibition and the other that assesses any additive affect due to metabolism [29, 34,35]. The intrinsic IC50 value may be obtained by incubating multiple concentrationsof test article with liver or cDNA-expressed P450microsomal protein, probe substrateat a concentration equal to KM and NADPH—just as is commonly done in routineassessment of CYP inhibition described earlier. In this case, the incubation is initiatedby addition of microsomal protein or more commonly NADPH. To generate the IC50value with the added effect of metabolism, a preincubation step is included. Here, themicrosomes, test article, and NADPH cofactor are incubated as above, but the probesubstrate is not added until completion of the preincubation period (typically 30 min).The IC50 value obtained without preincubation (or more conservatively, that obtainedwith a equal preincubation period in the absence of NADPH) is compared to the IC50value incorporating the preincubation step with NADPH. If this ratio is greater thana predefined threshold value (typically 1.5 or 2), the presence of TDI is indicated. Inaddition, the shifted IC50 value is generally well correlated with the kinact/KI ratio andso can be used to estimate the risk of TDI [36, 37]. The 30-min preincubation periodis somewhat arbitrary and thought to allow sufficient metabolism to occur to generateany “time-dependent” inhibitory metabolite(s). In practice, a 30-min period may notalways detect TDI and indeed longer incubations (e.g., 90 min or longer) may be


required [38]. In rare instances, TDI may be dependent on metabolic processes notfound in a simple microsomal incubation with NADPH cofactor, and the use of hepa-tocytes as the test systemmay be beneficial [32, 39]. Another approach to abbreviatedtesting is tracking the change in IC50 value with incubation time in a conventionalCYP inhibition assay. This is best accomplished by using fluorimetric assays in whichthe metabolite can be monitored in real time without stopping the reaction, simply bytaking plate readings at multiple intervals as the reaction progresses [40]. If the IC50value decreases upon successive readings, this indicates TDI. If there is no change, orIC50 values increase, this indicates no TDI or test article metabolic depletion, respec-tively. This method is probably not as sensitive as an IC50 shift assay in identifyingTDI, in part because probe substrate is included in the incubation, which in turn mayinhibit conversion of the test article to inhibitory metabolites. Although LC–MS/MSanalysis is not required, the need to generate multiple curves (e.g., ≥4) may requireoccupation of the plate reader throughout the incubation period. Another promisingabbreviated method was recently described by Zimmerlin et al. [37]. In this assay,the determination of inactivation rate was determined using a single concentrationof test article with only four preincubation time points. The concentration of testarticle (10 �M) was chosen to balance solubility with the likelihood to reach thisconcentration level in human blood at therapeutic doses. The authors found a goodcorrelation of inactivation rate with the kinact/KI ratio. Therefore, evaluation of riskand prioritization can occur early in drug discovery when it remains possible to mod-ify chemical structure. An excellent discussion of TDI assays in a drug discoverysetting andmerits of each for impactingmedicinal chemistry efforts has also appearedrecently [41].

25.3 CYTOCHROME P450 INDUCTION METHODS

Within the context of drug metabolism, cytochrome P450 induction may be definedas an increase in the amount of the enzyme, typically in response to a stimulus—suchas exposure to a drug or other xenobiotic. The net result is an increased capacityto metabolize substrates—often this can be the inducing chemical itself. Inductionis often viewed as an evolutionary defense mechanism because it can permit theaccelerated metabolism and excretion of chemicals that, if they accumulate, couldcause toxicity. However, if metabolites generated by induced enzymes are more toxicthan the parent molecule, induction could somewhat paradoxically lead to moretoxicity, not less. Induction of cytochrome P450 in humans has also been recognizedas a significant cause of DDI. Compared to CYP inhibition, DDI from induction areovert toxicities but therapeutic failures resulting from elevated rates of clearance.While serious, they are not as insidious as the sometimes life-threatening toxicity dueto accumulation of unmetabolized drug resulting from CYP inhibition. Indeed, thefinding of induction has historically been a rare cause of drug failures in the clinic[42]. However, induction, if found at concentrations near the intended therapeutic

CYTOCHROME P450 INDUCTION METHODS 461

levels, is generally undesirable and screening efforts to mitigate the effect are oftenwell rewarded.

25.3.1 Cytochrome P450 Induction as a Cause of Toxicity

Induction by chemical agents has been observed in experimental animals for decadesand can be associatedwith hepatotoxicity, carcinogenesis, and other adverse outcomes[43]. These adverse findings are often coincident with the findings of induction inpreclinical safety testing but are generally absent or insignificant in a clinical setting.This has prompted considerable debate on the toxicological relevance of inductionin humans. One recent consensus view has concluded that CYP induction in humansis generally considered an adaptive response with pharmacokinetic impact, but isnot a direct cause of toxicity per se [44]. However, it would not be surprising tofind indirect forms of toxicity as a downstream event of CYP induction. For exam-ple, idiosyncratic adverse reactions following carbamazepine treatment have beenattributed to the formation of reactive metabolites by CYP3A4 and CYP2C19, whichare both inducible [45]. Perhaps the best-studied example is that of the elevated riskof acetaminophen-induced hepatotoxicity occurring in patients with chronic alcoholintake [46, 47].Mechanistically, this observation can be explained by alcohol-inducedCYP2E1 catalyzing formation of a toxic reactive N-acetyl-p-benzoquinone imine(NAPQI) metabolite [48]. Although of relatively low significance in the metabolismof drugs, CYP2E1 may be the primary catalyst in the oxidative metabolism andbioactivation of many low-molecular-weight compounds, including ethanol, certainindustrial solvents, anesthetics, and dietary nitrosamines [49]. Besides induction asa cause of metabolite-induced toxicity, induction could cause premature catabolismand elimination of endogenous substrates such as steroids, bile acids, and lipids,thereby leading to potential disruption in homeostasis and indirect forms of toxicity[50, 51].

25.3.2 Cytochrome P450 Induction as a Cause of Drug Interactions

There are numerous examples of induction as a cause of DDI. A notable exampleis that of CYP3A4 induction following intake of St. John’s wort and its link toorgan transplant failure, resulting from accelerated elimination of immunosuppres-sant drugs [52]. St. John’s wort herbal supplement contains the powerful inducingagent hyperforin. A more recent example is that of mitotane [1-(2-chlorophenyl)-1-(4-chlorophenyl)-2,2-dichloroethane (o,p′-DDD)], which is used in the treatment ofCushing’s syndrome and adrenocortical carcinoma. In the above-mentioned report,two patients failing to respond to mitotane were given sunitinib in an experimen-tal regimen. These patients exhibited an aberrant pharmacokinetic profile of suni-tinib compared to other patients receiving sunitinib alone. Induction of CYP3A4 bymitotane was subsequently confirmed in these two patients as well as two additionalpatients by demonstration of profound induction of midazolam 1′-hydroxylase activ-ity [53]. Like other organochlorine compounds in its class (e.g., the insecticide DDT),


mitotane accumulates in adipose tissue. Therefore, the plasma elimination half-lifeis extremely long (18–159 days) and results in a long-lasting induction of CYP3A4.The example of mitotane is notable because the drug has been available and in usefor >50 years but was recognized only recently as an inducer.

25.3.3 Mechanisms of Cytochrome P450 Induction

Fortunately, the mechanisms of CYP induction are reasonably well characterized tothe point that there are generally accepted in vitromodels to reliably detect agents thatcan induce in vivo. In humans, major inducible isoforms that are relevant to xeno-biotic metabolism include CYP1A1, 1B1, 1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2E1,3A4, and 3A5. With the exception of CYP2E1 (see below), inducible expression ofCYP1, 2, and 3 family members is largely mediated by the aryl hydrocarbon recep-tor, constitutive androstane receptor, and pregnane X receptor, respectively [54–56].These receptors operate as classical ligand-dependent transcription factors. Follow-ing binding of ligand, the receptor translocates to the nucleus, undergoes proteinheterodimerization, and subsequently binds to a specific DNA recognition sequenceto activate target gene expression. Interestingly, AhR and CAR can also mediateinduction in a ligand-independent manner with omeprazole [57] and phenobarbi-tal [55] as notable examples, respectively. In these cases, transcription activationmediated by AhR can proceed by the action of tyrosine kinases, and CAR by phos-phatases. Significant overlap or crosstalk is found in the response of CYP2B, 2C, and3A subfamily members. This crosstalk can occur when inducing compounds activateboth CAR and PXR, or when these promiscuous nuclear receptors bind to identicalregulatory elements in CYP2B, 2C, and 3A target genes [58]. The mechanisms ofCYP2E1 induction are less well understood. Intake of various low molecular weightchemical agents, certain pathophysiological states (e.g., diabetes and obesity), andexposure to hypertonic environments can induce CYP2E1 levels in experimental ani-mals, humans, or human hepatocyte models [49, 59–61]. Themechanism of inductionof this enzyme by exogenous chemicals is generally ascribed to substrate-inducedprotein stabilization [62]. Transcriptional upregulation and mRNA stabilization mayalso be operative andmay depend on the inducing agent [61, 63]. Induction of CYP4Aenzymes that metabolize endogenous and exogenous medium and long-chain fattyacids is mediated in preclinical species by peroxisome-proliferator-activated receptor� [64] and this mechanism also appears to be relevant in humans [65]. The majoractive member in humans, CYP4A11, appears to be only weakly inducible [66],which is in marked contrast to CYP4A induction found in rodents. Certain CYP4Fsubfamily members are gaining increased interest not only because of their involve-ment in the metabolism of endogenous substrates (e.g., very long chain fatty acids)but also in that of xenobiotics [67, 68]. At least some of the CYP4F family membersare inducible by xenobiotics in various tissues and experimental models [69, 70].Nevertheless, it is generally recognized that induction of most CYP enzymes pertain-ing to xenobiotic metabolism is predominantly regulated by AhR, CAR, and PXRand, therefore, the strategy of testing for activation of these receptors or induction ofCYP1A2, 2B6, and 3A4 is viewed as a relatively comprehensive approach to detect


inducers of the major drug-metabolizing CYPs. Indeed, regulatory agencies such asthe US FDA and EMA advocate testing of CYP1A2, 2B6, and 3A4 [22, 23, 24].

25.3.4 Assays to Detect Human Cytochrome P450 Induction

Often, CYP induction assays are conducted at a later stage compared to CYP inhibi-tion assays. This strategy can be due to a greater risk tolerance to unfavorable assayoutcomes or because experiments tend to be more complex (e.g., cell based) andcostly. Human hepatocytes are generally accepted as the most robust and physiolog-ically relevant test system. Engineered cell lines are commonly used for discoveryassays and more recently, immortalized hepatocytes and hepatocyte-like cells derivedfrom tumors or of stem-cell origin are showing good promise as a potentially limitlesssupply of cells from a single individual.

25.3.4.1 Ligand Binding and Reporter Gene Assays Induction models based onnuclear receptor ligand binding in cell-free systems or transactivation (of a reportergene) in transfected cell linemodels have permitted theminiaturization of automation-compatible assays at acceptable costs. In ligand binding assays the displacement ofa labeled, high-affinity ligand to a qualified target is assessed using increasing con-centrations of test article. These assays include scintillation proximity assays (SPAs),fluorescence polarization, time-resolved fluorescence resonance energy transfer (TR-FRET), and yeast-two hybrid models. Zhu et al. (71) used a complex of biotinylatedcDNA-expressed ligand binding domain of the human PXR receptor and steroidreceptor coactivator-1 with the ligand [3H]SR12813 in a SPA. In this example ofa SPA, the test article displaced radiolabeled ligand binds to streptavidin-coatedpolyvinyl toluene beads incorporated with scintillant. The decay particle transfersenergy to the scintillant, which then emits fluorescent light detectable by a photo-multiplier tube. Alternatively, one can use scintillant-coated plates instead of beads.The concentration of receptor and ligand can be optimized to balance analyticalsensitivity with the convention of keeping overall concentrations low, so that theIC50 values will be close to KI values. Relative to fluorescence polarization, fluo-rescence resonance energy transfer assays have advantages of low background andreduced variability in response [72] and have successfully been used to identifyboth CAR and PXR ligands [73, 74]. While ligand binding assays have a numberof advantages over cell-based assays, including relative ease and the absence of acell membrane that could hinder access to the receptor, a significant disadvantage isthat they cannot discriminate between agonists (which bind and elicit an inductionresponse) and antagonist (which bind, but elicit a muted response or none at all).This limitation is overcome in transactivation assays, which feature not only bindingto a full-length receptor but demonstration of functionality by transcriptional acti-vation, typically in the form of a reporter gene such as luciferase [75]. Lehmann etal. (76) developed a reporter gene assay in HepG2 cells, which was subsequentlyfound to provide a reasonable correlation to human hepatocyte response [77]. How-ever, even transactivation assays should not be considered alone in the assessment of


CYP3A4 induction, as some compounds appear to induce through PXR-independentmechanisms [78, 79].

25.3.4.2 Methods to Investigate Cytochrome P450 Induction in Human Hepato-cytes The accepted “gold standard” approach to the assessment of CYP inductionpotential involves the use of plated human hepatocytes as the test system [80]. Inthe typical test, human hepatocytes are dispensed into plates coated with collagenI or another extracellular matrix protein [81]. After adherence, cells are exposed tomedia containing the test article. Ideally, the test article may be dissolved directlyin media, but because of solubility concerns and convenience it is often delivered ina solvent vehicle, most often DMSO. It is important to keep the total concentrationof solvent low (0.1% or less) and normalized across all wells to control for any effectson system response. The test article is replenished daily along with fresh media toachieve targeted exposure times, usually 2 or 3 days in total to take into considerationthe half-life of ∼36 h for CYP3A4 and other CYPs [80, 82]. With mRNA as anend point, robust but submaximal induction responses can be obtained within 24 h[83]. This latter approach would generally be reserved for screening applications,where turnaround times to results are critical and where researchers may tolerate alower dynamic range. Replenishment of test article with media changes is importantbecause the compound could be depleted due to chemical instability, metabolism,adherence to the vessel, or a combination of these. For a definitive test, six or moreconcentrations are usually tested to permit estimation of the EC50 and Emax values[23]. Determining these parameters provides an assessment of the affinity (EC50) forthe receptor involved and the efficacy or magnitude of effect (Emax), which can bebetter utilized to assess the drug-interaction risk profile. Testing more concentrationsalso helps to better evaluate compound cytotoxicity and solubility, but of course ismore resource intensive. After the exposure period, enzyme activity in a multiwellplate can be measured in situ by using probe substrates specific to each CYP enzyme.This measurement can be followed by either mRNA or protein expression analysisusing material prepared from the same cell culture. An example outline of selectedparameters for testing six inducible CYP isoforms with conventional probe substrateswith LC–MS/MS analyte detection is provided in Table 25.1.

25.3.5 The Measurement of Enzyme Activity as a Cytochrome P450Induction End Point

The measurement of enzyme activity in cells after the treatment period is typicallyperformed by quantifying metabolite formation directly in the well (in situ). Thedefinitive measurement is typically assessed using LC–MS/MS and probe substratesas recommended in regulatory guidance documents [23, 24]. Critical parametersinclude the length of exposure of cells to substrate as well as the substrate con-centration. The incubation period must be sufficient to allow adequate turnover ofsubstrate into a quantifiable range of metabolite, but short enough to avoid excessivesubstrate depletion (especially in positive control-inducer-treated cells) and possiblenonlinearity of metabolite formation with time. The substrate concentration chosen,

TA

BL

E25

.1Se

lect

edP

aram

eter

sfo

rC

ytoc

hrom

eP

450

Indu

ctio

nTe

stin

gin

Cul

ture

dH

uman

Hep

atoc

ytes

Enzyme

CYP1A2

CYP2B6

CYP2C8

CYP2C9

CYP2C19

CYP3A4

Positivecontrol

inducer

Omeprazole

Phenobarbital

Rifampicin

Rifampicin

Rifampicin

Rifampicin

Stock

concentration

50mM

100mM

10mM

10mM

10mM

10mM

Solvent

DMSO

10%DMSO

DMSO

DMSO

DMSO

DMSO

Final

concentration

50�M

1mM

10�M

10�M

10�M

10�M

Substratefor

enzyme

activity

Phenacetin

Bupropion

Amodiaquine

Diclofenac

(S)-mephenytoin

Testosterone

Stock

concentration

100mM

50mM

80mM

100mM

200mM

100mM

Solvent

DMSO

Methanol

DMSO

DMSO

Acetonitrile

DMSO

Final

concentration

100

�M

250

�M

100

�M

100

�M

200

�M

200

�M

Incubation

time(min)

6030

3030

120

15

Metabolite

Acetamidophenol

Hydroxybupropion

Desethylamodiaquine

4′-hydroxydiclofenac

4-hydroxy

(S)-mephenytoin

6�-hydroxytestosterone

Internal

standard

Acetamidophenol-

[13C2][15N]

Hydroxybupropion-

[D6]

Desethylamodiaquine-

[D3]

4′-Hydroxy

diclofenac-[13C6]

4′-HydroxyS-

Mephenytoin-[D3]

6�-Hydroxytestosterone-

[D7]

Stock

concentration

10�M

0.1

�M

0.1

�M

0.5

�M

0.5

�M

5�M

Solvent

Acetonitrileand

0.1%

formicacid

Acetonitrileand

0.1%

formicacid

Acetonitrileand0.1%

formicacid

Acetonitrileand0.1%

formicacid

Acetonitrileand

0.1%

formicacid

Acetonitrileand0.1%

formicacid

Final

concentration

2.5

�M

0.025

�M

0.025

�M

0.125

�M

0.125

�M

1.25

�M

Primerfor

qPCR

(TaqMan)

Hs00167927_m1

#Hs03044634-m1

#Hs00258314_m1

Hs00426397_m1

Hs00426380_m1

Hs00430021_m1

465


unlike that in CYP inhibition studies, is typically at least five-fold higher than theKm for the reaction of interest. This concentration helps to minimize or eliminatevelocity changes attributable to substrate loss and also increases reaction velocityand the amount of metabolite formed in a shorter time period. Overall throughputcan be enhanced by strategically pooling certain substrates, although the potentialfor interaction among the substrates and their metabolites (e.g., enzyme inhibition)in donors of differing drug-metabolizing enzyme phenotypes requires careful con-sideration [84, 85]. Further increases in throughput can be achieved by combiningassay miniaturization with higher-density multiwell plate formats. This approach canenable or facilitate the use ofmultichannel pipettes or liquid handling stations [86–89]and can have the added benefit of reducing hepatocyte consumption. Drawbacks tominiaturization include potentially reduced ability to maintain adequate precisionas pipette error increases and a higher level of analytical sensitivity is required forvery small sample sizes. The use of fluorimetric or luminogenic P450 substrates canhelp alleviate the latter concern as well as obviate the need for LC/MS quantitation,but these are not as extensively validated compared to conventional probe substrates[89, 90]. A major concern of enzyme activity assays to measure CYP induction isthe potentially confounding effects of enzyme inhibition by the test articles or theirmetabolites. This can result in false-negative outcomes. One of many examples of this“masking” effect is that of the protease inhibitor ritonavir, which is a CYP3A4 inducerin vitro, but exhibits a decrease in enzyme activity in a plated hepatocyte inductionassay [39, 77]. Direct measurement of protein by Western blotting or, more recently,LC/MS can circumvent this issue [91, 92]. However, the measurement of mRNA, atechnique which is not subject to the confounding effects of enzyme inhibition, isperhaps the most widely adopted method available to address this concern.

25.3.6 The Measurement of mRNA as a Cytochrome P450 End Point

Recent guidances from the US FDA and EMA require the measurement of mRNAfor assessing the induction potential of a new drug candidate (in Europe measure-ment of enzyme activity is also indicated if protein stabilization is suspected as amechanism of induction). A number of methods are available to assess induction ofmRNA including the ribonuclease protection assay [93], branchedDNA signal ampli-fication [94], and reverse-transcription quantitative PCR (RT-qPCR) assays [95, 96].In a recent survey of pharmaceutical companies, 85% of respondents indicated RT-qPCR as their preferred method of measuring induction of mRNA expression [88].In general, mRNA is a more sensitive measure of response than enzyme activityand although it is not a direct determination of functional enzyme, it is increasinglyaccepted as a reliable surrogate. Indeed, good correlations between CYP3A4 mRNAand activity were found when mechanism-based inhibitors or compounds formingmetabolites that were potent reversible inhibitors are excluded [86, 97, 98]. Despitethe confounding effects of enzyme inhibition on the induction assay, the side-by-sidecomparison of mRNA and activity as measures of induction offers valuable insightinto the net effect of these pharmacokinetically opposing properties. Relative quan-titation of mRNA can be determined as a stand-alone assay or in most cases coupled


with an in situ enzyme assay, in which the hepatocytes are subject to RNA isolationfollowing completion of the enzyme activity assay. Total RNA can be isolated withcommercially available kits, such as the RNeasy R© kit from Qiagen, which can beadapted to automation. Quantification of transcript is typically achieved using a one-or two-step method. In the one-step method, the reverse transcription of mRNA tocDNA and the PCR amplification occurs in a single well, whereas in the two-stepmethod these processes occur in separate wells. The one-step assay is less time-consuming, but is generally more costly and may be prone to higher variability. Thelatter may occur because of the potential for variable reaction efficiency of the reversetranscription [99]. Relative quantitation of mRNA expression requires normalizationof amount and quality of starting material using an endogenous housekeeping gene,such as�-actin, glyceraldehyde-3-phosphate dehydrogenase (GAPDH), or 18SRNA.After PCR analysis, fold induction of CYP mRNA isoform expression is determined.A commonly usedmethod is the comparative crossing threshold (Ct) or 2��Ct method[100]. Each target gene sample is normalized (�Ct) by subtracting theCt for its corre-sponding housekeeping gene. The��Ct is then determined by subtracting the�Ct ofthe vehicle-control samples from the test-article-treated samples. Common methodsof RT-qPCR often used in the assessment of CYP induction are the TaqMan R© assays(Life Technologies, Foster City, CA). This chemistry results in the development ofa fluorescence signal upon gene-specific probe hydrolysis, which can be quantifiedwithin the thermocycler instrument. TaqMan assays are preoptimized, but to ensurerobust performance verification experiments are typically performed at the begin-ning, for both housekeeping and target genes. The PCR amplification efficiency canbe determined by plotting Ct response as a function of amount of cDNA or RNA.An ideal assay assumes 100% efficiency, in which each cycle of PCR results in adoubling of product. Slopes of between –3.1 and –3.6 indicate robust performance.Although quantification and quality assessment of RNAmay be obtained by determin-ing the 260/280 absorbance ratio, this step may be eliminated in higher-throughputapproaches. Instead, unsatisfactory data may be simply flagged upon analysis of thefinal data. Finally, as for enzyme activity assays, multiple isoforms can be evaluatedin multiplex formats by using gene-specific probes with distinct fluorophores.Another strategy for increasing throughput in induction studies is to restrict assays

to CYP3A4, because the enzyme is the most abundant human isoform and is respon-sible for the majority of drugmetabolism. It can also serve as a surrogate for assessinginduction of CYP2B and CYP2C isoforms [97] due to the coinduction of CYP3A4with these enzymes via overlap in CAR and PXR receptor activation and crosstalkand interplay between these receptors. Moreover, findings of CYP1A2 inductionby drug candidates are relatively rare. Therefore, CYP1A2, 2B6, and 2C inductionassays might be deferred until drug candidates are further along in development.

25.3.7 Parameters Affecting Outcomes in Human HepatocyteInduction Assays

A number of parameters can affect the responses in the human hepatocyte induc-tion assay including hepatocyte plating density, percent live cell enrichment, period


following plating, choice of maintenance media, and use of cryopreserved versusfresh cells. Although many parameters are not standardized in the industry includingthose above, recent guidance from the FDA suggests general efforts to do so are underway [23].

25.3.7.1 Plating Density Typically, the density (of live cells) is usually targeted toprovide 100% confluence after the cells attach and flatten out. Seeding densities of150,000 to 225,000 cells/cm2 are typical [93, 97, 101, 102], although lower densi-ties (e.g., ∼125,000 cell/cm2) have been used with apparently good success [87]. Inpractice, 100% coverage (e.g., a monolayer) usually does not occur and visual deter-mination of 70% to 80% confluence is often adequate for a robust model, providedthat significant cell-to-cell contact and resulting cuboidal patterns are present.

25.3.7.2 Live Cell Enrichment Enriching the percentage of viable cells prior toplating typically provides a more satisfactory outcome because debris and contentfrom dead cells may elicit unwanted cell signaling responses leading to erratic results.In addition, plating predominantly live cells should provide a more predictable result.Live cell enrichment can be achieved by the use of PercollTM gradients or otherproprietary media, often obtainable from vendors that supply hepatocytes.

25.3.7.3 Post-plating Period After plating, cells attach approximately 4 h later,then flatten and adapt prior to treatment with test article. This period of adaptationvaries among labs, but in most cases is 1 to 2 days long. During this period and fornot well-understood reasons, native levels of CYP typically decline [80, 103, 104].This rate of decline can be donor- and CYP isoform-dependent. While the decline canbe a disadvantage in clearance, metabolite identification or CYP suppression studies,many researchers find this outcome beneficial in induction studies because of thelowered background and resulting increase in dynamic range.

25.3.7.4 Fresh versus Cryopreserved Cells The use of freshly isolated cells gener-ally results in high-quality plating outcomes, but has the drawbacks of unpredictabil-ity in supply and timing of delivery as well as unknown phenotype. The use ofcryopreserved hepatocytes as a model has expanded in recent years and is quicklybecoming the standard approach. The reasons include (1) improved liver tissue pro-cessing resulting in higher yields, viability, and availability, (2) general acceptancethat cryopreservation has no significant impact on induction response, (3) ability tobenchmark and establish quality control criteria from repeated use of the same donor,and (4) wide acceptance of their use by regulatory agencies.

25.4 CYTOCHROME P450 INHIBITION AND INDUCTION BYTHERAPEUTIC PROTEINS

The development and use of therapeutic proteins (monoclonal antibodies, peptides,full or partial length recombinant protein) or other large molecules (e.g., small

REFERENCES 469

molecule conjugates of protein) is now an increasingly important component ofthe pharmacopeia. These agents do not possess the ligand-like properties necessaryto cause inhibition of drug-metabolizing enzymes or transport proteins. Therefore,historically they have not been a cause for concern over CYP inhibition or induction-mediated DDI on the grounds of mechanistic implausibility. However, recent exam-ples of DDI involving therapeutic proteins have led to renewed interest [23, 105, 106],particularly for drug-induced changes in CYP levels resulting from either inductionor suppression. In one example, tocilizumab, a humanized monoclonal antibody tothe interleukin-6 receptor, caused an increased clearance of simvastatin [107]. Themechanism of this DDI was attributed to blockade of interleukin-6 signaling andits downstream suppressive effects on CYP levels. De-suppression and restorationto “normal” CYP levels thus led to increased clearance of the statin. It is clearthat this field continues to evolve as companies generate additional preclinical andclinical data to improve their mechanistic understanding of potential DDI involvingtherapeutic proteins.

25.5 CONCLUSIONS AND CONSIDERATIONSFOR FUTURE DIRECTIONS

In general, recent efforts aimed at higher-throughput testing for CYP inhibition andinduction have focused on adapting drug-registration-quality assays to a screeningmode, retaining as much as possible the assay conditions used, such as same enzymesource and same probe substrates. As regulatory agencies provide guidance that ismore prescriptive, this trend should continue. The integration of CYP inhibition andinduction data via modeling approaches represents another area likely to affect testingstrategies. This is because induction and inhibition can oppose each other, leading toa “net effect” [82] of the compound with respect to DDI potential. The importanceand challenges of integrating inhibition and induction data was recently underscored[108] and given prominence in the FDA’s recent guidance document for DDI [23].

ACKNOWLEDGMENTS

The authors wish to thank Audrey Vardy, Charles L. Crespi, Elke Perloff, and AndrewMason for their useful contributions and comments regarding this manuscript.

REFERENCES

1. Bleicher, K., Bohm, H., Muller, K., Alanine, A. (2003). Hit and lead generation: beyondhigh-throughput screening. Nature Reviews Drug Discovery, 2, 369–378.

2. Honig, P., Wortham, D., Zamani, K., Conner, D., Mullin, J., Cantilena, L. (1993).Terfenadine–ketoconazole interaction. Pharmacokinetic and electrocardiographic con-sequences. JAMA, 269, 1513–1518.


3. Kelly, C.M., Juurlink, D.N., Gomes, T., Duong-Hua, M., Pritchard, K.I., Austin, P.C.,Paszat, F.L. (2010) Selective serotonin reuptake inhibitors and breast cancer mortality inwomen receiving tamoxifen: a population based cohort study. British Medical Journal,340, c693. doi:10.1136/bmj.c693

4. Rodrigues, A.D., Kukulka, M.J., Surber, B.W., Thomas, S.B., Uchic, J.T., Rotert, G.A.,Michel, G., Thome-Kromer, B., Machinist, J.M. (1994). Measurement of liver micro-somal cytochrome P450 (CYP2D6) activity using [O-methyl 14C]dextromethorphan.Analytical Biochemistry, 219, 309–320.

5. Delaporte, E., Slaughter, D.E., Egan, M.A., Gatto, G.J., Santos, A., Shelley, J., Price,E., Howells, L., Dean, D.C., Rodrigues, A.D. (2001). The potential for CYP2D6 inhi-bition screening using a novel scintillation proximity assay-based approach. Journal ofBiomolecular Screening, 6, 225–231.

6. Crespi, C.L., Miller, V.P., Penman, B.W. (1997). Microtiter plate assays for inhibitionof human, drug-metabolizing cytochromes P450. Analytical Biochemistry, 248, 188–190.

7. Cali, J.J., Ma, D., Sobol, M., Simpson, D.J., Frackman, S., Good, T.D., Daily, W.J., Liu,D. (2006). Luminogenic cytochrome P450 assays. Expert Opinion on Drug Metabolism& Toxicology, 2, 629–645.

8. Lim, K.B., Ozbal, C.C., Kassel, D.B. (2010). Development of a high-throughput onlinesolid-phase extraction/tandemmass spectrometrymethod for cytochromeP450 inhibitionscreening. Journal of Biomolecular Screening, 15, 447–452.

9. Luippold, A.H., Arnhold, T., Jorg, W., Kruger, B., Sussmuth, R.D. (2011). Applicationof a rapid and integrated analysis system (RIAS) as a high-throughput processing tool forin vitro ADME samples by liquid chromatography/tandem mass spectrometry. Journalof Biomolecular Screening, 16, 370–377.

10. Sukumaran, S.M., Potsaid, B., Lee,M.Y., Clark, D.S., Dordick, J.S. (2009). Developmentof a fluorescence-based, ultra high-throughput screening platform for nanoliter-scalecytochrome P450 microarrays. Journal of Biomolecular Screening, 14, 668–678.

11. Krippendorff, B., Lienau, P., Reichel, A., Huisinga, W. (2007). Optimizing classificationof drug–drug interaction potential for CYP450 isoenzyme inhibition assays in early drugdiscovery. Journal of Biomolecular Screening, 12, 92–99.

12. Crespi, C.L. (1995). Xenobiotic-metabolizing human cells as tools for pharmacologicaland toxicological research. Advances in Drug Research, 26, 179–235.

13. Stresser, D.M. (2004). High-throughput screening of human cytochrome P450 inhibitorsusing fluorometric substrates: methodology for 25 enzyme substrate pairs. In: Optimiza-tion in Drug Discovery: In Vitro Methods (Yan, Z., Caldwell, G.W., Eds.) in the Series“Methods in Pharmacology and Toxicology” (Kang, Y.J., Series Ed.). Humana Press:Totowa, NJ.

14. Chauret, N., Dobbs, B., Lackman, R., Bateman, K., Nicoll-Griffith, D., Stresser, D.M.,Miller, V.P., Ackermann, J.M., Turner, S.L., Crespi, C.L. (2001). The use of 3-[2-(N,N-diethyl-N-methylammonium)ethyl]-7-methoxy-4-methylcoumarin (AMMC) as a spe-cific CYP2D6 probe in human liver microsomes. Drug Metabolism and Disposition, 29,1196–1200.

15. Stresser, D.M., Turner, S.D., Blanchard, A.P., Miller, V.P., Crespi, C.L. (2002).Cytochrome P450 fluorometric substrates: identification of isoform-selective probes forrat CYP2D2 and human CYP3A4. Drug Metabolism and Disposition, 30, 845–852.

REFERENCES 471

16. Sudsakorn, S., Skell, J., Williams, D.A., O’Shea, T.J., Liu, H. (2007). Evaluation of3-O-methylfluorescein as a selective fluorometric substrate for CYP2C19 in human livermicrosomes. Drug Metabolism and Disposition, 35, 841–847.

17. Stresser, D.M., Turner, S.D., Blanchard, A.B., Miller, V.P., Erve, J.C.L., Dandeneau,A.C., Crespi, C.L. (2000). Substrate-dependentmodulation of CYP3A4 catalytic activity:analysis of 27 test compounds with four fluorometric substrates. Drug Metabolism andDisposition, 28, 1440–1448.

18. Ito, K., Hallifax, D., Obach, S., Houston, J.B. (2005). Impact of parallel pathways ofdrug elimination and multiple CYP involvement on drug–drug interactions: CYP2D6paradigm. Drug Metabolism and Disposition, 33, 837–844.

19. Vandenbrink, B.M., Foti, R.S., Rock, D.A., Wienkers, L.C., Wahlstrom, J.L. (2012).Prediction of CYP2D6 drug interactions from in vitro data: evidence for substrate-dependent inhibition. Drug Metabolism and Disposition, 40, 47–53.

20. Fowler, S., Zhang, H. (2008) In vitro evaluation of reversible and irreversible cytochromeP450 inhibition: current status onmethodologies and their utility for predicting drug–druginteractions. AAPS Journal, 10, 410–424.

21. Veith, H., Southall, N., Huang, R., James, T., Fayne, D., Artemenko, N., Shen, M.,Inglese, J., Austin, C.P., Lloyd, D.G., Auld, D.S. (2009). Comprehensive characterizationof cytochrome P450 isozyme selectivity across chemical libraries.Nature Biotechnology,27, 1050–1055.

22. US FDA (2006). Guidance for Industry: Drug Interaction Studies—Study Design, DataAnalysis, and Implications for Dosing and Labeling. Draft Guidance. Available athttp://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/ucm072101.pdf; accessed July 22, 2011.

23. US FDA (2012). Guidance for Industry Drug Interaction Studies—Study Design,Data Analysis, Implications for Dosing, and Labeling Recommendations. DraftGuidance. Available at http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM292362.pdf; accessed May 23, 2012.

24. European Medicines Agency (2012). Guideline on the Investigation of Drug InteractionsCPMP/EWP/560/95/Rev. 1

25. Walsky, R.L., Obach, R.S. (2004). Validated assays for human cytochrome P450 activi-ties. Drug Metabolism and Disposition, 32, 647–660.

26. Peng, S.X., Barbone, A.G., Ritchie, D.M. (2003) High-throughput cytochrome P450inhibition assays by ultrafast gradient liquid chromatography with tandem mass spec-trometry using monolithic columns. Rapid Communications in Mass Spectrometry, 17,509–518.

27. Youdim, K.A., Lyons, R., Payne, L., Jones, B.C., Saunders, K. (2008). An automated,high-throughput, 384 well Cytochrome P450 cocktail IC50 assay using a rapid resolutionLC–MS/MS end-point. Journal of Pharmaceutical and Biomedical Analysis, 48, 92–99.

28. Youdim, K.A., Saunders, K.C. (2010) A review of LC–MS techniques and high-throughput approaches used to investigate drug metabolism by cytochrome P450s. Jour-nal of Chromatography B, 878, 1326–1336.

29. Perloff, E.S., Mason, A.K., Dehal, S.S., Blanchard, A.P., Morgan, L., Ho, T., Dandeneau,A., Crocker, R.M., Chandler, C.M., Boily, N., Crespi, C.L., Stresser, D.M. (2009).Validation of a cytochrome P450 time dependent inhibition assay: a two time point IC50shift approach facilitates kinact assay design. Xenobiotica, 39, 99–112.


30. Grimm, S.W., Einolf, H.J., Hall, S.D., He, K., Lim, H.K., Ling, K.H., Lu, C., Nomeir,A.A., Seibert, E., Skordos, K.W., Tonn, G.R., van Horn, R., Wang, R.W., Wong, N.Y.,Yang, T.J., Obach, R.S. (2009). The conduct of in vitro studies to address time-dependentinhibition of drug-metabolizing enzymes: a perspective of the pharmaceutical researchand manufacturers of America. Drug Metabolism and Disposition, 37, 1355–1370.

31. Kalgutkar, A.S., Obach, R.S., Maurer, T.S. (2007). Mechanism-based inactivation ofcytochrome P450 enzymes: chemical mechanisms, structure–activity relationships andrelationship to clinical drug–drug interactions and idiosyncratic adverse drug reactions.Current Drug Metabolism, 8, 407–447.

32. Parkinson, A., Kazmi, F., Buckley, D.B., Yerino, P., Ogilvie, B.W., Paris, B.L. (2010).System-dependent outcomes during the evaluation of drug candidates as inhibitorsof cytochrome P450 (CYP) and uridine diphosphate glucuronosyltransferase (UGT)enzymes: human hepatocytes versus liver microsomes versus recombinant enzymes.Drug Metabolism and Pharmacokinetics, 25, 16–27.

33. Zientek, M., Dalvie, D. (2012). Utilization of a multi-staged time-dependent inhibitionassay to assess the impact of intestinal metabolism on drug–drug interaction potential.Drug Metabolism and Disposition, 40, 467–473.

34. Berry, L.M., Zhao, Z. (2008). An examination of IC50 and IC50-shift experiments inassessing time-dependent inhibition of CYP3A4, CYP2D6 and CYP2C9 in human livermicrosomes. Drug Metabolism Letters, 21, 51–59.

35. Grime, K.H., Bird, J., Ferguson, D., Riley, R.J. (2009). Mechanism-based inhibition ofcytochrome P450 enzymes: an evaluation of early decision making in vitro approachesand drug–drug interaction prediction methods. European Journal of PharmaceuticalSciences, 36, 175–191.

36. Obach, R.S., Walsky, R.L., Venkatakrishnan, K. (2007). Mechanism-based inactivationof human cytochrome P450 enzymes and the prediction of drug–drug interactions. DrugMetabolism and Disposition, 35, 246–255.

37. Zimmerlin, A., Trunzer, M., Faller, B. (2011). CYP3A time-dependent inhibition riskassessment validated with 400 reference drugs. Drug Metabolism and Disposition, 39,1039–1046.

38. Henne, K.R., Thuy, B.T., Vandenbrink, B.M., Rock, D.A., Aidasani, D., Subramanian,R., Mason, A.K., Stresser, D.M., Teffera, Y., Wong, S.G., Johnson, M.G., Chen, X.,Tonn, G.R., Wong, B.K. (2012). Sequential metabolism of AMG 487, a novel CXCR3antagonist, results in formation of quinone reactive metabolites that covalently modifyCYP3A4 Cys239 and cause time-dependent inhibition of the enzyme. Drug Metabolismand Disposition, 40, 1429–1440.

39. McGinnity, D.F., Zhang, G., Kenny, J.R., Hamilton, G.A., Otmani, S., Stams, K.R.,Haney, S., Brassil, P., Stresser, D.M., Riley, R.J. (2009). Evaluation of multiple in vitrosystems for assessment of CYP3A4 induction in drug discovery: human hepatocytes,pregnane X receptor reporter gene, and Fa2N-4 and HepaRG cells. Drug MetabolismDisposition, 37, 1259–1268.

40. Yan, Z., Rafferty, B., Caldwell, G.W., Masucci, J.A. (2002). Rapidly distinguishingreversible and irreversible CYP450 inhibitors by using fluorometric kinetic analyses.European Journal of Drug Metabolism and Pharmacokinetics, 27, 281–287.

41. Orr, S.T.M., Ripp, S.L., Ballard, T.E., Henderson, J.L., Scott, D.O., Obach, R.S., Sun,H., Kalgutkar, A.S. (2012). Mechanism-based inactivation (MBI) of cytochrome P450

REFERENCES 473

enzymes: structure–activity relationships and discovery strategies to mitigate drug–druginteraction risks. Journal of Medicinal Chemistry, 55, 4896–4933.

42. Smith, D.A. (2000). Induction and drug development. European Journal of Pharmaceu-tical Sciences, 11, 185–189.

43. Conney, A.H. (1967). Pharmacological implications of microsomal enzyme induction.Pharmacological Reviews, 19, 317–366.

44. Mohutsky, M.A., Romeike, A., Meador, V., Lee, W.M., Fowler, J., Francke-Carroll, S.(2010). Hepatic drug-metabolizing enzyme induction and implications for preclinicaland clinical risk assessment. Toxicological Pathology, 38, 799–809.

45. Pearce, R.E., Lu, W., Wang, Y., Uetrecht, J.P., Correia, M.A., Leeder, J.S. (2008). Path-ways of carbamazepine bioactivation in vitro III: the role of human cytochrome P450enzymes in the formation of 2,3-dihydroxycarbamazepine. Drug Metabolism and Dis-position, 36, 1637–1649.

46. Seeff, L.B., Cuccherini, B.A., Zimmerman, H.J., Adler, E., Benjamin, S.B. (1986).Acetaminophen hepatotoxicity in alcoholics. A therapeutic misadventure. Annals ofInternal Medicine, 104, 399–404.

47. Schmidt, L.E., Dalhoff, K., Poulsen, H.E. (2002). Acute versus chronic alcohol con-sumption in acetaminophen-induced hepatotoxicity. Hepatology, 35, 876–882.

48. Thummel, K.E., Slattery, J.T., Ro, H., Chien, J.Y., Nelson, S.D., Lown, K.E., andWatkins, P.B. (2000). Ethanol and production of the hepatotoxic metabolite of acet-aminophen in healthy adults. Clinical Pharmacology and Therapeutics, 67, 591–599.

49. Lieber, C.S. (1997). Cytochrome P-4502E1: its physiological and pathological role.Physiological Reviews, 77, 517–544.

50. Amacher, D.E. (2010). The effects of cytochrome P450 induction by xenobiotics onendobiotic metabolism in pre-clinical safety studies. Toxicology Mechanisms and Meth-ods, 20, 159–166.

51. Wang, Z., Lin, Y.S., Zheng, X.E., Senn, T., Hashizume, T., Scian, M., Dickmann, L.J.,Nelson, S.D., Baillie, T.A., Hebert,M.F., Blough, D., Davis, C.L., Thummel, K.E. (2012).An inducible cytochrome P450 3A4-dependent vitamin D catabolic pathway.MolecularPharmacology, 81, 498–509.

52. Ruschitzka, F., Meier, P.J., Turina, M., Luscher, T.F., Noll, G. (2000). Acute hearttransplant rejection due to Saint John’s wort. Lancet, 355, 548–549.

53. van Erp, N.P., Guchelaar, H.J., Ploeger, B.A., Rominjn, J.A., denHartigh, J., Gelderblom,H. (2011). Mitotane has a strong and a durable inducing effect on CYP3A4 activity.European Journal of Endocrinology, 164, 621–626.

54. Denison, M.S., Soshilov, A.A., He, G., DeGroot, D.E., Zhao, B. (2011). Exactly thesame but different: promiscuity and diversity in the molecular mechanisms of action ofthe aryl hydrocarbon (dioxin) receptor. Toxicological Sciences, 124, 1–22.

55. Kodoma, S., Negishi, M. (2006). Phenobarbital confers its diverse effects by activatingthe orphan nuclear receptor CAR. Drug Metabolism Reviews, 38, 75–87.

56. Stanley, L.A., Horsburgh, B.C., Ross, J., Scheer, N., Wolf, C.R. (2006). PXR and CAR:nuclear receptors which play a pivotal role in drug disposition and chemical toxicity.Drug Metabolism Reviews, 38, 515–597.

57. Backlund, M., Ingelman-Sundberg, M. (2005). Regulation of aryl hydrocarbon receptorsignal transduction by protein tyrosine kinases. Cellular Signalling, 17, 39–48.


58. Lin, J.H. (2006). CYP induction-mediated drug interactions: in vitro assessment andclinical implications. Pharmaceutical Research, 23, 1089–1116.

59. Emery, M., Fisher, J.M., Chien, J.Y., Kharasch, E.D., Dellinger, E.P., Kowdley, K.V.,Thummel, K.E. (2003). CYP2E1 activity before and after weight loss in morbidly obesesubjects with nonalcoholic fatty liver disease. Hepatology, 38, 428–435.

60. Song, B.J., Veech, R.L., Saenger, P. (1990). Cytochrome P450IIE1 is elevated in lympho-cytes from poorly controlled insulin-dependent diabetics. Journal of Clinical Endocrinol-ogy & Metabolism, 71, 1036–1040.

61. Ito, T., Asakura, K., Tougou, K., Fukuda, T., Kubota, R., Nonen, S., Fujio, Y., Azuma,J. (2007). Regulation of cytochrome P450 2E1 under hypertonic environment throughTonEBP in human hepatocytes. Molecular Pharmacology, 72, 173–181.

62. Gonzalez, F.J. (2007). The 2006 Bernard B. Brodie Award Lecture. CYP2E1. DrugMetabolism and Disposition, 35, 1–8.

63. Woodcroft, K.J., Hafner, M.S., Novak, R.F. (2002). Insulin signaling in the transcrip-tional and posttranscriptional regulation of CYP2E1 expression. Hepatology, 35, 263–273.

64. Hardwick, J.P., Osei-Hyiaman, D., Wiland, H., Abdelmegeed, M.A., Song, B.J. (2009).PPAR/RXR regulation of fatty acid metabolism and fatty acid omega-hydroxylase(CYP4) isozymes: implications for prevention of lipotoxicity in fatty liver disease. PPARResearch doi:10.1155/2009/952734

65. Savas, U., Machemer, D.E., Hsu, M.H., Gaynor, P., Lasker, J.M., Tukey, R.H., Johnson,E.F. (2009). Opposing roles of peroxisome proliferator–activated receptor alpha andgrowth hormone in the regulation of CYP4A11 expression in a transgenic mouse model.Journal of Biological Chemistry, 284, 16541–16552.

66. Raucy, J.L., Lasker, J., Ozaki, K., Zoleta, V. (2004). Regulation of CYP2E1 by ethanoland palmitic acid and CYP4A11 by clofibrate in primary cultures of human hepatocytes.Toxicological Sciences, 79, 233–241.

67. Kalsotra, A., Strobel, H.W. (2006). Cytochrome P450 4F subfamily: at the crossroads ofeicosanoid and drug metabolism. Pharmacology & Therapeutics, 112, 589–611.

68. Jin, Y., Zollinger, M., Borell, H., Zimmerlin, A., Patten, C.J. (2011). CYP4F enzymes areresponsible for the elimination of fingolimod (FTY720), a novel treatment of relapsingmultiple sclerosis. Drug Metabolism and Disposition, 39, 191–198.

69. Hsu, M.H., Savas, U., Lasker, J.M., Johnson, E.F. (2011). Genistein, resveratrol, and5-aminoimidazole-4-carboxamide-1-�-d-ribofuranoside induce cytochrome P450 4F2expression through an AMP-activated protein kinase-dependent pathway. Journal ofPharmacology and Experimental Therapeutics, 337, 125–136.

70. Zhao, Z., He, X., Bi, Y., Xia, Y., Tao, N., Li, L., Ma, A. (2009). Induction of CYP4F3 bybenzenemetabolites in humanwhite blood cells in vivo in human promyelocytic leukemiccell lines and ex vivo in human blood neutrophils. Drug Metabolism and Disposition,37, 282–291.

71. Zhu, Z., Kim, S., Chen, T., Lin, J.H., Bell, A., Bryson, J., Dubaquie, Y., Yan, N.,Yanchunas, J., Xie, D., et al. (2004). Correlation of high-throughput pregnane X receptor(PXR) transactivation and binding assays. Journal of Biomolecular Screening, 9, 533–540.

72. Raucy, J.L., Lasker, J. (2010) Current in vitro high throughput screening approaches toassess nuclear receptor activation. Current Drug Metabolism, 11, 806–814.

REFERENCES 475

73. Moore, L.B., Parks, D.J., Jones, S.A., Bledsoe, R.K., Consler, T.G., Stimmel, J.B.,Goodwin, B., Liddle, C., Blanchard, S.G., Willson, T.M., Collins, J.L., Kliewer, S.A.(2000). Orphan nuclear receptors constitutive androstane receptor and pregnane X recep-tor share xenobiotic and steroid ligands. Journal of Biological Chemistry, 275, 15122–15127.

74. Shukla, S.J., Nguyen, D.T., Macarthur, R., Simeonov, A., Frazee, W.J., Hallis, T.M.,Marks, B.D., Singh, U., Eliason, H.C., Printen, J., Austin, C.P., Inglese, J., Auld, D.S.(2009). Identification of pregnane X receptor ligands using time-resolved fluorescenceresonance energy transfer and quantitative high-throughput screening. Assay and DrugDevelopment Technologies, 7, 143–169.

75. Raucy, J.L., Warfe, L., Yueh, M.-F., Allen, S.W. (2002). A cell-based reporter gene assayfor determining induction of CYP3A4 in a high volume system. Journal of Pharmacologyand Experimental Therapeutics, 303, 412–423.

76. Lehmann, J.M., McKee, D.D., Watson, M.A., Willson, T.M., Moore, J.T., and Kliewer,S.A. (1998). The human orphan nuclear receptor PXR is activated by compounds thatregulate CYP3A4 gene expression and cause drug interactions. Journal of ClinicalInvestigation, 102, 1016–1023.

77. Luo, G., Cunningham, M., Kim, S., Burn, T., Lin, J., Sinz, M., Hamilton, G., Rizzo, C.,Jolley, S., Gilbert, D., Downey, A., Mudra, D., Graham, R., Carroll, K., Xie, J., Madan,A., Parkinson, A., Christ, D., Selling, B., LeCluyse, E., Gan, L.S. (2002). CYP3A4induction by drugs: correlation between a pregnane X receptor reporter gene assay andCYP3A4 expression in human hepatocytes. Drug Metabolism and Disposition, 30, 795–804.

78. Prueksaritanont, T., Richards, K.M., Qiu, Y., Strong-Basalyga, K., Miller, A., Li, C.,Eisenhandler, R., Carlini, E.J. (2005). Comparative effects of fibrates on drug metabo-lizing enzymes in human hepatocytes. Pharmaceutical Research, 22, 71–78.

79. Li, L., Sinz, M.W., Zimmermann, K., Wang, H. (2012). An insulin-like growth fac-tor 1 receptor inhibitor induces CYP3A4 expression through a pregnane X receptor-independent, noncanonical constitutive androstane receptor-related mechanism. Journalof Pharmacology and Experimental Therapeutics, 340, 688–697.

80. Maurel, P. (1996). The use of adult humanhepatocytes in primary culture and other in vitrosystems to investigate drug metabolism in man. Advanced Drug Delivery Reviews, 22,105–132.

81. Pichard, L., Fabre, J.M., Domergue, J., Fabre, G., Saint Aubert, B., Mourad, G., Mau-rel, P. (1990). Molecular mechanism of cyclosporine A drug interactions: inducers andinhibitors of screening for inducers and inhibitors of cytochrome P450 (cyclosporine oxi-dase) in primary culture of human hepatocytes and in livermicrosomes.Drug Metabolismand Disposition, 18, 595–606.

82. Fahmi,O.A., Hurst, S., Plowchalk,D., Cook, J., Guo, F., Youdim,K., Dickins,M., Phipps,A., Darekar, A., Hyland, R., Obach, R.S. (2009). Comparison of different algorithmsfor predicting clinical drug-drug interactions, based on the use of CYP3A4 in vitrodata: predictions of compounds as precipitants of interaction. Drug Metabolism andDisposition, 37, 1658–1666.

83. Zhang, J.G., Ho, T., Callendrello, A.L., Crespi, C.L., Stresser, D.M. (2010). A multi-endpoint evaluation of cytochrome P450 1A2, 2B6 and 3A4 induction response in humanhepatocyte cultures after treatment with�-naphthoflavone, phenobarbital and rifampicin.Drug Metabolism Letters, 4, 185–194.


84. Rhodes, S.P., Otten, J.N., Hingorani, G.P., Hartley, D.P., Franklin, R.B. (2011). Simul-taneous assessment of cytochrome P450 activity in cultured human hepatocytes forcompound-mediated induction of CYP3A4, CYP2B6, and CYP1A2. Journal of Phar-macological and Toxicological Methods, 63, 223–226.

85. Feidt, D.M., Klein, K., Hofmann, U., Riedmaier, S., Knobeloch, D., Thasler, W.E.,Weiss, T.S., Schwab, M., Zanger, U.M. (2010). Profiling induction of cytochrome p450enzyme activity by statins using a new liquid chromatography-tandemmass spectrometrycocktail assay in human hepatocytes. Drug Metabolism and Disposition, 38, 1589–11597.

86. Kamiguchi, N., Aoyama, E., Okuda, T., Moriwaki, T. (2010). A 96-well plate assay forCYP4503A induction using cryopreserved human hepatocytes. Drug Metabolism andDisposition, 38, 1912–1916.

87. Kaneko, A., Kato, M., Endo, C., Nakano, K., Ishigai, M., Takeda, T. (2010). Predictionof clinical CYP3A4 induction using cryopreserved human hepatocytes. Xenobiotica, 40,791–799.

88. Chu, V., Einolf, H.J., Evers, R., Kumar, G., Moore, D., Ripp, S., Silva, J., Sinha, V., Sinz,M., Skerjanec, A. (2009). In vitro and in vivo induction of cytochrome P450: a survey ofthe current practices and recommendations: a pharmaceutical research andmanufacturersof America perspective. Drug Metabolism and Disposition, 37, 1339–1354.

89. Larson, B., Moeller, T., Banks, P., Cali, J.J. (2011). Automated triplexed hepatocyte-based viability and CYP1A and -3A induction. Journal of Biomolecular Screening, 16,895–902.

90. Doshi, U., Li, A.P. (2011) Luciferin IPA-based higher throughput human hepatocytescreening assays for CYP3A4 inhibition and induction. Journal of Biomolecular Screen-ing, 16, 903–909.

91. Williamson, B.L., Purkayastha, S., Hunter, C.L., Nuwaysir, L., Hill, J., Easterwood, L.,Hill, J. (2011). Quantitative protein determination for CYP induction via LC–MS/MS.Proteomics, 11, 33–41.

92. Schaefer, O., Ohtsuki, S., Kawakami, H., Inoue, T., Liehner, S., Saito, A., Sakamoto,A., Ishiguro, N., Matsumaru, T., Terasaki, T., Ebner, T. (2012). Absolute quantificationand differential expression of drug transporters, cytochrome P450 enzymes, and UDP-glucuronosyltransferases in cultured primary human hepatocytes. Drug Metabolism andDisposition, 40, 93–103.

93. Surry, D.D., McAllister, G., Meneses-Lorente, G., Evans, D.C. (1999). High throughputribonuclease protection assay for the determination of CYP3A mRNA induction incultured rat hepatocytes. Xenobiotica, 29, 827–838.

94. Hartley, D.P., Klaassen, C.D. (2000). Detection of chemical-induced differential expres-sion of rat hepatic cytochrome P450 mRNA transcripts using branched DNA signalamplification technology. Drug Metabolism and Disposition, 28, 608–616.

95. Roymans, D., van Looveren, C., Leone, A., Parker, J.B., McMillian, M., Johnson,M.D., Koganti, A., Silber, P., Mannens, G., Meuldermans, W. (2004). Determinationof cytochrome P450 1A2 and cytochrome P450 3A4 induction in cryopreserved humanhepatocytes. Biochemical Pharmacology, 67, 427–437.

96. Nishimura, M., Narimatsu, S., Naito, S. (2009). Evaluation of induction potency of newdrug candidates on CYP1A2 and CYP3A4 using real-time one-step RT-PCR in primary

REFERENCES 477

cultures of cryopreserved human hepatocytes. Drug Metabolism and Pharmacokinetics,24, 446–450.

97. Fahmi, O.A., Kish, M., Boldt, S., Obach, R.S. (2010). Cytochrome P450 3A4 mRNAis a more reliable marker than CYP3A4 activity for detecting pregnane X receptor-activated induction of drug-metabolizing enzymes. Drug Metabolism and Disposition,38, 1605–1611.

98. Shou, M., Hayashi, M., Pan, Y., Xu, Y., Morrissey, K., Xu, L., Skiles, G.L. (2008). Mod-eling, prediction, and in vitro in vivo correlation of CYP3A4 induction.Drug Metabolismand Disposition, 36, 2355–2370.

99. Wong, M.L., Medrano, J.F. (2005). Real-time PCR for mRNA quantitation. BioTech-niques, 39, 75–85.

100. Livak, K.J., Schmittgen, T.D. (2001). Analysis of relative gene expression data usingreal-time quantitative PCR and the 2−��C

T Ct method. Methods, 25, 402–408.

101. LeCluyse, E.L. (2001). Human hepatocyte culture systems for the in vitro evaluationof cytochrome P450 expression and regulation. European Journal of PharmaceuticalSciences, 13, 343–368.

102. Hariparsad, N., Carr, B.A., Evers, R., Chu, X. (2008). Comparison of immortalizedFa2N-4 cells and human hepatocytes as in vitro models for cytochrome P450 induction.Drug Metabolism and Disposition, 36, 1046–1055.

103. Silva, J.M., Morin, P.E., Day, S.H., Kennedy, B.P., Payette, P., Rushmore, T., Yergey,J.A., Nicoll-Griffith, D.A. (1998). Refinement of an in vitro cell model for cytochromeP450 induction. Drug Metabolism and Disposition, 26, 490–496.

104. Grover, G.S., Brayman, T.G., Voorman, R.L., Ware, J.A. (2007). Development of in vitromethods to predict induction of CYP1A2 and CYP3A4 in humans. Assay and DrugDevelopment Technologies, 5, 793–804.

105. Kraynov, E., Martin, S.W., Hurst, S., Fahmi, O.A., Dowty, M., Cronenberger, C., Loi,C.-M., Kuang, B., Fields, O., Fountain, S., Awwad, M., Wang, D. (2011). How currentunderstanding of clearance mechanisms and pharmacodynamics of therapeutic proteinscan be applied for evaluation of their drug–drug interaction potential? Drug Metabolismand Disposition, 39, 1779–1783.

106. Girish, S., Martin, S.W., Peterson, M.C., Zhang, L.K., Zhao, H., Balthasar, J., Evers, R.,Zhou, H., Zhu, M., Klunk, L., Han, C., Berglund, E.G., Huang, S.-M., Joshi, A. (2011).AAPSworkshop report: strategies to address therapeutic protein–drug interactions duringclinical development. AAPS Journal, 13, 405–416.

107. Schmitt, C., Kuhn, B., Zhang, X., Kivitz, A.J., Grange, S. (2011). Disease drug–druginteraction involving tocilizumab and simvastatin in patients with rheumatoid arthritis.Clinical Pharmacology & Therapeutics, 89, 735–740.

108. Kirby, B.J., Collier, A.C., Kharasch, E.D., Whittington, D., Thummel, K.E., Unadkat,J.D. (2011). Complex drug interactions of HIV protease inhibitors 1: inactivation, induc-tion, and inhibition of cytochrome P450 3A by ritonavir or nelfinavir. Drug Metabolismand Disposition, 39, 1070–1078.

26HIGH-THROUGHPUT YEAST-BASEDASSAYS TO STUDYRECEPTOR-MEDIATED TOXICITY

Johanna Rajasarkka and Marko Virta

26.1 INTRODUCTION

The yeast Saccharomyces cerevisiae is a unicellular eukaryotic organism that hasbeen a subject of numerous biological studies. It is easy to cultivate, and there arenumerous tools available for the genetic manipulation of yeast including differentselection markers. Yeast has become a popular organism to study nuclear receptor(NR) functioning and ligand binding. To date, several yeast-based assays utilizingdifferent eukaryotic NRs have been developed. The first and most studied yeast NRassays have been developed for the human estrogen receptor. Assays for the androgen,progesterone, aryl hydrocarbon, retinoid, thyroid, glucocorticoid, and mineralocorti-coid receptors, among others, have also been developed.In addition to studying the receptor biology, the assays have proven their applica-

bility in screening for chemicals and mixtures with endocrine-disrupting activity, andin environmental and food safety monitoring. Some yeast-based hormonal bioassayshave already been internationally validated for hormonal activity testing purposes[1, 2]. Yeast-based NR assays have several advantages. They are cost-efficient andsuitable for routine high-throughput screening (HTS) programs. The assays are sim-ple and robust and do not require any specialized performers, high-tech cultivation, ordetection equipment. They are suitable for different and even complex sample matri-ces, including mixture effect studies. Yeast NR assays are especially suitable forrobust initial screening of chemicals and samples for further verification. However,


479

480 HIGH-THROUGHPUT YEAST-BASED ASSAYS

because yeast NR assays only detect the binding and transactivation potency of acompound to a receptor, no conclusions about toxicity endpoints can be drawn.Most of current yeast-based NR assays have been designed mainly for 96-well

plate format, but some have already been miniaturized to 384- and even 1536-wellplate formats [3, 4]. Since the 384-well plate format is already routinely used in othercell-based assays, it is certain that in the future more yeast-based NR assays willbecome adapted to higher-density well plates.

26.2 YEAST AS A NUCLEAR RECEPTOR STUDY ORGANISM

NRs form a large superfamily of receptors. In the human genome, 48 NRs have beendetected so far. The receptors are divided into three subclasses [5]: type I for steroidhormone receptors (e.g., estrogen and androgen receptors), type II for other receptors(e.g., thyroid and retinoid receptors), and type III for the so-called orphan receptors,which lack a known natural ligand. The receptors share a common mode of action:upon ligand binding the receptor usually forms a dimer with another receptor and thentranslocates to the cell nucleus in order to regulate transcription of genes controlled byresponse elements of the particular receptor. NRs control many important signalingpathways involved in growth, reproduction, and development processes.Several man-made chemicals and environment pollutants have been shown to

mimic or suppress the action of natural hormones primarily via binding to the recep-tors. The compounds share some structural similarities with the natural hormones.The International Programme on Chemical Safety [6] defined these compounds asendocrine-disrupting chemicals (EDCs). The adverse health effects of EDCs havebeen reported in numerous wildlife and laboratory animal studies. These includehormone-dependent cancers, effects in neuronal and immune systems, impairedreproduction and development, and embryotoxicity. It is suspected that EDCs alsoaffect human health and reproduction [6, 7].Bothmammalian-cell- and yeast-cell-based assays are popularmethods to identify,

detect, and study EDCs. Although no endogenous NRs or NR-specific cofactors havebeen detected in yeast cells, different studies have demonstrated that functionallyactive human and other vertebrate NRs are successfully expressed in yeast cells[8–11]. In fact, the lack of an NR-dependent signaling system in yeast can be seen asan advantage since there is no risk of crosstalk between the studied and endogenousreceptors, as can be the case in mammalian cells.Yeast assays have also other advantages over mammalian-cell-based assays. Yeast

is a fast-growing organism, easy to cultivate, and it does not require serum for growth.Yeast assays are more robust, cost-efficient, and require no specialized performersor instruments. Yeast cells are also usually more resistant to toxic effects, especiallywhen testing environmental samples.Although the function of NRs in yeast cells is very similar to that in mam-

malian cells, there are differences between the assays. The major difference is theability to distinguish between agonistic, i.e., “receptor activating,” and antagonistic,i.e., “receptor repressing,” molecules. In mammalian cells, many partial and pure

YEAST AS A NUCLEAR RECEPTOR STUDY ORGANISM 481

antagonists are known, but in yeast cells receptor-binding molecules usually showonly agonistic effects. The distinction between antagonists and agonists in mam-malian cells is based on a conformational change of the receptor upon moleculebinding. Depending on the adopted conformation of the molecule-bound receptor,interplaying receptors, and the available cofactor proteins in the particular cell type,the receptor functions either as an activating or repressing transcription factor. In fact,a class of compounds called selective nuclear receptor modulators (SNRs) can actas both agonists and antagonists, depending on the tissue. Since no cofactor proteinsare present in yeast cells, antagonists and SNRs can behave very differently in yeastassays, showing agonistic or antagonistic activity, or both. It has been proposed thatin screening hormonal compounds, the insensitivity of yeast assays to antagonistsmight even lower the chance of false-negatives [12].Many of the current yeast assays are at least in part able to detect the antagonistic

effect of chemicals [13–19]. Usually the tested chemical is assayed in the presenceof the native hormone, and its effect on the transcriptional activation of a reporterprotein is measured. According to Bovee et al. [18], it is advisable to measureantagonism using concentrations of the native hormone giving both half and nearmaximal response. In this way, both strong and weak antagonistic effects can bediscerned. In addition, the possible additive effect of agonists can be detected atthe half-maximal concentration of the native hormone. These methods are a goodchoice if more information on the agonistic–antagonistic mechanism of toxicity ofthe chemical is needed.Differences in metabolism can also cause varying results between mammalian and

yeast cell assays. For example, metabolic conversion is sometimes needed to inacti-vate or activate the assayed chemical [20–22]. However, it is possible to incorporatea yeast NR assay with a mammalian cell-derived metabolic activation step in orderto detect prohormonal activity of chemicals [23, 24]. In addition, hormonal effects inmammalian cells that do not directly involve receptor binding, for example, enhancedreceptor degradation [25], are naturally not detected in yeast.Another well-characterized difference between mammalian and yeast assays is

that mammalian cell assays are usually somewhat more sensitive than yeast assays[26–28]. It is possible that some compounds are pumped out of yeast cells viaefficient transporters, reducing the sensitivity of the assay [16, 29]. This can beavoided by removing a multidrug transporter from the yeast genome, thus helpingto retain the chemicals in the cell [30]. In addition, differential permeability ofcompounds through yeast cell wall compared to mammalian cell membrane can limitthe sensitivity of yeast assays [31]. However, this effect seems not to be an obstaclewith most of the tested compounds in yeast assays even within a short incubation time[32]. In addition, the sensitivity of the modern yeast assays is usually sufficient forscreening purposes.In spite of the differences between mammalian and yeast cells assays, it has been

shown that the accuracy of a yeast assay in predicting agonistic and antagonisticNR-dependent activity of chemicals is nearly comparable to that of a mammalian-cell-based assay [19]. Thus, yeast assays offer a valuable supplement to the batteryof NR toxicity screening methods.


26.3 PRINCIPLE OF NUCLEAR RECEPTOR YEAST ASSAYS

Yeast NR assays can be designed to detect NR-mediated transcriptional activation ofa reporter gene, or ligand-dependent interaction of the receptor with a cofactor protein(the yeast two-hybrid method). In transcriptional activation assays (Fig. 26.1a), anintact receptor is used, and the reporter gene is controlled by a promoter with hormonereceptor responsive elements (HREs). The ligand-activated receptor translocates tothe nucleus, binds to the HRE, and activates the transcription of the reporter gene.In the yeast two-hybrid method (Fig. 26.1b), the ligand binding domain (LBD) ofthe receptor is fused either to the activation domain (AD) or to the DNA-bindingdomain (BD) of the Gal4 transcription factor. In this method, the ligand-boundreceptor first interacts with the cofactor protein. The complex then translocates intothe nucleus, where it binds to the Gal4-specific upstream activating sequence (UAS)on the promoter of the reporter gene, and activates its transcription.TheNRs and reporter protein constructs can be expressed in yeast from plasmids or

they can be integrated into the yeast genome. Integration into the genome is preferredbecause the constructs can be more stably maintained and the number of transformedreceptor or reporter genes remains uniform in all cells.In order to detect the compound binding to NRs in yeast assays, a reporter protein

is needed. The reporter protein produces a quantifiable signal that is dependent onthe compound concentration. Thus, a dose–response curve is obtained, and effectiveor inhibitory concentrations for the compound can be calculated. Many differentreporter proteins are used in yeast assays today. Some of them are more preferablefor HTS applications. An ideal reporter for HTS purposes requires minimal handling.Additional steps like centrifugation, extra incubations, and addition of substrate or

Ligand(a) (b)

HRE HRE

Reporter

NR

Signal

Ligand

UAS

ReporterNR+BD

CF+AD

Signal

FIGURE 26.1 Principle of yeast assays. (a) Nuclear receptor transactivation assay. NR,nuclear receptor; HRE, hormone receptor responsive element. (b) Yeast two-hybrid assay withnuclear receptor and Gal4 transcription factor domains. BD, binding domain; CF, cofactorprotein; AD, activation domain; UAS, upstream activating sequence.

PRINCIPLE OF NUCLEAR RECEPTOR YEAST ASSAYS 483

cell-disrupting agents are poorly suited for HTS because they increase the assay timeand risk of cross-contamination.The majority of the yeast-based NR assays developed so far use a lac-Z reporter

gene encoding the enzyme �-galactosidase. �-Galactosidase catalyzes the break-down of a chromogenic substrate (e.g., chlorophenol red-�-D-galactopyranoside oro-nitrophenyl-�-D-galactoside), and the absorbance resulting from the color changeis then measured.Despite its popularity, �-galactosidase-based assays have several drawbacks. The

assay time may take several hours due to the time-consuming enzyme secretion.This can be improved by cell lysis, which shortens the assay time and produces abetter signal [26], but also requires the addition of lysis chemicals, and many timesharvesting of cells by centrifugation. Balsiger et al. [30] refined the �-galactosidaseassay by replacing the chromogenic substrate with a commercial bioluminescentsubstrate. In this way, the assay time was reduced, and no cell lysis was needed.However, an extra incubation time with the substrate was needed. In addition, thefluorescent substrate fluorescein-di-(�-D-galactopyranoside) for �-galactosidase hasalso been used, although in combination with cell lysis [3].Luciferases are enzymes that catalyze the oxidation of their substrate and sub-

sequently produce bioluminescence. Luciferases are many times favored in HTSbecause of their high sensitivity and negligible natural background [33]. Fireflyluciferase has been used in yeast-based NR assays, but until recently these assaysalso required cell lysis [34]. Utilization of a luciferase lacking the peroxisome target-ing peptide signal was shown to produce high levels of luminescent signal in intactyeast cells [35]. Although the assays developed using this luciferase still require theaddition of substrate [14, 15, 36], it has been shown that in the case of short incubationmethods the substrate D-luciferin can be added to the cell suspension already beforeanalyte exposure, removing the postincubation substrate addition step [4].In addition to firefly luciferase, bacterial luciferase-encoding lux operon has also

been used as reporter in yeast-based NR assays [37, 38]. In this method, no sub-strate addition is needed. However, the signal is lower compared to firefly luciferasemethods, giving a six- to eightfold induction at maximum level.Green fluorescent protein (GFP) has successfully been used as reporter in yeast

NR assays. A codon-optimized GFP for yeast, the yeast-enhanced GFP (yEGFP),has been developed [39] and used in several yeast NR assays [17, 32, 34, 40, 41]. Nosubstrate addition is needed for signal detection, which is an obvious advantage ofthe yEGFP assays. However, the high natural background fluorescence can affect thesignal-to-noise ratio [17]. For this reason, long analyte exposure times are usuallyneeded to obtain sufficient signal level.Less used reporter proteins in yeast-based NR assays are the so-called growth

reporter proteins. These are usuallymetabolically essential enzymes, which have beenremoved from the yeast original genome, for example, imidazole glycerol phosphatedehydratase (His3), which synthesizes the amino acid histidine. Yeast cells devoid ofexternally added histidine are able to grow only in the presence of a ligand that canactivate the receptor. Because these methods are based on observation of cell growth,they are well suitable for HTS. However, the growth step is relatively time-consuming


compared to some other reporters. Growth reporter proteins have mainly been used inyeast-basedNRassays to screen forNRmutantswith desired ligand affinities [42–45].Toxicity and other effects of test chemicals and solvents on yeast growth and

reporter protein function need to be corrected in order to avoid false-negative and-positive results. One popular approach is to measure cell density after incubationwith chemicals [13, 17, 32, 34, 40]. Another way is to generate a control yeast strainthat expresses the reporter protein constitutively and to incubate this yeast strainwith the same chemicals as the NR yeast strain [14, 15, 36, 38]. Both approachesare useful, but they have their limitations. On the one hand, measuring cell densityonly reflects the toxicity of the sample and not the possible inhibitory effect of thechemical on the reporter protein. On the other hand, using an extra control strainrequires additional well plates. This is also the case of a cell density measurement ifit cannot be done with the same well plate as the reporter signal measurement.There is a large selection of different NR assay setups in yeast. Although fairly

similar, there are differences in the reporter proteins used, sensitivity, assay time,and established well plate format. All of these factors need to be considered whenchoosing an assay for HTS. In Section 26.4, some of the current yeast-based NRassays are presented.

26.4 CURRENT NUCLEAR RECEPTOR YEAST ASSAYS

To date, yeast assays with many different NRs have been developed. Their perfor-mance has been tested with numerous known hormonally active chemicals as wellas with different complex environmental samples. Many assays are also capable ofdetecting antihormonal activity of chemicals.

26.4.1 Estrogen Receptor Assays

The estrogen receptor (ER) is a steroid hormone receptor, which regulates genesinvolved in growth, differentiation, development, and activity of various tissues. Thenatural ligands for ER are 17�-estradiol, estrone, and estriol. In humans two subtypesof ER are known: ER� and ER�. Estrogenic compounds differ in their potency andefficacy toward ER� and ER�. Phytoestrogens like genistein bind to ER� moreefficiently than to ER� [46]. When both receptors are expressed in the same cell type,they may form heterodimers. In fact, ER� seems to act as a coregulator and inhibitorof ER� [47].Numerous compounds that exhibit estrogenic or antiestrogenic activity are known.

In addition to natural phytoestrogens, man-made chemicals like diethylstilbestrol,bisphenol A, and parabens bind to ER. Some compounds, such as tamoxifen andICI 182780, act as antiestrogens: they bind to the receptor, but they do not allow thereceptor to activate gene transcription.Many yeast-cell-based ER assays have been developed to date. Some examples

are presented in Table 26.1. Most of the assays are developed for the 96-well plateformat, but some have also been miniaturized into high-density 384- and 1536-wellplates [3, 4].

TA

BL

E26

.1E

stro

gen

Rec

epto

rY

east

Ass

ays

17�-Estradiol(nM)

Method

Reporter

Analyte

Exposure

Time

AnalysisFormat

LOD

EC50

Lysis

Postincubation

Substrate

Addition

Antagonism

Detection

References

hER

�transactivation

�-galactosidase

3–4d

96-wellplate

–0.22

No

Yes

Yes

[20,48]

hER

�transactivation

�-galactosidase

Overnight

50mLtubes/

96-wellplate

–0.225

Yes

Yes

nd[50,85]

Y2H:hER

�LBD

andTIF2

�-galactosidase

4h

96-wellplate

0.1nM

–Yes

Yes

nd[49,77]

Y2H:hER

�LBD

andSRC1

�-galactosidase

12–18h

Testtubes/

96-wellplate

–<1

Yes

Yes

Yes

[16,55]

hER

�transactivation

�-galactosidase

4h

96-wellplate

–0.145

Yes

Yes

nd[30]

hER

�transactivation

Bacterial

luciferase

6h

96-wellplate

0.045

0.24

No

No

nd[37]

hER

�transactivation

Fireflyluciferase

2.5h

96-wellplate

0.03

0.5

No

Yes

Partial

[15]

hER

�transactivation

Fireflyluciferase

2.5h

96-wellplate

0.1

0.5

No

Yes

Partial

[15]

hER

�transactivation

Fireflyluciferase

3h

384-wellplate

0.024

0.45

No

No

nd[4,15]

hER

�transactivation

Fireflyluciferase

3h

384-wellplate

0.56

5.6

No

No

nd[4,15]

hER

�transactivation

Fireflyluciferase

4.5h

1536-wellplate

0.076

0.79

No

No

nd[4,15]

hER

�transactivation

Fireflyluciferase

4.5h

1536-wellplate

0.31

0.7

No

No

nd[4,15]

hER

�transactivation

yEGFP

4h

96-wellplate

–0.4

No

No

Yes

[32,34,51]

hER

�transactivation

yEGFP

4h

96-wellplate

0.06

0.6

No

No

nd[32]

hER,humanestrogenreceptor;Y2H,yeasttwo-hybrid;LOD,limitofdetection;nd,notdetermined.

485


The first yeast-cell-based ER assays were so-called yeast estrogen screen (YES)assays utilizing �-galactosidase as a reporter of transactivation [48]. Nowadays,there are ER assays having firefly luciferase [15], yEGFP [32, 34], and even bacterialluciferase [37] as reporter proteins.Yeast assays are relatively sensitive and are able to detect sub-nanomolar con-

centrations of the natural ligand 17�-estradiol (Table 26.1). In addition, severalother chemicals have been shown to activate the ER-dependent transcription ina concentration-dependent manner. These chemicals include, for example, naturalsteroid hormones [4, 13, 15, 26, 27, 32, 34, 37, 49, 50], phytoestrogens [15, 32,49–52], diethylstilbestrol [4, 13, 26, 32, 37, 50, 52], bisphenol A [4, 13, 15, 27, 37,49, 50, 52, 53], alkyl phenolic compounds [54], plasticizers [27, 52], parabens [15,26], polychlorinated biphenyl (PCB) compounds [55, 56], as well as pesticides andpharmaceuticals [52].Some assays have also demonstrated partial or full antagonistic effect of some

chemicals. For example, tamoxifen, hydroxytamoxifen, and ICI 182780 have shownto be partial or full antagonists in several yeast assays [13, 15, 16, 19, 51], whereassome assays have only classified these chemicals as agonists [19, 51, 57]. However,Kolle et al. [19] showed that the yeast estrogen screen (YES) assay is capable ofscreening agonists and antagonists of both ER and androgen receptorwith an accuracycomparable to that of the mammalian-cell-based HeLa test. Bovee et al. [18] havealso demonstrated the capability of their ER yeast assay in screening ER antagonists.In addition to screening pure chemicals, ER yeast assays have been used tomeasure

the estrogenic activity of environmental samples. A popular application has been todetect estrogenic activity of wastewater [27, 30, 53, 58–60] and natural water [27,58, 61]. Further matrices like sediment extracts [62], moisturizer lotions [15], anddiverse biological samples [1, 55, 63] have also been successfully analyzed.

26.4.2 Androgen Receptor Assays

The androgen receptor (AR) is responsible for regulating genes acting in male repro-duction and development. The endogenous ligands of AR are testosterone and dihy-drotestosterone. Xenobiotic compounds binding to AR seem to act mainly in anantiandrogenic way. Several AR-binding antiandrogenic compounds are known, forexample, the fungicide vinclozolin as well as DDT and its metabolites. In addition,many estrogenic compounds have been shown to act as antiandrogens [13].The androgen receptor shares similar DNA-responsive elements with the proges-

terone, mineralocorticoid, and glucocorticoid receptors [17, 64]. The signaling path-ways of these receptors are thus likely to interfere with each other. In this respect,yeast assays are particularly useful, since there is no risk of crosstalk with thesereceptors.Some of the yeast-based AR assays are presented in Table 26.2. As in the case

of ER yeast assays, transcriptional activation AR yeast assays are most popular, butalso yeast two-hybrid assays have been developed. There is also a wide selection ofreporter genes. The yeast AR assay based on the yEGFP reporter developed by Boveeet al. [17] is the only yeast-based AR assay that has been internationally validated [2].

TA

BL

E26

.2A

ndro

gen

Rec

epto

rY

east

Ass

ays

DHT(nM)

Method

Reporter

Analyte

Exposure

Time

AnalysisFormat

LOD

EC50

Lysis

Postincubation

Substrate

Addition

Antagonism

Detection

References

hAR transactivation

�-galactosidase

40h

96-wellplate

–∼1

No

Yes

Yes

[11,13]

hAR transactivation

�-galactosidase

16h

96-wellplate

0.1

4Yes

Yes

Yes

[67]

Y2H:hARLBD

andGRIP1

�-galactosidase

2h

96-wellplate

–13

Yes

Yes

Yes

[65,84]

Y2H:ARLBD

andSRC1

�-galactosidase

4h

96-wellplate

∼1>10

Yes

Yes

nd[49]

hAR transactivation

�-galactosidase

Overnight

50mLtubes/

96-wellplate

–3.5

Yes

Yes

nd[50]

hAR transactivation

Bacterial

luciferase

3–4h

96-wellplate

2.5

9.7

No

No

nd[38]

hAR transactivation

Fireflyluciferase

2.5h

96-wellplate

0.5

5.5

No

Yes

Yes

[14,15]

hAR transactivation

Fireflyluciferase

3h

384-wellplate

0.46

7.1

No

No

nd[4,14]

hAR transactivation

Fireflyluciferase

4.5h

1536-wellplate

0.27

9No

No

nd[4,14]

hAR transactivation

yEGFP

24h

96-wellplate

333

No

No

Yes

[2,17]

hAR,humanandrogenreceptor;Y2H,yeasttwo-hybrid;DHT,dihydrotestosterone;LOD,limitofdetection;nd,notdetermined.

487


Several compounds have been shown to act as AR antagonists in yeast assays[13, 18, 19, 51], while fewer exhibit agonistic effects. Thus, when screening forAR-binding compounds, it is especially important that the assay of choice can detectantagonists as well as agonists.The yeast-based AR assays have been used to screen for chemicals with

(anti)androgenic properties [18, 19, 52, 65]. These include, for example, steroidhormones, plant-derived compounds, flame retardants, pesticides, and phenolic com-pounds. In addition to measuring pure chemicals, AR yeast assays have been used,for example, to measure antiandrogenic activity of steroid esters in hair samples[63], anabolic steroids in dietary supplements [66], androgenic activity in calf urineand feed [2], androgenic activity in waste water before and after treatment [14], and(anti)androgenic activity in pulp and paper mill effluents [67].

26.4.3 Progesterone Receptor Assays

Progesterone is a hormone that plays a key role inmany functions such as reproductionas well as preparation and maintenance of uterus in pregnancy (reviewed by Li et al.[68]). It regulates cell proliferation and differentiation in reproductive tissues togetherwith estrogens. Two isoforms of the human progesterone receptor (PR) are known,PR-A and PR-B, both acting in a distinct manner depending on the tissue type. PR-Acan even act as a repressor of transcription.The few chemicals tested so far have mainly been shown to be PR antagonists.

Thus, it seems that in the case of PR assays, as in the case of AR assays, it is importantto screen for antagonistic effects in addition to agonistic effects. Some of the existingPR assays are presented in Table 26.3. It has been shown that high progesteroneconcentrations might inhibit yeast growth in a receptor-dependent manner [50]. Forthis reason, the exposure times of progesterone and related compounds should bekept as short as possible. One of the few yeast-based NR assays established in high-density well plates has been developed for PR [3]; in this �-galactosidase-basedassay, a fluorescent substrate was used.The PR yeast assays have been used in many cases. Death et al. [69] demonstrated

with their PR yeast assay that the novel steroid tetrahydrogestrinone used for dopingpurposes is a PR agonist. Chatterjee et al. [41] tested various different compoundssuch as pesticides and their metabolites, nonylphenol and endosulfan, which werefound to be PR antagonists. The PR assay developed by Gaido et al. [50] was used tomeasure agonistic and antagonistic effects of waste water treatment of plant effluents[70]. In addition, Li et al. [65] tested some organochlorine pesticides and found thatsome of them are PR antagonists.

26.4.4 Aryl Hydrocarbon Receptor Assays

The aryl hydrocarbon receptor (AhR) has been linked to dioxin-like chemicalsbecause it binds and is activated by coplanar aromatic substances such as 2,3,7,8-tetrachlorodibentzo-p-dioxin (TCDD). Other well-known AhR-activating chemicalsare, for example, benzo[a]pyrene, PCBs, and polycyclic aromatic hydrocarbons

TA

BL

E26

.3F

urth

erN

ucle

arR

ecep

tor

Yea

stA

ssay

s

Method

Reporter

Analyte

Exposure

Time

AnalysisFormat

Reference

Compound

EC50

Lysis

Postincubation

Substrate

Addition

Antagonism

Detection

References

hPR transactivation

�-galactosidase

4h

24-wellplate

Progesterone

4.5nM

Yes

Yes

Yes

[69]

hPR transactivation

�-galactosidase

2h

96-wellplate

0.5nM

Yes

Yes

Yes

[65]

hPR transactivation

�-galactosidase

4h

96-wellplate

0.5nM

Yes

Yes

Yes

[50,70]

hPR transactivation

�-galactosidase

18h

96-wellplate

130nM

Yes

Yes

nd[64]

hPR-Btransactivation

yEGFP

24h

96-wellplate

1nM

No

No

Yes

[41]

hPR transactivation

�-galactosidase

3h

96-,384-,and

1536-wellplatePromegestone

<0.5– ∼50nM

Yes

Yes

nd[3]

hAhR transactivation

Fireflyluciferase

3.5h

96-wellplate

Benzo[a]pyrene0.19mM

No

Yes

nd[4,36]


Fireflyluciferase

3h

384-wellplate

0.34mM

No

No

nd[4,36]


Fireflyluciferase

4.5h

1536-wellplate

0.36mM

No

No

nd[4,36]


�-galactosidase

18h

Testtubes

40nM

Yes

Yes

nd[71]

hAhR

+hER

transactivation

�-galactosidase

4–6d

96-wellplate

180nM

No

Yes

nd[75] (c

onti

nued)

489

TA

BL

E26

.3(C

ontin

ued)

Method

Reporter

Analyte

Exposure

Time

AnalysisFormat

Reference

Compound

EC50

Lysis

Postincubation

Substrate

Addition

Antagonism

Detection

References

hRXR

�,�,and

�transactivation

�-galactosidase

16h

96-wellplate

9-ci

s-RA

50–150

nMYes

Yes

nd[86]

hRAR

�transactivation

�-galactosidase

16h

96-wellplate

100nM

Yes

Yes

nd[86]

Y2H:hRXR

�,�,

and

�andTIF2

�-galactosidase

4h

96-wellplate

>100nM

Yes

Yes

nd[82]

Y2H:hRXR

�andGRIP1

�-galactosidase

2h

96-wellplate

150nM

Yes

Yes

Yes

[84]

Y2H:hRAR

�andTIF2

�-galactosidase

4h

96-wellplate

All-

tran

s-RA

(EC×10)

5.4nM

Yes

Yes

nd[83]

Y2H:hTR

�and

TIF2

�-galactosidase

4h

96-wellplate

Triiodothyronine55nM

Yes

Yes

Yes

[49,56,87]

hMR transactivation

�-galactosidase

18h

96-wellplate

Aldosterone

1�M

Yes

Yes

nd[64]

hGR transactivation

�-galactosidase

18h

96-wellplate

Dexamethasone

10�M

Yes

Yes

nd[64]

hGR

�transactivation

yEGFP

24h

96-wellplate

120

�M

No

No

Yes

[40]

hPR,humanprogesteronereceptor;hAhR,humanarylhydrocarbonreceptor;hRXR,humanretinoidXreceptor;hRAR,humanretinoicacidreceptor;hTR,human

thyroidhormonereceptor;hGR,humanglucocorticoidreceptor;Y2H,yeasttwo-hybrid;RA,retinoicacid;LOD,limitofdetection;nd,notdetermined.

490

CURRENT NUCLEAR RECEPTOR YEAST ASSAYS 491

(PAHs). In addition, compounds such as tryptophan, indole, indole acetic acid,tryptamine, and indole-3-carbinol have been shown to activate AhR [71].The AhR controls the expression of drug metabolizing enzymes such as

cytochrome P450 1A1, which is involved in the activation/inactivation of a num-ber of xenobiotics. It is also known that the AhR signaling pathway can crosstalkwith other NR pathways (e.g., the ER pathway), thereby causing estrogenic andantiestrogenic effects [72–74].Some of the AhR yeast assays developed so far are presented in Table 26.3.

Kawanishi et al. [75] constructed an ER and AhR yeast assay to show that AhR canalso transactivate the ER signaling pathway in yeast. Compounds binding to AhR butnot to the ER could activate ER-dependent gene transcription via AhR. This assaywas used to measure water samples, and it was shown that AhR activity in watersamples also increases ER activity.The method developed by Leskinen et al. [36] has been used to measure AhR-

binding compounds in sediment samples heavily pollutedwith PCBs, polychlorinateddibenzodioxins and dibenzofurans, and diphenyl ethers. Kamata et al. [76] used themethod developed by Miller [71] to test hydroxylated and non-hydroxylated PCBsto evaluate the possible increase of toxicity of metabolized PCBs. They refined themethod by shortening the incubation time and using a chemiluminescent substratefor �-galactosidase. This method was also used by Allinson et al. [77] to determineriver water quality.

26.4.5 Retinoid Receptor Assays

Natural retinoids, such as vitamin A and its metabolites, act via two NRs, the retinoicacid receptor (RAR) and the retinoic X receptor (RXR) (reviewed by Inoue et al. [78]and Janosek et al. [73]). They control many functions such as growth and embryonicdevelopment, vision, apoptosis, bone development, the nervous and the immunesystem. There are three different subtypes (�, �, and � ) of both receptors. RAR andRXR are able to form heterodimers with each other, and RXR also with other nuclearreceptors, for example, with peroxisome proliferator-activated receptors, vitamin Dreceptors, and liver X receptors. This leads to numerous different regulatory modesof action of the receptors, depending on the cellular environment.The functioning of RARs and RXRs in yeast was demonstrated by Hall et al. [79]

and Heery et al. [80, 81]. All-trans-retinoic acid and 9-cis-retinoic acid have beenshown to bind RAR, whereas only 9-cis-retinoic acid seems to bind to RXR. Perhapsthe best-known exogenous compounds binding to RXR are organotin compoundssuch as tributyltin (TBT) and triphenyltin (TPT), which are used as antifoulingagents. They have been shown to cause, for example, imposex in marine organisms[82]. In addition, TBT and TPT seem to be more potent activators of RXRs than9-cis-retinoic acid.Some of the yeast assays utilizing RAR and RXR are presented in Table 26.3. The

assays have been used to determine the RAR and RXR agonism and antagonism ofseveral chemicals. Kamata et al. [83] screened over 500 chemicals for RAR activity,including pesticides, parabens, and alkylphenols. Li et al. [84] also tested several


chemicals, for example, phthalates, organochlorine pesticides, and phenols regardingtheir possible agonistic and antagonistic effects on human RXR. Nishikawa et al. [82]showed that TBT and TPT activate RXRs in a yeast two-hybrid assay. In addition,this yeast assay has been used to measure agonistic activity in waste water [60].

26.4.6 Further Nuclear Receptor Assays

Yeast-based assays have been developed for other NRs in addition to the ones dis-cussed above. These include, for example, the thyroid hormone, glucocorticoid, min-eralocorticoid, and ER-related receptors. Some of them are presented in Table 26.3.There is a wide selection of different yeast assays for different nuclear receptors.By including these assays in HTS assay batteries, it is possible to screen for severaldifferent NR-mediated effects of chemicals at the same time.

26.5 YEAST NUCLEAR RECEPTOR ASSAYS ANDHIGH-THROUGHPUT SCREENING

Themajority of the yeast-basedNRassays developed so far have only been establishedin the 96-well plate format. Only few of them have been miniaturized into higher-density 384- and 1536-well plates [3, 4]. Thesewell plate formats are already routinelyused in other cell-based HTS assays [88], and it is for sure that in the near futurethere will be a growing need for miniaturized yeast-based NR assays.In order to retain good data quality in HTS, the assay should have both a high

signal-to-background ratio and a low standard deviation [89]. Miniaturization canaffect the sensitivity of the yeast-based NR assay, both negatively and positively[3, 4]. It seems that a high signal-to-background ratio of an assay reflects its suitabilityto miniaturization: assays with low signal-to-background ratios seem to have higherdetection limits and lower data quality in high-density well plates [4]. It has alsobeen reported that miniaturization of an assay can increase the standard deviation[4]. For this reason, assays with high signal-to-background ratios are most suited tolow-volume 384- and 1536-well plate formats.Assay time is another important issue in HTS. The modern yeast-based NR assays

require relatively short chemical exposure times, even as short as 2.5 h [15]. Yeast cellcultivation is also fast when compared to mammalian cell cultures. For HTS, in whichhigh numbers of samples and high-density well plates are used, implementation ofautomated liquid handling becomes necessary. Yeast cells are generally well suitedfor robotic pipetting, and automation will also reduce assay time [4]. However, furthersteps, in which chemicals are added, will prolong the assay time, and for this reason,assays, in which a further substrate or reagent does not need to be added, are to befavored [4].Yeast assays are generally relatively cost-efficient because of inexpensive media

and other reagents. Cell cultivation is easy and does not require a special expertise.In miniaturized assays, the need of reagents and consumables such as well plates is

REFERENCES 493

reduced, and thus costs are even lower. Yeast-based NR assays have several propertiesthat make them well suited for HTS. They are robust, inexpensive, and generally easyand fast to perform. They have the potential to be applied to high-density wellplate formats and automated assays. Yeast assays are a good option for the initialscreening of large compound libraries to detect potential NR-mediated toxic effectsof chemicals.

REFERENCES

1. Bovee, T.F.H., Bor, G., Becue, I., Daamen, F.E.J., van Duursen, M.B.M., Lehmann, S.,Vollmer, G., De Maria, R., Fox, J.E., Witters, H., Bernhoeft, S., Schramm, K., Hoogen-boom, R.L.A.P., Nielen, M.W.F. (2009). Inter-laboratory comparison of a yeast bioassayfor the determination of estrogenic activity in biological samples. Analytica Chimica Acta,637, 265–272.

2. Bovee, T.F.H., Bor, G., Heskamp, H.H., Lasaroms, J.J.P., Sanders, M.B., Nielen, M.W.F.(2009). Validation and application of a yeast bioassay for screening androgenic activity incalf urine and feed. Analytica Chimica Acta, 637, 225–234.

3. Berg, M., Undisz, K., Thiericke, R., Moore, T., Posten, C. (2000). Miniaturization ofa functional transcription assay in yeast (human progesterone receptor) in the 384-and1536-well plate format. Journal of Biomolecular Screening, 5, 71–76.

4. Rajasarkka, J., Virta, M. (2011). Miniaturization of a panel of high throughput yeast cell-based nuclear receptor assays in 384-and 1536-wellmicroplates.Combinatorial Chemistry& High Throughput Screening, 14, 47–54.

5. Jacobs, M., Dickins, M., Lewis, D. (2003). Homology modelling of the nuclear receptors:human oestrogen receptor � (hER �), the human pregnane-X-receptor (PXR), the Ahreceptor (AhR) and the constitutive androstane receptor (CAR) ligand binding domainsfrom the human oestrogen receptor� (hER�) crystal structure, and the human peroxisomeproliferator activated receptor � (PPAR �) ligand binding domain from the human PPAR� crystal structure. Journal of Steroid Biochemistry and Molecular Biology, 84, 117–132.

6. Damstra, T., Barlow, S., Bergman, A., Kavlock, R., Van der Kraak, G. (2002). GlobalAssessment of the State-of-the-Science of Endocrine Disruptors. WHO publication No.WHO/PCS/EDC/02.2. World Health Organization: Geneva, Switzerland.

7. Sharpe, R., Irvine, D. (2004). How strong is the evidence of a link between environmentalchemicals and adverse effects on human reproductive health? British Medical Journal,328, 447–451.

8. Metzger, D., White, J., Chambon, P. (1988). The human estrogen-receptor functions inyeast. Nature, 334, 31–36.

9. Schena, M., Yamamoto, K. (1988). Mammalian glucocorticoid receptor derivativesenhance transcription in yeast. Science, 241, 965–967.

10. Mak, P., McDonnell, D., Weigel, N., Schrader, W., O’Malley, B. (1989). Expression offunctional chicken oviduct progesterone receptors in yeast (Saccharomyces cerevisiae).Journal of Biological Chemistry, 264, 21613–21618.

11. Purvis, I., Chotai, D., Dykes, C., Lubahn, D., French, F.,Wilson, E., Hobden, A. (1991). Anandrogen-inducible expression system for Saccharomyces cerevisiae. Gene, 106, 35–42.


12. Connolly, L., Ropstad, E., Verhaegen, S. (2011). In vitro bioassays for the study ofendocrine-disrupting food additives and contaminants. Trends in Analytical Chemistry,30, 227–238.

13. Sohoni, P., Sumpter, J. (1998). Several environmental oestrogens are also anti-androgens.Journal of Endocrinology, 158, 327–339.

14. Michelini, E., Leskinen, P., Virta, M., Karp, M., Roda, A. (2005). A new recombinant cell-based bioluminescent assay for sensitive androgen-like compound detection. Biosensors& Bioelectronics, 20, 2261–2267.

15. Leskinen, P., Michelini, E., Picard, D., Karp, M., Virta, M. (2005). Bioluminescent yeastassays for detecting estrogenic and androgenic activity in differentmatrices.Chemosphere,61, 259–266.

16. Liu, J., Jeannin, E., Picard, D. (1999). The anti-estrogen hydroxytamoxifen isa potent antagonist in a novel yeast system. Biological Chemistry, 380, 1341–1345.

17. Bovee, T.F.H., Helsdingen, R.J.R., Hamers, A.R.M., van Duursen, M.B.M., Nielen,M.W.F., Hoogenboom, R.L.A.P. (2007). A new highly specific and robust yeast andro-gen bioassay for the detection of agonists and antagonists. Analytical and BioanalyticalChemistry, 389, 1549–1558.

18. Bovee, T.F.H., Thevis, M., Hamers, A.R.M., Peijnenburg, A.A.C.M., Nielen, M.W.F.,Schoonen, W.G.E.J. (2010). SERMs and SARMs: detection of their activities withyeast based bioassays. Journal of Steroid Biochemistry and Molecular Biology, 118,85–92.

19. Kolle, S.N., Kamp, H.G., Huener, H.A., Knickel, J., Verlohner, A., Woitkowiak, C., Land-siedel, R., van Ravenzwaay, B. (2010). In house validation of recombinant yeast estrogenand androgen receptor agonist and antagonist screening assays. Toxicology In Vitro, 24,2030–2040.

20. Beresford, N., Routledge, E., Harris, C., Sumpter, J. (2000). Issues arising when inter-preting results from an in vitro assay for estrogenic activity. Toxicology and AppliedPharmacology, 162, 22–33.

21. Charles, G., Bartels, M., Gennings, C., Zacharewski, T., Freshour, N., Gollapudi, B.,Carney, E. (2000). Incorporation of S-9 activation into an ER-alpha transactivation assay.Reproductive Toxicology, 14, 207–216.

22. Hoogenboom, L., De Haan, L., Hooijerink, D., Bor, G., Murk, A., Brouwer, A. (2001).Estrogenic activity of estradiol and its metabolites in the ER-CALUX assay with humanT47D breast cells. APMIS, 109, 101–107.

23. Morohoshi, K., Yamamoto, H., Kamata, R., Shiraishi, F., Koda, T., Morita, M. (2005).Estrogenic activity of 37 components of commercial sunscreen lotions evaluated by invitro assays. Toxicology In Vitro, 19, 457–469.

24. Rijk, J.C.W., Bovee, T.F.H., Groot, M.J., Peijnenburg, A.A.C.M., Nielen, M.W.F. (2008).Evidence of the indirect hormonal activity of prohormones using liver S9 metabolicbioactivation and an androgen bioassay. Analytical and Bioanalytical Chemistry, 392,417–425.

25. Wijayaratne, A., McDonnell, D. (2001). The human estrogen receptor-alpha is a ubiq-uitinated protein whose stability is affected differentially by agonists, antagonists, andselective estrogen receptor modulators. Journal of Biological Chemistry, 276, 35684–35692.

REFERENCES 495

26. Schultis, T.,Metzger, J. (2004). Determination of estrogenic activity by LYES-assay (yeastestrogen screen-assay assisted by enzymatic digestion with lyticase). Chemosphere, 57,1649–1655.

27. Murk, A., Legler, J., van Lipzig, M., Meerman, J., Belfroid, A., Spenkelink, A., van derBurg, B., Rijs, G., Vethaak, D. (2002). Detection of estrogenic potency in wastewater andsurface water with three in vitro bioassays. Environmental Toxicology and Chemistry, 21,16–23.

28. Svobodova, K., Cajthaml, T. (2010). New in vitro reporter gene bioassays for screening ofhormonal active compounds in the environment.Applied Microbiology and Biotechnology,88, 839–847.

29. Dudley, M., Sheeler, C., Wang, H., Khan, S. (2000). Activation of the human estrogenreceptor by the antiestrogens ICI 182,780 and tamoxifen in yeast genetic systems: impli-cations for their mechanism of action. Proceedings of the National Academy of Sciencesof the United States of America, 97, 3696–3701.

30. Balsiger, H.A., de la Torre, R., Lee, W., Cox, M.B. (2010). A four-hour yeast bioassayfor the direct measure of estrogenic activity in wastewater without sample extraction,concentration, or sterilization. Science of the Total Environment, 408, 1422–1429.

31. Lyttle, C., Damian-Matsumura, P., Juul, H., Butt, T. (1992). Human estrogen-receptorregulation in a yeast model system and studies on receptor agonists and antagonists.Journal of Steroid Biochemistry and Molecular Biology, 42, 677–685.

32. Bovee, T., Helsdingen, R., Rietjens, I., Keijer, J., Hoogenboom, R. (2004). Rapid yeastestrogen bioassays stably expressing human estrogen receptors � and �, and green fluo-rescent protein: a comparison of different compounds with both receptor types. Journalof Steroid Biochemistry and Molecular Biology, 91, 99–109.

33. Fan, F., Wood, K.V. (2007). Bioluminescent assays for high-throughput screening. Assayand Drug Development Technologies, 5, 127–136.

34. Bovee, T., Helsdingen, R., Koks, P., Kuiper, H., Hoogenboom, R., Keijer, J. (2004). Devel-opment of a rapid yeast estrogen bioassay, based on the expression of green fluorescentprotein. Gene, 325, 187–200.

35. Leskinen, P., Virta, M., Karp, M. (2003). One-step measurement of firefly luciferaseactivity in yeast. Yeast, 20, 1109–1113.

36. Leskinen, P., Hilscherova, K., Sidlova, T., Kiviranta, H., Pessala, P., Salo, S., Verta, M.,Virta, M. (2008). Detecting AhR ligands in sediments using bioluminescent reporter yeast.Biosensors & Bioelectronics, 23, 1850–1855.

37. Sanseverino, J., Gupta, R., Layton, A., Patterson, S., Ripp, S., Saidak, L., Simpson,M., Schultz, T., Sayler, G. (2005). Use of Saccharomyces cerevisiae BLYES expressingbacterial bioluminescence for rapid, sensitive detection of estrogenic compounds. Appliedand Environmental Microbiology, 71, 4455–4460.

38. Eldridge, M.L., Sanseverino, J., Layton, A.C., Easter, J.P., Schultz, T.W., Sayler, G.S.(2007). Saccharomyces cerevisiaeBLYAS, a new bioluminescent bioreporter for detectionof androgenic compounds. Applied and Environmental Microbiology, 73, 6012–6018.

39. Cormack, B., Bertram, G., Egerton, M., Gow, N., Falkow, S., Brown, A. (1997). Yeast-enhanced green fluorescent protein (yEGFP): a reporter of gene expression in Candidaalbicans. Microbiology, 143, 303–311.

40. Bovee, T.F.H., Helsdingen, R.J.R., Hamers, A.R.M., Brouwer, B.A., Nielen, M.W.F.(2011). Recombinant cell bioassays for the detection of (gluco)corticosteroids and


endocrine-disrupting potencies of several environmental PCB contaminants. Analyticaland Bioanalytical Chemistry, 401, 873–882.

41. Chatterjee, S., Kumar, V., Majumder, C.B., Roy, P. (2008). Screening of some anti-progestin endocrine disruptors using a recombinant yeast based in vitro bioassay. Toxicol-ogy In Vitro, 22, 788–798.

42. Chen, Z., Zhao, H. (2005). Rapid creation of a novel protein function by in vitro coevolu-tion. Journal of Molecular Biology, 348, 1273–1282.

43. Miller, N., Whelan, J. (1998). Random mutagenesis of human estrogen receptor ligandbinding domain identifies mutations that decrease sensitivity to estradiol and increasesensitivity to a diphenol indene-ol compound: basis for a regulatable expression system.The Journal of Steroid Biochemistry and Molecular Biology, 64, 129–135.

44. Schwimmer, L.J., Rohatgi, P., Azizi, B., Seley, K.L., Doyle, D.F. (2004). Creation anddiscovery of ligand-receptor pairs for transcriptional control with small molecules. Pro-ceedings of the National Academy of Sciences of the United States of America, 101,14707–14712.

45. Rajasarkka, J., Hakkila, K., Virta, M. (2011). Developing a compound-specific receptorfor bisphenol A by directed evolution of human estrogen receptor �. Biotechnology andBioengineering, 108, 2526–2534.

46. Morito, K., Hirose, T., Kinjo, J., Hirakawa, T., Okawa, M., Nohara, T., Ogawa, S., Inoue,S., Muramatsu, M., Masamune, Y. (2001). Interaction of phytoestrogens with estrogenreceptors � and �. Biological & Pharmaceutical Bulletin, 24, 351–356.

47. Riggs, B., Hartmann, L. (2003). Drug therapy: selective estrogen-receptor modulators –Mechanisms of action and application to clinical practice. New England Journal ofMedicine, 348, 618–629.

48. Routledge, E., Sumpter, J. (1996). Estrogenic activity of surfactants and some of theirdegradation products assessed using a recombinant yeast screen. Environmental Toxicol-ogy and Chemistry, 15, 241–248.

49. Nishikawa, J., Saito, K., Goto, J., Dakeyama, F., Matsuo, M., Nishihara, T. (1999). Newscreening methods for chemicals with hormonal activities using interaction of nuclearhormone receptor with coactivator. Toxicology and Applied Pharmacology, 154, 76–83.

50. Gaido, K., Leonard, L., Lovell, S., Gould, J., Babai, D., Portier, C., McDonnell, D.(1997). Evaluation of chemicals with endocrine modulating activity in a yeast-basedsteroid hormone receptor gene transcription assay. Toxicology and Applied Pharmacology,143, 205–212.

51. Bovee, T.F.H., Schoonen,W.G.E.J., Hamers, A.R.M., Bento,M.J., Peijnenburg, A.A.C.M.(2008). Screening of synthetic and plant-derived compounds for (anti)estrogenic and(anti)androgenic activities. Analytical and Bioanalytical Chemistry, 390, 1111–1119.

52. Sanseverino, J., Eldridge, M.L., Layton, A.C., Easter, J.P., Yarbrough, J., Schultz, T.W.,Sayler, G.S. (2009). Screening of potentially hormonally active chemicals using biolumi-nescent yeast bioreporters. Toxicological Sciences, 107, 122–134.

53. Rutishauser, B., Pesonen, M., Escher, B., Ackermann, G., Aerni, H., Suter, M., Eggen,R. (2004). Comparative analysis of estrogenic activity in sewage treatment plant effluentsinvolving three in vitro assays and chemical analysis of steroids.Environmental Toxicologyand Chemistry, 23, 857–864.

54. Routledge, E., Sumpter, J. (1997). Structural features of alkylphenolic chemicals associ-ated with estrogenic activity. Journal of Biological Chemistry, 272, 3280–3288.

REFERENCES 497

55. Garritano, S., Pinto, B., Calderisi, M., Cirillo, T., Amodio-Cocchieri, R., Reali, D. (2006).Estrogen-like activity of seafood related to environmental chemical contaminants. Envi-ronmental Health, 5, 9.

56. Shiraishi, F., Okumura, T., Nomachi, M., Serizawa, S., Nishikawa, J., Edmonds, J., Shi-raishi, H., Morita, M. (2003). Estrogenic and thyroid hormone activity of a series ofhydroxy-polychlorinated biphenyls. Chemosphere, 52, 33–42.

57. Legler, J., Dennekamp,M., Vethaak, A., Brouwer, A., Koeman, J., van der Burg, B., Murk,A. (2002). Detection of estrogenic activity in sediment-associated compounds using invitro reporter gene assays. Science of the Total Environment, 293, 69–83.

58. Fine, T., Leskinen, P., Isobe, T., Shiraishi, H., Morita, M., Marks, R., Virta, M. (2006).Luminescent yeast cells entrapped in hydrogels for estrogenic endocrine disrupting chem-ical biodetection. Biosensors & Bioelectronics, 21, 2263–2269.

59. Salste, L., Leskinen, P., Virta, M., Kronberg, L. (2007). Determination of estrogens andestrogenic activity in wastewater effluent by chemical analysis and the bioluminescentyeast assay. Science of the Total Environment, 378, 343–351.

60. Inoue, D., Nakama, K., Sawada, K., Watanabe, T., Matsui, H., Sei, K., Nakanishi, T., Ike,M. (2011). Screening of agonistic activities against four nuclear receptors in wastewa-ter treatment plants in Japan using a yeast two-hybrid assay. Journal of EnvironmentalSciences (China), 23, 125–132.

61. Pinto, B., Garritano, S., Reali, D. (2005). Occurrence of estrogen-like substances in themarine environment of the Northern Mediterranean Sea. Marine Pollution Bulletin, 50,1681–1685.

62. Grund, S., Higley, E., Schoenenberger, R., Suter, M.J., Giesy, J.P., Braunbeck, T., Hecker,M., Hollert, H. (2011). The endocrine disrupting potential of sediments from the UpperDanube River (Germany) as revealed by in vitro bioassays and chemical analysis. Envi-ronmental Science and Pollution Research, 18, 446–460.

63. Becue, I., Bovee, T.F.H., Van Poucke, C., Groot, M.J., Nielen, M.W.F., Van Peteghem,C. (2011). Applicability of a yeast bioassay in the detection of steroid esters in hair.Analytical and Bioanalytical Chemistry, 399, 1031–1039.

64. Miller, Charles A., III, Tan, X., Wilson, M., Bhattacharyya, S., Ludwig, S. (2010). Singleplasmids expressing human steroid hormone receptors and a reporter gene for use in yeastsignaling assays. Plasmid, 63, 73–78.

65. Li, J., Li, N., Ma, M., Giesy, J.P., Wang, Z. (2008). In vitro profiling of the endocrinedisrupting potency of organochlorine pesticides. Toxicology Letters, 183, 65–71.

66. Rijk, J.C.W., Bovee, T.F.H., Wang, S., Van Poucke, C., Van Peteghem, C., Nielen, M.W.F.(2009). Detection of anabolic steroids in dietary supplements: the added value of an andro-gen yeast bioassay in parallel with a liquid chromatography-tandem mass spectrometryscreening method. Analytica Chimica Acta, 637, 305–314.

67. Chatterjee, S., Majumder, C.B., Roy, P. (2007). Development of a yeast-based assay todetermine the (anti) androgenic contaminants from pulp and paper mill effluents in India.Environmental Toxicology and Pharmacology, 24, 114–121.

68. Li, X., Lonard, D., O’Malley, B. (2004). A contemporary understanding of progesteronereceptor function. Mechanisms of Ageing and Development, 125, 669–678.

69. Death, A., McGrath, K., Kazlauskas, R., Handelsman, D. (2004). Tetrahydrogestrinone isa potent androgen and progestin. Journal of Clinical Endocrinology & Metabolism, 89,2498–2500.


70. Li, J., Chen, M., Wang, Z., Ma, M., Peng, X. (2011). Analysis of environmental endocrinedisrupting activities inwastewater treatment plant effluents using recombinant yeast assaysincorporated with exogenous metabolic activation system. Biomedical and EnvironmentalSciences, 24, 132–139.

71. Miller, C. (1997). Expression of the human aryl hydrocarbon receptor complex in yeast –Activation of transcription by indole compounds. Journal of Biological Chemistry, 272,32824–32829.

72. Zacharewski, T., Harris, M., Safe, S. (1991). Evidence for the mechanism of action ofthe 2,3,7,8-tetrachlorodibenzo-p-dioxin-mediated decrease of nuclear estrogen-receptorlevels in wild-type and mutant mouse Hepa-1c1c7 cells. Biochemical Pharmacology, 41,1931–1939.

73. Janosek, J., Hilscherova, K., Blaha, L., Holoubek, I. (2006). Environmental xenobioticsand nuclear receptors – Interactions, effects and in vitro assessment. Toxicology In Vitro,20, 18–37.

74. Safe, S. (2001). Molecular biology of the Ah receptor and its role in carcinogenesis.Toxicology Letters, 120, 1–7.

75. Kawanishi, M., Kondo, M., Shiizaki, K., Chu, W.L., Terasoma, Y., Yagi, T. (2008).Construction of a reporter yeast strain to detect estrogen receptor signaling througharyl hydrocarbon receptor activation. Environmental Science & Technology, 42, 6897–6902.

76. Kamata, R., Shiraishi, F., Nakajima, D., Takigami, H., Shiraishi, H. (2009). Mono-hydroxylated polychlorinated biphenyls are potent aryl hydrocarbon receptor ligands inrecombinant yeast cells. Toxicology In Vitro, 23, 736–743.

77. Allinson, M., Shiraishi, F., Kamata, R., Kageyama, S., Nakajima, D., Goto, S., Allinson,G. (2011). A pilot study of the water quality of the Yarra River, Victoria, Australia, usingin vitro techniques. Bulletin of Environmental Contamination and Toxicology, 87, 591–596.

78. Inoue, D., Sei, K., Ike, M. (2010). Disruption of retinoic acid receptor signaling byenvironmental pollutants. Journal of Health Science, 56, 221–230.

79. Hall, B., Smit-McBride, Z., Privalsky, M. (1993). Reconstitution of retinoid-X receptorfunction and combinatorial regulation of other nuclear hormone receptors in the yeastSaccharomyces cerevisiae. Proceedings of the National Academy of Sciences of the UnitedStates of America, 90, 6929–6933.

80. Heery, D., Zacharewski, T., Pierrat, B., Gronemeyer, H., Chambon, P., Losson, R. (1993).Efficient transactivation by retinoic acid receptors in yeast requires retinoid-X receptors.Proceedings of the National Academy of Sciences of the United States of America, 90,4281–4285.

81. Heery, D., Pierrat, B., Gronemeyer, H., Chambon, P., Losson, R. (1994). Homodimersand heterodimers of the retinoid-X receptor (RXR) activate transcription in yeast. NucleicAcids Research, 22, 726–731.

82. Nishikawa, J., Mamiya, S., Kanayama, T., Nishikawa, T., Shiraishi, F., Horiguchi, T.(2004). Involvement of the retinoid X receptor in the development of imposex caused byorganotins in gastropods. Environmental Science & Technology, 38, 6271–6276.

83. Kamata, R., Shiraishi, F., Nishikawa, J., Yonemoto, J., Shiraishi, H. (2008). Screeningand detection of the in vitro agonistic activity of xenobiotics on the retinoic acid receptor.Toxicology In Vitro, 22, 1050–1061.

REFERENCES 499

84. Li, J., Ma, M., Wang, Z. (2008). A two-hybrid yeast assay to quantify the effects ofxenobiotics on retinoid X receptor-mediated gene expression. Toxicology Letters, 176,198–206.

85. Pham, T., Hwung, Y., Santiso-Mere, D., McDonnell, D., O’Malley, B. (1992). Ligand-dependent and ligand-independent function of the transactivation regions of the humanestrogen receptor in yeast. Molecular Endocrinology, 6, 1043–1050.

86. Allegretto, E., McClurg, M., Lazarchik, S., Clemm, D., Kerner, S., Elgort, M., Boehm,M., White, S., Pike, J., Heyman, R. (1993). Transactivation properties of retinoic acid andretinoid X receptors in mammalian cells and yeast. Correlation with hormone-binding andeffects of metabolism. Journal of Biological Chemistry, 268, 26625–26633.

87. Terasaki, M., Kosaka, K., Kunikane, S., Makino, M., Shiraishi, F. (2011). Assessmentof thyroid hormone activity of halogenated bisphenol A using a yeast two-hybrid assay.Chemosphere, 84, 1527–1530.

88. Clemons, P.A. (2009). Cell-Based Assays for High-Throughput Screening – Methods andProtocols. Humana Press: New York.

89. Zhang, J., Chung, T., Oldenburg, K. (1999). A simple statistical parameter for use inevaluation and validation of high throughput screening assays. Journal of BiomolecularScreening, 4, 67–73.

27EVALUATING THE PEROXISOMALPHENOTYPE IN HIGH CONTENTTOXICITY PROFILING

Jonathan Z. Sexton and Kevin P. Williams

27.1 INTRODUCTION

Due to the innatemultivariate nature of high content screening (HCS) and high contentanalysis (HCA), cytotoxicity endpoints can be built into virtually every screen as apreliminary indicator of drug-induced cytotoxicity in both rapid single-concentrationruns and potency runs. HCS is increasingly being used for primary probe/drug discov-ery to yield therapeutically relevant phenotypic endpoints, as opposed to conventionalmolecular target-driven drug discovery, which produces information about biochem-ical events that may not translate to meaningful biologically active probes. Theinclusion of cytotoxicity indicators along with therapeutically relevant phenotypicendpoints in high content assays allows for simultaneous interrogation of intendedactivity and toxicity, so that lead compounds may be chosen judiciously.In a wide range of probe discovery efforts for numerous therapeutic areas, mul-

tivariate “cell-health” parameters should be considered in the majority of screens totake full advantage of the data-rich high content bioimaging environment. Herein wedescribe the adaptation and generalization of the assay described by Abraham et al. in2008 [1] for prioritizing lead compounds in avoidance of pronounced hepatotoxicityand for broad use in primary HCS as an indicator of subacute toxicities.A major challenge in the development and implementation of cell-based high con-

tent assays is the optimization of the “assay window” involving the maximization ofcompound effects and reduction in natural variability yielding a high signal-to-noise


501

502 PEROXISOMAL PHENOTYPE IN HIGH CONTENT TOXICITY PROFILING

(S/N) ratio environment for the robust identification of interesting biological effects.In an HCS setting, often highly variable bioassays dictate that compounds/conditionsmust be tested in triplicate to ensure adequate S/N ratio. This widely accepteddrawback to HCS is attributed to normal biological variability, and great effortsare made to minimize these effects. In this chapter, we describe exploratory dataanalysis to stratify biological variability from technical/experimental variability toenhance the understanding of high content assay limits and for development of rationalscoring systems.

27.2 EVALUATING PEROXISOME PHENOTYPEIN CYTOTOXICITY TESTING

Peroxisomes are ubiquitous subcellular organelles, a major site of �-oxidation offatty acids [2], and the only site of very-long-chain fatty acid (VLCFA) �-oxidation[3]. Assessing/controlling peroxisome phenotype and function has had value inboth the diabetes/metabolic syndrome therapeutic areas and in evaluating cellularstress and oxidative burden. Peroxisomes are also intricately coupled to the cellularstress response to xenobiotics, especially those affecting oxidative burden and lipo-toxicity [4].Therapeutically targeting peroxisome biogenesis follows from the hypothesis that

increasing the capacity for fatty acid oxidation could be beneficial for many aspects ofmetabolic syndrome by decreasing lipotoxicity, particularly in sensitive tissues likethe liver. In addition, consumers of Western high-fat diets may benefit from increasedfatty acid oxidation capacity as elevated plasma free fatty acids can contribute toinsulin resistance [5]. Indeed, it has long been known that many structurally unrelatedcompounds that increase peroxisome biogenesis in rodents are also successful intreating symptoms of dyslipidemia and metabolic syndrome in rodents [6].

27.2.1 Other Beneficial Peroxisomal Functions and Human Disease

The vital functions of peroxisomes were defined with the molecular identificationof mutations in a peroxisomal biogenesis disorder (PBD) identified in 1992 [7].Four fatal clinical disease syndromes are PBDs—Zellweger syndrome (ZS), neona-tal adrenoleukodystrophy (NALD), infantile Refsum disease (IRD), and rhizomelicchondrodysplasia punctata (RCDP). In addition to oxidizing fatty acids, peroxisomeshelp synthesize vital plasmalogens [8], special ether phospholipids that include sph-ingolipids, a major component of both myelin and lipid rafts. Peroxisomes are alsoa major site of reactive oxygen species (ROS) inactivation and thus provide ben-eficial protection against free radicals. Peroxisomal catalase decomposes hydrogenperoxide, while other peroxisomal enzymes metabolize harmful ROS in the cell [9].Further, another function of the peroxisome is the retroconversion of VLCFAs todocosahexaenoic acid (DHA) (found in fish oil), which may mediate numerous ben-eficial effects of very long-chain n-3 (�-3) fatty acids for treating cardiovasculardisease including reduced mortality and myocardial infarction [10]. Indeed, DHA

EVALUATING PEROXISOME PHENOTYPE IN CYTOTOXICITY TESTING 503

synthesis is impaired in ZS fibroblasts lacking peroxisomes [11]. Forward geneticscreens in yeast have identified a series of mutations in conserved “PEX” (PEroXin)genes required for peroxisome biogenesis. The importance of these genes has beenconfirmed by human mutations found in 13 different conserved PEX genes that causeZS, NALD, IRD, or RCDP [12]. In total, there are 20 different conserved PEX genesbetween yeast and humans.

27.2.2 eGFP-RSKL Reporter and Peroxisome Plasticity in HepG2 Cells

We have previously described the development of a stable cell line for HCAof peroxisomal phenotype in HepG2 cells using a genetically encoded enhancedperoxisomal-targeting fluorescent reporter (HepG2-EPTFR), which exactly colocal-izes with endogenous peroxisomes using immunohistochemical analysis of perox-isomes via anti-PMP-70 staining [13]. The EPTFR can be assayed in live cells bysimple fluorescent methods in real time. The reporter is based on a well-characterizedcarboxy terminal three amino acid peptide peroxisome targeting sequence 1 (PTS1)motif sufficient to target polypeptides to the peroxisomal membrane. Generally, anSKL (serine-lysine-leucine) peptide at the extreme carboxy terminus is sufficient.However, we have found that adding a particular fourth amino acid (RSKL) increasestargeting efficiency in human cells.Both HepG2 cells expressing or not expressing our peroxisomal reporter show

the same morphologies and numbers of peroxisomes. HepG2 cells have previouslydemonstrated plasticity for peroxisome biogenesis [14] and liver cells represent idealtargets for peroxisome regulation both for therapeutic potential and as a metric forcytotoxicity, as liver cells in vivo show the most peroxisome plasticity. We haveanalyzed both the stable and wildtype cell lines with flow cytometry and have con-cluded that the stable (G418 selected) cell line is monophasic, has remarkable GFPintensity allowing for short microscopic exposure times, and is ideal for screeningpurposes. In addition, we created a low-passage EPTFR-HepG2 master cell bankin order to reduce the biological variability from day-to-day continuous culture andplate cells directly from freezer stocks in 384-well plates while maintaining very high(>95%) viability.

27.2.3 Alternative Cell Models

Whilst our current assay utilizes HepG2 cells to assess compound effects on peroxi-somal biogenesis, ultimately leading to changes in cellular lipid phenotype, this cellline is not optimal for broad assessment of compound effects on normal hepatocytefunction. Primary human hepatocytes have long been used for detailed metabolicstudies due to their mature hepatic phenotype, but have limited availability and dif-ficulty in maintaining consistency when cultured in vitro due to metabolic changeswhen cultured and also from donor-to-donor variability. The utilization of cryopre-served hepatocellular carcinoma cells (HCCs) have advanced high-throughput toxi-city testing, but show substantial differences in genotype and metabolic phenotypeas compared with primary hepatocytes [15]. As an alternative to primary human


hepatocyte cells, HepaRG cells are a terminally differentiated hepatoma cell linedeveloped [16] to provide a more realistic hepatocyte model that can be expandedprior to differentiation. HepaRG cells retain many of the metabolic characteristics ofhuman hepatocytes and stably express key metabolic proteins for several weeks inculture. HepaRG cells in contrast to HepG2 show high expression of cytochromeP450 enzymes and possess liver-specific activities comparable to HCC [17].HepaRG cells appear to be a convenient and reproducible system for assessingcompound induction of P450 [17] and other metabolism and safety applications [18]and offer a promising alternative to using HCCs. As yet, there are no reports onperoxisomal biogenesis in HepaRG cells. The use of HepaRG cells in our currentassay format will allow us to further broaden our small molecule profiling and leadprioritization through more effective analysis of compound hepatotoxicity. Due tothe terminally differentiated/nonproliferating nature of the HepaRG cell platform,reporter systems should be delivered through viral transduction, where efficienciescan be above 90%, to enable direct high content imaging 24–48 h after transductionwithout the need for generation of stable cell lines.

27.3 ASSAY ASSEMBLY AND CONSIDERATIONS AFTER APRODUCTION SCREEN: HepG2 PEROXISOME BIOGENESIS ASSAY

27.3.1 Cell Plating

EPTFR-HepG2 cells were plated directly from a frozen stock at 3200 cells/well asdescribed by Sexton et al. [13]. In brief, 20 �l of prewarmed media were depositedin each well of a poly-D-lysine-coated 384-well plate (Corning #3712), followedby 20 �l of cell suspension with a Multidrop-384 (Thermo Scientific). Cells wereallowed to equilibrate at room temperature for 30 min prior to being placed in a37◦C incubator overnight to assure cell attachment prior to treatment as recom-mended by Lundholt et al. [19]. We discovered, through trial and error and fromcolloquial high content knowledge, that predispensing 20 �L of prewarmed mediahelped achieve a homogeneous cell density on the image-able surface in the centerof the well. Our hypothesis is that by predepositing media in the plate, the ini-tial capillarity draws the cell suspension up the walls of the well, making a tallercolumn of cell suspension around the rim of the well during the initial sedimenta-tion time, whereas predepositing media passivates the capillarity effects with plainmedia and an even column of cell suspension is sedimented on the plate bottom(Fig. 27.1). Corners of wells are subject to prominent stagnation zones, which fur-ther enhances this effect by minimizing mixing in the areas subject to the mostcapillarity draw.Cell dispensing into plates is done with a MultiDrop-384 automated liquid dis-

pensing system, which along with any other liquid handling system needs to beperiodically quality controlled to ensure that accurate volumes/cells are being trans-ferred to the plate. We have found that routine quality control, especially for toxicityassays that rely on cell counting as the assay window, helps to ensure consistency in

ASSAY ASSEMBLY AND CONSIDERATIONS AFTER A PRODUCTION SCREEN 505

FIGURE 27.1 Cell sedimentation upon addition to a 384-well plate. Left panel shows cellsdispensed in one aliquot resulting in greater cell density around the rim. Right panel showsthe homogenous sedimentation by predispensing media, followed by cells, leading to an evencolumn of cell suspension across the cross-sectional area of the well-bottom.

day-to-day assay performance and will improve overall assay statistics. One simplemethod for routine quality control is to predispense 20 uL of media containing phenolred dye into the well (as described above for achieving homogenous plating density)and monitoring the absorbance of phenol red at 550 nM in a plate reader to assessaccuracy and precision in media transfer. If the coefficient of variation (standarddeviation/mean) for the entire plate is less than 5% for phenol red absorbance, thenthe liquid handling is working adequately and plates will be seeded with cells. Forplates that require consistency in cell counting for toxicity or proliferation assays, weroutinely perform QC using this simple media volume quantitation at the beginningof each run.

27.3.2 Compound Addition

Several libraries were screened in the course of assay development, the librariesconsisting of known peroxisomal proliferators, FDA-approved drugs, and toxic com-pounds. The general compound addition strategy was changed from pipette tips topin tools mid-assay due to problems detected with spontaneous compound precipita-tion in intermediate dilution plates using pipette tips. A more detailed discussion ofthese effects will be described in a forthcoming manuscript (Sexton et al., manuscriptin preparation). In short, we found that to achieve an overall 1000-fold dilution,using pipette tips on a Biomek-NX, we chose to perform a 100-fold dilution ofcompounds into serum-free media, followed by a 10-fold dilution to 1× screen-ing concentration (1–10 �M typically). During assay validation and initial roundsof dose–response experiments, we noticed a random pattern of zero-effect wells ina dilution series. This was quickly alleviated by a direct 1000-fold dilution with


a 50 nL pin-tool array (V&P Sciences). Our hypothesis is that dilution of smalldrug-like molecules from DMSO into a complex milieu can cause nucleation of pre-cipitates in a nonconcentration-dependent manner, contrary to colloquial knowledgethat precipitation strongly depends on concentration. Rather, precipitation was notcorrelated with concentration, but random, suggesting that the random presence ofparticulates in wells can spontaneously precipitate compounds, which ultimately willnot be transferred into the assay/cell plate. After compound addition, plates wereincubated for 72 h before imaging.

27.4 IMAGE PROCESSING STRATEGY

During assay development, a 0.75NA Olympus UAPO 20×/340 microscope objec-tive was selected to provide adequate peroxisomal resolution for spot detection whilemaintaining satisfactory cell counts within a single field. Camera resolution wasapproximately equivalent to digital resolution by 2-fold binning, resulting in imagedimensions of 672 by 512 for a single field. A conventional cell detection/spotcounting image processing strategy was employed to generate accurate nuclear, cyto-plasmic, and peroxisomal masks for feature extraction.

27.4.1 Hoechst-33342 for Nuclear and Cytoplasmic Delineation

The open source CellProfiler software (Anne Carpenter, Broad Institute) was usedfor image processing in this study and is a flexible and extensible image processingplatform [20]. The general strategy for tabulating cytoplasmic spots is to first seg-ment the nuclei as primary objects, then dilate out to the edge of the cytoplasmicfluorescence intensity to generate a whole-cell mask, and subtract the two masks togenerate the cytoplasmic mask.This is accomplished using a single Hoechst-33342 image upon conventional fix-

ing and staining (with a relatively high—10 �g/mL—Hoechst concentration), whichbrightly stains the nucleus and is accompanied by a homogeneous low-level fluores-cent background in the cytoplasm. While this cytoplasmic nucleic acid backgroundstain is the bane of flow cytometrists, HCA can take advantage of the spatial resolu-tion and use the background cytoplasmic Hoechst intensity to delineate the cytoplasmas shown in Figure 27.2.Figure 27.2a shows the bright nuclear regions in the Hoechst image, where the

foreground intensity is exceedingly bright (over 20-to-1 S/N) as to obscure the cyto-plasmic background (2-to-1 S/N) when normal brightness/contrast autoscaling isapplied. When brightness/contrast display parameters are adjusted to enhance thelower end of the intensity histogram, one can clearly observe the cytoplasmic area inthe majority of cell lines. This should be evaluated as an alternative to multiplexinga cytoplasmic stain (Fig. 27.2b) and offers several advantages: using a single dyefor nuclear and cellular segmentation reduces the number of colors multiplexed inthe assay, thereby opening up the spectral space for additional dyes/endpoints or cansave time by reducing the number of images acquired in the well.

IMAGE PROCESSING STRATEGY 507

(a) (b)

(c) (d)

FIGURE 27.2 Segmentation of nuclei, cells, and peroxisomes using a single dye. Two-passthresholding of the Hoechst-33342 channel allows for accurate segmentation of both nucleiand cytoplasm in the same image. (a) Hoechst image with high threshold correction factor forsegmenting the nuclei, (b) the same image as in (a), but with a low threshold correction factordelineates the cytoplasm, (c) result of nuclear and cytoplasmic segmentation showing lightgray peroxisomes inside, and (d) results of the peroxisomal spot detection algorithm.

Dual segmentation is achieved by adjusting the threshold correction factor, a staticoffset from the optimally determined threshold using a chosen method (typicallyOtsu global with three intensity classes). To enhance the brighter nuclei, the thresholdcorrection factor is pushed greater than 1.0 for tight thresholding around the nuclearperimeter and is then lowered to below 1.0 for the subsequent identification ofthe secondary objects (whole cell) in the same Hoechst image. This is shown inFigure 27.2c with the green-peroxisome channel displayed with peroxisomes fullycontained by the cytoplasmic mask.Since the S/N in the cytoplasmic region is lower than inside the nucleus, greater

care must be taken in background correction/image flattening as any backgroundcurvature results in significantly smaller identified cytoplasmic regions in the imageextremities. This is easily achieved on a per-image basis using CellProfiler’s illumi-nation correction modules with a final 90-pixel Gaussian-blur to ensure a smoothillumination function for subtraction from the original Hoechst image prior tosegmentation.Many high content imaging systems have the ability to take multiple seamless

images per well (montaging). This is principally done when more cell observationsare requiredwhilemaintaining high image resolution, orwhenmorphological featuresare expansive (as in the case of neurite/tubule detection). This poses a problem forbackground illumination correction because the illumination function is repeated in


every subsequent frame. Conventional background subtraction algorithms fail whenflattening these “montaged” images due to the discontinuities at the field boundaries.When trying to fit a smooth function to a repeating illumination function, the qualityof fit is typically poor because of the abrupt changes at the field edges. To overcomethis issue for segmenting in low S/N images and to achieve greater precision infeature extraction, the individual images are cropped and separated (de-montaged),illumination correction is performed, and the images are re-montaged as part of theautomated image processing pipeline.

27.4.2 Spot Detection

Peroxisomal detection is performed inside the cytoplasmic mask as shown in Fig-ure 27.2d. The RSKL-tagged GFP is targeted to the peroxisomal membrane (theenhanced PTS1 motif [13]) and was found to exactly colocalize with endogenousperoxisomes using immunocounterstaining to visualize the peroxisomal membraneprotein-70 (PMP-70). We occasionally observed diffuse background GFP intensityin the cytoplasm, which may be due to recently synthesized GFP and which hasyet to colocalize, or the peroxisomal membrane is relatively saturated with GFP andequilibrium shifts some of the GFP to an unbound state. Diffuse background caninterfere with accurate detection of peroxisomes and should be removed througha spot enhancement technique. In this study, we chose to use a white-tophat filterprovided by CellProfiler in the “EnhanceOrSupressFeatures” module using a pixelradius slightly larger than the largest observed peroxisomes. This has the effect ofsuppressing the diffuse background, making for a more robust algorithm but caninject bias into the spot measurement if the biological effects favor making spots thatare much larger than the chosen spot radius. Spots much larger than the chosen spotradius will be flattened along with the background intensity and may not be counted.One of the many benefits of the high content platform is that the images may be ana-lyzed in multiple passes to yield different endpoints/information. For conditions thatappear to enlarge peroxisomes from the primary peroxisomal detection algorithm,we use a secondary algorithm that is more computationally intensive and does notuse any image enhancement.

27.5 ASSAY VALIDATION

A suitable positive control was identified in the literature [21] and confirmed by doseresponse in the EPTFR-HepG2 reporter cell line and by Western blotting demon-strating increased levels of a standard peroxisomal membrane marker (PMP-70). AMin-Mid-Max (DMSO/Carrier, EC50, EC90) experiment was conducted to analyzethe S/N and assay robustness. We performed inter- and intraplate Min-Mid-Maxexperiments to identify potential problems with plate-to-plate variability and plate-image variability (Fig. 27.3).4-Phenylbutyrate (PBA) was selected as the positive control because of the recent

reports that this FDA-approved drug, a PPAR-independent peroxisome proliferator

HIGH CONTENT INVESTIGATION OF PROTEIN DYNAMICS 509

0–20–10

0102030405060708090

100110

50 100 150 200Well

Mul

tivar

iate

res

pons

e (A

rb)

250 300 350

FIGURE 27.3 Assay validation run showing DMSO (carrier) negative controls (+ ) andthe chosen maximum concentration for 4-phenylbutyrate (PBA) as the assay positive control.Reproduced with permission from Sexton et al. [13].

and chemical chaperone, treats symptoms of the metabolic syndrome and type 2diabetes in ob/ob (obese diabetic), db/db (diabetic) [22, 23], and B6/HFD (high-fatdiet) mice [24]; this includes restoration of glucose homeostasis, enhanced insulinsensitivity, and decreased fatty liver disease. PBA is a nonclassical peroxisome pro-liferator promoting peroxisome biogenesis in human cells in vitro and rodent liverin vivo [21] and serves as an adequate positive control compound as a peroxisomeproliferator in our human liver cell line in our high content assay. PBA has beenshown to be effective experimentally in cells and mice, when administered at 9 mMand 1 g/kg, an extremely high dosage compared to rationally designed drugs. Weobserved some level of cell death after 72-h treatment with PBA, but with the desiredincrease in the number of peroxisomes per cell, the compound ultimately led to arobust positive control.

27.6 IMPLEMENTATION OF NOVEL PHOTOSWITCHABLE/PHOTOBLEACHABLE FLUORESCENT REPORTERS FOR HIGHCONTENT INVESTIGATION OF PROTEIN DYNAMICS

With the explosion of available fluorescent reporter proteins, a wide choice of colors isavailable for multiplexing in high content imaging. Another area is the developmentof new functionality in fluorescent reporters such as the recent photoswitchable,photobleachable, pH-sensitive fluorescent proteins. Despite their wide availability,the implementation of these novel tools has yet to make a big impact on the highcontent imaging field.As with many high content screens that seek to quantify protein concentra-

tion/localization within the cell, there is often a highly variable background that is


difficult to dissect from effects induced by perturbing agents. Peroxisomes are highlyvariable per cell and correlate with cell size and cell cycle; daughter cells asymmet-rically inherit seed peroxisomes from the mother cell, from which additional perox-isomes bud [25]. The photoswitchable constructs (e.g., Dendra2 and PSm-Orange)allow for easy in situ labeling of preexisting peroxisomes through photoconversionprior to compound addition. We have recently developed a Dendra2-RSKL constructthat colocalizes with peroxisomes exactly as the eGFP-based construct. TheDendra2-RSKL construct labels peroxisomes green as synthesized, but can be converted to redwith illumination of 488 nm or UV (340 nM). This can provide a valuable increaseof the S/N of the assay if all pre-existing peroxisomes can be labeled red by simplyimaging with a 340 nM excitation filter and a substantial exposure time (typicallyseveral seconds using a 20× objective). After compound addition and incubation,red and green images are taken, and the two can be compared to identify peroxi-somes that were pre-existing before compound addition (red and green) and ones thatare nascent (just green). Unfortunately with Dendra2, the photoconversion processis slow and impossible to completely eradicate the green signal without impartinglethal cellular phototoxicity. With the Dendra2 reporter system, peroxisomes mustbe classified as being new or old based on the ratio of green-to-red fluorescenceintensity. In practice, it is challenging to photoconvert enough Dendra2 to enable arobust new/old classification process, but can be easily used to identify new versus.old cells, as new cells have zero red intensity and old cells have mixed green/redintensity. We are currently exploring the use of readily photobleachable fluorescentproteins, such as mOrange (Clonetech), to eliminate the pre-existing peroxisome sig-nal. This can help to differentiate between compounds that increase biogenesis versuscompounds that decrease turnover. Both an increase in biogenesis and a decrease inturnover would result in more peroxisomes. The photoswitchable/photobleachablereporter proteins can offer unique utility to improve the information content inhigh content assays to better infer mechanism of action and protein dynamics inendpoint assays.An important aspect of the detection of nascent fluorescent protein using pho-

toconversion or photobleached background subtraction is the cellular distributionof the recovered signal. When the pre-existing signal is removed by photoconver-sion or photobleaching, the subsequent recovery can be small if compared to theinitial signal. For this kind of studies, we typically choose to have the fluores-cent protein localized in the cell, such as with peroxisomes in this study, whichaids in the detection of the recovered signal. In the case that the nascent fluores-cent protein is synthesized and it is untargeted with normal homogeneous cytoso-lic distribution, the recovered signal is spread out over the entire cell area. Whenthe fluorescent protein is targeted to peroxisomes and forms puncta, the signalis concentrated in these spots and is much more easily quantifiable because thecross-section of background intensity, which the signal area spans, is small com-pared to homogeneous cytoplasmic distribution. This helps to increase the S/Nand ultimately the sensitivity of detection (data not shown). In this scenario, oneshould tabulate the response using the max-pixel intensity per cell instead of themean intensity.

COMPOUND LIBRARIES FOR ASSAY DEVELOPMENT 511

27.7 COMPOUND LIBRARIES FOR ASSAY DEVELOPMENT

During initial assay development, after a suitable positive control was implemented,the commercially available Prestwick-FDA library (Prestwick Chemical, Inc.) wasscreened for concentration range finding. The Prestwick library, an FDA-approveddrug library similar to the common LOPAC library, was screened due to the diverseand biologically active nature of the compounds. The Prestwick library consisting of1120 compounds with a significant number of diabetes drugs and toxic compounds ismostly comprised of FDA-approved cancer drugs. Screening this library first allowedto discover interesting early “hits” for immediate follow-up, gave an indication of theinitial hit rate, and allowed to investigate different discrete modalities (therapeutic,toxic, or both) for effective triage of future hits based on this “training-set” of data.After this set was screened in triplicate, image analysis was performed and well-averages were computed for all extracted cellular features.In this initial pilot screen, we observed compounds that increased peroxisome

content and appeared to be nontoxic and compounds that were clearly toxic butshowed significant increases in peroxisomal content. To follow up this interest-ing modality of toxicity-related increases in peroxisome content, we screened twoadditional small libraries of toxic compounds: 13 histone-deacetylase inhibitors (agracious donation from the CHDI Foundation) and ToxCast-1 (a gracious donationfrom the EPA), which mostly consists of pesticides [26] with extensive animal- andcell-based assay toxicity data. After screening these libraries, we selected six com-pounds with a highly similar toxicity-related phenotypic fingerprint for follow-up,as shown in Figure 27.4. These compounds were suberoylanilide hydroxamic acid(SAHA)—a histone deacetylase inhibitor, doxorubicin—an anthracycline antibioticused in cancer chemotherapy, oxyfluorfen—a diphenyl-ether herbicide, fentin—anorganotin fungicide, cis-captafol—a dicarboximide fungicide, and chlorothalonil—a

FIGURE 27.4 Toxic compounds selected for significant alterations of the peroxisomalphenotype.


chlorinated phthalonitrile also used as a broad-spectrum fungicide (see Section 27.8for further analysis of the EPA-ToxCast dataset).

27.8 EXPLORATORY MULTIVARIATE DATA ANALYSIS

After initial pilot screening is performed and image analysis extracts roughly 50–100 primary features per cell, shifting through cellular attributes and prioritizingthem can be challenging. We have found that preliminary multivariate data analysiscan provide keen insight into different response modalities and can lead to effectivescoring systems. Specifically, principal component analysis (PCA) can be performedeasily with little or no programming experience with tools like JMP (SAS), theR-statistical software package, or SIMCA (Umetrics Software) to discover the pri-mary cellular attributes that have the largest variance in single-concentration screen-ing of known highly biologically active compounds. Importantly, initial PCA canalso indicate which parameters are the least significant and thus ignored to focus thescoring system on important cellular features and can give valuable feedback on theoptimization and validation feature extraction.An interesting result of PCA in this peroxisome assay was the discovery of a

correlation between reduced cell count perwell (a robust indicator of cytotoxicity) andincreased peroxisome area/larger peroxisomes. We chose to follow up this interestingeffect between cell viability, nuclear size/DNA content, integrated peroxisomal area,and peroxisome count as shown in Figures 27.5 and 27.6 by screening 320 compounds

0

0.25

0.5

0.75

1

1.25

1.5

1.75

2

323271

123274

0.4

0.5

0.6

0.7

0.8123274

271323

300

400

500

600

700

800

900

1000

323

123

271

274

30

40

50

60

70

80

90

100

110

Peroxisomeintensity

Peroxisomecount Viability DNA content

323271

274

FIGURE 27.5 Histograms and boxplots for four extracted features from the EPA Toxcast-1library screen showing a single set of recurring outliers (diamonds). The boxplot shows themedian as a horizontal line inside the box, the boundaries of the box are the interquartile range,and data points shown outside thewhiskers are considered outliers. This set of toxic compounds(outlier data points as diamonds) represents a uniquemodality that causes increased peroxisome(GFP) intensity, lowered viability, lowered DNA content, and fewer detected peroxisomes.

EXPLORATORY MULTIVARIATE DATA ANALYSIS 513

Via

bilit

y

0.25

0.5

0.75

1

1.25

1.5

DN

A c

onte

ntP

erox

cou

nt

400

300100

90

80

70

60

50

40

30

0.4 0.5 0.6

GFP intensity Viability DNA content

0.7 0.8 0.25 0.5 0.75 1 1.25 1.5 300 400 500 600 700 800 900

500

600

700

800

900

FIGURE 27.6 Scatterplot matrix showing correlations between extracted features. Previ-ously identified outliers are shown as diamonds and well-averaged data points with low-to-highviability as dots.

in the ToxCast-1 library at a single concentration for extracted feature distributionanalysis. Figure 27.5 shows the distribution analysis for four extracted features:peroxisome intensity—GFP intensity per cell normalized to the positive controlPBA, viability—calculated by the ratio of cells in the compound treated well to theaverage number of cells in 32 negative control (carrier DMSO) wells, DNA content—the integrated Hoechst-33342 intensity in relative fluorescent units, and the per-cellperoxisome count.Through the analysis of these distributions side by side using the JMP-9 software,

we selected outliers in the boxplot and observed a set of four compounds that wereoutliers in all four extracted parameters, representing a unique biological modality.The effect of these compounds can clearly be classified as toxic due to the decreasedcell counts in the compound-treated wells, and by the Sub-G0 DNA content that isindicative of fragmented DNA leaking out of permeabilized cells. The peroxisomalphenotype is unique, in that the peroxisome area is increasing while the numbersof peroxisomes go dramatically down, strongly indicating that the peroxisomes are


merging/fusing. The GFP expression in this cell line is regulated by a CMV promoter,and the brightness is generally indicative of the status of protein translational capacity,rather than any peroxisomal functional output. However, the GFP signal delineatesthe organelle boundaries for accurate assessment of the peroxisomal area.Another vital aspect of preliminary data analysis toward developing a scoring

system is to investigate the correlation between extracted features. One can gain atremendous amount of biological insight into an assay by a standard cross-correlationanalysis. The most powerful way to visualize and detect correlations in extractedfeatures is through a scatterplot matrix as shown in Figure 27.6. After the featuredistribution analysiswas performed and outlierswere detected andflagged (Fig. 27.5),a scatterplot matrix shows significant correlation/anticorrelation between extractedfeatures. For example, Figure 27.6 shows a strong anticorrelation between GFPintensity and viability.

27.9 DISCUSSION AND SUMMARY

The mechanism for peroxisomal degradation called micro- and macropexophagy haslargely been described by Subramani and coworkers [27–30], in which the ultimatefate of a peroxisome is to be engulfed through a peroxisome-specific autophagy-related process and sent to the lysosome or vacuole. Autophagic processes have beenobserved in cells and tissue regions undergoing programmed cell death in responseto cellular stress. The exact relation between pexophagy and xenobiotic toxicitiesremains to be elucidated. However, our data show a strong correlation between celldeath and enlarged peroxisomes, and one potential explanation is the induction ofautophagic cell death resulting in large vacuolized peroxisome masses.The main purpose of this assay is to find compounds that increase peroxisomal

functionality through increased biogenesis. This exploratory data analysis providedcritical information on a unique false-positive modality that was clearly related totoxicity effects. Empowered with this knowledge, we were able to prioritize hits fromsingle-point screening in dose response by selecting compounds with the greatestdistance between the onset of effectiveness (EC50) and the onset of toxicity effects(where peroxisome intensity increases and numbers per cell decrease). The validityof this prioritization has been verified in the C57-DIO mouse model for diet-inducedobesity and type 2 diabetes through careful examination of liver histology in treatedanimals. We have observed a clear correlation of observed toxicity in this HepG2-peroxisome assay and severity and zonal location of hepatic steatosis, inflammatorycell infiltration, lobular inflammation, and extent of fibrosis in the C57-DIO model(data not shown), thus suggesting that the assay is a valid platform for in vitromodeling of xenobiotic liver injury.Additionally, this assay/peroxisome targeting construct is easily adaptable to Hep-

aRG and C3A cell lines and can also be used in primary human hepatocytes to get amore complete picture of xenobiotic liver toxicity potential. HepG2 cells are certainlynot a perfect model for human liver due to the low or lack of expression of certaincritical toxicity-related metabolic enzymes, e.g., cytochrome P450s [31]. However,

REFERENCES 515

we have found an excellent correlation between predicted and in vivo hepatotoxicityusing the EPTFR-HepG2 cell line.

ACKNOWLEDGMENTS

Dr. Vahri Beaumont from the CHDI Foundation is thanked for kind donation of anHDAC library used in this study, and Dr. Keith Houck from the US EnvironmentalProtection Agency is thanked for kind donation of the Toxcast-1 library and forinsightful discussions.

REFERENCES


2. Schrader, M. (2006). Shared components of mitochondrial and peroxisomal division.Biochimica et Biophysica Acta, 1763, 531–541.

3. Brown 3rd, F.R., McAdams, A.J., Cummins, J.W., Konkol, R., Singh, I., Moser,A.B., Moser, H.W. (1982). Cerebro-hepato-renal (Zellweger) syndrome and neonataladrenoleukodystrophy: similarities in phenotype and accumulation of very long chainfatty acids. Johns Hopkins Medical Journal, 151, 344–351.

4. Elsner, M., Gehrmann, W., Lenzen, S. (2011). Peroxisome-generated hydrogen peroxideas important mediator of lipotoxicity in insulin-producing cells. Diabetes, 60, 200–208.

5. Boden, G. (2006). Fatty acid-induced inflammation and insulin resistance in skeletalmuscle and liver. Current Diabetes Reports, 6, 177–181.

6. Reddy, J.K., Krishnakantha, T.P. (1975). Hepatic peroxisome proliferation: induction bytwo novel compounds structurally unrelated to clofibrate. Science, 190, 787–789.

7. Shimozawa, N., Tsukamoto, T., Suzuki, Y., Orii, T., Shirayoshi, Y., Mori, T., Fujiki,Y. (1992). A human gene responsible for Zellweger syndrome that affects peroxisomeassembly. Science, 255, 1132–1134.

8. Heymans, H.S., Schutgens, R.B., Tan, R., van den Bosch, H., Borst, P. (1983). Severeplasmalogen deficiency in tissues of infants without peroxisomes (Zellweger syndrome).Nature, 306, 69–70.

9. Schrader, M., Fahimi, H.D. (2004). Mammalian peroxisomes and reactive oxygen species.Histochemistry and Cell Biology, 122, 383–393.

10. Wang, C., Harris, W.S., Chung, M., Lichtenstein, A.H., Balk, E.M., Kupelnick, B., Jor-dan, H.S., Lau, J. (2006). n-3 Fatty acids from fish or fish-oil supplements, but notalpha-linolenic acid, benefit cardiovascular disease outcomes in primary- and secondary-prevention studies: a systematic review. American Journal of Clinical Nutrition, 84,5–17.

11. Ferdinandusse, S., Denis, S., Mooijer, P.A., Zhang, Z., Reddy, J.K., Spector, A.A., Wan-ders, R.J. (2001). Identification of the peroxisomal �-oxidation enzymes involved in thebiosynthesis of docosahexaenoic acid. Journal of Lipid Research, 42, 1987–1995.


12. Steinberg, S.J., Dodt, G., Raymond, G.V., Braverman, N.E., Moser, A.B., Moser, H.W.(2006). Peroxisome biogenesis disorders. Biochimica et Biophysica Acta, 1763, 1733–1748.

13. Sexton, J.Z., He, Q., Forsberg, L.J., Brenman, J.E. (2010). High content screening for non-classical peroxisome proliferators. International Journal of High Throughput Screening,1, 127–140.

14. Schrader, M., Krieglstein, K., Fahimi, H.D. (1998). Tubular peroxisomes in HepG2 cells:selective induction by growth factors and arachidonic acid. European Journal of CellularBiology, 75, 87–96.

15. Jennen, D.G., Magkoufopoulou, C., Ketelslegers, H.B., van Herwijnen, M.H., Kleinjans,J.C., van Delft, J.H. (2010). Comparison of HepG2 and HepaRG by whole-genome geneexpression analysis for the purpose of chemical hazard identification. Toxicological Sci-ences, 115, 66–79.

16. Gripon, P., Rumin, S., Urban, S., Le Seyec, J., Glaise, D., Cannie, I., Guyomard, C.,Lucas, J., Trepo, C., Guguen-Guillouzo, C. (2002). Infection of a human hepatoma cellline by hepatitis B virus. Proceedings of the National Academy of Sciences U.S.A., 99,15655–15660.

17. Andersson, T.B. (2010). The application of HepRG cells in evaluation of cytochrome P450induction properties of drug compounds. Methods in Molecular Biology, 640, 375–387.

18. Jackson, J.P., Edwards, M., Deibert, E., Ferguson, S.S. (2011). Cryopreserved HepaRGTM

cells: an alternative in vitro screening tool for human hepatic drug metabolism,induction of metabolism, & safety applications. Available at http://tools.invitrogen.com/content/sfs/brochures/ISSX2011CryopreservedHepaRGCellsAnAlternativeInVitroScreeningToolforHumanHepaticDrugMetabolism.pdf.

19. Lundholt, B.K., Scudder, K.M., Pagliaro, L. (2003). A simple technique for reducing edgeeffect in cell-based assays. Journal of Biomolecular Screening, 8, 566–570.

20. Carpenter, A., Jones, T., Lamprecht, M., Clarke, C., Kang, I., Friman, O., Guertin, D.A.,Chang, J.H., Lindquist, R.A., Moffat, J., Golland, P., Sabatini, D.M. (2006). CellProfiler:image analysis software for identifying and quantifying cell phenotypes.Genome Biology,7, R100.

21. Gondcaille, C., Depreter,M., Fourcade, S., Lecca,M.R., Leclercq, S.,Martin, P.G., Pineau,T., Cadepond, F., ElEtr, M., Bertrand, N., Beley, A., Duclos, S., De Craemer, D., Roels,F., Savary, S., Bugaut, M. (2005). Phenylbutyrate up-regulates the adrenoleukodystrophy-related gene as a nonclassical peroxisome proliferator. Journal of Cell Biology, 169,93–104.

22. Ozcan, U., Yilmaz, E., Ozcan, L., Furuhashi, M., Vaillancourt, E., Smith, R.O., Gorgun,C.Z., Hotamisligil, G.S. (2006). Chemical chaperones reduce ER stress and restore glucosehomeostasis in a mouse model of type 2 diabetes. Science, 313, 1137–1140.

23. Ozcan, L., Ergin, A.S., Lu, A., Chung, J., Sarkar, S., Nie, D., Myers Jr., M.G., Ozcan,U. (2009). Endoplasmic reticulum stress plays a central role in development of leptinresistance. Cell Metabolism, 9, 35–51.

24. Basseri, S., Lhotak, S., Sharma, A.M., Austin, R.C. (2009). The chemical chaperone 4-phenylbutyrate inhibits adipogenesis bymodulating the unfolded protein response. Journalof Lipid Research, 50, 2486–2501.

25. Fagarasanu, A., Mast, F.D., Knoblach, B., Rachubinski, R.A. (2010). Molecular mecha-nisms of organelle inheritance: lessons from peroxisomes in yeast.Nature Reviews Molec-ular Cell Biology, 11, 644–654.

REFERENCES 517


27. Nazarko, T.Y., Farre, J.C., Subramani, S. (2009). Peroxisome size provides insights intothe function of autophagy-related proteins.Molecular Biology of the Cell, 20, 3828–3839.

28. Farre, J.C., Subramani, S. (2004). Peroxisome turnover bymicropexophagy: an autophagy-related process. Trends in Cell Biology, 14, 515–523.

29. Sakai, Y., Koller, A., Rangell, L.K., Keller, G.A., Subramani, S. (1998). Peroxisomedegradation by microautophagy in Pichia pastoris: identification of specific steps andmorphological intermediates. Journal of Cell Biology, 141, 625–636.

30. Subramani, S. (1998). Components involved in peroxisome import, biogenesis, prolifera-tion, turnover, and movement. Physiological Reviews, 78, 171–188.

31. Guillouzo, A., Corlu, A., Aninat, C., Glaise, D., Morel, F., Guguen-Guillouzo, C. (2007).The human hepatoma HepaRG cells: a highly differentiated model for studies of livermetabolism and toxicity of xenobiotics. Chemico-Biological Interactions, 168, 66–73.

28A PANEL OF QUANTITATIVE CALUX R©REPORTER GENE ASSAYS FORRELIABLE HIGH-THROUGHPUTTOXICITY SCREENING OFCHEMICALS AND COMPLEXMIXTURES

Bart van der Burg, Sander van der Linden, Hai-yen Man,Roos Winter, Lydia Jonker, Barbara van Vugt-Lussenburg,and Abraham Brouwer

28.1 BACKGROUND

Adequate assessment of toxicity of new chemicals and pharmaceuticals, either aloneor in complex mixtures as present in our food and the environment, is essentialin protecting the environment and human health. While pharmaceuticals are scru-tinized carefully before entering the market, unfortunately still large gaps exist inour knowledge on the effects of commonly used chemicals alone, not to speak ofthe complex mixtures we are exposed to. A major reason for this is not the lack ofprotective measures, but particularly the speed and capacity to efficiently test thetoxicity of chemicals, thereby calling for new methods to be introduced. Because ofthe extremely low throughput in the risk assessment of chemicals, there is a lack ofknowledge about the properties of most of the more than 100,000 existing chemicals,while new chemicals are entering the market each year [1]. The new REACH leg-islation has been put in place to greatly speed up the procedure of risk assessment.However, to be successful the introduction of new high-throughput (HTP) screening


519

520 A PANEL OF QUANTITATIVE CALUX R© REPORTER GENE ASSAYS

methods for toxicity assessment seems essential, since the capacity and speed of theconventional animal tests used is low.It is becoming evident that not only acutely toxic chemicals are posing human and

ecotoxicological risks but even chronic exposure to low concentrations of pollutantscan have profound effects as well. In particular, pollutants mimicking endogenous sexsteroid hormones have been shown to have strong effects on aquatic vertebrates suchas fish, even at ng/L concentrations. Chemical analysis of such types of compoundsis generally inadequate, since in complex environmental, food, and feed mixturesthey typically act in concert [2]. Moreover, because of our lack of knowledge onthe toxicological properties of most chemicals [1], it is impossible to perform a riskassessment on complex chemical mixtures as present in such samples, even if theexact chemical composition would be known. The problem of identifying the risks ofenvironmental mixtures of chemicals using classic approaches is further enhanced bythe fact that chemicals are converted into many metabolites with unknown biologicalactivities and thus unknown hazards. Therefore, other more comprehensive methodsare needed to assess the risk of chemicals and chemicalmixtures. Bioassaysmeasuringthe biological response rather than the exact nature of chemicals in amixture are primecandidates, but classical bioassays use experimental animals, which have proved tobe extremely low in throughput and usually quite unreliable as an analytical toolbecause of the large variability of the results. However, recent technological advanceshave now circumvented these drawbacks to a large extent, allowing the generationof highly quantitative and reproducible bioassays at a much higher throughput. Here,we describe the generation of a comprehensive panel of assays that can be usedfor quantitative and reproducible toxicity assessment of single chemical entitiesand complex mixtures such as those present in the environment, food, and exposedindividuals. Still, themethod is kept relatively simple to allow cost-effective operationand straightforward data interpretation, with the aim of a widespread distribution.

28.2 GENERAL APPROACH

Rapid advance in our understanding of the mechanism of action of toxicants throughmolecular research has led to a number of relatively simple cellular assays thatmeasure definedmechanistic endpoints, that is, activation of toxicity pathways. Basedon the expectation that mechanisms of toxicity have a molecular basis that is relatedbetween organisms, application of mechanistic assays can not only facilitate theprediction of complex endpoints such as carcinogenesis or reproductive toxicity butalso their interspecies extrapolation. To this end, we have started to develop panels ofbioassays using a standardized approach, with the aim of generating a comprehensivetoxicity screening panel. The first step in establishing a more comprehensive panelhas been the choice of a human U2-OS cell line as its basis. This panel aimed atmeasuring endocrine active and disruptive compounds, and information on the panelwas first published in 2005 by Sonneveld et al [3]. Although relatively simple, theseassays have proven to accurately predict in vivo test results [4, 5] and are now beingvalidated as alternative tests to in vivo assays for the detection of endocrine-disrupting

PRINCIPLE OF CALUX R© REPORTER GENE ASSAYS 521

compounds (EDCs) [6, 7]. A very comparable approach has recently been identifiedas the way forward to generate rapid and efficient toxicity screens that can replacelengthy and costly animal testing to a large extent in the 21st century [8].

28.3 PRINCIPLE OF CALUX R© REPORTER GENE ASSAYS TOMEASURE MAJOR TOXICITY PATHWAYS

Exposure to toxic chemicals will lead to a change in cellular behavior including achange in gene expression, which is generally decisive in eliciting a toxic response.To be able to measure reliably and quantitatively such changes in gene transcrip-tion and subsequent mRNA translation into proteins, we developed the CALUX R©

reporter gene assays for major toxicity pathways (Fig. 28.1). Transcriptional acti-vation in a cell is mediated through transcription factors, which in turn can beactivated by toxicants and other bioactive molecules. For instance, upon binding ofthe steroid hormone 17�-estradiol or EDCs, the estrogen receptor (ER) becomesactivated and binds to recognition sequences in promoter regions of target genes, theso-called estrogen-responsive elements (EREs). To generate the first version of theso-called ERCALUX R© assay, three of these EREswere linked to aminimal promoterelement (the TATA box) and the gene of an easily measurable protein (in this caseluciferase). The obtained reporter gene was then stably introduced into human T-47Dcells, which endogenously express ER. In this way, the ligand-activated receptor willactivate luciferase transcription, and the transcribed luciferase protein will emit light

CHEMICAL(mix)

Receptor

LUCIFERASE protein

LUCIFERASE

BIOLOGICAL EFFECT

ENDOGENOUS GENE

Receptor-binding elements

LUCIFERASE mRNA

Light signalproportional

to amount of biologicalactive chemical in

sample

FIGURE 28.1 Principle of the CALUX R© reporter gene assay. See text for details.


when a substrate is added. The signal will dose dependently increase as a result ofincreasing concentrations of ligand [9].

28.4 SELECTIVE REPORTER GENE ASSAYS IN THE CALUX R© PANEL

A main part of our strategy is to use assays that are highly selective for specificmolecular pathways, as has been typically the approach in the setup of the CALUX R©

battery of U-2 OS-based reporter gene assays [3, 4]. This is attained by using minimalpromoter elements coupled to multimerized, highly selective response elements,which in turn drive a minimal promoter (i.e., one that cannot significantly influencetranscription by itself) linked to luciferase. Multimerization of the specific responseelement will boost the responsiveness of the reporter gene assay. Since the U-2 OScells show very low expression levels of endogenous receptors, by introducing thecognate receptor/reporter gene combination highly selective and responsive assaysare generated (Fig. 28.2). When profiling chemical activities by making use of abattery of tests, this strategy has the advantage that it produces a very low level offalse-positives. By doing this, we aim to avoid the use of multiple assays per singleendpoint as is the case in the large-scale US EPA ToxCast program [10], therebyreducing the amount of tests and simplifying the testing and data analysis. The clearmechanistic base of the assays avoids crosstalk and artifacts, thereby facilitatingdata interpretation and risk assessment. This may also facilitate extrapolation of testresults to other species, which is important for environmental risk assessment.

FIGURE 28.2 Approach to generate reporter genes used in CALUX R© assays with lowbackground, high selectivity, and inducibility.

METABOLISM 523

28.5 GENERATION OF A PANEL OF CALUX R© REPORTERGENE ASSAYS

Ligands for nuclear hormone receptors need to pass through cellular membranes andtherefore generally are small lipophilic molecules. Nuclear hormone receptor ligandscomprise natural hormones, but also many pharmaceuticals and toxic chemicalsincluding EDCs. Since the mechanistic base of steroid hormone action is well known,CALUX R© assays based on this knowledge have been developed. The first assaystargeted the sexual hormone receptors as well as the glucocorticoid receptor [3].Another early and very well-applied assay is the DR CALUX R© assay measuringinterference with the dioxin receptor pathway (reviewed by Van Vugt-Lussenburget al. in chapter 29 of this book). Later assays included assays to identify, amongothers, ligands interactingwith thyroid hormone receptors, retinoic acid receptors, andperoxisome proliferator receptor ligands [11] (see Table 28.1 for an overview). Allof these assays typically produce S-shaped dose–response curves, starting at minuteligand concentrations (micromolar to picomolar concentrations). This property canbe used for accurate quantitative and sensitive measurements.An additional approach taken to develop reporter gene assays for toxicity

screening is to use central intracellular pathways that are involved in regulatingtranscription in a mammalian cell [12]. Moreover, a whole range of assays to coversuch pathways have been developed including reporter gene assays for AP1 (involvedin carcinogenic response), NF�B (involved in inflammatory response), p53 (involvedin DNA damage response), p21 (involved in cell cycle arrest), HIF1� (involved inhypoxia/angiogenesis), and XBP1 (involved in the endoplasmatic reticulum stressresponse). Clearly, most of these pathways are involved in quite generic responsesto toxicants, including those leading to cytotoxicity, apoptosis, and genotoxicity, butoften these generic effects are also important in more specific processes. Table 28.1summarizes currently available CALUX R© assays and Figure 28.3 their typicalresponses. As can be seen, these latter CALUX R© assays measuring more generalstressors activate the pathways involved at relatively high concentrations. Because ofthis, generally dose–response curves are shorter and aborted at higher concentrationsbecause cytotoxicity or other forms of nonspecific inhibition of gene expressionis being induced. Therefore, a control cell line named cytotox CALUX R©, whichconstitutively expresses the same luciferase gene under the control of the sameexpression plasmid that is used to generate reporter gene assays for steroid receptors,has been generated (34).

28.6 METABOLISM

Metabolism and pharmacokinetics of compounds are complex and differ betweenspecies, thereby contributing to a large extent to extrapolation problems whenwanting to perform a risk assessment for humans based on animal experiments[13]. Important differences between in vitro and in vivo results can occur whenmetabolism is neglected. This often is true for straightforward reporter gene assays


TABLE 28.1 Currently Available Stable CALUX R© Assays

Name Pathway Typical Ligands/Modulators

DR CALUX Dioxin receptor Dioxins and dioxin-likechemicals

PAH CALUX Dioxin receptor Carcinogenic PAHsER CALUX Estrogen receptor mix Estrogens, EDCsERalpha CALUX Estrogen receptor alpha Estrogens, EDCsERbeta CALUX Estrogen receptor beta (Phyto)estrogens, EDCsAR CALUX Androgen receptor Androgens, EDCsPR CALUX Progesterone receptor Progestins, EDCsGR CALUX Glucocorticoid receptor Glucocorticoids, EDCsTR CALUX Thyroid hormone receptor Thyroid hormones, EDCsRAR CALUX Retinoic acid receptor RetinoidsPPARgamma1

CALUXPeroxisome proliferator �1receptor

Antidiabetic thiazolidinediones,obesogens

PPARgamma2CALUX

Peroxisome proliferator �2receptor

Antidiabetic thiazolidinediones,obesogens

PPARalpha CALUX Peroxisome proliferator �receptor

Various, hypolipidemic fibrates

kappaB CALUX NF�B activation Proinflammatory cytokinesp21 CALUX p21 activation Cytotoxic/cytostatic agentsp53 CALUX p53 transcriptional activity Genotoxic agentsNrf2 CALUX Nrf2 transcriptional activity Electrophiles, oxidative stressTCF TCF transcriptional activity �-catenin/involved in

development andcarcinogenesis

AP1 CALUX AP1 transcriptional activity Carcinogens, UVHIF1alpha CALUX HIF1� transcriptional

activityHypoxia-mediated angiogenesis

ER stress CALUX XBP1 transcriptionalactivity

Endoplasmatic reticulumstressors

Cytox CALUX Constitutive transcriptionalactivity

Cytotoxic agents, nonspecificluciferase modulators

PAHs, polycyclic aromatic hydrocarbons; EDCs, endocrine-disrupting compounds.Note: Several assays are available in different cellular backgrounds (e.g., HEK293 or HEPG2 cells), whilemany additional assays are under construction.

[4], but also applies to much more complex primary in vitro systems aiming tomimic more closely tissue-specific responses. The complexity of metabolic routesmakes it extremely difficult to incorporate all these steps into individual testsystems, in particular when the test systems themselves are heterogeneous and havean endogenous, heterogeneous, poorly characterized, and uncontrolled metaboliccapacity, as is the case of many of the complex assay systems that are meant asalternatives to animal experiments. We are currently exploring various ways tointroduce metabolism in CALUX R© assays, and results with short S9 exposures havebeen promising, while new methods to express the enzymes are presently being

Fold induction

150

100 50 0 –1

4–1

2–1

0AR

CA

LUX

–8–6

–4

[Dih

ydro

test

oste

rone

] (lo

g M

)

150

100 50 0 –1

4–1

2–1

0

TR

beta

CA

LUX

–8–6

–4

[Trii

odot

hyro

nine

] (lo

g M

)

150

100 50 0 –1

4–1

2–1

0

RA

Ral

pha

CA

LUX

–8–6

–4

[All-

tran

s re

tinoi

c ac

id] (

log

M)

150

100 50 0 –1

4–1

2–1

0

ER

alph

a C

ALU

X

–8–6

–4

[Est

radi

ol] (

log

M)

150

100 50 0 –1

4–1

2–1

0PR

CA

LUX

–8–6

–4

[Org

2058

] (lo

g M

)

150

100 50 0 –1

4–1

2–1

0GR

CA

LUX

–8–6

–4

[Dex

amet

haso

ne] (

log

M)

150

100 50 0 –1

4–1

2–1

0

PPA

Rga

mm

a C

ALU

X

–8–6

–4

[Ros

iglit

azon

e] (

log

M)

150

100 50 0 –1

4–1

2–1

0

Nfk

appa

B C

ALU

X

–8–6

–4

[TN

Falp

ha] (

log μg

/L)

150

100 50 0 –1

4–1

2–1

0DR

CA

LUX

–8–6

–4

[2,3

,7,8

-TC

DD

] (lo

g M

)

150

100 50 0 –1

4–1

2–1

0

PAH

CA

LUX

–8–6

–4

[Ben

zo[a

]pyr

ene]

(lo

g M

)

150

100 50 0 –1

4–1

2–1

0

p53

CA

LUX

(U

2OS

)p5

3 C

ALU

X (

Hep

G2)

–8–6

–4

[Act

inom

ycin

D] (

log

M)

150

100 50 0 –1

4–1

2–1

0–8

–6–4

[Act

inom

ycin

D] (

log

M)

[Act

inom

ycin

D] (

log

M)

[Act

inom

ycin

D] (

log

M)

[tBhQ

] (lo

g M

)

150

100 50 0 –1

4–1

2–1

0p21

CA

LUX

–8–6

–4

150

100 50 0 –1

4–1

2–1

0

cyto

tox

CA

LUX

–8–6

–4

150

100 50 0 –1

4–1

2–1

0

Nrf

2 C

ALU

X

–8–6

–4

150

100 50 0 –1

4–1

2–1

0

TC

F C

ALU

XA

p1 C

ALU

X

–8–6

–4

[Lith

ium

chl

orid

e] (

log

M)

150

100 50 0 –1

4–1

2–1

0–8

–6–4

[TPA

] (lo

g M

)[T

unic

anm

ycin

] (lo

g μg

/L)

[Cob

alt c

hlor

ide]

(lo

g M

)[C

adm

ium

chl

orid

e] (

log

M)

150

100 50 0 –1

4–1

2–1

0

ES

RE

CA

LUX

–8–6

–4

150

100 50 0 –1

4–1

2–1

0

Hif1

alph

a C

ALU

X

–8–6

–4

150

100 50 0 –1

4–1

2–1

0

Met

al C

ALU

X

–8–6

–4

FIG

UR

E28

.3TypicalresponsesofthenovelCALUX

R©assaypanel.

525


evaluated (van Vugt-Lussenburg et al., unpublished results). However, in the contextof the initial CALUX R© screening metabolism is not yet included as a default. Itshould be noted, however, that metabolic activation is relevant in some of the assaysthat are available, such as DR CALUX R©, and in some of those under construction,such as PXR CALUX R© and Nrf2 CALUX R©. In addition to these experimentalmethods, methods to calculate exposure in vitro and in vivo are being explored tomore accurately predict the biological activity of chemicals [14, 35].

28.7 APPLICATIONS OF PANELS OF CALUX R© REPORTERGENE ASSAYS

Applications of CALUX R© reporter gene assays when used alone or in panels aremanifold and range from toxicity screening of chemicals and pharmaceuticals tosport doping, food and environmental monitoring, water quality assessment, greenchemistry, epidemiology, clinical research, and the determination of health risks aswell as benefits of functional foods. These assays can very accurately and quanti-tatively measure the combined effect exerted by all chemicals present in complexmixtures. Several of the assays, in particular the DR CALUX R© assay for dioxinsand the ER CALUX R© and AR CALUX R© assays for sex hormones and EDCs, haveproven to be powerful and reliable analytical tools for more than a decade. Here wewill concentrate on providing particular examples on the more novel applications thatrequire higher throughput and panels of assays.Our work leading to the HTP application described below originates from the early

establishment of the highly inducible and specific ER CALUX R© assay for estrogeniccompounds [9]. The reporter cell line used is a human breast cancer cell line stablytransfected with an estrogen-responsive luciferase reporter gene expressing estrogenreceptors endogenously. At that time, the specificity, high inducibility, and reliabilityof this assay led to numerous applications requiring considerable throughput of theassay, typically run in a 96-well format. Early applications concentrated particularlyon the identification of estrogenic EDCs as pure compounds [9, 15, 16] and in theenvironment. It was the method of choice to determine the occurrence of estrogenicactivity in the Dutch surface waters and sewage treatment effluents [17, 18]. Thisassay has also been successfully applied to biotic samples and for the identification ofactive fractions using toxicity identification and evaluation (TIE) procedures [19–21].Even nowadays, the ER CALUX R© assay remains a powerful assay, particularly inthe context of water quality monitoring, standing out favorably in an independentlarge-scale international evaluation by the Global Water Research Council [22].To further expand the applicability of our assay panel, we chose to work with

double-transfected cell lines (expressing a reporter gene and an expression plasmid forthe cognate receptor) rather than the single reporter gene transfectant ER CALUX R©

assay. This has the added value of being able to establish highly standardized panelsof assays, avoiding complications in the interpretation of assay results when multipleendogenous receptors are being expressed. Our first step toward this aim was the

APPLICATIONS OF PANELS OF CALUX R© REPORTER GENE ASSAYS 527

development of a panel of four human U-2 OS cell-based highly selective assays forestrogens, androgens, progestins, and glucocorticoids, respectively [4]. This panelhas proven to be even more powerful in water quality monitoring [23], enablingthe identification of pollutants and novel sources of pollutants [24]. Furthermore,assays to identify the sport doping activities of compounds and complex mixtures[25] as well as many other applications that go beyond the scope of the currentchapter have been established. Because of the specificity of these assays, they areparticularly suited to evaluate and quantify activities in complex mixtures containingpotential interfering activities, as has clearly been demonstrated when using humanserum, cord blood, and urine samples [25–27]. A recent study elegantly illustratesthe significance of the findings using CALUX R© assays and the powerful analyticalpossibilities, particularly when combined with instrumental methods. Suzuki et al.[28] demonstrated the occurrence of antiandrogenic activity in Baikal seals; by usingTIE methods they showed that 1,1-dichloro-2,2-bis(p-chlorophenyl)ethylene (p,p′-DDE) contributed with almost 60% to the antiandrogenic activity present in theBaikal seal tissues.The novel CALUX R© methods also proved to be very suitable to profile endocrine

activity of chemicals and pharmaceuticals [4–7, 25, 29–32]. Interestingly, theresponses in these assays have proven to be highly predictive for hormonal activ-ities in experimental animals (Fig. 28.4; [4, 5]) and have been used in more extensivepanels of assays predicting reproductive toxicity of chemicals [30]. This latter studyshowed that even with a limited panel of assays (n=14), prediction of the effect ofchemicals on extremely complex biological processes such as the entire reproductive

–3

–3 –2 –1

AR CALUX ERa CALUX

0 1

–2

–1

Adr

ogen

icity

in r

ate 2

1

–4

–4 –3 –2 –1

ER

(a) (b) (c)

AR

PR

0.87

0.46

0.84

<0.0001 104

103

104

<0.0001

<0.0001

0 1

–3

–1

–2

Est

roge

nici

ty in

mic

e 0

1

1–4

–4 –3 –2 –1PR CALUX

0 1

–3

–1

–2

Pro

gest

agen

ic a

ctiv

ity in

rab

bits

0

1

FIGURE 28.4 Comparison of CALUX R© assay results with data in experimental animals.


cycle can be predicted surprisingly well. This panel of assays was quite heteroge-neous, relied on assays with very low throughput, and included HTP assays only to avery limited extent (two reporter gene assays for estrogens [ERalpha CALUX R© andthe MELN cell line] and two for androgens [AR CALUX R© and the PALM cell line]).With this in mind, the collaborative FP7 project ChemScreen (www.chemscreen.eu;[33]) was started with the aim of using more extensive HTP and in silico methodsto efficiently predict reproductive toxicity. This project involves a transatlantic col-laboration with the US EPA ToxCast program. There is a great need to generatethese in vitro screening models to speed up chemical risk assessment and also topredict other types of toxicity in a HTP manner. To reach this aim, we expandedour panel of assays, and up to now more than 20 different cell lines are avail-able, totaling over 40 different assays if agonistic and antagonistic effects are takeninto account.

28.8 HIGH-THROUGHPUT (HTP) SCREENING

Mechanism-based bioassays are extensively used for the HTP screening of newdrug candidates. This has led to a great expansion of automated systems that canbe used to dilute samples and handle cells and multiwell plates. These methodsgenerally are used as prescreens to select a number of promising drug candidatesfrom large libraries of chemicals and often are relatively rough, not quantitative, andoften not validated to the extent required for regulatory acceptance. In addition, drugscreens are quite narrow in their scope and usually address endpoints that are notof general relevance and thus are not of prime importance in regulatory toxicology.The US EPA has launched the ToxCast program using HTP technologies to test largenumbers of chemicals in a short period of time, very much in line with the strategyto replace animal experiments proposed by the US National Research Council [21].For this purpose, it uses HTP screening with a wide range of assays and extensivedata mining and interpretation tools. Signatures and pathways that strongly correlatewith specific toxicity endpoints are sought. The approach taken is technically highlyadvanced and ambitious. Our aim is to implement in parallel a more simply focusedbut strongly widespread activity. To achieve this aim, we have active licensing andtraining programs, including service and support.An important step in increasing the throughput of CALUX R© screening was the

establishment of a liquid handling system. We have selected the Hamilton Starletsystem with an incubator for automated handling and incubation of 42 plates. Wedeliberately have chosen a relatively small and cost-effective medium-throughputsystem to allow widespread dissemination. This platform has been operating verysatisfactorily with results similar to those obtained by a trained technician, but withlower error rates and much higher throughput. Currently, we can run hundreds ofcompounds per week on multiple assays using 96- and more recently 384-well plateformats. Currently, the first large-scale screening activities are being initiated forprojects like ChemScreen.

QUALITY CONTROL AND VALIDATION OF HIGH-THROUGHPUT METHODS 529

28.9 QUALITY CONTROL AND VALIDATIONOF HIGH-THROUGHPUT METHODS

Obviously proper validation and quality assurance is one of the main requirementsfor any analytical method. In the case of HTP methods, this is not different, but someaspects need specific attention. One drawback of using panels of cells in a procedureis that all cell lines need to be of sufficient quality at the start of the analysis. To avoidproblems with one of the lines performing suboptimally and thereby influencing theperformance of the whole battery, we have now established and optimized the use offrozen cells to initiate tests, having a constant quality each time the assay is run (vanVugt-Lussenburg et al., unpublished results). Further controls include a priori testingof compound solubility in culture medium to avoid visual inspection of wells, whichbecomes impractical or even not feasible during screening procedures.Another issue when automating the assay panel is that preferably the whole panel

should be run with the same readout, because this avoids disrupting the automatedworkflow. Since at high concentrations compounds will become cytotoxic or mayinfluence reporter gene activity in a nonspecific way by other means, an importantcontrol cell line has recently been established. This is a particularly important pointregarding genotoxicity tests, in which high concentrations of compounds are used(34). To this end, the above-mentioned cytotox CALUX R© cell line, which constitu-tively expresses the same luciferase gene under the control of the same expressionplasmid which is used to generate reporter gene assays for steroid receptors (i.e., theplasmid pSG5; 34), has been generated as a control. This assay has been proven to bea very suitable HTP control for nonspecific repression of reporter gene activity whencompounds become cytotoxic or when transcription is repressed by other means athigher concentrations of the compounds.The advantage of HTP methods is that data processing needs to be automated to

remain efficient. This greatly reduces sources of error, but the large amount of datapose another problem when not inspected individually. For quantitative data analysis,proper quality controls and curve-fitting procedures, which can run automatically,need to be installed. The shape of the dose–response curves when using differentassays and compounds, however, can be very different (see Fig. 28.3). This point,which can be a challenge, needs to be taken into account by automated procedures.An issue that sometimes is overlooked or neglected when using high-tech

methodologies is the need to validate methods including the possibility to transferthem to other laboratories. This is particularly important when methods are to be usedin a harmonized fashion for regulatory purposes. Therefore, an essential point in ourapproach was to start with the use of assays that are well underway of being acceptedas alternatives to animal experiments and build up the test system from there on usingsimilar building blocks. This involves the use of receptor-based reporter gene assays,such as CALUX R© assays from BDS, for the detection of androgenic and estrogeniccompounds, which are being evaluated in an ECVAM/OECD-guided validationeffort [6, 7]. Moreover, CALUX R© assays have been successfully established in over50 laboratories worldwide. Responses in several of these assays have proven to be


highly predictive for hormonal activities in experimental animals (Fig. 28.4). Wehope that this can create a head start and pave the way for new mechanism-basedassays to be included.

ACKNOWLEDGMENTS

This work was carried out in part with financial support from the Commission of theEuropean Communities, the collaborative project ChemScreen (GA244236), and theEcogenomics/EcoLinc project.

REFERENCES

1. Commission of the European Communities (2001). White Paper on the Strategy for aFuture Chemicals Policy. COM (2001) 88 final.

2. McLachlan, J.A. (2001). Environmental signaling: what embryos and evolution teach usabout endocrine disrupting chemicals. Endocrine Reviews, 22, 319–341.

3. Sonneveld, E., Jansen, H.J., Riteco, J.A.C., Brouwer, A., van der Burg, B. (2005). Devel-opment of androgen- and estrogen-responsive bioassays, members of a panel of humancell line-based highly selective steroid responsive bioassays. Toxicological Sciences, 83,136–148.

4. Sonneveld, E., Riteco, J.A.C., Jansen, H.J., Pieterse, B., Brouwer, A., Schoonen,W.G., vander Burg, B. (2006). Comparison of in vitro and in vivo screening models for androgenicand estrogenic activities. Toxicological Sciences, 89, 173–187.

5. Sonneveld, E., Pieterse, B., Schoonen, W., van der Burg, B. (2011). Validation of in vitroscreeningmodels for progestagenic activities: Inter-assay comparison and correlation within vivo activity in rabbits. Toxicology In Vitro, 25, 545–554.

6. van der Burg, B., Winter, R., Man, H., Vangenechten, C., Weimer, M., Berckmans, P.,Witters, H., van der Linden, S. (2010). Optimization and prevalidation of the in vitro ARCALUX method to test androgenic and antiandrogenic activity of compounds. Reproduc-tive Toxicology, 30, 18–24.

7. van der Burg, B., Winter, R., Weimer, M., Berckmans, P., Suzuki, G., Gijsbers, L., Jonas,A., van der Linden, S., Witters, H., Aarts, J., Legler, J., Kopp-Schneider, A., Bremer,S. (2010). Optimization and prevalidation of the in vitro ER� CALUX method to testestrogenic and antiestrogenic activity of compounds. Reproductive Toxicology, 30, 73–80.

8. National Research Council (2007). Toxicity Testing in the 21st Century: A Vision and aStrategy. The National Academy of Sciences: Washington, DC.

9. Legler, J., van den Brink, C.E., Brouwer, A., Murk, A.J., van der Saag, P.T., Vethaak, A.D.,van der Burg, B. (1999). Development of a stably transfected estrogen receptor-mediatedluciferase reporter gene assay in the human T47D breast cancer cell line. ToxicologicalSciences, 48, 55–66.

10. Kavlock, R., Dix, D. (2010). Computational toxicology as implemented by the U.S. EPA:providing high throughput decision support tools for screening and assessing chemicalexposure, hazard and risk. Journal of Toxicology and Environmental Health Part B:Critical Reviews, 13, 197–217.

REFERENCES 531

11. Gijsbers, L., Man, H.Y., Kloeta, S.K., De Haan, L.H.J., Keijer, J., Rietjens, I.M., van derBurg, B., Aarts, J.M. (2011). Stable reporter cell lines for PPAR� -mediated modulationof gene expression. Analytical Biochemistry, 414, 77–83.

12. Hanahan, D., Weinberg, R.A. (2000). The hallmarks of cancer. Cell, 100, 57–70.

13. Hengstler, J.G., van der Burg, B., Steinberg, P., Oesch, F. (1999). Interspecies differencesin toxicology. Drug Metabolism Reviews, 31, 917–970.

14. Heringa, M.B., Schreurs, H.M.M., Busser, F., Van der Saag, P.T., van der Burg, B.,Hermens, J.L. (2004). Toward more useful in vitro toxicity data with measured freeconcentrations. Environmental Science & Technology, 38, 6263–6270.

15. Meerts, I.A.T.M., Letcher, R.J., Hoving, S., Marsh, G., Bergman, A., Lemmen, J.G., vander Burg, B., Brouwer, A. (2001). In vitro estrogenicity of polybrominated diphenyl ethers(PBDEs) and hydroxylated PBDEs. Environmental Health Perspectives, 109, 399–407.

16. Letcher, R.J., Lemmen, J.G., van der Burg, B., Brouwer, A., Bergman, A., Giesy, J.P., vanden Berg, M. (2002). In vitro antiestrogenic effects of aryl methyl sulfone metabolitesof polychlorinated biphenyls and 2,2-bis(4-chlorophenyl)-1,1-dichloroethene on 17�-estradiol-induced gene expression in several bioassay systems. Toxicological Sciences,69, 362–372.

17. Murk, A.J., Legler, J., van Lipzig, M.M., Meerman, J.H., Belfroid, A.C., Spenkelink,A., van der Burg, B., Rijs, G.B., Vethaak, D. (2002). Detection of estrogenic potency inwastewater and surface water with three in vitro bioassays. Environmental Toxicology andChemistry, 21, 16–23.

18. Vethaak, A.D., Lahr, J., Schrap, S.M., Belfroid, A.C., Rijs, G.B., Gerritsen, A., de Boer,J., Bulder, A.S., Grinwis, G.C., Kuiper, R.V., Legler, J., Murk, T.A., Peijnenburg, W.,Verhaar, H.J., de Voogt, P. (2005). An integrated assessment of estrogenic contaminationand biological effects in the aquatic environment of The Netherlands. Chemosphere, 59,511–524.

19. Houtman, C.J., van Oostveen, A.M., Brouwer, A., Lamoree, M.H., Legler, J. (2004).Identification of estrogenic compounds in fish bile using bioassay-directed fractionation.Environmental Science & Technology, 38, 6415–6423.

20. Houtman, C.J., Booij, P., Jover, E., Pascual del Rio, D., Swart, K., van Velzen, M., Vreuls,R., Legler, J., Brouwer, A., Lamoree, M.H. (2006). Estrogenic and dioxin-like compoundsin sediment from Zierikzee harbour identified with CALUX assay-directed fractionationcombined with one and two dimensional gas chromatography analyses. Chemosphere, 65,2244–2252.

21. Houtman, C.J., van Houten, Y.K., Leonards, P.E., Brouwer, A., Lamoree, M.H., Legler, J.(2006). Biological validation of a sample preparation method for ER-CALUX bioanalysisof estrogenic activity in sediment usingmixtures of xeno-estrogens.Environmental Science& Technology, 40, 2455–2461.

22. Leusch, F.D., de Jager, C., Levi, Y., Lim, R., Puijker, L., Sacher, F., Tremblay, L.A.,Wilson, V.S., Chapman, H.F. (2010). Comparison of five in vitro bioassays to measureestrogenic activity in environmental waters. Environmental Science & Technology, 44,3853–3860.

23. van der Linden, S., Heringa, M.B., Man, H.Y., Sonneveld, E., Puijker, L., Brouwer, A.,Van der Burg, B. (2008). Detection of multiple hormonal activities in waste water effluentsand surface water, using a panel of steroid receptor CALUX bioassays. EnvironmentalScience & Technology, 42, 5814–5820.


24. Schriks, M., van Leerdam, J.A., van der Linden, S.C., van der Burg, B., van Wezel, A.P.,de Voogt, P. (2010). High-resolution mass spectrometric identification and quantificationof glucocorticoid compounds in various wastewaters in the Netherlands. EnvironmentalScience & Technology, 44, 4766–4774.

25. Houtman, C.J., Sterk, S.S., Van de Heijning, M.P., Brouwer, A., Stephany, R.W., Van derBurg, B., Sonneveld, E. (2009). Detection of anabolic androgenic steroid abuse in dopingcontrol using mammalian reporter gene bioassays. Analytica Chimica Acta, 637, 247–258.

26. Brouwers, M.M., Besselink, H., Bretveld, R.W., Anzion, R., Scheepers, P.T., Brouwer,A., Roeleveld, N. (2011). Estrogenic and androgenic activities in total plasma measuredwith reporter-gene bioassays: relevant exposure measures for endocrine disruptors inepidemiologic studies? Environment International, 37, 557–564.

27. Pedersen, M., Halldorsson, T.I., Mathiesen, L., Mose, T., Brouwer, A., Hedegaard, M.,Loft, S., Kleinjans, J.C., Besselink, H., Knudsen, L.E. (2010). Dioxin-like exposures andeffects on estrogenic and androgenic exposures and micronuclei frequency in mother-newborn pairs. Environment International, 36, 344–351.

28. Suzuki, G., Tue, N.M., van der Linden, S., Brouwer, A., van der Burg, B., Lamoree,M., vanVelzen, M., Someya, M., Takahashi, S., Isobe, T., Tajima, Y., Yamada, T., Takigami, H.,Tanabe, S. (2011). Identification of major dioxin-like compounds and androgen receptorantagonist in acid-treated tissue extracts of high trophic-level animals. EnvironmentalScience & Technology, 45, 10203–10211.

29. Rizner, T.L., Brozic, P., Doucette, C., Turek-Etienne, T., Muller-Vieira, U., Sonneveld, E.,Van der Burg, B., Bocker, C., HusenB. (2011). Selectivity and potency of the retroproges-terone dydrogesterone in vitro. Steroids, 76, 607–615.

30. Schenk, B., Weimer, M., Bremer, S., van der Burg, B., Cortvrindt, R., Freyberger, A.,Lazzari, G., Pellizzer, C., Piersma, A., Schafer, W.R., Seiler, A., Witters, H., Schwarz,M. (2010). The ReProTect Feasibility Study, a novel comprehensive in vitro approach todetect reproductive toxicants. Reproductive Toxicology, 30, 200–218.

31. Schreurs, R.H., Sonneveld, E., Jansen, J.H., Seinen,W., van der BurgB. (2005). Interactionof polycyclic musks and UV filters with the estrogen receptor (ER), androgen receptor(AR) and progesterone receptor (PR) in reporter gene bioassays. Toxicological Sciences,83, 264–272.

32. Schreurs, R.H., Sonneveld, E., van der Saag, P.T., van der Burg, B., Seinen, W. (2005).Examination of the in vitro (anti)estrogenic, (anti)androgenic and (anti)dioxin-like activ-ities of tetralin, indane and isochroman derivatives using receptor-specific bioassays.Toxicology Letters, 156, 261–275.

33. van der Burg, B., Kroese, E.D., Piersma, A.H. (2011). Towards a pragmatic alternativetesting strategy for the detection of reproductive toxicants. Reproductive Toxicology, 31,558–561.

34. Van der Linden, S.C., von BerghA.R.M., Van Vugt-LussenburgB., JonkerL., BrouwerA.,TeunisM., KrulC.A.M., Van der BurgB. Development of a panel of high throughputreporter gene assays to detect genotoxicity and oxidative stress, submitted.

35. Piersma, A.H., Schulpen, S.H.W., Uibel, F., Van Vugt-Lussenburg, B., Bosgra, S.,Hermsen, S.A.B., Roelofs, M.J.E., Man, H., Jonker, L., Van der Linden, S., Van Duursen,M.B.M., Wolterbeek, A.P.M., Schwarz, M., Kroese, E.D., Van der Burg, B. Evaluation ofan alternative in vitro test battery for detecting reproductive toxicants, submitted.

29DR-CALUX R©: A HIGH-THROUGHPUTSCREENING ASSAY FOR THEDETECTION OF DIOXIN ANDDIOXIN-LIKE COMPOUNDS IN FOODAND FEED

Barbara van Vugt-Lussenburg, Harrie T. Besselink, Bartvan der Burg, and Abraham Brouwer

29.1 DIOXINS

29.1.1 Dioxin Contaminations

Dioxins belong to a heterogeneous group of recalcitrant polyhalogenated aromaticcompounds (PHAHs) that are widespread environmental contaminants includingpolychlorinated dibenzo-p-dioxins (PCDDs), dibenzofurans (PCDFs), biphenyls(PCBs), and diphenyl ethers (PCDEs). PCBs have been widely used in diverseindustrial applications as plasticizers, flame retardants, paint additives, adhesives,immersion oils, sealants, and waxes. PCBs accumulate in the environment mainlydue to their physicochemical properties, their biostability, and their hydrophobicnature. The PCDDs, PCDFs, and PCDEs on the other hand are mainly formed asby-products during industrial processes or combustion [1], with a large contributionfrom municipal incinerators [2]. Dioxin residues have been detected in a wide vari-ety of matrices including soil, sediment, water, fish, wildlife, human adipose tissue,serum, and milk. Therefore, although dioxin emission levels have decreased by morethan 80% since the 1980s in most countries [3], exposure to dioxin-like compounds


533

534 DR-CALUX R©

through contaminated soil or sediments and accumulation in the food chain are stilla problem today.

29.1.2 Dioxin Toxicity and the Arylhydrocarbon Receptor

Exposure to and bioaccumulation of halogenated aromatic compounds have beenobserved to produce a variety of species- and tissue-specific effects, such as tumor pro-motion, lethality, birth defects, hepatotoxicity, immunotoxicity, dermal toxicity, alter-ations in endocrine homeostasis, and induction of xenobiotic metabolizing enzymessuch as cytochrome P4501A1 (CYP1A1) [4–6].The broad spectrum of species- and tissue-specific biochemical and toxic effects

caused by PHAHs is primarily mediated by binding of these compounds to thecytosolic aryl hydrocarbon receptor (AhR) [7]: mice with low-affinity AhR allelesare less susceptible to the effects of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) [8],and AhR-null mice are resistant to the prototypical toxicities induced by TCDD andrelated ligands [9]. AhR is a ligand-dependent DNA transcription factor (Fig. 29.1).Upon ligand binding, the ligand–AhR complex translocates to the nucleus. Afterdissociation of the cytosolic components (Hsp90, XAP2, p23), the AhR dimerizeswith the AhR nuclear translocator (ARNT). This results in a conformational changeof AhR to its DNA-binding form, and the heterodimer binds to specific genomicsequences of responsive genes, the so-called dioxin-responsive elements (DREs), tomodulate gene expression.Adverse effects are thought to be caused by inappropriate or untimely AhR-

mediated regulation of gene expression. AhR binding results in induction of theexpression of several xenobiotic metabolizing enzymes, including the phase Ienzymes CYP1A1, CYP1A2, and CYP1B1. The endogenous role of phase I enzymesis to increase the water solubility of xenobiotics to facilitate urinary and biliary excre-tion, resulting in detoxification. However, in several cases these enzymes catalyzethe metabolic activation of various procarcinogens to ultimate carcinogens, and thisis thought to be the cause of AhR-mediated chemical carcinogenesis. Further targetgenes of AhR include the phase II metabolic enzymes UDP-glucuronosyl transferase,NADPH:oxidoreductases, and GST-Ya and the genes involved in cell proliferation(TGF-�, IL-1�, PAI-2), cell cycle regulation (p27, jun-B), and apoptosis (Bax) [10].

29.1.3 Toxic Equivalency Factors and European Regulation of Foodand Feed Dioxin Contamination

PHAHs are present as complex mixtures. These compounds can be halogenated onmultiple positions, with different substitution patterns and degrees of halogenation,meaning that in theory up to 209 different PCBs, 135 PCDFs, and 75 PCDDs couldbe formed [11]. All these compounds, isomers, and congeners have vastly differentchemical, physical, and toxicological properties. For example, in the case of PCDDs,it has been observed that the tetra- to hexachloro- 2,3,7,8-substituted analogs areconsiderably more active than the congeners with less than four lateral substituents[12]. TheWorldHealthOrganization (WHO) uses several criteria to label a compound

DIOXINS 535

CYTOSOL

TCDD

Ligand binding Add luciferine: light signalproportional to [TCDD]

Luciferase protein

NUCLEUSNuclear translocation

Dimerization DNA binding

DR CALUX® SITUATION

ENDOGENOUS SITUATION

AhR

AhR

AhR

AhR

ARNT

AhR

ARNT

AhR

ARNT

AhR

ARNT

AhR

ARNT

DRE DRE DRE LUCIFERASE

AhR

ARNT

AhR

ARNT

AhR

ARNT

DRE DRE DRE GENE

FIGURE 29.1 AhR pathway, endogenous situation (below the dotted line), and DR-CALUX R© situation.

as dioxin-like. It must be structurally related to PCDD/PCDFs, bind to the AhR, elicitAhR-mediated toxic responses, be persistent, and accumulate in the food chain. Atpresent, a list of 29 dioxin-like compounds has been established: 7 PCDDs, 10PCDFs, and 12 PCBs (Table 29.1 and EU Commission Directives 2002/70/EC and2005/7/EC). Since these compounds differ enormously in potency, a weighted factornamed toxic equivalency factor (TEF) has been established by the WHO to be able toestimate the total toxic activity of a sample (Table 29.1 and [13]). The TEF of TCDD,the most potent dioxin-like compound, is set to 1. The toxic potency of each dioxin ordioxin-like congener is related to the most toxic dioxin, that is, TCDD, and thereforetheir TEF values represent ratios of toxic potency relative to TCDD. In order to obtainthe total toxic potency of all dioxin and dioxin-like congeners in a certain sample, theconcentration of each congener identified in a sample is multiplied by its TEF valueand added up to result in a total toxic equivalence quotient (TEQ). In 2002, maximumlevels were set for PCDD/PCDFs, and for the sum of PCDD/PCDFs and dioxin-likePCBs in feed, food, and ingredients (EU Commission Directive 2002/32/EC).

536 DR-CALUX R©

TABLE 29.1 WHO TEFs of 29 Dioxins and Dioxin-like PCBs (Adapted from [13])

Congener TEF Value Congener TEF Value

Chlorinated Dibenzo-p-Dioxins Non-Ortho-Substituted PCBs2,3,7,8-TCDD 1 PCB 77 0.00011,2,3,7,8-PeCDD 1 PCB 81 0.00031,2,3,4,7,8-HxCDD 0.1 PCB 126 0.11,2,3,6,7,8-HxCDD 0.1 PCB 169 0.031,2,3,7,8,9-HxCDD 0.11,2,3,4,6,7,8-HpCDD 0.01 Mono-ortho-substituted PCBsOCDD 0.0003 PCB 105 0.0003

PCB 114 0.0003Chlorinated dibenzo-p-furans PCB 118 0.00032,3,7,8-TCDF 0.1 PCB 123 0.00031,2,3,7,8-PeCDF 0.03 PCB 156 0.00032,3,4,7,8-PeCDF 0.3 PCB 157 0.00031,2,3,4,7,8-HxCDF 0.1 PCB 167 0.00031,2,3,6,7,8-HxCDF 0.1 PCB 189 0.00031,2,3,7,8,9-HxCDF 0.12,3,4,6,7,8-HxCDF 0.11,2,3,4,6,7,8-HpCDF 0.011,2,3,4,7,8,9-HpCDF 0.01OCDF 0.0003

TEF, toxic equivalency factor; PCBs, polychlorinated biphenyls; T, tetra; Pe, penta; Hx, hexa; Hp, hepta;O, octa.

29.2 DETECTION OF DIOXINS

29.2.1 Traditional Detection Methods

Because the maximum allowed levels of dioxin-like compounds are in the lowerpg WHO-TEQ/g range, detection methods need to be very sensitive. Traditionally,dioxin levels are measured by capillary gas chromatography and high-resolutionmass spectrometry (HRGC-MS). Because this equipment is very sensitive to con-taminations, complete removal of the sample matrix is necessary, and the detectionis preceded by a laborious selective extraction and cleanup step of the samples [14].The procedure is sensitive, but also very time-consuming and therefore less suitablefor routine monitoring purposes, involving screening of large numbers of samples,or in crisis situations, when speed is essential. Apart from the high investment costs,the laboratory capacities are simply not sufficient in many countries of the EuropeanUnion. Therefore, faster, simpler, and cheaper screening methods are required.

29.2.2 The DR-CALUX R© Reporter Gene Assay

Since more than 90% of human exposure to dioxins and related compounds occursthrough food (mainly meat, dairy products, and fish), food and feed safety is a high

DETECTION OF DIOXINS 537

priority issue. Stringent EU limit values are in force for dioxins in foodstuffs andfeedingstuffs (EU Commission Directives 2003/57/EC, 2002/69/EC, 2002/70/ECand Council Regulation No 2375/2001). Therefore, monitoring programs for foodand feed contamination using high-throughput systems are of great importance. Oneexample of a high-throughput assay for the detection of dioxins is the DR-CALUX R©

reporter gene assay [15, 16]. This assay can be performed in 96-wells plates. Thecells are seeded on day 1, sample is added to the cells on day 2, and the plates aremeasured on day 3. At BioDetection Systems BV, one plate generally contains aTCDD calibration curve, several positive and negative controls, and four samplesin three dilutions. Prior to exposure, the samples are subjected to a shake-solventextraction, followed by acidic silica cleanup. With this method, up to 60 samplesper week can be processed by one technician. Because the cells are not as sensitiveto contaminations as HRGC-MS equipment, this simple sample work-up procedureis sufficient for the DR-CALUX R© assay. Optionally, the PCDD/PCDF and PCBfractions can be separated and reported separately using a Carbon, Alumina, orFlorisil column.“CALUX” is an acronym for Chemically Activated LUciferase eXpression. The

DR-CALUX R© assay (Fig. 29.1) used at BioDetection Systems consists of an H4IIErat hepatoma cell line containing the firefly luciferase gene under the control of DREs.This was achieved using a recombinant expression plasmid containing the luciferasegene under the control of part of themouseCyp1a1 promoter. This part of the promoter(bases –1301 to –819) contains four DREs with the consensus sequence TNGCGTG.When the cells are exposed toTCDD-like chemicals, luciferase is expressed in a time-,dose-, and AhR-dependent manner (Fig. 29.1). The luciferase expression can easilybe quantified by adding luciferin, which is metabolized by the luciferase enzyme toproduce light. The emitted light can be accurately detected using a luminometer.Stable transfection of the vector into various cell lines has produced a series

of species-specific cell bioassay systems. Originally, the mouse hepatoma cell lineHepa1c1c7 was used to create the DR-CALUX R© assay. However, rat H4IIE cells arecurrently being used at BioDetection Systems, since they have several advantagesover other cell systems. Some practical considerations were taken into account, suchas the robustness of the cells and the fact that they are easy to culture. Additionally,the initial TEF studies were performed by taking into account binding affinities torat AhR [11], making H4IIE a logical choice at the time. Most importantly, it wasshown that several PCBs elicited an antagonistic effect in Hepa1c1c7 cells, which inturn could lead to false-negative results [17]. This phenomenon was not as prominentin H4IIE cells.The advantage of a reporter gene assay over an analytical method such as HRGC-

MS is that the total toxic potency of a sample is measured directly, while HRGC-MSmeasures the concentration of individual dioxins and dioxin-like congeners, whichthen needs to be multiplied by their TEF to ultimately end up with the total toxicpotency [18]. With DR-CALUX R©, it is also possible to discover novel dioxin-likecompounds [19, 20] or to identify unknown additive effects of several compoundsin a mixture. To permit bioassays to be used for screening of foodstuffs and feed-ingstuffs, the EU has laid down general requirements for the determination of dioxins

538 DR-CALUX R©

and dioxin-like PCBs in foodstuffs and feedingstuffs and specific requirements forcell-based bioassays (EU Commission Directives 2002/69/EC and 2002/70/EC). Anessential prerequisite for bioassays to be used as screening tool is that they haveto be validated, and the laboratories performing the bioassay should be ISO17025accredited. The DR-CALUX R© has been validated for a wide variety of matrices [21],including milk [22], blood [15], sediments [23], pore water [24], butter fat [25], fish[26], emission gas/ashes [27], and various food- and feed-related matrices [28, 29].Generally, the CALUX analysis results are in good agreement with GC-MS–derivedTEQs. These studies showed that the DR-CALUX R© bioassay meets the harmo-nized quality criteria set by the EU for measuring dioxins/furans and dioxin-likePCBs in foodstuffs and feedingstuffs (EU Commission Directives 2002/69/EC and2002/70/EC) [28–30]. Typically, the repeatability and reproducibility of the extrac-tion, cleanup, and bioanalysis using the DR-CALUX R© bioassay for food and feedmaterials is below 15 and 25% respectively.Besides being validated, accredited, and reliable, bioassays need to be sensitive

enough to be able to quantify the levels of dioxins and related compounds at thepg level. Detection limits depend not only on the sensitivity of the assay, but alsoon the format (microtiter plate type), the extraction procedure, and the amount ofsample extracted. In 96-well format, the limit of detection (LOD) is 0.3 pM, whilethe limit of quantification (LOQ) is 1 pM. With the current standard sample volumesand the extraction and cleanup methods used, an LOQ at least five times below themaximum EU limit for dioxins/furans in various food and feed items can be achieved.For example, the LOQ for processing 9 g of feedingstuff is 0.11 ng TCDD TEQ/kgproduct (EU limit: 0.75 ng TCDD TEQ/kg).Furthermore, to ensure the reliability and performance of the DR-CALUX R© bioas-

say for monitoring foodstuffs and feedingstuffs, the participation in interlaboratorycomparison studies (ring tests) is mandatory. In 2005, the first international interlab-oratory comparison study with DR-CALUX R© for food and feed was performed [31].In this extensive three-phase study, 21 laboratories analyzed a set of pure compounds,prepared extracts, and raw food/feed material that had to be extracted by the partici-pants. The results were found to be in close agreement with HRGC-MS results in themajority of participating laboratories (97%), thereby obtaining z-scores below 2.

29.2.3 Perspectives of the DR-CALUX R©

Over the last decades, incidents with dioxins have frequently occurred (Table 29.2).In a crisis situation, the demand for dioxin analyses drastically increases. The DR-CALUX R© enables the analysis of large numbers of samples (e.g., up to 500 samplesper week at BioDetection Systems) and requires only minimal investment costscompared to HRGC-MS, which makes the DR-CALUX R© particularly suitable forless wealthy countries. The DR-CALUX R© has successfully been used in many dioxinincidents in the past (Table 29.2, in bold). In the Belgian dioxin crisis in 1999,for example, more than 2000 samples were analyzed. Eighty-nine percent of thesesampleswere below the relevant cut-off level, which is defined as 25%of the regulatedlevel in the European Union. Re-analysis of negative samples using HRGC-MS

DETECTION OF DIOXINS 539

TABLE 29.2 Examples of Incidents Involving Dioxins in the Past Decadesa

Year Location Situation

1996 USA Contaminated ball clay used as anti-caking agent infeed for poultry and catfish

1998 Netherlands,Germany

Contaminated lime used in Brazil as drying agentfor citrus pulp used in feed

1999 Belgium Technical PCB oil used for chicken and pig feed1999 Austria, Netherlands Contaminated clay used as anti-caking agent in

vitamin mixes for feed1999 Germany Grass meal (feed) was dried on open fire of

dioxin-containing waste wood2000 Spain Contaminated saw dust mixed with provitamin

choline chloride in feed2003 Germany,

NetherlandsBakery waste (feed) was dried on open fire of

dioxin-containing waste wood2004 Netherlands Contaminated clay mixed with potato peels fed to

lactating cows2006 Belgium, Netherlands Contaminated HCl used to produce fat from

gelatin2008 Chile Contaminated recycled zinc used in pork feed2008 Ireland Bakery waste (feed) dried on open fire with diesel

containing PCB oil2009 Italy Contaminated buffalo mozzarella due to buffaloes

grazing in waste deposit area2009 Netherlands,

GermanyContaminated corn from the Ukraine, source

unknown2011 Germany Contaminated technical fat from biodiesel

production used in animal feed2011 Germany Contaminated sugar melasse2012 Netherlands Natural red colorant

aDR-CALUX R© was used during the incidents indicated in bold.

identified only 0.7% false-negative results. Also in the case of the dioxin crisis inGermany in 2011, more than 1800 feed/food samples were tested by BioDetectionSystems. Over 97% of these samples were compliant [32]. According to currentEuropean Union legislation for feed/food, samples exceeding the threshold have tobe confirmed using HRGC-MS. Therefore, the DR-CALUX R© is most useful in thosesituations in which most of the samples score negative, as was the case in the twoabove-mentioned examples. In the latter case, cost savings compared to HRGC-MS have been approximately 70%, and the turnaround time during this particularcrisis has been 48 to 120 h for DR-CALUX R© versus 10 to 20 days for HRGC-MS[32]. These unusually long turnaround times for HRGC-MS were mainly due tocapacity problems.In addition to large-scale incidents, there are also general problems such as high

levels of dioxins in eggs from free-range hens and eel from polluted rivers. This

540 DR-CALUX R©

stresses the need for routine monitoring. DR-CALUX R© is used for regulatory pur-poses by several governments, for example, those of Chile, China, Germany, Ireland,Italy, Slovakia, Spain, The Netherlands, and Taiwan. In addition, because of the factthat the typical market prices for DR-CALUX R© are 2.5 to 4 times lower than thosefor HRGC-MS, many private companies and producers are also making use of theDR-CALUX R© for routine monitoring.

29.2.4 Improvements and Further Development

The original DR-CALUX R© cell line was already constructed in the early 1990s. Sincethen, several variations and improvements have been made. The first improvementinvolved the luciferase protein. The pGL2-based luciferase is targeted to the per-oxisomes and subsequently degraded, resulting in transient expression. Therefore, a“CAFLUX” assay was designed, in which the DREs were coupled to the more stable“enhanced green fluorescent protein” (EGFP) [33, 34]. Measurement of EGFP doesnot require cell lysis or the use of a substrate like luciferin; it is therefore possibleto measure in real time if desired. The CAFLUX assay was reported to be at least assensitive as the DR-CALUX R©. However, a drawback is that the EGFP protein has along half-life, and, therefore, the CAFLUX cells slowly obtain an increased amountof basal fluorescence due to background exposure.Second, attempts have been made to improve the sensitivity of the assay in order to

facilitate the quantification of dioxins and dioxin-like compounds in small amountsof samples, or samples containing low levels of dioxins, such as blood or serum. Toaccomplish this, the number of DREs in the DR-CALUX R© G2 plasmid [35–37] orin the original DR-CALUX R© plasmid [38] was increased by inserting five repeats ofthe mouse Cyp1a1 promoter, resulting in 20 DREs instead of the original 4 DREs.The vectors were tested in several cell lines including mouse Hepa1c1c7 and ratH4IIE.With the 20× DRE-G2 plasmid, although the maximal luciferase activity wasdramatically increased, the baseline signal was equally increased, thereby resultingin similar induction factors. The EC50 and LOQ values decreased 3-fold, resultingin a slightly more sensitive cell line [39]. However, this cell line is still not assensitive as the original DR-CALUX R© in H4IIE cells. With the 20× DRE-originalDR-CALUX R© plasmid in H4IIE cells, the baseline signal remained similar, while themaximum response increased, resulting in increased induction. However, the EC50and LOQ values were not improved. In conclusion, the increment of the number ofDREs does not significantly improve the sensitivity of the assay, but rather increasesthe response maximum.Another area, in which improvements have been made, is the selectivity of the

assay. The original DR-CALUX R© plasmid pGudLuc1.1 contains part of the mouseCyp1a1 promoter and anMMTV long terminal repeat. This plasmid possibly containsadditional responsive elements for other nuclear receptor pathways besides the fourknown DREs. For example, it has been observed that the DR-CALUX R© respondsto the glucocorticoid receptor (GR) agonists dexamethasone, hydrocortisone, andcorticosterone [40]. This type of behavior could result in a false-positive responsewhen analyzing samples that (also) contain GR agonists. Therefore, Sonneveld et al.

CONCLUSIONS AND OUTLOOK 541

[41, 42] created an H4IIE cell line with a minimal promoter construct containing fourmultimerized DREs in front of a TATA promoter coupled to the enhanced luciferase.This cell line is not only more selective for AhR agonists, but also has a 10-foldincreased maximum response to TCDD. Additionally, the more stable luciferasevariant makes this assay suitable for the detection of unstable AhR agonists such asPAHs. However, since GR agonists are usually removed in the cleanup step and aretherefore very unlikely to end up in the final extract, the practical added value of thiscell line is limited.In addition to sensitivity and specificity, the issue of fast turnover and high capacity

has also been addressed. This is particularly important in a crisis situation, when stillattempting to track the source of the contamination or when large numbers of sampleshave to be analyzed to map the extent of the contamination and to clear products frompossible dioxin contamination for export. The DR-CALUX R© assay was originallyperformed in 24-well plates, but improved extraction methods resulted in sufficientsensitivity to switch to 96-well plates. In this format, DR-CALUX R© is three to fivetimes faster than HRGC-MSwhen considering a set of 100 samples [3]. Furthermore,the assay is very well suited for automation. A liquid handling system coupled toa CO2 incubator and a luminometer can dilute the samples, prepare the exposuremedium, retrieve the cell plates from the incubator, add the exposure medium, storethe plates for 24 h, lyse the plates after incubation, and feed them to the luminometer,all without user intervention. Not only is a robotic system often faster, more reliable,and more constant in its performance than laboratory personnel, theoretically such asystem could function round the clock, 7 days a week. Of course in practice, there willbe limitations, such as the available space in the incubator or on the deck of the roboticsystem. At BioDetection Systems, the DR-CALUX R© assay has been automated on aHamilton STARlet pipetting robot coupled to a CO2 incubator. The performance ofthe system has been tested by comparing a TCDD dose–response curve experimentperformed by the robot to a manual experiment (Fig. 29.2a). After optimization,the curves overlapped, thus showing that the results obtained with the robot andmanually were identical. Subsequently, samples with known concentrations wereanalyzed by the robot. The calculated concentrations turned out to perfectly matchtheir actual concentration (Fig. 29.2b). This system is able to analyze 200 samples perday in a 96-well format. Additionally, the DR-CALUX R© can be performed in 384-well plates (Fig. 29.2c), resulting in the same EC50 and LOD/LOQ values as in the96-well format.

29.3 CONCLUSIONS AND OUTLOOK

The DR-CALUX R© is at present one of the most promising screening methods fordioxins and dioxin-like PCBs in food and feed and has been widely used in routinemonitoring and during larger incidents for over two decades. The high frequencyof incidents proves that this is still necessary. Compared to traditional HRGC-MSmethods, theDR-CALUX R© assay is faster, simpler, and cheaper. Furthermore, the useof bioassays in addition to chemical methods allows the discovery of potential novel

542 DR-CALUX R©

% o

f max

imum

res

pons

e

100

50

0R

LU

60,000

(a) (b) (c)

40,000

20,000

0–14 –13 –12

LOG[TCDD] (M) LOG[TCDD] (M)–11 –10 –9 %

of m

axim

um r

espo

nse

100

50

0–14 –13 –12

LOG[TCDD] (M)–11 –10 –9–13.0 –12.5–12.0 –11.5–11.0

FIGURE 29.2 Automation of the DR-CALUX R©. (a) The TCDD dose–response curvesobtained with the robot (closed triangles, continuous line) and manual (open triangles, dottedline) overlap. (b) When samples of known concentrations (closed triangles) are measured withthe robot, the data points fit perfectly on the TCDD dose–response curve (open squares withfitted sigmoidal curve). (c) The TCDDdose–response curves obtained in 384- (closed triangles,continuous line) and 96-well (open triangles, dotted line) overlap.

AhR agonists and the identification of non-additive interactions (synergistic effects,potentiation, inhibition) between chemicals [43]. Therefore, cell-based technologiesremain the ultimate system to evaluate complex cocktails of PXDDs/PXDFs/PXBs(X = Cl, Br, I, F).The sensitivity of the DR-CALUX R© is already high enough to enable the detection

of dioxin levels five times below the maximum EU limit in food and feed, and plasmadioxin levels in blood can be determined with an LOQ of 14 pg WHO-TEQ/g fat(average values in the Western world are 20–40 pg WHO-TEQ/g) with as little as1–1.5 g blood. Regarding throughput, efficiency, as well as standardization of theextraction and cleanup steps, large improvements have been made over the pastdecades. Because efficient extraction and cleanup are the main determinants of theoutcome of the analysis, it is important that different laboratories use the sameprotocols [44]. In recent years, combined and automated systems for extraction andcleanup have been widely evaluated and applied. Several methods for the separationof dioxin-like PCBs and PCDD/PCDFs have also been introduced (Carbon, Alumina,Florisil). However, although the DR-CALUX R© extraction procedure is relatively fastand simple compared to HRGC-MS, there is still room for further improvement.A major improvement in recent years has been the automation of the DR-

CALUX R© using liquid handling robots. Not only have the reproducibility, robustness,and throughput been greatly increased by automation, a robotic system also allowsminiaturization, thereby making use of 384-well plates. The advantage of a 384-wellformat is the requirement of smaller sample and reagent volumes and less cells andconsumables, which makes the assay cheaper and more practical.In addition to the DR-CALUX R©, in recent years several reporter gene assays

with other endpoints based on the same principle have been developed. Examplesare AR CALUX (for the detection of androgens), ER CALUX (for the detection ofestrogens), GR CALUX (for the detection of glucocorticoids), and PPAR� CALUX

REFERENCES 543

(for the detection of peroxisome proliferators) [45, 46]. With the automated CALUXassay it is possible to analyze multiple relevant endpoints in parallel with the sameextract. This screening approach is already being used for toxicological profiling inseveral projects, such as ChemScreen and Tox21.Overall, the DR-CALUX R© is a sensitive and efficient high-throughput bioassay

for the detection of dioxins and dioxin-like molecules in a wide variety of matrices.It is well suited for routine monitoring as well as in crisis management and hasinteresting perspectives as part of a screening panel for toxicological profiling.

ACKNOWLEDGMENTS

This work was carried out in part with financial support from the Commission of theEuropean Communities, the collaborative project ChemScreen (GA244236), and theEcogenomics/EcoLinc project.

REFERENCES

1. Rappe, C., Buser, H.R., Bosshardt, H.-P. (1979). Dioxins, dibenzofurans and other poly-halogenated aromatics: use, formation, and destruction. Annals of the New York Academyof Sciences, 320, 1–18.

2. Jimenez, B., Gonzalez, M.J., Hernandez, L.M. (1991). Presence of PCDDs and PCDFs infood for human consumption: a review. Food Chemistry, 39, 257–271.

3. Behnisch, P.A. (2005). Rapid methods for dioxins in food and feed. In: Rapid Methods forBiological and Chemical Contaminants in Food and Feed (van Amerongen, A., Barug,D., Lauwaars, M., Eds). Wageningen Academic Publishers: Wageningen; pp. 303–319.

4. Poland, A., Knutson, J.C. (1982). 2,3,7,8-Tetrachlorodibenzo-p-dioxin and related halo-genated aromatic hydrocarbons: examination of the mechanism of toxicity. Annual Reviewof Pharmacology and Toxicology, 22, 517–554.

5. Safe, S.H. (1994). Polychlorinated biphenyls (PCBs): environmental impact, biochemi-cal and toxic responses, and implications for risk assessment. CRC Critical Reviews inToxicology, 24, 87–149.

6. Giesy, J.P., Ludwig, J.P., Tillitt, D.E. (1994). Dioxins, dibenzofurans, PCBs and colonial,fish-eating water birds. In: Dioxins and Health (Schecter, A., Ed.). Plenum Press: NewYork; pp. 249–307.

7. Denison, M.S., Vella, L.M., Okey, A.B. (1986). Structure and function of the Ah receptorfor 2,3,7,8-tetrachlorodibenzo-p-dioxin. Species difference in molecular properties of thereceptors from mouse and rat hepatic cytosols. Journal of Biological Chemistry, 261,3987–3995.

8. Okey, A.B., Vella, L.M., Harper, P.A. (1989). Detection and characterization of a lowaffinity form of cytosolic Ah receptor in livers of mice nonresponsive to induction ofcytochrome P1-450 by 3-methylcholanthrene. Molecular Pharmacology, 35, 823–830.

9. Mimura, J., Yamashita, K., Nakamura, K., Morita, M., Takagi, T.N., Nakao, K., Ema,M., Sogawa, K., Yasuda, M., Katsuki, M., Fujii-Kuriyama, Y. (1997). Loss of teratogenic

544 DR-CALUX R©

response to 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) in mice lacking the Ah (dioxin)receptor. Genes Cells, 2, 645–654.

10. Mimura, J., Fujii-Kuriyama, Y. (2003). Functional role of AhR in the expression of toxiceffects by TCDD. Biochimica et Biophysica Acta, 1619, 263–268.

11. Safe, S.H. (1990). Polychlorinated biphenyls (PCBs), dibenzo-p-dioxins (PCDDs), diben-zofurans (PCDFs), and related compounds: environmental andmechanistic considerationswhich support the development of toxic equivalency factors (TEFs).CRC Critical Reviewsin Toxicology, 21, 51–84.

12. Kociba, R.J., Schwetz, B.A. (1982). Toxicity of 2,3,7,8-tetrachlorodibenzo-p-dioxin(TCDD). Drug Metabolism Reviews, 13, 387–406.

13. van den Berg, M., Birnbaum, L.S., Denison, M.S., De Vito, M., Farland, W., Feelev,M., Fiedler, H., Hakansson, H., Hanberg, A., Haws, L., Rose, M., Safe, S., Schrenk,D., Tohyama, C., Tritscher, A., Tuomisto, J., Tysklind, M., Walker, N., Peterson, R.E.(2006). The 2005 World Health Organization reevaluation of human and mammaliantoxic equivalency factors for dioxins and dioxin-like compounds. Toxicological Sciences,93, 223–241.

14. Reiner, E.J., Clement, R.E., Okey, A.B., Marvin, C.H. (2006). Advances in analyti-cal techniques for polychlorinated dibenzo-p-dioxins, polychlorinated dibenzofurans anddioxin-like PCBs. Analytical and Bioanalytical Chemistry, 386, 791–806.

15. Aarts, J.M.M.J.G., Cenijn, P.H., Blankvoort, B.M.G., Murk, A.J., Brouwer, A., Bovee,T.F.H., Traag, W.A., Hoogenboom, L.A.P., Patandin, S., Weisglas-Kuperus, N., Sauer,P.J.J., Denison, M.S. (1996). Application of the chemical-activated luciferase expression(CALUX) bioassays for quantification of dioxin-like compounds in small samples ofhuman milk and blood plasma. Organohalogen Compounds, 27, 285–290.

16. Garrison, P.M., Tullis, K., Aarts, J.M.M.J.G., Brouwer, A., Giesy, J.P., Denison, M.S.(1996). Species-specific recombinant cell lines as bioassay systems for the detection of2,3,7,8-tetrachlorodibenzo-p-dioxin-like chemicals. Fundamentals and Applied Toxicol-ogy, 30, 194–203.

17. Aarts, J.M.M.J.G., Denison, M.S., Cox, M.A., Schalk, M.A., Garrison, P.M., Tullis, K.,De Haan, L.H.J., Brouwer, A. (1995). Species-specific antagonism of Ah receptor actionby 2,2′,5,5′-tetrachloro- and 2,2′,3,3′4,4′-hexachlorobiphenyl. European Journal of Phar-macology, 293, 463–474.

18. Behnisch, P.A., Hosoe, K., Sakai, S. (2001). Bioanalytical screening methods for dioxinsand dioxin-like compounds a review of bioassay/biomarker technology. EnvironmentInternational, 27, 413–439.

19. Behnisch, P.A., Hosoe, K., Sakai, S. (2003). Brominated dioxin-like compounds: in vitroassessment in comparison to classical dioxin-like compounds and other polyaromaticcompounds. Environment International, 29, 861–877.

20. Sakiyama, T., Weber, R., Behnisch, P.A., Nakano, T. (2011). Preliminary assessmentof dioxin-like compounds in/from chlorpyrifos – A potential precursor of the pyridineanalogue of 2,3,7,8-TCDD. Organohalogen Compounds, 73, 146–149.

21. Behnisch, P.A., Hosoe, K., Brouwer, A., Sakai, S. (2002). Screening of dioxin-like toxicityequivalents for various matrices with wildtype and recombinant rat hepatoma H4IIE cells.Toxicological Sciences, 69, 125–130.

22. Bovee, T.F.H., Hoogenboom, L.A.P., Hamers, A.R.M., Traag, W.A., Zuidema, T.,Aarts, J.M.M.J.G., Brouwer, A., Kuiper, H.A. (1998). Validation and use of the

REFERENCES 545

CALUX R©-bioassay for the determination of dioxins and PCBs in bovine milk. FoodAdditives and Contaminants, 15, 863–875.

23. Besselink, H.T., Schipper, C., Klamer, H., Leonards, P., Verhaar, H., Felzel, E.F., Murk,A.J., Thain, J., Hosoe, K., Schoeters, G., Legler, J., Brouwer, A. (2004). Intra- andinterlaboratory calibration of the DR-CALUX bioassay for the analysis of dioxins anddioxin-like chemicals in sediments. Environmental Toxicology and Chemistry, 23, 2781–2789.

24. Murk, A.J., Jonas, A., Brouwer, A., Leonards, P.E.G., Denison, M.S. (1996). Applicationof the CALUX (chemically activated luciferase gene expression) assay for measuringTCDD-equivalents in sediment, pore water, and blood samples. Organohalogen Com-pounds, 27, 291–296.

25. Bovee, T.F.H., Hoogenboom, L.A.P., Traag, W.A., Zuidema, T., Hortsman, J.H.J., Aarts,J.M.M.J.G., Murk, T.J., Brouwer, A., Denison, M.S., Kuiper, H.A. (1996). Biologicalscreening of Ah receptor agonist activity in butter fat and coconut oil by means ofchemical-activated luciferase expression in a genetically engineered cell line (CALUX R©).Organohalogen Compounds, 27, 303–308.

26. Besselink,H.T., Leonards, P., Felzel, E.F., Brouwer,A. (2002).Analysis of polychlorinateddibenzo-p-dioxins (PCDD), dibenzofurans (PCDF) and biphenyls (PCB) in fish using DRCALUX R© and GC/MS: A comparison. Organohalogen Compounds, 58, 413–415.

27. Behnisch, P.A., Hosoe, K., Brouwer, A., Sakai, S. (2001). Cross-validation study ofthe DR CALUX R© bioassay and chemical analyses. Organohalogen Compounds, 54,81–85.

28. Besselink, H.T., Jonas, A., Pijnappels, M., Swinkels, A., Brouwer, A. (2004). Validationof extraction, clean-up and DR Calux R© bioanalysis. Part I: Feedingstuff. OrganohalogenCompounds, 66, 677–681.

29. Besselink, H.T., Jonas, A., Pijnappels, M., Swinkels, A., Brouwer, A. (2004). Validationof extraction, clean-up and DR Calux R© bioanalysis. Part II: Foodstuff. OrganohalogenCompounds, 66, 682–686.

30. Scippo, M.L., Eppe, G., Maghuin-Rogister, G., de Pauw, E. (2003). Evaluation of theDR-CALUX bioassay for the determination of dioxins in food and feed accordingto the requirements of the Commission Directives 2002/69 and 2002/70 respectively.Organohalogen Compounds, 60, 235–238.

31. Besselink, H.T., Felzel, E.F., Jonas, A., Brouwer, A. (2006). Results of the first interna-tional inter laboratory DR CALUX R© by BDS comparison study for food and feed (BICS2005). Organohalogen Compounds, 68, 249–252.

32. Behnisch, P.A., Besselink, H., Brouwer, B. (2011). German dioxin crisis 2011 – Expressanalysis of PCB- and PCDD/F-TEQ in food and feed matrices by the HTPSDRCALUX R©

method. Organohalogen Compounds, 73, 457–460.

33. Aarts, J.M.M.J.G., Jonas, A., van den Dikkenberg, L.C., Brouwer, A. (1998). CAFLUX,a simplified version of the CALUX assay for Ah receptor (ant)agonist, based on enhancedgreen fluorescent protein (EGFP) reporter gene expression. Organohalogen Compounds,37, 85–88.

34. Han, D., Nagy, S.R., Denison, M.S. (2004). Comparison of recombinant cell bioassaysfor the detection of Ah receptor agonists. Biofactors, 20, 11–22.

35. He, G., Baston, D.S., Denison, M.S., Tsutsumi, T. (2008). Third generation CALUXluciferase reporter vectors – amplification of dioxin responsive elements dramatically

546 DR-CALUX R©

increases CALUX bioassay sensitivity and responsiveness. Organohalogen Compounds,70, 772–775.

36. He, G., Tsutsumi, T., Zhao, B., Baston, D.S., Denison, M.S. (2009). Species-specificthird generation (G3) luciferase cell bioassays show dramatically increased sensitivityand magnitude of response to TCDD and other Ah receptor agonists. OrganohalogenCompounds, 71, 2399–2403.

37. He, G., Tsutsumi, T., Zhao, B., Baston, D.S., Zhao, J., Heath-Pagliuso, S., Denison, M.S.(2011). Third-generation Ah receptor-responsive luciferase reporter plasmids: amplifica-tion of dioxin-responsive elements dramatically increases CALUX bioassay sensitivityand responsiveness. Toxicological Sciences, 123, 511–522.

38. Van Vugt-Lussenburg, B.M.A., Man, H.Y., Farber, K.A., van der Burg, B., Brouwer, A.(2010). Generation of dioxin responsive CALUX H4IIE cell lines containing multipledioxin responsive elements. Organohalogen Compounds, 72, 669–671.

39. van Langenhove, K., Croes, K., Denison, M.S., Elskens, M., Baeyens, W. (2011). TheCALUX bio-assay: Analytical comparison between mouse hepatoma cell lines with alow (H1L6.1c3) and high (H1L7.5c1) number of dioxin response elements. Talanta, 85,2039–2046.

40. Hoogenboom, L.A.P., Hamers, A.R.M., Bovee, T.F.H. (1999). Bioassays for the detectionof growth-promoting agents, veterinary drugs and environmental contaminants in food.Analyst, 124, 79–85.

41. Sonneveld, E., van denBrink,C.E., Zeinstra, L., Jansen, J.H.J., van der Saag, P.T., Brouwer,A., van der Burg, B. (2002). Development of improved DR-CALUX bioassay for sen-sitive measurement of aryl hydrocarbon receptor activating compounds. OrganohalogenCompounds, 58, 369–372.

42. Sonneveld, E., Jonas, A., Meijer, O.C., Brouwer, A., van der Burg, B. (2007).Glucocorticoid-enhanced expression of dioxin target genes through regulation of the rataryl hydrocarbon receptor. Toxicological Sciences, 99, 455–469.

43. Behnisch, P.A., Hosoe, K., Sakai, S. (2001). Combinatorial bio/chemical analysis ofdioxin and dioxin-like compounds in waste recycling, feed/food, humans/wildlife and theenvironment. Environment International, 27, 495–519.

44. Windal, I., Denison, M.S., Birnbaum, L.S., van Wouwe, N., Baeyens, W., Goeyens, L.(2005). Chemically activated luciferase gene expression (CALUX) cell bioassay analysisfor the estimation of dioxin-like activity: critical parameters of the CALUX procedurethat impact assay results. Environmental Science and Technology, 39, 7357–7364.

45. van derLinden, S.C.,Heringa,M.B.,Man,H.Y., Sonneveld, E., Puijker, L.M.,Brouwer,A.,van der Burg, B. (2008). Detection of multiple hormonal activities in wastewater effluentsand surface water, using a panel of steroid receptor CALUX bioassays. EnvironmentalScience and Technology, 42, 5814–5820.

46. Gijsbers, L., Man, H.Y., Kloet, S.K., de Haan, L.H., Keijer, J., Rietjens, I.M., van der Burg,B., Aarts, J.M. (2011). Stable reporter cell lines for peroxisome proliferator-activatedreceptor � (PPAR� )-mediated modulation of gene expression. Analytical Biochemistry,414, 77–83.

INDEX

ACDC assay, see adherent cell differentiationand cytotoxicity assay

ACToR, see Aggregated ComputationalToxicology Resource

adherent cell differentiation and cytotoxicityassay, 360

adverse outcome pathway, 25Aggregated Computational Toxicology

Resource, 8, 17Alamar BlueTM assay, 42, 43, 310, 311, 312,

315, 345, 346, 350, 404Ames liquid fluctuation assay, 195Ames liquid format assay, 195, 203, 204, 205,

206, 207, 209, 210Ames testassay characteristics, 238assay for genotoxicity screening, 38, 241,244, 245, 246, 247, 254, 256, 258, 303,303, 437

assay for photogenotoxicity screening,180, 181

comparison with VitotoxTM assay, 214,215, 216, 221, 222, 223, 224, 226, 227,257, 259

validation, 197, 198, 205Ames II test, 38, 41, 302Ames liquid fluctuation assay, 195Ames liquid format assay, 195, 203, 204, 205,

206, 207, 209, 210

androgen receptor yeast assay, 486, 488antimicrobial, 6, 10, 20, 177AOP, see adverse outcome pathwayapoptosisAh receptor-mediated induction, 534assay characteristics, 123caspase activation, 119correlation with ACDC assay, 360drug toxicity, 396, 397, 398, 401H2O2-mediated cell death, 336high-content imaging readout, 146,147

live imaging, 150, 151mechanisms, 40, 130, 143, 149, 249, 250,491

multiplexed assay, 138, 139, 140multiplex imaging, 145pathways, 129, 523phosphorylated H2AX, 289quantification by flow cytometry, 136screening assay, 43, 45, 117, 120, 131,132, 133, 135

single cell imaging cytometry, 386, 387,388, 389, 390

time-lapse microscopy, 149aryl hydrocarbon receptor yeast assay, 488,

489, 490, 491ATP assay, 114, 115, 120ATP-LiteTM assay, 43


547

548 INDEX

automated patch clamp technique, 404, 405,414

automationAmes liquid format mutagenicity assay,198, 205

Comet assay, 286DR-CALUX.5 R© assay, 541, 542embryonic stem cell test, 344flow cytometric apoptosis analysis, 130,138

fluorimetric detection of alkaline DNAunwinding assay, 288

GreenScreen assay, 276human embryonic stem cell-cardiomyocytemicroelectrode array, 412

induction models, 463, 467micronuclei scoring, 39ReProGlo assay, 345, 350, 352soft agar colony formation assay, 312staining of �H2AX fociVitotoxTM assay, 221yeast-based nuclear receptor assays, 492

BALB/c 3T3 cell line, 182, 317, 318, 319,324, 336

Bhas 42 cell linecell growth assay, 327characteristics, 317handling, 324transformation assay, 318, 319, 321, 322,332, 333, 334, 335, 336

bioactivity profiling, 4, 11, 25, 26biocolloid reactor, 437, 438, 440, 441, 443,

445, 446, 448, 449branched DNA signal amplification assay, 466

C3H10T1/2 cell line, 317, 324calcein-acetoxymethyl assay, 43CALUX.5 R© assay, 519, 521, 522, 523, 524,

525, 526, 527, 528, 529cardiotoxicitycause of drug failure, 35, 213, 403embryonic stem cell test, 99, 100evaluation, 98, 99, 414high-content screening analysis, 399high-throughput in vitro screening, 405induced pluripotent stem cells, 409, 412microelectrode array assay, 410quantitative high-throughput analysis, 391single cell imaging cytometry, 385, 386,387

surface plasmon resonance, 416caspaDAP, 133, 134, 135, 136, 137, 138, 140caspaseantibody-based staining, 149apoptosis pathways, 129, 130, 143

caspaDAP, 134, 135, 136combined, viability/caspase activationassay, 119

fluorochrome-labeled inhibitors ofcaspases, 132

high-throughput assay, 131, 133, 135multiplexed assay, 138, 139same-well, optically averaged multiplex,123

substrates, 43, 119, 123, 129, 130, 131,132, 133, 134, 135, 136, 138, 139, 143,149, 150, 151

time-lapse microscopy, 149, 150, 151caspase activation assay, 119cell growth assayBhas 42 cell transformation assay, 320,321, 322, 323, 324, 326, 328, 329, 330,331, 332

initiation assay, 327promotion assay, 329

CellTiter-BlueTM assay, 110CellTiter-FluorTM assay, 110CellTiter-GloTM assay, 43, 110CellTox-GreenTM assay, 117cell transformation assay, 317, 318, 321, 322,

327, 336CFU assay, see colony forming unit assaychemical library, 9, 10, 17, 20, 80, 90, 377colony forming unit assay, 423, 424, 425,

427Comet assayapplications, 301, 303, 305assay principle, 240, 241, 285, 286assay validation, 290false-positive results, 298HepG2 cells, 250high-throughput assay, 295, 296, 297, 300,301, 303, 304

in vivo assay, 304phototoxicity assessment, 180, 181testing of pharmaceuticals, 237, 242testing of plant extracts, 228

computational toxicology, 4, 7, 8, 9, 15, 16, 27cytochrome P450 inducer, 462, 463, 464, 465,

466cytochrome P450 induction, 460, 461, 462,

463, 464, 465cytochrome P450 inhibition, 453, 454, 456,

468cytochrome P450 inhibitor, 46, 47, 454, 455,

458, 466Cyto-LiteTM assay, 43cytotox CALUX.5 R© cell line, 523, 525, 529cytotoxicity assay, 99, 110, 111, 123, 124,

298, 360cytotoxicity screening, 41, 42, 102

INDEX 549

cytotoxicity test, 42, 296cytotoxicity testing, 45, 102, 109, 113, 502

DAP, see differential anchorage probeDDI, see drug-drug interactionDEREK, 180, 244design of experiment testing, 112developmental toxicitycause of drug failure, 235embryonic stem cell test, 363, 364, 365,366

human embryonic stem cells, 98, 101legislative guidelines, 357omics technologies, 361, 362, 365, 366pesticide active ingredients, 6targeted testing, 25Toxicity Reference Database, 16, 20zebrafish, 371, 374, 379

differential anchorage probe, 133, 134dosimetry, 77, 79, 82, 84, 85, 89, 90DR-CALUX.5 R© assay, 533, 535, 536, 537,

538, 539, 540, 541, 542, 543drug-drug interaction, 80, 440, 443, 453, 456,

458, 459, 460, 461, 469drug metabolism profiling, 433

EFPTFR-HepG2 cell line, 503, 504, 508, 515embryonic stem cell test, 55, 98, 344, 357,

358, 359, 360, 361, 362, 363, 364, 365,366

embryotoxicity, 36, 37, 51, 52, 53, 55, 59, 353,364, 480

EST, see embryonic stem cell testESTc, 361, 362, 363, 365, 366ESTn, 361, 362, 363, 365estrogen receptor screen, 486estrogen receptor yeast assay, 479, 484, 486

FACS-EST, 359FADU assay, see fluorimetric detection of

alkaline DNA unwinding assayflow cytometryapoptosis analysis, 130, 131, 132, 133,134, 135, 136, 140, 397

evaluation of hematotoxicity of anticanceragents, 423

GreenScreen assay, 274phospholipidosis analysis, 44, 45

fluorescence resonance energy transfer, 132,391, 463

fluorimetric detection of alkaline DNAunwinding assay, 224, 286, 287, 288,290

fluorometric microculture cytotoxicityassay-granulocyte macrophage, 424,425, 426

FMCA-GM, see fluorometric microculturecytotoxicity assay-granulocytemacrophage

food additive, 7, 8, 10, 193food ingredient, 7, 535forward dosimetry, 82FRET, see fluorescence resonance energy

transfer

genotoxicity assayAmes test, 193Comet assay, 296comparison of different assays, 223, 263comparison with carcinogenicity assays,238

GreenScreen assay, 252, 279, 281human cell-based assays, 248, 249, 250,251, 252

metabolic activation system, 194micronucleus test, 240, 251, 305sensitivity, 255, 257, 259, 279specificity, 255, 257, 259, 279testing of pharmaceuticals, 242, 245,263

genotoxicity profiling, 433genotoxicity screening, 38, 195, 245, 262,

263, 272, 276, 278, 446, 448GreenScreen assay, 222, 247, 271, 272, 276,

277, 278, 280, 281

�H2AX, 40, 41, 146, 147, 148, 289, 290HALO.5 R© assay, see hematotoxicity assay via

luminescence outputHCS, see high-content screeninghematotoxicity, 421, 422, 423, 424, 426,

427hematotoxicity assay via luminescence output,

424, 425, 426HepaRG cell line, 504, 514hepatic clearance, 79, 81, 83, 84hepatotoxicity, 35, 89, 135, 213, 234, 235,

461, 501, 504, 515, 534HepG2 cell lineACuteTox project, 42CALUX assays, 524, 525caspaDAP, 133, 135, 137, 138clastogenic screening, 40Comet assay, 300cytochrome P450, 47, 49, 53, 249, 463cytochrome P450 induction, 463genotoxicity screening, 248, 249, 251, 252,253, 254, 255, 256, 257, 258, 259, 262,263

high-content screening, 39, 43, 44, 251high-throughput assays, 245, 246high-throughput screening, 43

550 INDEX

HepG2 cell line (Continued )luminescence-based reporter assays, 252,253, 254, 255, 256, 257, 258, 259, 262,263

metabolic capacity, 41, 248, 249, 251, 253nuclear receptor activation assay, 46, 56peroxisome biogenesis assay, 504, 514,515

peroxisome plasticity, 503phospholipidosis, 45steatosis, 45toxicogenomics, 57, 58

HepG2 cystatin A_luc assay, 246, 253, 254,255, 257, 258, 262

HepG2 in vitro micronucleus assay, 246, 253,254, 259, 262

HepG2 p53_luc assay, 246, 253, 254, 255,257, 258, 262

HepG2 peroxisome biogenesis assay, 504, 514HepG2 RAD51C_luc assay, 246, 252, 253,

254, 255, 257, 258, 262HERG, see human ether-a-go-go-related genehigh-content imaging, 131, 143, 504, 507, 509high-content screeninganalysis of apoptosis, 132, 133, 149analysis of hepatotoxicity, 501, 502analysis of toxicity pathways, 144cardiotoxicity testing, 386, 391, 399cytotoxicity screening, 42, 43, 44drug development, 34, 153end-point assays 147genotoxicity testing, 39, 41, 248, 251, 252,253, 254, 259, 262, 263

luminescence-based reporter assays, 246multiparametric methods, 118multiplexing, 145phospholipidosis, 44, 45sensitivity, 44, 45soft packages, 147specificity, 44, 45steatosis, 45

high-throughput equilibrium dialysis, 81high-throughput in vitro assay, 5, 90high-throughput screeningAmes IITM assay, 195analysis of apoptosis, 131, 132analysis of toxicity pathways, 143, 144cardiotoxicity, 99, 386, 387, 403, 404, 405,406, 414

cytotoxicity, 42, 43, 44, 109, 111, 113,115, 116, 117, 123

drug development, 12, 13, 34, 78, 245drug metabolizing enzymes, 50embryotoxicity, 55, 57genotoxicity testing, 38, 39, 40, 222, 229hematotoxicity, 421, 423, 425hepatotoxicity, 101

human dosimetry, 77, 79, 89human embryonic stem cells, 98induced pluripotent stem cells, 409malignant cell transformation, 309multiparametric methods, 119multiplexed assays, 120mutagenicity testing, 213, 214neurotoxicity, 100phospholipidosis, 44phototoxicity, 177, 178, 180, 184, 187receptor-mediated toxicity, 479resazurin assays, 114same-well, optically averaged multiplex,118

single cell imaging cytometry, 389, 391skin sensitization, 159soft agar colony formation assay, 310steatosis, 45surface plasmon resonance, 416time-lapse microscopy, 152ToxCast, 5, 6, 8, 9, 11, 15, 20, 22, 23, 26,27, 42, 78, 99

toxicogenomics, 57viability assays, 111yeast nuclear receptor activation assays,479, 482, 483, 484, 492, 493

zebrafish, 372hit optimization, 34, 35HOMO-LUMO gap, 180HTS, see high-throughput screeninghuman embryonic stem cell, 97, 98, 99, 100,

101, 102, 359, 410, 413human ether-a-go-go-related gene, 12, 391,

392, 394, 399, 402, 412

IC50 shift assay, 459, 460induced pluripotent stem cell, 99, 102, 153,

409, 412, 417industrial chemical, 7, 8, 9, 80, 193, 305initiation assay, 318, 321, 322, 325, 327, 328,

330, 331in vitro-to-in vivo extrapolation, 80, 84iPSC, see induced pluripotent stem cellIVIVE, see in vitro-to-in vivo extrapolation

KeratinoSens assay, 159, 161, 162, 163, 165,166, 167, 168, 169, 170, 171

lactate dehydrogenase, 42, 116, 123, 130, 143,404, 406

LDH, see lactate dehydrogenaselead optimizationcarcinogenicity testing, 244drug development phases, 233, 245genotoxicity testing, 41, 235, 244, 245,246, 247, 251, 259, 262, 263

human embryonic stem cells, 98

INDEX 551

in vitro screening strategy, 34receptore activation assays, 236toxicity screening, 37, 235

liquid culture assay, 423, 424live cell protease, 110, 115luminometry, 50, 455Luxcel assay, 43

MEA, see microelectrode arraymechanism-based inhibition, 458metabolic biomarker, 113metabolic profiling, 443, 444, 448metabolic stability assay, 81metabolite profiling, 437metabolomics, 42, 364microelectrode array, 404, 406, 407, 409, 410,

411, 412, 413, 414, 417microfluidic system, 389micronucleus test, 222, 228, 238, 240, 241,

249, 271, 301, 302, 304MitoXpress assay, 43mobility shift microfluidics assay, 12MPF Ames test, 246MTS assay, 113, 143MTT assay, 42, 43, 113, 130, 143, 164, 423multiparametric methods, 118, 122, 124multiplexed assay, 118, 123, 124, 135, 138,

139, 140, 145, 506multiplex imaging, 145

NBT assay, 43, 183neurotoxicity, 100, 235, 396NIH/3T3 cell line, 317, 324, 374nonmetabolic biomarker, 113normalization, 121, 122, 123, 151, 274, 467Nrf2assay sensitivity, 246assay specificity, 246CALUX.5 R© assay, 524, 525, 526HepG2 cell-based assay, 250, 252, 253,254, 255, 256, 257, 258, 262

KeratinoSens assay, 162, 163, 167, 168,170

reactive oxygen species assessment, 43

P450-GloTM assay, 47patch clamp technique, 386, 387, 400, 405,

414peroxisomal biogenesis, 502, 503, 504peroxisomal phenotype, 503, 511, 513peroxisomal proliferator, 505pesticideactive ingredients, 6Ames test, 193androgen receptor yeast assay, 488chemical libraries, 9, 10, 511estrogen receptor yeast assay, 486

human oral exposure, 82inerts, 8progesterone receptor yeast assay, 488retinoic acid receptor yeast assay, 491retinoic X receptor yeast assay, 492ToxCast, 80, 377Toxicity Reference Database, 16, 17, 20water contaminants, 8zebrafish embryo development, 377

pharmaceuticalapoptotic potential, 135CALUX.5 R© assay, 523, 526, 527carcinogenicity testing 244cardiac safety evaluation, 403chemical libraries, 10, 11, 280Comet assay, 301embryonic stem cell test, 358estrogen receptor yeast assay 486genotoxicity testing, 213, 228, 236, 237,242, 263, 271, 272, 278, 279, 280

GreenScreen assay, 279, 280, 281MEIC study, 42micronucleus test, 305OECD Guidelines, 194peroxisome proliferator-activated receptoractivators, 26

phototoxicity, 177, 178, 179, 183, 185, 187VitotoxTM assay, 221zebrafish neuromast damage, 373

phospholipidosis, 44phototoxicity, 177, 178, 179, 180, 181, 182,

183, 185, 186, 187, 510planar patch clamp technique, 414predictive model, 4, 5, 20, 22, 24, 27predictive modeling, 17, 22predictive signature, 12, 20, 25, 26, 89prioritization, 5, 8, 24, 25, 27, 78, 87, 90, 91,

460, 504, 514procaspase, 130progesterone yeast assay, 488, 489, 490, 524promotion assay, 318, 321, 322, 325, 328, 329,

332, 335protease release assay, 116proteomics, 42, 359, 361, 364

RABIT, see Rapid Automated BiodosimetryTool

RadarScreen assay, 221, 222, 223, 246, 247,254, 256, 257, 258, 259, 262

Rapid Automated Biodosimetry Tool, 290ReProGlo assay, 343, 344, 345, 346, 347, 348,

349, 350, 351, 352, 353, 359, 360resazurin assay, 114retinoid receptor assay, 480, 491, 492reverse dosimetry, 79, 82, 89, 90ribonuclease protection assay, 466RNAi screening, 150

552 INDEX

robotic system, 144, 541, 542RT-qPCR assay, 366, 367

same-well, optically averaged multiplex, 118SCGE assay, see single cell gel electrophoresis

assayscintillation proximity assay, 463single cell gel electrophoresis assay, 240, 285,

295single cell imaging cytometry, 386, 387, 388,

389, 390, 393, 394, 395, 396, 398, 399,400

single parameter assay, 113skin sensitization, 159, 160, 161, 166, 170soft agar colony formation assay, 309, 310,

311, 312, 313, 314, 317, 335, 336SOS chromotest, 214, 216, 223, 224, 246SPA, see scintillation proximity assaySPR, see surface plasmon resonancesurface plasmon resonance, 11, 404, 416SWOAM, see same-well, optically averaged

multiplex

3T3 neutral red uptake phototoxicity assay,179, 180, 181, 182, 183, 185, 186

3T3 NRU PT, see neutral red uptakephototoxicity assay

targeted testing, 5, 24, 25, 26TDI, see time-dependent inhibitiontime-dependent inhibition, 458, 459, 460time-lapse microscopy, 150Tox21, 5, 8, 10, 11, 543ToxCastACDC assay, 360activity:exposure ratios, 88biological activity profiling, 12, 23, 25CALUX.5 R© assay, 522, 528chemical libraries, 9, 10, 11, 12, 22, 79, 81,511, 512, 513

GreenScreen assay, 280oral equivalent dose, 82, 83, 84testing battery, 14, 15, 16, 21, 24zebrafish embryo development, 377

toxicity prediction model, 6toxicity profiling, 425, 427, 501Toxicity Reference Database, 16, 17, 18, 19,

20, 21, 25, 374toxicity screeningattrition rate reduction, 35biocolloid reactor particles, 445, 446,448

CALUX.5 R© assay, 523, 526cytotoxicity, 41, 42, 43, 44, 45drug development, 36, 37

genotoxicity, 38, 39, 40, 41, 195, 245, 262,263, 272, 276, 278

GreenScreen assay, 272, 276, 278high-content screening, 144human embryonic stem cells, 98, 102lead optimization, 37, 40, 144, 245, 371,445, 448, 449, 481, 519, 520, 523, 526

yeast nuclear receptor assays, 481zebrafish, 371

toxicogenomics, 42, 53, 57, 363ToxRefDB, see Toxicity Reference Databasetransactivation assay, 13, 463, 482transcriptomics, 100, 359, 361, 364troponin, 404, 406, 416troponin SPR assay, 416trypan blue, 117, 423

viability assay, 42, 99, 110, 114, 115, 116,123, 387, 389

VitotoxTM assayapplications, 221, 222comparison of genotoxicity tests, 223, 224,225

examples of test results, 218, 219, 220genotoxicity screening, 38, 41, 246, 247,254, 256, 257, 258, 259, 262, 263

high-throughput assays, 213, 214principle, 214, 215, 216procedure, 217sensitivity, 246, 259specificity, 246, 259testing of environmental samples, 226testing of medicinal plant extracts, 228testing of nonionizing radiation, 228, 229testing of smoke-water, 227

water contaminant, 8Wnt, 344, 345, 346, 348, 351, 352, 353, 359WST-1 assay, 43WST-8 assay, 318, 320, 321, 322, 323, 328,

329, 331, 332, 333

XTT assay, 43, 113

yeast assay, 276, 480, 481, 482, 484, 485, 486,487, 488, 489, 491, 492, 493

yeast estrogen screen, 486yeast-two hybrid method, 482, 485, 486, 487,

490, 492yeast-two hybrid model, 463

zebrafish, 56, 371, 372, 373, 374, 375, 376,377, 378, 379

Z’ factor, 114, 132, 140

GPCR

Transcription factors

Cytokine/receptor

Nuclear receptor

Cytochrome P450

Other

Cellular phenotype

Protease

Cell adhesion molecule

Other enzyme

Ion channel

Gene expression

Phosphatase

Transporter

Other receptor

Kinase

FIGURE 1.1 Distribution of assays categories in the ToxCast Phase I testing battery.


(a)

(b)

FIGURE 2.9 Hierarchical clustering of various nuclear receptor agonists for the aryl hy-drocarbon receptor (AhR), constitutive androstane receptor (CAR), liver X receptor (LXR),peroxisomal proliferation activating receptor � (PPAR�) and pregnane X receptor (PXR) (a)and a magnification of the specific CAR- and PXR-mediated gene activation and inactivationof CYP2B and CYP3A in particular (b). In the white rectangles A, B and C, the specificup-regulated genes mediated by CAR, PPAR�, and LXR, respectively, are shown, while in Dand E the specific PXR- and CAR-modulated CYP3A and CYP2B are visible.

Genotoxic Non-genotoxic

FIGURE 2.10 Hierarchical clustering of gene fragments in HepG2 cells treated for 24 or72 h with genotoxic (camptothecin [CAMP], ethinyl estradiol [EE], methyltestosterone [MT],cisplatin [Cis-Pt], doxorubicin (DOX), benzo[a]pyrene [B(a)P], all marked in yellow), andnon-genotoxic but cholestatic (�-naphthylisothiocyanate [ANIT], chlorpromazine [Cl-Pro]) ornecrotic compounds (tetracarbon chloride [CCl4], iproniazid [Ipro], acetaminophen [APAP],all marked in blue).

Time (min) (STS 1μM)

Cyt c -GFP

CaspaDAP Red

Overlay

Fluorescent protein TMD DEVD

Cyt c

CaspaDAP Red Overlay

125’ 128’ 131’ 134’

(a)

(b)

(c)

(d)

Untreated

STS 30 nM

STS 3 μM

2520

15C

ount

sC

ount

sC

ount

s10

50

85

42

010

50

100

100 101 102 103 104

100 101 102 103 104

101 102

FL1-Height

FL1-Height

FL1-Height

103 104

M1

M1

M1

FIGURE 6.1 A differential anchorage probe for effector caspase activity (CaspaDAP). (a)Schematic representation of caspaDAPwith its cleavable DEVDmotif; (b) Time lapse imagingof HeLa cells stably expressing GFP-cytochrome c (Cyt c-GFP) (kind gift of Dr. D.R. Green),transiently transfected with caspaDAP Red (the fluorescent protein is Hcred-tamdem fromEvrogen) and incubated with 1 �M staurosporine (STS); (c) Cytochrome c immunofluores-cence in HeLa cells transiently transfected with Red caspaDAP and treated 6 h with 1�MSTS.Both experiments show the correlation between cytochrome c release and redistribution of cas-paDAP; (d) Flow cytometric profiles of caspaDAP Green (the fluorescent protein is TurboGFPfrom Evrogen) after plasma membrane permeabilization of HeLa cells stably expressing theconstruct, either untreated or treated with 30 nM or 3 �M STS for 24 h.

AraC

5-FU

Vincr

istin

eEto

poside

Campto

thec

in CisPlat

in Arsen

ite MG-132

benzo

ylSTS

Doceta

xel

103

104

105

106

107

108

109

Pro

lifer

atio

n 24

H

Cas

pase

24

h

Cyt

olys

is 2

4 h

Pro

lifer

atio

n 48

H

Cas

pase

48

h

Cyt

olys

is 4

8 h

No

t se

nsi

tive

EC50

(a)

NT

Fla

vop

irid

ol

NT

020406080100

124816

10–9

10–8

10–7

10–6

10–5

10–4

T

OP

91.7

9

T

OP

92.5

6

T

OP

6.95

114

.63

% o

f cel

ls D

EV

Das

e+ 2

4 h

4.88

4e-0

07%

of c

ells

DE

VD

ase+

48

h4.

254e

-007

prol

ifera

tion

24 h

1.61

3e-0

07pr

olife

ratio

n 48

h1.

798e

-007

M

% of cells with DEVDase activity

Proliferation index

NT

Milt

efo

sin

e

NT

020406080100

124816

10–8

10–7

10–6

10–5

10–4

10–3

T

OP

~ 1

00.0

89.6

36.

686

14.1

9

%of

cel

ls D

EV

Das

e+ 2

4 h

0.00

0235

8%

of c

ells

DE

VD

ase

+ 4

8 h

4.55

8e-0

05pr

olife

ratio

n 24

h1.

870e

-006

prol

ifera

tion

48 h

~ 2

.463

e-00

6

M


Proliferation index

NT

Pac

litax

el

NT

020406080100

124816

10–9

10–8

10–7

10–6

10–5

T

OP

39.8

2

T

OP

73.3

56.

499

13.9

4

% o

f cel

ls D

EV

Das

e +

24

h6.

327e

-008

% o

f cel

ls D

EV

Das

e +

48

h6.

173e

-008

prol

ifera

tion

24 h

3.17

9e-0

08pr

olife

ratio

n 48

h2.

304e

-008

M


Proliferation index

(b)

FIG

UR

E6.

4Characterizationofthecytotoxiceffectsofdrugsbymultiplexedassays.(a)SchematicrepresentationofEC50valuesforcaspase(DEVDase)

activity,proliferation,andcytolysiscalculatedfrom

dose–responsecurvesintwomultiplexedassays(apoptosis-proliferationandapoptosis-cytolysis)for

cellstreatedfor24or48hwith10differentcompounds.(b)Dose–responsecurvesregardingcaspaseactivityandproliferationarrestforthreeanticancer

drugs(flavopiridol,paclitaxel,andmiltefosine)assessedafter24or48hoftreatment.

Gen

otox

ic

Fal

se p

ositi

ves

Non

-gen

otox

ic

Test systemAmes Vito


egulatory

In vitro m

ammalian regulatory

HCS IVMN CHO

HCS IVMN-HepG2

Cystatin A_luc


Nrf2_luc

Test systemAmesVito


egulatory

In vitro m

ammalian regulatory

HCS IVMN CHO

HCS IVMN-HepG2

Cystatin A_luc


Nrf2_luc

Test system Ames Vitotox Radarscreen

In vivo regulatory

In vitro mammalian regulatory

HCS IVMN CHO

HCS IVMN-HepG2

Cystatin A_luc


Nrf2_luc

2-A

AF

o-A

nthr

anili

c ac

idA

mitr

ole

Afla

toxi

n B

1B

enzy

l alc

ohol

Am

pici

llin

trih

ydra

teA

zido

thym

idin

et-

But

ylhy

droq

uino

net-

But

yl a

lcoh

olB

enzo

[a]p

yren

e (B

[a]P

)C

urcu

min

n-B

utyl

chl

orid

e C

adm

ium

chl

orid

e2,

4-D

ichl

orop

heno

l(2

-chl

oroe

thyl

)trim

ethy

l-NH

4Cl

Chl

oram

phen

icol

1,3-

Dih

ydro

xybe

nzen

eC

yclo

hexa

none

Cis

plat

inE

thio

nam

ide

N,N

-Dic

yclo

hexy

l thi

oure

aC

yclo

phos

pham

ide

Eth

yl a

cryl

ate

Die

than

olam

ine

2,4-

Dia

min

otol

uene

2-E

thyl

-1,3

-hex

aned

iol

Di-(

2-et

hylh

exyl

)pht

hala

te7,

12-D

MB

AE

ugen

olE

phid

rine

sulp

hate

Dim

ethy

lnitr

osam

ine

Isob

utyl

alde

hyde

Ery

thro

myc

in s

tear

ate

EN

U

D,L

-Men

thol

Flu

omet

ron

Eto

posi

dep-

Nitr

ophe

nol

Hex

achl

oroe

than

eH

ydro

quin

one

Pro

pyl g

alla

teD

-Lim

onen

eIQ

Pht

halic

anh

ydrid

eD

-Man

nito

lM

MS

Sod

ium

sac

char

inM

elam

ine

p-C

hlor

oani

line

Sod

ium

xyl

ene

sulfo

nate

M

ethy

l car

bam

ate

PhI

P.H

Cl

Sul

fisox

azol

eP

hena

nthr

ene

Sod

ium

ars

enite

Ure

aP

henf

orm

in H

CL

Taxo

lP

roge

ster

one

Pyr

idin

eTr

is(2

-eth

ylhe

xyl)p

hosp

hate

Tris

odiu

m E

DTA

trih

ydra

te

FIG

UR

E12

.4SchematicoverviewoftheresultsoftheECVAMcompoundlistintheAmestest,VitotoxTMtest,RadarScreentest,

invi

voregulatory

genotoxicitytests,

invi

tromammalianregulatorygenotoxicitytests,HCSIVMNCHO-K1assay,HCSIVMNHepG2assay,luciferase-basedreporterassays

withHepG2cellsforcystatinA,p53,RAD51CandNrf2,andtheGADD45a-GFPGreenScreenHCassay.Apositiveresultisindicatedinred,while

equivocalY/Nfindingisindicatedinyellow,noeffectingreen,andnodataavailableinwhite.

Genotoxic compounds

Test

sys

tem

Ames

Vito

tox

Rad

arsc

reen

In v

ivor

egul

ator

yA

In v

itro

mam

alia

n re

gula

tory

Cys

tatin

A_l

uc

p53_

luc_

luc

Rad

51C

_luc

Nrf

2_lu

c

Test

sys

tem

Ames

Vito

tox

Rad

arsc

reen

In v

ivo r

egul

ator

y

In v

itro

mam

alia

n re

gula

tory

Cys

tatin

A_l

uc

p53_

luc_

luc

Rad

51C

_luc

Nrf

2_lu

c

1-Nitropyrene 16α-Hydroxy-Estrone

2,4-Dinitrophenol 17α-Methyltestosterone

2,7-Dinitrofluorene Acetaminophen

*** see below Allylestrenol

2-Hydroxy-Estradiol Bromobenzene

2-Hydroxy-Estrone Canreonate K+

2-Methoxy-Estradiol Carbon tetrachloride

2-Methoxy-Estrone Chlormadinone

3-Methylcholantrene Chlormadinone acetate

4-Hydroxy-Estradiol Chlorpromazine

4-Hydroxy-Estrone Chlorprothixene citrate

4-Nitroquinoline Oxide Clomiphene

Dacarbazine Colchicine

Dantrolene Cyproterone acetate

Doxorubicin Cytarabine

Ellipticin Dexamethasone

Hydralazine Diclofenac

Hydrogen peroxide Diethylstilbestrol

β-Naphthoflavone Diethylthio carbamic acid

Melphalan Dihydroergotamine

Methampyrone Drospirenone

Nitrofurantoin Dydrogesterone

Org 2249 Equilin

Org 2408 Equilin-7α-Methyl

Org 2508 Estradiol-17αOrg 3240 Estradiol-17βOrg 4122 Estriol (E3)

Org 4330 Estrone (E1)

Org 5694 Ethinylestradiol-17β (EE)

Org 5695 Hexachlorobutadiene

Org 5697 Hydrochlorothiazide

Org 5710 Hydroxychloroquine sulfate

Org 5741 ICI 164.384

Org 5784 Imipramine HCl

Org 5796 Levonorgestrel

Org 5867 Methoxyprogesterone

Org 5907 Megestrol acetate

Org 7797 MENT-Bucyclate

Org 9063 Mestranol

Org 9150 Methotrexate

Org 9250 Moxestrol

Org 9252 Norethisterone

Org 20494 Norethynodrel

Org 32018 Noscapine HCl

Org 42671 Org 10325

Salicylamide Org 30029

Tacrine Org 30251

UK-57400 Org 39735

Uramustine Org 4433

5-Fluorouracil Oxymethalone

Rifampicin

***=2-amino-3-methyl-3H-imidazo-[4,5-f]-quinoline Rotenone

Stanozolol

Sulfinpyranone

Tamoxifen

Testosterone

Tolcapone

Trenbolone

FIGURE 12.5 Schematic overview of the results for the 108 in-house and reference com-pounds of the additional compound list from the Legacy Organon that are genotoxic in at leastone of the regulatory genotoxicity assays. Results are shown for the Ames test, VitotoxTM test,RadarScreen test, in vivo regulatory tests, in vitro mammalian regulatory genotoxicity tests,and luciferase-based reporter assays with HepG2 cells for cystatin A, p53, RAD51C, and Nrf2.A positive result is indicated in red, while equivocal Y/N finding is indicated in yellow, noeffect in green, and no data available in white. Chemical nomenclature of Org compounds isgiven in Table 12.7.

Non-genotoxic or no data available

Test

sys

tem

Ames

Vito

tox

Rad

ascr

een

C

ysta

tin A

_luc

p53_

luc_

luc

Rad

51C

_luc

Nrf

2_lu

c

Test

sys

tem

Ames

Vito

tox

Rad

arsc

reen

In v

ivo

regu

lato

ry

In v

itro

mam

alia

n re

gula

tory

In v

ivo

regu

lato

ry

In v

itro

mam

alia

n re

gula

tory

Cys

tatin

A_l

uc

p53_

luc_

luc

Rad

51C

_luc

Nrf

2_lu

c

2,5-Hexanedione Org 13011

7α-Methylnorethisterone Org 20091

Acetylsalicylic aci Od rg 20223

Aminophylline Org 20241

Amiodarone Org 20350

Antazoline mesylate Org 20660

Atamestane Org 30189

Atropine sulfat Oe rg 30535

Bishydroxycoumari On rg 30659

CERM 1188 O4 rg 30701

CERM 1306 O1 rg 31710

Clozapine Org 32608

Corticosteron Oe rg 32782

Cortisol Org 34037

Dehydroepiandrosterone Org 34694

Dimethisterone Org 34850

Dopamine Org 34901

Erythromycin Org 37445

Ethacrynic acid Org 42788

Ethionine Orphenadrine citrate

Ferrous sulphate Papaverine HCl

Fluoxymesterone - 17α-CH3 Perhexiline

Flutamide Perphenazine

Furafylline Phentolamine mesylate

Gentamycin A Propylmesterone

Indomethacin Quinidine

Iodoacetate Quinidine sulfate

Iproniazid R1881

Isoprenaline Raloxifen

Ketoconazole Reserpine

Labetalol RU 58668

L-DOPA Strychnine

Lilopristone +17α-CH3 Sulfamoxole

Methadone Sulfaphenazole

Naphazoline nitrate Tularik 0191317

N-ethylmaleimide Digoxin

Onapristone Org 30002

Org 3362 Org 9217

Org 4060 Org 9935

Org 4428

Org 4874

Org 5168

Org 7258

Org 9340

Org 10490

FIGURE 12.6 Schematic overview of the results for the 84 in-house and reference com-pounds from the additional compound list of the Legacy Organon that are nongenotoxic orhave no genotoxicity data in the regulatory genotoxicity tests. Results are shown for the Amestest, VitotoxTM test, RadarScreen test, in vivo regulatory tests, in vitro mammalian regulatorygenotoxicity tests, and luciferase-based reporter assays with HepG2 cells for cystatin A, p53,RAD51C, and Nrf2. A positive result is indicated in red, while equivocal Y/N finding is indi-cated in yellow, no effect in green, and no data available in white. Chemical nomenclature ofOrg compounds is given in Table 12.7.

Control valuesAbsolute DNA

quantity

Control valuesPhysiological

unwinding

Values for various

grades of damage

SybrGreen background

SybrGreen

SybrGreen

Neutr.

Neutr.

SybrGreen

Neutr.

SybrGreen

NaOH

NaOH

Lysis NaOH

NaOH

Neutr.

NoDamage

P0

DamagePx

HighDamage

B

Lysis

Lysis

NoDamage

T Lysis

FIGURE 14.1 Scheme of the principle of the FADU assay. The yellow boxes on the left sideof the figure represent the cells. In the middle the double-stranded DNA with increasing levelsof damage (gray boxes) and increasing extent of unwinding (blue boxes) is presented. The smallcircles (dark gray boxes) represent the fluorescent dye SybrGreen R© (yellow, no fluorescencesignal; green, fluorescence signal). T, P0, and B are controls to be run in parallel with theexperimentally treated cells. In P0 samples alkaline unwinding is allowed and represents theDNA strand breaks under physiological conditions (i.e., without exogenous DNA damage). InT samples the neutralization immediately follows the lysis and therefore, upon addition of thealkaline solution, unwinding cannot take place. T samples provide a measure of total DNAcontent and yield a fluorescence signal defined as 100%. In B samples the cellular DNA hascompletely been unwound as a result of very high damage level, and the resulting fluorescencerepresents the background. Px samples (P1, P2, P3, P4 . . . Px) are the experimental samples tobe analyzed and may represent any extent of damage between P0 and B levels.

FIGURE 15.1 Comet formation: (a) no damage; (b) low damage; (c) high damage.

(a) (b)

Living transformed cellsDead normal cellsTransformed cellsNormal cells

FIGURE 17.2 Selection of transformed Bhas 42 cells by treatment with hydrogen peroxide.(See text for full caption.)

0.1% DMSO 1 μg/mL MCA 0.1% DMSO 1 μg/mL MCA

alamarBlue

WST-8

H2O2＋ Dyes Fixation and Giemsa staining

FIGURE 17.7 Concordance between wells colored by dyes and wells with transformed fociin the same plates. (See text for full caption.)

9080706050403020101 1 60

5000

10000

15000

20000

25000

Wells

Fluo

resc

ence

0.1% DMSO

241017260504030201010

5000

10000

15000

20000

25000

Wells

Fluo

resc

ence

0.1 μg/mL MCA

40302010156504030201010

5000

10000

15000

20000

25000

Wells

Fluo

resc

ence

0.3 μg/mL MCA

56504030201014030201010

5000

10000

15000

20000

25000

Wells

Fluo

resc

ence

1 μg/mL MCA

FIGURE 17.8 Dose-dependent increase of alamarBlue fluorescence values correlates withinduction of transformed foci by treatment with MCA in Bhas 42 cells. (See text for fullcaption.)

FIGURE 21.1 (a) Scheme of single cell imaging cytometry. (See text for full caption.)

FIGURE 21.3 HCS of drug-induced changes in the permeability of the potassium channeland the concentration of intracellular sodium ion in H9c2(2-1) cells. (See text for full caption.)

Sodium ion

Superimposition

Superimposition

Apoptosis

Necrosis

579nm

AOTF(Multicolor

imaging system)

FIGURE 21.4 Combined use of assays to determine cell death and the concentration ofsodium ion in cardiomyocytes. The fluorescence spots of the sodium ion completely overlapwith those of necrotic cells. The figure is reproduced from Reference 13 with permission fromthe Royal Society of Chemistry.

-100

(d)

0 100 200 300 400 500

0

CTRL

300 pM

1 nM

3 nM

10 nM

30 nM

100 nM

300 nM

1 μM

3 μM

10 μM

30 μM

μV

t/ms

FIGURE 22.2 (d) Recordings of the field potential for a drug at 11 different concentrationsusing the MEA device (reproduced with permission from Multi Channel Systems GmbH).

(a)

(b)

(c)

(d)

(e)

FIGURE 22.3 Field potential recordings with the MEA and the hESC-derived cardiomy-ocyte cell line hES-CMC 002. MEA data for: (a) lidocaine, (b) quinidine, (c) nifedipine, (d)amiloride and (e) acetaminophen. The open dots indicate the relative field potential durationbased on the scale on the right side. The filled dots represent the relative block of depolarizationpeak indicated by the axis on the left side of the diagrams; 100% refers to control conditionsin the absence of the drug. Overlay plots in the left panel illustrate modulation of the signalshape over increasing compound concentrations as indicated by the color code (reproducedfrom Reference 27 with permission from John Wiley & Sons, Ltd.).

(a)

(b)

(c)

(d)

(e)

(f)

FIG

UR

E22

.4FieldpotentialrecordingswiththeMEAandthehumanembryonicstem

cell-derivedcardiomyocytecelllineHES03from

the

UniversityofLeiden(Leiden,TheNetherlands).(a)Sodium

peakamplitude–doserelationshipforhESC-CMinthepresenceofincreasingamountsof

lidocaine.(b)Exampleoffieldpotentialrawdata;notethereductionofamplitudeathigherconcentrationsoflidocaine.(c)FPD–doserelationshipfor

hESC-CMinthepresenceofincreasingamountsofnifedipine.(d)Exampleoffieldpotentialrawdata;notethereductionoffieldpotentialduration

athigherconcentrationsofnifedipine.(e)FPD–doserelationshipforhESC-CMinthepresenceofincreasingamountsofE-4031.(f)Exampleof

fieldpotentialrawdata;notetheprolongationoffieldpotentialdurationinresponsetoE-4031(reproducedfrom

Reference26withpermissionfrom

Elsevier).

FIGURE 24.9 Tamoxifen (TAM) results: (a) LC chromatogramwith UV detection (blue) forstandards: a, �-OHTAM; b, 4-OHTAM; c, N-des TAM; d, TAM; e, TAMN-oxide. Microsome-DNA biocolloid reactions are presented in red (RLM) and green (HLM) (×3) for 1 min reac-tion of 25 �M tamoxifen using NADPH regenerating system. (b) Multiple reaction monitoring(MRM) chromatograms for formation of �-(N2-deoxyguanosinyl)tamoxifen using RLM withmass transition from 319 to 261. Chromatogram 1 obtained from control with 25 �M tamox-ifen only. Chromatograms 2, 3, and 4 represent reactions for 1, 2, and 3 min, respectively.Inset is peak area vs. time, normalized as peak area/wt. RLM (red) and peak area/wt. HLM(blue). (c) Daughter ions of 319, with fragmentation pattern illustrated in the inset. (d) MRMchromatogram (m/z 564 protonated �-OHTAM glucuronide to m/z 370 tamoxifen) confirmingformation of �-OHTAM O-glucuronide using HLM/DNA films after 1 min reaction. Adaptedwith permission from Reference 40 (copyright 2009 American Chemical Society).

(a)

(b)

(d)

(c)

FIGURE 24.10 Features of the bioanalytical system for high-throughputmetabolic profiling.(a) Bioreactor assembly: a layer of the cationic polymer polydiallyldimethylammonium chlo-ride (PDDA) was initially deposited on silica nanoparticles, followed by a layer of oppositelycharged microsomes. (b) A reaction/filtration 96-well plate equipped with 10,000 Da-cutoffmass filters showing the liquid chromatography-mass spectrometry (LC–MS)-ready sample-collection plate underneath. (c) Schematic illustration of simultaneous enzyme reaction designfeaturing a 96-well plate. (d) LC–MS/MS analysis with an autosampler. Abbreviations: HLM,human liver microsome; RLM, rat liver microsome.

Date post:	17-Sep-2018
Category:	Documents
Upload:	lekhanh
View:	225 times
Download:	0 times