An Integrated Approach To Drug Discovery Using Parallel Synthesis

An integrated approach to drug discovery using parallel synthesis Graham F Smith Sandwich Chemistry Pfizer Global Research and Development

Overview Enabled Hit Follow Up & Productive lead Discovery File Enrichment Chemo informatics High Throughput Chemistry HTS

File Enrichment – our early thoughts Drug discovery fails more often than it succeeds Some very difficult drug targets Fewer targets than before Some targets in the past, the industry has succeeded but Pfizer hasn’t One possible cause Industry file said to be approx 3,000,000 in 1999 Pfizer’s was approx 500,000 Narrow range of structural types George M Milne Jr (Pfizer), Annual Reports in Medicinal Chemistry, 2003, 38, 383-396

Rationale for file enrichment 70% of PGRD project leads originate as HTS hits Quality of drug candidate is reflected in quality of the lead matter One significant source of Discovery and Development attrition is chemical in nature Screening file attributes will drive downstream processes, efficiency and overall success in Medicinal Chemistry Taking attrition early is good BUT suffering attrition on approaches early can be costly, Therefore having more series options critical Testing the mechanism is the key Gunther Wess (Aventis), Drug Discovery Today, 7 (10), 2002, 533-535.

File enrichment strategy: consider attrition from outset We know what chemotypes are more likely to fail in development We know clinical candidates are similar to leads Build this knowledge into library design make and screen drug-like or lead-like compounds hit lead candidate drug

Beautiful compound concept parallel chemistry no toxicophores rule of 5 compliant pure and stable Pfizer exclusive Lipinski, C.A. Chris Lipinski Discusses Life and Chemistry after the Rule of Five. Drug Discovery Today 2003 , 8 , 12-16.

File Enrichment Initiative Aim to make a 3,000,000 compound pure compound file by compound design and acquisition Hits enabled by having established parallel chemistry protocols The largest screening file of any pharmaceutical company More hits on difficult targets – better choices on easier targets Faster / cheaper lead discovery in Pfizer enabled for the long term

File Enrichment 1 – “the big four” Neurogen Informatics based design and synthesis technology Predicts the next // chemistry compound to make ArQule Broad experience of high throughput parallel chemistry Delivering ~500,000 compounds and associated protocols Evotec Ultra High Throughput Screening uHTS Aurora Ultra High Throughput Screening uHTS

The Evotec ultra high throughput screening system Capacity to analyse >100,000 compounds per day Assay volume of 1  l - saving in reagent and compound usage 2080 wells on a single assay plate

File Enrichment 2002-2006 ArQule 800,000 Developing ~200 new chemistry protocols Tripos 440,000 compounds 12,000 privileged monomers Chembridge 500,000 compounds 4,000 privileged monomers Chem RX (Discovery partners ) 500,000 compounds Evotec Technology Developed in collaboration Created uHTS centres of emphasis 1 mergers and 1 acquisition with associated legacy compound file acquisition Warner Lambert Pharmacia Alan Procter (Pfizer), Drug Discovery and Development, December, 2003 See various collaboration press releases ~$1 Billion FE investment

File Enrichment – managing diversity La Jolla Groton Ann Arbor Sandwich Generation of ideas for library protocols Global Ideas overlap resolution La Jolla Groton Ann Arbor Sandwich Prioritization of ideas for development Library chemistry development Library idea Management Tool La Jolla Library production and purification

Hit Follow Up (2006…) A strategic modification of File Enrichment contracts Hit Follow-up – (HF) Med Chem Support Emphasis on Pfizer design Pfizer monomers and protocols Custom synthesis of key monomers and templates Development and optimisation of // chemistry Fast turnaround a contractual priority Compound supplied enough for project screening and Pfizer global HTS activities Supported by local Pfizer Parallel Chemistry Core expertise

Enhancing Chemo-informatics Enable designs Monomers Protocols Chemical Knowledge Enumeration Filtering Prioritise lead series Share Designs Visualise large libraries Use common desktop tools which aid collaboration

The Pfizer Global Virtual Library - PGVL All resources connected in one application All Pfizer Monomers stored in one location All Pfizer // synthesis reactions and protocols Stored in one location With enumeration and structure checking With knowledge of prior parallel chemistry reactions All Pfizer in silico properties available Filtering, clustering tools and visualisation All parallel chemistry HTS Hit Follow up work comes from within this tool A research sharing and visualisation tool Links to Excel, Spotfire etc. Enables choice from ~10,000,000,000,000 potential products This virtual space grows with each new chemistry protocol and monomer discovered PGVL ~ (PGRL) 2

HTS Hit triage protocol – Pipeline Pilot Raw data Compound Name % inhibition Structures Physical properties (Mw, clogP, etc) Potential false positives/negatives (Bayesian statistics) Library protocol if any Custom substructures (known series etc) Clustering of actives by Murcko frame Known aggregators, reactive groups Ligand efficiency M Murcko (Vertex), J. Med. Chem. 1999, 42, 5095-5099 A L Hopkins (Pfizer), Drug Discovery Today. 2004 May 15;9(10):430-1

Pipeline Pilot tree HTS Actives HTS Hits

HTS library output multiple lead options for most targets

Global Protocol database intranet access All Parallel Chemistry searchable in one place All 9 PGRD site use this tool All File enrichment research captured in the global database A link from compound to parallel synthesis protocol

Architecture based on Formal Reactions Protocols are associated into Formal Reactions VRXN-2-00001 Reductive amination as a family of protocols aldehydes nucleophilic amines Two protocols of broadest scope cover the largest range of potential ---- thus potential for re-use: Capturing these is what really builds the asset in global Chemical Knowledge Bases VRXN-2-00001 is defined by the Formal Reaction scheme and spans the space covered by all suitable amines & aldehydes in the linked inventory db

Formal Reactions as Families of Protocols 21 VRXN-3-00029 4 VRXN-3-00007 2 VRXN-2-00081 3 VRXN-2-00018 7 VRXN-2-00017 9 VRXN-2-00016 8 VRXN-2-00013 15 VRXN-2-00011 39 VRXN-2-00010 12 VRXN-2-00009 9 VRXN-2-00007 16 VRXN-2-00005 4 VRXN-2-00004 11 VRXN-2-00002 13 VRXN-2-00001 Source (each color represents one of the Pfizer R&D sites) # of explicit protocols within reaction family Name of Formal Reaction (broad scope rxns)

Chemist’s Desktop Access to 543 (and growing) Formal Reactions Scope & Limitations for the thousands of Protocols under those Formal Reactions is scientific basis for Round-the-clock automated monomer mining PGVL server-side automated search engine system uses thousands of stored query components running on Pipeline Pilot Linked to global inventory Search engine defines the precise VL for each protocol and (by concatenation) the overall VL for each Formal Reaction PGVL exceeds 10 trillion compounds 10 13 + PGVL Hub, an internally developed application

PGVL Hub on the Chemist’s Desktop

Access to virtual libraries from the desktop Suitable monomers retrieved for each reaction protocol Properties calculated Inventory and location searched Clustering and filtering pre and post enumeration

HTS at Pfizer 10 years ago A 500,000 screening file Of mixed quality and purity Early days of parallel chemistry and automation in chemistry and purification Synthesis capacity > purification capacity > chemo informatics ability Typical HTS gave 10% of lead matter being parallel chemistry friendly Crude informatics tools for HTS follow up Mostly Excel The parallel chemistry friendly hits given more effort and therefore typically start more projects Singleton hits - higher chemistry cost and risk Parallel chemistry hits already have SAR, starting materials and high speed method

HTS at Pfizer now uHTS capacity with Evotec and other platforms Automated triage process based on Clustering of actives and analysis of near neighbours Hot monomer analysis Parallel chemistry protocols In Silico predicted physicochemical and ADME properties Several libraries initiated on priority series Several series / chemotypes progressed to reduce risk of project attrition Monomer and template synthesis initiated where needed Protocol development where needed

HTS at Pfizer now – cont. Iterations of library design are synthesised to address series issues versus project criteria Designs include global and project specific in silico models

Measures of success: Sandwich 96% of projects using parallel synthesis 138 libraries of >96 compounds each across the Sandwich discovery portfolio in 2003 Can identify multiple series to pursue from most HTS Several leads from HTS & parallel synthesis currently in lead to CAN phase and early development 33% of clinical candidates resulted from FE enabled hits applying multiple libraries (including structure based drug design) & traditional medicinal chemistry synthesis

IKK2 inhibitors from HTS hits to multiple project leads Weak HTS hits from early FE library 2 active custom templates 5.6 billion potential analogues 4880 library inactives in HTS 6 rounds of library synthesis Potent IKK2 inhibition in acidic, basic and neutral series found

The Hardware – 1st generation 6 x 2M Tecan Genesis 8 x Waters LCMS+ELSD 2 x Bohdan Weighing Stations 12 x Gilson 215 based HPLC systems

Assorted plate types continued

Tecan Genesis with Argon Blanketing

Prep HPLC stacking to save space

Barcode based fraction tracking

Crystal Structure of UK-411,930 bound to CDK2 - SBLD 2.3A resolution data SRS, Daresbury. Key pyrazole Donor Acceptor Donor interaction

Overview of Closed-loops for IKK Activity goes up with library iteration Number of actives increases with library iteration 0 10 20 30 40 50 60 0 1 2 3 4 5&6 Closed-loop iteration % of compounds active at 10  M 1 10 100 1000 10000 activity of most potent compound (nM)

Novel low MWt PDE-5 inhibitors from closed loop chemistry on HTS hits Closed loop (CL) involves iterative design, synthesis, purification, sample logistics and screening loops Combining singleton and exploratory library synthesis demonstrated that series was not flawed Closed-loop synthesis identified 2-aminopyridine as novel PDE5 pharmacophore 2 closed loops from CP-X (HTS hit) eliminated toxicophore, reduced MWt by over 200 Da, raised ligand efficiency, good selectivity over 1, 9 and 6 found 5 other active templates in addition to UK-C Aim: novel potent, selective, low clogP templates for o.d. PDE-5 i

Structure Based Library Design Templates Favoured monomers e.g. 2-aminopyridines Pendant ionisable centre for solubility and SAR probe Library design incorporated the positive structural filters Negative filters of cLogP of 4 and MW of 400 were applied 264/576 targeted compounds made.

PDE5 Library lead: UK-469,413 12-24 hours 18, ER<1 1.9, 3.3 367 32v6, 5v10, 5v11 71nM UK-469413 >12 hours Pred. Hu T1/2 >10, ER<1 Caco abs. 1-2, <3 LogD, LogP <400 MWt >10x vs PDE’s Selectivity  50nM PDE5 IC50 Lead Target Criteria

PDE5 2 nd generation series UK-469,413 Gln775 close to C7- aminopyridine methyl Gln817 H- bonds to C7- aminopyridine Mike Palmer et al , Current Topics in Medicinal Chemistry, 2006, in press.

Profiles of selective PDE5 leads Criteria PDE5 IC50 PDE6 Sel. PDE11 Caco2 TPSA N,O count Lead 2 0.80nM 42x 10x 18, ER 1 93 9 Lead 3 0.50nM 104x 160x 2, ER>10 122 11 Great pharmacology, but… efflux problems.

Oxytocin hit–to-lead chemistry HTS hits from FE project Open ring by product of Ugi Benzdiazepinedione library 2-Aminonicotinic acids yield non-cyclised materials Library synthesis yielded 284 compounds Most potent compounds confirmed as functional antagonists Keating, T.A. & Armstrong, R.W. (1995) "Molecular diversity via a convertible isocyanide in the Ugi four-component condensation," J. Am. Chem. Soc. 117 : 7842-7843

File enrichment library issue Cyclisation of anthranilic acid derivatives formed using Ugi chemistry 1 provided targeted Benzdiazepinedione Use of 2-aminonicotinic acid yielded non-cyclised by-products R = Ph 2 avoids use of volatile, smelly isonitrile and yields 4-phenylcyclohexanone, readily separable from products by HPLC. 1) Ugi et al. Chem. Ber., 1961, 94, 2802, Armstrong et al. J. Am. Chem. Soc. 1996, 118 , 2574 2) Martens et al. Bio-organic & Medicinal Chemistry 2000, 8, 1343

Resolution of Oxytocin library actives Resolution of library actives provided compounds meeting the target potency and selectivity criteria Series accepted as leads for further development.

Long Acting Beta 2 agonists Beta2 0.74 nM EC50 HTS hit Parallel synthesis friendly project strategy Enabled rapid progress of the project Partnering TA with PMC group Protected head groups Optimised for potency and duration Salmeterol like potency and DOA found Exciting potential for long DOA in humans

mGluR1 antagonist A number of weak and simple HTS hits identified Both side of amide were optimised to give some potent amide analogues Amides were high logD and high CL Chose to diversify chemistry into PGVL space 6000 FE analogues were available for screening The available VL was still huge 4.2 Billion products

mGluR1 Leads optimised within PGVL Within a larger VL the logD and Cl properties could be better optimised Several 1000s of analogues were easily made in the program Progress was rapid and used outsourcing at Tripos, Bude, UK Compounds with good potency and low Cl were obtained Project identified tools for further development Dafydd Owen* et al , Bioorganic Medicinal Chemistry Letter, 2006 In Press.

Maraviroc - Faster to CAN with parallel synthesis UK-107,543 (HTS hit) 400 nM CCR5 blocker no antiviral activity CYP450 inhibitor UK-427,857 antiviral IC 90 1 nM selective no CYP450 inhibition jun00 dec97 Dorr et al , Antimicrob Agents Chemother ., 2005 , 49 , 11 Armour & Wood, Progress in Medicinal Chemistry , 2005 , 43 , 239, WO2003084954 965 analogues

CCR5 antagonist - Maraviroc Most final products made by HSA Maraviroc first made in a filter tube with solid phase reagent Blood rotator used for agitation Crude products purified by prep HPLC OA FractionLynx

NEP Parallel Chemistry Strategy Initially retain a small R1(S1‘) based on known Candoxatrilat SAR start with nPr and start racemic for expediency Rapid exploration of R2 using parallel chemistry (1ml 96wp) diverse set of amines based on Desired drug properties Crude products sent for prep HPLC purification As SAR emerges, incorporate chiral and diverse R1 S1'

NEP inhibitor SAR Successful and rapid SAR exploration UK-447,841 discovered Short half life prn dose regimen required T1/2 7-10h High absorption (F) David C. Pryde et al , J. Med. Chem. 2005 WO2002002513 R2 dNEP (nM) MWt. LogD Caco-2 24 339 0.5 9/14 R1 6 393 -0.2 - 100 375 1.0 6/10 n Pr Me (CH 2 ) 2 OMe (CH 2 ) 2 OMe (CH 2 ) 2 OMe (CH 2 ) 2 OMe 4 333 0.8 21/21 3 379 0.5 12/17 0.5 396 1.0 12/14

Conclusions HTS hit follow up is supported by huge global resources FE created a 3MM compound pure parallel chemistry rich file Global Chemo informatics resources able to quickly mine 1 trillion potential molecules Contract research to support early stage lead optimisation and SAR generation

Additional Slides Pfizer Overview Industry pressures

Pfizer Overview World’s leading healthcare company Biggest investor in R&D worldwide 2004 R&D investment: $7.5 billion The largest foreign owned R&D investor in the UK 122,000 employees across 60 sites in 31 countries sales in over 150 countries: 12,000 R&D employees worldwide 2,700 people in R&D in UK 6900 Pfizer UK employees Biggest supplier of medicines to the NHS

INDUSTRY AVERAGES: 100 Screens Produce 12 Candidates 1 in 12 survival rate during development Only 1 of 4 products shows significant profit 12 years from idea to market 12-13 years of market exclusivity Up to $800 million investment per product Pre-Clin Phase I Phase II Phase III Registration MARKET Animal Toxicity, Chemical Stability, Superior Compound Human PK, Toleration, Formulation Efficacy, Safety Differentiation , Dose Long-term Safety Approval?

The Attrition Curve 1 2 3 4 5 >12 years 60% 25% Ideas  Leads  Dev. Candidates  PhIIa Clin.  Products 25% 100 Projects 1 Product % Projects 25%

An Integrated Approach To Drug Discovery Using Parallel Synthesis

More Related Content

What's hot (20)

Viewers also liked (20)

Similar to An Integrated Approach To Drug Discovery Using Parallel Synthesis (20)

More from Graham Smith (6)

Recently uploaded (20)

An Integrated Approach To Drug Discovery Using Parallel Synthesis