The Application of Non-Combinatorial Chemistry to Lead Discovery Dr Graham F. Smith Library Design and Production Group Pfizer Global R & D Sandwich, UK See Drug Discovery Today, 2001, volume 6, No. 15, p779-785

Parallel synthesis is a powerful technique The application of automation, parallelisation and miniaturization of Organic Chemistry has radically changed Chemistry Departments in Pharmaceutical companies. Combinatorial Chemistry is only one possible use of parallel synthesis

Our Groups Initial Goals To create a technology group which could make large numbers of HPLC pure single compounds To create systems and processes which allow single compounds from an array to be selected and made in parallel To combine the above with compound design at the individual compound level These are then combined to make a library To perform all of the above in a high throughput environment for file enrichment and later “lead optimisation”

To create a technology group which could make large numbers of HPLC pure single compounds

To create systems and processes which allow single compounds from an array to be selected and made in parallel

To combine the above with compound design at the individual compound level

These are then combined to make a library

To perform all of the above in a high throughput environment for file enrichment and later “lead optimisation”

Non Combinatorial Chemistry Parallel synthesis of single compounds Not all combinations of monomers No artificial constraints on array shape or dimensions (e.g 2000 from 200 X 150 x 30) Allows “cherry picking” of desired products from an array based on: diversity similarity calculated property e.g. Rule of 5 predicted activity gut feeling, crazy hunch, any of the above in combination….. Allows maximum diversity or similarity from a given number of compounds Uses all the locally available monomers

Parallel synthesis of single compounds

Not all combinations of monomers

No artificial constraints on array shape or dimensions (e.g 2000 from 200 X 150 x 30)

Allows “cherry picking” of desired products from an array based on:

diversity

similarity

calculated property e.g. Rule of 5

predicted activity

gut feeling, crazy hunch, any of the above in combination…..

Allows maximum diversity or similarity from a given number of compounds

Uses all the locally available monomers

Diversity File enrichment for hit seeking requires maximal diversity in drug space and limited redundancy populate with low diversity and small cluster sizes Hit follow up requires more similarity and redundancy populate with high diversity and large cluster sizes Need metrics which map chemical similarity to biological similarity

File enrichment for hit seeking requires maximal diversity in drug space and limited redundancy

populate with low diversity and small cluster sizes

Hit follow up requires more similarity and redundancy

populate with high diversity and large cluster sizes

Need metrics which map chemical similarity to biological similarity

Large clusters high similarity Cluster algorithm from Mike Miller - PGRD Groton USA Small clusters low similarity Spotfire Shenghua Shi

IT for Non Combinatorial Chemistry - LiCRA Li brary C reation R egistration and A nalysis Oracle database, batch based not array (plate) based MACCS database for structures VB interface Migration to Java and xml Q1 2002 Zhenwei Peng Inputs simple and general SDF file with batch numbers comma separated file with batch numbers & starting materials e.g. Batch#1,benzylamine#1,benzoic acid#1 synthesis plate format & monomer racks are created on the fly by analysis of library product composition Output = complete registration file to corporate DB structure, batch#, product weight, chemist, precursors etc. Stores, analyses and reports history of synthesis and QC data

Li brary C reation R egistration and A nalysis

Oracle database, batch based not array (plate) based

MACCS database for structures

VB interface

Migration to Java and xml Q1 2002

Zhenwei Peng

Inputs simple and general

SDF file with batch numbers

comma separated file with batch numbers & starting materials

e.g. Batch#1,benzylamine#1,benzoic acid#1

synthesis plate format & monomer racks are created on the fly by analysis of library product composition

Output = complete registration file to corporate DB

structure, batch#, product weight, chemist, precursors etc.

Stores, analyses and reports history of synthesis and QC data

LDP - Technology 96 well 1ml Teflon plates Robbins FlexChem solid phase filter plates Tecan Genesis 8 needle x 6 dispensing LLE SPE resin washing Hamilton SPE x 2 Bohdan tube weighing ( for products) x 3 Genevac (HT12 & series 2) x 6 Gilson 215 based HPLC x 16 UV fraction detection 9ml 48 Well plate bar coded collection plates

96 well 1ml Teflon plates

Robbins FlexChem solid phase filter plates

Tecan Genesis 8 needle x 6

dispensing

LLE

SPE

resin washing

Hamilton SPE x 2

Bohdan tube weighing ( for products) x 3

Genevac (HT12 & series 2) x 6

Gilson 215 based HPLC x 16

UV fraction detection

9ml 48 Well plate bar coded collection plates

Technology in use (2) MicroMass Platform MS or TOF x 6 4 way Mux recently added for pre QC ELSD (Polymer labs & Aztec) Beckman ORCA 2M track robot based fraction selector (Anachem ~$1MM) Molecular ion detection ELSD quantification standard at beginning and end of every plate multi detector rule based purity decision on the fly product plate reformatting (red, amber, green) on line progress monitoring from remote PC ~60,000 crude fraction samples in 2002 24 x 7 operation with maintenance (1 FTE)

MicroMass Platform MS or TOF x 6

4 way Mux recently added for pre QC

ELSD (Polymer labs & Aztec)

Beckman ORCA 2M track robot based fraction selector (Anachem ~$1MM)

Molecular ion detection

ELSD quantification

standard at beginning and end of every plate

multi detector rule based purity decision

on the fly product plate reformatting (red, amber, green)

on line progress monitoring from remote PC

~60,000 crude fraction samples in 2002

24 x 7 operation with maintenance (1 FTE)

Unforeseen advantages of Non Combinatorial Chemistry and HPLC library purification HPLC purification of plates ordered by clogP gives easier purification method development Large monomers sets define bigger and better protocols after several iterations Can take inputs from combinatorial and non combinatorial designs together Leads are easily followed by discovery chemistry Much larger monomer set = diversity weighing more monomers manually intensive inventory of more monomers needs resources (IT, $$ and FTE) Less reliance on the activity or chemical success of single monomers in a library

HPLC purification of plates ordered by clogP gives easier purification method development

Large monomers sets define bigger and better protocols after several iterations

Can take inputs from combinatorial and non combinatorial designs together

Leads are easily followed by discovery chemistry

Much larger monomer set = diversity

weighing more monomers manually intensive

inventory of more monomers needs resources (IT, $$ and FTE)

Less reliance on the activity or chemical success of single monomers in a library

Summary Non combinatorial chemistry is easily achieved within our LiCRA IT system High Throughput semi prep reverse phase HPLC is capable of >50,000 product samples per annum delivering ~6mg per sample Optimisation of library derived leads is more efficient with non combinatorial chemistry File enrichment is more efficiently achieved with non combinatorial chemistry Sparse matrix, cherry picked libraries produce HTS actives “ hit rate” greater than the average screening file no false positives due to “artefacts” or mixtures these actives have opportunity of efficient // synthesis follow up from a large chemically validated VL

Non combinatorial chemistry is easily achieved within our LiCRA IT system

High Throughput semi prep reverse phase HPLC is capable of >50,000 product samples per annum delivering ~6mg per sample

Optimisation of library derived leads is more efficient with non combinatorial chemistry

File enrichment is more efficiently achieved with non combinatorial chemistry

Sparse matrix, cherry picked libraries produce HTS actives

“ hit rate” greater than the average screening file

no false positives due to “artefacts” or mixtures

these actives have opportunity of efficient // synthesis follow up from a large chemically validated VL

Acknowledgements Mark Gardner Chris Selway Mike Stace Andrew Morrell Ian Parsons Jason Robson Derek Hearn Andrew Berridge Mark Lord Mike Snarey Frank Pullen Martin Edwards Jeremy Everett Nick Terrett Adrian Wright Andy Kemp (Aitken Scientific) Ian Hibbard (Anachem) Grant Cameron (Anachem)

Mark Gardner

Chris Selway

Mike Stace

Andrew Morrell

Ian Parsons

Jason Robson

Derek Hearn

Andrew Berridge

Mark Lord

Mike Snarey

Frank Pullen

Martin Edwards

Jeremy Everett

Nick Terrett

Adrian Wright

Andy Kemp (Aitken Scientific)

Ian Hibbard (Anachem)

Grant Cameron (Anachem)