Drug Discovery

1
Drug Discovery GPCR Models
Sheila DeWitt, PhDVP Discovery
ManufacturingOctober 25, 2007
2
Outline

• Overview of EPIX Product Portfolio
• Drug Discovery Strategy
• Case Study 5HT1A

3
Clinical Portfolio Internally Discovered
Three Drug Candidates in Phase 2 Development
Phase 3
NDA
Approved
Phase 2
Phase I
IND/GLP Tox
Lead Optimization
Lead Discovery
Target
Product
PRX-08066
(5-HT2B)
Pulmonary Hypertension w/ COPD
Depression
PRX-00023
(5-HT1A)
Alzheimer's Disease (GSK has exclusive option)
PRX-03140
(5-HT4)
Obesity, Cognitive Impairment
PRX-07034
(5-HT6)
COPD Chronic Obstructive Pulmonary Disease
4
Proprietary Drug Discovery Technology

• GPCRs Strategic drug development targets
• Embedded proteins in surface membrane of all
  cells
• Mediate biological signaling in health/disease
• Commercially validated - 40 of top 100 drugs
• Never crystallized 3D Structures Unknown
• SAR a hit-or-miss exercise requiring years
• Side effect / selectivity issues remain
  problematic
• Opportunity for EPIX
• Proprietary modeling / screening technologies
• Commitment to discovery triad
• Computational and medicinal chemistry integrated
  with biology

5
Outline

• Overview of EPIX Product Portfolio
• Drug Discovery Strategy
• Case Study 5HT1A

6
EPIX Discovery Strategy
Drug Discovery
Screening
Hit Characterization
Lead Optimization
Preclinical Development
Model Development

• Modeling Novel GPCR modeling methodology
  (PREDICT)
• Screening in silico screening gt 4 Mil
  commercially available cmpds
• Hit Charact 3D Models Purchased SAR (pSAR) to
  prioritize scaffolds
• Lead Opt 3D Models, Biology, and Med Chem to
  optimize

7
EPIX Discovery Strategy
Drug Discovery
Screening
Hit Characterization
Lead Optimization
Preclinical Development
Model Development

• Modeling Novel GPCR modeling methodology
  (PREDICT)
• Screening in silico screening gt 4 Mil
  commercially available cmpds
• Hit Charact 3D Models Purchased SAR (pSAR) to
  prioritize scaffolds
• Lead Opt 3D Models, Biology, and Med Chem to
  optimize

8
Modeling GPCRs with PREDICT

• Unique de novo GPCR structure prediction
  algorithm
• Based on scientific understanding of GPCR folds
• from experiments, simulations and theory
• Folds the protein within its membrane environment
• Does not rely on rhodopsin x-ray structure
• Does not use homology modeling
• Applicable (in principle) to any GPCR

9
G-Protein Coupled Receptors (GPCRs)
X-ray structure of Rhodopsin

• Seven transmembrane a-helices
• Alternating cytoplasmic and extracellular loops
• N-terminus is extracellular
• C-terminus is cytoplasmic
• Role is to transduce extracellular response via
  activation of hetero-trimeric G-proteins

10
PREDICT Modeling Process
GPCR sequence
11
PREDICTTM Step I - Build 7 TMs

• Represent each helix by a 2D dial
• Generate all closed 2D configurations of 7 dials
• under geometrical constraints
• Optimize each 2D configuration
• to maximize hydrophobic moment in the direction
  of the membrane (introduce experimental
  constraints)

Binding pocket
12
PREDICTTM Step II Translate 2D to 3D

• Extend each optimized 2D configuration into a 3D
  representation and optimize in 3D

13
EPIX Discovery Strategy
Drug Discovery
Screening
Hit Characterization
Lead Optimization
Preclinical Development
Model Development

• Modeling Novel GPCR modeling methodology
  (PREDICT)
• Screening in silico screening gt 4 Mil
  commercially available cmpds
• Hit Charact 3D Models Purchased SAR (pSAR) to
  prioritize scaffolds
• Lead Opt 3D Models, Biology, and Med Chem to
  optimize

14
EPIX in silico Screening Process
Data collection
Target modeling
Library generation
Selection of virtual hits
15
EPIX Screening Libraries

• Size 4 million drug-like compounds
• Source Catalogues of 30 reputable vendors
• Updates Continuously (before new projects)
• Criteria Availability for immediate purchase
• Advantages
• Diverse
• Rapid access to newest compounds (30 change per
  year)
• Cheap to obtain and to maintain
• Quick registration (buy only what is actually
  needed)
• Limitations
• Non-standard targets may not be represented well
• Need to improve IP properties since hits will be
  in public domain

16
in silico Screening Hit Characterization

1. Datamine collection of gt4 Mil commercially
   available cmpds
2. Select focused cmpd library for target
   (100,000400,000)
3. In silico screening of focused library against
   target protein
4. Scoring selection of prioritized cmpds (200-300
   virtual hits)
5. Purchase and test virtual hits in biological
   assay
6. Hit criteria Ki/IC50 lt 10mM (validated dose
   response)
7. Datamine around hits to generate pSAR
8. Prioritize scaffold for Lead Optimization
9. Further optimize model for specific scaffold
   using pSAR

17
EPIX Discovery Strategy
Drug Discovery
Screening
Hit Characterization
Lead Optimization
Preclinical Development
Model Development

• Modeling Novel GPCR modeling methodology
  (PREDICT)
• Screening in silico screening gt 4 Mil
  commercially available cmpds
• Hit Charact 3D Models Purchased SAR (pSAR) to
  prioritize scaffolds
• Lead Opt 3D Models, Biology, and Med Chem to
  optimize

18
EPIX Paradigm for Lead Optimization

• Integrated MedChemCompChem teams (21 ratio)
• Extensive use of computational tools (3D
  structures, predictive ADMET) to navigate the
  multiple possible optimization pathways
• Suggest/prioritize what to synthesize
• Suggest/prioritize what NOT to synthesize
• Efficient process, robust, agnostic to the
  receptor class

19
Efficient and Effective Discovery Engine
Industry Standards
EPIX
Hits in silico screen 4M compounds 6 months
Hits wet assay screen lt1M compounds 12 months
Lead Optimization 1,000 compounds 2-5 years to
clinical candidate
Lead Optimization 100 compounds or less 6-12
months to clinical candidate
20
Current Pipeline - in silico LO Track Record
(1) Conformational analysis (2) IC50 from
functionality assay (3) Pharma collaboration
(4) Pharmacophore screening (5) Ki estimated
from IC50
OM Becker et al, PNAS 101 (2004), 11304-11309
21
EPIX Lead Optimization Track Record
(1) Estimated
22
Outline

• Overview of EPIX Product Portfolio
• Drug Discovery Strategy
• Case Study 5HT1A

23
PRX-00023 Depression

• 5-HT1A partial agonist, proven mechanism of
  action
• Estimated world market for treatments 20
  billion
• 35M in US (more than 16 of the population)
  suffer from depression severe enough to warrant
  treatment at some time in their lives
• Substantial commercial opportunity for a
  selective, better tolerated alternative
• No withdrawal symptoms, sexual dysfunction,
  weight changes or sleep disturbances as observed
  with SSRIs
• Lacks the addictive and sedative effects of the
  benzodiazepines
• Does not have side effects of azapirones
• Initiated Phase 2b trial March 2007 results
  expected 1H08
• Achieved significant results on depression in
  Phase 3 anxiety clinical trial

• Source National Institute of Mental Health, 2003
  National Comorbidity Study, Sponsored by the
  National Institutes of Health

24
Mechanisms of Other Drug Classes

• SSRI / SNRI
• Mechanism results in increased levels of
  serotonin (5-HT), norepinephrine (NE)
• Affects 5-HT (14), NE (gt6) receptors
• Affects sleep, sexual function, appetite
• Withdrawal symptoms
• Black box warning

• Azapirones
• 5-HT1A agonists
• Affinity for off-target GPCRs
• Dopamine D2, alpha-1, alpha-2
• Nausea, lightheadedness, headache, restlessness
• Slow dose escalation requirements

25
Mechanism of Action PRX-00023

• Potential advantages
• Highly selective for 5-HT1A
• No sexual dysfunction
• No effects on sleep or appetite
• No withdrawal symptoms
• Do not expect black box warning
• Well tolerated compared to azapirones, with
  minimal dose escalation required

PRX-00023
26
PRX-00023 Superior to other 5-HT1A Agonists

• Azapirones other 5-HT1A agonists have
  selectivity issues and metabolic liabilities
• PRX-00023
• Very high affinity for 5-HT1A (Ki 5nM)
• Better selectivity
• minimal binding to alpha-1 (Ki 1600nM), alpha-2
  (gt 3000nM) and dopamine D2 (Ki gt 2000nM)
  receptors compared to Azapirones
• Not metabolized to 1-(2-pyrimidimyl)-piperazine,
  a potent alpha2-adrenergic modulator
• Better selectivity results in superior
  tolerability and no need to a few weeks of
  multi-step dose titration
• Once daily dosing
• No significant inhibition of CYP450 or hERG
• Well tolerated in three Phase I and two Phase II
  clinical trials
• No significant nausea / lightheadedness vs.
  azapirones

27
PRX-00023 Phase 2b in Depression in Progress
Data in 1H08

• Double-blind, randomized, placebo-controlled dose
  clinical trial of PRX-00023 in major depressive
  disorder (MDD)
• 8-week study with 120mg twice daily flexible
  dosing
• Approximately 330 MDD patients
• Randomized 11 drug vs. placebo
• Primary endpoint
• Change from baseline in MADRS compared to placebo
  
• Secondary endpoints
• Changes in the Hamilton Anxiety Score (HAM-A)
• Clinical Global Impressions Improvement Scale
  (CGI-I)
• Clinical Global Severity of Illness Scale (CGI-S)

28
Outline

• Overview of EPIX Product Portfolio
• in silico Modeling Strategy
• Discovery Case Study 5HT1A