Rapid Chemical Space Exploration –p p pApplications from Discovery to Manufacturing

Neal SachSenior Principal Scientist10777 Science Centre DriveLa Jolla, CA 92121, USAneal.sach@pfizer.com

ACS T h i l A hi t iACS Technical Achievements in Organic ChemistryPhiladelphia, USAAugust 22nd 2012August 22 2012

Contents

A I t d ti t R ti O ti i ti W kflAn Introduction to Reaction Optimization Workflows

Reaction Optimization in Process Chemistry

Adapting to Discovery Chemistry

Case Examples

Solving Chemistry Across the Portfolio

Discover Drugable Molecules S l D S b tSupply Drug SubstanceDevelop Commercial ProcessesTransfer Technology to Production

P P Supply ChainDiscovery ProcessResearch

ProcessDevelopment Manufacturing

Covalent Bonds Covalent BondsReaction Conditions

Covalent BondsReaction Conditions Reaction Conditions

How we do this in less time and at a lower cost?

Reaction Conditions Reaction ConditionsWork Up Conditions

Reaction ConditionsWork Up ConditionsVariations Variations

How we do this in less time and at a lower cost?

Adapting to Industry Pressures

Technology must ‘fill the gap’

W kl dWorkload

Resources

199X 20XX

The First Reaction Optimization Tool

In 1998 a Process Research Technology Group was created with the aim of speeding Process Developmentthe aim of speeding Process DevelopmentThe SK233 was the first tool of its kind, 10 simultaneous reactions with integrated analytics 1g per reaction, 10 reactions, 10-50ml volume

100

59

76

90 9296

60708090

100OH ON

LiOH (10eq.)

14

43

1020304050

N N

Cl

NH2 NH2

N N

Cl

NH NH

LiOH (10eq.)

MeOH / H2O

0 00 0 0 0 0 1 2 2 3010

0 5 10 15Time

NH2 NH2NH2 NH2

UK-427,387 – The Tipping Point

Despite a few successes, the general adoption of the p , g pSK233 and reaction optimization technologies was slow… …but that was to change…gUK-427,387 was nominated as a follow up PDE5 inhibitor to Sildenafil (Viagra)( g )

UK-427,387 Pyrazole Alkylation

25kg Required for Phase 1 Parallel Synthesis Robot (PSR)250 ti 50 ti250mg per reaction, 50 reactions, 5-25ml volume

NH

N

O

NH2N

PhN

N

O

NH2

Na2CO3(1.1eq.)NaI (1.1eq.)THF / Water(9:1@5ml/g)80C / 7 days

NO2N

N

OMs

Ph Ph

NPh

NO2N

Issues (where to start?!)30% overall yieldy2:1 undesired : desired pyrazole7 days reaction time, complex conditions (4 solids 2 solvents)conditions (4 solids, 2 solvents)5-14% undesired alkylated isolated

UK-427,387 Pyrazole Alkylation

25kg Required for Phase 1 New ConditionsNa2CO3(1.1eq.)

BeforeN

PhN

N

O

NH2

N

NN

ONH2

O2N

PhPh

NH

N

O

NH2

O2N OMsN

Ph

PhN

N

O

NH2

O2N

NaI (1.1eq.)THF / Water(9:1@5ml/g)80C / 7 daysN

H

N

O

NH2

O2N OMsN

Ph

PhN

N

O

NH2

O2N

LiOtBu / MeCN80C / 6hrs

NH

N

O

NH2

O2N

NPhO2N

1:2 N1:N22

N

Ph Ph

Ph22

N

Ph Ph

Ph2

AfterScreen120 reactions run over 3 weeksResults30:1 selectivity achieved

AfterN

Ph

PhN

N

O

NH2

O2N

30:1 N1:N212 bases x 10 solventsStrong base required Lithi t i d d

y12hrs vs. 6 days reaction timeScaled to 32kg, 71% isolated yield of 99% purity

30

Lithium count-ion neededMeCN solvent found to be key

99% purityQ. Would this obscure hit have been found in the short time frame allowed without reaction screening technology?

Development of High Throughput Experimentation (HTE) Workflows 2002-2005( )

StatisticalMethods

SolidsDispensing

DataAnalysis

Reaction Execution

Data Capture andReactionRational

y

Reaction FastExecutionand Analysis

Capture and VisualisationPlatformDesign

ReactionPlatform

Fast Conclusions

Chemists LiquidsDispensing

DataMining

5Higginson, Paul D.; Sach, Neal W. High-Throughput Experimentation in Pharmaceutical Process R&D: Developing a New Software Workflow to Overcome Downstream Data-Analysis Bottlenecks and Improve Productivity. Organic Process Research & Development (2004), (8), 1009-1014.

2002-2005 Investment in Technology

By 2005 the HTE group in Process had expanded to 5 FTE’s with state of the art equipment at their disposalwith state of the art equipment at their disposal25mg per reaction, 50 reactions, 1-2ml volume

One 2 Many Weighing iChemExplorer Reaction Platform 2-4 minute HPLC l ti

2mlReactionVolume

Fast HPLC High-throughput

cycle-time per sample

Many 2 Many Weighing Gaseous Chemistry

Fast HPLCAnalysis Platform

High-throughput analytics and generic methods are critical to

HTE Modular Workflows

Application of High Throughput Chemistry WorkflowsAcross the Portfolio

Discovery Process Supply ChainDiscovery Development Supply Chain

ApplicationsApplications

Solubility Profiling

Solid Form Identification

Reaction Screening Start

Resolution ScreeningResolution Screening

Example Workflow - Resolution Screening

Scalable access to chiral molecules via diastereoisomeric salt formationdiastereoisomeric salt formation NH2

NH

O

O

OHOH

O

OH

OH

O

OH

O

OH

NN

CH 2HS R

S

SR

H

Resolution Coupling Protocol :96 Chiral Acids or Bases

1 Solvent

NH

OH

O

O

OH

NH

Ph

Ph

NH 2

OHS

OS

O

H

10-20mg per reaction : 2g Total0.5 FTE Days

1 Day Start to Finish per Solvent

O

OH

O

NH

OHNH 2

NH 2

O

HO

S

OHO

y p

O

OH

O

O

Cl

NH 2

OH

RS NH 2

S

POOH

O2

Example Results of Resolution Screens

R R

NH2

O NHNH2

OH

OH

OO

NR

O

Ar

OH

O

NR

O

Ar

OH

O

O

1st Crop : 92%ee2 d C t lli ti 99%

MeCN / H2O 13:140 Volumes

N

NH2NOH

OHOH

1st Crop : 98%eeYield : 45%

MeOH / Water 10:110 Volumes

2nd Crystallisation : 99%eeOverall yield : 32%

NH OHO

O

N OH

OHOH

OHOH

N

O O

OH

N N

OO

OH O

Tol Tol

O

O OOH 1st Crop 96%ee

Yield : 36%

MeOH / H2O 8:145 Volumnes

N

NHN

NHEtOH / MeOH 1:140 volumes

1st Crop : 99%eeYield : 31%

Zhou, Ru; Bi, Chris; He, Mingying; Hoffman, Robert; Jalaie, Mehran; Kath, John; Kupchinsky, Stan; Ling, Tony; Lu, Jihong; Marx, Matthew; Richardson, Paul; Sach, Neal; Tran, Khanh; Barbour, Nicole; Cho-Schultz, Su. Asymmetric synthesis of Pyrrolodino-pyridine based CXCR4. Abstracts of Papers, 240th ACS National Meeting, Boston, MA, United States, August 22-26, 2010 (2010)

Resolution screening is successful in 80% of cases

Example Workflow - Solubility ScreeningIsolating good quality material can be as difficult as forming the covalent bondsProblematic isolations may result in largeProblematic isolations may result in large quantities of organic and aqueous waste streams and loss of product Traditional Reaction SequenceTraditional Reaction Sequence

1.

Design Reaction

2.

Run Reaction

3.

Analyse Reaction

4.

Design Work-up

5.

Work-up Reaction

Reaction Solvent

Low Solubilityof Product

Bottom-up approach to Reaction Design

1.

Design Reaction

2.

Run Reaction

3.

Analyse Reaction

4.

Design Work-up

5.

Work-up Reaction

High Solubilityof Bi-products

Driven by solubility screening

Example Results of Solubility Screens

Objective: Solvent

Solubility of Impurity @ RT

Solubility of product @ RT

Map solubility of all synthetic intermediates in range of process solvents

HTE Workflow:

Solvent p y @mg/ml

p @mg/ml

Water <1 34Methanol 14.00 86Acetonitrile 61.00 15

20 solvents screened as standard chosen by chemistSolution saturated and liquors analysed

Ethanol 10.00 74Acetone >100 78Acetic acid >100 >1002-Propanol 7.00 152 B t 85 00 65Solution saturated and liquors analysed

Quantitative LC2-Butanone 85.00 65Tetrahydrofuran >100 34Ethylacetate 43.00 16Butyl Acetate 40.00 14t-amyl OH 4 00 17

Typically applied to entire synthesis250mg-1g per screen in 0.5 Day

t-amyl OH 4.00 17Dichloromethane >100 36Diisopropylether 5.00 12Toluene 22.00 8Heptane <1 3epta e 31M HCl >100 >201M NaOH <1 >201M NaHCO3 <1 <1

2007 - Bringing HTE to Discovery

Reaction Optimization Projects 2002-2007 Discovery 20071 FTE!!

LJ 2007-2011120

140

160

Process Projects2002-2006

1 FTE!!

Sandwich60

80

100

120

Projects

2002-20065 FTE’s

Sandwich2002-2006

0

20

40

60

02002 2003 2004 2005 2006 2007

Year

…the workflow was not designed for this throughput and was too slow…

Reaction Optimization in Process

Experimental Set-Up

x1x2X nExperiment

p

ReactionProject Team Answerx1x2X nExperiment

DesignReaction Analysis

Team Problem

Answer

Cycle 2 =Cycle 3 =Cycle 1 =Results

Interpretation Cycle 2 4 days

Cycle 3 6 days

Cycle 1 2 days

2007 - The Discovery Experience

Discovery chemists want answers…f d i l i l d i h d l…faster and using less material compared with process development

…and they have no product markers!

SDS – Screening / Designed Synthesis

LD – LeadDevelopment

CS – CandidateSeeking

ESD – ExploratoryScreen Development

Early Development

DiscoverySpeed █

Process DevelopmentScalable █Speed █

Diversity █Yield █

Scalable █Yield █Isolation █

Isolation █Scalable █

Speed █Diversity █

Speeding the Workflow

Challenge over the past three years has been to speed the workflow by removing the bottle necksthe workflow by removing the bottle necksWhere is the time spent?

Manual and slow

Semi-Automated

and fast

Semi-automated and slo

Reaction Reaction Analysis

and slow 50%

and fast20%

and slow30%

Design Executiony

andVisualisation

These steps needed addressing

Designing a Reaction Screen in Process

StatisticalStatisticalMethods Literature

Surveillance

Intelligent ReactionDesign

Chemistsand

Green Chemistry and

CeNChemistryPrincipals

Screening Chemical Space…searching the literature on a per screen basis helps us move into the

“correct” chemical space but if the chemistry doesn’t work in this initial screen we spend a lot of time searching or settling at local maximascreen we spend a lot of time searching or settling at local maxima…

…what if we moved to a more generic one-size fits all type template… that would save time but…the chemical space covered increases dramatically L R tidramatically…

Unchartered territory

Large Reaction Template

Cycle 1 = Bas

e

Literature Says

territory

Cycle 2 = Cycle 3 = 2 days

Catalyst

BLiterature Says Start Here 4 days6 days…besides...was we ever really comfortable excluding this space?

will you ever find something new?

Literature Misdirection161/161 references utilize Pd(Ph)4)3or Pd(P(Ph)3)Cl2

Sonogashira Reaction

BrSi

CatalystWhat have we really learnt?

Precedent (we knew that)N

N

N

Br

Ar

R N

N

NR

Ar

CatalystSolventBase

Project team report poor yields using standard conditions

%Area UV50%conditions

Literature SearchSi

50˚C 100˚CNN

Si

X 161 literature conditionsP P

PdClCl

ProductReactant/ReagentHNHN Mass Ion

CountPd-132Un-precendented

Increasing Chemical Screening Space vs. Material Demands

…discovery project teams cant afford >500mg of material for each screen, in-fact they would like to reduce the amounteach screen, in fact they would like to reduce the amount required…

WE NEED TO MINATURIZE (but without diluting)( g)

120uL1.2mlReactionVolume2007

to

ReactionVolume

to2011

20 Reactions=100mg 200 Reactions=100mg

Reducing Scale – Air/Moisture Sensitivity

8x30mm120uL

12x32mm1.2ml

Scale (400MW) 7.5mg, 0.02mmol 0.75mg, 0.002mmol

Catalyst (400MW 10mol%) 0 4mg 0 002mmol 0 04mg 0 0002mmolCatalyst (400MW, 10mol%) 0.4mg, 0.002mmol 0.04mg, 0.0002mmol

With just 0.04mg of catalyst present the workflow had to be moved into a glove box with a controlled O2 (<10ppm) and g 2 ( pp )H2O atmosphere (<20ppm)

Projects by Reaction Type

Buchwald12%

OtherReaction TypeReaction Type Templated :

Suzuki Couplings12%

Biotransformation0%

Oxidation0%

13%Other26%

p gBuchwald CouplingsAmide Couplings

Suzuki 17%

Flow Chemistry15%

Amide CouplingsHeck CouplingsCarbonylations

TemplatedBiotransformation

2%

Oxidation1%

yClassical Resol. Alkylations Hydrogenation

Sonogashira5%Heck

0%CH ActivationAmideResolution

Library Validation8%

Templated60%Flow Chemistry

9%

HydrogenationSonogashiraUllman Couplings0%CH Activation

3%Hydrogenation

8%Alkylation

11%Carbonylation

4%

Halogenation1%

Amide1%

Resolution0%Library Validation

3%

p gHalogenationsetc...

Speeding the Workflow

For templated chemistry (60%) the workflow is significantly fastersignificantly faster….…but analytical samples have increased by a magnitudemagnitude…

Semi Semi-Templated Chemistry

10%

Semi-Automated

and fast

Semi-Automated

but slow

ReactionDesign

Reaction Execution Analysis and

10% 30% 60%

Design Execution Visualisation

Data Analysis, Capture and Visualisation

Capturing data, identifying products, analysing trends and reaching quick conclusionsreaching quick conclusions

iChemExplorer Data Capture

Rapid Resolution 1200 LC/MSD

50-500 data points per

reaction screen.Email report

and then

Rapid Resolution 1200 LC/MSDPer sample cycle-time

dropped from 4 to 0.8minsreaction screen.

100 reactions take 1.5hrs

capture in CeN as well as in-

house reaction screening

Spotfire Visualization with Data Display

Data to knowledge in 1 hourscreening database

6Mathews, B. T.; Higginson, P. D.; Lyons, R.; Mitchell, J. C.; Sach, N. W.; Snowden, M. J.; Taylor, M. R.; Wright, A. G. Improving quantitative measurements for the evaporative light scattering detector.Chromatographia (2004), 60(11/12), 625-633.

Discovery Orientated Screening Workflow

…combining reaction templates with miniaturization and glove box technologies has enabled a reduction in materialglove box technologies has enabled a reduction in material demands and reduced screen cycle times leading to an increased applicability to Discovery Chemistry

317 35926.1

20 3 20 7

30400 Reaction Optimization Project 2002-2011

reen

158 195 24018.320.3

18.020.7

20200ro

jects

Projects

per S

cr

20 28 54 81 882.3 1.9 1.5 1.2 1.0

0

10

0

Pr Days per Screen

Day

s p

002002 2003 2004 2005 2006 2007 2008 2009 2010 2011

A templated reaction screen of >100 conditions is now turned around in 24hrs

Examples

Amide Bond Formations

Carboamidation Technology gy

C-N Couplings

Amide Bond Formation – A Discovery Problem

25% of all reactions in discovery involve amide couplingsy p g

Reaction Step detailsPiperidine formationMitsunobu couplingBuchwald amine arylation i f i

O-alkylationCarbon acylation

P lidi f ti

Large numbers of coupling agents are commercially availableTriazole formation

Buchwald amine arylation

Reduction aminationPurinone formation

Thiourea formationPyrazole formationPyridone formation

Pyrimidine formation

Nucleophilic addition

Enamime formationImidazole formation

Thermal decarboxylation

Pyrrolidinone formation

Buchwald ether coupling

Deprotection

Which one?Imine formation

Diketohydrazone formation

NO2 reduction

Imine cyclization

Boronic acid Mannich rearrangement

Urea formation

Sulfonamide formation

Dianiline 5 ring

N lk l tiWhich one?Reductive alkylation

N-alkylation

Nucleophilic substitution

Suzuki coupling

UgiQuinoline formation

Amide formation

Difficult Amide Bond Formations - Discovery

FSterically

70%1. T3P

60%1. Cyanuric Fluoride

ArCl

OH

OAr

Cl

NH

OR

NOH N

RR

NH

FF

F

FSelectivity

Stericallyhindered2. 1Cl2MPyI

1. Cyanuric Fluoride2. 2C-DMI-TFB

NAr

OH

NAr

NH

NH2

NH

OH

ONH RAmide Coupling Protocol :

36 Coupling Agents2 S l t 2 T t 1 Additi 50%

1. (COCl)22 CDMT

2 Solvents, 2 Temperatures, 1 Additive150 reactions = 90mg total (300MW)

0.5 FTE Days 2 D St t t Fi i h

60%1. TPTU 2. CDMT

3. iBuCOCl4. HATU

SO S

NR SO

OH

O

NH

Ar

NHO

Electron poor

amineSelectivity

2 Days Start to Finish2. TDBTU3. TOTT

R S

R Ar

OH NNR

R Ar

NHS

NN

NH2

ROH

Ar NN

NH2

NNH

R

The Amide Bond – Expanding Monomer Space

The amide bond remains a popular connection in drug di t l SARdiscovery to explorer SAR….

Ar O

OH

ANHCoupling AgentSolvent Core

Core

ArCOOH ArCl, Br, I, OTfAr1 O

NR3

Ar1

O

NHR3+

AcidAmide

Base

Amine Traditional Coupling

O l 725 (3 5%)

6441 14136

Ar1X

NH NAr1+

CatalystBaseSolvent

Core Core

but there are other way of making an amide bond and a

Overlap 725 (3.5%)R3N

R3OAmine Amide

X=Cl, Br, I, OTf Carboamidation Coupling

…but there are other way of making an amide bond and a monomer analysis reveals the advantages…

Carboamidation – Accessing New Chemical SpaceCarbonylation Screen

NH2N

N

NNH2

NN

NNH2

NN

NNH2CO Pressure

CatalystBaseS l t

H H

Details

NN

N

Br

N

ONH

N

ON O

NH

N N

SolventP

HHHH

Method required for primary, secondary and non-nucleophilic amines78 Conditions (0.005mmol, 1mg per reaction)

NH2NH2

N N

NH

( , g p )– 12 catalyst/ligand (combinations (all monomers)– 56 catalyst/ligand combinations (one monomer)– DMAc (0.03M), DIPEA (5eq.), 100C, 8Bar CO

NHTechnology NN

NNH2

ONH

NN

NNH2

O

O NH

O

Carbonylation Library ValidationCarbonylation Screen

N NNH2

NNH2

N NNH2CO Pressure

Catalyst

P

H H

HHHH

$73/gN

N

NNH2

Br

NN

N

ONH

NN

N

ON

N

ONH

N N

yBaseSolvent

ABS

P

NH2NH2

N N

NH

DetailsTake top eight ligands from initial screen and run under actual library conditions Conclusions

PFe$640/gN

run under actual library conditions78 Conditions (0.05mmol=10mg per reaction)

8 catalyst/ligand combinations3 monomers (3eq.)

ConclusionsLigand system works across all demonstration monomersLibrary 1 = 88 designed products

DMAc (0.1M), DIPEA (5eq.), 100C, 4Bar CO

CommentsSM is not soluble at this concentration at 25C.

Library 1 = 88 designed products82 / 88 achieved = 93% success rate @ 0.05mmol scale. How would acid compare?500 products have since been prepared

Solids weighing robot is used to weigh 80 x 10mg

500 products have since been prepared using these conditions (2 weeks)

Suzuki Optimization%Area UV

80% NN

ClCl

NF

F N

Br

R

NF

F

CatalystSolventBase

+

MeOH / CsF(aq.) /Pd-132 // 80C / 18hrsP P

PdClCl

Pd-132

Issue

FBOH OH

N

NH2

OR F

N

NH2

OROriginal

Conditions

Team tried DME / H2O / Pd(dppf)2Cl2/ 80°C but gave only 50% conversion after 2 days Results

Mass IonCounty

Screen Design96 conditions (0.003mmol, 1.13mg)

Results1/96 conditions gave >80% in-situ yieldMeOH, CsF and Pd-132 are critical. Three f t i t ti h d t fi d O f t

Product– 110mg for screen– 6 catalyst/ligand combinations– 4 bases and 4 solvents

factor interactions are hard to find. One factor at a time optimizations wont work here.Request to Results Cycle Time = 1day

– 80C, 18hrs, glove box Scaled to 200mg, 90% isolated yieldRequest to Registration Time = 2days

Des Bromo

Conclusions

1 / 100 / 10 / 1 / 0 05 /

20071998 2001 2004 2012

1g/rxnX 10

100mg/rxnX 50

10mg/rxnX 100

1mg/rxnX 200

0.05mg/rxn.X 1000

KNOWLEDGE PER GRAM PER UNIT TIME

20071998 2001 2004 2012

Maximize information per reactionMinimize material consumptionMinimize time to explore any given chemical space

Using HTE Workflows, optima within a given parameter g p g pspace can be quickly identified to focus resource at any stage across the portfolio

Acknowledgements

TechnologyC l B i

ChemistryKevin Bunker

Special ThanksSimon BaileyCarol Baines

David BernhardsonKlaus DressD id D

Kevin BunkerLaurence HarrisBob HoffmanPeter Huang

Simon BaileyRob CrookPeter DunnKlaus DressDavid Damon

Bill FarellStuart FieldSt F ll

Peter HuangSarah LewandowskiSacha NinkovicDan Richter

Klaus DressMartin EdwardsPaul HigginsonJennifer LafontaineSteve Fussell

Rob MaguireBen MatthewsAndre Morrell

Dan RichterHong ShenJohn TatlockJohn Tatlock

Jennifer LafontaineVan MartinIvan MarzianoSteve TrudelAndrew Morrell

Paul RichardsonSteve TrudelAle Wilder

John Tatlock Steve TrudelPaul RichardsonPete SpargoDavid WaiteAlex Wilder

Katie WilfordAlex Yanovsky

David WaiteMike WilliamsAlex Yanovsky