Automated data analysis in NMR fragment based lead discovery targeting LARG-RhoA

pathway

Ke RuanSchool of Life Science

University of Science and Technology of China

ENC/ACD, April, 2012

Fragment Based Screening

Key techniques for drug screening

1960s: Natural products

1970s: Computational chemistry, QSAR

1980s: Structure based drug discovery

1990s: Combinational chemistry, HTS

2000s: fragment based drug discovery

Typical binding pocket of Kinase

“hot spot” of protein-protein interaction

• High hit rate

• Weak but high quality interaction detected by

biophysical methods, e.g., NMR and SPR

• Applicable to novel PPI target

FBS vs. HTS

Small is better (sample bigger chemical space, higher hit rate, higher Ligand Efficiency, less false positive results, more intuitive to medicinal chemists)

Target

Synthetic library

Fragment growth link merge

Fragments:

Fragment based screening

High throughput screening~106 compounds

FBS

~1000 compounds

MW ~ 250DaPossible "drug-like" compounds: 109

MW ~ 500DaPossible "drug-like"compounds: 1020~10200

Fragment Based Lead Discovery

PPI targets

NMR fragment screening at700MHz spectrometer equippedwith 96 well autosampler

fragment library

Fragment screening and cross validation

Flurescencepolarization

N

NH

CH3

CH3

Br

CH3CH3

O

Hits design,synthesisand optimization

cancer cells

moleculetargetinteraction

Target validation

liver cancers, breast

cancers, etc.

Specific and potentcompound

0 50 100 150 200

0.00

0.05

0.10

0.15

0.20

Pol

ariz

atio

n(m

P)

Concentration (mM)

in vitro and in vivoassay

N

NH

CH3

CH3

Br

CH3CH3

O

Clinical candidates of fragment-derived compounds

Compound Company Status Target Therapy areas

ABT-263 Abbott Genentech

Phase II Bcl-xL Small-cell lung cancer, CML, Lymphoma, Hematological neoplasm Cancer

ABT-869 Abbott Phase II VEGF and PDGF receptor tyrosine kinase family members

Non-small-cell lungcancer, Myelodysplasticsyndrome, Acutemyelogenous leukemiaRenal cell carcinomaHepatocellularcarcinoma, Breast tumorCancer, Colorectal tumor

AT-7519 Astex Phase II CDK family members

Multiple myeloma cancer

AT-9283 Astex Phase II Aurora kinase family membersFlt3 tyrosine kinase, Jak2 tyrosine kinaseAbl tyrosine kinase

HematologicalneoplasmSolid tumor

Compound Company Status Target Therapy areas

DG-051 deCODE Phase II Leukotriene A4hydrolase

Myocardial infarction

PLX-204 Plexxikon Phase II PPAR alpha, delta, gamma

Inflammatory diseaseCardiovascular diseaseNon-insulin dependentdiabetes

LY-517717 Lilly Tularik Phase II Factor Xa Thrombosis

NVP-AUY-922 Vernalis Novartis

Phase II ATPase Hsp 90 Cancer Solid tumor

ABT-518 Abbott Phase I GelatinaseMetalloprotease-2Metalloprotease-9

Solid tumor

IC-776 Lilly ICOS Phase I CD11a, ICAM Inflammatory disease, Psoriasis, Autoimmune disease

AT-13387 Astex Phase I Hsp 90 Cancer

Compound Company Status Target Therapy areas

PLX-4032 Plexxikon Roche

Phase I Raf B protein kinase

Melanoma Cancer

PLX-5568 Plexxikon Roche

Phase I Raf protein kinase Pain, Polycystic kidney disease

SNS-314 Sunesis Phase I Aurora protein kinase

Cancer solid tumor

AT-13148 Astex, ICR CRT AstraZeneca

Phase I AKT protein kinase Cancer

SGX-393 Lilly (SGX) IND for Phase I

Abl tyrosine kinase Cancer

Expert Opin. Drug Discov. (2009) 4:1125

>10 fragment derived compounds in clinical phase II from pharmaceutical companies, e.g., Novartis, Lily, Abbott, Genentech, Astex, Vernalis, Plexxikon.>20 fragment derived compounds in clinical phase I in 2000s.

Fragment Library Buildup

RO3: 110 < MW <250 ~ 300; cLogP < 2 ~ 3 (or cLogD < 2 ~ 3); 2 < N+O < 6

logS > -4.5; TPSA < 110

Final: 1008 fragments condensed from Chembridge Library

commercial fragment library

RO3 Filter

Remove compounds w/o aromatic peaks

Remove compounds with a pairwiseTanimoto similarity score beyond threshold

Prepare new library in stock

Phase I

Fragment QC

Load FIDs/Parameters & FT

Acquire 1D solvent suppressed proton spectra

Build multiplets

Remove solvent peaks

Pseudo combined verification

Hmax=Hex+HAOI+prediction errorHmin=HAOI-prediction error

Find min of integration accuracy

measure concentration by intensity/proton

Upload to database for user scrutiny

Hmax=HtotalHmin=1

pass

ed

faile

d

Phase II

Processed data

Method [Protein] NMR measurement

Applicability

Transverse Relaxation

High 1H R2 Peak overlap

PRE Low 1H R2 spin label attached to protein or ligands, knowledge of protein structure

19F relaxation low 19F R2 Less background noise, high sensitivity (~1H), better dispersion (900 ppm), specific compounds.

STD ~1 µM/tube 1H intensity Most widely applied, large MW target, KD of 100nM to mM

WaterLOGSY > STD 1H sign(invert H2O)

Hydrophilic protein, control waterLOGSY for free ligand, KD>0.1µM

Transferred NOE

high NOESY,sign change

KD of 100nM to mM, less sensitive than STD and WaterLOGSY

Ligand-Based NMR Screening Methods

Saturation Transfer Difference (STD)

WaterLOGSY

Work flow of NMR fragment based screening

Phase III

screening compounds:solubility>0.1mM; consistent structure

Replace the compound having overlappingscore above threshold with a later compoundtill all compounds in the same corktail haveoverlapping score below threshold. Repeatfor the rest corktails

Acquire proton/STD/WaterLOGSY for every corktailwithout (negative control) and with target proteins

Process spectra and database those spectra withreference spectra, one record per corktail

User scrutiny and conversion to excel report

Optimized FBS corktail

Random selected

Optimized by automation script

Hits to leads

Evolving Fragments

NH

NOH

OH O

O

NO

O

NHOH

OHN

OOH

OH CO2Me

OOH

OH

O

O

O

HH

O

ClOH

OH Me

Known Ligands

Detailed DesignNVP-AUY922Phase I/II

Merging

Virtual Screen

Linking

NH

OS

O O

NH

S

N

ON

N

Cl

SO

O

F

F

F

ABT-263

Phase II

OHO

F

OH

BA

Growing

NH

NNH

N

AT9283

Phase IINH

NNH

NNHO

NH

NNH

NNH

NH

ON

O

Fragment 1 µµµµM 0.07 µµµµM 0.003 µµµµM

Structure details especially orientations of fragments are critical for FBDD

Target Validation via Traffic Light System

Frearson et al. Trends in Parasitology 2007

Target Validation

No or weak evidence that the target is essential for growth, survival or invasion & metastasis

Either genetic or chemical evidence

Genetic and chemical evidence

Assay Feasibility

No in vitro assay developed and/or significant problems

In vitro assay exists with plate format feasible, but not achieved

Assay ready in plate format and protein supply assured

Druggability

Novel target class and no known inhibitors or substrate analogues exist

Know Target class / inhibitors of the class

known target class and inhibitors with drug-like properties / a clinical precedent within the gene family

Toxicity Issues

Host homologue present and little or no structural, chemical or other evidence that selective inhibition is possible

Host homologue present, but some structural, chemical or other evidence

Tumour specific or host homologue known to be non-essential

SBDD suitability

No structure of target or closely related homologue

Structure without ligand available/poor resolution (>2.3 Å)/ homology model

Ligand bound structure of target available at high resolution (< 2.3 Å)

● ●●

LARG Pathway

LARG Structure

Conformation changed upon peptide binding for LARG PDZ domain

Soaking may not become feasible.

Liu et al, Protein Sci 2008

LARG DH-PH/RhoA X-ray structure

Kristelly et al. JBC 2004

New NMR techniques in fragment assembling

Specific Aim 1. Hit identification by NMR fragment based screening

Specific Aim 2. Fragment assembling assisted by novel NMR techniques, e.g., RDC & PCS

Specific Aim 3. New ligand synthesis and bioassay. Small-molecule probe to the molecular basis of tumor growth and metastasis

Residual Dipolar Couplings

Residual Dipolar Couplings (RDC) are measured as contributions to lines splittings in anisotropic solution relative to isotropic solution.

RDCs of LARG PDZ domain

-15 -10 -5 0 5 10 15

-12

-8

-4

0

4

8

12

16

Dexp(C12E5)

a

b

c

d

Dca

lc(C

12E

5)

-8 -6 -4 -2 0 2 4 6 8

-6

-4

-2

0

2

4

6

Dca

lc(p

ositi

ve S

AG

)

Dexp(positive SAG)

a. Eletropotential surface of PDZ domain (PDB: 2OMJ). b. Electro-potential surface after refined by two RDC datasets.c. The predicted RDCs derived from structure before (black) and after (red) using data in C12E5 alignment media. d. Predicted RDCs before(black) and after refinement are compared with experiment RDCs in positive strain induced acrylamide gel not used in refinement.

Pseudocontact Shift

Pintacuda etal. Acc Chem Res 2007Su etal, JACS, 2008/2009

Ln tags:

CYVDTNNDGAYEGDEL

NO

OHO

OH

SH

NO

O-O

O-

N

O O-

O

O-

N

O

O-

O O-

Ln

Preliminary PCS data

Y(DPA)33- is used as the PCS reference

Pseudocontact chemical shift under the titration of Tb(DPA)3

3- to PDZ domain

Weak binder to the hot spot

15N

ppm

10 9 8 7130

120

110

T23

F20

A79

1H ppm

b

O

O

OH

[compound]/[PDZ]01020 G21

analogy to DPA specifically binds to PDZ/Plexin-B1 hot spot, residuesin binding sites have been highly in red squares.

Acknowledgments

John LikosClaude Jones Shengtian YangGary LuekerKarey VanSantWei Wang

David SnydermanSergey GolotvinRyan Sasaki

Yunyu Shi

Jihui Wu

Jia Gao

Rongsheng Ma

$$$ support from 973 program, NSFC & Fundamental Research Funds for the Central Universities