Week 3a PDF

Challenges
in Fragment Based

Drug Discovery for
Protein Kinases

Stephen K. Burley, M.D., D.Phil.
Center for Integra@ve Proteomics Research
Department of Chemistry and Chemical Biology
Rutgers, The State University of New Jersey
Cancer Ins@tute of New Jersey

February 7th 2013
1
Outline
•  Presenta@on Goals and Background Informa@on
•  Useful Concepts/Tools for Drug Discovery Teams
•  Introduc@on to Fragment-‐Based Drug Discovery
•  FBDD/SBDD Case Study: MET Inhibitor
Non-Confidential Presentation 2
Copyright© 2012 Stephen K. Burley M.D.
Presenta@on Goals
•  Understand some of the challenges in small-‐

molecule drug discovery
•  Understand some of the advantages offered by

fragment-‐based drug discovery
•  Appreciate the possibility of unforeseeable

piValls

Drug Discovery Strategies
“Distinct Target “Biological Systems
Hypotheses” Hypotheses”
Molecular Biochemical Cellular In Vivo
Fragment Target Directed Phenotypic

Drug Discovery Screening Drug Discovery
Molecules built Repurpose/modify
Uncover molecule signatures
for purpose existing molecules
•  HT Crystallography •  Gene family platforms •  High Content Imaging •  SCORE

•  SPR •  Diversity/iterative screening (HTS) •  Advanced informatics •  Zebra fish
•  HDX •  Compound libraries •  Alternative diversity •  In vivo imaging
•  Fragment libraries •  Computational models/informatics •  Advanced cellular assays
•  High conc. assays •  Structural Biology •  Stem cells
•  Cellular and biochemical assays
Causes of AWri@on in Pharma
How can MedChemists help address ~1/3 of Pharma Attrition?
Kola and Landis(2004)

Statement of the Problem
# CNOF-containing Cpds with MW<500 ~ 1060
Subset with Pharmacologic Activity ~ <1026
Subset with Good ADMET Properties ~ 10??
•  Lipinski et al. (1997) Poor Absorp@on or Permeability more likely if
# H-‐bond don>5, # H-‐bond acc>10, MW>500, cLogP>5
àBroad Adop@on of the “Rule of 5”
•  Medicinal Chemists need to bias designàGreen Subset
Hydrophobic Effect
ΔG = ΔH -‐ T ΔS
–  Ligand displacement of par@ally ordered waters from a

hydrophobic surface feature is favorable!
–  Lipophilicity does contribute to target binding
–  Too much lipophilicityàUnwanted off-‐target binding

Why Lipophilicity MaWers (cLogP<3)!
•  Median cLogP versus Promiscuity 2,133 Cpds in 200 Cerep

assays
•  Promiscuity = # Cpds with >30%

inhibi@on at [10 µM]

•  Sigmoidal rela@onship (r = 0.84)

•  cLogP<3 more favorable!
Leeson and Springthorp (2007)

LLE (Lipophilic Ligand Efficiency)
LLE: Potency without too much Grease!
LLE= – Log(IC50) – cLogP

Exemplar Values: LLE>5; IC50<10nM and cLogP<3
•  Adding grease is an easy way to accrue “bogus” Potency
•  Must consider the efficiency of each lipophilic addi@on
•  LLE allows iden@fica@on/tracking of “good” Potency

(i.e., not driven by greaseàhigh protein binding)

Leeson and Springthorp (2007)
Why Size MaWers (MW<400)!
•  Lower MW Cpds have superior Pr(oral agent approval)!

450
440
430
420
410
400
390
380
370
Phase 1 Phase 2 Phase 3 Launch
LEAN: Drive for Potency at the Right Price!
LEAN= – log(IC50)/n [n=# of non H atoms]

Exemplar Values: LEAN>0.27; IC50<10nM and MW<400
Wenlock et al. (2003); Hopkins et al. (2004)
LEAN versus LLE Trends
•  Op@miza@on of both LEAN and LLE in drive from LEADàCS
•  Example: Kinase Project Lean vs. LLE
Lead Candidate
•  Target Values: Zone Zone
•  Lead:
–  MW≤400/IC50 ≤100nM L
àLEAN ≥~0.23
–  cLogP≤3/IC50 ≤100nM E
àLLE≥~4-‐5 A
•  Candidate: N
–  MW≤400/IC50 ≤10nM
àLEAN ≥~0.27
–  cLogP≤3/IC50 ≤10nM
àLLE ≥~5-‐8
Dot Color Scheme: RLM - Green <33%; Orange >33%, <66%; Red >66% LLE
Guidelines for Oral Drug Candidates
MW < 400
cLogP < 3
Lean > 0.27
LLE > 5
•  These are the ‘gold zone’ targets
•  There is a ‘gray zone’ but be mindful that
operating there may come with more risk
•  “Make Drugs looks like Natural Products”
Fragment-‐Based Drug Discovery
•  Fragment library screening Hits
•  Exploit literature precedents
H1
•  Structure-guided triage process
•  Pick “promising” opportunities
•  Drive Potency/SelectivityàCandidate
H3 H2
•  Perceived/Real Challenges
mM-µM Fragment Hit
•  Lower affinity starting points
•  Modest fragment libraries (1000s)
•  Hydrophobic Effect: Friend or Foe?
•  Lipophilicity Matters!
•  Size does Matter!
•  Structure is essential
nM Drug Candidate •  Are fragments selective per se?

mM-‐µM Affinity Star@ng Points
ΔGtotal = ΔGintrinsic + ΔGrigid (ΔGrigid ~3.5-5 kcal/mol)
ΔGrigid penalty due to loss of entropy
# Heavy Kd ΔGtotal ΔGintrinsic
Atoms/MW (kcal/mol) (kcal/mol)
Scaffold A 11/140Da 100µM -‐5.4 -‐9 to -‐10
A
CandidateAC 30/400Da 3nM -‐11.4 -‐15 to -‐16
A C
MW ↑ ~3-foldàAffinity ↑ ~33000-fold
Scaffold A contributes significantly to ΔGintrinsic!
Only well anchored scaffolds show up in fragment screens
Murray and Verdonk (2002)
Crystallographic Fragment Screening
Crystal soaking and

data collection Br
Hit rate~1-5%
•  Fragment library (1500-‐2000 cpds) is divided into pools of 10 shape diverse
compoundsàsoaked into preformed crystals
•  Bound fragment is iden@fied from shape of electron density features
•  En@re library can be screened within a few days at a synchrotron
Fragment Library Design/Diversity/U@lity
R
O
•  ~2000 fragments (scaffolds)
N
•  Scaffolds have ~3 Handles
N H
Br H amenable to modifica@on
Linear Library 1 (LL1) •  1000s-‐10000s of commercial
Handle
H3C R groups/Handle
1 (H1) O
N •  Poten@al Diversity Enormous
N H
Br H (H3)
–  2x103*10003à2x1012
(H2)
LL3
–  2x103*100003à2x1015
LL2
H3C H3C CHFNOcpds(<400Da)~1030
O
N N
O
•  Fragment Libraries support rapid
N H N hit SAR explora@on with
R H Br H R automated synthesis
FBDD Generates Mul@ple Opportuni@es
Multiple Targets Multiple Binding Sites Multiple Chemotypes
Active
Site
Allosteric
Site
MEK1
Wealth of Opportunities!
MET Receptor Tyrosine Kinase Inhibitor
•  MET (HGFR) is the receptor for V13L
hepatocyte growth factor/scaWer E168D Extracellular
factor I316M Domain
•  Ac@va@ng muta@ons occur in N375S
S I638L
various human cancers

–  Hereditary Papillary Renal Cell
Carcinoma (HPRCC) V941L
–  Sporadic, e.g., PRCC, non-‐small cell

lung cancer (NSCLC)
R988C
–  Ac@va@ng MET muta@ons have been P1009S
S1058P
observed in metastases V1110I
H1112Y
H1124D
•  MET gene amplifica@on seen in M1149T
V1206L
other tumors
L1213V
V1238I
D1246N
–  Gastric cancer Kinase Y1248C

K1262R
M1268T
–  EGFR inhibitor-‐resistant lung cancer Domain

Gastric Cancer
Lung Cancer
Kidney Cancer
Fragment Hit Evalua@on/Lead Op@miza@on
N N
R N
R
N
N N
MET IC50 (nM) 156 114
GTL16 IC50 (nM) 558 179
MW 343 337
HO
N N
MET IC50 = 712 nM cLogP 3.5 3.6
N
N
GTL16 IC50 = 1.37 µM
MW = 302; LEAN = 0.28
cLogP 3.0; LLE = 3.1 LEAN 0.27 0.27
Good ligand efficiency AND LLE 3.3 3.3

Novel binding mode
Achieving LEAN>0.27/LLE>5
N R
N
H
N N N
N N
R O N
F
MET IC50 (nM) 25 61
GTL16 IC50 (nM) 14 151
MW 412 341
Exploring the Cleft cLogP 2.5 1.5
Goal: Reduce cLogP LEAN 0.25 0.28
LLE 5.1 5.7

Design: Phenyl variants
and Heterocycles Non-Confidential Presentation
20
Reducing Clearance (<10mL/min/kg)
N
N
Benzylic center known N
N
N X
site of oxidation N
N
GTL16 In vitro Cl T1/2

Compound X
IC50 (nM) (mL/min/kg) (hr)
CH2 151 High No Data
CF2 20 5.1 2.2
CH(CH3) 38 11 1.7
SGX523 S 23 7.2 2.7
SGX523: In-‐vitro Profiling
Enzyme Assay: IC50 SGX523
MET Cytoplasmic portion 4 nM
Cell Assay: IC50

GTL16 cell proliferation (XTT) 24 nM
BaF3/TPR-MET (pMET ELISA) 17 nM
Cell assay specificity control > 10,000 nM
Liver Microsome Metabolism: T1/2
Rat 41 min
Human 279 min
CYP Inhibition (IC50)
1A2 > 10 µM
2C9 10 µM
2C19 > 10 µM
2D6 > 10 µM
3A4 > 10 µM
hPXR 3A4 induction > 10 µM
Safety Pharmacology
hERG > 10 µM
AMES/Micronucleus Neg./Neg.
SGX523: Unique Binding Mode
•  RON is MET’s nearest neighbor

•  Active site: 1 residue difference Y1248L
•  RON Inhibition [1µM] = 9%
SGX523: Human Kinome Profiling I
% Inhibition
Kinase Tested IC50 (nM)
%Inhibition @ 1µM for 213 kinases (1 µM SGX523)
MET 92 4
<50% Inhibition BRAF V599E 36 >66,667
RAF1 (cRAF) 27 >66,667
>50% Inhibition ABL 26 >7,407
•  MET and MET M1250T MAPK14 (p38α) 25 >200,000
ABL1 Y253F 20 >7,407
B a r C h a rt
K ina s e T e s te d
SGX523: Human Kinome Profiling II
•  Active site
residues in
common
Non-Confidential
25 Presentation 25
SGX523: MET versus MER
•  MET (blue) inhibited by SGX523 (yellow)
•  Homology model of MER (magenta)
•  MET-‐AlaàMER-‐Ser blocks binding
SGX523: MET versus AXL
•  MET (blue) inhibited by SGX523 (yellow)
•  Homology model of AXL (green)
•  MET-‐AlaàMER-‐Ser and MET-‐LeuàAXL-‐Met block binding
SGX523: In vivo Efficacy
1000
900
Average Tumor Vol (mm3)

Active in two mouse models 800
GTL16
700
•  Human gastric carcinoma cell 600
xenograft (GTL16 cell implant) 500
400
•  Activated MET xenograft (TPR-MET 300
200
BaF3 cell implant)
100
Inhibition of pharmacodynamic marker 0

0 2 4 6 8 10
(pMET) 0 2 4 6 8 10 12 14 16 18 20
Time (days post tumor implant)
Time (days post implant)
Average Tumor Volume (mm3)
500
450
X
400 TPR-MET Vehicle ctrl
X
30 mg/kg bid
350 10 mg/kg bid 60 mg/kg qd
300 20 mg/kg bid 100 mg/kg bid
250
200
150 SGX523 PO 6 Hours (GTL16)
100
50
Vehicle
0
6 Hours 10 mg/kg 20 mg/kg 30
30 mg/kg 100 mg/kg
mg/kg 60 mg/kg 100 mg/kg
0 3 5 6 8 10 12
0 2 4 6 8 10 12
Time (days post tumor implant)
Time (days post implant) pMET
SGX523: Phase I Clinical Outcome
•  Lowest dose group tolerated 40 mg per day

•  >80 mg dosing
–  Rapid onset of renal failure
–  Dosing suspended
–  Kidney func@on in affected pa@ents restored

•  Obstruc@ve crystal nephropathy observed in monkeys
Representative photomicrograph of renal histology: Crystals observed

SGX MET Inhibitor Summary
•  TZP LeadsàLEAN>0.27, LLE>5
•  SGX523
–  MET IC50 = 4 nM
–  > 1000-‐fold selec@vity for MET
versus 211 human protein kinases
–  Human/primate metabolite à
crystals in distal tubule à
obstruc@ve nephropathy
•  Other SGX TZPs à Lilly
•  TZP variant à Drug Candidate

Useful References!
–  Cavalli et al. (2002) J. Med. Chem. 45, 3844
–  Fink et al. (2005) Angew. Chem. Int. Ed. 44. 1504
–  Gleeson (2008) J. Med. Chem. 51, 817
–  Henkel et al. (1999) Angew. Chem. Int. Ed. 38, 643
–  Hopkins et al. (2004) Drug Disc.Today 9, 430
–  Hughes et al. (2008) BMCL 18, 4872
–  Johnson et al. (2009) BMCL 19, 5560
–  Kelder et al. (1999) Pharm. Res. 16, 1514
–  Kola and Landis (2004) Nature Rev./Drug Disc. 3, 711
–  Leeson and Springthorp (2007) Nature Rev./Drug Disc. 6, 881
–  Leeson and St-‐Gallay (2011) Nature Rev./Drug Disc. 10, 749
–  Lipinski et al. (1997) Adv. Drug Del. Rev. 23, 3-‐25
–  Lovering et al. (2009) J. Med. Chem. 52, 6752
–  Paolini et al. (2006) Nature Biotech. 24, 805
–  Ploemen et al. (2004) Exp. Toxic Pathol. 55, 347
–  Price et al. (2009) Expert Opin. Drug Metabol. Toxicol. 5, 921
–  Varma et al. (2010) J. Med. Chem. 53, 1098
–  Waring (2009) BMCL 19, 2844
–  Waring (2010) Expert Opin. Drug Discov. 5, 235
–  Waring et al. (2006) BMCL 17, 1759
–  Wenlock et al. (2003) J. Med. Chem. 46, 1250


Week 3a PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Week 3a PDF

Uploaded by

Copyright:

Available Formats

Challenges

in Fragment Based

• Presenta@on Goals and Background Informa@on

• Useful Concepts/Tools for Drug Discovery Teams

• Introduc@on to Fragment-­‐Based Drug Discovery

• FBDD/SBDD Case Study: MET Inhibitor

• Understand some of the challenges in small-­‐

• Understand some of the advantages oﬀered by

• Appreciate the possibility of unforeseeable

Molecular Biochemical Cellular In Vivo

Fragment Target Directed Phenotypic

• HT Crystallography • Gene family platforms • High Content Imaging • SCORE

How can MedChemists help address ~1/3 of Pharma Attrition?

Kola and Landis(2004)

# CNOF-containing Cpds with MW<500 ~ 1060

Subset with Pharmacologic Activity ~ <1026

Subset with Good ADMET Properties ~ 10??

– Ligand displacement of par@ally ordered waters from a

– Lipophilicity does contribute to target binding

– Too much lipophilicityàUnwanted oﬀ-­‐target binding

• Median cLogP versus Promiscuity 2,133 Cpds in 200 Cerep

• Promiscuity = # Cpds with >30%

Leeson and Springthorp (2007)

LLE: Potency without too much Grease!

LLE= – Log(IC50) – cLogP

• Must consider the eﬃciency of each lipophilic addi@on

• LLE allows iden@ﬁca@on/tracking of “good” Potency

• Lower MW Cpds have superior Pr(oral agent approval)!

LEAN: Drive for Potency at the Right Price!

LEAN= – log(IC50)/n [n=# of non H atoms]

nM Drug Candidate • Are fragments selective per se?

CandidateAC 30/400Da 3nM -­‐11.4 -­‐15 to -­‐16

Crystal soaking and

Multiple Targets Multiple Binding Sites Multiple Chemotypes

various human cancers

– Sporadic, e.g., PRCC, non-­‐small cell

– Gastric cancer Kinase Y1248C

– EGFR inhibitor-­‐resistant lung cancer Domain

MET IC50 (nM) 156 114

GTL16 IC50 (nM) 558 179

Good ligand efficiency AND LLE 3.3 3.3

MET IC50 (nM) 25 61

GTL16 IC50 (nM) 14 151

Exploring the Cleft cLogP 2.5 1.5

Goal: Reduce cLogP LEAN 0.25 0.28

LLE 5.1 5.7

GTL16 In vitro Cl T1/2

CH2 151 High No Data

CF2 20 5.1 2.2

SGX523 S 23 7.2 2.7

MET Cytoplasmic portion 4 nM

Cell Assay: IC50

• RON is MET’s nearest neighbor

Average Tumor Vol (mm3)

Inhibition of pharmacodynamic marker 0

• Lowest dose group tolerated 40 mg per day

Representative photomicrograph of renal histology: Crystals observed

• Other SGX TZPs à Lilly

• TZP variant à Drug Candidate

You might also like

•  Presenta@on Goals and Background Informa@on

•  Useful Concepts/Tools for Drug Discovery Teams

•  Introduc@on to Fragment-‐Based Drug Discovery

•  FBDD/SBDD Case Study: MET Inhibitor

•  Understand some of the challenges in small-‐

•  Understand some of the advantages oﬀered by

•  Appreciate the possibility of unforeseeable

•  HT Crystallography •  Gene family platforms •  High Content Imaging •  SCORE

–  Ligand displacement of par@ally ordered waters from a

–  Lipophilicity does contribute to target binding

–  Too much lipophilicityàUnwanted oﬀ-‐target binding

•  Median cLogP versus Promiscuity 2,133 Cpds in 200 Cerep

•  Promiscuity = # Cpds with >30%

•  Must consider the eﬃciency of each lipophilic addi@on

•  LLE allows iden@ﬁca@on/tracking of “good” Potency

•  Lower MW Cpds have superior Pr(oral agent approval)!

nM Drug Candidate •  Are fragments selective per se?

CandidateAC 30/400Da 3nM -‐11.4 -‐15 to -‐16

–  Sporadic, e.g., PRCC, non-‐small cell

–  Gastric cancer Kinase Y1248C

–  EGFR inhibitor-‐resistant lung cancer Domain

•  RON is MET’s nearest neighbor

•  Lowest dose group tolerated 40 mg per day

•  Other SGX TZPs à Lilly

•  TZP variant à Drug Candidate