Controlling Cellular Behavior with Synthetic Biology

ETH Library

Receptors, Small Moleculesand Physical Cues - ControllingCellular Behavior with SyntheticBiology

Doctoral Thesis

Author(s):Strittmatter, Tobias

Publication date:2021

Permanent link:https://doi.org/10.3929/ethz-b-000493914

Rights / license:In Copyright - Non-Commercial Use Permitted

This page was generated automatically upon download from the ETH Zurich Research Collection.For more information, please consult the Terms of use.

https://doi.org/10.3929/ethz-b-000493914

http://rightsstatements.org/page/InC-NC/1.0/

https://www.research-collection.ethz.ch

https://www.research-collection.ethz.ch/terms-of-use

DISS. ETH NO. 27358

RECEPTORS, SMALL MOLECULES AND PHYSICAL CUES - CONTROLLING CELLULAR BEHAVIOR WITH

SYNTHETIC BIOLOGY

A thesis submitted to attain the degree of

DOCTOR OF SCIENCES of ETH ZURICH

(Dr. sc. ETH Zurich)

presented by

TOBIAS STRITTMATTER

Dipl. Biochem., Eberhard Karls Universität Tübingen

born on 20.08.1987

citizen of Germany

accepted on the recommendation of

Prof. Dr. Martin Fussenegger

Prof. Dr. Randall Platt

Prof. Dr. Ivan Martin

Prof. Dr. Renato Paro

2021

Tobias Strittmatter i

Table of Contents:

Abstract ........................................................................................................................................................ 1

Zusammenfassung ........................................................................................................................................ 3

Introduction .................................................................................................................................................. 5 Concepts of Synthetic Biology ........................................................................................................................... 5 The Synthetic Biology Toolbox ........................................................................................................................... 6 Applications of Synthetic Biology in Biotechnology ........................................................................................... 9 Synthetic Biology-Enabled (Future) Medicine ................................................................................................. 10 Contributions of this work ............................................................................................................................... 13

Chapter 1: Programmable DARPin-based Receptors for the Detection and Treatment of Thrombotic Events15 Author Contributions ....................................................................................................................................... 15 Contributions to each figure ............................................................................................................................ 15 Author Affiliations ........................................................................................................................................... 15 Abstract ........................................................................................................................................................... 17 Introduction ..................................................................................................................................................... 19 Results ............................................................................................................................................................. 23

General receptor design strategy ............................................................................................................... 23 DARPin selection ........................................................................................................................................ 25 Designing AMBERs in homodimeric, heterodimeric and tandem configurations ...................................... 26 AMBER screening setup ............................................................................................................................. 27 Control experiments show no unspecific induction ................................................................................... 28 AMBER screening ....................................................................................................................................... 28 Characterization of candidate AMBERs ...................................................................................................... 31 Generating monomeric tandem AMBERs .................................................................................................. 32 Linking fibrinogen activation to induction .................................................................................................. 34 D-dimer- and fragment E-mediated activity ............................................................................................... 36 Generating mammalian designer cells for the detection of coagulation ................................................... 39 Preventing coagulation with hirudin-producing designer cells .................................................................. 41

Discussion ........................................................................................................................................................ 43 Acknowledgements ......................................................................................................................................... 45 Materials & Methods ...................................................................................................................................... 47

Chemicals and proteins .............................................................................................................................. 47 Selection and screening of DARPin binders specific for FDP ...................................................................... 47 Sequence analysis ...................................................................................................................................... 48 Plasmid preparation ................................................................................................................................... 48 SDS-PAGE and Coomassie staining ............................................................................................................. 48 Cell culture ................................................................................................................................................. 49 Transfections and stable cell line generation ............................................................................................. 49 Reporter Assays .......................................................................................................................................... 50 Size-exclusion chromatography of reconstituted plasma .......................................................................... 51 Mass spectrometry ..................................................................................................................................... 51 tPA activity measurements ........................................................................................................................ 51 Statistics ..................................................................................................................................................... 51

Tobias Strittmatter ii

Supplementary Information ............................................................................................................................ 53 Table S1: Plasmids used in this work .......................................................................................................... 53 Table S2: Plasmids transfected in each experiment ................................................................................... 61

Supplementary Figures .................................................................................................................................... 69 References ....................................................................................................................................................... 93

Chapter 2: Gene switch for L-glucose-induced biopharmaceutical production in mammalian cells ............... 99 Author Contributions ....................................................................................................................................... 99 Contributions to each figure ............................................................................................................................ 99 Author Affiliations ........................................................................................................................................... 99 Abstract ......................................................................................................................................................... 101 Introduction ................................................................................................................................................... 103 Results ........................................................................................................................................................... 107

Design of an L-glucose-responsive gene circuit ........................................................................................ 107 Testing of metabolic enzymes .................................................................................................................. 110 Functionality of ON- vs OFF-switch .......................................................................................................... 112 L-Glucose-induced rituximab production ................................................................................................. 112

Discussion ...................................................................................................................................................... 119 Acknowledgements ....................................................................................................................................... 121 Materials & Methods .................................................................................................................................... 123

Cell culture ............................................................................................................................................... 123 Plasmid generation and transfection ....................................................................................................... 123 Generation of stably expressing cell lines ................................................................................................ 124 Induction with L-glucose .......................................................................................................................... 124 Assessment of reporter gene expression ................................................................................................. 125 Measurement of rituximab production .................................................................................................... 125 Analysis of glycosylation patterns ............................................................................................................ 126 Statistical Analysis .................................................................................................................................... 127

Supplementary Information .......................................................................................................................... 129 Table S1: Plasmid Information ................................................................................................................. 129

References ..................................................................................................................................................... 133

Chapter 3: Controlling gene expression through mechanical cues ............................................................... 137 Author Contributions ..................................................................................................................................... 137 Contributions to each figure .......................................................................................................................... 137 Author Affiliations ......................................................................................................................................... 137 Abstract ......................................................................................................................................................... 139 Introduction ................................................................................................................................................... 141 Results ........................................................................................................................................................... 143

Endogenous and engineered depolarization activated gene switches .................................................... 143 Amplifiers of calcium response ................................................................................................................ 144 Population-based induction of shear stress through turbulent flow ....................................................... 148

Discussion ...................................................................................................................................................... 155 Materials & Methods .................................................................................................................................... 157

Cell culture ............................................................................................................................................... 157 Plasmid generation and transfection ....................................................................................................... 157

Tobias Strittmatter iii

Stimulation of shear stress using the piston device ................................................................................. 157 Generation of stable cell lines .................................................................................................................. 158 Assessment of reporter gene expression ................................................................................................. 158 Statistical Analysis .................................................................................................................................... 159

Supplementary Information .......................................................................................................................... 161 Table S1: Plasmid Information ................................................................................................................. 161 Table S2: Plasmids transfected in each experiment ................................................................................. 167

References ..................................................................................................................................................... 171

Chapter 4: Mechano-Regulation of Wnt/β-catenin Signaling to Control Paraxial Versus Lateral Mesoderm Lineage Bifurcation .................................................................................................................................... 177

Author Contributions ..................................................................................................................................... 177 Contributions to each figure .......................................................................................................................... 177 Author Affiliations ......................................................................................................................................... 177 Abstract ......................................................................................................................................................... 179 Introduction ................................................................................................................................................... 181 Results ........................................................................................................................................................... 185 Discussion ...................................................................................................................................................... 191 Materials and Methods ................................................................................................................................. 193

Plasmid preparation ................................................................................................................................. 193 Cell culture ............................................................................................................................................... 193 Directed differentiation of hiPSCs ............................................................................................................ 194 Mechano-regulation of Wnt/β-catenin signaling ..................................................................................... 195 Analytical assays ....................................................................................................................................... 195 Gene expression profiling by RT-qPCR ..................................................................................................... 195 Statistical analyses .................................................................................................................................... 196

Supplementary Information .......................................................................................................................... 197 Table S1: Plasmid Information ................................................................................................................. 197 Table S2: List of primers for qRT-PCR ....................................................................................................... 199

Supplementary Figures .................................................................................................................................. 200 References ..................................................................................................................................................... 203

Discussion .................................................................................................................................................. 207 New receptors to expand the range of targetable disease ........................................................................... 207 Controlling cellular behavior with small molecules ....................................................................................... 210 Mechanical cues to control gene expression and lineage commitment ........................................................ 212 Developments and future applications of mammalian synthetic biology ..................................................... 214

Conclusion ................................................................................................................................................. 219

Additional References ................................................................................................................................ 221

Acknowledgements .................................................................................................................................... 233

Curriculum Vitae ................................................................................................. Error! Bookmark not defined.

Tobias Strittmatter iv

Abstract

Tobias Strittmatter 1/240

Abstract Synthetic biology applies an engineering mindset to the classical field of biology in

order to facilitate the design of new proteins and cellular systems. Synthetic biology-inspired

solutions and products are already a part of modern-day biotechnology and medicine.

Drawing from genetic components from all kingdoms of life, engineers are able to design

sophisticated genetic circuits that respond to various physical and chemical inputs

One key motivation of synthetic biological engineering is the creation of tools to

control cellular behavior typically through genetic circuits. However, the generation of gene

circuits that respond to extracellular soluble ligands remains limited by the availability of

suitable receptors. Although there is a vast number of endogenous receptors at hand, some

proteins, especially disease markers, lack a native sensor module. To tackle this problem our

group designed a novel receptor platform based on the established erythropoietin receptor

scaffold that can accommodate various binding domains to detect small molecules as well as

larger proteins. We took this idea further by employing a high-throughput selection for the

generation of designed ankyrin repeat proteins (DARPins) to detect an even broader range of

ligands in a streamlined fashion as described in chapter 1. We tested this pipeline by creating

receptors that can detect sub-pathological concentrations of fibrin degradation products

arising from e.g., thrombotic events.

In order to detect a ligand without known receptor it is not necessarily the receptor

that needs to be engineered – instead, the ligand itself can be modified to enable its

detection. We show in chapter 2 how we designed an inducible gene switch that responds to

the inert-to-human small molecule L-glucose. Here, we harnessed a metabolic process found

in the soil bacterium Paraccocus species 43P to convert L-glucose into the active ligand of the

bacterial transcriptional regulator LgnR. By fusing transcriptional modulator domains to LgnR,

we could control gene expression from a synthetic promoter in an L-glucose responsive

manner to enable production of the model biopharmaceutical rituximab.

Physical stimuli like mechanical cues are omnipresent in our surroundings and aid cells

to make informed decisions on propagation, tissue organization and cell fate. In chapter 3 we

present results from our work on implementing a mechanically triggered gene switch in

mammalian cells. With the help of a custom-made shear stress induction device we screened

a small library of genetic circuits, transcription factors as well as accessory channels and

Abstract


receptors for their potential to improve shear stress induced calcium responses. This

approach allowed us to identify key components that sensitize cells to mechanical cues to

enable the exogenous control of gene expression.

In chapter 4 we used the same device to study the effects of shear stress on lineage

commitment of human induced pluripotent stem cells (hiPSCs) into the three germ layers. We

were able to show that commitment to the ectoderm and mesoderm lineages was sensitive

to shear stress through modulation of the Wnt/β-catenin signaling pathway while endoderm

was not affected. More detailed analysis revealed that the effect is reversible and can be used

to modulate gene expression levels and lineage commitment even in further matured cells of

the mesoderm lineage.

Zusammenfassung


Zusammenfassung Synthetische Biologie wendet die Denkweise der Ingenieurswissenschaften auf das Gebiet der

klassischen Biologie an, um die Entwicklung von neuen Proteinen und Zellsystemen zu

vereinfachen. Durch synthetische Biologie inspirierte Lösungsansätze und Produkte sind

bereits Teil moderner Biotechnologie und Medizin. Durch die Verwendung von genetischen

Komponenten verschiedenster Spezies können aufwändige genetische Schaltkreise

konstruiert werden, die auf unterschiedliche physikalische und chemische Eingaben

reagieren. Eines der wichtigsten Ziele der Synthetischen Biologie ist das Entwickeln von

Werkzeugen, um das Zellverhalten mittels genetischer Schaltkreise zu kontrollieren. Die

Entwicklung solcher Schaltkreise, die auf extrazelluläre Liganden reagieren können, wird

durch die Verfügbarkeit geeigneter Rezeptoren begrenzt. Obwohl es eine grosse Zahl

endogener Rezeptoren gibt, fehlt zu einigen Proteinen, speziell zu krankheitsrelevanten, ein

natürliches Sensormodul. Um dieses Problem zu lösen, hatte unsere Gruppe bereits eine

neuartige Rezeptorplattform etabliert die auf dem Gerüst des Erythropoietinrezeptors

basiert. Diese Plattform erlaubt es mit Hilfe verschiedener austauschbarer Bindungsdomänen

gleichermassen kleine Moleküle wie auch grosse Proteine zu erkennen. Wie in Kapitel 1

beschrieben entwickelten wir die Idee weiter indem wir eine Hochdurchsatz-

Selektionsmethode zur Generierung von künstlichen repetitiven Ankyrinproteinen (designed

ankyrin repeat proteins; DARPins) verwendeten, um die Bandbreite an möglichen Liganden

auf effiziente Art und Weise noch zu vergrössern. Wir testeten unsere Pipeline durch die

Konstruktion von Rezeptoren, welche Abbauprodukte des Fibrins, die in der Folge von

thrombotischen Ereignissen auftreten, bereits vor dem Erreichen des pathologischen

Grenzwerts erkennen können.

Wie wir in Kapitel zwei darlegen, muss man jedoch nicht immer den Rezeptor

verändern, um einen Liganden erkennen zu können, der keinen bekannten Rezeptor hat,

sondern kann auch den Liganden an den Rezeptor anpassen. Hierzu verwendeten wir einen

metabolischen Prozess aus dem Bodenbakterium Paraccocus species 43P, um den für

Menschen inerten Stoff L-Glucose in einen aktiven Liganden für das bakterielle

Regulatorprotein LgnR umzuwandeln. Indem wir LgnR mit Domänen zur Modulation der

Transkription verbanden, konnten wir die Expression eines synthetischen Promoters L-

Zusammenfassung


Glucose-abhängig kontrollieren, um die Produktion von Rituximab als Modell für ein

biopharmazeutisches Produkt, zu steuern.

Physikalische Reize wie zum Beispiel mechanische Signale sind omnipräsent in unserer

Umgebung und helfen Zellen informierte Entscheidungen bezüglich Zellteilung,

Gewebeorganisation und Zellentwicklung zu treffen. In Kapitel drei präsentieren wir die

Ergebnisse unserer Arbeit zur Umsetzung eines mechanisch aktivierten genetischen Schalters

in Säugerzellen. Mit Hilfe eines speziell angepassten Scherstressgenerators waren wir in der

Lage, einen Pool an genetischen Schaltkreisen, Transkriptionsfaktoren sowie zusätzlichen

Ionenkanälen und Rezeptoren auf ihren Nutzen zur Verbesserung der Signalqualität hin zu

untersuchen.

In Kapitel vier verwendeten wir denselben Scherstressgenerator, um Auswirkungen

von Scherstress auf das Differenzierungsverhalten von menschlichen induzierten

Stammzellen (human pluripotent stem cells; hiPSCs) in die drei Keimblätter zu studieren. Wir

konnten zeigen, dass die Festlegung auf Ektoderm und Mesoderm durch Scherstress

beeinflusst wird, wohingegen die Entwicklung hin zu Endoderm nicht betroffen ist. In einer

detaillierteren Untersuchung fanden wir heraus, dass der Effekt abhängig vom

Induktionszeitpunkt sogar die Entwicklung von Zellen des lateralen Mesoderms oder des

paraxialen Mesoderms begünstigt.

Introduction


Introduction This work aims to expand on the available tools of synthetic biology to control cellular

behavior by engineered receptors, conversion of small-molecule inducers through metabolic

pathways as well as through mechanical cues. This introduction provides an overview of the

underlying design principles of synthetic biology-inspired engineering citing examples of how

synthetic biology is used in industrial biotechnology as well as current and future applications

in medicine.

References included in this introduction, as well as the closing discussion and

conclusion sections can be found at after the conclusion at the end of this work.

Concepts of Synthetic Biology

Synthetic biology applies concepts of classical engineering to the field of biology by

describing functional biological entities as independent modules. By recombining these

modules, synthetic biologists create novel functions in RNAs (Auslander et al. 2014; Auslander

et al. 2016), proteins ( Scheller et al. 2018; Chassin et al. 2019) and whole cells (Gitzinger et

al. 2012).

The modular approach on biological engineering draws from discoveries in protein

biochemistry that established protein domains as independently folding entities with distinct

functions. Domains that are of particular interest for synthetic biology applications are

signaling components as well as metabolically active domains. Bacterial signaling pathways

can generally be very simple, compared to sophisticated interdependent phosphorylation

cascades that can be found in mammalian systems. Instead, prototypic bacterial gene circuits

are based on induced binding or dissociation of transcriptional regulators to the target

sequence on the DNA. Therefore, chimeric transcription factors comprising fusion proteins of

bacterial regulators and mammalian transactivator domains have become a mainstay in

biological research and bioengineering. The most prominent implementation of such a

chimeric transcription factor is the tetracycline transactivator (tTA) (Gossen et al. 1992),

which still is the current gold standard for orthogonal gene expression systems.

The Synthetic Biology Toolbox



The modules that are used for synthetic biology-inspired engineering are often

referred to as the synthetic biology toolbox. This toolbox comprises items from all levels of

cellular organization in the form of protein domains, DNA sequences and RNA elements which

can be rationally assembled to generate new functions. For example, structural

rearrangements of protein domains allow for the creation of novel protein scaffolds used in

binder development for biopharmaceutical products. Cell-based applications, in turns, often

need signaling functions that are not naturally available but that can be engineered by

recombining existing modules to form novel regulatory gene networks.

Figure I: The mammalian synthetic biology toolbox consists of receptors, signal mediators

(NFAT1c) and promoters from all kingdoms of life that enable triggering or rewiring of

endogenous as well as orthogonal signaling pathways.

Such gene networks can be designed on the basis of native or engineered receptors

and channels residing in the plasma membrane that integrate the signal transduction events

through endogenous or orthogonal signaling cascades (figure I). There is a multitude of native

receptors to be found in all kingdoms of life. However, it can prove to be challenging to

integrate transmembrane proteins from bacteria, fungi or plants into mammalian systems

due to their specific requirements on the local membrane environment. Nevertheless,

synthetic biology toolbox

RE RERE promoter GOI

receptors, e.g.- catenines- channels- RTKs- GPCRs- TCRs

signal mediators- signaling cascades- synthet ic transcript ion factors- dCas9

promoter- endogenous- synthet ic

Introduction


mammalian cells are already equipped with specific receptors that can sense a diverse set of

inputs, including temperature (Bai et al. 2019; Zhang et al. 2018), ion concentrations (Yoast

et al. 2020), membrane potential (Helton et al. 2005; Kerstein et al. 2013) or mechanical stress

(Nonomura et al. 2018; Saotome et al. 2018). Additionally, the largest class of cellular sensors,

the family of G-protein-coupled receptors (GPCRs) contains members that respond to small

molecules (Muller et al. 2017) as well as light (Mansouri et al. 2019).

Artificial membrane receptors can be designed with customizable in- and outputs by

equipping existing membrane receptor scaffolds with new binding moieties to trigger

endogenous (Scheller et al. 2018; Kuwana et al. 1987) or orthogonal signaling cascades

(Schwarz et al. 2017). Native receptors are heavily evolved to excel at their respective task in

cell signaling. The sometimes sophisticated, fine-tuned conformational changes that lead to

receptor activation hamper attempts to engineer new functionalities into the existing

receptor scaffold. The class of dimerization dependent receptors like the erythropoietin

receptor (EpoR) however employs a rather simple mechanism of activation. EpoR is expressed

as a dimer of two identical receptor chains. The receptor is activated by binding of its ligand

erythropoietin to the extracellular domains of each receptor chain, which induces

dimerization and a conformational change that translates into a rotation of the intracellular

signaling domains towards each other. Through this turning motion, the associated Janus

kinases (JAKs) move into close proximity leading to mutual phosphorylation of the kinases as

well as of the signaling domains of the receptor chains which triggers downstream STAT

signaling (Watowich, 2011). Because of this comparably simple mechanism of activation,

EpoR has been the target for engineering artificial signaling pathways already.

In a recent study, we were able to demonstrate a generalized extracellular molecule

sensor (GEMS) platform for programming cellular behavior based on EpoR (Scheller et al.

2018). This platform allows for the modular addition and switching of extracellular ligand-

binding domains as well as intracellular signal transduction domains enabling the detection

of a selection of diverse soluble ligands ranging from small molecules to protein cancer

markers that could be used to trigger different signaling cascades.

Native GPCRs signal through binding of small G-proteins that are specific for each

pathway and receptor. In an attempt to alter the signaling from a given GPCR, chimeric G-

proteins have been created that still bind to the respective GPCR but trigger an alternative

pathway (Coward et al. 1999). GPCRs were also engineered through the more cumbersome



method of directed evolution to alter their binding specificity to form receptors activated

solely by synthetic ligands (RASSLs) (Redfern et al. 1999; Armbruster et al. 2007; Conklin et al.

2008).

A more flexible rewiring of endogenous signaling cascades via a modular approach

was described recently (Krawczyk et al. 2020). Here, a nuclease deficient Cas9 was targeted

to endogenous promoters on the genome via specific guide RNAs (gRNAs). Activation of gene

expression was mediated by the interaction of a bacteriophage MS2 coat protein (MCP)

moiety with an MS2-binding RNA loop in the gRNA (figure I). By fusing the MCP to a pathway-

specific regulatory protein, this system can be used to either activate specific pathways or

rewire input from one pathway to the expression of an unrelated target. In addition to this

modular rewiring of endogenous signaling pathways, dCas9 can also be engineered to

respond to light, temperature changes as well as small molecules, constituting a highly flexible

architecture for the design of artificial transcription factors (Richter et al. 2017).

Alternative approaches tackle detection of small molecules from a different angle by

avoiding membrane receptors and rely solely on cytosolic modules instead. To this end,

different approaches have been employed, including light-sensitive dimerization-dependent

gene switches (Levskaya et al. 2009), small molecule-dependent gene switches based on

bacterial regulators (Gitzinger et al. 2009; Wang et al. 2015) or endogenous temperature-

sensitive promoter elements (Miller et al. 2018).

Besides controlling the expression of a given gene, the resulting transcript can also be

engineered to tune mRNA concentrations (Auslander et al. 2016), localize the expression to

predefined regions of the cell (Berkovits et al. 2015), allow for the interaction with specific

RNA-binding proteins like L7Ae (Auslander et al. 2014) or even to sense temperature

(Giuliodori et al. 2010).

On another note, the modular architecture of proteins can not only be exploited for

the design of epitope-binding proteins like designed ankyrin repeat proteins (DARPins)

(Schilling et al. 2014) and antibodies (Liu et al. 2017) but also allows to directly control protein

concentrations inside the cell by addition of destabilizing domains, so-called degrons, that

trigger proteasomal degradation (Chassin et al. 2019).

Cells themselves can be regarded as the biggest available module in synthetic biology

as they can also serve as signaling entities within higher-order cellular systems. Such systems

employ communication between different cell populations via the common metabolite

Introduction


tryptophan (Bacchus et al. 2012), small molecules dopamine and histamine (Auslander et al.

2018) or by direct surface contact through the modular syn-notch receptor platform (Morsut

et al. 2016). Especially the syn-notch platform is nowadays considered a promising tool to

alter and study tissue generation in genetically modified stem cells. With the help of cell-cell

contact-mediated signaling, researchers aim to recapitulate the in vivo situation where cells

receive multiple inputs from their neighbors within a specific developmental niche.

Applications of Synthetic Biology in Biotechnology

Synthetic biology is present on many levels in biotechnological production guiding

protein design as well as genetic modifications of the production cell line. For example,

engineering of antibodies is feasible through their modular structure that enables the

generation of truncated versions comprising only domains essential for epitope binding (Liu

et al. 2017). Fusion of different binding domains for the simultaneous targeting of multiple

epitopes (Cuesta et al. 2010) or addition of T-cell recruiting affinity domains (Lejeune et al.

2020) are further examples of synthetic biology-inspired engineering of biopharmaceuticals.

With the rise of gene therapy, adeno associated viruses (AAVs) are getting broader

attention as safe DNA delivery vehicles in vivo. By engineering capsid proteins of the AAV hull,

scientists are now able to tune the specificity of AAVs for several tissues and cell types (Li et

al. 2020). Novel holistic screening approaches covering all steps from DNA modification, AAV

production and assessment of tissue specificity in live animals are poised to speed up this

process even more, enabling the design of targeted next-generation designer drugs for gene

therapy (Weinmann et al. 2020).

Inducible gene expression is commonly used in microbial fermentations for small to

large scale production of peptides and proteins in lab and industry settings (Huang et al.

2014). Large scale production can be divided into two phases: (i) the growth phase where the

cells are grown to a critical density before (ii) proliferation is limited and the cells enter a

stationary phase. Induced protein production offers a way to time biopharmaceutical

production to the second phase to reduce stress on the culture during the growth phase.

Thereby excess production of harmful metabolites which would lead to a decrease in product

quality can be avoided. In mammalian systems those issues have been tackled by traditional

process optimization rather than genetic engineering. However, inducible biopharmaceutical

Synthetic Biology-Enabled (Future) Medicine


production is slowly making its way into mammalian bioreactors for the production of hard-

to-express proteins that show sophisticated post-translational modifications (Kallunki et al.

2019).

Besides the integration of genetic circuits to respond to exogenous triggers for timed

expression of a transgene - which is still in its infancy for mammalian systems in industrial

applications - researchers have early on tried to improve the productivity of the host cell

through means of genetic engineering. To this end, expression of metabolic enzymes is either

reduced or shut down to limit excess production of metabolites that pose a danger to the

host cell or the product in order to improve overall productivity of the cell line. A prime

example of metabolic engineering is the ongoing effort to reduce overflow metabolism and

lactate production in Chinese hamster ovary (CHO) cells by genetically tuning the activity of

glucose metabolism (Kim et al. 2007; Toussaint er al, 2016; Zhou et al. 2011). Specific post-

translational modifications in the cell can be achieved by ectopic expression of additional

enzymes to either expand the range of products that can be expressed or to reduce unwanted

potentially immunogenic modifications instead (Amann et al. 2019).

A complementary route for increased cellular productivity is offered by protocols to

increase the copy number of the transgene in the genome of stable cell lines. Increased copy

number can be achieved by selection protocols employing glutamine synthetase (GS)-

mediated gene amplification (Fan et al. 2012; Yu et al. 2018). Here, GS-deficient cells can be

adapted for higher expression of a GS-coupled transgene by adding increasing concentrations

of the GS-inhibitor methionine sulfoximine (MSX).


Treatments based on small molecules and antibodies are still the backbone of

modern-day medicine. However, with the approval of engineered T-cells equipped with

chimeric antigen receptors (CAR-T cells), which were developed already in the last century

(Kuwana et al. 1987), the field of cellular therapies saw its first application outside clinical

trials in 2017. In light of these advances, the application of cellular therapies in patients

becomes more likely. There are already multiple potential treatments based on human

designer cells for the treatment of psoriasis (Schukur et al. 2015) or gouty arthritis (Kemmer

et al. 2010). Equipped with specific receptors to sense disease-related markers and

Introduction


metabolites in patients’ blood, engineered human cells can take over lost body functions to

counteract disease symptoms and potentially restore the quality of life for people with e.g.

diabetes (Scheller et al. 2019; Teixeira et al. 2019) (figure II).

Figure II: Therapeutic designer cells allow the detection of disease markers, integrate the

input and respond in a predefined manner. Amongst others, therapeutic designer cells have

been described that sense and respond to elevated blood glucose or fibrin degradation

products.

Inspired by electrical engineering, synthetic biology recombines interdependent

signaling pathways and transcription factors for the implementation of logic gates in living

cells (Auslander et al. 2018). By doing so, a cell can integrate multiple inputs to cater to the

more complex environment of potential disease niches. For example, logic AND gates based

on gene circuits as well as natural signal-integrating cytokine receptors for the simultaneous

detection of two inflammatory blood markers have been demonstrated to be useful to fight

auto-inflammatory diseases (Schukur et al. 2015; Chassin et al. 2017).

In the new field of electrogenetics, the connection between electrical engineering and

synthetic biology is taken to a new level. A one-way electrical interface was developed to

trigger a prosthetic gene network inside a cell implant for the remote-controlled release of

insulin in diabetic mice (Krawczyk et al. 2020). Besides electricity, other physical stimuli can

be used as well as there are applications for gene switches triggered by light (Mansouri et al.

2019), temperature (Bai et al. 2019; Miller et al. 2018; Vilaboa et al. 2005) and even

ultrasound (Y. Pan et al. 2018).

In conclusion there are currently numerous therapeutic applications in the making

that are directly drawing from synthetic biology-inspired engineering. The designer cells used

disease markers- blood glucose- coagulation products- ...

designer cell therapeutic output- insulin- fibrinolyt ics- ...

detect process respond



in these therapies are envisioned to enable autonomous treatment of disease in patients and

– once implanted – would require little or no intervention (Kitada et al. 2018).

Introduction


Contributions of this work

The work presented here aims to improve our understanding of mammalian synthetic

biology by adding new modules to the synthetic biology toolbox to control cellular behavior.

Chapter 1 comprises a protocol for the acceleration of receptor development against targets

without a known natural receptor that is based on the well-characterized erythropoietin

receptor. We also describe how an inaccessible small molecule can be harnessed for induction

of gene expression by employing a metabolic process to convert it into an active ligand of a

bacteria-derived artificial transcription factor (chapter 2).

We furthermore present our work on how mechanical cues can potentially be used to

control gene expression in a non-invasive, ligand-free manner in chapters 3 and 4. We show

how mechanical stimuli can alter calcium signaling and downstream gene expression (chapter

3) as well as how shear stress can change morphogen signaling in induced pluripotent stem

cells (iPSCs) to modulate lineage commitment (chapter 4).

Contributions of this work


Chapter 1: Programmable DARPin-based Receptors for the Detection and Treatment of Thrombotic Events


Chapter 1: Programmable DARPin-based Receptors for the Detection

and Treatment of Thrombotic Events

Chapter 1 describes a work in progress that has not yet been submitted for publication.

Author Contributions

Tobias Strittmatter1, Leo Scheller2 and Martin Fussenegger1,6 designed the project together

with Jonas V. Schaefer3 and Andreas Plueckthun4. Tobias Strittmatter1, Pascal Stuecheli1,

Patrick C. Freitag3, Jonas V. Schaefer4, Thomas Reinberg3 and Dimitrios Tsakiris5 performed

experiments and analyzed the results. Leo Scheller2 performed initial experiments that are

not covered in this manuscript and provided plasmids.

Contributions to each figure

Tobias Strittmatter1: figures 1a-c, 2, 3, 4a-e, S1-6, S8-14

Jonas V. Schaefer3, Thomas Reinberg4 and Andreas Plueckthun4: figures 1d, S3

Patrick C. Freitag4: prepared binders for figure S7

Dimitrios Tsakiris5: figure 4f

Jens Sobek4: performed SPR measurements in figure S7

Steven Schmitt1: assisted with HPLC in figure S11

Ulrike Lanner6 as well as Alexander Schmidt6: performed mass spectrometry for figure S11

Author Affiliations 1Department of Biosystems Science and Engineering, ETH Zurich, Basel, Switzerland. 2Laboratory of Protein Design & Immunoengineering, École Polytechnique Fédérale de

Lausanne | EPFL, Lausanne, Switzerland 3Present Address: Novartis Institutes for BioMedical Research (NIBR), Basel, Switzerland 4University of Zurich, Zürich, Switzerland 5Diagnostic Hematology, University Hospital Basel, Basel, Switzerland 6University of Basel, Faculty of Science, Basel, Switzerland.

Author Affiliations




Abstract Sophisticated design strategies have been developed for engineering mammalian receptors

with different intracellular signaling capabilities. However, cellular therapies remain

constrained by the limited availability of sensors for the relevant disease states and markers,

because detection of new markers currently relies on pre-existing binders. To address the

need for an integrated target-to-receptor approach, we present here a pipeline for the

generation of a customizable advanced modular bispecific extracellular receptor (AMBER)

that combines our generalized extracellular molecule sensor (GEMS) system with a high-

throughput platform for generating designed ankyrin repeat proteins (DARPins). For proof of

concept, we chose human fibrin degradation products (FDPs) as markers with high clinical

relevance and screened a DARPin library for FDP binders. We built AMBERs equipped with 19

DARPins selected from 160 hits, and found 4 of them to be functional as heterodimers in

combination with a known scFv binder. Tandem receptors consisting of combinations of the

validated DARPins are also functional. We demonstrate an application of these AMBER

receptors by constructing designer cell lines that detect pathological concentrations of FDPs

and respond with the production of a therapeutic anti-thrombotic protein.

Abstract




Introduction

Cell-based therapies are still hampered by the limited availability of cellular receptors

for the detection of clinical disease markers. To address this issue, various types of receptors

have been employed to change cellular behavior (Scheller and Fussenegger 2019). These

include well-known chimeric antigen receptors (CARs) that rely on endogenous signaling of T

cells to drive T-cell responses to tumors. Other chimeric receptors that consist of fusions of

extracellular receptor domains and intracellular signaling domains of native receptors have

been engineered to reroute cellular responses (Mohammed et al. 2017). In addition, synthetic

dimerization-dependent receptors can drive transgene expression by employing bacterial

transcription factors (Schwarz et al. 2017). However, all of these receptor systems require

careful fine-tuning for each new ligand, and the ligand space is limited by the characteristics

of the existing binding proteins.

The recently developed generalized extracellular molecular sensor (GEMS) platform

allows for more robust switching of extracellular binding domains as well as intracellular

signaling domains (Scheller et al. 2018). GEMS receptors consist of an extracellular target-

binding moiety linked to the erythropoietin receptor (EpoR) extracellular domain. Activated

receptors amplify the incoming signal through signal transduction domains, such as the

intracellular domain of interleukin receptor 6B, which activates the signal transducer and

activator of transcription 3 (STAT3) pathway and triggers expression of a protein of interest

from STAT3-responsive promoters (fig. 1a). However, changing the specificity of the receptor

still relies on the availability of a suitable target-binding domain.

Most engineered receptors in mammalian synthetic biology use well-characterized

protein-protein interactions of native receptor-ligand interactions or single chain variable

fragments (scFv). However, the ligand space for native receptor-ligand interactions is severely

limited, and scFvs are typically derived from antibodies (AB) that are generated by laborious

immunizations of animals. Besides Fabs or scFvs obtained by classic AB selection combined

with protein engineering (Helma et al. 2015), there are various other kinds of protein scaffolds

that can be modified to change their binding properties. Some of these scaffolds have been

used in high-throughput screenings and for selecting binders to non-native targets. Examples

include fibronectin III (FN3)-derived monobodies (Karatan et al. 2004), αRep proteins

containing HEAT-like repeats (Guellouz et al. 2013), camelid-derived VHH domains, termed

nanobodies (Skerra et al. 1988), and designed ankyrin repeat proteins (DARPins) (Binz et al.

Introduction


2004; Schilling et al. 2014). While monobodies and αRep are generated by phage-display

(Guellouz et al. 2013) and yeast-surface display (Uchanski et al. 2019), respectively, DARPins

(Binz et al. 2004; Dreier et al. 2012) and nanobodies (Bencurova et al. 2015) have been shown

to work robustly with cell-free ribosome display selection systems that enable the screening

of bigger libraries and can be combined with error-prone PCR to increase sequence variation.

DARPins are small proteins of 14 kDa that preferentially bind to three-dimensional features

of target proteins (Pluckthun 2015). The DARPin technology enables the production of high-

affinity binders in an efficient manner. However, screening efforts generally aim to select just

a few high-affinity binders for a given target, and may discard potentially valuable binders.

In the context of synthetic dimerization-dependent receptors, selection of multiple

binders in parallel tackles an intrinsic bottleneck, i.e., that at least two independent binding

modules are required for a given target. Therefore, we aimed to combine the GEMS platform

with the DARPin screening technology to generate a customizable advanced modular

bispecific extracellular receptor (AMBER). To demonstrate the feasibility of this strategy, we

decided to develop and characterize the AMBER system by focusing first on cellular sensors

of human fibrin degradation products (FDPs), which are used as a clinical marker to exclude

pathologic coagulation events (Adam et al. 2009) and are also a candidate prognostic marker

for the severity of viral infections (Yao et al. 2020).

In the physiological context, blood coagulation is necessary to prevent hemorrhages

and the intrusion of pathogens during injury by forming stable clots consisting of fibrin fibrils.

Fibrin formation from its precursor, fibrinogen, is mediated by thrombin, which is activated

via the proteolytic blood coagulation cascade (Smith et al. 2015). Excessive coagulation can

happen due to stalling of blood flow (e.g., in the case of atrial fibrillation (Watson et al. 2009)),

reaction of the immune system to foreign material (Xu et al. 2014) (e.g., artificial heart valves

(Iung et al. 2014), stents (Luscher et al. 2007)), hereditary disorders (e.g. hyperactive factor V

Leiden (Dahlback et al. 1993; Bertina et al. 1994)), or cancer (Dipasco et al. 2011). Pathological

blood coagulation events can cause serious disability and even death (Benjamin et al. 2019),

so a cell-therapy approach employing autonomous detection of FDPs and delivery of a

therapeutic response directly in patients at risk could help to improve patient outcomes and

reduce mortality.

Therefore, we set out to demonstrate the translational potential of the AMBER system

by engineering mammalian cells equipped with AMBERs to detect pathological coagulation



events and to respond by triggering the secretion of a therapeutic protein, either

tenecteplase (TNK), which is an engineered version of the fibrinolytic protein tissue

plasminogen activator (tPA), or the anticoagulant hirudin, which naturally is found in leeches.

To slow down or inhibit blood coagulation, either direct (e.g. with hirudin (Sohn et al.

2001)) or indirect inhibition (e.g. with heparin (Baglin et al. 2006)) of thrombin can be

employed. However, even very recently developed drugs are constitutively active and hence

increase the risk of bleeding, or can lead to induced thrombocytopenia (Bonaca et al. 2020;

Blackmer et al. 2009). An integrated cellular system able to measure efficacy and toxicity in

parallel could improve the lead times of innovative anticoagulant therapies. We envision that

the presented DARPins and the receptors developed here will be useful for the benchmarking

of existing and novel blood-thinning agents, as well as inspiring the development of next-

generation designer cells to reduce the deleterious effects of thrombotic events.

The AMBER pipeline provides streamlined and predictable methodology for the

generation of new cellular receptors for soluble targets of choice. We believe this technology

will facilitate research on cellular signaling as well as the development of future cellular

therapies to treat diseases for which no relevant cellular receptors are known. Accordingly,

this study aims to expand the available repertoire of receptors against soluble targets by

introducing a new type of binder as well as to improve reliability and predictability of receptor

design as a whole. To this end a high-throughput DARPin-selection protocol was included to

open up new epitope spaces and streamline receptor development. In a proof-of-concept

experiment functionality of the resulting receptors for use in therapeutic settings was

assessed by targeting fibrin degradation products (FDPs) as clinical markers for thrombosis.

Introduction




Results

General receptor design strategy

The AMBER system employs the GEMS receptor architecture. We focused here on the

interleukin 6 signal transduction domain (IL6ST) as the intracellular domain. IL6ST signals

through the JAK/STAT pathway via signal transducer and activator of transcription 3 (STAT3),

and hence a synthetic reporter construct equipped with STAT3 response elements (RE) was

used to record STAT3 signaling (fig. 1a). Specificity of the sensor system is ensured by

introducing a previously described mutation (F93A) (Middleton et al. 1996) into the

extracellular domain of the EpoR scaffold to decrease native erythropoietin sensitivity.

We use this receptor scaffold in combination with an established high-throughput

DARPin selection method. This leads us to a four-step pipeline for the generation of functional

AMBERs (fig. 1b). Briefly, DARPins are first selected by ribosome display and then their

sequences are classified by means of phylogenetic comparison and pairwise amino acid

alignments. Distinct DARPins are fused to the receptor scaffold and tested for induction in a

homodimeric or heterodimeric receptor setup. Finally, a more compact tandem receptor

design may be applied to validated DARPin binders. Each step is described in detail below for

the case of xFDP receptors as an example.

Due to the dimeric nature of the AMBER platform, successful activation of the

receptors relies on binding of both receptor chains to their target. This requirement implies a

limitation concerning the distance between the epitopes. Based on the crystal structure of

activated EpoR (Livnah et al. 1996) we estimated the optimal theoretical distance between

the attachment sites of the C-termini of the two binders to the extracellular domains to be

less than 8 nm. Identical epitopes may be located along the whole 45 nm length of the dimeric

fibrinogen molecule (Kollman et al. 2009; Brown et al. 2000), providing a range of differently

spaced epitope pairs that should offer a good chance to find working receptors. However, the

large size of the molecule also reduces the likelihood that two epitopes would lie in close

proximity. These considerations led us to select cross-linked fibrin degradation products

(xFDPs) as the preferred target for demonstrating the feasibility of our receptor-building

pipeline; a choice that is supported by their established clinical relevance.

Results


Figure 1: a) Schematic illustration of the final coagulation-responsive cellular sensor system.

A sensor on the cell surface detects fibrin-derived fibrin degradation products (FDP) (grey

hexagons) via engineered extracellular binding domains (orange). Activation of generalized

extracellular molecular sensor (GEMS)-based advanced modular bispecific extracellular

receptor (AMBER) (green) triggers the signal transducer and activator of transcription 3

(STAT3; blue) transcription factor pathway, which is rerouted towards expression of a

reporter or a therapeutic output (tenecteplase, TNK or hirudin, Hir). b) A four-step pipeline

for the generation of functional AMBERs. First, binders (designed ankyrin repeat proteins;

DARPins) are generated by ribosome display and the genetic information of candidate

binders is analyzed. Next, genetically distinct binders are used to build AMBERs, which are

tested in cellular assays in homodimeric or heterodimeric configuration. After the

identification of functional binders, they can be used to build more compact tandem

a) The AMBER architecture

d) DARPin screening against xFPDs

b) The AMBER pipeline

c) Ribosome display for DARPin generation

ribosome

dissolvecomplex

reverse transcription

in vitrotranslation

translation stopsat last codon

mRNA library

nascentpeptide

stalled ribosome

reporter

STAT3

AMBER

inducer

binder

scaffold

signaling

hSAxF

DPs0

1

2

3

4

5H

TFR

ratio

relat

ive

to S

trept

avid

in [A

U]

1.45

118 = 31 %160 = 42 %

1.15

binder selection

(ribosome display)

sequence analysis

homodimeric

AMBERs

heterodimeric

AMBERs

tandem AMBERs

of validated hits



receptors with similar specificity. c) Overview of ribosome-display-based DARPin selection

to generate novel binding domains. An optimized DARPin library containing a mixture of

N3C-DARPins with randomized and non-randomized N- and C-caps is transcribed to mRNA,

which is translated in vitro. Since the mRNA lacks a stop codon, the ribosome is stalled on

the mRNA with the nascent peptide still attached upon completion of translation. The fully

folded DARPin can now bind to FDPs coupled to streptavidin-coated beads via their

biotinylated lysine residues. FDP-DARPin-ribosome-mRNA complexes are washed and

subsequently dissolved. The resulting mRNA is reverse-transcribed to yield an enriched

plasmid library, which is subjected to three additional rounds of the same selection cycle. d)

Results from high-throughput homogeneous time-resolved fluorescence assay (HTRF) to

assess binding of DARPins to human serum albumin (hSA) or cross-linked fibrin degradation

products (xFDPs) depicted as the ratio of maximum fluorescence of the sample versus the

streptavidin control. Streptavidin is used to couple biotinylated target protein to beads

during the selection run. DARPins are labeled via a FLAG-tag with an antibody coupled to a

fluorescent acceptor dye (d2) while xFDPs are labeled in a similar fashion via their biotin tag

with streptavidin carrying a Tb FRET donor. Binding of a DARPin to FDP protein hence results

in an increase of FRET efficiency (increased signal-to-noise ratio). AU: arbitrary units.

DARPin selection

To identify a suitable target for FDP binder screening we assessed the composition of

commercially available xFDPs enriched in D-dimers by means of SDS-PAGE. We confirmed

that the major component is D-dimers (fig. S1), accompanied with other degradation

products, most likely fragment E and fragment D of fibrin, as well as DD-E complexes, which

serve to broaden the available epitope space. xFDPs are protein complexes ranging from 120

kDa (D-E complex) to 220 kDa (DD-E complex) in size, and may include larger unprocessed

fragments of several hundred kDa. Those complexes have large surface areas that are likely

to present numerous epitopes. We biotinylated these xFDPs and applied them to an in vitro

ribosome display-based screening system for the generation of DARPins (fig. 1c). Briefly, an

optimized plasmid DARPin library comprising N3C-DARPins with randomized and non-

randomized N-terminal and C-terminal cap structures was transcribed into mRNA, and the

obtained mRNA was translated using isolated ribosomes. However, because the mRNA lacks

a stop codon, translation is stalled at the last codon, since the ribosome can neither progress

Results


nor complete the translation. Consequently, the mRNA is attached through the ribosome to

the corresponding nascent peptide. This peptide undergoes folding into a functional DARPin

that can bind to immobilized FDPs coupled to streptavidin-coated beads through biotinylated

lysine residues. The resulting FDP-DARPin-ribosome-mRNA complexes are purified and

disassembled to extract the enriched mRNA. The selected mRNA is used in a subsequent cycle

to further increase the affinities of DARPins in the pool. We employed four rounds of selection

under the same conditions.

We selected 380 DARPins through ribosome display and tested them for binding to

the target in a homogeneous time-resolved fluorescence (HTRF)-based assay (fig. 1d). We

excluded unspecific binding to human serum albumin (hSA), used as a stabilizing agent in the

xFDP preparation.

Of the 380 DARPins, 160 candidates (42 %) gave a signal sufficiently above background

in the HTRF assay (fold induction > 1.15). Of these, 118 (31 %) were considered good binders

(fold induction > 1.45). Next, DNA of a subset of 32 DARPins (8 %) was sequenced, leading to

the exclusion of 13 duplicated sequences. This left a set of 19 genetically unique DARPins (5

%). Phylogenetic analysis (fig. S2a) showed some of the 19 binders to have a high degree of

sequence similarity (e.g. DARPins A6, A7 and B4), which was confirmed by sequence

alignment (fig. S2b) and pairwise comparison of the hits (fig. S2c); DARPins with high similarity

would potentially bind to the same epitope. All 19 binders were further characterized by size-

exclusion chromatography (SEC) analysis to check for the presence of aggregates, which were

observed in four of the 19 DARPins, leaving 15 (4 %) DARPins with monomeric characteristics

(fig. S3). Nevertheless, we used all 19 binders in subsequent receptor designs, since we

hypothesized that constraints introduced by receptor fusion might affect dimerization in the

final receptor constructs.

Designing AMBERs in homodimeric, heterodimeric and tandem configurations

Different receptor configurations were explored to find the best setup to sense FDPs

and also to help us deduce some general guidelines for targeting other proteins of interest

(fig. 2a). A homodimeric receptor configuration in which all receptors are equipped with the

same binding module is the smallest and most straight-forward receptor architecture (fig. 2a,

i). However, homodimeric receptors rely on the presence of two epitopes in close proximity,

so that functional receptors require dimeric or symmetric ligands.



Heterodimeric receptors are pairwise combinations of receptors with different

binding moieties (fig. 2a, ii), and hence two independent receptor chains always need to be

expressed. Pairwise combinations grow exponentially with the number of available binders

(following the equation for pairwise combinations N of two out of n elements: N = 0.5 n2 - 0.5

n). Assuming a unique epitope for each receptor, this would greatly increase the likelihood of

finding active receptors, i.e. receptors that allow the binders to come into close proximity

upon ligand binding.

Tandem receptors consist of two binders fused in sequence to the N-terminus of the

receptor scaffold; this results in an increase of avidity and hence a possibly higher sensitivity

(fig. 2a, iii). Because the tandem approach involves only one receptor chain to build receptors

with two binding sites, all receptors presented on the cell surface are of the same

architecture, which increases the effective concentration. Tandem receptors also reduce the

size of the genetic construct (2.5-2.9 kb for the receptor alone) allowing it to be used in size-

restricted vectors, such as vectors derived from adeno-associated viruses (AAVs) (Grieger et

al. 2005). The drawback to building functional tandem receptors is the even greater

combinatorial space, which grows to the power of four for each binder involved (following

the equation for N pairwise combination of two out of n elements in pairs: N = 0.5 x2 - 0.5 x

with x = 0.5 n2-0.5 n, N = 0.125 n4 - 0.25 n3 - 0.125 n2 + 0.25 n).

For the AMBER platform, we therefore decided to screen homodimeric or

heterodimeric receptors prior to confirming the functionality of tandem receptors built with

the validated binders.

AMBER screening setup

To benchmark new FDP-DARPins as well as to explore potential benefits of cross-

platform combinations of binders, we used a previously described single-chain variable

fragment (scFv) targeting human D-dimers (Laroche et al. 1991). We equipped the AMBER

platform with both the anti-D-Dimer-scFv (scFv-AMBER) and the new repertoire of 19 FDP-

binding DARPins. This yielded a small library of 20 receptors, each bearing a single binding

moiety on the extracellular domain. HEK293T cells were co-transfected with plasmids each

encoding a single receptor under the control of a constitutive synthetic promoter derived

from human cytomegalovirus (hCMV) (PhCMV-receptor-pA), along with a plasmid encoding a

constitutively expressed STAT3-transcription factor (PhCMV-STAT3-pA) and a secreted human

Results


placental alkaline phosphatase (SEAP) reporter plasmid controlled by two STAT3 response

elements (RE) and a minimal version of the hCMV promoter (SEAP, RE2-PhCMVmin-SEAP-pA).

Control experiments show no unspecific induction

Insensitivity of the receptors to erythropoietin (EPO) was confirmed by measuring

EPO-mediated activation of the receptor scaffold (fig. S4a). A previously reported construct

based on an scFv responding exclusively to the non-toxic industrial dye reactive red (RR120)

(Scheller et al. 2018) was used as a negative control for testing potential receptor-

independent effects on reporter activity (fig. S4b). We assessed the impact of purified human

xFDPs on productivity and on the viability of HEK293T cells by measuring changes in

constitutive SEAP expression (fig. S4c) and reductive capacity via resazurin, which is reduced

by healthy mitochondria and is an indicator of metabolically active cells (fig. S4d). We found

xFDPs to be non-toxic at concentrations up to 10 µg/mL. We also tested for background

effects of human plasma and found that concentrations of up to 2 % (v/v) had little effect on

constitutive SEAP expression (fig. S4e) or viability (fig. S4f).

AMBER screening

We finally screened all 210 possible pairwise combinations of the available receptors

(including all 20 receptors as homodimers) for inducibility by 0.5 µg/mL xFDPs as well as by

0.5 % reconstituted plasma in cell culture (fig. S5a-d), and calculated the fold inductions

relative to the uninduced samples (fig. 2b). We divided the responses in two categories: weak

(fold induction >2) and strong (fold induction >5).

Induction with xFDPs yielded four receptors (1.9 %) in the weak and 3 (1.4 %) in the

strong category. Notably, all receptors responding strongly to xFDPs were heterodimers co-

expressed with the scFv receptor.



Figure 2: a) Engineered receptors based on the GEMS platform bind to fibrin degradation

products, including fragment E (blue), fragment D (green) and cross-linked D-dimers (green,

indicated by orange crosses), as well as complexes thereof. Binding induces STAT3-mediated

gene expression of therapeutic proteins or the reporter protein human secreted placental

a) schematic of different classes of generated AMBERs

c) homodimeric receptors

b) screening of homodimeric and heterodimeric receptors against xFDPs and plasma

f)

d) heterodimeric receptors

g)

e) tandem receptors

h)

1 10-3 1 10-2 1 10-1 1 1000

20

40

60

80

100

xFDP concentration [µg/mL]

SEAP

[UL-1

]

B4A7A6G1scFvRR120-receptor

1 10-3 1 10-2 1 10-1 1 1000

20

40

60

80

100

SEAP

[UL-1

]

B4 + scFvA7 + scFvA6 + scFvG1 + scFvRR120-receptor

xFDP concentration [µg/mL]1 10-3 1 10-2 1 10-1 1 100

0

20

40

60

80

100


SEAP

[UL-1

]

G1-A7-tandemG1-B4-tandemG1-G1-tandemscFv-A7-tandemscFv-B4-tandemscFv-G1-tandemRR120-receptor

1 10-3 1 10-2 1 10-1 1 1000

40

80

120

160

plasma concentration [%]

SEA

P [U

L-1]

B4A7A6G1scFvRR120-receptor

1 10-3 1 10-2 1 10-1 1 1000

40

80

120

160


SEA

P [U

L-1]

B4 + scFvA7 + scFvA6 + scFvG1 + scFvRR120-receptor

1 10-3 1 10-2 1 10-1 1 1000

40

80

120

160


SEA

P [U

L-1]

G1-A7-tandemG1-B4-tandemG1-G1-tandemscFv-A7-tandemscFv-B4-tandemscFv-G1-tandemRR120-receptor

SEAP

STAT3

heterodimeric receptor homodimeric tandem-receptorhomodimeric receptor

(i) (ii) (iii)

A12-A12

A10-A10

B12-B12

C4-C4

A12-scF

v

A10-F12

B12-F8

C4-A11

A12-G

1

A10-C11

B12-A6

C4-A7

A12-E11

A10-D11

B12-F10

C4-F9

A12-F6

A10-B4

B12-D12

C4-B10

A12-C4

A10-A12

B12-A10

C4-B12

A12-A11

A10-scF

v

B12-F12

C4-F8

A12-A7

A10-G

1

B12-C11

C4-A6

A12-F9

A10-E11

B12-D11

C4-F10

A12-B10

A10-F6

B12-B4

C4-D12

A12-B12

A10-C4

RR1200

10

20

30

40

fold

indu

ctio

n

xFDP plasma

Results


alkaline phosphatase (SEAP). (i) Homodimeric receptors present one binding moiety only

and rely on symmetrically presented epitopes on the target. (ii) Heterodimeric receptors

display two different binding moieties, similarly to (iii) homodimeric tandem receptors,

which comprise two binders that are fused on top of the receptor chains in a head-to-tail

fashion. b) Combinations of different receptors comprising different binding domains were

tested for their response to purified xFDP protein (blue dots) as well as plasma (orange

dots). Values are normalized to uninduced samples of the same receptors and represent

means ± SD of triplicates. c)-e) Functional systems started to respond well below the

clinically relevant concentration of 0.3 µg/mL (dotted line). f)-h) When tested for reactivity

against human plasma the sensitivity of the system increased remarkably. c)-h) Solid lines

are non-linear fits. Values for the highest concentration were excluded for the fit in

subfigure e)-f) to improve the fits for lower to medium concentrations. The RR120 receptor

serves as a negative control. For details, see the main text. The x-axis is log-scale, and all

values are means ± SEM of N=3 independent experiments performed in triplicates.

Human plasma was tested to evaluate the performance of the system in a more native

setting and to investigate responses to other species of FDPs that were underrepresented in

the DARPin screen. In the case of plasma induction, we found weak responses for 37 (17.6 %)

and strong responses for 49 (23 %) of the 210 receptors. Among the plasma-responsive

receptor combinations, 77% included one of four DARPins (A6, A7, B4, F6). We could not find

any correlation of HTRF signal intensity from DARPin selection with either the number of

functional receptors or the resulting receptor signal strength in a homodimeric setup (fig. S6).

More stringent selection of DARPins on the basis of HTRF values therefore seems not to

influence the likelihood of finding a functional receptor. However, receptor activity in a

homodimeric setting seems to be predictive for the total number of functional receptor

combinations (fold induction >2). Functional receptors in some cases comprised DARPins that

showed aggregation behavior as recorded by SEC for the A7- and F9-DARPins (fig. S2),

indicating that auto-dimerization of the binder does not necessarily limit its usefulness in a

receptor.



Characterization of candidate AMBERs

We characterized a set of four DARPin binders comprising candidate DARPins A6, A7,

B4 and the less active DARPin G1 as well as the scFv in a homodimeric receptor configuration.

Addition of increasing concentrations of xFDPs revealed an OFF-type switching behavior for

receptors based on A6, A7 and B4 DARPins (fig. 2c). While this could be explained by inhibition

of auto-dimerization by ligand binding, only the A7 DARPin showed such characteristics

during SEC analysis. Therefore, we speculate that the observed OFF-type switching behavior

is caused by the binding of large antigens to both receptor chains, thereby reducing basal

(“leaky”) activity by stabilizing pre-formed receptor dimers in an inactive conformation.

In contrast, we recorded ON-type switching behavior for the same DARPins with up to

110-fold induction in response to human plasma (fig. 2f). These results suggest successful

expression and display of DARPin-AMBERs. The increase in activity towards plasma indicates

that these DARPins are activated by another protein that is not present, or is present only at

a trace level, in the xFDP preparation (compare fig. S1).

We next examined the four candidate DARPins in combination with the scFv as

heterodimeric receptor pairs. In contrast to the homodimeric configuration, these receptors

dose-dependently respond to xFDPs with inductions of up to 25-fold and EC50 values of 0.15,

0.19 and 0.12 µg/mL for DARPins A6, A7 and B4, respectively (fig. 2d). The measured EC50

values for xFDPs are well below the pathological threshold for D-dimer protein of 0.3 µg/mL

in blood (Froehling et al. 2007) (indicated by a dotted line in fig. 2 c-e). Induction with plasma

yielded similar dose responses to those in the homodimeric configuration (fig. 2g). However,

the heterodimeric setup also allows for homodimeric receptors at the cell surface, and the

latter most likely dominate or at least strongly contribute to the response in the case of

plasma induction.

Binding affinities of DARPins A6, A7, B4 and G1 as well as the scFv were quantified by

surface plasmon resonance (SPR) (fig. S7). Due to the symmetric structure of the target

molecules, we stipulated a heterogenous binding modality resulting in two distinct binding

constants. All binding constants, except for the G1-DARPin, were within the nanomolar range

measuring 169/244 nM for the scFv and 14/37 nM, 44/91 nM, 23/59 nM and 602/1030 nM

for DARPins A6, A7, B4 and G1, respectively. Affinities for the G1-DARPin were up to 10- to

Results


50-fold lower compared to those of other DAPRPins which may explain the low performance

of the G1-DARin in a receptor context.

Generating monomeric tandem AMBERs

After identifying functional DARPin binders and scFv combinations in the homodimeric

and heterodimeric screen, we fused all 16 combinations of tested DARPins A7, B4, G1 and the

scFv to the receptor scaffold in a tandem orientation (see fig. 2a, iii). Screening of

homodimeric tandem receptors against xFDPs revealed three out of 16 (19 %) receptors to

be sensitive (fig. S8); all three contain a DARPin in combination with the scFv (fig. S8c,d). All

but one of the receptors based on DARPins (89 %) and two (29 %) of the scFv-DARPin

combinations were responsive to plasma. Notably, we found a working receptor for each of

the DARPins successfully tested before (fig. S8a,b). In contrast to the working DARPin tandem

receptors that contain a tandem repeat of a single DARPin, scFv alone did not respond to

neither plasma nor xFPDs in such a configuration, in accordance with the results for the

respective homodimeric receptors.

In a dose-response assay, DARPin-only tandem receptors detected xFDPs in a dose-

dependent manner, albeit with lower sensitivity compared to heterodimeric scFv

combinations, showing estimated EC50 values of 4.3 and 1.2 µg/mL for the G1-A7 and G1-B4

tandem receptors, respectively (fig. 2 e). Tandem receptors comprising DARPins and the scFv

were similarly sensitive as their heterodimeric counterparts, showing EC50 values of 0.31

µg/mL (vs 0.19 µg/mL) for the scFv-A7 receptor and 0.25 µg/mL (vs 0.12 µg/mL) for the scFv-

B4 receptor. Induction with plasma led to increased signal intensities for all tandem receptors,

even for the previously non-responding G1-G1 tandem receptor. These results indicate that

the tandem receptors exhibit full functionality and mostly reflect the features of the

corresponding heterodimeric receptors, while at the same time being more compact. The

question of whether tandem configurations also allow for different ligands to be detected will

be addressed later.



Figure 3: Overview of processes involved in fibrinolysis. Fibrinogen forms a head-to-head

dimer via the E-domain (blue) that is covalently linked via di-sulfide bonds. Upon activation

with thrombin, fibrinogen oligomerizes via interactions of its D-domains (green) with the E-

d) state of cross linking affects induction

e) attribution of ligand specificity of each receptor

a) Overview of fibrinolytic processes

b) fibrinogen induces AMBERs c) inhibitors of coagulation block induction

Fibrinogen

D-domainE-domain

disulf ide bond

factor XIIIa

tridegin

non-cross-linked fibrin f ibre non-cross-linkedDD-E complex

cross-linked fibrin

plasmin

plasmin

TXA

TXA

thrombin

plasma

plasma +

hirudin

plasma +

hepari

n

fibrin

ogen

fibrin

ogen + hiru

din

fibrin

ogen + hepa

rin

uninduced

0

5

10

15

SEAP

[U*L

-1]

B4B4 + scFvG1-B4scFv-B4

10g/m

L

2g/m

L

uninduced

10g/m

L

2g/m

L

uninduced

10g/m

L

2g/m

L

uninduced

10g/m

L

2g/m

L

uninduced

0

5

10

15

SEAP

[U*L

-1]

B4B4 + scFvG1-B4scFv-B4

uninduced

no thrombin

thrombin and fib

rinogen

+ TXA

+ trideg

in

+ plasmin

+ plasmin + tri

degin

uninduced

no thrombin

thrombin and fib

rinogen

+ TXA

+ trideg

in

+ plasmin

+ plasmin + tri

degin

uninduced

no thrombin

thrombin and fib

rinogen

+ TXA

+ trideg

in

+ plasmin

+ plasmin + tri

degin

uninduced

no thrombin

thrombin and fib

rinogen

+ TXA

+ trideg

in

+ plasmin

+ plasmin + tri

degin0.0

0.5

1.0

1.5

2.0

2.5

norm

alize

d S

EAP

[RU]

B4 B4 + scFv G1-B4 scFv-B4(i)

uninduced

no thrombin

thrombin and fib

rinogen

+ TXA

+ trideg

in

+ plasmin

+ plasmin + tri

degin0.0

0.5

1.0

1.5no

rmal

ized

SEA

P [R

U](i i) AMBER

B4+ specific induction

(ii) AMBERB4/scFv

:

D-dimers

(i) AMBERB4

:

fragment E

(iii)AMBERG1-B4

:

FDPs

(iv) AMBERscFv-B4

:

xFDPs

45nm 6.5nm 10nm

Results


domain to form an insoluble non-cross-linked fibrin mesh. The mesh is stabilized by factor

XIIIa, which introduces isopeptide bonds between adjacent D-domains. Fibrinolysis is

mediated by plasmin, which cleaves fibrin into smaller soluble fragments. Depending on

whether the fibrin was cross-linked prior to plasmin digestion, different types of fragments

arise. Degradation products include fragment E comprising the E-domain of fibrinogen

(blue), fragment D (green) and cross-linked D-dimers (green, indicated by orange crosses),

as well as complexes thereof. b) Reactivity of receptors towards fibrinogen was tested in a

minimal system using two different concentrations of fibrinogen and thrombin. c) Reactivity

towards fibrinogen and plasma can be blocked by addition of known anti-coagulants

heparin and hirudin. d) Cross-linking of fibrinogen is mediated by factor XIII (FXIII) and

blocked by its inhibitor tridegin. Reporter expression is modulated by addition of tridegin or

plasmin and TXA in a minimal system comprising purified fibrinogen and thrombin. Reporter

activity is normalized to a sample with constitutive expression of the reporter under the

same conditions to compensate for unspecific alterations. ii) Values of i) for B4+scFv are

normalized to values for B4 to emphasize differences in reporter expression between both

receptor configurations. All values are means ± SD of representative experiments in

triplicate. e) Summary of suggested receptor specificities. AMBERB4 may preferentially bind

to fragment E of fibrin, AMBERB4/scFv may bind to cross-linked D-dimers complexed with

fragment E (DD-E complex) or fragment E alone (because some parts are also presented in

the AMBERB4 configuration), AMBERG1-B4 may bind partially to non-cross-linked fibrin

degradation products (FDPs), and AMBERscFv-B4 favors cross-linked FDPs (xFDPs).

Linking fibrinogen activation to induction

To characterize the impact of potentially under-represented species of xFDPs in the

initial DARPin selection and to exclude off-target receptor activation, we next investigated

the receptor sensitivity to plasma. For all subsequent characterizations we focused on the

homodimeric B4-receptor (AMBERB4; PhCMV-B4-receptor-pA), the heterodimeric B4+scFv-

receptor (AMBERB4/scFv; PhCMV-B4-receptor-pA + PhCMV-scFv-receptor-pA), the tandem G1-B4-

receptor (AMBERG1-B4; PhCMV-G1-B4-receptor-pA), and the scFv-B4-receptor (AMBERscFv-B4;

PhCMV-scFv-B4-receptor-pA).

During coagulation and subsequent degradation of the fibrin mesh, various

fibrinogen-related complexes are formed (fig. 3a). In general, possible degradation products



that could serve as inducers of AMBER receptors include fragment E of fibrin, D-E and DD-E

complex, as well as D-dimers and larger, partially digested degradation products containing

the same epitopes. Fragment E of fibrin has a diameter of approximately 4 nm and a length

of 13 nm, while fragment D is at least 10 nm long and 5 nm wide, with D-dimer accordingly

being of 20 nm length and 5 nm diameter (Kollman et al. 2009).

We found that in addition to xFDPs and plasma, all four receptors could also be dose-

dependently induced by purified fibrinogen in the presence of FCS (fig. 3b). Plasmin

degradation products of fibrinogen, i.e., fragment D (fig. S9a) or fragment E (fig. S9b), or a

mixture of fragments D and E (Fig. S9c), did not trigger induction. We found that extensive

plasmin digestion of fibrinogen in the presence of thrombin decreased the signaling of

AMBERs, which indicates that high plasmin concentrations degrade fibrinogen faster than it

can be activated by thrombin, and these degradation products cannot be detected by the

receptors (fig. S10a).

Taken together, these results show that neither native fibrinogen nor its plasmin

cleavage products are inducers of AMBERB4, AMBERB4/scFv or AMBERG1-B4. Furthermore, the

activity of AMBERB4, AMBERB4/scFv and AMBERG1-B4 pre-incubated with fibrinogen for 24 h was

unchanged by additional incubation with plasmin for another 24 h (fig. S10b). This could

indicate that either the degradation products initially bound by the receptors cannot be

processed any further or that binding of (x)FDPs to the receptors blocks plasmin cleavage. In

accordance with these results, induction with xFDPs was also not modulated by plasmin (fig.

S10c).

To confirm that fibrinogen is the active molecule in plasma preparations, we

fractionated plasma by size-exclusion chromatography (SEC) to separate the proteins based

on their hydrodynamic radius. The run-times of the fractions that activated AMBERB4 and

AMBERB4/scFv corresponded well with reported run-times for fibrinogen (fig. S11a). The

presence of fibrinogen in the active fractions was confirmed by polyacrylamide gel-

electrophoresis followed by mass spectrometry (fig. S11b).

Furthermore, fibrinogen- and plasma-mediated induction was directly linked to

coagulation, because the induction was blocked by direct (hirudin) or indirect (heparin)

inhibition of thrombin. Both inhibitors specifically repressed induction of the system upon

addition of either 0.5 % (v/v) plasma or 10 µg/mL fibrinogen (fig. 3c, S12a).

Results


Taken together, these results establish thrombin and thrombin-containing FCS as

potent drivers of coagulation directly linked to receptor activation without off-target activity.

At this point, we wondered whether the system could be used to evaluate anti-

coagulant agents. Indeed, AMBERB4, AMBERB4/scFv and AMBERG1-B4 showed the expected dose-

related response to the thrombin inhibitor argatroban, in good agreement with previous

results for argatroban used against fibrin-bound thrombin (Berry et al. 1994) (fig. S12b). The

prodrug ximelagatran remained ineffective in our test up to concentrations of 100 µM (fig.

S12c).

Finally, we tested whether differences in fibrin preparations affect receptor

sensitivity, using insoluble, dried fibrin resuspended in cell culture media containing FCS (fig.

S12d) and fibrin that was solubilized at acidic pH and precipitated by neutralization (fig. S12e).

In both cases, we recorded a bell-shaped induction of all receptors tested (AMBERB4,

AMBERB4/scFv, AMBERG1-B4 and AMBERscFv-B4) with peak inductions at around 0.3 mg/mL and 3-

10 µg/mL, respectively. Given the mechanism of receptor activation, we expect each receptor

chain to bind to a separate ligand at high concentrations, thus stabilizing the receptor in an

inactive conformation and giving rise to a bell-shaped induction curve. Interestingly, the

response to each compound varied among the receptors. While this may be partly due to

differences in compound composition and epitope accessibility, it also raises the possibility

that the receptors bind to different proteins or protein complexes that are present at

different concentrations in each preparation.

D-dimer- and fragment E-mediated activity

To investigate potential differences in receptor specificity and to further exclude off-

target activity we tried to identify the active molecule(s) for candidate receptors AMBERB4,

AMBERB4/scFv, AMBERG1-B4 and AMBERscFv-B4. We hypothesized that the receptors can

distinguish fibrin-derived fragment D, fragment E, D-dimer and DD-E complexes (compare fig.

3a). DD-E complex and D-dimers originate from the degradation of mature fibrin formed by

cross-linking of activated fibrinogen molecules via isopeptide bonds catalyzed by factor XIIIa

(Lorand et al. 1968; Gupta et al. 2016). The leech-derived peptide inhibitor tridegin reportedly

blocks factor XIIIa activity through direct inhibition (Bohm et al. 2014) and was therefore used

to reduce the concentrations of cross-linked species such as D-dimer and DD-E complexes.



We investigated the effect of plasmin, the plasmin inhibitor tranexamic acid (TXA),

and tridegin on coagulation-mediated receptor activation. We used a setup without FCS and

initiated coagulation with fibrinogen and thrombin. Due to the unspecific effect of TXA on

SEAP expression (fig. S13), we normalized all samples to cells constitutively expressing the

reporter under the same conditions. Cells were washed with FCS-free medium before the

experiment to reduce residual serum components. However, as can be deduced from the

following results, trace amounts of plasmin as well as factor XIII remained, because inhibitors

reported to target these enzymes were still effective (fig. 3d, i). Possibly these enzymes are

even secreted by the cells themselves (Petryszak et al. 2016). While controls incubated with

only fibrinogen remained uninduced, addition of thrombin led to the expected increase of

SEAP reporter expression, indicating coagulation-induced receptor activation. We observed

stronger activation of tandem receptors AMBERG1-B4 and AMBERscFv-B4 than of homodimeric

or heterodimeric receptors AMBERB4 and AMBERB4/scFv (fig. 3d, i).

Interestingly, AMBERB4 activation was independent of TXA-mediated plasmin

inhibition. These results indicate that this receptor responds to early products of coagulation

(fig. 3d, i). In the presence of tridegin in the induction mix, AMBERB4 responded with an

increase of reporter expression, which also points towards more accessible, non-cross-linked

FDPs as inducers. This is consistent with the fact that AMBERB4 did not respond in previous

tests with D-dimer-enriched (fragment E depleted) cross-linked FDPs (compare fig. 2c).

Addition of plasmin again led to a decrease of signal intensities for all receptors; the decrease

was largest for AMBERB4 and AMBERB4/scFv (fig. 3d, i), likely because of excessive degradation

of fibrinogen by plasmin prior to coagulation (Hur et al. 2015) (compare fig. S9a). However,

further addition of tridegin to plasmin led to a marked increase of AMBERB4 activity

confirming the involvement of non-cross-linked degradation products (i.e. fragments D, E and

D-E complex) as ligands (fig. 3d, i). We also found earlier that the proteins detected by the

AMBERB4 and AMBERB4/scFv systems are most likely final products of plasmin digestion

(compare fig. S10b). Since fragment D is only symmetric if it is cross-linked, it only then

comprises two identical epitopes. Thus, fragment E is the sole candidate that fulfills all

requirements, and we conclude that DARPin-B4 (present in AMBERB4 and AMBERB4/scFv) binds

FDPs in a cross-linking-independent manner, and that possible binding partners are fragment

E as a single component or in D-E or DD-E complexes (fig. 3e, i).

Results


Expression of heterodimeric AMBERB4/scFv also allows for the assembly of AMBERB4

homodimers on the cell surface. Hence, results for AMBERB4/scFv are overlaid with the

activation pattern of AMBERB4 (fig. 3d, i). We therefore normalized the SEAP levels of

AMBERB4/scFv to AMBERB4 to analyze the contribution of the scFv receptor (Fig. 3d, ii). AMBERB4

and AMBERB4/scFv responded similarly under all conditions examined, except for the

combination of plasmin and tridegin, for which we recorded a decrease in reporter expression

for the AMBERB4/scFv receptor configuration. This tridegin-mediated inhibition may imply that

the AMBERB4/scFv receptor relies on cross-linking. These results are consistent with earlier

findings that AMBERB4/scFv but not its AMBERB4 counterpart proved sensitive to purified cross-

linked FDPs, enriched in D-dimer protein (compare fig. 2c). We therefore conclude that the

scFv moiety within AMBERB4/scFv binds cross-linked D-dimer protein in the complex with

fragment E (DD-E complex), which in turns bind to the B4-DARPin (fig. 3e, ii). This specificity

of the scFv is in line with that reported in the original publication(Laroche et al. 1991).

Probably due to a low epitope count or steric hindrance, receptors based on the scFv alone

are not able to detect D-dimers or coagulation. Nevertheless, we can build functional

receptors that combine DARPin binders and the scFv.

The tandem receptors AMBERG1-B4 and AMBERscFv-B4 were both sensitive to TXA

inhibition, indicating a dependence on plasmin activity to produce their respective ligands

(fig. 3d, i). While AMBERG1-B4 showed an increase in activity upon addition of tridegin, the

same treatment led to a decreased reporter output for AMBERscFv-B4, indicating binding to

non-cross-linked and cross-linked FDPs, respectively. Supplementation with plasmin again

resulted in decreased reporter expression and additional tridegin lowered the reporter

expression further. However, the effect of tridegin was notably stronger for AMBERscFv-B4 (fig.

3d, i). These results indicate that both tandem receptor configurations detect partially

digested, larger FDPs, with AMBERscFv-B4 preferentially binding cross-linked fragments and

AMBERG1-B4 preferentially binding non-cross-linked fragments (fig. 3e, iii, iv).

In conclusion, these results reveal that the AMBERs in the analyzed set have different

sensitivities and specificities towards fibrin degradation products, thus providing an

illustration of the versatility of the AMBER platform.



Generating mammalian designer cells for the detection of coagulation

To test the AMBER platform in a translationally relevant proof-of-concept format, we

generated mammalian designer cells to sense coagulation or xFDPs and respond by the

secretion of a therapeutic protein, TNK or hirudin. For this purpose, we employed the more

sensitive heterodimeric AMBERB4/scFv as well as the genetically more compact homodimeric

AMBERscFv-B4 tandem receptor (compare fig. 2d). Prior to generating stable cell lines, we

examined the sensitivity of different reporters containing two, four, 12, or 16 repeats of the

response element (fig. S14a), and selected a reporter element with four repeats.

We generated a plasmid for genomic integration of a reporter cassette based on the

four-repeat response element that drives induced expression of a secreted version of the

luminescent reporter nano-luciferase (Nluc) fused to the therapeutic protein TNK. We

genetically linked expression of TNK to the reporter Nluc via a furin cleavage site (RARYKR)

(figure 4a). The furin site is cleaved in the Golgi apparatus to deliver an equimolar ratio of

both proteins to the supernatant, hence providing a direct correlation of reporter activity with

potential therapeutic effect. We selected stable monoclonal HEK293T cells that express the

STAT3 transcription factor alongside AMBERscFv-B4 and the Nluc and TNK reporter cassette

(termed HEKscFv-B4/TNK). We also generated a polyclonal control cell line expressing the

receptor for RR120 instead (HEKRR120R/TNK); this does not respond to any of the coagulation-

dependent stimuli (fig. S14b).

Correlation of Nluc and TNK activity was confirmed by measuring the activity of TNK

and Nluc in the supernatant of the monoclonal cell line HEKscFv-B4/TNK in response to xFDPs (fig.

4 b). With an onset time of six hours, the system was responding within the limits of a

transcription-based gene switch (fig. S14c-d). Additionally, the system could detect xFDPs at

levels as low as 0.1 µg/mL, well below the clinical disease threshold of 0.3 µg/mL and was also

sensitive to plasma (fig. S14e-f).

Results


Figure 4: a) Schematic illustration of vectors used for the generation of stable coagulation-

sensing cells. Top: sensing and expression cassette for a secreted nano-luciferase (Nluc)

reporter fused to a murine Fc tag for enhanced stability in vivo (Nluc-mFc) and a

tenecteplase-mFc fusion construct (TNK-mFc). Equimolar production and separation of the

products during secretion is ensured by placing a furin cleavage site (RARYKR) between the

two proteins. Middle: construct for stable integration of the reporter cassette using the

Sleeping Beauty transposon system expressing a fluorescent protein (FP) to facilitate

fluorescence-activated cell sorting (FACS) coupled to antibiotic resistance markers for zeocin

(ZeoR) or blasticidin S (BlastR) for stable cell line selection. Bottom: construct for stable

a) schematic of reporter/TNK construct used

c) schematic of hirudin reporter construct d) schematic of hirudin activity assay

e) inhibition of coagulation by hirudin f) effect on thrombin clotting time (TCT)

b) activity of TNK correlates with reporter output

AMBERB4/scFv/hirudin B4

day 1: induction of hirudin expression

with xFDPs

4x STAT3-Op Igk-SS nLuc-mFc RARYKR TNK-mFc

4x STAT3-Op Igk-SS nLuc-mFc p2a natSS-Hirudin-HM2

FP-ZeoR

nLuc-T

NK

5’ DR

3’ DR

FP-BlastR

receptor

5’ DR

3’ DR

STAT3

+ 1g/m

L xFDPs

uninduced

30

35

40

45

Thro

mbi

n C

lotti

ng T

ime

[sec

]

p=0.0016

00.0

10.0

20.0

60.2

00.6

42.0

0

heparin

0

2

4

6

8

10

-5.0 105

0.0

5.0 105

1.0 106

1.5 106

2.0 106

2.5 106

SEAP

[U*L

-1]

nLucSEAP

nLuc Luminescence [A

U]

xFDPs [µg/mL]

01.0

03.1

610

.0131

.64

100.0

431

6.31

1000

.18

3162

.56

1000

0.00

0

1 106

2 106

3 106

4 106

5 106

0

2

4

6

8

xFDP concentration [ng/mL]

Nlu

c Lu

min

esce

nce

[AU]

Nluc activity

TNK activtiy [U

/L]

TNK activity



integration of the receptors as well as an overexpression module for signal transducer and

activator of transcription 3 (STAT3) transcription factor. DR, direct repeat; Nluc, nano-

luciferase; TNK, tenecteplase (htPA variant); mFc, murine antibody fragment for increased

stability. b) Measurements of xFDP- induced expression of reporter Nluc and TNK from

HEKscFv-B4/TNK cells show the two are well correlated and legitimates the use of Nluc

reporter production as a proxy for TNK activity. 45 000 cells were seeded per well and

incubation was done for 24 h. We subsequently induced those cells for 24 h with the

indicated concentrations of xFDP protein or plasma. Nluc and TNK activity were measured

in the supernatant. TNK activity of untreated cells was subtracted from the observed values.

c) Schematic illustration of the vector for the generation of the monoclonal stable cell line

mHEKB4/scFv/hirudin for inducible hirudin-HM2 production. d) Workflow to assess hirudin-

HM2 efficacy in cell culture. As in b), mHEKB4/scFv/hirudin cells were induced with xFDP

protein on day 1. Supernatant of induced mHEKB4/scFv/hirudin cells was mixed with 0.5 %

(v/v, final) plasma on day 2 and dispensed on coagulation-sensing cells expressing

AMBERB4. SEAP reporter expression from sensor cells and Nluc activity of

mHEKB4/scFv/hirudin are measured in the same sample with a multiplate reader on day 3.

e) Hirudin-HM2-induced inhibition of coagulation is sensed by reporter cells equipped with

AMBERB4 and expressing SEAP in response to coagulation. mHEKB4/scFv/hirudin cells show

a clear correlation of induction by higher concentrations of xFDP protein with reduced

coagulation. f) Thrombin time of mHEKB4/scFv/hirudin cells induced with xFDPs was

assessed in a clinical laboratory and compared to uninduced controls. Values are b) mean ±

SD of quadruplicate determinations, or e-f) means ± SD of triplicate determinations.

Preventing coagulation with hirudin-producing designer cells

In most cases, the prevention of coagulation is therapeutically preferable to the

induction of clot lysis. Anti-coagulants such as hirudin and heparin or its derivatives are

generally used, especially for long-term administration. We therefore set out to determine

the effectiveness of a designer cell-based treatment strategy for the production of hirudin. As

a first step, we re-designed the reporter construct used for inducible tenecteplase expression.

We replaced the tenecteplase and furin cleavage sequence with a p2a peptide for the

separation of the adjacent proteins during translation and used a codon-optimized sequence

Results


encoding secreted hirudin-HM2 from Poecilobdella manillensis (Mexican medical leech) (fig.

4c). We generated a stable monoclonal HEK293T cell line (mHEKB4/scFv/hirudin) harboring the

new reporter construct alongside AMBERB4/scFv. Next, we developed an assay to measure the

efficacy of the therapeutic output by harnessing the AMBERB4-based coagulation reporter

system. Figure 4d illustrates the workflow used in this experiment. mHEKB4/scFv/hirudin cells were

induced for 24 h with increasing doses of xFDPs. The supernatant was transferred to reporter

cells transfected with AMBERB4 as well as a SEAP reporter cassette and mixed with a final

concentration of 0.5 % (v/v) plasma to assess the anticoagulant effect of secreted hirudin.

Purified hirudin and heparin were added as a control. Hirudin expression was sufficient to

reduce the coagulation-induced signal of the reporter cells by 3.3-fold (fig. 4e), and

significantly (p=0.0016) prolonged the thrombin clotting time (TCT) by 10 % in an established

clinical laboratory test (fig. 4f), thus demonstrating functional secretion of the therapeutic

protein.

Taken together, the above results show that our modular receptor system works as

expected; we can detect coagulation at an early stage by employing the AMBERB4 system, or

we can detect early as well as late biomarkers of thrombotic events by using an integrated

sensor incorporating the scFv (AMBERB4/scFv or AMBERscFv-B4). Furthermore, our proof-of-

concept designer cell lines show that AMBER receptors can be coupled to therapeutic protein

output: either tenecteplase to boost clot lysis or hirudin to prevent further coagulation. These

results indicate that the AMBER pipeline has translational potential for cell-therapy

approaches to treat recurrent thrombosis or embolism in patients at risk.



Discussion In this work we have combined DARPin selection with our GEMS technology, and we

show that the resulting AMBER platform enables the generation of programmable receptors

with customizable sensitivity and specificity by screening a vast space of possible epitopes.

Such versatility is unprecedented in a receptor system for soluble ligands. We used the

created receptors to reliably detect FDPs, which are generated at the onset of blood

coagulation, in cell culture experiments. Furthermore, the reporter cells were successfully

used to evaluate the efficacy of anti-coagulant drugs. In another proof-of-concept, we

showed that secretion of a therapeutic protein alongside a luminescent reporter provides a

direct correlation of reporter activity with the expression of a therapeutic protein.

Notably, not all hits from the initial DARPin selection turned out to be good candidates

for building a functional receptor. This also appears to be the case for other binders described

in the literature, and emphasizes the importance of a diverse library as input for the AMBER

pipeline. In fact, with respect to induction by plasma, about 16 % of combinations with a

known scFv and 40 % of DARPin-only receptors were active in a heterodimeric receptor

configuration. Since about 83 % of these DARPin-only receptors are based on one out of four

of the initially characterized 19 DARPins, the overall success rate in finding suitable binders

was 21 %. With a symmetric target the generation of homodimeric receptors is of course

much more likely, and we expect the chance of finding such receptors for regular monomeric

targets to be lower. Nevertheless, the GEMS platform has been successfully used in

combination with pre-existing binders to detect the monomeric 26 kDa cancer antigen PSA

with the use of two scFvs (Scheller et al. 2018). Here, out of three scFvs tested, two were

shown to form a working receptor. As discussed earlier, we expect the number of working

combinations of FDP binders that satisfy the design constraints of the receptor architecture

to be lower due to the much greater size. While a large surface is not necessarily indicative of

successful binder generation, the number of non-overlapping epitopes is nevertheless

increased, and therefore so is the chance of finding suitable pairs of binders for receptor

generation. While one can use the overall design to build receptors similar to already reported

binders, an integrated pipeline to check a receptor library for activity in cell culture is easy to

parallelize and provides a time- as well as cost-efficient solution. Therefore, we expect the

versatile and highly streamlined DARPin technology in combination with the AMBER

technology to yield more receptors for a range of different targets. Targets may include

Discussion


soluble cancer antigens, cytokines or even cell-surface epitopes. AMBERs might be especially

useful for targets for which no natural receptor is known and available sequences for scFvs or

other binders are limited. Modular switching of the signaling domains of the underlying

receptor scaffold has already been reported (Scheller et al. 2018) and this could be used to

accelerate research in systems and stem-cell biology to rewire endogenous signaling

cascades.

In addition, the selected anti-FDP DARPins may be used to build novel diagnostic tools

as well as therapeutics by expanding the available epitope landscape for the development of

ELISA-like assays or DARPin-coupled fibrin-targeted drugs for clot lysis. Detection of fragment

E could offer more dynamic measurements compared to other degradation products,

especially D-dimers, in a potential therapeutic setting due to the shorter serum half-life

(Ardaillou et al. 1977; Tanswell et al. 1993).

Finally, ribosome display offers a powerful method to generate a highly diverse binder

library through enrichment of high-affinity binders in a cell-free setup. Irrespective of the

generation method, fewer rounds of binder selection might be beneficial for finding

functional receptors. Although the average binder affinity increases with each cycle, the

covered epitope space might be reduced at the same time. Expanding the receptor platform

to include other types of binders that can be generated via similarly tunable procedures such

as monobodies (Karatan et al. 2004), HEAT-like repeat proteins (Guellouz et al. 2013) and

nanobodies (Skerra et al. 1988),(Ward et al. 1989) would further broaden the range of

possible targets and applications, especially for the detection of small molecules (Lesne et al.

2019). Combinations of different binder selection platforms might also be a promising

strategy for difficult targets. The benefits of such a combination are already visible in this

study in terms of the change in binding site preference that was observed upon employing an

scFv in combination with a DARPin.

We hope that our system will inspire the clinical development of GEMS-based cellular

therapeutics to improve the treatment of patients at risk of thrombosis by reducing harmful

delays in intervention. We envision that FDP-targeted AMBERs will be useful in the

development of novel anti-coagulants as well as therapeutic cells. Our data already

demonstrates the feasibility of producing suitable therapeutics in a human cell line,

suggesting that the AMBER pipeline will be applicable to cell-based therapies in next-

generation medicine.



Acknowledgements

The authors thank Steven Schmitt for help with size-exclusion chromatography and

Ulrike Lanner as well as Alexander Schmidt from the Proteomics Core Facility (PCF) at the

University Basel for their support with mass spectroscopy. We also thank Jens Sobek from the

Functional Genomics Center Zurich (FGCZ) for providing affinity data. We are grateful to

Mariangela Di Tacchio, Alexandra Gumienny and Thomas Horn from the single-cell facility at

the D-BSSE for help with cell sorting. We thank David Fuchs, Adrian Bertschi, Viktor Hällman,

and Pratik Saxena, all from ETH Zurich, for critical readings of the manuscript and valuable

discussions. We thank Craig Hamilton from the University Bern and Birgit Dreier from the

University of Zurich for critical comments on the manuscript. Finally, we want to thank Astrid

Beerlage from the University Hospital Basel and Rebekka Strittmatter from the St. Elisabethen

Hospital Lörrach for advice on medical aspects.

Acknowledgements




Materials & Methods

Chemicals and proteins

Native human D-dimer protein (ab98311), recombinant hirudin protein (ab201396)

and native human plasmin (ab90928) were purchased from Abcam plc. trans-4-

(Aminomethyl)cyclohexanecarboxylic acid (tranexamic acid, TXA, 857653), heparin sodium

salt from porcine intestine (H3149), lyophilized human plasma (P9523), active thrombin from

human plasma (SRP6557), and fibrin from human plasma (F5386) was purchased from Sigma-

Aldrich. Human fibrinogen fragment E (HCI-0150E) and human fibrinogen fragment D (HCI-

0150D) were purchased from Haematologic Technologies, LLC. Soluble human fibrin (FIB-S-

1.0MG) was purchased from Molecular Innovations Inc. Tridegin (opr0062) was purchased

from Covalab S.A.S.

Selection and screening of DARPin binders specific for FDP

To generate DARPin binders for human FDP protein, the biotinylated target protein

was immobilized on either MyOne T1 streptavidin-coated beads (Pierce) or Sera-Mag

neutravidin-coated beads (GE), depending on the selection round. Ribosome display

selections were performed essentially as described (Dreier & Pluckthun, 2012), using a semi-

automatic KingFisher Flex MTP96 well platform.

The library includes a mix of N3C-DARPins with randomized and non-randomized N-

and C-caps respectively, and successively enriched pools were cloned as intermediates in a

ribosome display specific vector (Schilling et al. 2014). Selections were performed over four

rounds with decreasing target concentration and increasing washing steps to enrich high-

affinity binders.

The final enriched pool was cloned as fusions into a bacterial pQE40 derivative vector

with a N-terminal MRGSH8-and C-terminal FLAG tag via unique BamHI x HindIII sites

containing lacIq for expression control. After transformation into E. coli XL1-blue, 380 single

DARPin clones were expressed in 96-well format and lysed by addition of a concentrated Tris-

HCL-based HT-Lysis buffer containing OTG, lysozyme and nuclease. These bacterial crude

extracts of single DARPin clones were subjected to homogeneous time-resolved fluorescence

(HTRF)-based screening to identify potential binders. Binding of the FLAG-tagged DARPins to

streptavidin-immobilized biotinylated FDP protein was measured using FRET (donor:

Materials & Methods


streptavidin-Tb, acceptor: anti-FLAG-d2, Cisbio). Further HTRF cross reactivity measurements

against ‘No Target’ and ‘HSA’ allowed for discrimination of FDP-specific hits.

Among the identified binders, 32 were sequenced and 19 single clones were selected.

The DARPins were expressed on a small scale and purified using a 96-well IMAC column

(HisPurTM Cobald plates, Thermo Scientific). DARPins after IMAC purification were analyzed

at a concentration of 10 µM on a Superdex 75 5/150 GL column (GE Healthcare) using an

Aekta Micro system (GE Healthcare) with PBS containing 400 nM NaCl as the running buffer.

Absorbance at 280 nm was recorded. β-Amylase (200 kDa), bovine serum albumin (66 kDa),

carbonic anhydrase (29 kDa) and cytochrome c (12.4 kDa) were used as molecular mass

standards. 15 DARPins were identified as monomers.

Sequence analysis

Phylogenetic analysis of amino acid sequences was done with a simple phylogeny tool

(www.ebi.ac.uk/Tools/phylogeny/simple_phylogeny/). Sequence alignments were

performed using Clustal O (www.ebi.ac.uk/Tools/msa/clustalo/). Pairwise identity calculation

was carried out using the EMBL-Needle algorithm employing the BLOSUM62 matrix

(www.ebi.ac.uk/Tools/psa/emboss_needle/). All algorithms are part of the EMBL-EBI search

and sequence analysis tools API (Madeira et al. 2019).

Plasmid preparation

A comprehensive list of all plasmids used in this study can be found in table S1.

Plasmids were generated using conventional molecular cloning techniques.

Polymerase chain reaction (PCR) was performed using Phusion® polymerase (F530,

ThermoFisher Scientific) or Q5® high-fidelity polymerase (M0491, New England Biolabs, NEB)

following the manufacturers’ recommendations. Cleavage of PCR products and plasmids was

done with restriction endonucleases (New England Biolabs NEB, HF enzymes were used if

applicable) and ligated with T4 DNA ligase (EL0011, ThermoFisher Scientific).

SDS-PAGE and Coomassie staining

Samples were heated for 15 minutes at 70°C prior to loading. Up to 20 µL of sample

was loaded per well of a 12-well or 15-well polyacrylamide gel (Bolt 4-12% Bis-Tris Plus gels,



Invitrogen). SDS-PAGE was performed by using 1x Bolt MOPS SDS Running Buffer (Invitrogen)

in a XCell SureLock® Mini-Cell (Invitrogen) following the instructions of the manufacturer.

PageRuler Plus Prestained Protein Ladder (Invitrogen) was used as a size reference.

Gels were washed three times with water after the run to remove SDS from the running

buffer, then stained with Coomassie (17% (v/v) methanol, 3.3% (v/v) acetic acid, and 0.08%

(w/v) Coomassie Brilliant Blue G250) for >1 h at RT. Gels were washed and destained in water

for several days before use in mass-spectrometric analysis.

Cell culture

HEK293T/17 (HEK293T) cells (ATCC® CRL-11268™) were purchased from the American

Type Culture Collection and grown in high-glucose Dulbecco's modified Eagle’s medium

containing GlutaMAX (DMEM, high glucose, GlutaMAX Supp, 61965026, ThermoFisher

Scientific) supplemented with 10 % fetal bovine serum (FBS, F7524, Sigma-Aldrich) at 37°C in

a humidified incubator with an 7.5% CO2 atmosphere. Cells were grown in 10 mL of DMEM at

a seeding density of 1.5 * 106 cells in a 10 cm culture dish and kept at < 80 % confluency.

For transfection, cells were seeded in 15 mL of DMEM supplemented with 1x

penicillin/streptomycin (L0022, Biowest) in 96-well culture plates (15000 cells per well).

Stable HEK293T cells were generated using the Sleeping Beauty transposon system (Kowarz

et al. 2015). Donor plasmids bearing a selection marker (resistance genes for puromycin (pac),

blasticidin (bsr) or zeocin (ble)) driven by a constitutive promoter and flanked by recognition

sites of the Sleeping Beauty transposase were co-transfected with an expression plasmid for

Sleeping Beauty transposase. For selection of stable clones, the cell culture medium was

supplemented with puromycin (2 µg/mL), blasticidin (4 µg/mL) or zeocin (20 µg/mL) or

combinations of them, depending on the types of plasmids to be integrated.

Transfections and stable cell line generation

Transfection was done in a 96-well plate by mixing 200 ng of plasmid DNA and 1.2 µL

of 1 mg/mL polyethyleneimine (PEI, 24765-1, Polysciences Inc.) per well in FCS-free DMEM.

The mixture was incubated for 5-10 minutes, added to each well and incubated with the cells

overnight. For all transient expression experiments, 2 ng of each receptor plasmid was mixed

with 30 ng of plasmid pLS13 and 30 ng of pLS15. Amounts of DNA were adjusted to 200 ng

Materials & Methods


using the inert filler plasmid pDF101. A comprehensive list of plasmids used in each

experiment can be found in Table S2.

For stable cell line generation, 6-well plates were used. Here 250 000 cells per well

were seeded and transfected 24 h later. A total of 1 µg of plasmid DNA was mixed with 6 µL

of 1 mg/mL PEI in FCS-free DMEM. For two vectors to be integrated, 300 ng of each

integration vector were mixed with 200 ng of Sleeping Beauty expression vector pTS395 and

200 ng of pDF101.

For three integration vectors, 200 ng of each integration vector was mixed with 200

ng of Sleeping Beauty expression vector pTS395 and 200 ng of pDF101.

Cells were grown for one day after transfection prior to start selection, depending on the

selection marker used, with a final concentration of 2 µg/mL of puromycin, 4 µg/mL of

blasticidine S or 20 µg/mL of zeocine.

Stable monoclonal cell lines were produced from mixed populations by FACS-

mediated single-cell seeding using a BD Aria III FACS.

Reporter Assays

To measure the activity of secreted human placental alkaline phosphatase (SEAP), the

increase of absorbance at 405 nm due to hydrolysis of para-nitrophenyl phosphate (pNPP) in

the cell culture supernatant was followed (1 reading per minute for 25 minutes). To this end,

60-100 µL of supernatant was collected from each well and heat-inactivated for 30 minutes

at 65 °C. 180 µL of assay reagent containing 100 µL 2x SEAP buffer (20 mM homoarginine,

1 mM MgCl2, 21% (v/v) diethanolamine, pH 9.8), 20 µL of pNPP substrate solution (20 mM, in

2x SEAP buffer) and 60 µL of water were added to 20 µL of heat-inactivated supernatant. The

absorbance at 405 nm at 37 °C was monitored with a Tecan M1000 multiplate reader (Tecan

AG) and used to calculate SEAP activity.

Cell culture supernatant was also used to determine the activity of secreted nano-

luciferase with the Nano-Glo® Luciferase Assay System (N1110, Promega). To this end, 7.5 µL

of cell culture supernatant was mixed with 7.5 µL buffer/substrate mix per well in a 384-well

black-well plate. Luminescence of the reporter was measured using a Tecan M1000 multiplate

reader (Tecan AG).



Size-exclusion chromatography of reconstituted plasma

Reconstituted human plasma was filtered through a 0.2 µm syringe filter to clear

liposomes and loaded on a self-packed Sephadex-200 column equilibrated with D-PBS

containing 2 mM EDTA to prevent clotting of the column. The run was performed in the same

buffer and a total of 133 fractions of 1 mL were collected. Selected fractions were further

tested for their activity in cell culture as well as analyzed with SDS-PAGE and western blotting.

Mass spectrometry

In-gel digestion of relevant SDS-PAGE bands was performed by using trypsin in 100

mM ammonium bicarbonate buffer. Samples were purified by using C18 columns, dried, and

resuspended in 20 µL of a suitable buffer. Concentrations were adjusted to be approximately

0.1 µg/µL and 1 µg of the solution was subjected to mass spectrometric analysis on an

Orbitrap Elite system.

tPA activity measurements

A commercial kit was used to assess tPA activity following the manufacturer’s

guidelines (Biovision, #K178-100) with an adaptation of the protocol for 384-well plates. Here,

we adjusted the amount of sample to 6 µL and scaled down the amount of inhibitor cocktail

and substrate mix accordingly to 2 µL each. Plates were centrifuged in-between steps at 250

rcf to collect samples at the bottom of the wells. We followed the protocol to determine the

activity of tenecteplase (TNK) by recording the absorbance of the emerging pNA product at

405 nm over 1 h. A calibration curve was prepared using authentic pNA.

Statistics

All analyses were done with GraphPad Prism 8.4. Representative graphs showing n = 3

biologically independent samples are presented as bar diagrams ± SD. Unless indicated

otherwise, no statistical analysis was performed. Graphs in figure 2 show the mean ± SEM of

N=3 independent experiments performed in triplicates. No statistical analysis was performed.

To determine the level of significance in figure 4f a two-tailed unpaired t-test was performed

(t = 7.585, df = 4) using GraphPad Prism 8.4 and the p value was calculated to be 0.0016.

Materials & Methods




Supplementary Information Table S1: Plasmids used in this work

Plasmid Description Reference

005-546-

1295-

M_A6

Bacterial expression plasmid bearing an inducible promoter for

expression of DARPin-A6

This work

005-546-

1295-

M_A10



This work

005-546-

1295-

M_A12



This work

005-546-

1295-

M_C11


expression of DARPin-C11

This work

005-546-

1295-

M_F9


expression of DARPin-F9

This work

005-546-

1296-

M_A11



This work

005-546-

1296-

M_B4


expression of DARPin-B4

This work

005-546-

1296-

M_F6



This work

005-546-

1296-

M_G1


expression of DARPin-G1

This work

Supplementary Information


005-546-

1297-

M_A7



This work

005-546-

1297-

M_B10



This work

005-546-

1297-

M_B12



This work

005-546-

1297-

M_C4


expression of DARPin-C4

This work

005-546-

1297-

M_D12


expression of DARPin-D12

This work

005-546-

1297-

M_E11


expression of DARPin-E11

This work

005-546-

1298-

M_D11


expression of DARPin-D11

This work

005-546-

1298-

M_F8



This work

005-546-

1298-

M_F10



This work

005-546-

1298-

M_F12



This work



pDF101 Inert filler plasmid bearing a bacterial T7 promoter driving an

inactive ribozyme (PT7-SpAL-sTRSVac)

(Auslander

et al. 2016)

pLS13 STAT3-reporter plasmid with 2 STAT3 response elements (RE)

followed by a minimal promoter driving expression of human

secreted placental alkaline phosphatase (SEAP) (RE2-PhCMVmin-

SEAP-pA)

(Schukur et

al. 2015)

pLS15 Mammalian expression vector bearing the coding sequence of the

human STAT3 transcription factor under control of an hCMV

promoter (PhCMV-STAT3-pA)

(Schukur et

al. 2015)

pMM1 Mammalian expression vector with a modified MCS (PhCMV-MCS-

pA; MCS, EcoRI-ATG-SpeI-NheI-BamHI-STOP-XbaI-HindIII-FseI-

pA).

(Muller et

al. 2017)

pTS379 Mammalian expression vector bearing a PhCMV-driven EpoR

receptor equipped with an anti-D-dimer scFv described by

Laroche et al. 1991. pLeo644 (Scheller et al. 2018) was used as a

backbone. (PhCMV-Igk-scFv-EpoR-IL6st-pA)

This work


receptor equipped with a DARPin as a D-dimer binding moiety.

Additional labels indicate the DARPin used as the binding moiety.

pLeo644 (Scheller et al. 2018) was used as a backbone (PhCMV-Igk-

DARPin-EpoR-IL6st-pA).

This work

pTS395 PhCMV-driven Sleeping Beauty transposase mammalian expression

vector (PhCMV-SB100-pA).

This work

pTS441 STAT3-reporter plasmid with 2 STAT3 response elements (RE)



SEAP-pA) in pMM1 backbone.

This work





This work







This work





This work

pTS835 Stable Sleeping Beauty integration vector bearing three cassettes;

an PhCMV-driven EpoR receptor equipped with the anti-D-dimer

scFv, an PSV40-driven STAT3 and a PRPBSA-driven selection cassette

encoding the blue fluorescent protein mTagBFP2 fused via a p2a

peptide sequence to a puromycin resistance gene. (PhCMV-Igk-

scFv-EpoR-IL6st-pA-PSV40-STAT3-pA-PRPBSA-mTagBFP2-p2a-PuroR-

pA)

This work


receptor equipped with DARPin-A7. pMM1 was used as a

backbone. (PhCMV-Igk-DARPinA7-EpoR-IL6st-pA)

This work


receptor equipped with DARPin-B4. pMM1 was used as a

backbone. (PhCMV-Igk-DARPinB4-EpoR-IL6st-pA)

This work


receptor equipped with DARPin-G1. pMM1 was used as a

backbone. (PhCMV-Igk-DARPinG1-EpoR-IL6st-pA)

This work


receptor equipped with a tandem fusion of DARPin-A7 on top of

DARPin B4. pMM1 was used as a backbone. (PhCMV-Igk-DARPinA7-

DARPinB4-EpoR-IL6st-pA)

This work



This work



DARPin G1. pMM1 was used as a backbone. (PhCMV-Igk-DARPinA7-

DARPinG1-EpoR-IL6st-pA)


receptor equipped with a tandem fusion of DARPin-B4 on top of

DARPin A7. pMM1 was used as a backbone. (PhCMV-Igk-DARPinB4-

DARPinA7-EpoR-IL6st-pA)

This work



DARPin G1. pMM1 was used as a backbone. (PhCMV-Igk-DARPinB4-


This work


receptor equipped with a tandem fusion of DARPin-G1 on top of

DARPin A7. pMM1 was used as a backbone. (PhCMV-Igk-DARPinG1-


This work



DARPin B4. pMM1 was used as a backbone. (PhCMV-Igk-DARPinG1-


This work



DARPin A7. pMM1 was used as a backbone. (PhCMV-Igk-DARPinA7-


This work



DARPin B4. pMM1 was used as a backbone. (PhCMV-Igk-DARPinB4-


This work



DARPin G1. pMM1 was used as a backbone. (PhCMV-Igk-DARPinG1-


This work



pTS914 Stable Sleeping Beauty integration vector bearing two cassettes;

a PhCMV-driven EpoR receptor equipped with DARPin B4 and a

PRPBSA-driven selection cassette encoding a selection marker for

blasticidin resistance. (PhCMV-Igk-DARPinB4-EpoR-IL6st-pA-

PRPBSA-BlastR-pA)

This work


receptor equipped with the anti-D-dimer scFv. pMM1 was used as

a backbone. (PhCMV-Igk-scFv-EpoR-IL6st-pA)

This work


receptor equipped with a tandem fusion of the anti-D-dimer scFv

on top of DARPin-A7. pMM1 was used as a backbone. (PhCMV-Igk-

scFv-DARPinA7-EpoR-IL6st-pA)

This work



on top of DARPin-B4. pMM1 was used as a backbone. (PhCMV-Igk-

scFv-DARPinB4-EpoR-IL6st-pA)

This work



on top of DARPin-G1. pMM1 was used as a backbone. (PhCMV-Igk-

scFv-DARPinG1-EpoR-IL6st-pA)

This work


an PhCMV-driven EpoR receptor equipped with a tandem fusion of

the anti-D-dimer scFv on top of DARPin-B4, an PSV40-driven STAT3

and a RPBSA-driven selection cassette encoding the blue

fluorescent protein mTagBFP2 fused via a p2a peptide sequence

to a selection marker for puromycin resistance. (PhCMV-Igk-scFv-

DARPinB4-EpoR-IL6st-pA-PSV40-STAT3-pA-PPRBSA-mTagBFP2-p2a-

PuroR-pA)

This work


an PhCMV-driven EpoR receptor equipped with an scFv against the

industrial dye reactive red (RR120) as described in Scheller et al

This work



2018, an PSV40-driven STAT3 and a PRPBSA-driven selection cassette

encoding the blue fluorescent protein mTagBFP2 fused via a p2a

peptide sequence to a selection marker for puromycin resistance.

(PhCMV-Igk-scFv(RR120)-EpoR-IL6st-pA-PSV40-STAT3-pA-PRPBSA-

mTagBFP2-p2a-PuroR-pA)


four repeats of the STAT3 response element sequence followed

by a minimal promoter driving expression of a secreted nano-

luciferase fused to a murine Fc (mFc) tag separated by a cleavage

site for furin from an also Fc-stabilized copy of tenecteplase (TNK).

A second cassette drives expression of a selection marker for

zeocin resistance fused to the yellow fluorescent protein YPet via

a p2a peptide sequence.

(RE4-PhCMVmin-Igk-Nluc-mFc-Furin-TNK-mFc-pA-PRBSA-ZeoR-p2a-

YPet-pA)

This work


receptor equipped with an scFv against the industrial dye reactive

red (RR120) as described in Scheller et al 2018. This plasmid was

used as negative control. pMM1 was used as a backbone. (PhCMV-

Igk-scFv(RR120)-EpoR-IL6st-pA)

This work


four repeats of the STAT3 operator sequence followed by a

minimal promoter driving expression of a secreted nano-

luciferase fused to a murine Fc (mFc) tag separated by a p2a site

that separates translation of Nluc and hirudin-HM2. A second

cassette drives expression of a selection marker for zeocin

resistance fused to the yellow fluorescent protein YPet via a p2a

peptide sequence.

(RE4-PhCMVmin-Igk-Nluc-mFc-p2a-HIRM2-pA-PRBSA-ZeoR-p2a-

YPet-pA)




receptor equipped with a double tandem fusion of the anti-D-

dimer scFv. pMM1 was used as a backbone.

(PhCMV-D-dimer-scFv-D-dimer-scFv-receptor-pA)

This work



the anti-D-dimer scFv. pMM1 was used as a backbone.

(PhCMV-DARPinA7-D-dimer-scFv-receptor-pA)

This work




(PhCMV-DARPinB4-D-dimer-scFv(tandem)-receptor-pA)

This work


receptor equipped with a tandem fusion of DARPin-G1on top of


(PhCMV-DARPinG1-D-dimer-scFv(tandem)-receptor-pA)

This work



Table S2: Plasmids transfected in each experiment

Details of the transfection protocol can be found in the methods.

Figure Plasmids used

1 Plasmid library for loop DARPins

2

b)

pTS380-A12: PhCMV-DARPinA12-receptor-pA

pTS380-B4: PhCMV-DARPinB4-receptor-pA

pTS380-D11: PhCMV-DARPinD11-receptor-pA

pTS380-C11: PhCMV-DARPinC11-receptor-pA

pTS380-F12: PhCMV-DARPinF12-receptor-pA













pTS380-E11: PhCMV-DARPinE11-receptor-pA

pTS380-G1: PhCMV-DARPinG1-receptor-pA

pTS379: PhCMV-D-dimer-scFv-receptor-pA

pTS2011: PhCMV-RR120 -scFv-receptor-pA

2 ng of the indicated receptor plasmids,

30 ng of each pLS13 and pLS15 used to build the reporter system,

pDF101 to adjust plasmid DNA amount to 200 ng per transfection

2

c) – d)

f) – g)












2

e)

h)

pTS870: PhCMV-DARPinG1-A7(tandem)-receptor-pA

pTS871: PhCMV-DARPinG1-B4(tandem)-receptor-pA

pTS872: PhCMV-DARPinG1-G1(tandem)-receptor-pA

pTS930: PhCMV-D-dimer-scFv-DARPinA7(tandem)-receptor-pA

pTS931: PhCMV-D-dimer-scFv-DARPinB4(tandem)-receptor-pA

pTS932: PhCMV-D-dimer-scFv-DARPinG1(tandem)-receptor-pA

pTS2011: PhCMV-RR120-scFv-receptor-pA




3

b)-d)

pTS864: PhCMV-DARPinB4-receptor-pA







4

a) – b)

Stable cell line HEKscFv-B4/TNK was generated using the following plasmids

pTS941: PhCMV-D-dimer-scFv-DARPinB4(tandem)-receptor-pA-PSV40-STAT3-pA-

PhCMV-TagBFP2-p2a-PuroR-pA

pcTS992: RE4-PhCMV-IgkSS-Nluc- mFc-RARYKR-TNK- mFc-PhCMV-ZeoR-p2a-YPet-pA

pcTS395: PhCMV-SleepingBeauty-transposase-pA

pDF101 to adjust plasmid DNA amount to 1 µg per transfection

4

c) – e)

Stable cell line mHEKB4/scFv/hirudin was generated using the following plasmids



pTS835: PhCMV-D-dimer-scFv-receptor-pA-PSV40-STAT3-pA-PhCMV-TagBFP2-p2a-

PuroR-pA

pTS914: PhCMV-DARPinB4-receptor-pA-PhCMV-BlastR-pA

pTS2151: RE4-PhCMV-IgkSS-Nluc-mFc-p2a-HIRM2-pA-PhCMV-ZeoR-p2a-YPet-pA



S3 Plasmid library for loop DARPins

S4

a)





pLS13 and pLS15 used to build the reporter system


S4

b)

pLeo615: PhCMV-RR120-scFv-receptor-pA



S4

c)

e)

100 ng of each of the following plasmids was used per transfection

pSEAP2ctr: PSV40-SEAP-pA

pFS29: PSV40-mCherry-pA

S5 pTS380-A12: PhCMV-DARPinA12-receptor-pA

























S5

a) – d)

























S6 pTS380-A12: PhCMV-DARPinA12-receptor-pA





















S8

a)

b)

pTS863: PhCMV-DARPinA7-receptor-pA


pTS865: PhCMV-DARPinG1-receptor-pA

pTS866: PhCMV-DARPinA7-B4(tandem)-receptor-pA

pTS867: PhCMV-DARPinA7-G1 (tandem)-receptor-pA

pTS868: PhCMV-DARPinB4-A7(tandem)-receptor-pA

pTS869: PhCMV-DARPinB4-G1(tandem)-receptor-pA

pTS870: PhCMV-DARPinG1-A7(tandem)-receptor-pA




pTS872: PhCMV-DARPinA7-A7(tandem)-receptor-pA

pTS873: PhCMV-DARPinB4-B4(tandem)-receptor-pA

pTS874: PhCMV-DARPinG1-G1(tandem)-receptor-pA




S8

c)

d)


pTS2166: PhCMV-DARPinA7-D-dimer-scFv(tandem)-receptor-pA

pTS2167: PhCMV-DARPinB4-D-dimer-scFv(tandem)-receptor-pA

pTS2168: PhCMV-DARPinG1-D-dimer-scFv(tandem)-receptor-pA

pTS930: PhCMV-D-dimer-scFv-DARPinA7(tandem)-receptor-pA


pTS932: PhCMV-D-dimer-scFv-DARPinG1(tandem)-receptor-pA

pTS2165: PhCMV-D-dimer-scFv-D-dimer-scFv(tandem)-receptor-pA




S9

a)-c)








S10

a)









S10

b)

c)






S11

a)





S12

a)




pSEAP2control: PSV40-SEAP-pA


S12

b-c)







S12

d-e)









S13 pTS2011: PhCMV-RR120-scFv-receptor-pA





S14

a)



pTS539: PSV40-STAT3-pA

pLS13: RE2-phCMVmin-SEAP-pA

pTS441: RE2-phCMVmin-SEAP-pA





S14

b)

Stable cell line HEKscFv-B4/TNK and HEKRR120/TNK were generated using the following

plasmids

HEKscFv-B4/TNK:

pTS941: PhCMV-D-dimer-scFv-DARPinB4(tandem)-receptor-pA-PSV40-STAT3-pA-

PhCMV-TagBFP2-p2a-PuroR-pA

pcTS992: RE4-PhCMV-IgkSS-Nluc-mFc-RARYKR-TNK-mFc-PhCMV-ZeoR-p2a-YPet-pA

HEKRR120/TNK:

pTS942: PhCMV-RR120-scFv-receptor-pA-PSV40-STAT3-pA-PhCMV-TagBFP2-p2a-

PuroR-pA

pcTS992: RE4-PhCMV-IgkSS-Nluc-mFc-RARYKR-TNK-mFc-PhCMV-ZeoR-p2a-YPet-pA





Supplementary Figures

Supplementary Figure 1: Result of non-reducing polyacrylamide gel-electrophoresis of the

target protein (“native human D-dimer protein”, Abcam, ab98311) used in all experiments,

done with a 6 % gel. Analysis suggests that the mixture consists primarily of D-dimer protein

with significant amounts of other proteins. The putative identity of the protein in each band

is shown in the right-hand column. M: PageRuler Prestained Protein Ladder, 10 to 180 kDa.

180 kDa D-dimer

Fragment D

D-E fragments (fragment Y)

high MW FD P, fibrin

Fragment E

hSA

M 10 µ

g2

µg

130 kDa

100 kDa

70 kDa

55 kDa



005-546-1296-M_F6 MRGSHHHHHHHHGSDLGKKLLEAATSGQDDEVRILMANGADVNAMDHWGWTPLHLAAIEG 60005-546-1295-M_F9 MRGSHHHHHHHHGSDLGKKLLEAARAGQDDEVRILMANGADVNAQDIWGWTPLHLAAYAG 60005-546-1296-M_G1 MRGSHHHHHHHHGSDLGKKLLEAARAGQDDEVRILMANGADVNAT--------------- 45005-546-1297-M_A7 MRGSHHHHHHHHGSDLGKKLLEAARAGQDDEVRILMANGADVNAF--------------- 45005-546-1295-M_A6 MRGSHHHHHHHHGSDLGKKLLEAARAGQDDEVRILMANGADVNAE--------------- 45005-546-1296-M_B4 MRGSHHHHHHHHGSDLGKKLLEAARAGQDDEVRILMANGADVNAE--------------- 45 ************************ :******************

005-546-1296-M_F6 HQEIVEVLLKTGADVNAKDQWGATPLHLAAVVGHLEIV---------------------- 98005-546-1295-M_F9 HLEIVKVLLKTGADVNAYDDWGSTPLHLAAWIGHLEIVEVLLKAGADVNAWDVHGFTPLH 120005-546-1296-M_G1 ------------------DNWGDTPLHLAAWHGHLEIVEVLLKTGADVNAQDIIGATPLH 87005-546-1297-M_A7 ------------------DWYGTTPLHLAAFNGHLEIVEVLLKTGADVNAQDLFGNTPLH 87005-546-1295-M_A6 ------------------DWYGTTPLHLAAHNGHLEIVEVLLKTGADVNAQDLFGNTPLH 87005-546-1296-M_B4 ------------------DWYGTTPLHLAAHNGHLEIVEVLLKTGADVNAQDLFGNTPLH 87 * :* ******* ******

005-546-1296-M_F6 -----------EVLLKHGADVNAQDISGQTPFDLAAWHGNEDIAEVLQKAAKLNDYKDDD 147005-546-1295-M_F9 LAAIRGHLEIVEVLLKHGADVNAQDKFGKTPFDLAIDNGNEDIAEVLQKAAKLNDYKDDD 180005-546-1296-M_G1 LAAIMGHLEIVEVLLKAGADVNAQDKFGKTPFDLAIDNGNEDIAEVLQKAAKLNDYKDDD 147005-546-1297-M_A7 LAAWNGHLEIVEVLLKHGADVNAQDKFGKTPFDLAIDNGNEDIAEVLQKAAKLNDYKDDD 147005-546-1295-M_A6 LAAYEGHLEIVEVLLKHGADVNAQDKFGKTPFDLAIDNGNEDIAEVLQKAAKLNDYKDDD 147005-546-1296-M_B4 LAAWNGHLEIVEVLLKHGADVNAQDKFGKTPFDLAIDNGNEDIAEVLQKAAKLNDYKDDD 147 ***** ******** *:****** :**********************

005-546-1296-M_F6 DK* 149005-546-1295-M_F9 DK* 182005-546-1296-M_G1 DK* 149005-546-1297-M_A7 DK* 149005-546-1295-M_A6 DK* 149005-546-1296-M_B4 DK* 149 ***

b) CLUSTAL O (1.2.4) multiple sequence alignment

a) Simple phylogenetic analysis of DARPin amino acid sequence

identity similarityF6 86.0 % 90.0 %F9 74.9 % 78.7 %G1 92.0 % 94.0 %A7 97.3 % 98.0 %A6 100.0 % 100.0 %B4 98.7 % 99.3 %



Supplementary Figure 2: Characterization of DARPin binders. a) Simple phylogenetic

analysis of all DARPins based on their amino acid sequences. DARPins G1, F6, F9, A7, A6 and

B4 are boxed in red. DARPins G1, F6, F9, A7, A6 and B4 were further analyzed by b)

alignment of amino acid sequences using the Clustal Omega 1.2.4 algorithm and c)

similarity analysis using pairwise alignments employing the EMBOSS-needle algorithm.



Supplementary Figure 3: Size exclusion chromatography was performed to assess the

dimerization behavior of selected DARPins. Chromatograms in red were found to represent

non-monomeric candidates.



Supplementary Figure 4: Control experiments to assess the cross-reactivity of receptors and

inputs as well as the toxicity of inducers. HEK293T cells were transfected overnight with the

indicated receptors and the pLS13 STAT3 reporter plasmid alongside pLS15 for STAT3

transcription factor overexpression or a constitutive SEAP expressing plasmid (pSEAP2ctr),

as described in the methods. a) None of the receptor scaffolds used is responsive to

erythropoetin (EPO), which is the natural ligand for the native EPO receptor. Also, the STAT3

pathway is not induced by EPO administration. Cells were incubated with 10 ng/mL EPO

a) EPO insensitivity of AMBERs b) RR120-receptor as negative control

c) productivity is not influenced by xFDPs d) xFDPs are non-toxic at relevant concentrations

e ) plasma influences cell productivity f ) plasma shows dose-dependent toxicity

pTS380-B4

pTS380-B4 +

pTS379

pTS864

reporte

r ctr

0

5

10

15

SEAP

[UL-1

]+ EPO (10 ng/mL)

uninduced

1 % plasm

a

20g/m

L fibrin

ogen

5 µg/m

L xFDPs

100 n

g/mL RR12

0

10 ng/m

L RR120

2 ng/m

L RR120

uninduced

0

5

10

15

20

SEAP

[UL-1

]

10g/m

L

5g/m

L

2g/m

L

1g/m

L

0.5g/m

L

uninduced

0

50

100

150

SEAP

[UL-1

]

10g/m

L

5g/m

L

2g/m

L

1g/m

L

0.5g/m

L

uninduced

0

10000

20000

30000

40000

50000re

soru

fin [F

U]

50%

25%

20%

10% 5% 2.5

% 1% 0%0

10000

20000

30000

40000

reso

rufin

[FU]

50%

25%

20%

10% 5% 2.5

% 1% 0%0

10

20

30

40

SEAP

[UL-1

]



protein in complete DMEM containing 10 % FCS for 24 h and the supernatant was sampled

for analysis of SEAP reporter activity. b) The RR120 receptor used as the negative control

exclusively responds to its ligand RR120 and is not activated by plasma, fibrinogen or xFDPs.

c-f) Productivity and viability of HEK293T cells were assessed in c) and e) in terms of

constitutive expression of SEAP and in d) and f) by measuring resazurin reduction, which

leads to production of the fluorescent dye resorufin, in parallel. HEK293T cells were

transfected overnight with pSEAP2ctr prior to incubation with c)-d) xFDPs or e)-f) plasma at

the indicated concentrations. At 24 h after induction, the supernatant was sampled and

SEAP activity was measured. After sampling, the medium was exchanged for 100 µL of

complete DMEM containing 10 % FCS and resazurin. Cells were incubated with resazurin for

approximately 30 minutes and 60 µL of medium was transferred to a clear-bottomed assay

plate for fluorescence recording. All values are means ± SD of triplicates.



Supplementary Figure 5: Screening of homodimeric receptors. HEK293T cells were

transfected overnight with the indicated receptors and the pLS13 STAT3 reporter plasmid

alongside pLS15 for STAT3 transcription factor overexpression, as described in the methods.

Cells were incubated with 0.5 µg/mL xFDPs or 0.5 % (v/v) reconstituted human plasma in

complete DMEM containing 10 % FCS for 24 h, and supernatant was sampled for analysis

a) Overview

b)

d)

c)

A12-A12

A10-A10

B12-B12

C4-C4

A12-scF

v

A10-F12

B12-F8

C4-A11

A12-G

1

A10-C

11

B12-A6

C4-A7

A12-E11

A10-D

11

B12-F10

C4-F9

A12-F6

A10-B4

B12-D12

C4-B10

A12-C4

A10-A12

B12-A10

C4-B12

A12-A

11

A10-scF

v

B12-F12

C4-F8

A12-A7

A10-G

1

B12-C

11C4-A

6

A12-F9

A10-E11

B12-D

11

C4-F10

A12-B10

A10-F6

B12-B4

C4-D12

A12-B12

A10-C4

RR1200

20

40

60

SEAP

[U*L

-1]

plasma xFDP uninduced

A12-A12B4-B4

D11-D

11

C11-C

11

F12-F12

A10-A10

D12-D12

F10-F10A6-A6

F8-F8

B12-B12

B10-B10F9-F

9A7-A7

A11-A

11C4-C4

F6-F6

E11-E11G1-G

1

scFv

-scFv

A12-scFv

B4-A12

D11-B4

C11-D

11

F12-C11

A10-F12

D12-A10

F10-D12

A6-F10

F8-A6

B12-F8

B10-B12

F9-B10A7-F9

A11-A7

C4-A11F6-C

4

E11-F6

G1-E11

scFv

-G1

A12-G1

B4-scFv

D11-A

12

C11-B4

F12-D11

A10-C11

D12-F12

F10-A10

A6-D12

F8-F10

B12-A6

B10-F8

F9-B12

A7-B10

A11-F9C4-A7

F6-A11

E11-C4G1-F

6

scFv

-E11

A12-E11B4-G1

D11-scF

v

C11-A

12

F12-B4

A10-D11

D12-C11

F10-F12

A6-A10

F8-D12

RR120pS2ct

r0

20

40

60

SEAP

[U*L

-1]

xFDP uninducedplasma

B12-F10

B10-A6F9-F

8

A7-B12

A11-B

10C4-F9

F6-A7

E11-A11G1-C

4

scFv

-F6

A12-F6

B4-E11

D11-G1

C11-scF

v

F12-A12

A10-B4

D12-D11

F10-C11

A6-F12

F8-A10

B12-D12

B10-F10F9-A

6A7-F8

A11-B

12

C4-B10F6-F

9

E11-A

7

G1-A11

scFv

-C4

A12-C4B4-F6

D11-E

11

C11-G1

F12-scF

v

A10-A12

D12-B4

F10-D11

A6-C11

F8-F12

B12-A10

B10-D12

F9-F10A7-A6

A11-F8

C4-B12

F6-B10

E11-F9G1-A

7

scFv

-A11

A12-A11B4-C4

D11-F6

C11-E

11

F12-G1

A10-scFv

D12-A12

F10-B4

A6-D11

F8-C11

B12-F12

B10-A10

F9-D12

A7-F10

A11-A

6C4-F8

F6-B12

E11-B

10G1-F

9

scFv

-A7

RR120pS2ct

r

0

20

40

60

SEA

P [U

*L-1

]

plasma xFDP uninduced

A12-A7

B4-A11

D11-C4

C11-F6

F12-E11

A10-G1

D12-scFv

F10-A12A6-B4

F8-D11

B12-C11

B10-F12

F9-A10

A7-D12

A11-F10C4-A6

F6-F8

E11-B

12

G1-B10

scFv

-F9

A12-F9

B4-A7

D11-A

11

C11-C4

F12-F6

A10-E11

D12-G1

F10-scF

v

A6-A12F8-B

4

B12-D11

B10-C11

F9-F12

A7-A10

A11-D

12

C4-F10

F6-A6

E11-F8

G1-B12

scFv

-B10

A12-B10B4-F9

D11-A7

C11-A

11

F12-C4

A10-F6

D12-E11

F10-G1

A6-scFv

F8-A12

B12-B4

B10-D11

F9-C11

A7-F12

A11-A

10

C4-D12

F6-F10

E11-A6G1-F

8

scFv

-B12

A12-B12

B4-B10

D11-F9

C11-A7

F12-A11

A10-C4

D12-F6

F10-E11A6-G1

F8-scFvRR120

pS2ctr

0

20

40

60

SEAP

[U*L

-1]

xFDP uninducedplasma



of SEAP reporter activity. All values are means ± SD of triplicates. a) Overview of all

measurements. b-d) Subsets of a) of 70 combinations each. b) 1-70, c) 71-140, d) 141-210.

All panels include RR120 as a negative control and constitutive expression of SEAP as a

reference.



Supplementary Figure 6: Correlation of receptor activation by plasma of DARPin-AMBERs

in homodimeric configuration measured by SEAP reporter expression with the

corresponding signal intensities from HTRF (left y-axis) as well as the number of functional

receptor combinations (fold induction >5) for coagulation or xFDP detection (right y-axis).

HTRF signals are single values while receptor activation is represented by mean values of

triplicate determinations.

0 5 10 15 202.0

2.5

3.0

3.5

4.0

0

5

10

15

20

receptor signal (SEAP) [U*L-1]

HTR

F ra

tio [A

U]

# functional receptor comb.

HTRF signal

number of functional receptors - plasma

number of functional receptors - xFDPs



Supplementary Figure 7: Affinities of binders were assessed by surface plasmon resonance

(SPR) measurements on a Biacore T200 (GE healthcare) using a CMD200M chip. Duplicates

of full kinetic measurements were performed and fitted with a heterogenous ligand (hetlig)

model using BIAevaluation software (GE healthcare). Flow cell 1 (FC1) served as reference

for FC2 and was coated with 205 RU of target protein. Each concentration was measured

twice in ascending and descending order.

0 50 100 150 2000

20

40

60

t [sec]

dRU

[RU

]

0 200 4000

20

40

60

t [sec]

dRU

[RU

]

0 200 4000

20

40

60

t [sec]

dRU

[RU

]

0 100 200 3000

10

20

30

40

t [sec]

dRU

[RU

]

0 200 4000

10

20

30

40

Lo

remip

sum

dolor

sita

met ,cons ect e tue r adipi s cin g elit,

t [sec]

dRU

[RU

]

a) scFv b) DARPin A6

c) DARPin A7

e) DARPin G1

d) DARPin B4

kON

-1/2 = 633,084.2 / 16,502.9 M-1s-1

kOFF

-1/2 = 0.381 / 0.017 s-1

Kd-1/2 = 602 / 1031 nM

RUmax

= 14.6 / 15.2

kON

-1/2 = 146,625.2 / 1,410,638.3 M-1s-1

kOFF

-1/2 = 0.013 / 0.062 s-1

Kd-1/2 = 91 / 44 nM

RUmax

= 2.6 / 20.8

kON

-1/2 = 1,421,090.6 / 181,810.2 M-1s-1

kOFF

-1/2 = 0.033 / 0.011 s-1

Kd-1/2 = 23 / 59 nM

RUmax

= 20.2 / 5.1

kON

-1/2 = 2,515,538 / 375,788.4 M-1s-1

kOFF

-1/2 = 0.035 / 0.014 s-1

Kd-1/2 = 14 / 37 nM

RUmax

= 12.0 / 13.2

kON

-1/2 = 103,836.1 / 48,723 M-1s-1

kOFF

-1/2 = 0.025 / 0.008 s-1

Kd-1/2 = 244 / 169 nM

RUmax

= 12.6 / 12.9



Supplementary Figure 8: Screening of tandem receptors based on DARPins only. (a-b) or

scFv-DARPin combinations (c-d). HEK293T cells were transfected overnight with the

indicated receptors and the pLS13 STAT3 reporter plasmid alongside pLS15 for STAT3

transcription factor overexpression, as described in the methods. a) and c) Cells were

incubated for 24 h in complete DMEM containing 10 % FCS and xFDPs at 0.5 µg/mL or

a) DARPin-based tandem receptors induced with plasma and xFDPs

b) fold induction of DARPin-based tandem receptors induced with plasma and xFDPs

c) scFv-based tandem receptors induced with plasma and xFDPs

d) fold induction of scFv-based tandem receptors induced with plasma and xFDPs

A7B4

A7G1

B4A7

B4G1

G1A7G1B4

A7A7

B4B4

G1G1

RR120pS2ctr

0

20

40

60

80

SEA

P [U

*L-1

]

plasma

xFDP

uninduced

A7B4

A7G1

B4A7

B4G1

G1A7G1B4

A7A7

B4B4

G1G1

RR120pS2ctr

0

10

20

30

fold

indu

ctio

n

plasma

xFDP

scFv

A7-scFv

B4-scFv

G1-scFv

scFv-A

7

scFv-B

4

scFv-G

1

scFv-s

cFv

RR1200

50

100

150

SEAP

[UL-1

]

plasma

xFDP

uninduced

scFv

A7-scFv

B4-scFv

G1-scFv

scFv-A

7

scFv-B

4

scFv-G

1

scFv-s

cFv

RR1200

10

20

30

fold

indu

ctio

n

plasma

xFDP



plasma at a final concentration of 0.5 % (v/v) prior to assessment of SEAP reporter activity.

b) and d) Fold induction was calculated for each receptor by normalizing samples induced

with plasma or xFDP to uninduced control samples. All values are means ± SD of triplicate

determinations.



Supplementary Figure 9: Activation of receptors by plasmin degradation products of native

fibrinogen was assessed. HEK293T cells were transfected overnight with the indicated

a) fragment D of fibrinogen

b) fragment E of fibrinogen

c) mixture of fragment D and E of fibrinogen

1.000.3

20.1

00.0

01.0

00.3

20.1

00.0

01.0

00.3

20.1

00.0

01.0

00.3

20.1

00.0

01.0

00.3

20.1

00.0

01.0

00.3

20.1

00.0

00

10

20

30

40

fragment D of fibrinogen [ g/mL]

SEAP

[UL-1

]B4

B4 + scFv

G1-B4

scFv-B4

RR120-receptor

SEAP ctr

2.000.6

30.2

00.0

02.0

00.6

30.2

00.0

02.0

00.6

30.2

00.0

02.0

00.6

30.2

00.0

02.0

00.6

30.2

00.0

02.0

00.6

30.2

00.0

00

10

20

30

40

fragment E of fibrinogen [ g/mL]

SEAP

[UL-1

]

B4

B4 + scFv

G1-B4

scFv-B4

RR120-receptor

SEAP ctr

1.00 +

2.00

0.32 +

0.63

0.10 +

0.20.00

1.00 +

2.00

0.32 +

0.63

0.10 +

0.20.00

1.00 +

2.00

0.32 +

0.63

0.10 +

0.20.00

1.00 +

2.00

0.32 +

0.63

0.10 +

0.20.00

1.00 +

2.00

0.32 +

0.63

0.10 +

0.20.00

1.00 +

2.00

0.32 +

0.63

0.10 +

0.20.00

0

10

20

30

40

fragment D + E of fibrinogen [ g/mL]

SEAP

[UL-1

]

B4

B4 + scFv

G1-B4

scFv-B4

RR120-receptor

SEAP ctr



receptors and the pLS13 STAT3 reporter plasmid alongside pLS15 for STAT3 transcription

factor overexpression, as described in the methods. Cells were incubated for 24 h in

complete DMEM containing 10 % FCS and a) purified fragment D protein of fibrinogen at 1,

0.32, 0.1 or 0 µg/mL or b) purified fragment E protein of fibrinogen at 2, 0.63, 0.2 or 0 µg/mL

or c) a combination of both fragment D and E of fibrinogen at the indicated concentrations.

All values are means ± SD of triplicate determinations.



a) plasmin degrades fibrinogen at high concentrations

1.00000.1

0000.0

1000.0

0100.0

0001.0

0000.1

0000.0

1000.0

0100.0

0001.0

0000.1

0000.0

1000.0

0100.0

0001.0

0000.1

0000.0

1000.0

0100.0

0000

10

20

30

40

50

plasmin acitivity [mU/mL]

SEAP

[U*L

-1]

B4B4 + scFv

G1-B4RR120-receptor

c) plasmin on xFDPs

b) plasmin on fibrinogen

1.000.3

20.1

00.0

0

uninduced

1.000.3

20.1

00.0

0

uninduced

1.000.3

20.1

00.0

0

uninduced

1.000.3

20.1

00.0

0

uninduced

1.000.3

20.1

00.0

0

uninduced

1.000.3

20.1

00.0

0

uninduced

1.000.3

20.1

00.0

0

uninduced

1.000.3

20.1

00.0

0

uninduced

0

10

20

plasmin activitiy [mU/mL]

SEAP

[U*L

-1]

B4 - day1B4 - day2 + plasminB4+scFv - day1B4+scFv - day2 + plasminRR120-receptor - day1RR120-receptor - day2 + plasmin

10.0

3.16 1.00.3

20.1

00.0

320.0

10.0

0320.01

uninduced

10.0

3.16 1.00.3

20.1

00.0

320.0

10.0

0320.01

uninduced

10.0

3.16 1.00.3

20.1

00.0

320.0

10.0

0320.01

uninduced

0

5

10

15

plasmin activitiy [mU/mL]

SEAP

[U*L

-1]

B4B4 + scFv

RR120-receptor



Supplementary Figure 10: Additional characterization of inducer molecules. HEK293T cells

were transfected overnight with the indicated receptors and the pLS13 STAT3 reporter

plasmid alongside pLS15 for STAT3 transcription factor overexpression, as described in the

methods. Cells were incubated for 24 h in complete DMEM containing 10 % FCS and the test

substance. a) Fibrinogen-mediated activity of AMBERB4, AMBERB4/scFv or AMBERG1-B4 is

reduced by high activity of plasmin at 1 mU/mL. Fibrinogen was incubated at 5 µg/mL

together with the indicated concentrations of plasmin. Sensitivity to plasmin depends on

the receptor configuration. b) Cells expressing AMBERB4 or AMBERB4/scFv or RR120 receptor

were activated with fibrinogen at 5 µg/mL for 24 h prior to replacing the medium with

complete DMEM containing no fibrinogen but the indicated activity of plasmin instead. No

difference in activation can be observed between different concentrations of plasmin. c)

xFDP-activated expression of SEAP reporter is independent of plasmin activity in the

supernatant. Cells were incubated with 1 µg/mL xFDP protein and supplemented with the

indicated activities of plasmin. All values are means ± SD of triplicate determinations.



Supplementary Figure 11: Analysis of plasma fractionated by size-exclusion

chromatography. Size-exclusion chromatography (SEC) was used to assess the active

fraction of whole human plasma using a self-packed Sephadex 200 column. Plasma was

filtered through a 0.2 µm filter to remove most of the lipoprotein vesicles prior to loading.

a) UV detection of SEC experiment (blue) correlated with activity of each fraction in the cell

culture (red and orange). b) SDS-PAGE under denaturing conditions of relevant fractions

from the SEC experiment. Boxed bands were excised and subjected to mass spectrometry

for identification. The results of mass spectrometry are presented in boxes on the left-hand

side, including the molecular mass of the respective full-length protein along with the

a) activity analysis of size exclusion chromatography derived fractions

b) SDS-PAGE and mass-spectrometric analysis of relevant fractions

Mfract ion number - 6% denaturing SDS-PAGE

MDD 22 24 26 28 29 30 31 32 34 36 38 40

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 105 110 115 120 125 1300

5

10

15

20

25

30

35

40

45

0

10

20

30

40

50

60

fraction No.

SEAP

[U*L

-1]

UV-Absorbance

Activity - B4

Activity - B4 + scFv

UV-Absorbance [mA

U]



relative abundance of the indicated protein in the sample. Values of receptor activity in a)

are means ± SD of duplicate determinations.



Supplementary Figure 12: a) Unspecific activation of receptors under the conditions used

in figure 3b is excluded by measuring the responses with RR120-receptor-transfected

HEK293T cells harboring the pLS13 STAT3 reporter plasmid and pLS15 for STAT3

a) fibrinogen and plasma controls

plasma -

0.5 %

plasma +

hirudin

plasma +

hepari

n

fibrin

ogen -

10 g/m

L

fibrin

ogen -

2 g/mL

fibrin

ogen + hiru

din

fibrin

ogen + hepa

rin

uninduced

plasma -

0.5 %

plasma +

hirudin

plasma +

hepari

n

fibrin

ogen -

10 g/m

L

fibrin

ogen -

2 g/mL

fibrin

ogen + hiru

din

fibrin

ogen + hepa

rin

uninduced

0

10

20

30SE

AP [U

*L-1

]RR120

pS2ctr

b) Argatroban as an inhibitor

heparin 0.0

01.0

05.0

010

.00

100.0

00

10

20

30

Argatroban [ M]

SEA

P [U

L-1]

B4

B4 + scFv

G1-B4

RR120-receptor

c) Ximelegatran as an inhibitor

0.00

1.00

5.00

10.00

100.0

00

10

20

30

Ximelagatran [ M]

SEAP

[UL-1

]B4

B4 + scFv

G1-B4

RR120-receptor

heparin

d) induction with solid, insoluble fibrin e) induction with solublized, insoluble fibrin

1.0 10-2 0.1 1.0 10.0 100.00

10

20

30

40

5050

100

150

fibrin concentration [ g/mL] (soluble)

SEAP

[UL-1

]

B4

B4 + scFv

G1-B4

scFv-B4

RR120-receptor

pS2ctr

1.0 10-2 0.1 1.0 10.00

5

10

15

50100150

fibrin concentration [mg/mL] (solid)

SEAP

[UL-1

]

B4

B4 + scFv

G1-B4

scFv-B4

RR120-receptor

pS2ctr



transcription factor overexpression. Cells were incubated for 24 h in complete DMEM

containing plasma or fibrinogen and either 1 µg/mL hirudin or 15 USP/mL heparin.

Inhibition of coagulation by the thrombin inhibitors b) argatroban and c) ximelegatran was

assessed by measuring the SEAP reporter production of cells expressing AMBERB4,

AMBERB4/scFv or AMBERG1-B4. A 10-dose response plot of fibrin on AMBERB4, AMBERB4/scFv,

AMBERG1-B4 or AMBERscFv-B4 revealed bell-shape-like activation patterns for d) an insoluble

fibrin preparation and e) a solubilized fibrin preparation that becomes insoluble at neutral

pH. All values are means ± SD of triplicates.



Supplementary Figure 13: Reference samples for assessment of cross-linking and

fibrinolysis-modulating factors depicted in fig. 3d). HEK293T cells were transfected with

either the RR120 receptor in combination with pLS13 and pLS15 or pSEAP2ctr constitutively

expressing SEAP reporter. The latter showed an unspecific boost of cell productivity in

DMEM supplemented with 10 mg/mL TXA. All values are means ± SD of triplicate

determinations.

cell productivity is boosted by tranexamic acid

0

10

20

30

40

SEAP

[UL-1

]

RR120-receptor

SEAP ctr

FGNTHRB

TrideginTXAFXIII

plasmin

+ ++

+++

++

+

++

+

++

++

++

++

++

+

++

+

+

+ ++

+++

++

+

++

+

++

++

++

++

++

+

++

+

+



Supplementary Figure 14: Characterization of stable cell lines. a) Different reporter

constructs were evaluated for their inducibility with xFDPs in combination with

AMBERB4/scFv. HEK293T cells were transfected overnight with the indicated receptors and

the pLS13 STAT3 reporter plasmid alongside pLS15 for STAT3 transcription factor

overexpression, as described in the methods. Cells were incubated for 24 h in complete

DMEM containing 10 % FCS and the test substance. xFDPs were supplemented at the

a) reporter design for stable constructs b) stable RR120-receptor nLuc/TNK expression

c) time course with xFDP induction

e) dose-response against xFDP f) dose-response against plasma

d) time course with plasma induction

1 0.10.050.0

1 0 1 0.10.050.0

1 0 1 0.10.050.0

1 0 1 0.10.050.0

1 0 1 0.10.050.0

1 00

20

40

60

80

100


SEAP

[U*L

-1]

pLS13

pTS441

pTS566

pTS810

pTS824

uninduced

0.125 µ

g/mL xF

DPs

0.5 µg/m

L xFDPs

2.0 µg/m

L xFDPs

0.05 %

plasma

0

1 105

2 105

3 105

4 105

5 105

0

1

2

3

4

nLuc

[AU]

TNK activtiy [m

U/m

L]

TNK activity

N.D. N.D. N.D. N.D. N.D.

nLuc luminescence

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 360.0

5.0 105

1.0 106

1.5 106

2.0 106

time after induction [h]

nLuc

Lum

ines

cenc

e [A

U]

pcTS59

pcTS60

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 360.0

5.0 105

1.0 106

1.5 106

2.0 106

time after induction [h]

nLuc

Lum

ines

cenc

e [A

U]

pcTS59

pcTS60

10-2 10-1 100 1010

2 105

4 105

6 105

8 105

xFDP concentration [ g/mL]

nLuc

Lum

ines

cenc

e [A

U]

pcTS59

pcTS60

10-2 10-1 1000

2 105

4 105

6 105

8 105


nLuc

Lum

ines

cenc

e [A

U]

pcTS59

pcTS60



indicated concentrations, ranging from 1 µg/mL to 0.01 µg/mL. b) The stable polyclonal cell

line HEKRR120/TNK bears the inactive RR120 receptor controlling a STAT3-driven reporter

construct expressing reporter nano-luciferase (Nluc) coupled via a furin cleavage site to the

therapeutic protein tenecteplase (TNK). 45 000 cells were seeded per well and incubation

was done for 24 h. We subsequently induced these cells for 24 h with the indicated

concentrations of xFDPs or plasma. Nluc and TNK activity were measured in the

supernatant. The TNK activity of untreated cells was subtracted from the observed values.

c)-f) Stable monoclonal HEKscFv-B4/TNK cells and polyclonal HEKRR120/TNK were seeded at 15 000

cells per well and induced 24 h after seeding. c)-d) Time courses of reporter gene expression

after induction with either c) 1 µg/mL xFDPs or d) 1 % (v/v) reconstituted human plasma.

Reporter gene activity was measured at the indicated time points post induction. e)-f) Dose

responses of HEKscFv-B4/TNK and HEKRR120/TNK upon incubation with the indicated amounts of

inducer. A √10-dilution series was prepared with e) 5 µg/mL to 0.011 µg/mL xFDPs. In f)

stable cells were induced with plasma ranging from 2 % to 0.003 % (v/v). Values in a) are

means ± SD of triplicate determinations, values in b) are means ± SD of quadruplicate

determinations, values in c-f) are cumulative values of three independent measurements

performed in triplicate, N=9, shown as mean ± SD.





References

Adam, S. S., N. S. Key, and C. S. Greenberg. 2009. 'D-dimer antigen: current concepts and

future prospects', Blood, 113: 2878-87.

Ardaillou, N., L. Dray, A. Z. Budzynski, V. J. Marder, and M. J. Larrieu. 1977. 'The half-life of

plasmic degradation products of human fibrinogen in rabbits', Thromb Haemost, 37:

201-9.

Baglin, T., T. W. Barrowcliffe, A. Cohen, M. Greaves, and British Comm Stand Haematology.

2006. 'Guidelines on the use and monitoring of heparin', British Journal of

Haematology, 133: 19-34.

Bencurova, E., L. Pulzova, Z. Flachbartova, and M. Bhide. 2015. 'A rapid and simple pipeline

for synthesis of mRNA-ribosome-V(H)H complexes used in single-domain antibody

ribosome display', Mol Biosyst, 11: 1515-24.

Benjamin, E. J., P. Muntner, A. Alonso, M. S. Bittencourt, C. W. Callaway, A. P. Carson, A. M.

Chamberlain, A. R. Chang, S. Cheng, S. R. Das, F. N. Delling, L. Djousse, M. S. V. Elkind,

J. F. Ferguson, M. Fornage, L. C. Jordan, S. S. Khan, B. M. Kissela, K. L. Knutson, T. W.

Kwan, D. T. Lackland, T. T. Lewis, J. H. Lichtman, C. T. Longenecker, M. S. Loop, P. L.

Lutsey, S. S. Martin, K. Matsushita, A. E. Moran, M. E. Mussolino, M. O'Flaherty, A.

Pandey, A. M. Perak, W. D. Rosamond, G. A. Roth, U. K. A. Sampson, G. M. Satou, E. B.

Schroeder, S. H. Shah, N. L. Spartano, A. Stokes, D. L. Tirschwell, C. W. Tsao, M. P.

Turakhia, L. B. VanWagner, J. T. Wilkins, S. S. Wong, S. S. Virani, Amer Heart Assoc

Council Epid, Prevention Stat Comm, and Stroke Stat Subcomm. 2019. 'Heart Disease

and Stroke Statistics-2019 Update A Report From the American Heart Association',

Circulation, 139: E56-E528.

Berry, C. N., C. Girardot, C. Lecoffre, and C. Lunven. 1994. 'Effects of the synthetic thrombin

inhibitor argatroban on fibrin- or clot-incorporated thrombin: comparison with

heparin and recombinant Hirudin', Thromb Haemost, 72: 381-6.

Bertina, R. M., B. P. C. Koeleman, T. Koster, F. R. Rosendaal, R. J. Dirven, H. Deronde, P. A.

Vandervelden, and P. H. Reitsma. 1994. 'Mutation in Blood-Coagulation Factor-V

Associated with Resistance to Activated Protein-C', Nature, 369: 64-67.

References


Binz, H. K., P. Amstutz, A. Kohl, M. T. Stumpp, C. Briand, P. Forrer, M. G. Grutter, and A.

Pluckthun. 2004. 'High-affinity binders selected from designed ankyrin repeat protein

libraries', Nature Biotechnology, 22: 575-82.

Blackmer, A. B., M. D. Oertel, and J. M. Valgus. 2009. 'Fondaparinux and the Management of

Heparin-Induced Thrombocytopenia: The Journey Continues', Annals of

Pharmacotherapy, 43: 1636-46.

Bohm, M., C. A. Bauml, K. Hardes, T. Steinmetzer, D. Roeser, Y. Schaub, M. E. Than, A. Biswas,

and D. Imhof. 2014. 'Novel insights into structure and function of factor XIIIa-inhibitor

tridegin', J Med Chem, 57: 10355-65.

Bonaca, M. P., R. M. Bauersachs, S. S. Anand, E. S. Debus, M. R. Nehler, M. R. Patel, F. Fanelli,

W. H. Capell, L. Diao, N. Jaeger, C. N. Hess, A. F. Pap, J. M. Kittelson, I. Gudz, L. Matyas,

D. K. Krievins, R. Diaz, M. Brodmann, E. Muehlhofer, L. P. Haskell, S. D. Berkowitz, and

W. R. Hiatt. 2020. 'Rivaroxaban in Peripheral Artery Disease after Revascularization',

N Engl J Med.

Brown, J. H., N. Volkmann, G. Jun, A. H. Henschen-Edman, and C. Cohen. 2000. 'The crystal

structure of modified bovine fibrinogen', Proceedings of the National Academy of

Sciences of the United States of America, 97: 85-90.

Dahlback, B., M. Carlsson, and P. J. Svensson. 1993. 'Familial Thrombophilia Due to a

Previously Unrecognized Mechanism Characterized by Poor Anticoagulant Response

to Activated Protein-C - Prediction of a Cofactor to Activated Protein-C', Proceedings

of the National Academy of Sciences of the United States of America, 90: 1004-08.

Dipasco, P. J., S. Misra, L. G. Koniaris, and F. L. Moffat. 2011. 'Thrombophilic State in Cancer,

Part I: Biology, Incidence, and Risk Factors', Journal of Surgical Oncology, 104: 316-22.

Dreier, B., and A. Pluckthun. 2012. 'Rapid Selection of High-Affinity Binders Using Ribosome

Display', Ribosome Display and Related Technologies: Methods and Protocols, 805:

261-86.

Froehling, D. A., P. R. Daniels, S. J. Swensen, J. A. Heit, J. N. Mandrekar, J. H. Ryu, and P. L.

Elkin. 2007. 'Evaluation of a quantitative D-dimer latex immunoassay for acute

pulmonary embolism diagnosed by computed tomographic angiography', Mayo Clin

Proc, 82: 556-60.



Grieger, J. C., and R. J. Samulski. 2005. 'Packaging capacity of adeno-associated virus

serotypes: Impact of larger genomes on infectivity and postentry steps', Journal of

Virology, 79: 9933-44.

Guellouz, A., M. Valerio-Lepiniec, A. Urvoas, A. Chevrel, M. Graille, Z. Fourati-Kammoun, M.

Desmadril, H. van Tilbeurgh, and P. Minard. 2013. 'Selection of specific protein binders

for pre-defined targets from an optimized library of artificial helicoidal repeat proteins

(alphaRep)', PLoS One, 8: e71512.

Gupta, S., A. Biswas, M. S. Akhter, C. Krettler, C. Reinhart, J. Dodt, A. Reuter, H. Philippou, V.

Ivaskevicius, and J. Oldenburg. 2016. 'Revisiting the mechanism of coagulation factor

XIII activation and regulation from a structure/functional perspective', Scientific

Reports, 6.

Helma, J., M. C. Cardoso, S. Muyldermans, and H. Leonhardt. 2015. 'Nanobodies and

recombinant binders in cell biology', Journal of Cell Biology, 209: 633-44.

Hur, W. S., N. Mazinani, X. J. D. Lu, H. M. Britton, J. R. Byrnes, A. S. Wolberg, and C. J. Kastrup.

2015. 'Coagulation factor XIIIa is inactivated by plasmin', Blood, 126: 2329-37.

Iung, B., and J. Rodes-Cabau. 2014. 'The optimal management of anti-thrombotic therapy

after valve replacement: certainties and uncertainties', European Heart Journal, 35:

2942-9.

Karatan, E., M. Merguerian, Z. H. Han, M. D. Scholle, S. Koide, and B. K. Kay. 2004. 'Molecular

recognition properties of FN3 monobodies that bind the Src SH3 domain', Chemistry

& Biology, 11: 835-44.

Kollman, J. M., L. Pandi, M. R. Sawaya, M. Riley, and R. F. Doolittle. 2009. 'Crystal structure of

human fibrinogen', Biochemistry, 48: 3877-86.

Kowarz, E., D. Loscher, and R. Marschalek. 2015. 'Optimized Sleeping Beauty transposons

rapidly generate stable transgenic cell lines', Biotechnol J, 10: 647-53.

Laroche, Y., M. Demaeyer, J. M. Stassen, Y. Gansemans, E. Demarsin, G. Matthyssens, D.

Collen, and P. Holvoet. 1991. 'Characterization of a Recombinant Single-Chain

Molecule Comprising the Variable Domains of a Monoclonal-Antibody Specific for

Human Fibrin Fragment D-Dimer', Journal of Biological Chemistry, 266: 16343-49.

Lesne, J., H. J. Chang, A. De Visch, M. Paloni, P. Barthe, J. F. Guichou, P. Mayonove, A. Barducci,

G. Labesse, J. Bonnet, and M. Cohen-Gonsaud. 2019. 'Structural basis for chemically-

induced homodimerization of a single domain antibody', Scientific Reports, 9.

References


Livnah, O., E. A. Stura, D. L. Johnson, S. A. Middleton, L. S. Mulcahy, N. C. Wrighton, W. J.

Dower, L. K. Jolliffe, and I. A. Wilson. 1996. 'Functional mimicry of a protein hormone

by a peptide agonist: The EPO receptor complex at 2.8 angstrom', Science, 273: 464-

71.

Lorand, L., J. Downey, T. Gotoh, A. Jacobsen, and S. Tokura. 1968. 'The transpeptidase system

which crosslinks fibrin by gamma-glutamyle-episilon-lysine bonds', Biochem Biophys

Res Commun, 31: 222-30.

Luscher, T. F., J. Steffel, F. R. Eberli, M. Joner, G. Nakazawa, F. C. Tanner, and R. Virmani. 2007.

'Drug-eluting stent and coronary thrombosis: biological mechanisms and clinical

implications', Circulation, 115: 1051-8.

Madeira, F., Y. M. Park, J. Lee, N. Buso, T. Gur, N. Madhusoodanan, P. Basutkar, A. R. N. Tivey,

S. C. Potter, R. D. Finn, and R. Lopez. 2019. 'The EMBL-EBI search and sequence

analysis tools APIs in 2019', Nucleic Acids Research, 47: W636-W41.

Middleton, S. A., D. L. Johnson, R. Jin, F. J. McMahon, A. Collins, J. Tullai, R. H. Gruninger, L. K.

Jolliffe, and L. S. Mulcahy. 1996. 'Identification of a critical ligand binding determinant

of the human erythropoietin receptor. Evidence for common ligand binding motifs in

the cytokine receptor family', J Biol Chem, 271: 14045-54.

Mohammed, S., S. Sukumaran, P. Bajgain, N. Watanabe, H. E. Heslop, C. M. Rooney, M. K.

Brenner, W. E. Fisher, A. M. Leen, and J. F. Vera. 2017. 'Improving Chimeric Antigen

Receptor-Modified T Cell Function by Reversing the Immunosuppressive Tumor

Microenvironment of Pancreatic Cancer', Molecular Therapy, 25: 249-58.

Petryszak, R., M. Keays, Y. A. Tang, N. A. Fonseca, E. Barrera, T. Burdett, A. Fullgrabe, A. M.

Fuentes, S. Jupp, S. Koskinen, O. Mannion, L. Huerta, K. Megy, C. Snow, E. Williams,

M. Barzine, E. Hastings, H. Weisser, J. Wright, P. Jaiswal, W. Huber, J. Choudhary, H. E.

Parkinson, and A. Brazma. 2016. 'Expression Atlas update--an integrated database of

gene and protein expression in humans, animals and plants', Nucleic Acids Res, 44:

D746-52.

Pluckthun, A. 2015. 'Designed ankyrin repeat proteins (DARPins): binding proteins for

research, diagnostics, and therapy', Annu Rev Pharmacol Toxicol, 55: 489-511.

Scheller, L., and M. Fussenegger. 2019. 'From synthetic biology to human therapy: engineered

mammalian cells', Curr Opin Biotechnol, 58: 108-16.



Scheller, L., T. Strittmatter, D. Fuchs, D. Bojar, and M. Fussenegger. 2018. 'Generalized

extracellular molecule sensor platform for programming cellular behavior', Nature

Chemical Biology, 14: 723-+.

Schilling, J., J. Schoppe, and A. Pluckthun. 2014. 'From DARPins to LoopDARPins: Novel

LoopDARPin Design Allows the Selection of Low Picomolar Binders in a Single Round

of Ribosome Display', Journal of Molecular Biology, 426: 691-721.

Schilling, J., J. Schoppel, E. Sauer, and A. Pluckthun. 2014. 'Co-Crystallization with

Conformation-Specific Designed Ankyrin Repeat Proteins Explains the Conformational

Flexibility of BCL-W', Journal of Molecular Biology, 426: 2346-62.

Schwarz, K. A., N. M. Daringer, T. B. Dolberg, and J. N. Leonard. 2017. 'Rewiring human cellular

input-output using modular extracellular sensors', Nature Chemical Biology, 13: 202-

09.

Skerra, A., and A. Pluckthun. 1988. 'Assembly of a Functional Immunoglobulin-Fv Fragment in

Escherichia-Coli', Science, 240: 1038-41.

Smith, S. A., R. J. Travers, and J. H. Morrissey. 2015. 'How it all starts: Initiation of the clotting

cascade', Critical Reviews in Biochemistry and Molecular Biology, 50: 326-36.

Sohn, J. H., H. A. Kang, K. J. Rao, C. H. Kim, E. S. Choi, B. H. Chung, and S. K. Rhee. 2001. 'Current

status of the anticoagulant hirudin: its biotechnological production and clinical

practice', Applied Microbiology and Biotechnology, 57: 606-13.

Tanswell, P., W. Nieuwenhuizen, and E. Seifried. 1993. 'Pharmacokinetics of Degradation

Products of Fibrin and Fibrinogen during Alteplase Therapy of Acute Myocardial-

Infarction', Fibrinolysis, 7: 408-15.

Uchanski, T., T. Zogg, J. Yin, D. P. Yuan, A. Wohlkonig, B. Fischer, D. M. Rosenbaum, B. K.

Kobilka, E. Pardon, and J. Steyaert. 2019. 'An improved yeast surface display platform

for the screening of nanobody immune libraries', Scientific Reports, 9.

Ward, E. S., D. Gussow, A. D. Griffiths, P. T. Jones, and G. Winter. 1989. 'Binding activities of a

repertoire of single immunoglobulin variable domains secreted from Escherichia coli',

Nature, 341: 544-6.

Watson, T., E. Shantsila, and G. Y. H. Lip. 2009. 'Mechanisms of thrombogenesis in atrial

fibrillation: Virchow's triad revisited', Lancet, 373: 155-66.

Xu, L. C., J. W. Bauer, and C. A. Siedlecki. 2014. 'Proteins, platelets, and blood coagulation at

biomaterial interfaces', Colloids and Surfaces B-Biointerfaces, 124: 49-68.

References


Yumeng Yao, Jiatian Cao, Qingqing Wang, Kai Liu, Zhe Luo, Kaihuan Yu, Xiang Chen, Bijie Hu,

Zheyong Huang. 2020. 'D-dimer as a biomarker for disease severity and mortality in

COVID-19 patients: a case control study', Preprint at Researchsquare - Critical Care &

Emergency Medicine.

Chapter 2: Gene switch for L-glucose-induced biopharmaceutical production in mammalian cells


Chapter 2: Gene switch for L-glucose-induced biopharmaceutical

production in mammalian cells

Chapter 2 is based on a manuscript that has been submitted for publication. This work builds

upon work described in the master theses of Sabina Egli (ETH Stud. ID: 11-923-422) and

Richard Plieninger (ETH Stud. ID: 12-063-137), both of whom had been co-supervised by

Mingqi Xie and Martin Fussenegger.


Tobias Strittmatter1, Mingqi Xie4 and Martin Fussenegger1,5 designed the project. Tobias

Strittmatter1, Adrian Bertschi1, Sabina Egli2, Richard Plieninger1, Daniel Bojar3 and Mingqi Xie

performed experiments and analyzed the results. Tobias Strittmatter1, Adrian Bertschi1,

Mingqi Xie and Martin Fussenegger1,5 wrote the manuscript.


Tobias Strittmatter1: figures 1a-c, 2a,c,d, 3, 4

Sabina Egli2 and Mingqi Xie4: figure 1d

Adrian Bertschi1: figures 1e-f, 2b

Chia-wei Lin6: performed glycosylation analysis for figure 4c

Author Affiliations 1Department of Biosystems Science and Engineering, ETH Zurich, Mattenstrasse 26, CH-4058

Basel, Switzerland 2Present address: Roche Pharma (Schweiz) AG, Basel, Switzerland 3Present address: University of Gothenburg, Gothenburg, Sweden 4School of Life Sciences, Westlake University, 310024 Hangzhou, Zhejiang, China 5Faculty of Science, University of Basel, Mattenstrasse 26, CH-4058 Basel, Switzerland 6University of Zurich, Zürich, Switzerland

Author Affiliations




Abstract

In this work, we designed and built a gene switch that employs metabolically inert L-

glucose to regulate transgene expression in mammalian cells via D-idonate-mediated control

of the bacterial regulator LgnR. To this end, we engineered a metabolic cascade in mammalian

cells to produce the inducer molecule D-idonate from its precursor L-glucose by ectopically

expressing the Paracoccus species 43P- derived catabolic enzymes LgdA, LgnH and LgnI. To

obtain ON- and OFF-switches, we fused LgnR to the human transcriptional silencer domain

KRAB and the viral trans-activator domain VP16, respectively. Thus, these artificial

transcription factors KRAB-LgnR or VP16-LgnR modulated cognate promoters containing

LgnR-specific binding sites in a D-idonate-dependent manner as a direct result of L-glucose

metabolism. In a proof-of-concept experiment, we show that the switches can control

production of the model biopharmaceutical rituximab in both transiently and stably

transfected HEK-293T cells, as well as CHO-K1 cells. Rituximab production reached 5.9 µg/mL

in stably transfected HEK-293T cells and 3.3 µg/mL in stably transfected CHO-K1 cells.

Abstract




Introduction

The production of therapeutic biologics by engineered mammalian cells in bioreactors

is a key element of modern-day medicine to treat various diseases, which include but are not

limited to cancer (Reff et al. 1994; Pierpont et al. 2018; Muhsin et al. 2004), immune system

disorders (Belliveau 2005; Monaco et al. 2015) and inherited enzyme deficiencies (Brady et

al. 1973; Brady et al. 1974; Ioannou et al. 1992; Grabowski et al. 1998). Typically, the

production cassette is induced by small molecules after an initial growth phase in order to

reduce stress on the cells and to improve the purity of the product (Kopp et al. 2019; Huang

et al. 2014). For this purpose, various gene switches have been developed for the exogenous

control of gene expression in bacteria, yeast and mammalian cells in response to the addition

of small inducer molecules or even upon exposure to light (Kallunki et al. 2019; Mansouri et

al. 2019). However, some of these inducers can have undesired side effects (Ahler et al. 2013),

while in the case of light as an inducer, insufficient penetration especially into high-density

cultures can lead to inhomogeneous induction (Suh et al. 2003). Attempts to increase yields

have long been focused on process engineering improvements, but implementation of

induced gene expression in mammalian cells has recently attracted increasing interest (Lonza

2019). Reasons for this development include the greater ability of mammalian cells to

produce difficult-to-express proteins, such as large and complex recombinant proteins that

require post-translational modifications. Although bacterial and yeast production systems

have their advantages in productivity and efficiency, only mammalian cells are able to

efficiently introduce post-translational modifications that are essential for functionality of

some biopharmaceuticals (Amann et al. 2019). Post translational modifications such as

specific glycosylation patterns (Zhang et al. 2016; Stavenhagen et al. 2018), carboxylation

(Hansson et al. 2005), phosphorylation (Yalak et al. 2012) and sulfation (Mufarrege et al. 2020)

are unique to the production cell line and often require mammalian cell systems for

implementation. Especially for the production of therapeutics like clotting factors (Mannully

et al. 2018), functionalized AB as well as novel multispecific drugs (Deshaies 2020),

mammalian cell lines have become more important to ensure quality and safety of the

product. To date, the most commonly used mammalian cell lines employed for industrial

production of biopharmaceuticals are HEK-293T cells for human derived and CHO-K1 cells for

a multitude of recombinant proteins (Amann et al. 2019).

Introduction


All living organisms show homochirality toward organic compounds, such as amino

acids and sugars, where L-amino acids and D-sugars are typically abundant in nature (Shimizu

et al. 2012). L-Glucose is the enantiomer of D-glucose, and provides the same sensation of

sweetness as the common D-enantiomer, but is inert to degradation in mammals (Rudney

1940; Levin et al. 1995). Therefore, humans can neither store nor metabolize L-glucose to

produce energy (Rudney 1940; Sasajima et al. 1979). Thus, L-glucose seems to be an ideal

candidate for orthogonal control of gene expression in mammalian cells. In this work, we

established a gene switch for L-glucose-controlled expression of a model biopharmaceutical

in standard mammalian cell lines based on the D-idonate-responsive bacterial transcriptional

regulator LgnR (Shimizu et al. 2012). LgnR is a member of the IclR family of bacterial repressors

that control antibiotic resistance, metabolism, quorum sensing, and pathogenicity (Molina-

Henares et al. 2006). In its native context, LgnR drives the expression of the lgn operon, which

contains five enzymes involved in L-glucose catabolism. Those enzymes catalyze a chain of

reactions that further degrade the L-glucose oxidation product L-gluconate to the common

metabolites glycerinaldehyde-3-phosphate and pyruvate. In the initial step of L-glucose

degradation, L-glucose is oxidized to yield L-gluconate. This reaction is mediated by the

dehydrogenase LgdA, which is mainly involved in the degradation of scyllo-inositol (KM = 3.70

± 0.4 mM / Kcat = 705 ± 12 min-1), but can utilize L-glucose as an alternative substrate with

lower efficiency (KM = 59.7 ± 5.7 mM / Kcat = 1040 ± 28 min-1) (Shimizu et al. 2012). Following

its production from L-glucose, L-gluconate is isomerized to D-idonate by L-gluconate

dehydrogenase (L-GnDH; LgnH) and 5-keto-L-gluconate reductase (LgnI). Notably, only three

enzymes are needed to form D-idonate (figure 1a), which controls LgnR activity by inducing

dissociation of LgnR from its promoter (Shimizu et al. 2014). By fusing LgnR to the trans-

activator virus protein 16 (VP16) from herpes simplex virus (Jonker et al. 2005) or the silencing

domain Krüppel associated box (KRAB), (Margolin et al. 1994) we were able to build an L-

glucose-responsive synthetic trans-activator or trans-silencer, respectively. The resulting

artificial transcription factors were combined with synthetic promoters that include binding

sites for LgnR.

Thus, by employing components from the L-glucose degradation system, artificial

LgnR-based transcription factors and synthetic promoters, we were able to build L-glucose-

sensitive genetic ON- and OFF-switches to control transgene expression, as measured in

terms of production of reporter proteins. We show that these switches can control



production of the model biopharmaceutical rituximab in both transiently transfected and

stably transfected HEK-293T cells, as well as CHO-K1 cells.

Introduction




Results

Design of an L-glucose-responsive gene circuit

To implement L-glucose as an inducer of an orthogonal gene switch, we engineered

expression vectors for the three enzymes necessary for conversion of L-glucose into D-

idonate in mammalian cells. We also fused the D-idonate-responsive bacterial repressor

protein LgnR with either a KRAB silencer domain or a VP16 activator domain to obtain

synthetic D-idonate-controlled mammalian transcription factors. Since it is reported that the

canine sodium/myo-inositol transporter 1 (SMIT1; SLC5A3) is non-specifically permeable for

L-glucose (Hager et al. 1995), we also included SMIT1 in our setup. Thus, we tested whether

ectopic expression of SMIT1 would increase the induction fold of L-glucose-dependent

transgene expression. Co-expression of the KRAB-LgnR and VP16-LgnR fusions together with

the metabolic cascade would yield genetic L-glucose-dependent ON- (figure 1b, left) and OFF-

switches (figure 1b, right), respectively. For the ON-switch, binding of D-idonate by KRAB-

LgnR triggers the release of the trans-silencer from the operator site and hence enables

transcription. In contrast, the VP16-LgnR trans-activator triggers expression when bound to

the promoter in the uninduced state. Binding of the D-idonate ligand then causes the trans-

activator to dissociate from the DNA, shutting down expression of the transgene. Both

switches are triggered once D-idonate reaches sufficient levels in the cytosol of transfected

cells through the activity of the enzyme cascade.

Based on the DNA sequence of the intergenic region between LgnR and the adjacent

lgn gene cluster, we examined four potential binding motifs for LgnR by cloning them either

downstream of a constitutive promoter derived from simian virus 40 (PSV40) or upstream of a

minimal human cytomegalovirus promoter (PhCMVmin) (figure 1c). Binding motif 0 spans the

whole intergenic region, containing operator elements for both the lgn gene cluster (LgnA)

and the LgnR gene. Three shortened motifs contain either a binding sequence spanning the

described -35 and -10 elements controlling expression of LgnA of the lgn gene cluster (motif

A), the corresponding version on the LgnR operon (motif C) or a shortened sequence based

on motif C (motif B1). Two and three repeats of binding motifs A (A2, A3) and C (C2, C3) were

compared to their respective single binding sites (A1, C1).

Results


Figure 1: a) Overview of the L-glucose degradation pathway producing D-idonate. In the

first step, LgdA uses NAD+ to oxidize L-glucose to L-1,5-gluconolactone, which

spontaneously hydrolyzes to form L-gluconate. In two subsequent oxidation and reduction

reactions at the C5 atom, L-gluconate is epimerized to D-idonate using NAD/NADH as a

b) schematic of ON/OFF switcha) L-glucose catabolism

d) LgnR based expression

LgdA

L-glucose L-glucono-1,5-lactone

L-gluconateL-5-ketogluconateD-idonate

NAD+ NADH

LgnHLgnI

NAD+

H2O

NADHNAD+

NADH

c) LgnR binding sites

KRAB-LgnR

D-idonate

ON-system OFF-system

VP16-LgnR

L-glucoseSMIT1

PCMVmin SEAPOLgnRPCMV SEAPOLgnR

f) toxicity of L-glucose inducere) toxicity of intermediates

agtgcttagttcatattatgggaaccaatcccatgtcaatcagtgaacacaaatttctttcatatagtaatactgtgtgccataatatgggatattgaagatcacgaatcaagtataatacccttggttagggtacagttagtcacttgtgtttaaagaaagtatatcattatgacacacggtattataccctataacttct

-35-35-10 -10

pMX157 (0)pMX212/220 (A) pMX215/223 (B)

pMX216/224 (C)

LgnRLgnA LgnA - LgnR intergenic region

PCMVmin SEAPOLgnR

PSV40 SEAPOLgnR

L-glu 10

0mM

L-gluco

nate 1m

M

D-idon

ate 1mMblank

0

1 106

2 106

3 106

4 106

Lum

ines

cenc

e [A

U]

HEK-293T CHO-K1

0.00

0.010.0

30.1

00.3

21.0

03.1

610

.0031

.6310

0.00

0

2000

4000

6000

8000

0

5000

10000

15000

L-Glucose [mM|

Res

oruf

in (H

EK

-293

T) [F

U]

HEK-293T Resorufin (C

HO

-K1) [FU

]

CHO-K1

pcDNA3.1 pMX181 (VP16-LgnR)

pMX157

pMX220

pMX221

pMX222

pMX223

pMX224

pMX225

pMX226

0

40

80

120

SE

AP

(UL-1 ) 0 A1 A2 A3 B1 C1 C2 C3

pcDNA3.1 pMX183 (KRAB-LgnR)

pMX212

pMX213

pMX214

pMX215

pMX216

pMX217

pMX218

010203040

SEA

P (U

L-1) A1 A2 A3 B1 C1 C2 C3



cofactor by the dehydrogenase LgnH and reductase LgnI, respectively. b) Schematic of the

L-glucose-responsive genetic circuit. L-Glucose (dark hexagon) enters the cell with the help

of transporters (e.g., SMIT1) and is metabolized by the ectopically expressed enzymatic

cascade depicted in a) to produce D-idonate (orange triangle). D-idonate binds the bacterial

repressor protein LgnR, which then dissociates from the operator sequence of a synthetic

promoter. left: LgnR can be fused to the mammalian trans-silencing domain KRAB to

generate a signal in response to D-idonate binding, yielding an ON-switch by induced de-

repression of the strong PhCMV promoter. right: Alternatively, a fusion construct of LgnR and

the trans-activator VP16 induces expression from a weaker minimal promoter derived from

PhCMV (PhCMVmin) in the absence of D-idonate. Accordingly, the signal is quenched if D-

idonates forces LgnR to dissociate from the promoter, yielding an OFF-switch for induced

inactivation of the promoter. c) Overview of the intergenic region of LgnR and the lgn gene

cluster. The regulatory protein LgnR controls expression of both operons. The -10 (green

arrow) and -35 (red and orange arrow) elements are highlighted in bold. Parts of the

intergenic region that are used in response elements to report LgnR binding are indicated

by blue bars labeled with the corresponding genetic constructs. d) HEK-293T cells were co-

transfected with 125 ng of reporter plasmids (pMX212-218 and pMX157 and pMX220-226)

with 375 ng of effector plasmids harboring VP16-LgnR (pMX181-PSV40-VP16-LgnR-pA) or

KRAB-LgnR (pMX183-PhEF1a-KRAB-LgnR-pA) fusions in 48-well plates. e) The impact on

productivity of HEK-293T and CHO-K1 cells was monitored by measuring the luminescence

of secreted NLuc reporter produced from a constitutive PPGK promoter after 48 h. 20 ng of

pTS1034 (PPGK-Igk-NLuc-pA) and 180 ng of filler plasmid pDF101 were co-transfected in 96-

well plates. The impact on viability was assessed by measuring the ability of cells to reduce

resazurin to resorufin. d) The viability of wild-type HEK-293T and CHO-K1 cells was assessed

by resazurin assay after incubation with different amounts of L-glucose for 48 h. All values

are mean ± SD of n = 3 biologically independent samples.

HEK-293T cells co-transfected with 125 ng of PhCMVmin-based reporter constructs and

375 ng of the VP16-LgnR fusion (pMX181) exhibited inducible reporter gene expression for all

four binding motifs. Sequences based on binding motif 0 and A showed strongest induction

(figure 1d, top) and a two-fold repeat of motif A (pMX221) was chosen for further

characterization of the OFF-switch. Reporter constructs containing binding motifs A, B and C

Results


in combination with the constitutive PSV40 promoter (125 ng) were co-transfected with a

KRAB-LgnR fusion (pMX183, 375 ng) to evaluate their use in ON-switches. Here, the plasmid

containing binding motif B in pMX215 showed best inducibility and was used as LgnR-specific

operator site in all follow-up studies.

Testing of metabolic enzymes

Prior to testing the metabolic enzymes, the impact of intermediates of the metabolic

cascade on the cells was assessed. Toxicity was evaluated with the resazurin reduction assay

and metabolic integrity was assessed in terms of constitutive expression of NanoLuc

luciferase (NLuc) in transiently transfected HEK-293T and CHO-K1 (figure 1e) cells. None of

the compounds affected cell viability or productivity. Furthermore, the inducer L-glucose was

found to be non-toxic at concentrations up to 100 mM in the resazurin reduction assay (figure

1f).

To establish whether all components of the enzymatic L-glucose degradation pathway

are required to produce D-idonate in HEK293T cells, we used the LgnR ON-system and

introduced the respective enzymes in different combinations. We tested expression from two

constitutive promoters based on the strong promoter for human elongation factor 1 α (PhEF1α)

and the weaker promoter for human phosphoglycerate kinase (PPGK). We found increased

reporter protein expression from the D-idonate-responsive promoter only in the presence of

all three enzymes of the metabolic cascade (figure 2a). Intriguingly, only the use of the weaker

PPGK promoter, but not the stronger PhEF1α promoter, afforded a functional gene switch.

Additional expression of the transporter SMIT1 increased the fold switching in CHO-K1 cells,

in accordance with the anticipated facilitation of L-glucose transport across the plasma

membrane. To exclude unspecific effects, we tested the intermediates of the metabolic

cascade for induction of the system in HEK-293T cells (figure 2b, i) and CHO-K1 cells (figure

2b, ii). Here we found specific activation of reporter expression from the synthetic promoter

construct in combination with the KRAB-LgnR trans-silencer only when the final product of

the metabolic cascade, D-idonate, was added. In contrast, the complete ON-system was

triggered by L-glucose, L-gluconate and D-idonate, indicating the contribution of each

enzyme.

Taken together, these findings confirm that L-glucose and the ectopic L-glucose

degradation pathway are safe and effective for use in HEK-293T and CHO-K1 cells.



Figure 2: a) Combinations of metabolic enzymes LgdA, LgnH and LgnI, as well as SMIT1,

were tested for ON-switch functionality in HEK-293T and CHO-K1 cells. Enzymes are

expressed from PPGK promoters or PhEF1α promoters as described in the methods section.

Plasmids comprising the full ON-switch that were excluded from the transfection mix in this

experiment were replaced by filler plasmid pDF101. Induction is shown as the ratio of

induced to uninduced samples under each condition. b) The mechanism of action was

validated by transfecting i) HEK-293T and ii) CHO-K1 cells with either the promoter and

VP16-LgnR trans-activator alone (LgnR + rep) or the full ON-system. 65 ng of trans-activator

b) validation of induction mechanisma) essential parts for ON switch

i) HEK293-T cells ii) CHO-K1 cellsHEK-293T - PGK

HEK-293T - EF1aCHO-K1 - EF1a

CHO-K1 - PGK

LgnH/ILgdA

SMIT1

LgnH/I + LgdA

LgnH/I + SMIT1

LgdA + SMIT1

LgnH/I + SMIT1 +

LgdA

reporte

r ctr

0

1

2

3

4

fold

indu

ctio

n [R

U]

L-gluco

se 10

0mM

L-gluco

nate 1m

M

D-idon

ate 1mMblank

L-gluco

se 10

0mM

L-gluco

nate 1m

M

D-idon

ate 1mMblank

0

100

200

300

400

500

SEA

P [U

L-1 ]

ΔLgdA, ΔLgnH, ΔLgnIfull system

0

10

20

30

40

SEA

P [U

L-1 ]

ΔLgdA, ΔLgnH, ΔLgnIfull system

L-gluco

se 10

0mM

L-gluco

nate 1m

M

D-idon

ate 1mMblank

L-gluco

se 10

0mM

L-gluco

nate 1m

M

D-idon

ate 1mMblank

c) ON / OFF switch

i) HEK-293T cells ii) CHO-K1 cells

0.00

0.10

1.00

10.00

100.0

00

20

40

60

80

0

10

20

30

40

50

L-Glucose [mM]

SEA

P [U

L-1]

ON

OF

FS

EAP

[UL

-1]

0.00

0.10

1.00

10.00

100.0

00

100

200

300

400

500

0

10

20

30

40

50

L-Glucose [mM]

SEA

P [U

L-1]

ON

OF

FS

EAP

[UL

-1]

d) time course of inductioni) HEK-293T cells ii) CHO-K1 cells

0 12 24 36 48 60 72 84 96

10

20

30

time [h]

SE

AP

[UL-1

]

-

+

1.45

1.80

1.992.19 2.16 2.06

0 12 24 36 48 60 72 84 96

50

100

150

200

250

time [h]

SEA

P [U

L-1]

-

+1.42

1.551.59

1.63

1.18

Results


pMX183 in and 15 ng of pMX335 reporter were co-transfected with 20 ng of NLuc

expressing plasmid pTS1052 and 120 ng of filler plasmid pDF101. Both setups were tested

with key compounds of the L-glucose degradation pathway. Activation was monitored by

measuring SEAP levels in the supernatant after induction for 48 h. c) The responses of the

ON-switch (left y-axis) and OFF-switch (right y-axis) to various concentrations of L-glucose

were tested in i) HEK-293T and ii) CHO-K1 cells. The medium was exchanged again after 24

h and SEAP reporter activity was measured after 48 h. d) Expression of SEAP reporter from

the ON-switch induced with 10 mM L-glucose was recorded over 4 days in i) HEK-293T and

ii) CHO-K1 cells. Twice the standard volume of medium was used to allow for prolonged

incubation. Induction ratios of induced versus uninduced samples are indicated above the

relevant time points. All values are mean ± SD of n = 3 biologically independent samples.

Functionality of ON- vs OFF-switch

We next evaluated the performance of the ON-system comprising the trans-silencer KRAB-

LgnR and compared it with the OFF switch based on the trans-activator VP16-LgnR. Both

systems responded dose-dependently to increased inducer concentrations (figures 2c).

Induction of the ON-switch generated a clear reporter signal above the background, starting

at 24 h after induction (figure 2d). This signal was more pronounced in CHO-K1 cells than in

HEK-293T, owing to the higher fold induction in the former cell line. Although the fold-change

in reporter expression was higher in the case of the OFF-switch, we considered that higher

absolute expression levels would be preferable for biopharmaceutical production, so we

focused on the ON-switch in subsequent experiments.

L-Glucose-induced rituximab production

For proof-of-concept, we applied the L-glucose system to the production of rituximab

as a model biopharmaceutical. For this purpose, we created a reporter cassette for L-glucose-

inducible production of rituximab by placing a genetic fusion construct of the light and heavy

chain of the antibody under control of the ON-switch. To ensure an equimolar ratio of the

secreted chains, as well as minimal disturbance of the original amino acid sequence, we linked

both chains via a modified furin cleavage site (KRRKR) followed by a T2A sequence, similarly

to a previous study (Fang et al. 2007) (figure 3a).



a) rituximab production system

c) transient rituximab production

b) YPet expression in transient transfections

i) dose-response curve ii) peak rituximab production

i) dose-response curve ii) peak fluorescence

HEK-293T CHO-K10

100

200

300

400

500av

erag

e Y

Pet f

luor

esce

nce

[FU

]4.92 x

1.23 x

uninduced

induced

HEK-293T CHO-K1

0.0

0.1

0.2

0.3

IgG

[g/

mL]

8.1 x

28.5 x

uninduced

induced

0.0000

0.01000.0

3200.1

0000.3

1601.0

0003.1

620

10.0000

31.6230

100.0

000

0

1020

30

40

100200300400500

L-Glucose [mM|

mea

n Y

Pet f

luor

esce

nce

[FU

]

HEK-293T

CHO-K1

0.000

0.0100.0

320.1

000.3

161.0

003.1

63

10.000

31.626

100.0

000.0000.005

0.0100.015

0.020

0.1

0.2

0.3

L-Glucose [mM|

IgG

[g/

mL]

HEK-293T

CHO-K1

D-Idonate-dependent ON-switch

KRAB-LgnRPhEF1a

Rituximab

HC LCT2A IRESFurin YPetPCMV OLgnR

L-Glucose to D-idonate conversion

LgnH P2A LgnIPPGKLgdAPPGK SMIT1PPGK

Results


Figure 3: a) Schematics of constructs used for rituximab production. The same trans-silencer

as before was used in combination with a rituximab expression cassette coupled to the

expression of yellow fluorescent reporter YPet via an internal ribosome entry site (IRES).

Both light and heavy chains of rituximab are expressed in equimolar ratios via a furin

cleavage site fused to a T2A sequence to ensure a high-quality final antibody product. L-

Glucose-dependent expression of b) intracellular YPet reporter (as assessed by flow-

cytometry) and c) secreted rituximab (measured by ELISA for human IgG) from transiently

transfected HEK-293T and CHO-K1 cells was measured after 72 h of continuous induction

with 3.16 mM L-glucose. Plots in i) represent dose-response curves, while bar graphs in ii)

represent peak values at 3.16 mM L-glucose. Values above the bars indicate fold inductions

of induced versus uninduced samples. Values are mean ± SD of b) n = 3 and c) n = 2

biologically independent samples.

The use of this combination enables both chains to be secreted individually, since they are

separated upon translation by the T2A sequence. In addition, the furin cleavage site removes

all the additional amino acids but two from the C-terminal of the heavy chain upon transport

through the Golgi apparatus. The remaining basic amino acids are removed by

carboxypeptidases residing in the trans-Golgi during secretion. We added a yellow fluorescent

reporter protein (YPet) via an internal ribosome entry site (IRES) to the antibody sequence

(figure 3a) as a reporter of induction of the system. We next evaluated the dose-response

curves of YPet and rituximab production in HEK-293T and CHO-K1 cells. YPet fluorescence

(figure 3b) and rituximab concentration (figure 3c) increased concomitantly with increasing

concentration of L-glucose in transiently transfected HEK-293T and CHO-K1 cells. We

measured maximum YPet fluorescence (figure 3 b, ii) and rituximab production (figure 3 c, ii)

at 10 mM L-glucose and obtained peak rituximab production of around 0.25 µg/mL in HEK-

293T and approximately 0.015 µg/mL in CHO-K1 cells. Encouraged by these results, we next

sought to improve rituximab production by generating stable HEK-293T and CHO-K1 cell lines.

To this end, we engineered Sleeping Beauty transposon vectors (Kowarz et al. 2015) to

facilitate genomic integration of all four cassettes making up the ON-switch. Since the

expression levels of stably integrated transgenes can differ from those of transiently

expressed constructs, and at the same time the level of the trans-silencer construct KRAB-

LgnR seemed to be critical for the functionality of the ON-switch, we decided to screen both



the PhEF1α (HEK-293TEF1α and CHO-K1EF1α) and PPGK promoters (HEK-293TPGK and CHO-K1PGK) for

optimal expression of KRAB-LgnR. All the polyclonal cell lines examined showed superior

sensitivity and inducibility compared to transient transfection in both HEK-293T and CHO-K1

cells (figure 4a). Maximum induction of YPet fluorescence was reached at 1 mM L-glucose in

CHO-K1 cell lines (CHO-K1EF1α and CHO-K1PGK) and at 10 mM for HEK-293T cell lines (HEK-

293TEF1α and HEK-293TPGK) (figure 4a, ii). Rituximab production reached 2.4 µg/mL and 5.9

µg/mL in HEK-293TEF1α and HEK-293TPGK and 0.7 µg/mL and 3.3 µg/mL in CHO-K1EF1α and CHO-

K1PGK, respectively (figure 4b). IgG expression was increased up to 24-fold in HEK-293T and up

to 220-fold in CHO-K1 compared to transient expression, while at the same time reducing the

amount of inducer necessary for full induction of the system.

Finally, we assessed whether expression of the metabolic enzymes comprising the L-

glucose switch would alter the glycosylation pattern of the produced rituximab antibody upon

prolonged incubation with L-glucose. To this end, we selected polyclonal cell lines of HEK-

293T cells and CHO-K1 cells that stably express LgdA, LgnH, LgnI and SMIT1 from a PPGK

promoter while also expressing rituximab from a constitutive PhCMV promoter. Cells were

seeded and grown for three days as in previous experiments in standard culture medium or

in medium supplemented with 10 mM L-glucose. Glycosylation patterns were compared

between both culture conditions for both cell lines and found to be broadly similar (figure 4c).

Results


Figure 4: a) intracellular YPet reporter and b) secreted rituximab from stable HEK-293T and

CHO-K1 cells was measured after 72 h of continuous induction with various concentrations

of L-glucose. i) Dose-response curves were measured and ii) peak values of YPet and IgG

production are depicted in separate bar graphs. Expression of the metabolic cascade is

driven by a PPGK promoter. Expression of the synthetic transcription factor KRAB-LgnR is

a) YPet expression in stable HEK-293T and CHO-K1 cells

b) rituximab production in stable HEK-293T and CHO-K1 cells

i) dose-response curves ii) peak rituximab production

c) Assessment of glycosylation of rituximab

i) dose-response curves ii) peak YPet fluorescence

HEK-293

T EF1a

HEK-293

T PGK

CHO-K

1 EF1a

CHO-K

1 PGK

0

1000

2000

3000

4000

5000

aver

age

YPe

t flu

ores

cen

ce [F

U]

1.48 x

1.35 x

uninduced

induced

9.90 x

4.27 x

0

2

4

6

IgG

[g/

mL]

20.65 x

9.82 x2.37 x

2.57 xuninduced

induced

HEK-293

T EF1a

HEK-293

T PGK

CHO-K

1 EF1a

CHO-K

1 PGK

0.0000

0.0100

0.0320

0.1000

0.3160

1.0000

3.1630

10.0000

31.6260

100.0

000

0

1000

2000

3000

4000

L-Glucose [mM|

mea

n Y

Pet f

luor

esce

nce

[FU

]

HEK-293TEF1a

HEK-293TPGK

CHO-K1EF1a

CHO-K1PGK

0.000

0.010

0.032

0.100

0.316

1.000

3.163

10.000

31.626

100.0

000

2

4

6

L-Glucose [mM|

IgG

[g/

mL]

HEK-293TEF1a

HEK-293TPGK

CHO-K1EF1a

CHO-K1PGK

i) HEK-293T

10-3 10-2 10-1 100 101 10210-3

10-2

10-1

100

101

102

rel. abundance (no L-glucose) [%]rel.

abu

ndan

ce (1

0 m

M L

-glu

cose

) [%

]

R2=0.9978

ii) CHO-K1

10-3 10-2 10-1 100 101 10210-3

10-2

10-1

100

101

102

rel. abundance. (no L-glucose) [%]rel.

abu

ndan

ce. (

10 m

M L

-glu

cose

) [%

]

R2=0.9947



controlled by a strong PhEF1α promoter (HEK-293TEF1α and CHO-K1EF1α) or a weaker PPGK

promoter (HEK-293TPGK and CHO-K1PGK). Values above bars indicate induction ratios

between induced and uninduced states. Values are mean ± SD of b) n = 3 and c) n = 2

biologically independent samples. c) Correlation plot of results from MS-analysis to assess

glycosylation patterns of rituximab produced in i) HEK-293T and ii) CHO-K1 cells in presence

or absence of 10 mM L-glucose. Relative abundance of glycan modifications of rituximab

produced without L-glucose were plotted on the x-axis and correlated to abundancies found

in L-glucose treated sample. Straight lines indicate identity (f(x)=x). Detailed analysis is

provided in tables S3 and S4.

Results




Discussion

Here, we have designed, built and characterized a novel gene switch based on a

metabolic cascade starting from the small molecule L-glucose, which is functionally inert in

humans. We transferred all three enzymes, LgdA, LgnH and LgnI, needed for conversion of L-

glucose to D-idonate from Paracoccus species 43P into HEK-293T and CHO-K1 cells, and we

confirmed that D-idonate produced in these cells can control the bacterial regulator protein

LgnR. Thus, by fusing LgnR to the silencer domain KRAB or the trans-activator domain VP16,

we could build ON- and OFF-systems, respectively. In a proof-of-concept, the ON-system

could trigger expression of the model biopharmaceutical rituximab dose-dependently in

response to L-glucose in the cell culture medium without affecting glycosylation. Stable

integration of the engineered pathway afforded polyclonal HEK-293T and CHO-K1 cell lines in

which IgG production reached 5.9 µg/mL and 3.3 µg/mL, respectively.

A key feature of our system is the use of L-glucose as an extremely safe and orthogonal

inducer to control gene expression in mammalian cells. In addition, the use of a metabolic

cascade instead of a single-component switch could offer much greater opportunities for

expression control by adjusting the enzyme activity at each step of the process, enabling fine-

tuning of the system, and perhaps the incorporation of logic-based control.

Interestingly, performance of the ON-switch was much better in cells with stable expression

than in cells with transient expression. This is most likely due to epigenetic modifications of

the promoter region that reduce leaky expression and thereby increase the fold-switching

upon de-repression caused by dissociation of KRAB-LgnR from the promoter. Metabolic

adaptations of the host cell line, as well as suboptimal expression ratios of the components

in transient transfections, might also contribute to the increased functionality in the stable

setup. The cytosolic fluorescent reporter protein YPet provided an easy and convenient tool

for monitoring promoter activity by flow-cytometry. However, the fold-switching of the YPet

reporter signal was reduced in the stable system, possibly due to the long half-life of the

protein, which would lead to a build-up of background signal. If flow-cytometry is not

required, this issue could be avoided in future implementations of the system by using a

different reporter that is not a cytosolic fluorescent protein.

There are undoubtedly many ways in which the IgG production yield could be

increased from the levels obtained in this initial trial. Although both HEK-293T and CHO-K1

cell lines are routinely used for biopharmaceutical production (Dumont et al. 2016), the use

Discussion


of a suspension cell line would drastically increase the cell density in the production process

and hence improve production capacity. Furthermore, the output could potentially be

boosted by additional integrations of the rituximab expression cassette via the orthogonal

PiggyBac transposase system (Yusa et al. 2011) or by the use of the GS-system (Bebbington

et al. 1992; Brown et al. 1992; Fan et al. 2012). The stable cell lines could be further optimized

by selection of monoclonal cell lines.

We believe the successful implementation of this L-glucose-triggered gene switch

further expands the available array of inducers by adding a non-toxic, functionally inert small

molecule, providing a new option for biopharmaceuticals production.



Acknowledgements

The authors thank Pascal Stücheli and Viktor Hällman for fruitful discussions and

critical readings of the manuscript and Chia-wei Lin from the Functional Genomic Center

Zurich for analysis of glycosylation patterns of rituximab samples. This work was financed by

the Swiss National Centre of Competence in Research (NCCR) for Molecular Systems

Engineering.

Acknowledgements




Materials & Methods

Cell culture

HEK-293T cells and CHO-K1 cells were cultured in 10 cm cell culture dishes in

Dulbecco’s modified Eagle’s medium (DMEM, Gibco, #61965-026) supplemented with 10 %

FCS, 50 Units/mL penicillin and 50 µg/mL streptomycin (Gibco, #15070-063) in a humidified

atmosphere containing 7.5 % CO2. The medium was supplemented with 150 µM L-proline for

culturing CHO-K1 cells. Cells were routinely split using trypsin-EDTA (Gibco, #25300-054)

every two days or upon reaching 70 % confluency, and reseeded at 1.5 million cells per plate.

For cell culture experiments, cells were seeded at 15 000 cells per well in 96-well plates

(Corning, #3599) or 250 000 cells per well in a 6-well plate (Corning, #3516).

Plasmid generation and transfection

Plasmids used for ectopic expression of transgenes were obtained by standard

molecular cloning techniques. All plasmids used in this study are listed in table S1.

For transfection of HEK-293T and CHO-K1 cells with plasmid DNA, we used

polyethyleneimine (PEI, 24765-1, Polysciences Inc.) at a PEI:DNA ratio of 6:1. 24 h prior to

transfection, cells were seeded as described in the previous section. A solution of 200 ng

plasmid DNA in 150 µL of serum-free DMEM medium was mixed with 1.2 µg of PEI (1 mg/mL

stock) for each well of a 96-well plate. Transfection mixes were incubated at room

temperature for 5-10 minutes. The solution was mixed again and added to the cells in a drop-

wise manner. Transfected cells were were incubated overnight (>12 h) using standard culture

conditions. The next morning, medium was replaced with standard culture medium (as

described above) supplemented with inducer molecules as necessary.

Data in figure 1d was obtained from cells transfected in 48-well plate format using a

PEI:DNA ratio of 3:1 followed by incubation for only 6 h with the transfection mix prior to

medium exchange.

Experiments with the ON-switch were done using pMX183 (65 ng), pMX335 (15 ng),

pTS2181 (20 ng), pTS2183 (50 ng) and pTS2185 (50 ng) while for the OFF-switch plasmids

pMX181 (40 ng), pMX221 (40 ng), pTS2181 (20 ng), pTS2183 (50 ng) and pTS2185 (50 ng) were

used. In cases were the ON-switch was used to produce rituximab (figure 2) the reporter

Materials & Methods


plasmid pMX335 was replaced with the same amount of pTS2247 and trans-silencer construct

pMX183 was exchanged for pTS2174.

Generation of stably expressing cell lines

HEK-293T and CHO-K1 cells were seeded in 6-well plates and transfected as described

in the previous section by using 250 ng each of pTS2178, pTS2179 and pTS395 (PhCMV-SB100X-

pA) and the filler plasmid pDF101. Cells were selected with 2 µg/mL of puromycin (InvivoGen,

#ant-pr), 4 µg/mL blasticidin (InvivoGen, #ant-bl) in standard cell culture medium for 10 days.

The resulting stable cell lines were further modified by integration of the rituximab

production cassette including PhEF1α - or PPGK-driven trans-silencer KRAB-LgnR. 150 ng of

pTS2228 (PhEF1α) or pTS2230 (PPGK) was co-transfected with 200 ng of Sleeping Beauty

expression vector pTS395 and 650 ng of filler plasmid pDF101. Selection was performed in

standard cell culture medium supplemented with 2 µg/mL of puromycin (InvivoGen, #ant-pr),

4 µg/mL blasticidin (InvivoGen, #ant-bl) and 50 µg/mL Zeocin (InvivoGen, #ant-zn) for 10 days.

To analyze the impact of L-glucose on rituximab glycosylation, stably expressing HEK-293T

cells and CHO-K1 cells were generated by co-transfecting 200 ng of each sleeping beauty

vectors harboring PPGK-driven SMIT1 (pTS2178), LgdA, LgnH and LgnI (pTS2179) and a PhCMV-

driven rituximab-YPet expression cassette (pTS2248) together with 400 ng of a sleeping

beauty expression construct (pTS395). Cells were selected with 2 µg/mL of puromycin

(InvivoGen, #ant-pr), 4 µg/mL blasticidin (InvivoGen, #ant-bl) and 20 µg/mL Zeocin

(InvivoGen, #ant-zn) for 10 days.

Induction with L-glucose

Dose-response curves of reporter gene (SEAP or YPet) expression and rituximab

production were generated by incubating cells in standard cell culture medium (see above)

with the indicated concentrations of L-glucose in a 3.162-fold (√10) dilution series ranging

from 0.01 mM to 100 mM. A 1 M Stock solution of D-gluconate was prepared in D-PBS. D-

Idonate stock solution was prepared as 1 M D-idonic acid-1,4-lactone in D-PBS containing 1.1

M NaOH.



Assessment of reporter gene expression

Activity of the reporter protein, secreted alkaline placental phosphatase (SEAP), was

measured as previously described (Berger et al. 1988). Briefly, supernatant from cell culture

experiments was heat-treated for 30 min at 65 °C to inactivate endogenous phosphatases. 20

µL of heat-treated supernatant was mixed with 180 µL of assay reagent containing 60 µL of

water, 80 µL of 2x assay buffer (1 M diethanolamine pH 9.8, 0.5 mM MgCl2, 10 mM L-

homoarginine) and 20 µL of substrate solution (120 mM p-nitrophenyl phosphate in 2x assay

buffer). Reporter activity was assessed at 37 °C by following substrate turnover at 405 nm on

an Infinite M1000 microplate reader (Tecan Trading AG).

NanoLuc® luciferase activity was measured using a Nano-Glo® kit (Promega AG),

following the recommendations of the manufacturer. Volumes were adjusted to a total of 15

µL per reaction for use in a 384-well plate format.

For viability assays, the culture medium was replaced with medium containing 30

µg/mL (final) of resazurin. Cells were incubated for 15 minutes and fluorescence in the

supernatant was analyzed at 560 nm excitation and 590 nm emission on an Infinite M1000

microplate reader (Tecan Trading AG).

A yellow fluorescent protein (YPet) was used as a non-secreted, cytosolic reporter for

indirect read-out of rituximab expression. The coding sequence of YPet was coupled to the

coding sequence of rituximab with the help of an internal ribosome entry site (IRES).

Expression of YPet was recorded by flow-cytometry using 488 nm excitation and 510 nm

emission on a BD LSR Fortessa flow-cytometer equipped with a 96-well automatic sampler.

Cells were detached using 60 µL per well of trypsin-EDTA, incubated for 5-10 minutes at 37

°C, and then resuspended in 60 µL of D-PBS without Ca2+ (Gibco, #14190-094). 60 µL of the

cell suspension were measured in flow-cytometry experiments. Gates were adjusted based

on wild-type HEK-293T and CHO-K1 cells.

Analysis of flow-cytometry data was done using FlowJo 10 software.

Measurement of rituximab production

Levels of rituximab in cell culture media were measured using an enzyme-linked

immunosorbent assay (ELISA) kit for human IgG (Immunology Consultants Laboratory Inc., #E-

80G) following the manufacturer’s protocol. Samples from transient expression experiments

Materials & Methods


were diluted 1:5, while samples from stable cell lines were diluted 1:20 in 1x assay buffer

before measurement.

Analysis of glycosylation patterns

Rituximab was produced for three days from polyclonal HEK-293T or CHO-K1 cell lines

in standard cell culture medium or in the presence of 10 mM L-glucose. Cell culture

supernatant was collected on day 3, centrifuged for 5 min at 500 rcf, filtered through a 0.2

µm syringe filter, supplemented with cOmplete™, Mini, EDTA-free Protease Inhibitor Cocktail

(Roche, #04693159001) and stored refrigerated until further use.

The culture media was concentrated and buffer exchanged with an Amicon

ultrafiltration column (100 000KDa MWCO, Millipore). Purification of IgGs from the

concentrated solution was done with using a Protein A HP SpinTrap column (cytiva,

#28903132) following the manufacturer’s instructions. All purified IgGs were used for

analysis. Proteins were reduced and alkylated following a filter-assisted sample preparation

(FASP) protocol as described previously (Hang et al. 2015). In brief samples were treated with,

dithiothreitol (DTT) in 50 mM Tris buffer (pH 8.5) at 37⁰C for 1 h, followed by alkylation with

65 mM iodoacetamide (IAA) in 50mM Tris-buffer at 37⁰C in the dark for 1 h. DTT and IAA were

removed by centrifugation and proteins were subsequently digested with 1ug of trypsin. After

trypsin-digestion, samples were desalted using a C18 Zip-Tip. Peptides were resuspended in

10 µL 2.5% acetonitrile with 0.1% formic acid before analysis on a Q-Exactive mass

spectrometer (Thermo Fischer Scientific) coupled to a nano-Acquity UPLC system (Waters).

2µL of the solution was loaded onto a Acclaim PepMap 100 trap column (75 μm Å~ 20 mm,

100 Å, 3 μm particle size) and separated on a nanoACQUITY UPLC BEH130 C18 column (75 μm

Å~ 150 mm, 130 Å, 1.7 μm particle size), at a constant flow rate of 300 nL min-1, with a column

temperature of 50 °C and a linear gradient of 2−32% acetonitrile/0.1% formic acid in 59 min,

and then 32-45% acetonitrile/0.1% formic acid in 10 min, followed by a sharp increase to 98%

acetonitrile in 2 min and then held isocratically for another 10 min. The mass spectrometer

was operated under data dependent acquisition (DDA), one scan cycle was comprised of a full

scan MS survey spectrum, followed by up to 12 sequential HCD MS/MS on the most intense

signals above a threshold of 1e4. Full-scan MS spectra (600–2000 m/z) were acquired in the

FT-Orbitrap at a resolution of 70,000 at 400 m/z, while HCD MS/MS spectra were recorded in

a FT-Orbitrap at a resolution of 35,000 at 400 m/z. HCD was performed with a target value of



1e5 and normalization collision energy 25 was applied. AGC target values were 5e5 for full

FTMS. For all experiments, dynamic exclusion was used with a single repeat count, 15 s repeat

duration, and 30 s exclusion duration.

Glycopeptide were analysed by Byonic 3.6 (Protein Metrics, USA) and inspected

manually. The quantification for each glycan form on IgG was performed as described

previously (Hang et al. 2015).

Statistical Analysis

No statistical analysis was performed. All data are presented as means ± standard

deviation (SD) of triplicates or duplicates as indicated.

Materials & Methods





Table S1: Plasmid Information


pCDNA3.1 Filler plasmid used to replace PhCMV-driven active

components.

Invitrogen™

V79020

pMX157 Reporter construct used with VP16-LgnR. LgnR-operator

sequence (motif 0) is cloned 5’ of the minimal promoter

PhCMVmin (0-PhCMVmin-SEAP-pA).

This work

pMX181 Constitutive VP16-LgnR expression vector with a

constitutive PSV40 promoter (PSV40-VP16-LgnR-pA).

This work

pMX183 Constitutive KRAB-LgnR expression vector with a strong

PhEF1α promoter (PhEF1α-KRAB-LgnR-pA).

This work

pMX212 Reporter construct used with KRAB-LgnR. LgnR-operator

sequence (motif A1) is cloned 3’ of the constitutive

promoter PSV40 (PSV40-A1-SEAP-pA).

This work

pMX213 Reporter construct used with KRAB-LgnR. A tandem repeat

of LgnR binding motif A used in pMX212 is cloned 3’ of the

constitutive promoter PSV40 (PSV40-A2-SEAP-pA).

This work

pMX214 Reporter construct used with KRAB-LgnR. A three-fold

repeat repeat of LgnR binding motif A used in pMX212 is

cloned 3’ of the constitutive promoter PSV40 (PSV40-A3-SEAP-

pA).

This work


sequence (motif B1) is cloned 3’ of the constitutive

promoter PSV40 (PSV40-B1-SEAP-pA).

This work


sequence (motif C1) is cloned 3’ of the constitutive

promoter PSV40 (PSV40-C1-SEAP-pA).

This work

pMX217 Reporter construct used with KRAB-LgnR. A tandem repeat

of LgnR binding motif C used in pMX216 is cloned 3’ of the

constitutive promoter PSV40 (PSV40-C2-SEAP-pA).

This work



pMX218 Reporter construct used with KRAB-LgnR. A three-fold

repeat repeat of LgnR binding motif C used in pMX216 is

cloned 3’ of the constitutive promoter PSV40 (PSV40-C3-SEAP-

pA).

This work


sequence (motif A1) is cloned 5’ of the minimal promoter

PhCMVmin (A1-PhCMVmin-SEAP-pA).

This work

pMX221 Reporter construct used with VP16-LgnR. A tandem repeat

of LgnR binding motif A used in pMX220 is cloned 5’ of the

minimal promoter PhCMVmin (A2-PhCMVmin-SEAP-pA).

This work

pMX222 Reporter construct used with VP16-LgnR. A three-fold

repeat of LgnR bindng motif A used in pMX220 is cloned 5’

of the minimal promoter PhCMVmin (A3-PhCMVmin-SEAP-pA).

This work


sequence (motif B1) is cloned 5’ of the minimal promoter

PhCMVmin (B1-PhCMVmin-SEAP-pA).

This work


sequence (motif C1) used in pMX224 is cloned 5’ of the

minimal promoter PhCMVmin (C1-PhCMVmin-SEAP-pA).

This work

pMX225 Reporter construct used with VP16-LgnR. A tandem repeat

of LgnR binding motif C used in pMX224 is cloned 5’ of the

minimal promoter PhCMVmin (C2-PhCMVmin-SEAP-pA).

This work

pMX225 Reporter construct used with VP16-LgnR. A three-fold

repeat of LgnR binding motif C used in pMX224 is cloned 5’

of the minimal promoter PhCMVmin (C3-PhCMVmin-SEAP-pA).

This work

pMX335 Reporter construct for the ON-switch, to be used with

KRAB-LgnR. LgnR-operator sequence used in pMX223 is

cloned 3’ of a strong constitutive PhCMV promoter (PhCMV- B1-

SEAP-pA)

This work



pDF101 Inert filler plasmid bearing a bacterial T7 promoter driving

expression of an inactive ribozyme (PT7-SpAL-sTRSVac) used

to adjust plasmid amounts for transfections.

(Auslander et

al. 2016)

pTS1052 Vector for constitutive expression of secreted NanoLuc

luciferase (NLuc) and cytosolic blue fluorescent reporter

(mTagBFP) (PPGK-Igk-NLuc-P2A-mTagBFP-pA).

This work

pTS2174 Constitutive KRAB-LgnR expression vector (PhEF1α-KRAB-

LgnR-pA).

This work

pTS2178 Vector for Sleeping Beauty-mediated stable integration of

PPGK driven SMIT1 transporter and PRPBSA-driven Puromycine

and mTagBFP2 selection markers (PPGK-SMIT1-pA-PRPBSA-

mTagBFP2-P2A-PuroR-pA).

This work


PPGK driven metabolic enzymes LgnH-P2A-LgnI and LgdA

alongside PRPBSA-driven Blasticidine and mRuby2 selection

markers (PPGK-LgnH-P2A-LgnI-pA-PPGK-LgdA-pA-PRPBSA-

mRuby2-P2A-BlastR-pA).

This work

pTS2180 Constitutive expression vector for PhEF1α driven SMIT1

expression (PhEF1α -SMIT1-pA).

This work

pTS2181 Constitutive expression vector for PPGK driven SMIT1

expression (PPGK-SMIT1-pA).

This work

pTS2182 Constitutive expression vector for PhEF1α driven LgdA

expression (PhEF1α -LgdA-pA).

This work

pTS2183 Constitutive expression vector for PPGK driven LgdA

expression (PPGK-LgdA-pA).

This work

pTS2184 Constitutive expression vector for PhEF1α driven LgnH and

LgnI expression (PhEF1α -LgnH-P2A-LgnI-pA).

This work

pTS2185 Constitutive expression vector for PPGK driven LgnH and LgnI

expression (PPGK-LgnH-P2A-LgnI-pA).

This work


ON-system-controlled (PhCMV-B1) expression of rituximab

This work



and yellow fluorescent reporter YPet. A second cassette

constitutes PhEF1α -driven expression of KRAB-LgnR. The

third expression cassette harbors a PhCMV-driven Zeocin

selection marker (PhCMV-B1-rituximab-IRES-YPet-pA-PPGK-

KRAB-LgnR-pA-PhCMV-ZeoR-pA).


ON-system-controlled (PhCMV- B1) expression of rituximab

and yellow fluorescent reporter YPet. A second cassette

constitutes PPGK-driven expression of KRAB-LgnR. The third

expression cassette harbors a PhCMV-driven Zeocin selection

marker (PhCMV- B1-rituximab-IRES-YPet-pA-PPGK-KRAB-LgnR-

pA-PhCMV-ZeoR-pA).

This work

pTS2247 Production construct for L-glucose inducible production of

rituximab, to be used with KRAB-LgnR. LgnR-operator

sequence is cloned 3’ of a strong constitutive PhCMV

promoter. Both chains of rituximab are fused via a furin-T2A

sequence to ensure equimolar secretion of both chains and

a minimal addition of amino acids to the antibody

sequence. Yellow fluorescent reporter YPet is added via an

internal ribosome entry site (IRES) to allow for quick

assessment of induction (PhCMV-B1-rituximab_LC-Furin-T2A-

rituximab_HC-IRES-YPet-pA).

This work


PhCMV-driven rituximab. Both chains of rituximab are fused

via a furin-T2A sequence to ensure equimolar secretion of

both chains and a minimal addition of amino acids to the

antibody sequence. Yellow fluorescent reporter YPet is

added via an internal ribosome entry site (IRES) to allow for

quick assessment of induction (PhCMV-B1-rituximab_LC-

Furin-T2A-rituximab_HC-IRES-YPet-pA-PhCMV-ZeoR-pA).

This work



References

Ahler, E., W. J. Sullivan, A. Cass, D. Braas, A. G. York, S. J. Bensinger, T. G. Graeber, and H. R.

Christofk. 2013. 'Doxycycline alters metabolism and proliferation of human cell lines',

PLoS One, 8: e64561.

Amann, T., V. Schmieder, H. Faustrup Kildegaard, N. Borth, and M. R. Andersen. 2019. 'Genetic

engineering approaches to improve posttranslational modification of

biopharmaceuticals in different production platforms', Biotechnology and

Bioengineering, 116: 2778-96.

Bebbington, C. R., G. Renner, S. Thomson, D. King, D. Abrams, and G. T. Yarranton. 1992. 'High-

Level Expression of a Recombinant Antibody from Myeloma Cells Using a Glutamine-

Synthetase Gene as an Amplifiable Selectable Marker', Bio-Technology, 10: 169-75.

Belliveau, P. P. 2005. 'Omalizumab: a monoclonal anti-IgE antibody', MedGenMed, 7: 27.

Berger, J., J. Hauber, R. Hauber, R. Geiger, and B. R. Cullen. 1988. 'Secreted placental alkaline

phosphatase: a powerful new quantitative indicator of gene expression in eukaryotic

cells', Gene, 66: 1-10.

Brady, R. O., P. G. Pentchev, A. E. Gal, S. R. Hibbert, and A. S. Dekaban. 1974. 'Replacement

Therapy for Inherited Enzyme Deficiency - Use of Purified Glucocerebrosidase in

Gauchers-Disease', New England Journal of Medicine, 291: 989-93.

Brady, R. O., J. F. Tallman, W. G. Johnson, A. E. Gal, W. R. Leahy, J. M. Quirk, and A. S. Dekaban.

1973. 'Replacement Therapy for Inherited Enzyme Deficiency - Use of Purified

Ceramidetrihexosidase in Fabrys-Disease', New England Journal of Medicine, 289: 9-

14.

Brown, M. E., G. Renner, R. P. Field, and T. Hassell. 1992. 'Process-Development for the

Production of Recombinant Antibodies Using the Glutamine-Synthetase (Gs) System',

Cytotechnology, 9: 231-36.

Deshaies, R. J. 2020. 'Multispecific drugs herald a new era of biopharmaceutical innovation',

Nature, 580: 329-38.

Dumont, J., D. Euwart, B. Mei, S. Estes, and R. Kshirsagar. 2016. 'Human cell lines for

biopharmaceutical manufacturing: history, status, and future perspectives', Crit Rev

Biotechnol, 36: 1110-22.

References


Fan, L. C., I. Kadura, L. E. Krebs, C. C. Hatfield, M. M. Shaw, and C. C. Frye. 2012. 'Improving

the efficiency of CHO cell line generation using glutamine synthetase gene knockout

cells', Biotechnology and Bioengineering, 109: 1007-15.

Fang, J. M., S. L. Yi, A. Simmons, G. H. Tu, M. Nguyen, T. C. Harding, M. VanRoey, and K. Jooss.

2007. 'An antibody delivery system for regulated expression of therapeutic levels of

monoclonal antibodies in vivo', Molecular Therapy, 15: 1153-59.

Grabowski, G. A., N. Leslie, and R. Wenstrup. 1998. 'Enzyme therapy for Gaucher disease: the

first 5 years', Blood Reviews, 12: 115-33.

Hager, K., A. Hazama, H. M. Kwon, D. D. Loo, J. S. Handler, and E. M. Wright. 1995. 'Kinetics

and specificity of the renal Na+/myo-inositol cotransporter expressed in Xenopus

oocytes', J Membr Biol, 143: 103-13.

Hang, I., C. W. Lin, O. C. Grant, S. Fleurkens, T. K. Villiger, M. Soos, M. Morbidelli, R. J. Woods,

R. Gauss, and M. Aebi. 2015. 'Analysis of site-specific N-glycan remodeling in the

endoplasmic reticulum and the Golgi', Glycobiology, 25: 1335-49.

Hansson, K., and J. Stenflo. 2005. 'Post-translational modifications in proteins involved in

blood coagulation', J Thromb Haemost, 3: 2633-48.

Huang, M., J. Bao, and J. Nielsen. 2014. 'Biopharmaceutical protein production by

Saccharomyces cerevisiae: current state and future prospects', Pharmaceutical

Bioprocessing, 2: 167-82.

Ioannou, Y. A., D. F. Bishop, and R. J. Desnick. 1992. 'Overexpression of Human Alpha-

Galactosidase-a Results in Its Intracellular Aggregation, Crystallization in Lysosomes,

and Selective Secretion', Journal of Cell Biology, 119: 1137-50.

Jonker, H. R., R. W. Wechselberger, R. Boelens, G. E. Folkers, and R. Kaptein. 2005. 'Structural

properties of the promiscuous VP16 activation domain', Biochemistry, 44: 827-39.

Kallunki, T., M. Barisic, M. Jaattela, and B. Liu. 2019. 'How to Choose the Right Inducible Gene

Expression System for Mammalian Studies?', Cells, 8.

Kopp, J., A. M. Kolkmann, P. G. Veleenturf, O. Spadiut, C. Herwig, and C. Slouka. 2019.

'Boosting Recombinant Inclusion Body Production-From Classical Fed-Batch Approach

to Continuous Cultivation', Front Bioeng Biotechnol, 7: 297.

Kowarz, E., D. Loscher, and R. Marschalek. 2015. 'Optimized Sleeping Beauty transposons

rapidly generate stable transgenic cell lines', Biotechnol J, 10: 647-53.



Levin, G. V., L. R. Zehner, J. P. Saunders, and J. R. Beadle. 1995. 'Sugar Substitutes - Their

Energy Values, Bulk Characteristics, and Potential Health Benefits', American Journal

of Clinical Nutrition, 62: 1161-68.

Lonza. 2019. 'Lonza Chooses Synpromics’ Novel Inducible Promoter Platform to Manufacture

Biopharmaceuticals'.

Mannully, S. T., N. R. L, and K. K. Pulicherla. 2018. 'Perspectives on progressive strategies and

recent trends in the production of recombinant human factor VIII', Int J Biol Macromol,

119: 496-504.

Mansouri, M., T. Strittmatter, and M. Fussenegger. 2019. 'Light-Controlled Mammalian Cells

and Their Therapeutic Applications in Synthetic Biology', Adv Sci (Weinh), 6: 1800952.

Margolin, J. F., J. R. Friedman, W. K. Meyer, H. Vissing, H. J. Thiesen, and F. J. Rauscher, 3rd.

1994. 'Kruppel-associated boxes are potent transcriptional repression domains', Proc

Natl Acad Sci U S A, 91: 4509-13.

Molina-Henares, A. J., T. Krell, M. E. Guazzaroni, A. Segura, and J. L. Ramos. 2006. 'Members

of the IclR family of bacterial transcriptional regulators function as activators and/or

repressors', Fems Microbiology Reviews, 30: 157-86.

Monaco, C., J. Nanchahal, P. Taylor, and M. Feldmann. 2015. 'Anti-TNF therapy: past, present

and future', Int Immunol, 27: 55-62.

Mufarrege, E. F., E. L. Benizio, C. C. Prieto, F. Chiappini, M. C. Rodriguez, M. Etcheverrigaray,

and R. B. Kratje. 2020. 'Development of Magoh protein-overexpressing HEK cells for

optimized therapeutic protein production', Biotechnol Appl Biochem.

Muhsin, M., J. Graham, and P. Kirkpatrick. 2004. 'Bevacizumab', Nat Rev Drug Discov, 3: 995-

6.

Pierpont, T. M., C. B. Limper, and K. L. Richards. 2018. 'Past, Present, and Future of Rituximab-

The world's First Oncology Monoclonal Antibody Therapy', Frontiers in Oncology, 8.

Reff, M. E., K. Carner, K. S. Chambers, P. C. Chinn, J. E. Leonard, R. Raab, R. A. Newman, N.

Hanna, and D. R. Anderson. 1994. 'Depletion of B cells in vivo by a chimeric mouse

human monoclonal antibody to CD20', Blood, 83: 435-45.

Rudney, H. 1940. 'The Utilization of L-Glucose by Mammalian Tissues and Bacteria', Science,

92: 112-3.

Sasajima, K. I., and A. J. Sinskey. 1979. 'Oxidation of L-glucose by a Pseudomonad', Biochim

Biophys Acta, 571: 120-6.

References


Shimizu, T., and A. Nakamura. 2014. 'Characterization of LgnR, an IclR family transcriptional

regulator involved in the regulation of L-gluconate catabolic genes in Paracoccus sp.

43P', Microbiology, 160: 623-34.

Shimizu, T., N. Takaya, and A. Nakamura. 2012. 'An L-glucose catabolic pathway in Paracoccus

species 43P', J Biol Chem, 287: 40448-56.

Stavenhagen, K., H. M. Kayili, S. Holst, C. A. M. Koeleman, R. Engel, D. Wouters, S. Zeerleder,

B. Salih, and M. Wuhrer. 2018. 'N- and O-glycosylation Analysis of Human C1-inhibitor

Reveals Extensive Mucin-type O-Glycosylation', Mol Cell Proteomics, 17: 1225-38.

Suh, I. S., and S. B. Lee. 2003. 'A light distribution model for an internally radiating

photobioreactor', Biotechnol Bioeng, 82: 180-9.

Yalak, G., and V. Vogel. 2012. 'Extracellular Phosphorylation and Phosphorylated Proteins: Not

Just Curiosities But Physiologically Important', Science Signaling, 5.

Yusa, K., L. Q. Zhou, M. A. Li, A. Bradley, and N. L. Craig. 2011. 'A hyperactive piggyBac

transposase for mammalian applications', Proceedings of the National Academy of

Sciences of the United States of America, 108: 1531-36.

Zhang, L., S. Luo, and B. L. Zhang. 2016. 'The use of lectin microarray for assessing

glycosylation of therapeutic proteins', Mabs, 8: 524-35.

Chapter 3: Controlling gene expression through mechanical cues





Tobias Strittmatter1, Krzysztof Krawczyk2 and Martin Fussenegger1,3 designed the project.

Tobias Strittmatter1 performed the experiments, analyzed the results and wrote the

manuscript. Krzysztof Krawczyk2 performed initial experiments that are not covered in this

manuscript, provided plasmids and wrote the manuscript. Paul Argast1 and Peter Buchman1

adapted and constructed the shear stress induction device based on a previous publication.


Tobias Strittmatter1: figures 1-4

Author Affiliations 1Department of Biosystems Science and Engineering, ETH Zurich, Mattenstrasse 26, CH-4058 2Present address: Novartis Pharma AG, CH-4002 Basel, Switzerland

3Faculty of Science, University of Basel, Mattenstrasse 26, CH-4058 Basel, Switzerland

Author Affiliations




Abstract

In humans, mechanoperception contributes to the proper programming of cell fate during

embryonic development and also serves as the basis of touch and proprioception. Cellular

mechanoreceptors respond to their environment by mediating transient adjustments of ion

homeostasis or by triggering cellular responses through calcium-dependent alteration of gene

expression via specific signaling pathways such as the NFAT (nuclear factor of activated T-

cells) pathway. Application of mechanoreceptors to control gene expression in vivo have been

reported but their application requires supplements that need to be exogenously supplied.

Additionally, mechanoreceptors are drug targets for various diseases but current techniques

to study mechanically gated processes are often based on custom tailored microfluidic

systems which require special setups or have limited throughput. Here, we employ a

programmable, 96-well-format, shear-stress induction device to examine the effects of

imposing various mechanical loads on mammalian adherent cell lines, with the aim of

developing a platform to characterize shear-stress-triggered, calcium-mediated gene

expression. We show that a reporter construct utilizing the endogenous NFAT pathway

enables the detection of shear-stress-induced calcium responses. The presented method is

suitable for high-throughput experiments and provides a large tunable parameter space to

optimize conditions for different cell types. Our findings indicate that the device is an effective

tool to explore conditions in terms of frequency, intensity, intervals as well as extracellular

matrix composition alongside the evaluation of different combinations of mechanosensitive

proteins for mechanically activated gene expression. We also examined the influence of

culture-plate surface coatings and expression of accessory proteins on mechanically induced

gene expression. We believe our results can serve as a platform for further investigations into

shear stress-controlled gene expression in basic research and drug screening as well as the

development of truly autonomous mechanoperception for use in next-generation medicine.

Abstract




Introduction

Mechanical cues are omnipresent in the surroundings of cells. Being it single cells of

procaryotic or eucaryotic origin, adherent cells or free-floating cells, mechanical input is

always present to guide decisions of propagation and survival. In humans these cues provide

crucial information from early stages of embryonic cell development to guide patterning of

the emerging embryo and aid in cell differentiation (Nonomura et al. 2018; Vining and

Mooney 2017) but also in the adult to sense the environment (e.g. through senses of touch,

hearing or proprioception) as well as to control various physiological processes e.g. blood

pressure (Xu et al. 2018).

Key mediators of mechanosensation are mechanosensitive proteins in the plasma

membrane of cells. Bacterial channels like the mechanosensitive channel of small or large

conductance (MscS or MscL, respectively) provide a simple blueprint for such proteins. MscS

and MscL both consist of only a single transmembrane protein that oligomerizes in the plasma

membrane of bacteria to form mechanically gated pores that are activated by local distortions

of the membrane and serve as safety valves to protect the cell from osmotic stress (Booth et

al. 2012). This simple setup triggered research MscS and MscL to study the structures and

underlying principles (Chang et al. 1998; Liu et al. 2009). Due to the much larger and more

complex structure along with the fact that the corresponding gene remained elusive for a

long time, the crystal structures of two major mechanosensors have been solved more

recently: Piezo1, the major candidate for proprioception and potentially blood pressure

regulation (Saotome et al. 2018; Wang et al. 2018) and TMC1, a likely mediator for hearing

(Pan et al. 2018).

The purpose of mechanosensitive channels is the detection of mechanical cues for

transient adjustments of ion homeostasis or triggering of cellular responses through calcium-

dependent processes. Mechanoreceptor signaling is involved in tissue reorganization and

cell-cell alignment as seen for Piezo1 (McHugh et al. 2010; Nonomura et al. 2018). Some

alternative pathways of mechanosensation were found to modulate gene expression through

transmitting mechanical cues from the outside of the cell directly to the nucleus (Kirby et al.

2018). Here, the components of the cytoskeleton, actin filaments, microtubules or

intermediate filaments would relay the mechanical stress from their anchoring points at the

plasma membrane onto the lamin scaffold inside the nucleus. The mechanical force is

ultimately relayed via Nesprin and Sun proteins that span the nuclear envelope onto the

Introduction


chromatin where changes in chromatin accessibility are induced that lead to alterations in

gene expression.

On the other hand, the actin skeleton also contributes to the sensitivity of

mechanosensitive calcium channels by focusing mechanical stress sensed by the whole cell

directly onto the receptor (Zhang et al. 2015) or on focal adhesion points in the plasma

membrane (Hayakawa et al. 2008). This mechanism enables cells to adjust to changes in local

pressure and shear stress mediated by altered flow of extracellular fluids or changes in extra

cellular matrix (ECM) composition. In turns, changes in cell-cell contacts or adherence to the

ECM directly influence the mechanical properties of the tissue, which affects the ability of

each cell to sense mechanical cues.

This leaves us with several different pathways as well as mechanosensitive proteins

that could be harnessed for mechanically induced gene expression. To enable efficient

screening of multiple combinations of mechanosensitive components, we developed a

higher-throughput method to assess shear stress-induced gene expression. We employed this

method to explore ways to boost intrinsic as well as engineered mechanosensitivity of

mammalian cells to trigger gene expression from synthetic promoters. Parameters like

stimulus intensity, frequency, extra cellular calcium concentration and extra cellular matrix

composition as well as the presence of accessory proteins were assessed for their impact on

gene expression.

The presented findings serve as a platform for future explorations of mechano-

activated gene expression in mammalian cells.



Results

Endogenous and engineered depolarization activated gene switches

Calcium is one of the most important mediators for mechanical stress. Hence, most

mechanosensitive receptors and channels trigger changes in intracellular calcium either

directly or further downstream in the signal transduction cascade. We therefore tested a well-

established reporter construct for elevated cytosolic calcium based on NFAT-responsive

elements from the interleukin 4 (IL4) promoter cloned 5’ of a mammalian minimal promoter

(Xie et al., 2016) alongside a semi-orthogonal pathway based on an engineered calcium

sensitive tobacco etch virus (TEV) protease for use in our setup (figure 1a). In HEK-293T cells

addition of 40 mM KCl to the cell culture medium can be used to trigger a robust NFAT

response (figure 1b) most likely caused by depolarization-induced release of store-operated

calcium (Colella et al. 2008).

To enable orthogonal calcium responsive signaling that is independent of the NFAT

pathway we explored the capability of a synthetic mediator to trigger gene expression upon

KCl-mediated depolarization. This mediator consists of a split TEV protease and a calcium

responsive protein comprising a calmodulin (hCaM) moiety including the M13 helix. By fusing

the split TEV moieties on each end of the hCaM protein we expected the resulting protein to

fold up in response to calcium binding thereby reconstituting an active TEV protease similarly

to other designs (O'Neill et al. 2018b; Lee et al. 2017). To enable calcium responsive gene

expression, the reconstituted TEV protease cleaves off a tTA transcriptional activator that is

fused via a TEV cleavage site (TEVsite) to a transmembrane domain (figure 1a). Once the

TEVsite is cleaved, the tTA translocates to the nucleus activating gene expression from a

synthetic promoter equipped with 5’ response elements for tTA followed by a minimal

promoter derived from the human cytomegalo virus (PhCMV) that drives expression of a

secreted human plcental alkaline phosphatase (SEAP) reporter.

To test the TEV protease-based gene switch we first established activation of the

transcription factor in combination with a constitutively active TEV. To improve access of the

TEV protease to the membrane bound tTA, we tethered the protease to the plasma

membrane via a transmembrane domain (TM). Indeed, both, cytosolic as well as membrane

bound TEV yielded increased reporter gene expression (figure 1c). We wondered whether this

system could be also inverted by using the TEV as the mobile component while the tTA would

Results


reside in the nucleus. To this end we fused the transsilencer domain KRAB to the N- or C-

terminus of the tTA transactivator separated by a TEV cleavage site. This setup gave rise to

increased reporter gene expression upon co-expression of a nuclear-targeted TEV yielding a

functional ON-type gene switch (figure 1d). However, such an inverted system would further

complicate the experimental setup making it inferior to the membrane bound tTA system. In

the next step we replaced the constitutive TEV protease by a split TEV fused to the calcium

binding protein calmodulin (hCaM). This yielded a fusion protein comprising the C-terminal

fragment of the TEV protease (CTEV), the hCaM protein, the M13 helix peptide necessary for

hCaM function and the N-terminal TEV (NTEV) fragment (CTEV-hCaM-M13-NTEV). Upon calcium

binding, the hCaM at the core of the protein would bind to the M13 peptide and undergo a

confirmational change that would bring the ends of the protein into close proximity,

reconstituting an active TEV protease. Indeed, we were able to record increased reporter

levels in response to KCl stimulation when co transfecting this hCaM-TEV construct with the

membrane-bound tTA construct (figure 1e).

We next tested a set of known modulators of the calcium response for their ability to

boost KCl-induced calcium signaling. To this end we co-transfected HEK-293T cells with the

established NFAT reporter construct together with either single channels or combinations

thereof to probe for possible synergistic effects. Although it provides lower absolute

expression levels then the TEV-tTA gene switch, the NFAT reporter requires only one but two

components making it the reporter of choice for this screen.

Amplifiers of calcium response

Among the tested channels we also selected the voltage sensitive motor protein

prestin that is involved in the sensation of hearing (Santos-Sacchi et al. 2017). Upon

depolarization, prestin contracts alongside the membrane axis causing membrane tension to

increase. We reasoned that the heightened membrane tension would increase open

probabilities of other calcium channels like transient receptor potential (TRP) channels (Shen

et al. 2015) leading to an overall sensitization of the cell. Voltage gated calcium channel (CaV)-

mediated calcium flux across the membrane amplifies the calcium response triggered by

depolarization events. Inward rectifying potassium channels (Kir) are responsible for

decreasing the membrane potential as well as improving restoration of the potential after

depolarization (Kim et al. 2004). A stronger membrane potential has been reported to



increase calcium responses from CaV channels and this setup was essential for the

functionality of electrically induced gene switches based on HEK-293T cells in our group

before (Krawczyk et al. 2020).

Figure 1: a) schematic representation of calcium signaling by endogenous (left box) or

engineered (right box) signaling pathways. Either pathway is triggered by elevated cytosolic

a) graphical abstract

TM_TEVsit

e_tTA

+ CTEV_h

CaM_M

13_N

TEV

TM_TEVsit

e_tTA

TM_TEVsit

e_tTA

+ TEV

reporte

r only

0

5

10

15

SEAP

[UL-1

] + 40 mM KCl

uninduced

e) Calmodulin-controlled TEV

b) NFAT reporter activation by KCl

d) invertedTEV

KRAB_TEVsite

_tTA

tTA_T

EVsite_K

RABrep o

nly0

20

40

SEAP

[UL-1

]

uninduced

+ TEV-NLS

NFAT reporter0

5

10

15

SEAP

[UL-1

]

uninduced

+ 40mM KCl

c) TEV-inducible expression

TM-TEVsite-tT

A

reporte

r only

0

25

50

75

100

125

SEAP

[UL-1

]

+ cyto. TEV

+ TM TEV

uninduced

mechanical stress

Ca2+

promoter GOI promoter GOI

orthogonal signaling

cleavage

membrane bound tTA

tTA

protease

cut site

TM

NFAT

CaM

Results


calcium that is initiated by ion flux across a mechanosensitive channel like Piezo1 in the

plasma membrane. In the endogenous pathway the trimeric calcineurin A (CnA)-calcineurin

B (CnB)-calmodulin (CaM) complex is activated by calcium. The activated complex binds to

phosphorylated NFAT transcription factor and dephosphorylates it through the

phosphatase activity of CnA. Dephosphorylated NFAT can enter the nucleus and initiate

transcription from NFAT responsive promoters. In an alternative pathway we engineered a

membrane bound transcriptional activator based on the tetracycline transactivator (tTA)

protein. Here a calcium sensitive protease cleaves the tTA moiety from the membrane which

then translocates to the nucleus to initiate transcription from a corresponding synthetic

promoter. Additional channels like calcium release activated channels (CRACs) or store-

operated channels (SOC) that would amplify the calcium signal are not depicted for sake of

clarity. b) To test activity of the endogenous NFAT pathway we transfected HEK-293T cells

with a plasmid containing a NFAT responsive promoter (pMX59) that drives expression of

the reporter protein secreted human placental alkaline phosphatase (SEAP). We induced

theNFAT pathway by depolarization of the membrane by adding 40 mM of KCl to the cell

culture medium. c) Expression of a tTA responsive promoter in HEK-293T cells was recorded

in response to constitutively active tobacco etch virus (TEV) protease mediated cleavage of

a corresponding TEV cleavage site (TEVsite). A tTA transcriptional activator was fused to a

transmembrane domain (TM) via a TEVsite. Upon cleavage the tTA translocates to the

nucleus to initiate transcription. Cytosolic TEV (cyt. TEV) or membrane bound TEV (TM-TEV)

were tested. d) In an inverted setup the transsilencer domain of the Krüppel associated box

(KRAB) protein was fused to the N- or C-terminal end of a tTA via a TEVsite. Upon

localization of the TEV to the nucleus through means of an attached nuclear localization

signal (NLS) the transsilencer domain is cleaved off and tTA is able to initiate transcription.

e) By engineering a split TEV construct with its N-terminal (NTEV) or C-terminal (CTEV)

halves fused the human CaM (hCaM) protein and the M13 peptide that is essential for

stabilization of hCaM in its induced conformation, we could render TEV activity dependent

on calcium. All values are mean ± SD, N=3 biologically independent samples. Samples from

b) were taken after 24h of induction while samples from c)-e) were taken after 48h of

induction.



In contrast to calcium flowing into the cell from the extracellular side, store operated

calcium is a reservoir of calcium stored inside the ER which can be released into the cytosol,

triggering a calcium response. Therefore, store operated calcium works in synergy with

calcium flux across the plasma membrane. Members of the caffeine sensitive ryanoidin

receptor (RyR) family mediate calcium release from the ER (Lanner et al. 2010) while ATP

consuming pumps like sarcoplasmic/endoplasmic reticulum calcium ATPase 1 (ATP2A1) are

counteracting the release by shuttling cytosolic calcium back into the ER especially in muscle

cells (Salazar-Cantu et al. 2016).

We assessed prestin, hCaV 1.2, Kir2.1, a mutant version of RyR2 (RyR2-D189E) and

ATP2A1 as single components as well as in combination for their ability to boost NFAT

controlled reporter gene expression in HEK-293T (figure 2a). The single biggest contributor

was found to be the expression of hCaV1.2 in all combinations. Interestingly, RyR2-D189E also

seemed to cause superior inducibility compared to the reporter control.

a) modulators of calcium response

b) mutants of RyR2 c) overexpression of NFATc1 d) overexpression of NFATc2

reporte

r ctr

Prestin

hCaV1.2

Kir2.1

RyR2

ATP2A1

Prestin + hCaV

Prestin + Kir

Prestin + RyR

2

Prestin + ATP2A1

hCaV +

Kir

hCaV +

RyR2

hCaV +

ATP2A1

Kir + RyR

2

Kir + ATP2A1

RyR2 +

ATP2A1

Prestin + hCaV

+ Kir

Prestin + hCaV

+ RyR2

Prestin + hCaV

+ ATP2A1

Prestin + Kir +

RyR2

Prestin + Kir +

ATP2A1

Prestin + RyR

2 + ATP2A1

hCaV +

Kir + RyR

hCaV +

Kir + ATP2A1

hCaV +

RyR2 + ATP2A1

Kir + RyR

2 + ATP2A1

Prestin + hCaV

+ Kir + RyR

2

Prestin + Kir +

RyR2 + ATP2A1

Prestin + hCaV

+ RyR2 +

ATP2A1

Prestin + hCaV

+ Kir + ATP2A1

Prestin + hCaV

+ Kir + RyR

2 + ATP2A1

0

20

40

60

80

SEAP

[UL-1

]

uninduced

+ 40mM KCl

RyR2-w

T + hCaV + Kir

RyR2-E

189D +

hCaV + Kir

RyR2-R

4497

C + hC

aV + Kir

reporte

r contro

l0

2

4

6

SEAP

[UL-1

] + 40 mM KCl

uninduced

10 ng

5 ng

1 ng

0.1 ng

0 ng

0

50

100

SEAP

[UL-1

] + 40mM KCl

uninduced

20 ng

10 ng

5 ng

0 ng

0

10

20

30

40

50

SEAP

[UL-1

] + 40mM KCl

uninduced

Results


Figure 2: Modulators of calcium homeostasis were tested for their effect on NFAT-induced

gene expression. a) Prestin, human L-type voltage gated channel (hCaV1.2), inwardly

rectifying potassium channel 2.1 (Kir2.1), ryanodine receptor 2 (RyR2) and ATPase

sarcoplasmic/endoplasmic reticulum Ca2+ transporting 1 (ATP2A1) were tested as single

components or in combination for their ability to enhance KCl-induced NFAT signaling. b)

Mutants of RyR2 were tested in combination with hCaV1.2 and Kir2.1 for increased

inducibility of NFAT-mediated gene expression. Key transcription factors of the NFAT

pathway (c) NFATc1 and (d) NFATc2 were tested for beneficial effects on KCl-induced gene

expression from an NFAT-responsive synthetic promoter. All values are mean ± SD, N=3


The mutant version RyR2-D189E used in this setup was reported to result in higher

open probability (Jiang et al. 2010), however, mutant RyR2-R4496C was also found to yield

higher calcium sensitivity in previous studies (Fernandez-Velasco et al. 2009). We therefore

assessed next how the mutant versions compared to wildtype RyR2 in boosting the calcium

response (figure 2b). Indeed, both mutants showed increased calcium responses compared

to the wildtype with the strongest response being recorded using RyR2-R4496C.

We furthermore tested overexpression of key players of the NFAT pathway, NFATc1

and NFATc2 as a way to boost the sensitivity and robustness of the NFAT pathway in HEK-

293T cells. NFATc1 (figure 2c) and NFATc2 (figure 2d) both increased absolute expression

levels while slightly reducing fold switches when co-transfected with Kir2.1 and CaV1.2. We

argued that increased expression is favorable to enable detection of smaller changes in

calcium homeostasis in order to make the sometimes variable NFAT reporter performance

more robust.

Having characterized a working calcium responsive pathway in HEK-293T cells, we

now moved on to establish NFAT-mediated mechanically triggered gene expression through

application of shear stress in cell culture.

Population-based induction of shear stress through turbulent flow

Induction of shear stress on whole cell populations was achieved using a device

published for an shRNA screen targeting mechanosensitive proteins involved in calcium

signaling (Xu et al. 2018). The device consists of a 3D-printed array of 96 pistons adapted for



a regular 96-well format which is glued to the membrane of a loud speaker. The speaker in

turns is controlled by a signal generator based on an Arduino controller coupled to an

operational amplifier and a hi-fi stereo amplifier. When running, the pistons would move up

and down inside the cell culture wells to generate turbulent flow and hence shear stress on

the bottom of the well. First tests with the genetically encoded calcium indicator (GECI)

GCaMP6s (Chen et al. 2013) in cultured mouse neurons to record calcium signals in response

to mechanical stimulation proved the device was functional (Gaub et al. 2020).

a) shear-stress response dependent on frequency and reporter

pYL1pAT50

pHY30

pMX57

pMX58

tTA-base

d0

5

10

15

SEAP

[U/L

] 10 Hz

30 Hz

60 Hz

uninduced

b) shear-stress response of wT HEK

0

20

40

60

nLuc

/ SE

AP

[AU]

10 Hz

30 Hz

60 Hz

uninduced

4xNFAT-NLuc

c) mPiezo1 modulates response

0

20

40

60

nLuc

/ SE

AP

[AU]

10 Hz

30 Hz

60 Hz

uninduced

4xNFAT-NLuc

d) stable mPiezo1 modulates timing and duty-cycle dependent response

pcTS1 wT

0

20

40

60

80

100

nLuc

/ SE

AP

[AU]

pcTS1 wT

0

50

100

nLuc

/ SE

AP

[AU]

1.7 %

3.3 %

7.7 %

uninduced

(i) day 1 (ii) day 2

Results


Figure 3: A 96-well piston device was tested for the ability to modulate calcium induced

gene expression upon application of shear stress generated through turbulent flow

produced by oscillating pistons. a) To find the best calcium responsive promoter, a set of

published synthetic reporter constructs was tested with different frequencies. Induction was

done for 5 sec every 5 min at indicated frequencies for 24h and compared to corresponding

uninduced samples. b) To account for unspecific effects a dual-reporter setup was

introduced using NFAT-responsive expression of secreted NanoLuc luciferase (nLuc) and

constitutive expression of SEAP from a weak constitutive promoter derived from the simian

virus 40 (PSV40). By normalization of nLuc activity to SEAP expression a more robust signal

was generated that showed a clear dose dependent increase of NFAT activity in HEK-293T

cells upon stimulation with the piston device. c) Mechanosensitive channel Piezo1 was

overexpressed in HEK-293T cells grown for additional 24h after transfection under standard

culturing conditions to allow for proper expression of the channel prior to stimulation as

done in b). d) To ensure more homogenous expression of Piezo1, we generated a stable HEK-

293T cell line expressing Piezo1 (pcTS1). Induction of the cell line either (i) directly after

transfection or (ii) after 24h of regeneration was assessed. All values are mean ± SD, N=3


The used NFAT reporter construct is well characterized and proven to be sensitive to

depolarization induced perturbations in calcium homeostasis. However, stimuli from

mechanoreceptors might differ in terms of downstream signaling from KCl induced calcium

signaling. We therefore screened various published reporter constructs that employ different

combinations of NFAT binding sites derived from calcium responsive IL2 and IL4 promoters as

well as the serum response element (SRE) alongside our TEV-tTA circuit for their response to

different frequencies of shear stress inducing oscillations. We again found an IL4-derived

NFAT promoter (pMX57) to yield the biggest fold changes with a peak at 30 Hz induction

frequency (figure 3a). Unfortunately, the semi-orthogonal TEV-tTA-based gene switch that

proved to be sensitive to KCl-mediated depolarization failed to record induction from shear

stress. Hence, we focused on the NFAT pathway for further studies. Based on the promoter

used in the pMX57 construct we designed a new reporter cassette and switched the reporter

gene from SEAP to NanoLuc luciferase (nLuc). This improved sensitivity and enabled the use

of SEAP as a constitutive reference for internal normalization to compensate for possible



unspecific effects caused by the piston device. Functionality of the reporter system was

confirmed by shear stress induction (figure 3b).

We finally tested transient expression of Piezo1 in HEK-293T cells to boost

mechanosensitivity and hence NFAT-mediated mechanically triggered gene expression.

Piezo1 expression yielded higher overall expression but lower fold switches compared to cells

without overexpression of mechanoreceptors (figure 3c). Additionally, for Piezo1 transfected

cells we did not see an improvement when using 30 Hz or 60 Hz for induction.

a) NFAT overexpression improves signal

+ NFAT1/2

+ NFAT1

+ NFAT2

reporte

r only

0

50000

100000

150000

200000

nLuc

/ SE

AP [A

U]

uninduced

3.3 %

14.3 %

b) extracellular calcium levels

c) shear stress response in HEK-293T cells can be modulated by ECM and mechanoreceptors

d) shear stress response in CHO-K1 cells can be modulated by ECM and mechanoreceptors

1.8 m

M CaCl2

5 mM CaCl2

10 m

M CaCl2

0

200

400

600

nLuc

/ fL

uc

uninduced

+ KCl 40 mM

(i) HEK-293T (ii) HEK-293T + MscL (iii) HEK-293T + Piezo1

uncoated

+ 1:25

0 Matrig

el

+ 1:50

Matr

igel

+ 37.5

g/mL getre

x

+ 75g/m

L getrex

+ 150

g/mL getre

x0

5

10

15

nLuc

/ fL

uc

uninducedinduced

uncoated

+ 1:25

0 Matrig

el

+ 1:50

Matr

igel

+ 37.5

g/mL getre

x

+ 75g/m

L getrex

+ 150

g/mL getre

x0

200

400

600

800

nLuc

/ fL

uc

uncoated

+ 1:25

0 Matrig

el

+ 1:50

Matr

igel

+ 37.5

g/mL getre

x

+ 75g/m

L getrex

+ 150

g/mL getre

x0

200

400

600

800

(i) CHO-K1 (ii) CHO-K1 + MscL (iii) CHO-K1 + Piezo1

uninducedinduced

uncoated

+ 1:25

0 Matrig

el

+ 1:50

Matr

igel

+ 37.5

g/mL getre

x

+ 75g/m

L getrex

+ 150

g/mL getre

x0

5

10

15

uncoated

+ 1:25

0 Matrig

el

+ 1:50

Matr

igel

+ 37.5

g/mL getre

x

+ 75g/m

L getrex

+ 150

g/mL getre

x0

5

10

15

uncoated

+ 1:25

0 Matrig

el

+ 1:50

Matr

igel

+ 37.5

g/mL getre

x

+ 75g/m

L getrex

+ 150

g/mL getre

x0

200

400

600

800

1000

Results


Figure 4: a) NFATc1 (NFAT1) and NFATc2 (NFAT2) were tested as single components or in

combination (NFAT1/2) in response to 5 sec each 2.5 min or 5 sec each 0.5 min of

stimulation corresponding to 3.3 % or 14.3 % duty cycle for improved shear stress induced

gene expression. To improve robustness of the system and streamline future experiments,

stable cell lines were generated by designing a vector construct comprising NFAT-responsive

nLuc expression, constitutive expression of firefly luciferase (fLuc) and a selection cassette

conferring puromycine resistance as well as promoting expression of additional NFATc1

transcription factor from an attached IRES element. Resulting stable cell lines were used to

b) test different concentrations of extracellular calcium for improved responses to KCl

induction or shear stress (10 sec each 20 sec or 33.3 % duty cycle) mediated gene expression

in c) HEK-293T or d) CHO-K1 cell lines grown on different extra cellular matrix (ECM)-

simulating compounds upon additional expression of mechanoreceptors (ii) MscL or (iii)

Piezo1. All values are mean ± SD, N=3 biologically independent samples.

Since we experienced expression of Piezo1 to be slow and heterogenous within a

transiently transfected cell population, we next selected a HEK-293T cell line stably expressing

Piezo1 (pcTS1). We speculated that cell-cell adherence might be a major contributor to

improved mechanosensitivity and we therefore used this cell line to both confirm prolonged

pre-induction cultivation to boost induction folds and to test different induction protocols

using induction times of 5 seconds every 5, 2.5 or 1 minutes (translating to duty cycles of 1.7,

3.3 and 7.7 %, respectively). We found inducibility on day 2 to be superior with higher duty

cycles promoting higher fold switches in engineered as well as wild-type cells (figure 3d).

Drawing from previous experiments (figure 2c-d), we tested expression of additional

NFATc1 and NFATc2 to improve mechanosensitivity of the NFAT pathway. Similar to what we

saw from KCl induction, expression of NFATc1 was able to improve fold switches

tremendously in response to shear stress induction in a dose dependent manner, finally

offering a working setup for population-based mechanically induced gene expression in HEK-

293T cells (figure 4a).

To simplify the experimental design, we created a reporter construct that contains the

already established NFAT-reporter cassette as well as a constitutively expressed firefly

luciferase driven by a phosphoglycerate kinase promoter (PPGK) replacing the SEAP reporter

used in previous assays. In addition to the two reporter cassettes, we added a PRPBSA-driven



selection cassette to the construct that expresses a polycistronic RNA comprising an infra-red

fluorescent protein (iRFP) coupled to a puromycin resistance gene (puroR) via a p2a sequence

but also contains the CDS for NFATc1 which is driven by an internal ribosome entry site (IRES)

(PRPBSA-iRFP-p2a-puroR-IRES-NFATc1). By using cytosolic versions of the luciferase reporters,

we also aimed to reduce dependency on the secretory pathway to avoid artifacts caused by

expression of multiple membrane proteins like the channels used in this study. We tested a

polyclonal stable HEK-293T cell line by assessing the effect of three different concentrations

of calcium in the cell culture medium. Standard DMEM culture medium contains 1.8 mM of

calcium. We tested whether higher concentrations of calcium in the medium provides a

stronger trigger for KCl-induced NFAT signaling. In fact, we found medium supplemented with

5 or 10 mM calcium to promote higher fold switches compared to the standard formulation

(figure 4b). From there on we changed the standard medium composition to also contain a

final concentration of 10 mM calcium during shear stress induction.

Further experiments were conducted to characterize shear stress-responses of the

stable HEK-293T reporter cell line when grown on different surface coatings (Matrigel or

geltrex) or when transfected with mechanoreceptors MscL or Piezo1 (figure 4c). Induction

was performed using 30 Hz and a 10-sec-on-20-sec-off protocol (translating to 33 % duty

cycle). While surface coatings were ineffective in changing inducibility in untransfected stable

cells (figure 4c, i), low-density Matrigel coatings showed improved fold-switches in cells

transfected with MscL (figure 4c, ii). However, this could not be reproduced in cells

transfected with Piezo1 (figure 4c, iii), possibly due to a shortened incubation time of 24h.

In contrast, when conducting the same set of experiments in a stable CHO-K1 cell line,

we found opposite effects. For one, cell surface coatings showed increased performance even

in untransfected cells, with 37.5 µg/mL geltrex showing highest but 150 µg/mL geltrex lowest

inducibility (figure 4d, i). On the other hand, this cell line showed a strong increase in fold-

switches when transfected with mechanoreceptors MscL (figure 4d, ii) or Piezo1 (figure 4d,

iii) with low-density Matrigel or 37.5 µg/mL geltrex providing the biggest increase while

responses from cells grown in uncoated wells stayed the same.

Results




Discussion

This study provides a platform for further characterization of shear stress-triggered,

calcium-mediated gene expression in mammalian adherent cell lines. We present two

different gene circuits that can be used for this purpose based on the endogenous NFAT

pathway and an engineered split-TEV-calmodulin fusion protein which enables the detection

of KCl-induced calcium signals in a semi-orthogonal manner. Intriguingly, induction of calcium

signaling via shear stress seems to trigger different downstream pathways than KCl-mediated

depolarization as both methods induce a different set of known genetic reporters for calcium

signaling (figure 2a). The inability to record calcium responses via the TEV-tTA switch might

lie in a higher concentration of intracellular calcium needed for activation as well as prolonged

duration of calcium transients. The lack of an amplification step as it is typically seen in native

signaling pathways, including the NFAT pathway, furthermore makes the TEV-tTA switch

more dynamic but also less sensitive. Since a similar strategy was reported to work with low

fold switches in HEK-293T cells (Wang et al. 2017) but exceled in neurons (O'Neill et al. 2018a;

Wang et al. 2017), which exhibit much larger changes in calcium levels upon stimulation,

future iterations of the TEV-tTA design should aim to render the construct more sensitive to

work at fold changes that are within the range of non-neuronal cell types.

While the L-type calcium channel CaV1.2 was already successfully employed in

therapeutic settings using electrically induced or depolarized cells to treat diabetes and

chronic pain (Xie et al. 2016; Wang et al. 2018; Krawczyk et al. 2020), the potential of the

other big contributor to calcium signaling, RyR2 channels, has yet to be explored in this

regard. Since RyR2 channels synergize with the established CaV-Kir-combination one could

expect superior sensitivity of the resulting gene network. Each of the channel families again

contains additional members that could be characterized. Hence, the selection studied here

is by far not comprehensive. Because of the sophisticated architecture of some of the

channels that involves multiple large subunits (Zamponi et al. 2015), studying responses by

means of transient expression might be challenging.

The differences in reporter signals seen in response to shear stress versus KCl-

mediated depolarization already indicate different pathways being triggered by each

stimulation method. Therefore, reporters that tap into alternative signaling pathways e.g.

cytoskeleton-mediated pathways (Kirby et al. 2018; Zhang et al. 2015) might provide an

additional means to realize mechanically induced gene expression.

Discussion


Intriguingly extracellular matrix-simulating coatings like matrigel and geltrex provided

a strong boost in inducibility when used in combination with mechanoreceptors MscL in HEK-

293T as well as MscL and Piezo1 in CHO-K1 cells. This might indicate a potential mechanism

of activation of these receptors and should be carefully considered in subsequent

experiments. Dependency of mechanosensitivity on ECM was reported before for Piezo1

(Gaub et al. 2017).

Future research should focus on expanding the repertoire and knowledge about

mechanosensitive proteins that could be used to boost shear stress induced gene expression

as well as filling in the gaps of the parameter space of the induction protocol. The device itself

would greatly benefit from a way to calibrate the intensities of the provided oscillations to

reduce variability between devices. A further splitting of the array into subgroups that could

be individually controlled would be useful, too. Finally, other ways of inducing shear stress

that are not based on turbulent flow but would include laminar flow as well, should be

explored. Here, a small propeller or turbine could be considered that could be individually

controlled and would allow for rotation in both directions to induce laminar or turbulent flow

from the same setup.

We are convinced, that the presented method for shear stress mediated gene

expression can be useful in applications beyond basic research for example in screenings for

compounds targeting mechanosensitive signaling pathways and receptors in a more holistic

model that does not look at calcium transients alone but includes the underlying signaling

cascades as well. On the other hand, mechanosensitive circuits could not only be used in

combination with exogenous stimulation but could also sense physiological stimuli like flow

rate inside the vasculature. In potential future-medicine applications, cells that contain

mechanosensitive gene circuits, could potentially be employed to detect aberrant levels of

blood pressure in the context of cardio-vascular diseases or as part of stent implants to report

on or even counteract harmful occlusion events.



Materials & Methods

Cell culture

HEK293-T cells and CHO-K1 cells were cultured in 10 cm cell culture dishes in

Dulbecco’s Modified Eagle Medium (DMEM, gibco, #61965-026) supplemented with 10 %

FCS, 50 Units/mL Penicillin and 50 µg/mL of Streptomycin (gibco, #15070-063) in a humidified

atmosphere containing 7.5 % CO2. Medium was supplemented with 150 µM of L-proline for

culturing of CHO-K1 cells. All cell lines were split using Trypsin-EDTA (gibco, #25300-054)

every two days or when growing to more than 70 % confluency and reseeded at 1.5 million

cells. For cell culture experiments, cells were seeded at indicated concentrations (15 000, 30

000, 45 000 or 60 000 cells per well) in 96-well plates (Corning, #3599) or 250 000 cells per

well in a 6-well plate (Corning, #3516) for generation of stable cell lines.

Plasmid generation and transfection

All plasmids were cloned by standard molecular cloning techniques. See table 1 for a

comprehensive list of plasmids and the genetic elements they carry. Details on amounts of

transfected plasmids for each experiment can be found in table 2.

Polyethyleneimine (PEI, 24765-1, Polysciences Inc.) was used for transfection of

plasmid DNA in a ratio of PEI:DNA of 6:1. 24 h prior to transfection cells were seeded as

described above. 200 ng or 1 000 ng was mixed with 150 µL or 500 µL of serum-free DMEM

and 1.2 µg or 6 µg PEI (1.2 µL or 6 µL of a 1 mg/mL stock solution) was added for transfecting

one well of a 96-well plate or 6-well plate, respectively. The mixture was incubated at room

temperature for 5-10 minutes before adding it to the cells. Cells were incubated overnight

(>12 h) under standard culturing conditions. Medium was replaced by standard culture

medium (as described above) the next morning and inducer molecules were added as

necessary.

Stimulation of shear stress using the piston device

A description of the device and along with a schematic drawing can be found in the

main text. Each device is controlled by an Arduino controller coupled to an operational

amplifier with manual potentiometer. The resulting signal (intervals of signals of 3-60 Hz) is

fed into the speaker via a commercial hi-fi stereo amplifier. A gyrostat sensor attached to the

Materials & Methods


membrane of the speaker measures the force on the array during operation reporting

changes in acceleration in volt and thereby provides a means to calibrate and adjust induction

intensities. This setup allows for the generation of fully customizable stimulation patterns by

allowing for selection of specific frequencies, intensities and intervals.

Experiments with this shear stress induction device were carried out in standard cell

culture incubators. Pistons were cleaned before each use by incubation in 1 M NaOH for at

least 10 min. Pistons were rinsed with ddH2O and dried in a cell culture flow-hood with 30

min of UV-sterilization. Devices were pre-warmed for a couple of hours in the cell culture

incubator to allow condensation to evaporate.

Generation of stable cell lines

For generation of stable cell lines cells were seeded in 6-well plates and transfected

with donor vectors for the sleeping beauty transposase system as described above using 600

ng of donor vector and 400 ng of pTS395 (PhCMV-SB100X-pA). Selection was done using 1

µg/mL of puromycine (invivogen, #ant-pr) for at least 10 days.

HEK-293T cells stably expressing mammalian Piezo1 were generated by transfecting

pTS391 and pTS395 while HEK-293T and CHO-K1 cells carrying the NFAT-dependent NanoLuc

luciferase (NLuc) reporter, NFATc1 expression cassette and constitutive expression of firefly

luciferase (fLuc) were generated using pTS2258 and pTS395.

Assessment of reporter gene expression

Activity of the reporter protein secreted alkaline placental phosphatase (SEAP) was

performed as previously described (Berger et al. 1988). In short, supernatant from cell culture

experiments was heat treated for 30 min at 65 °C to inactivate endogenous phosphatases. 20

µL of heat-treated supernatant was mixed with 60 µL of water and 80 µL of 2x assay buffer (1

M diethanolamine pH 9.8, 0.5 mM MgCl2, 10 mM L-homoarginine). Reaction was started by

the addition of 20 µL substrate solution (120 mM p-nitrophenylphosphate in 2x assay buffer).

Reporter activity was assessed by measuring the reaction kinetics at 37 °C by following

substrate turn-over at 405 nm in an Infinite M1000 microplate reader (Tecan Trading AG).

NanoLuc Luficerase (NLuc) activity from cell culture supernatant was recorded as

recommended by the manufacturer. 7.5 µL of Nano-Glo® Assay Reagent was mixed with



7.5 µL of supernatant per well of a 384-well plate and luminescence was measured in an

Infinite M1000 microplate reader (Tecan Trading AG) with 100 ms integration time.

NanoLuc Luficerase (NLuc) and Firefly Luciferase (FLuc) activity was measured from

cell lysates using Nano-Glo® Luciferase Assay (Promega) and ONE-Glo™ Luciferase Assay

(Promega) kits. To this end supernatant of cells was aspirated and replaced by 100 µL of 1x

Passive Lysis Buffer (Promega). Cells were incubated for up to 30 min at room temperature

while shaking at 400 rpm. 7.5 µL of Nano-Glo® or One-Glo™ Assay Reagent was mixed with

7.5 µL of lysate per well of a 384-well plate. Luminescence was recorded with 100 ms

integration time on an Infinite M1000 microplate reader (Tecan Trading AG).

Statistical Analysis

No statistical analysis was performed. Generally, data is presented as bar graphs

representing means ± SD of N=3 or N=6 biologically independent samples.

Materials & Methods







mPiezo1-

IRES-

eGFP

Mammalian expression vector of mammalian Piezo1 coupled

to internal ribosome entry site (IRES)-mediated expression of

enhanced green fluorescent protein (eGFP) from a constitutive

promoter derived from the human cytomegalo virus (PCMV)

(PCMV-CaV1.2-pA)

(Coste et al.

2010)

hCav1.2 Mammalian expression vector for channel forming subunit α1

of mammalian voltage gated channel 1.2 (CaV1.2) of red

fluorescent protein from a constitutive promoter derived from

the human cytomegalo virus (PCMV) (PCMV-CaV1.2-pA)

(Helton, Xu,

& Lipscombe,

2005)

pKK5 Mammalian expression vector for inward rectifying potassium

channel 2.1 (Kir2.1) from a constitutive promoter derived from

the human cytomegalo virus (PCMV) (PCMV-Kir2.1-pA)

(Krawczyk et

al. 2020)

RyR2 Mammalian expression vector for expression of wild type

ryanodine receptor 2 (RyR2) from a constitutive promoter

derived from the human cytomegalo virus (PCMV) (PCMV-RyR2-

pA)

(Jiang et al.

2010)

RyR2-

E189D

Mammalian expression vector for expression of mutant

ryanodine receptor 2 E189D (RyR2-E189D) from a constitutive


(PCMV-RyR2-E189D-pA)

(Jiang et al.

2010)

RyR2-

R4496C

Mammalian expression vector for expression of mutant

ryanodine receptor 2 R4496C (RyR2-R4496C) from a

constitutive promoter derived from the human cytomegalo

virus (PCMV) (PCMV-RyR2-R4496C-pA)

(Jiang et al.

2010)

pFS29 Mammalian expression vector for expression of red fluorescent

protein mCherry from a constitutive promoter derived from

simian virus 40 (PSV40) (PSV40-mCherry-pA)

(Sedlmayer

et al. 2018)



pDF101 Inert filler plasmid bearing a bacterial T7 promoter driving an

inactive ribozyme (PT7-SpAL-sTRSVac)

(Auslander et

al. 2018)

pFOX6 Mammalian expression vector for expression of red fluorescent

protein tdTomato from a constitutive promoter derived from

the human cytomegalo virus (PCMV) (PCMV-tdTomato-pA)

unpublished

pFOX6 Mammalian expression vector for expression of yellow

fluorescent protein YPet from a constitutive promoter derived

from the human cytomegalo virus (PCMV) (PCMV-YPet-pA)

unpublished

pSEAP2-

Control

Constitutive SEAP expression vector (PSV40-SEAP-pA). Clonetech

pMX57 Reporter plasmid for NFAT-induced expression of SEAP from a

NFAT responsive synthetic promoter containing five response

elements derived from the IL4 promoter (PNFAT5). (PNFAT5-

Citrine-2A-SEAP-pA)

(Xie et al.

2016)

pMX58 Reporter plasmid for NFAT-induced expression of SEAP from a

NFAT responsive synthetic promoter containing seven

response elements derived from the IL4 promoter (PNFAT7).

(PNFAT7-Citrine-2A-SEAP-pA)

(Xie et al.

2016)

pDA326 Reporter plasmid for tTA induced expression of yellow

fluorescent protein citrine and SEAP from a TREB1 promoter. In

order to reduce leakiness of the induction a destabilizing

ribozyme (sTRSV) is added to the 3’ of the construct. (PTREBI-

Citrine-2A-SEAP-sTRSV-pA)

(Auslander et

al. 2018)

pPST218 Mammalian expression vector for expression of tobacco etch

virus protease (TEV) from a constitutive promoter derived from

the human cytomegalo virus (PCMV) (PCMV-TEV-pA)

unpublished

pLeo7 Mammalian expression vector for expression of tobacco etch

virus protease (TEV) bound to the membrane by secretion

signal (SS) mediated insertion of the transmembrane domain

(TM) of the platelet-derived growth factor receptor (PDGFR)

from a synthetic constitutive promoter made up of parts from

unpublished



a promoter derived from the human cytomegalo virus, the

chicken beta-actin promoter and the rabbit beta-globin splice

acceptor (CAG; PCAG) (PCAG-SS-TM-TEV-pA)

pMM545 Mammalian expression vector for expression of yellow

fluorescent protein citrine from a constitutive promoter

derived from the human cytomegalo virus (PCMV) (PCMV-citrine-

pA)

unpublished

pVH288 Mammalian expression vector for expression of yellow

fluorescent protein citrine and SEAP from the constitutive

phosphoglycerate kinase promoter (PPGK) (PPGK-Citrine-p2A-

SEAP-pA)

Haellmann et

al.

(unpublished)

pYL01 Calcium sensitive reporter plasmid expressing SEAP reporter

protein.

(Liu et al.

2018)

pHY30 Calcium sensitive reporter plasmid expressing SEAP reporter

protein.

(Ye et al.

2011)

pAT50 Calcium sensitive reporter plasmid expressing SEAP reporter

protein.

(Tastanova et

al. 2018)

pTS75 PCMV-driven mammalian expression vector for a calpain-

induced tTA transactivator. tTA is inserted and anchored in the

plasma membrane via a secretion signal (SS) fused to a

transmembrane domain (TM) and a linker sequence containing

a calpain cleavage site (CCS). (PCMV-SS-TM-CCS-tTA)

This work

pTS87 Mammalian expression vector for expression of rat calpain

from a weak constitutive promoter derived from simian virus

40 (PSV40). (PSV40-calpain-pA)

This work

pTS90 Mammalian expression vector for expression of the small

calpain subunit from a weak constitutive promoter derived

from simian virus 40 (PSV40). (PSV40-calpain-pA)

This work

pTS96 PCMV-driven mammalian expression vector for a potential TEV-

induced tTA transactivator. tTA is localized to the cytosol via a

nuclear export signal (NES) fused via linker sequence containing

This work



a TEV cleavage site (TEVsite) to a tTA carrying a nuclear

localization signal (NLS). (PCMV-NES-TEVsite-tTA-NLS)

pTS99 PCMV-driven mammalian expression vector for a TEV-induced

tTA transactivator. tTA is inserted and anchored in the plasma

membrane via a secretion signal (SS) fused to a transmembrane

domain (TM) and a linker sequence containing a TEV cleavage

site (TEVsite). (PCMV-SS-TM-TEVsite-tTA)

This work

pTS229 PCMV-driven mammalian expression vector for a split TEV

construct based on a human calmodulin (hCaM) that is fused to

the M13-peptide. C-terminal and N-terminal split TEV

fragments (CTev and NTev, respectively) are fused to the N-

terminus and C-terminus of the hCaM core protein. (PCMV-CTev-

hCaM-M13-NTev)

This work

pTS304 Mammalian expression vector of tobacco etch virus protease

(TEV) localized to the nucleus via a nuclear localization signal

(NLS) driven by a constitutive promoter derived from the

human cytomegalo virus (PCMV) (PCMV-TEV-NLS-pA)

This work

pTS305 PCMV-driven mammalian expression vector of TEV-induced tTA

transactivator. tTA-mediated expression is blocked by a KRAB

transsilencer domain fused to the N-terminus of tTA via a linker

sequence containing a TEV cleavage site (TEVsite) (PCMV-KRAB-

TEVsite-tTA)

This work

pTS307 PCMV-driven mammalian expression vector of TEV-induced tTA

transactivator. tTA-mediated expression is blocked by a KRAB

transsilencer domain fused to the C-terminus of tTA via a linker

sequence containing a TEV cleavage site (TEVsite) (PCMV-tTA-

TEVsite-KRAB)

This work

pTS391 Vector for stable integration of two expression cassettes

flanked by insertion and recognition sites of sleeping beauty

transposase (SB). Cassette one contains a PCMV-driven

mammalian Piezo1 (mPiezo1). Cassette two comprises a PRPBSA-

This work



driven blue fluorescent protein (BFP) coupled to a puromycin

resistance marker (puroR). (SB-(1)CMV-mPiezo1-pA-

(2)RPBSA_BFP_p2a_PuroR-pA-SB)

pTS395 Mammalian expression vector for SB100 sleeping beauty

transposase driven by a constitutive promoter derived from the

human cytomegalo virus (PCMV) (PhCMV-SB100-pA).

This work

pTS737 Mammalian expression vector for expression of prestin from a


virus (PCMV) (PCMV-prestin-pA) prestin was isolated from cDNA

clone HsCD00436673 (Yang et al. 2011)

This work

pTS774 Mammalian expression vector for expression of

sarcoplasmic/endoplasmic reticulum calcium ATPase 1

(ATP2A1) from a constitutive promoter derived from the

human cytomegalo virus (PCMV) PCMV-ATP2A1-pA. ATP2A1 was

extracted from cDNA clone HsCD00731696.

This work

pTS775 Mammalian expression vector for expression of nuclear factor

and activator of transcription c1 (NFATc1) from a constitutive


(PCMV-NFATc1-pA)

This work

pTS776 Mammalian expression vector for expression of nuclear factor

and activator of transcription c2 (NFATc2) from a constitutive


(PCMV-NFATc2-pA)

This work

pTS1054 Mammalian expression vector for expression of bacterial

mechanosensitive channel of large conductance (MscL) from a


virus (PCMV) (PCMV-MscL-pA)

This work

pTS2010 Reporter plasmid for NFAT-induced expression of secreted

nLuc reporter and blue fluorescent protein mTagBFP2 from a

NFAT responsive synthetic promoter containing four response

This work



elements derived from the IL4 promoter (PNFAT4). (PNFAT4-SS-

nLuc-P2A-mTagBFP2-pA)

pTS2258 Vector for stable integration of three expression cassettes flanked by insertion and recognition sites of sleeping beauty transposase (SB). First cassette carries a reporter construct for NFAT-induced gene expression of cytosolic nLuc reporter coupled to blue fluorescent protein (mTagBFP2). Second cassette contains infra-red fluorescent reporter protein (iRFP) coupled to a resistance gene for puromycin (puroR) driven by a synthetic RPBSA promoter (PRBSA) and an ORF encoding nuclear factor and activator of transcription c1 (NFATc1) driven by an internal ribosome entry site (IRES). The third cassette enables expression of cytosolic firefly luciferase (fLuc) reporter for internal normalization of nLuc expression from a constitutive phosphoglycerate kinase promoter (PPGK). (SB-(1)PNFAT4-nLuc-P2A-mTagBFP2-pA-(2)RPBSA-iRFP-P2A-PuroR-IRES-NFATc1-pA-(3)PPGK-fLuc-pA-SB)

This work



Table S2: Plasmids transfected in each experiment

Details on the transfection procedure are given in the methods section.

Figure Plasmids used ng per well Goal of

experiment

1 b) pMX58

pDF101

50

ad. to 200

Verify KCl-induced

expression of

NFAT-responsive

promoter

1 c) pTS99

pPST218/pLeo7

pDA326

pFS29

100

0/10

40

ad. to 250

Test inducible TEV

reporter construct

with cytosolic TEV

and membrane

bound TEV

1 d) pTS305/307

pTS304

pDA326

pMM545

20

40

50

ad. to 250

Test inverted

KRAB-based TEV

reporter construct

1 e) pTS299

pTS99

pDA326

pFOX6

50

16.7

50

ad. to 250

Test calmodulin-

based split TEV

calcium reporter

construct

2 a) pTS737/hCaV/pKK5

RyR2-E189D

pTS774

pMX58

pDF101

0/5

0/7.5

0/10

50

ad. to 200

Test various

channel proteins

for improved

performance of

NFAT reporter

upon KCl

induction

2 b) RyR2/RyR2-E189D/RyR2-R4496C

hCaV

pKK5

5

50

12.5

Test mutant

versions of RyR2

channel protein



pMX57

pFOX6

50

ad. to 200

for improved

performance of

NFAT reporter

upon KCl

induction

2 c) pTS775

hCaV

pKK5

pMX58

pDF101

20/10/5/0

5

5

50

ad. to 200

Test effects of

NFATc1

overexpression on

NFAT responses

with KCl induction

2 d) pTS775

hCaV

pKK5

pMX58

pDF101

10/5/1/0.1/0

5

5

50

ad. to 200

Test effects of

NFATc2

overexpression on

NFAT responses

with KCl induction

3 a) pMX57/58/pYL01/pHY30/pAT50

pDF101

50

ad. to 200

Screen for optimal

reporter construct

for shear stress

induced calcium-

dependent gene

expression

3 b) pTS2010

pSEAP2-Control

pDF101

50

50

ad. to 200

Dependency of

shear stress

induced NFAT-

mediated gene

expression on

frequency

3 c) mPiezo1-IRES-eGFP

pTS2010

pSEAP2-Control

pDF101

10

50

50

ad. to 200

Effect of mPiezo1

overexpression on

shear stress



induced gene

expression

3 d) pTS2010

pSEAP2-Control

pDF101

50

50

ad. to 200

Dependency of

shear stress

induced NFAT-

mediated gene

expression on

timing in stable

cell line and wT

cells

4 a) pTS2010

pVH288

pTS775

pTS776

pFS29

80

40

10/0

1/0

ad. to 200

Test effects of

NFATc1 or NFATc2

overexpression on

NFAT responses

under shear stress

4 c)-d) mPiezo1-IRES-eGFP/pTS1054/pFox8

pFS29

10

ad. to 200

Assess effects of

coatings and

overexpression of

mechanoreceptors

MscL and mPiezo1

on NFAT activity





References

Auslander, D., S. Auslander, X. Pierrat, L. Hellmann, L. Rachid, and M. Fussenegger. 2018.

'Programmable full-adder computations in communicating three-dimensional cell

cultures', Nat Methods, 15: 57-60.

Berger, J., J. Hauber, R. Hauber, R. Geiger, and B. R. Cullen. 1988. 'Secreted placental alkaline

phosphatase: a powerful new quantitative indicator of gene expression in eukaryotic

cells', Gene, 66: 1-10.

Booth, I. R., and P. Blount. 2012. 'The MscS and MscL families of mechanosensitive channels

act as microbial emergency release valves', J Bacteriol, 194: 4802-9.

Chang, G., R. H. Spencer, A. T. Lee, M. T. Barclay, and D. C. Rees. 1998. 'Structure of the MscL

homolog from Mycobacterium tuberculosis: a gated mechanosensitive ion channel',

Science, 282: 2220-6.

Chen, T. W., T. J. Wardill, Y. Sun, S. R. Pulver, S. L. Renninger, A. Baohan, E. R. Schreiter, R. A.

Kerr, M. B. Orger, V. Jayaraman, L. L. Looger, K. Svoboda, and D. S. Kim. 2013.

'Ultrasensitive fluorescent proteins for imaging neuronal activity', Nature, 499: 295-

300.

Colella, M., F. Grisan, V. Robert, J. D. Turner, A. P. Thomas, and T. Pozzan. 2008. 'Ca2+

oscillation frequency decoding in cardiac cell hypertrophy: Role of calcineurin/NFAT

as Ca2+ signal integrators', Proceedings of the National Academy of Sciences of the

United States of America, 105: 2859-64.

Coste, B., J. Mathur, M. Schmidt, T. J. Earley, S. Ranade, M. J. Petrus, A. E. Dubin, and A.

Patapoutian. 2010. 'Piezo1 and Piezo2 are essential components of distinct

mechanically activated cation channels', Science, 330: 55-60.

Fernandez-Velasco, M., A. Rueda, N. Rizzi, J. P. Benitah, B. Colombi, C. Napolitano, S. G. Priori,

S. Richard, and A. M. Gomez. 2009. 'Increased Ca2+ sensitivity of the ryanodine

receptor mutant RyR2R4496C underlies catecholaminergic polymorphic ventricular

tachycardia', Circulation Research, 104: 201-9, 12p following 09.

Gaub, B. M., K. C. Kasuba, E. Mace, T. Strittmatter, P. R. Laskowski, S. A. Geissler, A.

Hierlemann, M. Fussenegger, B. Roska, and D. J. Muller. 2020. 'Neurons differentiate

magnitude and location of mechanical stimuli', Proc Natl Acad Sci U S A, 117: 848-56.

References


Gaub, B. M., and D. J. Muller. 2017. 'Mechanical Stimulation of Piezo1 Receptors Depends on

Extracellular Matrix Proteins and Directionality of Force', Nano Letters, 17: 2064-72.

Hayakawa, K., H. Tatsumi, and M. Sokabe. 2008. 'Actin stress fibers transmit and focus force

to activate mechanosensitive channels', J Cell Sci, 121: 496-503.

Helton, T. D., W. F. Xu, and D. Lipscombe. 2005. 'Neuronal L-type calcium channels open

quickly and are inhibited slowly', Journal of Neuroscience, 25: 10247-51.

Jiang, D. W., P. P. Jones, D. R. Davis, R. Gow, M. S. Green, D. H. Birnie, S. R. W. Chen, and M.

H. Gollob. 2010. 'Characterization of a novel mutation in the cardiac ryanodine

receptor that results in catecholaminergic polymorphic ventricular tachycardia',

Channels, 4: 302-10.

Kim, T., J. Choi, S. Kim, O. Kwon, S. Y. Nah, Y. S. Han, and H. Rhim. 2004. 'The biochemical

activation of T-type Ca2+ channels in HEK293 cells stably expressing alpha1G and

Kir2.1 subunits', Biochem Biophys Res Commun, 324: 401-8.

Kirby, T. J., and J. Lammerding. 2018. 'Emerging views of the nucleus as a cellular

mechanosensor', Nature Cell Biology, 20: 373-81.

Krawczyk, K., S. Xue, P. Buchmann, G. Charpin-El-Hamri, P. Saxena, M. D. Hussherr, J. W. Shao,

H. F. Ye, M. Q. Xie, and M. Fussenegger. 2020. 'Electrogenetic cellular insulin release

for real-time glycemic control in type 1 diabetic mice', Science, 368: 993-+.

Lanner, J. T., D. K. Georgiou, A. D. Joshi, and S. L. Hamilton. 2010. 'Ryanodine Receptors:

Structure, Expression, Molecular Details, and Function in Calcium Release', Cold Spring

Harbor Perspectives in Biology, 2.

Lee, D., J. H. Hyun, K. Jung, P. Hannan, and H. B. Kwon. 2017. 'A calcium- and light-gated switch

to induce gene expression in activated neurons', Nat Biotechnol, 35: 858-63.

Liu, Y., G. Charpin-El Hamri, H. Ye, and M. Fussenegger. 2018. 'A synthetic free fatty acid-

regulated transgene switch in mammalian cells and mice', Nucleic Acids Res, 46: 9864-

74.

Liu, Z., C. S. Gandhi, and D. C. Rees. 2009. 'Structure of a tetrameric MscL in an expanded

intermediate state', Nature, 461: 120-4.

McHugh, B. J., R. Buttery, Y. Lad, S. Banks, C. Haslett, and T. Sethi. 2010. 'Integrin activation

by Fam38A uses a novel mechanism of R-Ras targeting to the endoplasmic reticulum',

J Cell Sci, 123: 51-61.



Nonomura, K., V. Lukacs, D. T. Sweet, L. M. Goddard, A. Kanie, T. Whitwam, S. S. Ranade, T.

Fujimori, M. L. Kahn, and A. Patapoutian. 2018. 'Mechanically activated ion channel

PIEZO1 is required for lymphatic valve formation', Proceedings of the National

Academy of Sciences of the United States of America, 115: 12817-22.

O'Neill, B. K., and S. T. Laughlin. 2018a. 'Neuronal Calcium Recording with an Engineered TEV

Protease', Acs Chemical Biology, 13: 1159-64.

———. 2018b. 'Neuronal Calcium Recording with an Engineered TEV Protease', ACS Chem

Biol, 13: 1159-64.

Pan, B., N. Akyuz, X. P. Liu, Y. Asai, C. Nist-Lund, K. Kurima, B. H. Derfler, B. Gyorgy, W.

Limapichat, S. Walujkar, L. N. Wimalasena, M. Sotomayor, D. P. Corey, and J. R. Holt.

2018. 'TMC1 Forms the Pore of Mechanosensory Transduction Channels in Vertebrate

Inner Ear Hair Cells', Neuron, 99: 736-53 e6.

Salazar-Cantu, A., P. Perez-Trevino, D. Montalvo-Parra, J. Balderas-Villalobos, N. L. Gomez-

Viquez, N. Garcia, and J. Altamirano. 2016. 'Role of SERCA and the sarcoplasmic

reticulum calcium content on calcium waves propagation in rat ventricular myocytes',

Archives of Biochemistry and Biophysics, 604: 11-19.

Santos-Sacchi, J., D. Navaratnam, R. Raphael, and D. Oliver. 2017. 'Prestin: Molecular

Mechanisms Underlying Outer Hair Cell Electromotility', Understanding the Cochlea,

62: 113-45.

Saotome, K., S. E. Murthy, J. M. Kefauver, T. Whitwam, A. Patapoutian, and A. B. Ward. 2018.

'Structure of the mechanically activated ion channel Piezo1', Nature, 554: 481-86.

Sedlmayer, F., D. Hell, M. Muller, D. Auslander, and M. Fussenegger. 2018. 'Designer cells

programming quorum-sensing interference with microbes', Nature Communications,

9.

Shen, B., C. O. Wong, O. C. Lau, T. Woo, S. Bai, Y. Huang, and X. Yao. 2015. 'Plasma membrane

mechanical stress activates TRPC5 channels', PLoS One, 10: e0122227.

Tastanova, A., M. Folcher, M. Muller, G. Camenisch, A. Ponti, T. Horn, M. S. Tikhomirova, and

M. Fussenegger. 2018. 'Synthetic biology-based cellular biomedical tattoo for

detection of hypercalcemia associated with cancer', Sci Transl Med, 10.

Vining, K. H., and D. J. Mooney. 2017. 'Mechanical forces direct stem cell behaviour in

development and regeneration', Nat Rev Mol Cell Biol, 18: 728-42.

References


Wang, H., M. Xie, G. Charpin-El Hamri, H. Ye, and M. Fussenegger. 2018. 'Treatment of chronic

pain by designer cells controlled by spearmint aromatherapy', Nature Biomedical

Engineering, 2: 114-23.

Wang, W., C. P. Wildes, T. Pattarabanjird, M. I. Sanchez, G. F. Glober, G. A. Matthews, K. M.

Tye, and A. Y. Ting. 2017. 'A light- and calcium-gated transcription factor for imaging

and manipulating activated neurons', Nat Biotechnol, 35: 864-71.

Wang, Y., S. Chi, H. Guo, G. Li, L. Wang, Q. Zhao, Y. Rao, L. Zu, W. He, and B. Xiao. 2018. 'A

lever-like transduction pathway for long-distance chemical- and mechano-gating of

the mechanosensitive Piezo1 channel', Nat Commun, 9: 1300.

Xie, M., H. Ye, H. Wang, G. C. E. Hamri, C. Lormeau, P. Saxena, J. Stelling, and M. Fussenegger.

2016. 'beta-cell-mimetic designer cells provide closed-loop glycemic control', Science,

354: 1296-301.

Xu, J., J. Mathur, E. Vessieres, S. Hammack, K. Nonomura, J. Favre, L. Grimaud, M. Petrus, A.

Francisco, J. Y. Li, V. Lee, F. L. Xiang, J. K. Mainquist, S. M. Cahalan, A. P. Orth, J. R.

Walker, S. Ma, V. Lukacs, L. Bordone, M. Bandell, B. Laffitte, Y. Xu, S. Chien, D. Henrion,

and A. Patapoutian. 2018. 'GPR68 Senses Flow and Is Essential for Vascular

Physiology', Cell, 173: 762-+.

Yang, X. P., J. S. Boehm, X. P. Yang, K. Salehi-Ashtiani, T. Hao, Y. Shen, R. Lubonja, S. R. Thomas,

O. Alkan, T. Bhimdi, T. M. Green, C. M. Johannessen, S. J. Silver, C. Nguyen, R. R.

Murray, H. Hieronymus, D. Balcha, C. Y. Fan, C. W. Lin, L. Ghamsari, M. Vidal, W. C.

Hahn, D. E. Hill, and D. E. Root. 2011. 'A public genome-scale lentiviral expression

library of human ORFs', Nature Methods, 8: 659-U80.

Ye, H., M. Daoud-El Baba, R. W. Peng, and M. Fussenegger. 2011. 'A synthetic optogenetic

transcription device enhances blood-glucose homeostasis in mice', Science, 332: 1565-

8.

Zamponi, G. W., J. Striessnig, A. Koschak, and A. C. Dolphin. 2015. 'The Physiology, Pathology,

and Pharmacology of Voltage-Gated Calcium Channels and Their Future Therapeutic

Potential', Pharmacological Reviews, 67: 821-70.

Zhang, W., L. E. Cheng, M. Kittelmann, J. F. Li, M. Petkovic, T. Cheng, P. Jin, Z. H. Guo, M. C.

Gopfert, L. Y. Jan, and Y. N. Jan. 2015. 'Ankyrin Repeats Convey Force to Gate the

NOMPC Mechanotransduction Channel', Cell, 162: 1391-403.



References


Chapter 4: Mechano-Regulation of Wnt/β-catenin Signaling to Control Paraxial Versus Lateral Mesoderm Lineage Bifurcation


Chapter 4: Mechano-Regulation of Wnt/β-catenin Signaling to Control

Paraxial Versus Lateral Mesoderm Lineage Bifurcation



Tobias Strittmatter1, Viktor Haellman1 and Martin Fussenegger1,3 designed the project and

wrote the manuscript. Tobias Strittmatter1 and Viktor Haellman1 contributed equally and

performed experiments, analyzed results and worked on the manuscript. Paul Argast1 and

Peter Buchman1 adapted and constructed the shear stress induction device based on a

previous publication.


Tobias Strittmatter1 and Viktor Haellman1: figures 1-3, S1

Author Affiliations 1Department of Biosystems Science and Engineering, ETH Zurich, Mattenstrasse 26, CH-4058 2Present address: Novartis Pharma AG, CH-4002 Basel, Switzerland

3Faculty of Science, University of Basel, Mattenstrasse 26, CH-4058 Basel, Switzerland

Author Affiliations




Abstract

Mechanical cues form the basis for many phenomena in biology. Mechanosensation enables

us to engage with our surroundings through senses of touch and hearing but also provides a

means for self-awareness (proprioception). Additionally, embryonic development and

patterning are aided by the mechanical properties of the surrounding tissue. In this chapter,

we explore what effects exogenous mechanical stimuli have on differentiation during germ

layer formation, with a specific focus on mesoderm differentiation. We specifically identified

mechanosensitive Wnt-β-catenin signaling to be modulated through shear stress leading to

alterations in lineage specification during ectoderm and mesoderm differentiation. We

further show that this effect is transient and reversible, and can be used to guide mesoderm

differentiation towards cardiac mesoderm in a stage-specific manner.

Abstract




Introduction

Mechanical cues enable adults to probe their environment through the sense of touch

and hearing but also allow for proprioception and control of various other physiological

processes (e.g. blood pressure) (Xu et al. 2018). During embryogenesis, they guide the

fundamental processes of patterning of the embryo (Nonomura et al. 2018) and support

correct differentiation of cells to form functional tissues and organs (Vining et al. 2017).

Establishing the stage-specific role of mechanical cues during differentiation is not only

important for understanding embryonic development, but may also provide clues for more

efficient cell differentiation in vitro towards high-value cell types needed for regenerative

medicine, drug screening and disease modelling.

Wnt-β-catenin signaling is a highly conserved cell-cell communication system

regulating stem cell renewal, proliferation, and differentiation during embryogenesis and in

adult tissues to regulate tissue homeostasis (Logan and Nusse 2004). In embryonic stem cells

(ESCs), Wnt-β-catenin signaling levels are involved in regulating the transition between the

naïve and the primed ESC state, with high levels of Wnt-β-catenin signaling promoting

pluripotency and self-renewal in the naïve state (ten Berge et al. 2011; Xu et al. 2016).

Oppositely, in the primed state, Wnt-β-catenin signaling plays a crucial role in controlling

differentiation (Davidson et al. 2012; Kurek et al. 2015), where the temporal and stage-

specific activation Wnt-β-catenin signaling is involved in specifying the lineage trajectory

(figure 1a). For example, during the initial germ layer specification, activation of Wnt-β-

catenin directs the differentiating cell towards primitive streak specification (mesoderm and

endoderm progenitors state) (Loh et al. 2014; Loh et al. 2016), while no activation or

inactivation directs the cell towards ectoderm specification (Tchieu et al. 2017).

External mechanical forces (shear stress), lateral cell-cell interactions, and mechanical

properties (stiffness) of the extracellular matrix are putative factors that modulate Wnt-β-

catenin signaling in diverse cells and tissues. In endothelial cells and colon cancer cells, shear

stress inhibits Wnt-β-catenin signaling through the induction of Dickkopf Wnt signaling

pathway inhibitor 1 (DKK1) expression, an antagonist of the Wnt-β-catenin pathway (Li et al.

2016; Avvisato et al. 2007). Oppositely, studies in endothelial cells and osteoblasts have

shown that shear stress can also activate Wnt-β-catenin signaling potentially through

increased secretion of Wnt ligands (Cha et al. 2016; Cha et al. 2018) or reduced degradation

of β-catenin through increased phosphorylation of both AKT and glycogen synthase kinase 3β

Introduction


(GSK-3β) (Norvell et al. 2004). In osteoblasts and mesenchymal stem cells, shear stress has

also been shown to increase Wnt-β-catenin signaling through disruption of cadherin/β-

catenin cell-cell interaction complexes, resulting in increased β-catenin nuclear translocation

(Norvell et al. 2004; Arnsdorf et al. 2009). During differentiation of ESCs, the relative

abundance of cadherin/β-catenin complexes, as determined by lateral cell-cell interactions

and the extracellular matrix's mechanical properties, contributes to reinforcing Wnt-β-

catenin signaling during subsequent differentiation (Azarin et al. 2012; Kinney et al. 2013;

Przybyla et al. 2016).

Various techniques exist to induce shear stress on cells utilizing microfluidic chips or

slides that allow for the application of tunable mechanical load on adherent or trapped

suspension cells in narrow channels. These techniques are often used in low-throughput

experiments to study single cells or small cell populations with great precision via atomic force

microscopy (Krieg et al. 2019), patch-clamp (Haswell et al. 2011) or microfluidic devices

(Huber et al. 2018). In this study, human induced pluripotent stem cells (hiPSCs) are exposed

to shear stress, i.e. turbulent flow, generated by a 96-well piston array adapted from a

previously developed device (Xu et al. 2018) (figure 1b). This device consists of a custom-

made 3D-printed array of 96 pistons driven by a membrane speaker controlled by an Arduino

controller coupled to an operational amplifier in series with a commercial hi-fi stereo

amplifier. Previous studies have demonstrated how this device can efficiently induce whole-

population shear stress stimulation on adherent cells grown in 384-well plates to screen for

mechanosensitive membrane receptors (Xu et al. 2018) or to study mechanically triggered

responses of isolated neurons in a 96-well format (Gaub et al. 2020)

When applied to hiPSCs during germ layer specification, we found differentiation of

hiPSC into ectoderm and mesoderm to be differentially affected by mechanical stimulation.

In contrast, stimulation during endoderm specification had no observable effect. To further

characterize the role that mechanical forces play in regulating Wnt-β-catenin signaling during

mesoderm (ME) specification, we generated a stable hiPSC reporter cell line equipped with

an established Tcf-3-responsive promoter (Molenaar et al 1996). By modulating shear stress

induction during ME specification, we show that shear stress can reversibly inhibit Wnt-β-

catenin signaling. Furthermore, we show that ME specification into paraxial or lateral

mesoderm lineages can be altered depending on the day of stimulation during differentiation.



Figure 1: a) iPSCs can differentiate into all three germ layers. While ectoderm derives from

early progenitor cells, endoderm and mesoderm lineages share cells from the primitive

streak as a common ancestor. Cells from the primitive streak can furthermore commit into

an anterior or middle primitive streak compartment with middle primitive streak giving rise

to lateral mesoderm and anterior primitive streak bifurcating into definite endoderm or

paraxial mesoderm. Wnt signaling inversely controls discrimination between paraxial and

lateral mesoderm. Paraxial mesoderm formation is dependent on Wnt signaling whereas

a) lineage committment

iPSC

primitivestreak

anter ior

middle

Wnt

Wnt

g) wnt repression is reversible

1 2 30.0

0.5

1.0

1.5

time [days]

fold

cha

nge

rela

tive

to c

ontro

l

OFF-ON-OFF

ON-OFF-ON

b) mechano-induction device

speaker

3D-printedpiston array

hi-fi amplifier

induced turbulent flow

gene

exp

ress

ion

lineage committment

c) modulation of ectoderm diff.

PAX6SOX1

SOX10

TFAP2A

SIX1

HAND10

1

24

4080

mR

NA le

vels

rela

tive

toun

indu

ced

cont

rol

*** ****

****

***

e) modulation of endoderm diff.

EOMES

SOX17

FOXA2CER

CXCR4GSC

0.0

0.5

1.0

1.5

2.0

mR

NA le

vels

rela

tive

toun

indu

ced

cont

rol

d) modulation of mesoderm diff.

TMIXL1

HAND1

NKX2.5TBX6

MSGN10.0

**** ***

****

0.5

1.0

1.5

2.0

2.5

mR

NA le

vels

rela

tive

toun

indu

ced

cont

rol

f) hiPSC wnt reporter cell line (TopFlash)

10050

010

0020

0040

0060

0080

001.7

1.8

1.9

2.0

0

50

100

150

200

CHIR [ M]

norm

alize

d flu

ores

cenc

e(B

FP/m

Cher

ry) [

RU]

nLuc Luminescence [A

U]

fluorescence NLuc

Introduction


formation of lateral mesoderm requires inhibition of Wnt. b) Schematic of a device for

induction of turbulent flow in cell culture. A 3D-printed array of 96 pistons is attached to the

membrane of a loud speaker that sits on top of a standard 96-well cell culture plate. A sine

signal (10 or 30 Hz) is generated by an Arduino controller in combination with a custom-

made operational amplifier that allows for adjustment of frequencies through a

potentiometer. The resulting signal is fed into a regular hi-fi stereo amplifier that drives the

speaker. The device is built into a chamber with connections for live support. During

operation the pistons move up and down inside each well generating turbulent flow. After

each run supernatant as well as RNA from cells can be harvested and analyzed. c)-e) hiPSCs

were differentiated using a commercial Tirlineage Differentiation Kit in the presence and

absence of turbulent flow and expression levels of selected representative genes were

measured using qPCR. Expression levels of key marker genes for c) ectoderm, d) mesoderm

and e) endoderm were normalized to an unstimulated control. f) A reporter hiPSC line for

activation of the beta-catenin/Wnt pathway was generated using an established Tcf-8-

responsive promoter driving expression of a cytosolic fast Fluorescent Timer (fast FT) and a

secreted NanoLuc luciferase (NLuc) reporter. fastFT matures from an early blue fluorescent

form into a red fluorescent reporter protein. Ratios of blue over red are plotted alongside

levels of secreted NLuc reporter. Activation of Wnt was simulated by titration of GSK-3

inhibitor CHIR99021. g) The reporter cell line characterized in f) was used to assess effects

of turbulent flow produced by the piston array on Wnt signaling over the course of three

days with daily media exchange. Values in b)-d) are means ± SD of N=4 biologically

independent samples. Values in e) are means ± SD of N=6 biologically independent samples

and values in f) are means ± SD of N=8 biologically independent samples. *: p < 0.05, **: p

< 0.01, ***: p < 0.001, ****: p < 0.0001.



Results

To assess the effect of shear stress during differentiation and how it affects lineage

commitment during generation of the three germ layers (ectoderm, endoderm and

mesoderm), hiPSCs were specified towards ectoderm for 5 days, endoderm for 3 days, or

mesoderm for 3 days (figure 1c-e). During differentiation, the cells were either left

undisturbed or cultured with turbulent flow applied by the piston array. Quantifying the

relative expression levels of key lineage markers in day 5 ectoderm cells revealed a distinctly

altered expression pattern suggesting increased differentiation into non-neural ectoderm

cells caused by the turbulent flow (figure 1c). Furthermore, day 3 mesoderm demonstrated

an apparent reduction of lateral mesoderm genes in favor of paraxial mesoderm (figure 1d).

Taken together, this suggests that piston array-induced shear stress might be directly

antagonizing Wnt-β-catenin signaling as both non-neural ectoderm and paraxial mesoderm

specification requires relatively lower levels of Wnt early during differentiation to deselect

alternative ectoderm or mesoderm lineages, respectively. This hypothesis is further

supported by the negligible influence that shear stress has on the expression levels in day 3

endoderm cells (figure 1e) since endoderm specification is comparably less dependent on the

levels of Wnt-β-catenin signaling.

To investigate the role of mechanical stress on Wnt-β-catenin signaling, we generated

a Wnt reporter hiPSC line based on an established Tcf-3-responsive promoter that controls

the expression of the secreted reporter protein NanoLuc luciferase (NLuc) coupled to a

fluorescent-timer (FT) fluorescent reporter protein. FT is expressed as a blue fluorescent

protein first (BFP) but changes fluorescence over time upon maturation into a red fluorescent

protein (RFP). The ratio of BFP over RFP can hence be used to study the dynamics of Wnt-β-

catenin signaling. The functionality of the reporter was confirmed by dose-dependent

inhibition of GSK-3 in stable polyclonal iPSCs by assessing both NLuc reporter activity as well

as BFP:RFP ratios (figure 1f). We further confirmed that mechano-regulation of Wnt-β-catenin

signaling is transient and reversible by applying piston stimulation on different days during

mesoderm differentiation (figure 1g). NLuc reporter activity was lower on days with

stimulation but recovered to levels similar to pre-induction when the device was turned off.

The established reversibility offers the option to apply mechano-regulation to modulate

lineage specification at specific stages during differentiation where Wnt-β-catenin signaling

is involved in regulating lineage bifurcations.

Results


Figure 2: a) Wnt reporter levels were monitored for four days during undisturbed mesoderm

differentiation of hiPSCs. b) Modulation of wnt reporter activity upon stimulation with the

piston device (indicated by arrows on top of days of stimulation). NLuc reporter levels were

quenched reversibly at the day of induction. c) Expression levels of key mesoderm marker

genes were assessed by qPCR during undisturbed mesoderm differentiation using STEMdiff

Mesoderm Induction Media. Samples were taken as indicated every day over the course of

four days and compared to reference samples extracted on day 0. Groups contain genes

associated with (i) pan-mesoderm (brown), (ii) lateral/cardiac mesoderm (green) and (iii)

paraxial mesoderm and early somites (blue). Values in a) and b) represent means ± SD of

(i) primitive streak

rela

tive

expr

essi

on le

vels

(iii) paraxial mesoderm(ii) lateral mesoderm

d1 d2 d3 d405

10152025

NKX2.5

****

****

**

d1 d2 d3 d402468

GATA4***

**

***

d1 d2 d3 d40

1000200030004000

FOXC2

**** ********

****

d1 d2 d3 d40

20406080

PAX3

**

*

*

d1 d2 d3 d40.00.51.01.52.02.5

TCF15

d1 d2 d3 d40

2

4

6MEOX1

** **

d1 d2 d3 d40

20

40

60ISL1

****

**

****

****

d1 d2 d3 d40

1000200030004000

PRRX1

******

****

****

d1 d2 d3 d40

500100015002000

MIXL1

****

****

****

****

d1 d2 d3 d40

50000

100000

150000HAND1

****

****

****

****

d1 d2 d3 d40

5000

10000

15000FOXF1

********

****

****

d1 d2 d3 d40

2000400060008000

T (Brachyury)****

******* ****

d1 d2 d3 d40

50100150200

TBX6

****

********

****

d1 d2 d3 d40

500010000150002000025000

MSGN1

****

************

b) reporter gene expression under shear stress induction

c) marker gene expression during undisturbed mesoderm differentiation

day 1day

2day

3day

4day

1day

2day

3day

4day

1day

2day

3day

4day

1day

2day

3day

4day

1day

2day

3day

40.0

0.51.01.52.0

rela

tive

lum

ines

cenc

e [R

LU] ind day 1

ind day 2

ind day 3 ind all days

ind day 4

a) NLuc during differentiation

day 1

day 2

day 3

day 4

0

200

400

600

800

Lum

ines

cenc

e [R

LU]



N=8 biologically independent samples whereas gene expression data in c) is represented as

means ± SD of N=4 biologically independent samples. *: p < 0.05, **: p < 0.01, ***: p <

0.001, ****: p < 0.0001.

To explore this hypothesis, we focused our efforts on mesoderm specification. We

first followed reporter expression of hiPSCs during mesoderm differentiation over four days

using a general mesoderm induction medium (figure 2a). NLuc reporter levels were increasing

through day 4 and additionally peaking on day 2. This pattern in Wnt-β-catenin signaling

presumably reflects changes in the cell population during the initial transition through the

primitive streak followed by the commitment to either paraxial or lateral mesoderm. In a

follow-up experiment, we assessed how stimulation with the piston device would alter

reporter expression (figure 2b). As seen before, cells undergoing mesoderm differentiation

showed reversible, mechano-induced quenching of NLuc reporter activity during stimulation.

Specifically, we observed quenching of the Wnt-β-catenin reporter activity upon shear stress

treatment on all but one day. The strongest reduction of Wnt-β-catenin signaling was

observed on day 2, which coincides with the highest signaling levels during undisturbed

differentiation (figure 2a). We speculated that the signaling levels are especially sensitive to

mechanical stimuli when at their peak. However, in contrast, the elevated reporter levels on

day 4 did not translate into higher sensitivity to shear stress. Instead, we found Wnt reporter

levels to be upregulated upon induction, suggesting that mechano-regulation of Wnt-β-

catenin signaling depends of the stage of differentiation.

We next sought to compare mesoderm differentiation with or without shear stress

induction. To this end, we first recorded expression patterns of key markers from a four-day

time course of undisturbed mesoderm differentiation of hiPSCs (figure 2c). We found that

iPSCs heterogeneously mature into a mixture of paraxial and lateral mesoderm-derived

lineages as indicated by elevated levels of lateral mesoderm markers FOXF1, HAND1, ISL1,

PRRX1, NKX2.5 and GATA4 and paraxial mesoderm markers TBX6, MSGN1, MEOX1, PAX3, and

FOXC2 on different days during the 4 days protocol. Accordingly, we observed the emergence

of potentially less mature mesoderm populations on day 2 and 3. For example, while markers

for paraxial mesoderm derived ventral (FOXC2) and dorsal (PAX3) somites peak on day 2 and

3, respectively, these populations seem to vanish by day 4. Furthermore, the cells also showed

high expression of a key marker of the lateral mesoderm-derived limb bud (PRRX1) on day 3,

Results


which subsequently declined on day 4. In combination, this pattern is suggesting cardiac

mesoderm represented by marker genes GATA4 and NKX2.5 to be the predominant cell line

by day 4 of undisturbed differentiation with contaminations of early somites from the paraxial

mesoderm lineage and lateral mesoderm limb bud cells.

a) piston-assisted differentiation protocol - 3-day protocol

(i) primitive streak (iii) paraxial mesoderm(ii) lateral mesoderm

rela

tive

expr

essi

on le

vels

ctrd0-1d1-2d2-3d0-30.00.51.01.52.02.5 T

***

ctrd0-1d1-2d2-3d0-30.00.51.01.52.0 MIXL1

**

ctrd0-1d1-2d2-3d0-301234 TBX6

**** ***

**

ctrd0-1d1-2d2-3d0-30.00.51.01.52.0 MSGN1

*** *

ctrd0-1d1-2d2-3d0-30.00.51.01.52.0 HAND1

*

*

**

ctrd0-1d1-2d2-3d0-30.0

0.5

1.0

1.5 FOXF1

**

ctrd0-1d1-2d2-3d0-301234 TCF15

*****

ctrd0-1d1-2d2-3d0-30.00.51.01.52.02.5 MEOX1

ctrd0-1d1-2d2-3d0-30.00.51.01.52.02.5 ISL1

*** **

*

ctrd0-1d1-2d2-3d0-30.00.51.01.52.0 NKX2.5

***

ctrd0-1d1-2d2-3d0-30

1

2

3 GATA4*

*

ctrd0-1d1-2d2-3d0-30.00.51.01.52.0 FOXC2

ctrd0-1d1-2d2-3d0-30.0

0.5

1.0

1.5 PRRX1

*******

*

ctrd0-1d1-2d2-3d0-30.00.51.01.52.02.5 PAX3

*****

b) piston-assisted differentiation protocol - 4-day protocol

(i) primitive streak

rela

tive

expr

essi

on le

vels

(iii) paraxial mesoderm(ii) lateral mesoderm

ctrd0-1d1-2d2-3d3-4d0-402468 PAX3

*

ctrd0-1d1-2d2-3d3-4d0-4012345 PRRX1

******

****

****

ctrd0-1d1-2d2-3d3-4d0-4036

50100150 FOXC2

**

****

ctrd0-1d1-2d2-3d3-4d0-401234 GATA4

***

****

****

ctrd0-1d1-2d2-3d3-4d0-401234 NKX2.5

****

**

***

ctrd0-1d1-2d2-3d3-4d0-4036

2040 ISL1

****

******

ctrd0-1d1-2d2-3d3-4d0-402468

10 MEOX1

**

****

ctrd0-1d1-2d2-3d3-4d0-40.00.51.01.52.0 TCF15

** * *

ctrd0-1d1-2d2-3d3-4d0-40.00.51.01.52.0 HAND1

*** **

**

ctrd0-1d1-2d2-3d3-4d0-40.00.51.01.52.02.5 MSGN1

********

ctrd0-1d1-2d2-3d3-4d0-40

1

2

3 TBX6

** **

* *

****

ctrd0-1d1-2d2-3d3-4d0-40.00.51.01.52.0 MIXL1

****

ctrd0-1d1-2d2-3d3-4d0-40.00.51.01.52.02.5 T

****

***

ctrd0-1d1-2d2-3d3-4d0-40.00.51.01.52.02.5 FOXF1

*** * ******

****



Figure 3: Impact of piston stimulation (30 Hz for 10 sec, 25 % duty cycle) on mesoderm

differentiation using STEMdiff Mesoderm Induction Media was assessed by qPCR during

the a) 3-day or b) 4-day differentiation protocol. All samples were normalized to an

untreated control (ctr). (i) Pan-mesoderm (brown), (ii) lateral, cardiac, and limb bud

mesoderm (green), and (iii) paraxial mesoderm and early somite genes (blue) were

analyzed. Values in represent means ± SD of N = 4 biologically independent samples. *: p <

0.05, **: p < 0.01, ***: p < 0.001, ****: p < 0.0001.

To test the impact on differentiation of piston-stimulation when run at different days,

we performed two independent experiments inducing cells on consecutive days over the

course of 3 (figure 3a) or 4 days (figure 3b). A switched-off array served as negative control

to quantify the relative changes in expression levels of key mesoderm markers. During the 3-

day run, pan-mesoderm markers T and MIXL1 stayed mostly constant with a slight increase

upon treatment on day 3. Lateral mesoderm differentiation seemed to be specifically

inhibited by shear stress induction on day two while being boosted by induction on day 3

(figure 3a, ii). In contrast, paraxial mesoderm commitment benefitted from induction during

day 1 or day 2 (figure 3a, iii). In the 4-day protocol, the specific drop in lateral mesoderm

markers upon induction on day 2 was less pronounced and only affected later genes, which

indicate that shear stress stimulation at this time hampered the progression of cells into the

lateral mesoderm lineage (figure 3b, ii). Oppositely, when shear stress was applied on day 3,

we observed increased expression of cardiac mesoderm markers ISL1 and NKX2.5 and limb

bud marker PRRX1. These results together with the higher levels of Wnt required to promote

mid versus anterior primitive streak and Wnt-β-catenin signaling inhibition required during

lateral mesoderm indicate that the cells are progressing through the primitive streak during

day 2 and subsequently specify towards lateral mesoderm on day 3. Hence, by transiently

applying shear stress on day 3 we can actively promote a more efficient transition towards

lateral mesoderm using the piston array. Shear stress induction on day 4 is similarly able to

promote upregulation of cardiac mesoderm markers ISL1 and NKX2.5 (figure 3b, ii).

Concurrent upregulation of GATA4 suggests that shear stress-mediated quenching of Wnt-β-

catenin signaling on day 4 might further aid in cardiac mesoderm differentiation (figure 3b,

ii). Activation or inhibition of Wnt-β-catenin signaling following lateral mesoderm

Results


differentiation discriminates between subsequent specification into the limb bud or cardiac

mesoderm lineage, respectively. In agreement, we observed relatively lower upregulation of

limb bud marker PRRX1 in response to shear stress applied on day 4 compared to when shear

stress was applied on day 3 (figure 3b, ii).

To study effects of shear stress on genes directly associated with or know to influence

Wnt-β-catenin signaling, we assessed the relative expression levels of key genes involved in

cell adherence (figure S1a), the canonical Wnt pathway (figure S1b) as well as common

inhibitors (figure S1c) and activators (figure S1d) of Wnt-β-catenin. Intriguingly, we found that

continued shear stress induction over 2 days during mesoderm differentiation did not affect

the expression of any of the genes neither on the first nor the second day. This seems to

indicate that the observed effect on lineage commitment does not involve significant

alterations of the expression levels of the signaling components themselves but is purely

restricted to effects mediated by the existing signaling setup on a protein level.

Taken together, our data suggest that shear stress can be utilized to modulate lineage

commitment during ectoderm and mesoderm differentiation through alterations of Wnt

signaling. Furthermore, initial characterization of key genes directly associated with or known

to influence Wnt-β-catenin signaling suggest that the effect of shear stress is due to

phenotypic changes on a post-translational level rather than a transcriptional or post-

transcription level. However, full clarification of the exact mechanism will require additional

characterization.



Discussion

We here present a new way of modulating lineage commitment of iPSCs in cell culture

based on induction of turbulent flow generated by an oscillating piston array. The piston array

used in this study provides a traceless, robust, reversible and easy to implement induction

method. The method was validated using an established Wnt-responsive promoter construct

as well as gene expression analysis of key marker genes during the differentiation of all three

lineages with a specific focus on mesoderm differentiation. Quenching of reporter activity

exclusively on days of piston induction and selection against Wnt-dependent cell types

support the idea that the observed modulation is based on inhibition of Wnt signaling. This

effect was confirmed by alterations in lineage specification during mesoderm differentiation

exemplified by enhanced expression of key markers of cardiac mesoderm upon stimulation

on days which are reported to depend on Wnt inhibition.

Although the general time scale of lineage commitment seen in this setup diverges

slightly from established protocols, the sequence of the stages remains the same and likewise

the resulting lineages. However, based on our initial analysis of gene expression of markers

involved in Wnt signaling and cell adherence, we were unable to determine the underlying

mechanism causing the quenching of Wnt-β-catenin signaling nor can we conclusively exclude

supplementary effects on other pathways. Both issues should be characterized in-depth in

future studies. Furthermore, subsequent studies should also explore how shear stress

synergizes with modern differentiation protocols where highly optimized differentiation

media precisely direct each differentiation stages.

The arduino-based, flexible and easy to program controls in combination with tunable

oscillation frequencies and intensities allows the generation of basically any induction profile

of interest. The downside of this flexibility is the plethora of possible stimulation patterns

which prohibits efficient screening for optimal induction conditions. Hence, our data does

most likely not cover the optimal conditions and alternative modulation patterns might

further enhance the observed effect. Future work should therefore aim to establish a more

comprehensive parameter landscape and test induction on combinations of several days for

potential beneficial effects on lineage commitment. We already provide preliminary evidence

that differentiation into lateral mesoderm is enhanced by stimulation during mid primitive

streak exit on day 3 or during the limb bud versus cardiac mesoderm bifurcation stage on day

4. Therefore, a good starting point for further studies might be to test whether continued

Discussion


stimulation from day 3 into day 4 can enrich for cardiac mesoderm or conversely, if

continuous stimulation on days 1 and 2 can enrich for paraxial mesoderm and somite

generation. Furthermore, since the level of Wnt signaling activation and inhibition is highly

stage dependent throughout human development future studies could also explore the

applicability of mechano-regulation to other differentiation stages and lineages.

The here described piston-based method is easy to implement for studying the effects

of shear stress on lineage commitment during differentiation of iPSCs into various lineages.

We hope that this technique will contribute to a broader understanding of the role of

mechanical cues in stem cell development as well as potentially facilitate the production of

high-quality cell lines for use in research as well as personalized medicine in the future.



Materials and Methods

Plasmid preparation

Plasmids were generated by digestion with standard restriction enzymes (New

England BioLabs), followed by dephosphorylation of the plasmid backbones with Antarctic

phosphatase (New England BioLabs, Cat. no. M0289), and then ligation of the backbones and

inserts with T4 DNA ligase (Thermo Scientific, Cat. no. EL0011). Phosphorylation of single

oligonucleotides was performed using T4 polynucleotide kinase (New England BioLabs, Cat.

no. M0201), according to the manufacturer’s instructions. Phosphorylation and annealing of

two oligonucleotides were performed using T4 polynucleotide kinase (New England BioLabs,

Cat. no. M0201) in T4 DNA ligase buffer (Thermo Scientific, Cat. no. B69). PCRs were

performed with Phusion high-fidelity DNA polymerase (Thermo Scientific, Cat. no. F530),

according to the manufacturer’s instructions. All plasmids used in this study are described in

Supplementary Table 1.

Cell culture

In house generated human induced pluripotent stem cells (hiPSCs) (Heng et al. 2013),

and derived stable lines, were propagated feeder-free in mTeSR1 medium (StemCell

Technologies, Cat. no. 85850) supplemented with NormocinTM (50 µg/mL, Invivogen, Cat.

no. ant-nr-1) on cell culture plates coated with Geltrex basement membrane matrix (Gibco,

Cat. no. A1413302). HiPSC lines were routinely passaged as aggregates using ReLeSR

(StemCell Technologies, Cat. no. 05872). All cells were maintained at 37°C in a humidified

atmosphere of 5% CO2 in air and routinely tested for mycoplasma contamination at an

accredited external sequencing facility (GATC Biotech AG).

To generate piggyBac transposon-mediated stable hiPSC lines, cells were seeded as

single cells and then transiently transfected by using Lipofectamine Stem (Invitrogen, Cat. no.

STEM00003) transfection reagent. In brief, 80% confluent hiPSCs were washed with PBS

(without calcium and magnesium, Gibco, Cat. no. 14190169) and dissociated to single cells

using StemPro Accutase (Gibco, Cat. no. A1110501) at 37°C for 5-10 min, or until the cells

were completely dissociated. Cells were harvested by using 4 volumes of DMEM/F12 (Gibco,

Cat. no. 31331028) and gently triturated to further dissociate them. The cell suspension was

centrifuged at 150 x g for 5 min, then the supernatant was aspirated, and the cells were gently



resuspended in mTeSR1 supplemented with rho kinase inhibitor (Y-27632; 100 µM, StemCell

Technologies, Cat. no. 72304). Cells were counted and then seeded on Geltrex-coated multi-

well plates. After 20-24 h, the spent medium was aspirated. The cells were washed with PBS

and fresh mTeSR1 was added. Cells were then transiently transfected by mixing the required

amount of DNA with Opti-MEM reduced serum media (Gibco, Cat. no. 31985070).

Lipofectamine Stem was added to the solution and mixed thoroughly. The transfection

mixture was incubated for 10-12 min at room temperature and then added dropwise to the

cells. For optimal transfection results, cells were transfected within 30 min after removing the

rho kinase inhibitor. At 24 h after transfection, the medium was replaced with fresh mTeSR1.

At 48 h after transfection, selection medium was added, consisting of mTeSR1 supplemented

with puromycin (0.5 µg/mL, Invivogen, Cat. no. ant-pr-1) for 5 days before increasing selection

2x-times (1.0 µL puromycin) and continuing selection of an additional 7 days. Then, the cells

were assessed for transgene expression and differentiation potential.

Directed differentiation of hiPSCs

To assess mechano-regulation of Wnt/β-catenin signaling during differentiation into

different lineages, we used the STEMdiff Trilineage Differentiation Kit (StemCell Technologies,

Cat. no. 05230) to generate the three germ layers (mesoderm, endoderm and ectoderm). In

brief, hiPSCs grown to 80% confluency were dissociated and seeded as single cells in mTeSR1

supplemented with Y-27632 (100 µM) at a density of 5.0 x 104 cells/cm2 for mesoderm

differentiation or 1.0 x 105 cells/cm2 for endoderm and ectoderm differentiation on 50

µg/cm2 Geltrex-coated multi-well pates. At 18 h after seeding, spent medium was replaced

with fresh mTeSR1 for 6 h to recover cells. Then cells were washed with PBS and lineage

specification was induced as follows: mesoderm differentiation was specified by using

STEMDiff Trilineage Mesoderm Medium for 3 days with daily medium changes; endoderm

differentiation was specified by using STEMDiff Trilineage Endoderm Medium for 3 days with

daily medium changes; ectoderm differentiation was specified by using STEMDiff Trilineage

Ectoderm Medium for 5 days with daily medium changes.

For further characterization of mechano-regulation of Wnt/β-catenin signaling during

mesoderm differentiation, hiPSCs were seeded as described above and differentiated using

STEMdiff Mesoderm Induction Media (StemCell Technologies, Cat. no. 05221) for 3 to 4 days

with daily media medium changes.



Mechano-regulation of Wnt/β-catenin signaling

A schematic of the shear stress induction setup can be found in the main text (figure

1b) and is derived from a previously described device used to screen for mechanosensitive

genes in 384-well format (J. Xu et al. 2018). The setup consists of a device comprising a

custom-made 3D-printed array of 96 pistons that are attached to a membrane speaker

controlled by an Arduino controller that is linked to an operational amplifier. The Arduino can

be programmed to generate a pulses and intervals of a voltage signal that is converted into a

sine signal of 10-60 Hz by the operational amplifier circuit. The resulting interval-modulated

sine signal is fed into a commercial hi-fi stereo amplifier that drives the speaker. This setup

allows for the generation of fully customizable stimulation patterns by allowing for selection

of specific frequencies, intensities and intervals.

Experiments were done in standard cell culture incubators in humidified atmosphere

containing 5% CO2. Pistons were regenerated prior to use treatment with 1 M NaOH for at

least 10 min. Pistons were rinsed with ddH2O and dried in a cell culture flow-hood with 30

min of UV-sterilization. Devices were pre-warmed for a couple of hours in the cell culture

incubator to allow condensation to evaporate.

Initial experiments using the Trilineage Differentiation Kit (figure 1c-e) were done

using 10 Hz for 5 seconds every 5 minutes. After evaluation of cell viability settings were

adjusted to 30 Hz for 10 seconds every 30 seconds for all subsequent experiments.

Analytical assays

NLuc production was measured in cell culture supernatants with the Nano-Glo

Luciferase Assay System (Promega, Cat. no. N1110). In brief, 7.5 µL of each sample of secreted

Nluc was incubated with 7.5 µL Nano-Glo buffer/substrate mixture (50:1) for 5 min in black

384-well plates. Luminescence was measured with a Tecan Infinite M1000 PRO multiplate

reader (Tecan AG).

Gene expression profiling by RT-qPCR

A Quick-RNA mini prep kit (Zymo Research, Cat. no. R1054) was used for RNA isolation,

and 250 ng of purified RNA was used as a template for cDNA synthesis (High-Capacity cDNA



Reverse Transcription Kit, Applied Biosystems, Cat. no. 4368814). Gene expression was

quantified using gene-specific primers and SsoAdvanced Universal SYBR Green Supermix (Bio-

Rad, Cat. no. 1725271) with an Eppendorf ep realplex Mastercycler (Eppendorf AG). Results

are presented as relative fold changes normalized to the expression of one or two

endogenous reference genes by using the DDCt method (Schmittgen & Livak, 2008). All qPCR

primers used in this study are listed in Supplementary Table 2.

Statistical analyses

All data represent the means ± SD of four or eight biologically independent samples

per condition. Statistical significance was assessed by unpaired t tests employing the two-

stage setup procedure of Benjamini, Krieger and Yekutieli using GraphPad Prism 9 software.

Significance is indicated as *: p < 0.05, **: p < 0.01, ***: p < 0.001, ****: p < 0.0001.





Plasmid Details Reference pCMV-hyPBase

Constitutive piggyBac expression vector (PhCMV-hyPBase-pA).

Sanger Institute, UK

pMM531

Vector encoding PhCMV-driven fast-type fluorescent timer (fastFT) (PhCMV-fastFT-pA)

(Chassin et al. 2019)

pTS1010 Vector encoding PhCMV-driven secreted Nluc expression vector (PhCMV-SP-Nluc-pA).

(Bojar, Fuhrer, & Fussenegger, 2019)

pTS1018 Cloning vector encoding a PhCMV-driven P2A sequence (PhCMV-P2A-pA). P2A was assembled by phosphorylating and annealing oligos OTS894 (5’ ctagTGGTGGTTC-TGGTGGAAGCGGAGCTACTAACTTCAGCCTGCTGAAGCAGGCTGGtGACGTGGAGGAGAACCCTGGACCTGCTAGCg 3’) and OTS895 (5’ gatcCGCTAGCAGGTCCAGGGTTCTCCTCCACGTCaCCAGCCTGCTTCAGCAGGCTGAAGTTAGTAGCTCCGCTTCCACCAGAACCACCa 3’) The annealed oligos were ligated into pTS1010 (SpeI/BamHI).

(Haellman et al. 2020) unpublished

pTS1019 piggyBac transposon vector with three insertion sites (A1, A2, A3) for stable transgene integration (5’ITR-A1-pA-A2-pA-A3-pA-3’ITR).


pTS1024 Vector encoding PRPBSA-driven yellow fluorescent protein (YPet) coupled to a Puromycine resistance gene PuroR via a P2A sequence (PRPBSA-YPet-P2A-PuroR-pA)


pTS1052 Vector encoding PPGK-driven secreted NLuc reporter coupled to a gene blue fluorescent protein (mTagBFP) via a P2A sequence (PPGK-SP-NLuc-P2A-mTagBFP-pA)


pTS1053 Cloning vector encoding a fusion protein of P2A and a fast-type fluorescent timer (fastFT) (PhCMV-P2A-fastFT-pA). fastFT was excised from pMM531 (SpeI/BamHI) and ligated into pTS1018 (NheI/BamHI).

This work

pTS2106 Reporter vector encoding PTCF8-driven secreted Nluc expression coupled to a fast-type fluorescent timer (fastFT) fluorophore via P2A sequence (OTCF8-Pmin-SP-NLuc-P2A-fastFT-pA). P2A-fastFT was excised from pTS1053 (SpeI/BamHI) and ligated into pVH487 (NheI/BamHI).

This work

pTS2107 piggyBac transposon vector encoding a PTCF8-driven SP-Nluc-2A-fastFT and a constitutive YPet-2A-PuroR expression cassette (5’ITR-OTCF8-Pmin-SP-NLuc-P2A-fastFT-pA-PmPGK1-YPet-2A-PuroR-pA-3’ITR). OTCF8-Pmin-SP-NLuc-P2A-fastFT was excised from pTS2106 (MluI/HindIII) and ligated into the corresponding acceptor site A1 (MluI/HindIII) of pTS1019. PmPGK1 was excised from pTS1052 (MluI/EcoRI) and YPet-2A-PuroR was excised from pTS1024

This work



(EcoRI/HindIII) and both fragments were jointly ligated into the corresponding acceptor site A3 (BsmBI) of pTS1019

pVH444 Reporter vector encoding Pmin-driven secreted Nluc expression

(Pmin-SP-NLuc-pA). Pmin was assembled by phosphorylating and annealing oligos OVH798 (5’ tcgaGTAGAGGGTATATAATGGAAGCTCGACTTCCAGGAGCTCTTCG- AAGCGG 3’) and OVH799 (5’ aattCCGCTTCGAAGAGCTCCTGGAAGTC- GAGCTTCCATTATATACCCTCTAC 3’). The annealed oligos were ligated into pTS1010 (XhoI/EcoRI).

This work

pVH487 Reporter vector encoding PTCF8-driven secreted Nluc expression (OTCF8-Pmin-SP-NLuc-pA). OTCF8 was assembled by phosphorylating and annealing oligos OVH862 (5’ cgcgTAGATCAAAGGGTAGCGTGGTAAGATCAAAGGGTAGCGTGGT- AAGATCAAAGGGTAGCGTGGTAAGATCAAAGGGTAGgtac 3’) and OVH863 (5’ CTACCCTTTGATCTTACCACGCTACCCTTTGATCTTACCAC- GCTACCCTTTGATCTTACCACGCTACCCTTTGATCTA 3’), and OVH864 (5’ CATAGATCAAAGGGTAGCGTGGTAAGATCAAAGGGTAGCGTGGTAAG- ATCAAAGGGTAGCGTGGTAAGATCAAAGGGGGC 3’) and OVH865 (5’ tcgaGCCCCCTTTGATCTTACCACGCTACCCTTTGATCTTACCACGCTAC- CCTTTGATCTTACCACGCTACCCTTTGATCTATGgtac 3’). The annealed oligos were ligated into pVH444 (MluI/XhoI).

This work

Oligonucleotides: Annealing base pairs contained in oligonucleotide sequences are shown in capital letters. Restriction endonuclease-specific sites and sticky-ends generated through oligonucleotide annealing are underlined. Abbreviations and additional information: P2A, self-cleaving peptide; 5’/3’ITR, 5’/3’inverted terminal repeat; hyPBase, hyperactive piggyBac transposase; Nluc, NanoLuc luciferase engineered from Oplophorus gracilirostris, OTCF8, T cell factor (TCF)-specific operator with 8 response elements; pA, polyadenylation signal; PhCMV, strong constitutive CMV promoter PCMVmin-1, strong hCMV minimal promoter; Pmin, pGL4.23-derived minimal promoter; PmPGK1, mouse phosphoglycerate kinase 1 promoter; PRPBSA, strong constitutive synthetic RPBSA promoter PuroR, puromycin resistance gene puromycin N-acetyl-transferase; SP, secretion signal peptide derived from IgG kappa light chain leader sequence; YPet, improved version of Venus derived from Aequorea victoria.



Table S2: List of primers for qRT-PCR

Gene Forward Primer Reverse Primer TBP TGTATCCACAGTGAATCTTGGTTG GGTTCGTGGCTCTCTTATCCTC GAPDH GTCTCCTCTGACTTCAACAGCG ACCACCCTGTTGCTGTAGCCAA GATA4 TCCCTCTTCCCTCCTCAAATT CAGCGTGTAAAGGCATCTG CER1 CAGGACAGTGCCCTTCAGCCA ACAGTGAGAGCAGGAGGTATGG CXCR4 CTCCTCTTTGTCATCACGCTTCC GGATGAGGACACTGCTGTAGAG EOMES AAATGGGTGACCTGTGGCAAAGC CTCCTGTCTCATCCAGTGGGAA FOXA2 GGAACACCACTACGCCTTCAAC AGTGCATCACCTGTTCGTAGGC FOXC2 CCTCCTGGTATCTCAACCACA GAGGGTCGAGTTCTCAATCCC FOXF1 AGCAGCCGTATCTGCACCAGAA CTCCTTTCGGTCACACATGCTG GSC GCACCATCTTCACTGACGAGCA TTTGGCGCGGCGGTTCTTAAAC HAND1 CAAGGATGCACAGTCTGGCGAT GCAGGAGGAAAACCTTCGTGCT ISL1 GCAGAGTGACATAGATCAGCCTG GCCTCAATAGGACTGGCTACCA MEOX1 GAGATTGCGGTAAACCTGGACC TCTGAACTTGGAGAGGCTGTGG MIXL1 GGAAGGATTTCCCACTCTGACG CCCGACATCCACTTGCGCGAG MSGN1 AGAGGGAGAAGCTCAGGATGAG GTGTCTGGATCTTGGTGAGAGG NKX2.5 AAGTGTGCGTCTGCCTTTCCCG TTGTCCGCCTCTGTCTTCTCCA PAX3 CTCCACGCTCCGGATAGTTC ATCTTGTGGCGGATGTGGTT PAX6 CTGAGGAATCAGAGAAGACAGGC ATGGAGCCAGATGTGAAGGAGG PRRX1 TGATGCTTTTGTGCGAGAAGA AGGGAAGCGTTTTTATTGGCT SIX1 AGGTCAGCAACTGGTTTAAGAACC GAGGAGAGAGTTGGTTCTGCTTG SOX1 GAGTGGAAGGTCATGTCCGAGG CCTTCTTGAGCAGCGTCTTGGT SOX10 ATGAACGCCTTCATGGTGTGGG CGCTTGTCACTTTCGTTCAGCAG SOX17 ACGCTTTCATGGTGTGGGCTAAG GTCAGCGCCTTCCACGACTTG T (Brachyury) CCTTCAGCAAAGTCAAGCTCACC TGAACTGGGTCTCAGGGAAGCA

TBX6 TCATCTCCGTGACAGCCTACCA CCGCAGTTTCCTCTTCACACGG TCF15 GCGGGCAGTGCCAAGGGCG CCCTCACCTTCAAGCAGCTGC TFAP2A GACCTCTCGATCCACTCCTTAC GAGACGGCATTGCTGTTGGACT




Supplementary Figure 1: Perturbation of general factors of wnt signaling by piston

induction-mediated shear stress treatment over 2 days was assessed by qPCR. Genes

involved in a) cell adherence and b) beta-catenin signaling as well as c) genes encoding

a) cadherins and cadherin - interacting protein p120

day 1

day 2

0.00.51.01.52.0

E-cadherin

day 1

day 2

05

101520

N-cadherin

day 1

day 2

0.00.51.01.52.02.5

p120-catenin

rela

tive

expr

essi

on le

vels

b) beta-catenin - interacting proteins

day 1

day 2

0.00.51.01.52.0

beta-catenin

day 1

day 2

0500

100015002000

LEF1

day 1

day 2

0.00.51.01.5

CTNNBIP1

rela

tive

expr

essi

on le

vels

c) inhibitors of wnt signaling

day 1

day 2

0100020003000

DKK1

day 1

day 2

01000020000300004000050000

DKK4

rela

tive

expr

essi

on le

vels

d) wnt family

day 1

day 2

05

10152025

Wnt3

day 1

day 2

0100200300400

Wnt5b

day 1

day 2

05

101520

Wnt9b

day 1

day 2

050

100150200250

Wnt5a

day 1

day 2

0102030

Wnt3a

day 1

day 2

01000200030004000

Wnt6

day 1

day 2

0.00.51.01.52.02.5

Wnt4

day 1

day 2

0500

10001500

Wnt8a

rela

tive

expr

essi

on le

vels

uninducedshear stress






known wnt inhibitors and c) genes from the wnt family itself were analyzed after 1 or 2 days

as indicated. Samples were normalized to an untreated reference before induction (d0).

Data is depicted as means ± SD of N=4 biologically independent samples.





References

Arnsdorf, E. J., P. Tummala, and C. R. Jacobs. 2009. 'Non-canonical Wnt signaling and N-

cadherin related beta-catenin signaling play a role in mechanically induced osteogenic

cell fate', PLoS One, 4: e5388.

Avvisato, C. L., X. Yang, S. Shah, B. Hoxter, W. Li, R. Gaynor, R. Pestell, A. Tozeren, and S. W.

Byers. 2007. 'Mechanical force modulates global gene expression and beta-catenin

signaling in colon cancer cells', J Cell Sci, 120: 2672-82.

Azarin, S. M., X. Lian, E. A. Larson, H. M. Popelka, J. J. de Pablo, and S. P. Palecek. 2012.

'Modulation of Wnt/beta-catenin signaling in human embryonic stem cells using a 3-

D microwell array', Biomaterials, 33: 2041-9.

Bojar, D., T. Fuhrer, and M. Fussenegger. 2019. 'Purity by design: Reducing impurities in

bioproduction by stimulus-controlled global translational downregulation of non-

product proteins', Metab Eng, 52: 110-23.

Cha, B., X. Geng, M. R. Mahamud, J. Fu, A. Mukherjee, Y. Kim, E. H. Jho, T. H. Kim, M. L. Kahn,

L. Xia, J. B. Dixon, H. Chen, and R. S. Srinivasan. 2016. 'Mechanotransduction activates

canonical Wnt/beta-catenin signaling to promote lymphatic vascular patterning and

the development of lymphatic and lymphovenous valves', Genes Dev, 30: 1454-69.

Cha, B., X. Geng, M. R. Mahamud, J. Y. Zhang, L. Chen, W. Kim, E. H. Jho, Y. Kim, D. Choi, J. B.

Dixon, H. Chen, Y. K. Hong, L. Olson, T. H. Kim, B. J. Merrill, M. J. Davis, and R. S.

Srinivasan. 2018. 'Complementary Wnt Sources Regulate Lymphatic Vascular

Development via PROX1-Dependent Wnt/beta-Catenin Signaling', Cell Reports, 25:

571-84 e5.

Davidson, K. C., A. M. Adams, J. M. Goodson, C. E. McDonald, J. C. Potter, J. D. Berndt, T. L.

Biechele, R. J. Taylor, and R. T. Moon. 2012. 'Wnt/beta-catenin signaling promotes

differentiation, not self-renewal, of human embryonic stem cells and is repressed by

Oct4', Proc Natl Acad Sci U S A, 109: 4485-90.

Gaub, B. M., K. C. Kasuba, E. Mace, T. Strittmatter, P. R. Laskowski, S. A. Geissler, A.

Hierlemann, M. Fussenegger, B. Roska, and D. J. Muller. 2020. 'Neurons differentiate

magnitude and location of mechanical stimuli', Proc Natl Acad Sci U S A, 117: 848-56.

Haswell, E. S., R. Phillips, and D. C. Rees. 2011. 'Mechanosensitive Channels: What Can They

Do and How Do They Do It?', Structure, 19: 1356-69.

References


Heng, B. C., K. Heinimann, P. Miny, G. Iezzi, K. Glatz, A. Scherberich, H. Zulewski, and M.

Fussenegger. 2013. 'mRNA transfection-based, feeder-free, induced pluripotent stem

cells derived from adipose tissue of a 50-year-old patient', Metab Eng, 18: 9-24.

Huber, D., A. Oskooei, I. Solvas X. Casadevall, deMello Andrew, and G. V. Kaigala. 2018.

'Hydrodynamics in Cell Studies', Chem Rev, 118: 2042-79.

Kinney, M. A., C. Y. Sargent, and T. C. McDevitt. 2013. 'Temporal modulation of beta-catenin

signaling by multicellular aggregation kinetics impacts embryonic stem cell

cardiomyogenesis', Stem Cells Dev, 22: 2665-77.

Krieg, M., G. Flaschner, D. Alsteens, B. M. Gaub, W. H. Roos, G. J. L. Wuite, H. E. Gaub, C.

Gerber, Y. F. Dufrene, and D. J. Muller. 2019. 'Atomic force microscopy-based

mechanobiology', Nature Reviews Physics, 1: 41-57.

Kurek, D., A. Neagu, M. Tastemel, N. Tuysuz, J. Lehmann, H. J. G. van de Werken, S. Philipsen,

R. van der Linden, A. Maas, IJcken W. F. J. van, M. Drukker, and D. Ten Berge. 2015.

'Endogenous WNT signals mediate BMP-induced and spontaneous differentiation of

epiblast stem cells and human embryonic stem cells', Stem Cell Reports, 4: 114-28.

Li, M., X. Liu, Y. Zhang, M. Di, H. Wang, L. Wang, Y. Chen, X. Liu, X. Cao, R. Zeng, Y. Zhang, and

M. Zhang. 2016. 'Upregulation of Dickkopf1 by oscillatory shear stress accelerates

atherogenesis', J Mol Med (Berl), 94: 431-41.

Logan, C. Y., and R. Nusse. 2004. 'The Wnt signaling pathway in development and disease',

Annu Rev Cell Dev Biol, 20: 781-810.

Loh, K. M., L. T. Ang, J. Zhang, V. Kumar, J. Ang, J. Q. Auyeong, K. L. Lee, S. H. Choo, C. Y. Lim,

M. Nichane, J. Tan, M. S. Noghabi, L. Azzola, E. S. Ng, J. Durruthy-Durruthy, V.

Sebastiano, L. Poellinger, A. G. Elefanty, E. G. Stanley, Q. Chen, S. Prabhakar, I. L.

Weissman, and B. Lim. 2014. 'Efficient endoderm induction from human pluripotent

stem cells by logically directing signals controlling lineage bifurcations', Cell Stem Cell,

14: 237-52.

Loh, K. M., A. Chen, P. W. Koh, T. Z. Deng, R. Sinha, J. M. Tsai, A. A. Barkal, K. Y. Shen, R. Jain,

R. M. Morganti, N. Shyh-Chang, N. B. Fernhoff, B. M. George, G. Wernig, R. E. A.

Salomon, Z. Chen, H. Vogel, J. A. Epstein, A. Kundaje, W. S. Talbot, P. A. Beachy, L. T.

Ang, and I. L. Weissman. 2016. 'Mapping the Pairwise Choices Leading from

Pluripotency to Human Bone, Heart, and Other Mesoderm Cell Types', Cell, 166: 451-

67.



Molenaar, M., M. van de Wetering, M. Oosterwegel, J. Peterson-Maduro, S. Godsave, V.

Korinek, J. Roose, O. Destree, and H. Clevers. 1996. 'XTcf-3 transcription factor

mediates beta-catenin-induced axis formation in Xenopus embryos', Cell, 86: 391-9.





Norvell, S. M., M. Alvarez, J. P. Bidwell, and F. M. Pavalko. 2004. 'Fluid shear stress induces

beta-catenin signaling in osteoblasts', Calcif Tissue Int, 75: 396-404.

Przybyla, L., J. N. Lakins, and V. M. Weaver. 2016. 'Tissue Mechanics Orchestrate Wnt-

Dependent Human Embryonic Stem Cell Differentiation', Cell Stem Cell, 19: 462-75.

Schmittgen, T. D., and K. J. Livak. 2008. 'Analyzing real-time PCR data by the comparative C(T)

method', Nat Protoc, 3: 1101-8.

Tchieu, J., B. Zimmer, F. Fattahi, S. Amin, N. Zeltner, S. Chen, and L. Studer. 2017. 'A Modular

Platform for Differentiation of Human PSCs into All Major Ectodermal Lineages', Cell

Stem Cell, 21: 399-410 e7.

ten Berge, D., D. Kurek, T. Blauwkamp, W. Koole, A. Maas, E. Eroglu, R. K. Siu, and R. Nusse.

2011. 'Embryonic stem cells require Wnt proteins to prevent differentiation to

epiblast stem cells', Nature Cell Biology, 13: 1070-5.

Vining, K. H., and D. J. Mooney. 2017. 'Mechanical forces direct stem cell behaviour in

development and regeneration', Nat Rev Mol Cell Biol, 18: 728-42.

Xu, J., J. Mathur, E. Vessieres, S. Hammack, K. Nonomura, J. Favre, L. Grimaud, M. Petrus, A.

Francisco, J. Y. Li, V. Lee, F. L. Xiang, J. K. Mainquist, S. M. Cahalan, A. P. Orth, J. R.

Walker, S. Ma, V. Lukacs, L. Bordone, M. Bandell, B. Laffitte, Y. Xu, S. Chien, D. Henrion,

and A. Patapoutian. 2018. 'GPR68 Senses Flow and Is Essential for Vascular

Physiology', Cell, 173: 762-+.

Xu, Z., A. M. Robitaille, J. D. Berndt, K. C. Davidson, K. A. Fischer, J. Mathieu, J. C. Potter, H.

Ruohola-Baker, and R. T. Moon. 2016. 'Wnt/beta-catenin signaling promotes self-

renewal and inhibits the primed state transition in naive human embryonic stem cells',

Proc Natl Acad Sci U S A, 113: E6382-E90.

References


Discussion


Discussion

This work covers different ways to control cellular behavior via classical receptor-

mediated signal transduction pathways, metabolic activation of a small molecule as well as

physical stimuli in the form of mechanical load and shear stress. Each approach contributes

to resolving prevalent issues of biological research on cell signaling and differentiation. The

following sections provide a short summary of each chapter and a discussion of the broader

implications of the results.

New receptors to expand the range of targetable disease

In the first chapter of this work, we presented a pipeline for the generation of novel

receptors based on the established EpoR scaffold. We used the pipeline to generate new

receptors to detect degradation products of fibrin from human blood following thrombotic

events. The receptors function as homodimers, heterodimers, or with tandem-binders fused

to the extracellular domains. Cells equipped with these receptors could also be modified to

secrete fibrinolytic as well as anti-coagulative therapeutics. This approach is not limited to a

single type of protein binder but can, in theory, be used in combination with any binder

selection platform of choice which expands the range of potential receptor ligands even

further. The presented work-flow does not only validate the initial GEMS platform design but

also increases reliability of receptor design in general. Although the presented data only

covers the generation of receptors against a single albeit highly relevant target, we are

confident that the results allow for an extrapolation on other targets as well. With the current

improvements, it is now possible to more reliably and predictably generate receptors against

a soluble target of choice.

A combination of binders derived from different selection strategies would increase

the amount and variation of available binders with different epitope preferences and

requirements for protein binding. In turn, an increasing number of binders also produces a

higher number of receptors and receptor combinations that need to be evaluated in a cell

culture screen. A more stringent selection for binders might therefore be considered to

reduce the size of the receptor library. However, such an approach would also dramatically

shrink the available epitope space. In addition, we did not find a correlation of binding affinity

and the success rate of a binder in a receptor context. Hence, affinities of selected binders

New receptors to expand the range of targetable disease


seem to be less important than the epitope space that they cover. One way to “rescue” non-

dominant binders that attach to less-affine epitopes from a selection run is given by blocking

the dominant epitopes during selection (Tsui et al. 2002). This method could be used to

increase the amount of lower-affinity binders by adding the best binders of a first run to a

second round of selection against the same target. Skimming the binder pool with

combinatorial competition assays could be used to rid the library of “duplicate” binders i.e.

binders against the same epitope. However, neither of the approaches does account for the

steric context that the binders are introduced to once they are embedded in the receptor

scaffold. Yet, chances are that there are binders that bind the same epitope but do so in

different angles. Removing all duplicates might therefore also lead to elimination of binders

with just the necessary orientation. Here, the use of membrane-bound selection systems like

yeast-display (Uchanski et al. 2019) might help to increase the number of relevant binders by

screening them in a receptor-like context. In conclusion, priority should be given to

maintaining diversity of the binder pool instead of applying more stringent selection

conditions. Still, an increased variety also provokes a much larger library of receptors that has

to be assessed in cell-based assays.

Fortunately, the EpoR scaffold already provides a solution to this issue which is at least

applicable to larger targets. Binding of a larger protein to either or both of the extracellular

domains leads to a drop in residual activity that results in an OFF-type switch, as can be seen

in our evaluation of homodimeric receptors. Hence binding of the target in the native context

can be detected and functional receptors and their binders can be selected for further

characterization in a combinatorial screen. Modern automation solutions could aid in this task

to enable screening of larger libraries to reduce development time and increase robustness

of the presented pipeline. The unsophisticated cloning strategy would possibly even allow for

an integration into a lab automation environment for fully autonomous generation of

functional receptors.

While high-throughput binder selection now provides a vast space of possible

epitopes, the EpoR architecture itself remains limited by the underlying mechanism of

activation which requires the expression of two receptor chains that each need to bind to the

target individually. This involves the presence of two identical epitopes on the target molecule

with suitable spacing or the generation of two different receptor chains that are each

equipped with a different binding domain. While finding binders against a given target and

Discussion


expression of two receptor chains in each cell is less of an issue for larger targets and easy-

to-transfect cell lines, detection of smaller molecules in primary cells might prove to be

difficult.

Unfortunately, the mechanism of activation of EpoR hampers its use for detection of

cell-surface markers and its signal transduction relies on endogenous signaling pathways.

However, complementary receptor architectures are available for use in cell-cell

communication that could potentially fill the gaps in EpoR-based signaling. The degree of

orthogonality depends on the signaling moieties of the respective receptor and can be

selected according to the desired application. For example, the modular extracellular sensor

architecture (MESA) (Schwarz et al. 2017) promises fully orthogonal signaling via the tTA

system but requires tedious fine tuning of expression levels of each component. However,

existing “off-the-shelf” MESA-receptors should be easy to integrate. Harnessing the signal

amplification capabilities of the cell’s native signal transduction cascades provides greater

sensitivity without the need for additional components. Chimeric G-proteins (Coward et al.

1999), GPCR-derived RASSLs (Redfern et al. 1999), chimeric antigen receptors (CARs) (Kuwana

et al. 1987) and synthetic Notch (synNotch) receptors (Morsut et al. 2016) all reroute a given

input signal to an existing endogenous pathway. In this context, the ability to reroute

additional inputs through the AMBER platform described in chapter 1 can be expected to

provide valuable synergistic effects. In combination with the existing receptor technologies,

it would be possible to specify disease conditions more precisely to reduce side-effects in

potential clinical applications of cell therapies. Also, it would be possible to mirror

developmental niches by mimicking the effects of more intricate growth factor combinations.

Obviously, the ultimate receptor for mammalian cell signaling applications is a small

monomeric receptor with a single transmembrane domain that would be activated by binding

of a single ligand molecule while downstream signaling would be modular to allow for entirely

orthogonal signaling or activation of native signal transduction pathways. Since there is no

such receptor to be found in nature, a possible solution might lie in a computationally

designed receptor scaffold. So far, fully de-novo designed new-to-nature receptors for

orthogonal cell signaling are still science-fiction. However, with recent progress in resolution

of protein structures in silico (Senior et al. 2020) and protein engineering (Yang et al. 2021) in

combination with ever improving computing power such an approach might lead to

functional designs in the not too far future.

Controlling cellular behavior with small molecules


EpoR based receptors equipped with a single DARPin are as small as 2.5 kb while scFv

versions account for a 3 kb payload. Both receptor types could hence be used in gene therapy

vehicles like AAVs for example to enhance CAR-T cells by activation of the IL6/STAT3 pathway

(Li et al. 2018; Breuer et al. 2020) or in situ to provide protection of cardiomyocytes against

hypoxia-induced damage during ischemia through simulation of VEGF signaling (Parvin et al.

2014).

Controlling cellular behavior with small molecules

The second chapter of this thesis details the use of the inert-to-human small-molecule

inducer L-glucose in a novel gene switch. There are already a variety of small-molecule

induced gene switches that come in different flavors. ON- and OFF-type switches in

mammalian cells were reported using doxycycline (Gossen et al. 1992), vanillic acid (Gitzinger

et al. 2012), glucocorticoids (James et al. 2000) or cumate (Mullick et al. 2006) just to name a

few. Those systems are commonly used to control gene expression in vitro and there is a

growing number of in vivo applications (Gitzinger et al. 2012; Wang et al. 2020; Sakamoto et

al. 2020).

Compared to these examples, the use of L-glucose as inducer molecule is unique in

the sense that it does not employ the inducer molecule directly but relies on a metabolic

process to convert the pre-inducer into an active ligand. When designing the L-glucose gene

switch we found that there is no natural transcription factor or receptor available that would

be activated directly by L-glucose, hence L-glucose can be used as an orthogonal inducer.

Metabolic engineering typically aims to augment host cell function in order to increase

product yield and quality as detailed in the introduction. In this case, however, the metabolic

functions that were integrated did not target endogenous processes but were used to enable

conversion and detection of the small molecule inducer L-glucose, which makes this system

a metabolically activated gene switch. In a recent study from our group a new reporter system

for continuous monitoring of promoter activity was designed by introducing metabolic

functions into HEK-293T cells. Here, the production of the fluorescent dye betaxanthin was

possible by the ectopic expression of fungal enzymes leading to fluorescent cells upon

expression of the key enzyme of betaxanthin synthesis (Stucheli et al. 2020). Both studies

show-case a cross-kingdom implementation of synthetic biology and might prompt further

Discussion


efforts in metabolic engineering in mammalian systems beyond improvements in

biopharmaceutical production to engineer artificial reporter systems and orthogonal gene

switches.

Implementing novel small-molecule induced gene switches is especially relevant since

inducers for some of the established gene switches are reported to affect cell viability and

productivity as seen for the doxycycline system (Moullan et al. 2015) or are not orthogonal

like the glucocorticoid-based gene switch. That said, there still remains a couple of gene

switches that show similar characteristics like the presented L-glucose gene circuit. Therefore,

we see the presented L-glucose gene switch as an extension of the existing toolbox that can

be freely combined with other gene switches that are controlled by e.g. vanillic acid or cumate

to control expression of multiple transgenes independently. A combination of small-molecule

inducers vanillic acid and benzoate was already used to control expression of a transgene

independently as well as in combination as part of a dual-input system (Xie et al. 2014).

Because the inducer L-glucose cannot be degraded by mammalian cells and gets

readily cleared from the blood stream through the kidneys, it is an ideal non-toxic compound

with respect to safety concerns (Raymer et al. 2003). On the downside of this quick clearance

rate and the resulting low concentrations of L-glucose in the blood stream, implementations

of L-glucose responsive gene switches for use in in vivo applications become challenging.

However, as we were able to show, these characteristics still allow for use of L-glucose in

closed systems like a bioreactor.

The orthogonal character of L-glucose metabolism also allows to speculate on the use

of L-glucose as an energy source to enable antibiotics free selection. By expression of the

three missing enzymes LgnE, LgnF and LgnG from Paraccocus sp.43P the L-glucose

degradation pathway could be completed. Thereby, L-glucose could be effectively converted

into pyruvate and glycerine-aldehyde-3-phosphate (GAP) which are common intermediates

of glycolysis and energy-rich substrates in eucaryotes. Transfecting cells with plasmids

encoding a transgene of choice that is genetically linked to the expression of all necessary

catabolic enzymes could enable the specific selection of transgenic cells under D-glucose

starvation and continuous supplementation with L-glucose. Such an orthogonal selection

process would omit the use of antibiotics and simplify downstream processing of

biopharmaceuticals. The use of other orthogonal energy-rich substrates that are cheaper and

easier to produce in higher quantities could further improve usability of the approach.

Mechanical cues to control gene expression and lineage commitment


Mechanical cues to control gene expression and lineage commitment

In chapter 3 and 4 we studied effects of mechanical cues on calcium signaling and wnt

signaling, respectively. Calcium-activated nuclear factor and activator of transcription (NFAT)

was assessed besides an artificial orthogonal calcium-sensitive pathway for gene expression

based on the transcriptional activator tTA in combination with a protease from tobacco etch

virus (TEV). We also screened accessory receptors, channels and transcription factors

involved in calcium homeostasis for their potential to boost NFAT-activation. In order to

provide a higher-throughput protocol to study effects of shear stress, we implemented a

device that was based on a 3D-printed piston array attached to the membrane of a speaker

to allow for the generation of turbulent flow in a 96-well cell culture plate. We used this

device to explore the dependency of mechanically induced gene expression on extracellular

matrix density and composition as well as on the density of the cell population. Mechanical

cues generated from the same device were also used to modulate lineage commitment in

human induced pluripotent stem cells (hiPSCs).

In its native context, mechanosensation is involved in regulation of blood pressure

through sensation of shear stress in the vasculature (Wang et al. 2016). In addition,

mechanical properties of the nearby extracellular space are known cues in stem cell and tissue

development (Kerstein et al. 2013). More specifically, shear stress is a known modulator of

Wnt signaling (Arnsdorf et al. 2009; Jansen et al. 2010; Cha et al. 2016). Hence, we studied

the impact of mechanical stimulation on Wnt signaling by means of a stable iPSC reporter cell

line. In addition to a reversible quenching of Wnt signaling, we also found alterations in

differentiation into ectoderm and mesoderm lineages under standard differentiation

conditions. The presented protocol is non-invasive, does not require changes in growth factor

composition and can be easily combined with established methods.

Recent advances using focused ultrasound for remote gene expression based on

thermal effects in the cell might offer an alternative induction method for mechanical cues

that can be more readily applied to clinical settings (Liu et al. 2006). Control of gene

expression by temperature has been achieved through the use of heat-shock-protein (HSP)

70B-induced promoters (Miller et al. 2018). Alternatively, heat sensitive coiled-coil protein

domains derived from the bacterial regulator TlpA were successfully implemented to control

gene expression in a tunable manner (Piraner et al. 2017; Piraner et al. 2019). A rather

disputed (Meister 2016) study claimed induction of calcium responses in neurons by magnetic

Discussion


ferritin tethered to the cytosolic side of the TRPV4 channel (Wheeler et al. 2016). Recent

experiments indicate that direct induction of heat-sensitive ion channels like transient

receptor potential cation channel subfamily V member 1 (TRPV1) might be feasible although

at temperatures not relevant for use in humans (Zhang et al. 2018).

Although thermal cues might seem more amenable to control cellular behavior in vivo

because the necessary technologies for thermal induction are omnipresent, thermally-

controlled gene switches suffer from unspecific or non-orthogonal induction. Hence the use

of such switches is limited to applications outside of the patient or to occasions when

systemically elevated temperatures are assessed i.e. in case of fever. On the other hand,

elevated temperatures were reported to promote differentiation of neural stem cells (Wang

et al. 2017) and hence can be considered save with respect to cell viability. Although, this

finding should not be generalized, temperature induced gene switches might indeed provide

a means for modulation of stem cell differentiation in vitro.

In contrast, purely mechanically induced gene expression offers the possibility to

detect endogenous signals while also allowing for specific induction of gene expression via

non-invasive artificial external stimuli like ultrasound or shear stress. Studies on

mechanogenetics for the remote control of cellular behavior have recently shown promising

results. In a first potential therapeutic application of ectopically expressed Piezo1, ultrasound

has been used to induce gene expression in T-cells to enhance their cytotoxic response to

cancer cells (Pan et al. 2018). However, this non-invasive stimulation method is still hampered

by requiring additional supplements (microbubbles) being present in the very vicinity of the

cells. Against this background, gene circuits and cells that are triggered by pure mechanical

stimulation in a therapy setting become indeed achievable. Such mechanogenetic systems

could be used in basic research as well as in next-generation medicine to monitor shear stress

in the patient’s body.

We hope that the results and methods described in chapter 3 and 4 facilitate research

on calcium signaling and cell differentiation including the channels that are involved as well

as on the role that cell-cell contact and cell adherence play. Since the shear stress used in this

experiment provides a general stimulus that works on the endogenous set of signaling

pathways, this method could be of special interest in stem cell research to omit the use of

costly growth factors used as supplements in established differentiation media. The so

reduced complexity would lead to a more robust differentiation protocol. Improving on the

Developments and future applications of mammalian synthetic biology


generation of high-quality cells from patient-derived iPSCs which contribute to next-

generation personalized medicine in two ways. First, the generated cells could directly be

used in regenerative medicine. Second, the produced cell types could also be employed for

drug screening and testing in a patient-centered approach to account for patient-specific

anomalies in drug metabolism. Such processes are already examined for use in cardiovascular

disease (Paik et al. 2020) e.g. to assess liver toxicity (Gough et al. 2020) and rely on established

protocols for cell line differentiation.


The projects presented in this thesis contribute to the overall goal of mammalian synthetic

biology of strengthening our understanding of cellular systems to ultimately being able to

improve (cell) therapies and biotechnological processes (figure III). Key to this goal is a way to

control cellular behavior in response to various inputs. In turns, controlling cellular behavior

through means of synthetic biology requires a solid understanding of the underlying

processes. To this end synthetic biology provides a rich toolbox of receptors, transcription

factors and response elements to alter and reroute cellular signaling to synthetic promoters.

In order to speed up the process of synthetic biological engineering, the BioBrick standard

(Constante et al. 2011) and the standardized European vector architecture (SEVA) (Silva-

Rocha et al. 2013) were established for genetic engineering of bacteria. SEVA is now

commonly used and continuously updated (Martinez-Garcia et al. 2020). Although it took a

while, similar initiatives were started that specifically cater to the needs of mammalian cells

with the goal of a unified genetic platform to stream-line synthetic biology-driven

engineering. To this end COmposable Mammalian Elements of Transcription (COMET) to

control gene expression via designed orthogonal transcription factors (Donahue et al. 2020)

were developed alongside a Versatile plasmid Architecture for Mammalian Synthetic Biology

(VAMSyB) (Haellman et al., unpublished) to provide a common vector and cloning strategy to

enhance reproducibility and exchange of modules between research groups. Lower costs and

increased turn-around times for custom gene synthesis are beginning to make in-house

cloning a thing of the past. Especially methods like Gibson cloning that require additional time

for synthesis of oligonucleotides in each step (Gibson et al. 2009) would be replaced by gene

synthesis approaches in the near future. Still, any modular cloning strategy that makes use of

Discussion


an existing repertoire of parts and modules that can be recombined quickly with high

efficiency still has an edge against de novo synthesis. Therefore, we will keep seeing SEVA- or

VAMSyB-like platforms around for a while.

Recent improvements and adaptations of the CAR-architecture as well as the original

CARs themselves serve as prime examples of synthetic biology applications. Feasibility of

DARPins as binding moieties in CARs has been demonstrated (Hammill et al. 2015; Patasic et

al. 2020) and this now allows to target a growing range of diseases by CAR-T cell therapy. We

expect CAR-T cells to be used more commonly even outside cancer therapy to fight persistent

viral and fungal infections e.g. by HIV (Leibman et al. 2017; Patasic et al. 2020) or Aspergillus

fumigatus (Kumaresan et al. 2014; Seif et al. 2019). Additionally, immune-modulatory

interventions based on chimeric autoantibody receptor (CAAR)-T cells that target auto-

reactive B-cells (Ellebrecht et al. 2016) or CAR-Treg cells that bind an allergen and subsequently

quench over-shooting immune responses were described (Elinav et al. 2008; Brunstein et al.

2016). In this regard, we are confident that our AMBER system could aid in the development

of next-generation cell therapies to further improve specificity and safety of cell therapies.

Multiple studies have been performed already to reach this goal by implementing logic gates

on various levels of T-cell activation. For one, it was shown that the synNotch-platform could

functionally replace CARs in cytotoxic T-cell response, providing an alternative for targets that

are not accessible to the T-cell receptor (Roybal et al. 2016). In an extension to this work,

CARs could be conditionally expressed upon binding of synNotch receptors to cancer-defining

cell-surface markers (Roybal et al. 2016).

In split, universal, and programmable (SUPRA)-CARs, the T-cell receptor is truncated and

lacks an antigen binding domain. Instead, SUPRA-CARs are activated by exchangeable

adapters that bind to cell-surface antigens on one side and to the T-cell receptor on the other

side thereby enabling targeting of various different cell-surface markers by the same cell (Cho

et al. 2018). Recently, a system for protein logic gates, termed “colocalization-dependant

latching orthogonal cage/key proteins” (Co-LOCKR), was shown to recruit T-cells only if a

specified set of surface markers is present thereby performing logic operations on a protein

level directly on the cell (Lajoie et al. 2020). Both Co-LOCKR and the underlying concept used

in SUPRA-CARs could potentially be adapted for use in the AMBER platform. Here, the

extracellular domains of the EpoR could serve as binding sites for Co-LOCKR proteins to enable

protein logics to control receptor activity. Similar to SUPRA-CARs, the EpoR extracellular



domains could also recruit adaptors to mediate activation of the same receptor by multiple

different targets. Furthermore, SUPRA-CARs and protein logics could greatly benefit from the

addition of de novo designed protein logic gates to further widen the available combinatorial

space (Chen et al. 2020).

As an additional safety measure, timed activation of CARs was achieved by activation of

CAR-T cells upon binding of a small-molecule. In two different designs the small molecule

triggers either the complementation of the intracellular signaling domain (Wu et al. 2015) or

mediates the installation of an extracellular antigen binding moiety (Zajc et al. 2020). A similar

functionality could be generated with EpoR-based synthetic receptors to control expression

of the effector CARs by a small molecule or disease marker. Hence, the receptor platform

described in chapter 1 could be combined with any of the described CAR modifications by

providing an additional control layer for increased specificity and safety.

Induced gene expression provides advantages beyond the field of CAR-T cell therapy

and by now also sees some applications in biotechnological production. Simple logic gates are

used not only for the production of proteins that show extensive post-translational

modifications (Lonza 2019) but also in stem cell technologies like the opti-ox™ system to

control the differentiation of stem cells into a cell type of choice upon addition of doxycyline

(Pawlowski et al. 2017). A combination of multiple inducers at once was used in a synthetic

lineage control network described for the generation of beta-like cells from iPSCs (Saxena et

al. 2016). Such lineage control networks offer more robust differentiation and greater

homogeneity of the resulting cell population thereby improving the overall efficiency of the

process enabling commercial use of iPSC-derived cell lines for basic research and personalized

medicine in the future.

Most available orthogonal gene switches are based on bacterial regulators that lack

the signal amplification systems of most eucaryotic systems similar to the L-glucose gene

switch described in chapter 2. However, by now the use of dimerizing bacterial kinases was

demonstrated which offer orthogonal signal amplification through phosphorylation of

multiple effector proteins by a single activated kinase (Scheller et al. 2020; Maze et al. 2020).

Cell-based therapies are poised to transform the field of personalized medicine. A

number of proof-of-concept studies on cell-based therapies are lined up for clinical evaluation

already ranging from simple one-input-one-output systems for the detection and treatment

of gouty arthritis (Kemmer et al. 2010) to logic gates targeting psoriasis (Schukur et al. 2015),

Discussion


allergic disease (Chassin et al. 2017) or liver damage (Bai et al. 2016). Progress in the field of

electrogenetics now enables a direct communication between electronic hardware and cells.

A pioneering study proved the feasibility of electrogenetic cell therapy for the treatment of

diabetes via induced gene expression as well as fast release of insulin from pre-formed

vesicles (Krawczyk et al. 2020).

Figure III: Synthetic biology can be used to engineer cells with new functionalities or to

augment existing cell lines. Engineered cells can be used in drug discovery to enable efficient

screening of compounds and assay toxicity (left), for direct use in patients to treat disease

(middle) or to produce and improve production of antibodies or enzyme-replacement

therapies (right).

Discussion


Conclusion


Conclusion The field of synthetic biology and related cellular therapies is advancing at a high pace.

With more genomes of species from all kingdoms of life being sequenced and more protein

structures being solved each year by the help of advanced automation techniques (Chavas et

al. 2015) we now know more parts and modules for synthetic biology-inspired applications

than ever before. Furthermore, this pace might even increase in the future since already

today protein structures can be predicted in silico with high accuracy (Senior et al. 2020) and

new gene switches can be designed computationally (Chen et al. 2020).

However, despite ongoing efforts to improve the underlying hardware used for

cellular implants (Bose et al. 2020) the number of real-world applications outside of stem cell

transplants is still limited. The approval of CAR-T cells and AAV-mediated gene therapies and

vaccines alongside major investments of big pharmaceutical companies might mark a turning

point in this regard with more money and human resources getting involved to accelerate the

transition of cell therapies from lab benches to patients.

In summary one can expect synthetic biology to have a growing impact on

biotechnological production of classic and modern biopharmaceuticals like antibodies and

gene therapy vehicles as well as on personalized medicine via the generation of patient-

derived disease models and autologous cell therapies in the future (figure III).

Conclusion


Additional References


Additional References Amann, T., V. Schmieder, H. Faustrup Kildegaard, N. Borth, and M. R. Andersen. 2019. 'Genetic

engineering approaches to improve posttranslational modification of

biopharmaceuticals in different production platforms', Biotechnol Bioeng, 116: 2778-

96.

Armbruster, B. N., X. Li, M. H. Pausch, S. Herlitze, and B. L. Roth. 2007. 'Evolving the lock to fit

the key to create a family of G protein-coupled receptors potently activated by an

inert ligand', Proc Natl Acad Sci U S A, 104: 5163-8.

Arnsdorf, E. J., P. Tummala, and C. R. Jacobs. 2009. 'Non-canonical Wnt signaling and N-

cadherin related beta-catenin signaling play a role in mechanically induced osteogenic

cell fate', PLoS One, 4: e5388.

Auslander, D., S. Auslander, X. Pierrat, L. Hellmann, L. Rachid, and M. Fussenegger. 2018.

'Programmable full-adder computations in communicating three-dimensional cell

cultures', Nat Methods, 15: 57-60.

Auslander, S., D. Fuchs, S. Hurlemann, D. Auslander, and M. Fussenegger. 2016. 'Engineering

a ribozyme cleavage-induced split fluorescent aptamer complementation assay',

Nucleic Acids Res, 44: e94.

Auslander, S., P. Stucheli, C. Rehm, D. Auslander, J. S. Hartig, and M. Fussenegger. 2014. 'A

general design strategy for protein-responsive riboswitches in mammalian cells', Nat

Methods, 11: 1154-60.

Bacchus, W., M. Lang, M. D. El-Baba, W. Weber, J. Stelling, and M. Fussenegger. 2012.

'Synthetic two-way communication between mammalian cells', Nature Biotechnology,

30: 991-+.

Bai, P., Y. Liu, S. Xue, G. C. E. Hamri, P. Saxena, H. F. Ye, M. Q. Xie, and M. Fussenegger. 2019.

'A fully human transgene switch to regulate therapeutic protein production by cooling

sensation', Nature Medicine, 25: 1266-+.

Bai, P., H. Ye, M. Xie, P. Saxena, H. Zulewski, G. Charpin-El Hamri, V. Djonov, and M.

Fussenegger. 2016. 'A synthetic biology-based device prevents liver injury in mice', J

Hepatol, 65: 84-94.

Berkovits, B. D., and C. Mayr. 2015. 'Alternative 3 ' UTRs act as scaffolds to regulate membrane

protein localization', Nature, 522: 363-+.



Bose, S., L. R. Volpatti, D. Thiono, V. Yesilyurt, C. McGladrigan, Y. Y. Tang, A. Facklam, A. Wang,

S. Jhunjhunwala, O. Veiseh, J. Hollister-Lock, C. Bhattacharya, G. C. Weir, D. L. Greiner,

R. Langer, and D. G. Anderson. 2020. 'A retrievable implant for the long-term

encapsulation and survival of therapeutic xenogeneic cells', Nature Biomedical

Engineering.

Brunstein, C. G., J. S. Miller, D. H. McKenna, K. L. Hippen, T. E. DeFor, D. Sumstad, J. Curtsinger,

M. R. Verneris, M. L. MacMillan, B. L. Levine, J. L. Riley, C. H. June, C. Le, D. J. Weisdorf,

P. B. McGlave, B. R. Blazar, and J. E. Wagner. 2016. 'Umbilical cord blood-derived T

regulatory cells to prevent GVHD: kinetics, toxicity profile, and clinical effect', Blood,

127: 1044-51.

Cha, B., X. Geng, M. R. Mahamud, J. Fu, A. Mukherjee, Y. Kim, E. H. Jho, T. H. Kim, M. L. Kahn,

L. Xia, J. B. Dixon, H. Chen, and R. S. Srinivasan. 2016. 'Mechanotransduction activates

canonical Wnt/beta-catenin signaling to promote lymphatic vascular patterning and

the development of lymphatic and lymphovenous valves', Genes Dev, 30: 1454-69.

Chassin, H., B. Geering, L. Schukur, D. Auslander, B. Lang, and M. Fussenegger. 2017. 'Sensing

and responding to allergic response cytokines through a genetically encoded circuit',

Nature Communications, 8.

Chassin, H., M. Muller, M. Tigges, L. Scheller, M. Lang, and M. Fussenegger. 2019. 'A modular

degron library for synthetic circuits in mammalian cells', Nat Commun, 10: 2013.

Chavas, L. M. G., L. Gumprecht, and H. N. Chapman. 2015. 'Possibilities for serial femtosecond

crystallography sample delivery at future light sources', Structural Dynamics-Us, 2.

Chen, Z., R. D. Kibler, A. Hunt, F. Busch, J. Pearl, M. Jia, Z. L. VanAernum, B. I. M. Wicky, G.

Dods, H. Liao, M. S. Wilken, C. Ciarlo, S. Green, H. El-Samad, J. Stamatoyannopoulos,

V. H. Wysocki, M. C. Jewett, S. E. Boyken, and D. Baker. 2020. 'De novo design of

protein logic gates', Science, 368: 78-84.

Cho, J. H., J. J. Collins, and W. W. Wong. 2018. 'Universal Chimeric Antigen Receptors for

Multiplexed and Logical Control of T Cell Responses', Cell, 173: 1426-38 e11.

Conklin, B. R., E. C. Hsiao, S. Claeysen, A. Dumuis, S. Srinivasan, J. R. Forsayeth, J. M. Guettier,

W. C. Chang, Y. Pei, K. D. McCarthy, R. A. Nissenson, J. Wess, J. Bockaert, and B. L.

Roth. 2008. 'Engineering GPCR signaling pathways with RASSLs', Nature Methods, 5:

673-78.



Constante, M., R. Grunberg, and M. Isalan. 2011. 'A Biobrick Library for Cloning Custom

Eukaryotic Plasmids', PLoS One, 6.

Coward, P., S. D. Chan, H. G. Wada, G. M. Humphries, and B. R. Conklin. 1999. 'Chimeric G

proteins allow a high-throughput signaling assay of Gi-coupled receptors', Anal

Biochem, 270: 242-8.

Cuesta, A. M., N. Sainz-Pastor, J. Bonet, B. Oliva, and L. Alvarez-Vallina. 2010. 'Multivalent

antibodies: when design surpasses evolution', Trends in Biotechnology, 28: 355-62.

Donahue, P. S., J. W. Draut, J. J. Muldoon, H. I. Edelstein, N. Bagheri, and J. N. Leonard. 2020.

'The COMET toolkit for composing customizable genetic programs in mammalian

cells', Nature Communications, 11.

Elinav, E., T. Waks, and Z. Eshhar. 2008. 'Redirection of regulatory T cells with predetermined

specificity for the treatment of experimental colitis in mice', Gastroenterology, 134:

2014-24.

Ellebrecht, C. T., V. G. Bhoj, A. Nace, E. J. Choi, X. Mao, M. J. Cho, G. Di Zenzo, A. Lanzavecchia,

J. T. Seykora, G. Cotsarelis, M. C. Milone, and A. S. Payne. 2016. 'Reengineering

chimeric antigen receptor T cells for targeted therapy of autoimmune disease',

Science, 353: 179-84.

Fan, L. C., I. Kadura, L. E. Krebs, C. C. Hatfield, M. M. Shaw, and C. C. Frye. 2012. 'Improving

the efficiency of CHO cell line generation using glutamine synthetase gene knockout

cells', Biotechnology and Bioengineering, 109: 1007-15.

Gibson, D. G., L. Young, R. Y. Chuang, J. C. Venter, C. A. Hutchison, and H. O. Smith. 2009.

'Enzymatic assembly of DNA molecules up to several hundred kilobases', Nature

Methods, 6: 343-U41.

Gitzinger, M., C. Kemmer, M. D. El-Baba, W. Weber, and M. Fussenegger. 2009. 'Controlling

transgene expression in subcutaneous implants using a skin lotion containing the

apple metabolite phloretin', Proceedings of the National Academy of Sciences of the

United States of America, 106: 10638-43.

Gitzinger, M., C. Kemmer, D. A. Fluri, M. D. El-Baba, W. Weber, and M. Fussenegger. 2012.

'The food additive vanillic acid controls transgene expression in mammalian cells and

mice', Nucleic Acids Research, 40.



Giuliodori, A. M., F. Di Pietro, S. Marzi, B. Masquida, R. Wagner, P. Romby, C. O. Gualerzi, and

C. L. Pon. 2010. 'The cspA mRNA Is a Thermosensor that Modulates Translation of the

Cold-Shock Protein CspA', Molecular Cell, 37: 21-33.

Gossen, M., and H. Bujard. 1992. 'Tight control of gene expression in mammalian cells by

tetracycline-responsive promoters', Proc Natl Acad Sci U S A, 89: 5547-51.

Gough, A., A. Soto-Gutierrez, L. Vernetti, M. R. Ebrahimkhani, A. M. Stern, and D. L. Taylor.

2020. 'Human biomimetic liver microphysiology systems in drug development and

precision medicine', Nat Rev Gastroenterol Hepatol.

Hammill, J. A., H. VanSeggelen, C. W. Helsen, G. F. Denisova, C. Evelegh, D. G. Tantalo, J. D.

Bassett, and J. L. Bramson. 2015. 'Designed ankyrin repeat proteins are effective

targeting elements for chimeric antigen receptors', Journal for Immunotherapy of

Cancer, 3: 55.

Helton, T. D., W. F. Xu, and D. Lipscombe. 2005. 'Neuronal L-type calcium channels open

quickly and are inhibited slowly', Journal of Neuroscience, 25: 10247-51.

Huang, M., J. Bao, and J. Nielsen. 2014. 'Biopharmaceutical protein production by

Saccharomyces cerevisiae: current state and future prospects', Pharmaceutical

Bioprocessing, 2: 167-82.

James, R. I., J. P. Elton, P. Todd, and D. S. Kompala. 2000. 'Engineering CHO cells to overexpress

a secreted reporter protein upon induction from mouse mammary tumor virus

promoter', Biotechnology and Bioengineering, 67: 134-40.

Jansen, J. H., M. Eijken, H. Jahr, H. Chiba, J. A. Verhaar, J. P. van Leeuwen, and H. Weinans.

2010. 'Stretch-induced inhibition of Wnt/beta-catenin signaling in mineralizing

osteoblasts', Journal of Orthopaedic Research, 28: 390-6.

Kallunki, T., M. Barisic, M. Jaattela, and B. Liu. 2019. 'How to Choose the Right Inducible Gene

Expression System for Mammalian Studies?', Cells, 8.

Kemmer, C., M. Gitzinger, M. Daoud-El Baba, V. Djonov, J. Stelling, and M. Fussenegger. 2010.

'Self-sufficient control of urate homeostasis in mice by a synthetic circuit', Nat

Biotechnol, 28: 355-60.

Kerstein, P. C., B. T. Jacques-Fricke, J. Rengifo, B. J. Mogen, J. C. Williams, P. A. Gottlieb, F.

Sachs, and T. M. Gomez. 2013a. 'Mechanosensitive TRPC1 Channels Promote Calpain

Proteolysis of Talin to Regulate Spinal Axon Outgrowth', Journal of Neuroscience, 33:

273-85.



———. 2013b. 'Mechanosensitive TRPC1 channels promote calpain proteolysis of talin to

regulate spinal axon outgrowth', Journal of Neuroscience, 33: 273-85.

Kim, S. H., and G. M. Lee. 2007. 'Down-regulation of lactate dehydrogenase-A by siRNAs for

reduced lactic acid formation of Chinese hamster ovary cells producing

thrombopoietin', Appl Microbiol Biotechnol, 74: 152-9.

Kitada, T., B. DiAndreth, B. Teague, and R. Weiss. 2018. 'Programming gene and engineered-

cell therapies with synthetic biology', Science, 359.

Krawczyk, K., L. Scheller, H. Kim, and M. Fussenegger. 2020. 'Rewiring of endogenous signaling

pathways to genomic targets for therapeutic cell reprogramming', Nat Commun, 11:

608.

Krawczyk, K., S. Xue, P. Buchmann, G. Charpin-El-Hamri, P. Saxena, M. D. Hussherr, J. W. Shao,

H. F. Ye, M. Q. Xie, and M. Fussenegger. 2020. 'Electrogenetic cellular insulin release

for real-time glycemic control in type 1 diabetic mice', Science, 368: 993-+.

Kumaresan, P. R., P. R. Manuri, N. D. Albert, S. Maiti, H. Singh, T. Mi, J. Roszik, B. Rabinovich,

S. Olivares, J. Krishnamurthy, L. Zhang, A. M. Najjar, M. H. Huls, D. A. Lee, R. E.

Champlin, D. P. Kontoyiannis, and L. J. Cooper. 2014. 'Bioengineering T cells to target

carbohydrate to treat opportunistic fungal infection', Proc Natl Acad Sci U S A, 111:

10660-5.

Kuwana, Y., Y. Asakura, N. Utsunomiya, M. Nakanishi, Y. Arata, S. Itoh, F. Nagase, and Y.

Kurosawa. 1987. 'Expression of chimeric receptor composed of immunoglobulin-

derived V regions and T-cell receptor-derived C regions', Biochem Biophys Res

Commun, 149: 960-8.

Lajoie, M. J., S. E. Boyken, A. I. Salter, J. Bruffey, A. Rajan, R. A. Langan, A. Olshefsky, V.

Muhunthan, M. J. Bick, M. Gewe, A. Quijano-Rubio, J. Johnson, G. Lenz, A. Nguyen, S.

Pun, C. E. Correnti, S. R. Riddell, and D. Baker. 2020. 'Designed protein logic to target

cells with precise combinations of surface antigens', Science, 369: 1637-43.

Leibman, R. S., M. W. Richardson, C. T. Ellebrecht, C. R. Maldini, J. A. Glover, A. J. Secreto, I.

Kulikovskaya, S. F. Lacey, S. R. Akkina, Y. Yi, F. Shaheen, J. Wang, K. A. Dufendach, M.

C. Holmes, R. G. Collman, A. S. Payne, and J. L. Riley. 2017. 'Supraphysiologic control

over HIV-1 replication mediated by CD8 T cells expressing a re-engineered CD4-based

chimeric antigen receptor', PLoS Pathog, 13: e1006613.



Lejeune, M., M. C. Kose, E. Duray, H. Einsele, Y. Beguin, and J. Caers. 2020. 'Bispecific, T-Cell-

Recruiting Antibodies in B-Cell Malignancies', Frontiers in Immunology, 11: 762.

Levskaya, A., O. D. Weiner, W. A. Lim, and C. A. Voigt. 2009. 'Spatiotemporal control of cell

signalling using a light-switchable protein interaction', Nature, 461: 997-1001.

Li, B., L. L. Jones, and T. L. Geiger. 2018. 'IL-6 Promotes T Cell Proliferation and Expansion

under Inflammatory Conditions in Association with Low-Level RORgammat

Expression', Journal of Immunology, 201: 2934-46.

Li, C., and R. J. Samulski. 2020. 'Engineering adeno-associated virus vectors for gene therapy',

Nat Rev Genet, 21: 255-72.

Liu, H., A. Saxena, S. S. Sidhu, and D. Wu. 2017. 'Fc Engineering for Developing Therapeutic

Bispecific Antibodies and Novel Scaffolds', Frontiers in Immunology, 8: 38.

Liu, Y., T. Kon, C. Li, and P. Zhong. 2006. 'High intensity focused ultrasound-induced gene

activation in solid tumors', J Acoust Soc Am, 120: 492-501.

Lonza. 2019. 'Lonza Chooses Synpromics’ Novel Inducible Promoter Platform to Manufacture

Biopharmaceuticals'.

Mansouri, M., T. Strittmatter, and M. Fussenegger. 2019. 'Light-Controlled Mammalian Cells

and Their Therapeutic Applications in Synthetic Biology', Adv Sci (Weinh), 6: 1800952.

Martinez-Garcia, E., A. Goni-Moreno, B. Bartley, J. McLaughlin, L. Sanchez-Sampedro, H. P.

Del Pozo, C. P. Hernandez, A. S. Marletta, D. De Lucrezia, G. Sanchez-Fernandez, S.

Fraile, and V. de Lorenzo. 2020. 'SEVA 3.0: an update of the Standard European Vector

Architecture for enabling portability of genetic constructs among diverse bacterial

hosts', Nucleic Acids Res, 48: 3395.

Maze, A., and Y. Benenson. 2020. 'Artificial signaling in mammalian cells enabled by

prokaryotic two-component system', Nature Chemical Biology, 16: 179-87.

Meister, M. 2016. 'Physical limits to magnetogenetics', Elife, 5.

Miller, I. C., M. Gamboa Castro, J. Maenza, J. P. Weis, and G. A. Kwong. 2018. 'Remote Control

of Mammalian Cells with Heat-Triggered Gene Switches and Photothermal Pulse

Trains', Acs Synthetic Biology, 7: 1167-73.

Morsut, L., K. T. Roybal, X. Xiong, R. M. Gordley, S. M. Coyle, M. Thomson, and W. A. Lim.

2016. 'Engineering Customized Cell Sensing and Response Behaviors Using Synthetic

Notch Receptors', Cell, 164: 780-91.



Moullan, N., L. Mouchiroud, X. Wang, D. Ryu, E. G. Williams, A. Mottis, V. Jovaisaite, M. V.

Frochaux, P. M. Quiros, B. Deplancke, R. H. Houtkooper, and J. Auwerx. 2015.

'Tetracyclines Disturb Mitochondrial Function across Eukaryotic Models: A Call for

Caution in Biomedical Research', Cell Reports, 10: 1681-91.

Muller, M., S. Auslander, A. Spinnler, D. Auslander, J. Sikorski, M. Folcher, and M.

Fussenegger. 2017. 'Designed cell consortia as fragrance-programmable analog-to-

digital converters', Nature Chemical Biology, 13: 309-16.

Mullick, A., Y. Xu, R. Warren, M. Koutroumanis, C. Guilbault, S. Broussau, F. Malenfant, L.

Bourget, L. Lamoureux, R. Lo, A. W. Caron, A. Pilotte, and B. Massie. 2006. 'The cumate

gene-switch: a system for regulated expression in mammalian cells', Bmc

Biotechnology, 6.





Paik, D. T., M. Chandy, and J. C. Wu. 2020. 'Patient and Disease-Specific Induced Pluripotent

Stem Cells for Discovery of Personalized Cardiovascular Drugs and Therapeutics',

Pharmacological Reviews, 72: 320-42.

Pan, Y., S. Yoon, J. Sun, Z. Huang, C. Lee, M. Allen, Y. Wu, Y. J. Chang, M. Sadelain, K. K. Shung,

S. Chien, and Y. Wang. 2018. 'Mechanogenetics for the remote and noninvasive

control of cancer immunotherapy', Proc Natl Acad Sci U S A, 115: 992-97.

Parvin, A. A., A. R. Pranap, U. Shalini, A. Devendran, J. E. Baker, and A. Dhanasekaran. 2014.

'Erythropoietin Protects Cardiomyocytes from Cell Death during Hypoxia/Reperfusion

Injury through Activation of Survival Signaling Pathways', PLoS One, 9.

Patasic, L., J. Seifried, V. Bezler, M. Kaljanac, I. C. Schneider, H. Schmitz, C. Tondera, J.

Hartmann, A. Hombach, C. J. Buchholz, H. Abken, R. Konig, and K. Cichutek. 2020.

'Designed Ankyrin Repeat Protein (DARPin) to target chimeric antigen receptor (CAR)-

redirected T cells towards CD4(+) T cells to reduce the latent HIV(+) cell reservoir',

Med Microbiol Immunol, 209: 681-91.

Pawlowski, M., D. Ortmann, A. Bertero, J. M. Tavares, R. A. Pedersen, L. Vallier, and M. R. N.

Kotter. 2017. 'Inducible and Deterministic Forward Programming of Human



Pluripotent Stem Cells into Neurons, Skeletal Myocytes, and Oligodendrocytes', Stem

Cell Reports, 8: 803-12.

Piraner, D. I., M. H. Abedi, B. A. Moser, A. Lee-Gosselin, and M. G. Shapiro. 2017. 'Tunable

thermal bioswitches for in vivo control of microbial therapeutics', Nature Chemical

Biology, 13: 75-80.

Piraner, D. I., Y. Wu, and M. G. Shapiro. 2019. 'Modular Thermal Control of Protein

Dimerization', Acs Synthetic Biology, 8: 2256-62.

Raymer, G. S., D. E. Hartman, W. A. Rowe, R. F. Werkman, and K. L. Koch. 2003. 'An open-label

trial of L-glucose as a colon-cleansing agent before colonoscopy', Gastrointestinal

Endoscopy, 58: 30-35.

Redfern, C. H., P. Coward, M. Y. Degtyarev, E. K. Lee, A. T. Kwa, L. Hennighausen, H. Bujard,

G. I. Fishman, and B. R. Conklin. 1999. 'Conditional expression and signaling of a

specifically designed Gi-coupled receptor in transgenic mice', Nat Biotechnol, 17: 165-

9.

Richter, F., I. Fonfara, R. Gelfert, J. Nack, E. Charpentier, and A. Moglich. 2017. 'Switchable

Cas9', Curr Opin Biotechnol, 48: 119-26.

Roybal, K. T., L. J. Rupp, L. Morsut, W. J. Walker, K. A. McNally, J. S. Park, and W. A. Lim. 2016.

'Precision Tumor Recognition by T Cells With Combinatorial Antigen-Sensing Circuits',

Cell, 164: 770-9.

Roybal, K. T., J. Z. Williams, L. Morsut, L. J. Rupp, I. Kolinko, J. H. Choe, W. J. Walker, K. A.

McNally, and W. A. Lim. 2016. 'Engineering T Cells with Customized Therapeutic

Response Programs Using Synthetic Notch Receptors', Cell, 167: 419-32 e16.

Sakamoto, K., P. D. Radler, B. L. Wehde, A. A. Triplett, H. Shrestha, R. M. Ferraiuolo, F. Amari,

V. Coppola, A. Klinakis, A. Efstratiadis, and K. U. Wagner. 2020. 'Efficient tissue-type

specific expression of target genes in a tetracycline-controlled manner from the

ubiquitously active Eef1a1 locus', Scientific Reports, 10.

Saotome, K., S. E. Murthy, J. M. Kefauver, T. Whitwam, A. Patapoutian, and A. B. Ward. 2018.

'Structure of the mechanically activated ion channel Piezo1', Nature, 554: 481-86.

Saxena, P., B. C. Heng, P. Bai, M. Folcher, H. Zulewski, and M. Fussenegger. 2016. 'A

programmable synthetic lineage-control network that differentiates human IPSCs into

glucose-sensitive insulin-secreting beta-like cells', Nature Communications, 7.



Scheller, L., and M. Fussenegger. 2019. 'From synthetic biology to human therapy: engineered

mammalian cells', Curr Opin Biotechnol, 58: 108-16.

Scheller, L., M. Schmollack, A. Bertschi, M. Mansouri, P. Saxena, and M. Fussenegger. 2020.

'Phosphoregulated orthogonal signal transduction in mammalian cells', Nat Commun,

11: 3085.

Scheller, L., T. Strittmatter, D. Fuchs, D. Bojar, and M. Fussenegger. 2018. 'Generalized

extracellular molecule sensor platform for programming cellular behavior', Nature

Chemical Biology, 14: 723-+.

Schilling, J., J. Schoppe, and A. Pluckthun. 2014. 'From DARPins to LoopDARPins: Novel

LoopDARPin Design Allows the Selection of Low Picomolar Binders in a Single Round

of Ribosome Display', Journal of Molecular Biology, 426: 691-721.

Schukur, L., B. Geering, G. Charpin-El Hamri, and M. Fussenegger. 2015. 'Implantable synthetic

cytokine converter cells with AND-gate logic treat experimental psoriasis', Sci Transl

Med, 7: 318ra201.

Schwarz, K. A., N. M. Daringer, T. B. Dolberg, and J. N. Leonard. 2017. 'Rewiring human cellular

input-output using modular extracellular sensors', Nature Chemical Biology, 13: 202-

09.

Seif, M., H. Einsele, and J. Loffler. 2019. 'CAR T Cells Beyond Cancer: Hope for

Immunomodulatory Therapy of Infectious Diseases', Frontiers in Immunology, 10:

2711.

Senior, A. W., R. Evans, J. Jumper, J. Kirkpatrick, L. Sifre, T. Green, C. L. Qin, A. Zidek, A. W. R.

Nelson, A. Bridgland, H. Penedones, S. Petersen, K. Simonyan, S. Crossan, P. Kohli, D.

T. Jones, D. Silver, K. Kavukcuoglu, and D. Hassabis. 2020. 'Improved protein structure

prediction using potentials from deep learning', Nature, 577: 706-+.

Silva-Rocha, R., E. Martinez-Garcia, B. Calles, M. Chavarria, A. Arce-Rodriguez, A. de las Heras,

A. D. Paez-Espino, G. Durante-Rodriguez, J. Kim, P. I. Nikel, R. Platero, and V. de

Lorenzo. 2013. 'The Standard European Vector Architecture (SEVA): a coherent

platform for the analysis and deployment of complex prokaryotic phenotypes', Nucleic

Acids Research, 41: D666-D75.

Stucheli, P., S. Sieber, D. W. Fuchs, L. Scheller, T. Strittmatter, P. Saxena, K. Gademann, and

M. Fussenegger. 2020. 'Genetically encoded betaxanthin-based small-molecular

fluorescent reporter for mammalian cells', Nucleic Acids Res, 48: e67.



Teixeira, A.P., and M. Fussenegger. 2019. 'Engineering mammalian cells for disease diagnosis

and treatment', Curr Opin Biotechnol, 55: 87-94.

Toussaint, C., O. Henry, and Y. Durocher. 2016. 'Metabolic engineering of CHO cells to alter

lactate metabolism during fed-batch cultures', J Biotechnol, 217: 122-31.

Tsui, P., M. A. Tornetta, R. S. Ames, C. Silverman, T. Porter, C. Weston, S. Griego, and R. W.

Sweet. 2002. 'Progressive epitope-blocked panning of a phage library for isolation of

human RSV antibodies', Journal of Immunological Methods, 263: 123-32.

Uchanski, T., T. Zogg, J. Yin, D. P. Yuan, A. Wohlkonig, B. Fischer, D. M. Rosenbaum, B. K.

Kobilka, E. Pardon, and J. Steyaert. 2019. 'An improved yeast surface display platform

for the screening of nanobody immune libraries', Scientific Reports, 9.

Vilaboa, N., M. Fenna, J. Munson, S. M. Roberts, and R. Voellmy. 2005. 'Novel gene switches

for targeted and timed expression of proteins of interest', Molecular Therapy, 12: 290-

8.

Wang, H., H. Ye, M. Xie, M. Daoud El-Baba, and M. Fussenegger. 2015. 'Cosmetics-triggered

percutaneous remote control of transgene expression in mice', Nucleic Acids Res, 43:

e91.

Wang, L., Y. Deng, D. Duan, S. Sun, L. Ge, Y. Zhuo, T. Yuan, P. Wu, H. Wang, M. Lu, and Y. Xia.

2017. 'Hyperthermia influences fate determination of neural stem cells with lncRNAs

alterations in the early differentiation', PLoS One, 12: e0171359.

Wang, S., R. Chennupati, H. Kaur, A. Iring, N. Wettschureck, and S. Offermanns. 2016.

'Endothelial cation channel PIEZO1 controls blood pressure by mediating flow-induced

ATP release', Journal of Clinical Investigation, 126: 4527-36.

Wang, Y. D., S. Y. Liao, N. Z. Guan, Y. X. Liu, K. L. Dong, W. Weber, and H. F. Ye. 2020. 'A

versatile genetic control system in mammalian cells and mice responsive to clinically

licensed sodium ferulate', Science Advances, 6.

Watowich, S. S. 2011. 'The erythropoietin receptor: molecular structure and hematopoietic

signaling pathways', J Investig Med, 59: 1067-72.

Weinmann, J., S. Weis, J. Sippel, W. Tulalamba, A. Remes, J. El Andari, A. K. Herrmann, Q. H.

Pham, C. Borowski, S. Hille, T. Schonberger, N. Frey, M. Lenter, T. VandenDriessche,

O. J. Muller, M. K. Chuah, T. Lamla, and D. Grimm. 2020. 'Identification of a myotropic

AAV by massively parallel in vivo evaluation of barcoded capsid variants', Nat

Commun, 11: 5432.



Wheeler, M. A., C. J. Smith, M. Ottolini, B. S. Barker, A. M. Purohit, R. M. Grippo, R. P.

Gaykema, A. J. Spano, M. P. Beenhakker, S. Kucenas, M. K. Patel, C. D. Deppmann, and

A. D. Guler. 2016. 'Genetically targeted magnetic control of the nervous system', Nat

Neurosci, 19: 756-61.

Wu, C. Y., K. T. Roybal, E. M. Puchner, J. Onuffer, and W. A. Lim. 2015. 'Remote control of

therapeutic T cells through a small molecule-gated chimeric receptor', Science, 350:

aab4077.

Xie, M., H. Ye, G. C. Hamri, and M. Fussenegger. 2014. 'Antagonistic control of a dual-input

mammalian gene switch by food additives', Nucleic Acids Res, 42: e116.

Yang, C., F. Sesterhenn, J. Bonet, E. A. van Aalen, L. Scheller, L. A. Abriata, J. T. Cramer, X. Wen,

S. Rosset, S. Georgeon, T. Jardetzky, T. Krey, M. Fussenegger, M. Merkx, and B. E.

Correia. 2021. 'Bottom-up de novo design of functional proteins with complex

structural features', Nature Chemical Biology.

Yoast, R. E., S. M. Emrich, X. Zhang, P. Xin, M. T. Johnson, A. J. Fike, V. Walter, N. Hempel, D.

I. Yule, J. Sneyd, D. L. Gill, and M. Trebak. 2020. 'The native ORAI channel trio underlies

the diversity of Ca2+ signaling events', Nature Communications, 11.

Yu, D. Y., S. Y. Lee, and G. M. Lee. 2018. 'Glutamine synthetase gene knockout-human

embryonic kidney 293E cells for stable production of monoclonal antibodies',

Biotechnol Bioeng, 115: 1367-72.

Zajc, C. U., M. Dobersberger, I. Schaffner, G. Mlynek, D. Puhringer, B. Salzer, K. Djinovic-

Carugo, P. Steinberger, A. De Sousa Linhares, N. J. Yang, C. Obinger, W. Holter, M. W.

Traxlmayr, and M. Lehner. 2020. 'A conformation-specific ON-switch for controlling

CAR T cells with an orally available drug', Proc Natl Acad Sci U S A, 117: 14926-35.

Zhang, F., A. Jara-Oseguera, T. H. Chang, C. H. Bae, S. M. Hanson, and K. J. Swartz. 2018. 'Heat

activation is intrinsic to the pore domain of TRPV1', Proceedings of the National

Academy of Sciences of the United States of America, 115: E317-E24.

Zhou, M., Y. Crawford, D. Ng, J. Tung, A. F. Pynn, A. Meier, I. H. Yuk, N. Vijayasankaran, K.

Leach, J. Joly, B. Snedecor, and A. Shen. 2011. 'Decreasing lactate level and increasing

antibody production in Chinese Hamster Ovary cells (CHO) by reducing the expression

of lactate dehydrogenase and pyruvate dehydrogenase kinases', J Biotechnol, 153: 27-

34.



Acknowledgements


Acknowledgements I would like to thank …

… Professor Dr. Martin Fussenegger, for the opportunity to pursue my doctoral thesis

in his lab and all the helpful insights that I gathered during this time.

… my thesis committee, Professor Dr. Renato Paro, Professor Dr. Ivan Martin and

Professor Dr. Randall Platt for their great support and interesting discussions.

I also would like to thank Professor Dr. Niko Beerenwinkel for chairing the defense.

… all authors and co-workers for the fruitful collaborations, most important the

Plückthun group of University of Zurich and Benji and Krishna from the Müller lab of

D-BSSE.

... the single-cell facility of the BSSE, the functional genomics center Zürich (FGCZ) and

the proteomics facility of the University of Basel for their essential support and

expertise.

... the active members and alumni of the Fussenegger Group for a great time and many

(un-)interesting discussions.

... my family for their backing me in this adventure and for providing an anchor during

the PhD-life.

... Rebekka, for everything.

... Theo, for a little bit more.

Date post:	09-Feb-2023
Category:	Documents
Upload:	khangminh22
View:	0 times
Download:	0 times

Controlling Cellular Behavior with Synthetic Biology

Documents