pseudo-dynamic combinatorial libraries: a …digitool.library.mcgill.ca/thesisfile85059.pdf ·...

PSEUDO-DYNAMIC COMBINATORIAL LIBRARIES:

A RECEPTOR-ASSISTED COMBINATORIAL CHEMISTRY APPROACH TO

DRUG DISCOVERY

Jeremy D. Cheeseman

A thesis submitted to the Faculty of Graduate Studies and Research

of Mc Gill University in partial fulfilment of the requirements of the degree of

Doctor of Philosophy

Department of Chemistry

McGill University

Montréal, Québec, Canada

May 2004

© Jeremy D .. Cheeseman 2004

1+1 Library and Archives Canada

Bibliothèque et Archives Canada

Published Heritage Branch

Direction du Patrimoine de l'édition

395 Wellington Street Ottawa ON K1A ON4 Canada

395, rue Wellington Ottawa ON K1A ON4 Canada

NOTICE: The author has granted a nonexclusive license allowing Library and Archives Canada to reproduce, publish, archive, preserve, conserve, communicate to the public by telecommunication or on the Internet, loan, distribute and sell th es es worldwide, for commercial or noncommercial purposes, in microform, paper, electronic and/or any other formats.

The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission.

ln compliance with the Canadian Privacy Act some supporting forms may have been removed from this thesis.

While these forms may be included in the document page count, their removal does not represent any loss of content from the thesis.

• •• Canada

AVIS:

Your file Votre référence ISBN: 0-494-06283-5 Our file Notre référence ISBN: 0-494-06283-5

L'auteur a accordé une licence non exclusive permettant à la Bibliothèque et Archives Canada de reproduire, publier, archiver, sauvegarder, conserver, transmettre au public par télécommunication ou par l'Internet, prêter, distribuer et vendre des thèses partout dans le monde, à des fins commerciales ou autres, sur support microforme, papier, électronique et/ou autres formats.

L'auteur conserve la propriété du droit d'auteur et des droits moraux qui protège cette thèse. Ni la thèse ni des extraits substantiels de celle-ci ne doivent être imprimés ou autrement reproduits sans son autorisation.

Conformément à la loi canadienne sur la protection de la vie privée, quelques formulaires secondaires ont été enlevés de cette thèse.

Bien que ces formulaires aient inclus dans la pagination, il n'y aura aucun contenu manquant.

"'1-fe who makes a 6east of himseg: 8ets rid of the yain of 6ein8 a man."

-I}-{unter s. rfhomyson

ABSTRACT

Emerging methods of combinatorial chemistry involve receptor assistance to combine

synthesis and screening. Binding to the receptor alters either the thermodynamics or

kinetics of synthesis. Dynamic combinatorial chemistry uses reversible synthesis where

binding to the receptor shifts the equilibrium to make more of the best binders. In target

accelerated synthesis, binding of the starting materials to the receptor speeds up the

synthesis of the best-binding compounds. We report a new receptor-assisted method

-pseudo-dynamic combinatorial chemistry- where binding to a receptor slows the

destruction of the best-binding compounds. In pseudo-dynamic libraries, synthesis and

destruction of library members are separate, irreversible reactions. Extending the

destruction reaction amplifies binding differences similar to a kinetic resolution of

enantiomers. Initial libraries of two to eight dipeptides, sorne containing an aryl

sulfonamide moiety that binds to carbonic anhydrase, showed that a ratio of> 100: 1 of the

best binding dipeptide over the next best was possible. These experiments also suggested

that the selectivity is related to the number of compounds in a library, with more library

members producing higher selectivity (a highly desirable result opposite that seen in

traditional dynamic libraries). Expansion of these libraries to include compounds

containing sulfonamides, aryl sulfonamides, sulfamates and hydroxamic acids further

support postulations as to the origins of the high selectivity of these systems, and take the

number of compounds screened by a pseudo-dynamic library closer to practical levels for

drug discovery.

RÉSUMÉ

Les nouvelles méthodes de la chimie combinatoire impliquent une cible biologique pour

combiner la synthèse et la détection des résultats d'une librairie. La liaison d'un composé

à la cible change les aspects thermodynamiques, ou cinétiques de la synthèse. Les

librairies combinatoires dynamiques utilisent une synthèse réversible pour déplacer

l'équilibre en faveur des composés qui se lient à la cible. Dans les systèmes où la cible

favorise la synthèse, la vitesse de synthèse est améliorée pour les composés qui se lient le

plus fortement à la cible. On a créé un nouveau système combinatoire -les librairies

combinatoires pseudo-dynamiques- dont la liaison d'un composé à la cible ralentisse sa

destruction. Dans les systèmes pseudo dynamiques, la synthèse et la destruction de la

bibliothèque sont des processus séparés et irréversibles. Dans nos premières expériences,

on a créé des inhibiteurs dipeptidiques contre l'anhydrase carbonique. La synthèse

utilisait des esters activés qui se couple avec des aminoacides. La destruction était

l'hydrolyse catalysée par Pronase. En utilisant un processus similaire aux résolutions

cinétiques, le prolongement de la destruction amplifie les différences des affinités des

composés pour la cible. Ces systèmes peuvent distinguer entre les composés ayant une

différence dans les constantes de liaison de moins d'un facteur de deux. Les expériences

récentes ont augmenté le nombre des composés, et démontrent que dans les systèmes

pseudo-dynamiques le fait d'avoir plus des composés dans le système améliore la

sélectivité de la cible pour son inhibiteur le plus puissant.

ii

ACKNOWLEDGEMENTS

1 would like to thank my research supervisor, Prof essor Romas J. Kazlauskas,

whose extensive knowledge and multifaceted approach to chemistry has in great part

shaped my own scientific style. 1 am also grateful (tongue in cheek) for his technological

savvy in the form of video conferencing, database support and email, which more than

compensated for his lack of physical presence during the greater part of my studies. But

in aIl seriousness, 1 am grateful for the opportunity to have worked so independently.

1 also wish to thank Professor James L. Gleason, with whom we collaborated on

this project. He always pushed me to my limits, without which 1 would have foundered

early on in my studies.

1 want to thank Andrew D. Corbett for his often tireless synthetic efforts, for

editing this thesis, and for taking sorne of the nights off my hands when 1 just couldn't

stay awake for another shift on the library. Without his efforts, this project would not

have been possible.

1 would like to thank David Soriano deI Arno, who has taken over the job of

supplying me with molecules to play with, and who l'm sure will take this project to the

nextlevel.

The department and especially my lab have been quite dynamic, with many

changes in personnel and location in a short time and l'm grateful to aIl who came and

went. l'd especially like to thank Seongsoon Park, Chris Savile, Paul Mugford, Krista

Morely and James Ashenhurst along with the rest of the Gleason lab for a great time, at

least while we were aIl in the same countries.

iii

l would like to thank Renée Charron, Chantal Marotte, Fay Nurse, Paulette

Henault, Carol Brown and Sandra Aerssen for great administrative support.

Financial assistance from McGill University and the department of chemistry in

the form of Alma Mater Travel Grants, the Parsini Diwan prize in chemistry, and the

Robert Zamboni Prize in chemistry, and from NSERC and CIHR for post-graduate

fellowships is gratefully acknowledged.

iv

TABLE OF CONTENTS

Abstract

Résumé

Acknowledgements

Table of Contents

Abbreviations

List of Figures

List of Schemes/Tables/Miscellaneous

General Contribution of Authors

Chapter One Introduction to Receptor Assisted Combinatorial

Chemistry in Drug Discovery

1.1 Combinatorial Chemistry

1

11

111

v

X11

XV111

XXIV

xxvii

1

2

v

1.2 Objectives, Methods and Terms in Drug Discovery Applications of Receptor-

Assisted Combinatorial Chemistry

1.3 Receptor Assisted Systems Under Thermodynamic Control:

1.4

Dynamic Combinatorial Libraries

1.3.1 DCLs requiring "lock in" reactions

1.3.2 Systems with in situ detection

1.3.3 Limitations to DCLs

Receptor Assisted Systems That Add Kinetic Control

1.4.1 Receptor-Accelerated Synthesis

1.4.2 Limitations In RAS

1.4.3 Affinity Chromatography

References

5

10

11

15

21

25

26

28

29

32

ChapterTwo Amplification of Screening Sensitivity Through Selective

Destruction: Theory and Screening of a Library of Carbonic

Anhydrase Inhibitors 36

Abstract 37

2.1 Introduction 37

vi

2.2 Theory: Finding the Best Inhibitor by Destruction of Poorer

Inhibitors

2.3 Results

2.3.1 Synthesis of 4'-Sulfonamidophenylalanine Dipeptides

2.3.2 Inhibition of Carbonic Anhydrase

2.3.3 Selective Extraction of Inhibitors By Carbonic Anhydrase

2.3.4 Selective Protection oflnhibitors by Carbonic Anhydrase

from Hydrolysis

2.3.4.1 Screening ofProteases

2.3.4.2 Selective Protection oflnhibitors

2.4 Discussion/Conclusions

Contribution of Authors

References

Chapter Three First Generation Pseudo-Dynamic Libraries

Abstract

40

43

43

44

46

49

49

49

54

58

59

61

62

vii

3.1 Designing the First Pseudo-Dynamic Combinatorial Library

3.1.1 Synthesis of the Library

62

64

3.1.2 Binding to the Receptor and Destruction ofUnbound Library Members

66

3.1.3 Recycling Destruction Products and Iteration of Synthe sis 69

3.2 Results of the Integrated Processes in the First Pseudo-Dynamic Combinatorial

Library 70

3.3 Discussion/Conclusion 75

Contribution of Authors 77

References 77

Chapter Four Pseudo-Dynamic Combinatorial Libraries: A New Receptor-

Assisted Approach For Drug Discovery 78

Abstract 79

4.1 Introduction 79

4.2 Experimental Design 80

viii

4.3 Results 83

4.4 Discussion/Conclusion 86


References 88

Chapter Five Amplification and Selectivity in, Expansion and Modeling of

Pseudo-Dynamic Combinatorial Libraries

91

Abstract

5.1 Origins of the Amplification Maxima in the Eight-Memberedp-DCL

5.1.1 Introduction

5.1.2 Testing Enzyme Purity

5.1.3 Lowering the Concentration ofNucleophiles

92

92

92

95

97

5.1.4 Discussion: A Steady State Concentration of the Strongest Binding

Dipeptide 102

ix

5.2 The Effect of the Number of Inhibitors on Amplification and Selectivity in

5.3

5.4

5.5

5.6

Pseudo-Dynamic Combinatorial Libraries

5.2.1 Introduction

104

104

5.2.2 Two- and Three-Inhibitor 16 h Cycle P-DCLs 104

5.2.3 Discussion: Fewer Inhibitors Increase Amplification, but Decrease

Se1ectivity in P-DCLs 106

Expansion ofthe Pseudo-Dynamic Combinatorial Library 107

5.3.1 Introduction: New Library Members and P-DCL Scheme 107

5.3.2 Results

5.3.3 Discussion

Preliminary Modeling of Pseudo-Dynamic Combinatorial Libraries

5.4.1 Introduction

5.4.2 Modeling Synthesis

5.4.3 Mode1ing Receptor-Binding

5.4.4 Modeling Destruction 117

110

111

112

112

113

115

5.4.5 Discussion: The Integrated Model and its Comparison to Experiment

119

Overall Conclusions

Future Endeavours

5.6.1 Improving the p-DCL Model

5.6.2 Fundamental Experimentation inp-DCLs

122

124

124

125

x

5.6.3 Expansions and Miniaturizations 125


References 126

Chapter Six Experimental Section 129

6.1 Experimental Section for Chapter Two 130

6.2 Experimental Section for Chapter Three 141

6.3 Experimental Section for Chapter Four 145

6.5 Experimental Section for Chapter Five 149

References 156

Contribution to Knowledge 157

Appendix 1 Amplification of Screening Sensitivity through Selective Destruction:

Theory and Screening of a Library of Carbonic Anhydrase Inhibitors

xi

Journal of the American Chemical Society 2002, 124,5692-5701 © 2002

American Chemical Society

Appendix II Pseudo-Dynamic Combinatorial Libraries: A Receptor-Assisted Approach

for drug Discovery Angewandte Chemie International Edition English

2004,43,2432-2436 © 2004 Wiley InterScience

XII

ABBREVIATIONS

AA amino acid

Ac acetyl

AcCN acetonitrile

AChE acetylcholine esterase

Alasa [3-sulfonamidoalanine

Aq aqueous

Ar aryl

Arg argmme

Asp aspartic acid

Asn asparagme

BICINE N,N-bis(2-hydroxyethyl)glycine

Bn benzyl

BOC tert -butoxycarbony 1

BSA bovine serum albumine

Bu butyl

Bz benzoyl

oC degree Celsius

CA carbonic anhydrase

cat. catalystlcatalytic

Cbz benzyloxycarbonyl

CC combinatorial chemistry

Xlll

Con A concanavalin A

d doublet

DCL dynamic combinatoriallibrary

dd doublet of doublets

ddd doublet of doublet of doublets

de diastereomeric excess

DMAP 4-( dimethy lamino )pyridine

DMF N,N-dimethylformamide

DMSO dimethy lsulfoxide

DNA deoxyribonucleic acid

DTT dithiothreitol

EDC-HCI 1-(3-dimethylamino )propyl-3-~-

ethy lcarbodiimide hydrochloride

Etoc ethoxycarbonyl

ee enantiomeric excess

eq. equivalents

Et ethyl

g gram(s)

Gal galactose

GC gas chromatography

Glc glucose

GluNHoH L-glutamic acid y-monohydroxamate

Gly glycine

xiv

h hour(s) .~-

HMPA hexamethylphosphoramide

HPLC high performance liquidchromatography

HTS high-throughput screening

Hz hertz

i iso

l inhibitor

ICso concentration required for 50% inhibition

J coupling constant

Ka affinity constant

kI destruction rate constant

Kd dissociation constant

Kr inhibition constant

ks synthetic rate constant

Ks equilibrium constant of synthesis

L litre

LCMS high performance liquid

chromatograhy /mass spectrometry

Leu leucine

Lyssa lysine-E-sulfamide

m meta

Il micro

m milli, multiplet

xv

M moles per litre

Man mannose

mCPBA meta-chloroperoxybenzoic acid

Me methyl

MIC minimum inhibitory concentration

mL millilitre

mmol millimole

mol mole

MS mass spectrometry

m/z mass to charge ratio

n normal

NANA N-acetylneuraminic acid ~-

ND not determined

NHSSu N-hydroxysulfosuccinamide

n nano

NMR nuclear magnetic resonance

p para or pseudo

PEG polyethylene glycol

PFP pentafluorophenol

pH -log [H+]

Ph phenyl

PhMe toluene

Phe phenylalanine

xvi

Phesa

pKa

Pr

Pro

R

RACe

RAS

RNA

rt

s

s

SM

STP

t

T

t

tert

tetra-FP

TEA

tri-FP

TFA

THF

Tyr

4' -sulfonamidophenylalanine

-log(Ka)

propyl

proline

receptor

receptor-assisted combinatorial chemistry

receptor-accelerated synthesis

ribonucleic acid

room temperature

secondary

singlet

starting material

4-sulfo-2,3,5,6-tetrafluorophenol

tertiary

biological target

triplet

tertiary

tetrafluorophenol

triethylamine

trifluorophenol

trifluoroacetic acid

tetrahydrofuran

tyrosine

XVll

UV

Val

w/w

WGA

Xyl

ultraviolet

Valine

weight by weight comparison

wheat germ agglutinin

xylose

xviii

LIST OF FIGURES

Figure 1.1: ParaUel Synthesis 3

Figure 1.2: Split-Pool Synthesis 4

Figure 1.3: Three receptor-assisted combinatorial methods 10

Figure 1.4: The first dynamic combinatoriallibrary 12

Figure 1-5: An example of a DCL from O. Ramstrom and l-M. Lehn 13

Figure 1-6: Creation of an imine library against neuraminidase using a diamine and

several ketones 15

Figure 1-7: A DCL made from NANA aldolase catalyzed aldol formation in the

presence of wheat germ agglutinin 16

Figure 1-8: An oxindole library created in the presence of cyclin-dependent kinase-2

(CDK-2) crystals 17

Figure 1-9: An example of "tethering" usmg DCL principles from Sunesis

pharmaceuticals 18

xix

Figure 1-10: Extended tethering with caspase-3, a protease with several binding pockets

(SI-S4) 20

Figure 1-11. The dynamic combinatoriallibrary equilibria yield a higher amount of

good inhibitor 22

Figure 1-12. Theoretical effect of a DCL in which the combination of three different

starting materials (A, B and C) yields a library of trimers 25

Figure 1-13. Nicolaou's vancomycin monomer scaffold 27

Figure 1-14. An example ofreceptor-accelerated synthesis from W. G. Lewis et al.

28

Figure 1-15. The bis(salicylaldimiato)zinc DNA-binding complex affinity column

experiment 30

Figure 1.16. The affinity column-UV generator loop by Nelen and Eliseev 32

Figure 2-1. Predicted ratio of the total (bound and unbound) concentrations of two

hypothetical inhibitors 43

xx

Figure 2-2. Selective concentration of the sulfonamide 4a over non-inhibitor 5 into the

carbonic anhydrase-containing compartment of a two compartment vessel

47

Figure 2-3. Selective concentration of sulfonamides 4a-d over non-inhibitor 5 into the

carbonic anhydrase-containing compartment of a two compartment vessel

48

Figure 2-4. Reaction design for the selective destruction experiments 50

Figure 2-5. Selective protection from destruction of sulfonamide 4a over non-inhibitor

5 by carbonic anhydrase 51

Figure 2-6. Selective protection from destruction of sulfonamide 4a over 4 b by

carbonic anhydrase 52

Figure 2-7. Selective protection from destruction of sulfonamide 4a over 4c by

carbonic anhydrase 53

Figure 2-8. Selective protection from destruction of sulfonamides by carbonic

anhydrase 54

xxi

Figure 2-9. The graph shows theoretical and experimental ratios for the screening

experiments 56

Figure 3-1. Schematic of a hypothetical pseudo-dynamic combinatoriallibrary 64

Figure 3-2. Solid phase, aqueous synthesis of a library of dipeptides 65

Figure 3-3. The dipeptide library can now interact with the receptor, carbonic

anhydrase 67

Figure 3-4. The destruction of unbound library members 68

Figure 3-5. The three-chambered p-DCL experimental set up 69

Figure 3-6. P-DCL oftwo cycles, 24 h each 73

Figure 3-7. Two 24 h cycles, with a delay in the addition of the destruction chamber

74

Figure 3-8. Four 12 h cycles with half the amount of dipeptide formed per cycle

75

xxii

Figure 4-1. Schematic of the pseudo-dynamic combinatoriallibrary experiment

82

Figure 4-2. Structures and competitive inhibition constants for the CA-catalyzed

hydrolysis of p-nitrophenyl actetate 84

Figure 4-3. Pseudo-dynamic library experiments 86

Figure 5-1. Quantification of CA binding sites 96

Figure 5-2. Six 16-h cycles of an eight-membered p-DCL with [nucleophiles] = 4 mM

99

Figure 5-3. Twelve 16-h cycles of an eight-membered p-DCL with [nucleophiles] = 4

mM 100

Figure 5-4. Twelve 16-h cycles of an eight-membered p-DCL with [nucleophiles] = 4

mM 101

Figure 5-5. Results of a four membered p-DCL, with two inhibitors 105

Figure 5-6. Results of a six membered p-DCL over six 16-hour cycles with three

inhibitors 106

xxiii

Figure 5.7. The expanded library members 109

Figure 5.8. Expanded library scheme 110

Figure 5.9. Uncorrected UV absorbance of the final two remaining members of a 30

memberedp-DCL 111

Figure 5-10. General scheme for library synthe sis 114

Figure 5-11. Two inhibitors lA and lB compete for a biological target T 115

Figure 5-12. The destruction of any unbound inhibitor 1 govemed by the inhibition

constant and the kinetic rate constant 118

Figure 5-13. Theoretical curve for the ratio oftwo inhibitors in apDCL 119

LIST OF SCHEMESrrABLES/MISCELLANEOUS

Box 1.1. Terms in receptor-assisted combinatorial chemistry. 6

Table 1.1. Drug discovery approaches using receptor-assisted combinatorial '

chemistry 9

xxiv

Scheme 2-1. Aryl sulfonamide-based dipeptide libraries as inhibitors of carbonic

anhydrase 39

Scheme 2-2. Destruction of inhibitors 40

Scheme 2-3. Preparation of 4'sulfonamidophenylalanine dipeptides 44

Table 2.1.

Table 3.1.

Table 3.2.

Inhibition of carbonic anhydrase by sulfonamides 1 and 2, sulfonamide

dipeptides 4a-d, and dipeptide 5 45

Distribution of products from the library described in Figure 3.1 66

Inhibition constants for the components of the first generation p-DLCs

71

Scheme 4.1. Creation of a pseudo-dynamic library of dipeptides 81

Table 5.1.

Table 5.2.

Effect on synthetic rate and yield due to changing the nucleophile

concentration 98

Effect on yield and selectivity due to changing the nucleophile,

electrophile and number of cycles in the eight-membered library 101

xxv

Table 5.3. Parameters used to generate Figure 5.6 127

Table 6.1. Amounts of (1) used in static library experiments 151

Table 6.2. Variations of the eight-membered, 16 hour library 152

xxvi

General Contribution of Authors

The research project described in this thesis was carried out as a coIlaborative

effort between the research groups of professors James L. Gleason and Romas J.

Kazlauskas of the department of chemistry of Mc Gill University. Ronghua Shu was

responsible for the initial protease screening in section 2.3.4.1. Andrew D. Corbett,

Jonathan Croteau and David Soriano deI Arno, under the supervision of professor J. L.

Gleason, developed aIl che mi cal syntheses described in this thesis. R. J. Kazlauskas

developed the theory in section 2.2. Andrew D. Corbett or David Soriano with the author

performed sorne library experiments together. The author developed and carried out aIl

other work, including library design, described in this thesis. Chapters two and four are

adapted from published manuscripts reprinted in the appendices 1 and II respectively.

xxvii

CHAPTERONE

INTRODUCTION: RECEPTOR-ASSISTED COMBINATORIAL CHEMISTRY

IN DRUG DISCOVERY

1

1.1 Combinatorial Chemistry

For millennia, human beings have used products from nature to alleviate

symptoms from, or to cure disease. Little known to our ancestors, leaves, bark, and other

natural sources were effective because of active ingredients within them. Purifying out

the active ingredients became possible as chemistry developed starting in the

Renaissance. With the onset of globalization in the late 1800s, researchers realized that

there were many natural products in use as pharmaceuticals in various parts of the world,

but attaining these compounds in levels that would meet demand was difficult or

impossible. In the late 1800s, acetylsalicylic acid (Aspirin™) was synthesized in a

chemicallaboratory, and an alternate source to the natural one (willow bark) was finally

available. This advancement began what is now known as drug discovery. Throughout

the 1900s, new synthetic methods have been developed and numerous drugs have been

synthesized.

In the 1980's, with the onset ofmacromolecular modeling and rapid advances in

numbers of known biological structures, rationally designing drugs based on careful

inspection of pharmacophores aspired to create potent pharmaceuticals. However, tailor

making the perfect drug proved far more difficult than initially anticipated. Additionally,

creating new, potent drugs one at a time was not supplying demand. If one could not then

successfully design a tight binder from scratch, could one be created by chance?

Combinatorial chemistry emerged in the early 1990's as a way to create many

compounds in a short period of time. A plethora of potential drugs are made at the same

time (called a library), providing hundreds and even thousands of new molecules in the

2

same time it had taken to make one. Potential drug leads are usually synthesized using

one of two techniques: parallel synthe sis (Figure 1.1) or split pool synthesis (Figure 1.2).

After the library is made it is screened against a biological receptor (a membrane

receptor, enzyme, DNA, RNA etc.) in a separate step. This screening usually involves a

specialized assay that needs to be designed for each new receptor.

0101.1 ... <:::> : ! 1 !

a

C;;i~i+i"'? r

<jiliti"'t Figure 1.1: Parallei synthesis. Starting materials (represented by an open circle, square, black circle and

open oval) are aIl subjected to the same set of reaction conditions in separate vessels, resulting in

modification "a". Performing subsequent reactions and purifications ( ... n) yields products that can then be

screened against a biological receptor, R, and detected using a specific assay (represented by "D"). Because

aIl reactions happened in separate vials, aIl products are separate at the end of the synthetic procedure and

identifying the ones that bind to the receptor is straightforward. Another of this method's strengths is that it

is easy to automate. Drawbacks of parallei synthesis include the difficulty in creating libraries greater than

hundreds of compounds, and the need to have a separate, specific assay to detect binding to the receptor.

3

Figure 1.2: Split-Pool combinatorial synthesis (two steps shown) and library screening. In split pool

synthesis an initial mixture of starting materials (usually on solid support) are reacted with a variety of

compounds under various conditions (represented on the far left by three reaction arrows with a circle,

triangle or square representing three hypothetical sets of conditions). This results (in this illustration) in

three vessels, one containing products from the first reaction, one containing products of the second and so

on (the first "split"). The three products are then mixed together (pooled), split into three portions each of

which are then again subjected to three different sets of reactions. Each new round of splitting and pooling

exponentially increases the number of compounds in the library. In the example shown, three initial

compounds undergo three reactions in two split/pool rounds to give 27 different products. Step "a"

represents addition of the receptor, R. Step "b" represents addition of a detection method (symbolized by

"D"). Step "c" represents deconvolution of library and identification of drug lead. Split/pool techniques can

make libraries of millions of compounds, however deconvoluting which compounds bind to the receptor

can be very difficult. This method also requires a separate, receptor-specific screening step.

This completely random approach has not, in large part, proven successful. The

libraries were simply not large enough, or did not contain high enough diversity to

generate potent receptor-binding compounds. Since increasing the size of the library

renders its screening more and more difficult, a shift back towards a more rational design

is taking place, only now using element of combinatorial synthesis.

4

1.2 Objectives, Methods and Terms in Drug Discovery Applications of Receptor-

Assisted Combinatorial Chemistry

Combinatorial chemistry methods currently favour focused libraries, which use

templates or functional groups known to bind to the desired receptor.[1,2] An emerging

method in combinatorial chemistry, receptor-assisted combinatorial chemistry (RACC),

not only uses focused libraries, but also adds stoichiometric amounts of the receptor

during the synthesis of the library. Addition of the receptor biases the synthesis toward

the best binding compounds, thereby combining the synthesis and screening into one

step. In addition, analysis avoids specific receptor assays, but detects increased amounts

of the best-binding compounds with established chemical methods such as HPLC, MS,

NMR or even X-ray crystallography.

Three main RACC methods have emerged: dynamic combinatorial libraries,

receptor-accelerated synthesis, and a new method, pseudo-dynamic combinatorial

libraries (Table 1.1, Figure 1.3). These methods respectively use thermodynamic control,

kinetic control or both to increase the relative amounts of the best binding compounds

during synthesis. This relative increase can be defined as amplification (see Box 1.1 for

definitions). The amplification should reflect the relative binding constants of the library

members and selectivity is defined so as to measure this relationship. The selectivity for a

pair of library members is their amplification divided by their respective binding

constants. Methods based on thermodynamic control are in equilibrium, so the

amplification does not normally exceed the binding constants and the maximum

selectivity is usually below one. On the other hand, methods based on kinetic control can

5

yield selectivities weIl beyond one. Selectivities beyond one are called enhancements.

Enhancements enable methods to easily distinguish library members with similar binding

constants.

amplification - ratio of the amount of a library member synthesized in the presence of

a receptor as compared to its amount in the absence of the receptor, [IA]receptor/ [IA]no

receptor. In pseudo-dynamic libraries, aIl compounds are destroyed in the absence of the

receptor, so the amplification is infinite. In these cases, it is more useful to compare

yield, which is the concentration of a compound in the binding chamber compared to

the concentration of the receptor, [IA]/[R]. The maximum yield is 100%.

casting vs. molding - casting forms a small molecule using a receptor binding site as a

template. Drug discovery seeks to cast a drug lead using the receptor. Molding forms a

receptor by surrounding a small molecule target. For example, mol ding could form a

crown ether-like macro cycle around an ion.

enhancement - selectivity beyond one; that is, the ability to amplify library members

beyond their relative binding constants. Enhancement allows RACC to distinguish

between library members with similar binding constants.

receptor - entity to which the library members should bind. In drug discovery

applications, receptors can be cell membrane receptors, enzymes, interfaces for

protein-protein interaction, sites on RNA or DNA, etc. In supramolecular receptor

building applications, receptors are the supramolecules that are evolved to bind the

guest molecule.

selectivity - in DCL and RAS selectivity is the amplification of two library members

6

lA and lB, where lA is the stronger binder, compared to their relative binding constants,

[([IA]receptor/[IA]no receptor)/([IB]receptor![IB]no receptor)][(Ka )/(Ka )rl. In p-DCL, since aIl

lA lB

compounds are destroyed in the absence of the receptor, any detectable amount of a

library member would result in an amplification ofinfinity. For p-DCL, the selectivity

is [[IA]receptor![IB]receptor] [(Ka ) /(Ka )rl, which compares the relative concentrations of

lA lB

the inhibitors to their relative binding constants.

tethering - a dynamic combinatorial method involving formation of disulfides where

one component is a cysteine residue on the receptor. Tethering focuses the binding to

the region near the cysteine.

Box 1.1: Terms in Receptor-assisted combinatorial chemistry.

Dynamic combinatorial libraries (DCLs) use reactions in equilibrium to form

mixtures of library members. The receptor binds the tightest-binding library members,

removing them from solution. The synthetic equilibrium shifts to increase the amounts of

these tightest binding library members according to LeChâtelier' s principle. As

mentioned above, the amplification does not normaIly exceed the relative binding

constants of the library members. In other words, the selectivity rarely exceeds one and

there is little or no enhancement in DCLs.

In receptor-accelerated synthesis (RAS) the receptor binds several starting

components and promotes their coupling due to proximity, forming a new, tighter-

binding species. The rate-acceleration of this coupling reaction identifies the best binding

compounds and determines the amplification. Selectivity arises from two factors: binding

of the starting components and the ability of the receptor to catalyze the reaction. The

7

binding of the starting components is presumably an equilibrium, but the binding

constants of the components may not match their contribution to binding in the final

products. If the product exhibits tighter binding than do the composite starting materials,

then this binding increases selectivity, but if the binding decreases in the final product,

then selectivity is decreased. The ability of the receptor to accelerate coupling is essential

for amplification and selectivity because there will be no amplification if the library

components also form products outside the receptor's binding pocket. Since the receptor

is not normally a catalyst for this coupling reaction, this can be a challenging

requirement. This coupling reaction should be irreversible, thus, the synthesis is

kinetically controlled. This kinetic control permits high selectivity and enhancements,

although the level of enhancement is difficult to predict.

A new receptor-assisted method, pseudo-dynamic combinatorial chemistry, uses

an irreversible library synthe sis, combined with an irreversible destruction reaction that

regenerates sorne of the starting materials (Figure 1.3). These starting materials are then

re-used in a new round of synthesis. Thermodynamically controlled binding to a receptor

protects strong binding library members from the kinetically controlled destruction

process. Amplification results from the receptor protecting bound library members from

the destruction because in the absence of receptor, the destruction reaction removes aIl

library members. Iterative synthetic cycles allow the tightest binding library members to

build up in the system, thus increasing the absolute amounts. The selectivity cornes

partially from the initial, reversible binding of the library members to the receptor, but

mainly from the kinetically controlled destruction of the weaker binders. The selectivity

can exceed one and increases as the destruction reaction proceeds.

8

The purpose of this introduction will be to give an overview of first dynamic

combinatorial libraries, and then of receptor-accelerated synthesis, with examples as to

how they have been applied to drug discovery. A detailed description of the development

of the destruction process in pseudo-dynamic combinatorial libraries (P-DCLs) will be

the focus of chapter two. The first generation of p-DCLs will be described in chapter

three. The optimization of amplification and selectivity in a p-DCL will comprise chapter

four. Finally, chapter five will deal with both the expansion of this library and theoretical

characterizations of the processes involved inp-DCL systems.

Table 1.1 Drug Discovery Approaches using Receptor-Assisted Combinatorial Chemistry

Approach Description How ~~Gbinding is How How selectivity arises (is used amplification enhancement possible)

arises Dynamic Reversible reaction creates Thermodynamic Bound Stronger binders are bound

--- combinatorial a library; binding to the approach; shift products are to relatively stronger chemistry receptor shifts the equilibrium synthesized degrees (no enhancement)

equilibrium more to balance eguilibrium

Receptor- Binding of components to Kinetic approach; Receptor Stronger binding starting accelerated receptor accelerates increase rate of accelerates materials are more synthesis synthesis ofbest inhibitors synthesis coupling of available for receptor-

starting induced coupling materials due (enhancement difficult and to their impossible to predict) proximity when bound

Pseudo- Formation and destruction Kinetic 1. Receptor 1. Receptor selects stronger dynamic of library are separate destruction protects binders combinatorial irreversible reactions; enhancesthe binders from 2. Destruction culls weak chemistry binding slows destruction thermodynamic destruction binders

reactions selectivity 2. Iterations of 3. Iterations introduce synthesis library members again, allows strong binders have chance compounds to to take the place ofweaker build up (high levels of

enhancement)

9

DCL: Thermodynamic ,- (receptor binding shifts synthetic equilibrium)

0-0 00 Library Synthesis t>--O Receptor

C> 0-<] Re<,pto--l

RAS: Kinetic (receptor binding accelerates synthesis)

... Recepto~ ~ l ........ Recep~---l~~ Receptor-",,,,,

pDCL: Thermodynamic + Kinetic (receptor b. inding slows destruction) (J ~ 0 LibrarySynthesis 9 r Receptor

~'o r-'''O~ Re<eptofh

D Figure 1.3: Three receptor-assisted combinatorial methods. In dynamic combinatorial libraries (DCLs),

because library synthesis and binding to the receptor are reversible, thermodynamic pressures govern both

amplification and selectivity. In receptor-accelerated synthesis (RAS), starting materials with strong

receptor-affinity come together in the active site and form products due to their proximity, forming a

stronger inhibitor due to kinetic pressures. In pseudo-dynamic combinatoriallibraries (P-DCLs), the term

pseudo-dynamic refers to an irreversible library synthesis, coupled with a complimentary irreversible

destruction that takes the place of the reversible library synthesis in traditional DCLs. Kinetic destruction

enhances the thermodynamic selectivity in these systems.

1.3 Receptor Assisted Systems Under Thermodynamic Control:

Dynamic Combinatorial Libraries

In the early 20th century LeChatelier showed that secondary reactions could shift

an equilibrium. U sing this princip le to discover tight binding molecules is more recent. In

the early 1990's, researchers used it to discover DNA-binding molecules, where binding

10

to the biomolecule shifted a synthetic equilibriumP1 Similarly, equilibrating starting

materials can surround a template and form a receptor.[41

1.3.1 DeLs Requiring "Lock-In" Reactions

Ivan Huc and Jean-Marie Lehn first used LeChâtelier's princip le to identify

enzyme inhibitors and identified the key requirements of a dynamic combinatoriallibrary

(DCL) in 1997.[51 The creation of a library should be reversible, occur in aqueous media,

with a detectable shift in equilibrium. Huc and Lehn chose a well-characterized,

inexpensive receptor, carbonic anhydrase, for this proof of principle study.

Arylsulfonamides are known to inhibit carbonic anhydrase by binding to its active site

zinc ion. Their DCL consisted of 12 imines that were formed from the reversible

combinations of four amines and three aldehydes in water at pH 6. One of the aldehydes

contained an aryl sulfonamide that gave sorne products receptor-affinity. To detect these

imines by HPLC without shifting the equilibrium, they used a "lock-in" reaction -

irreversible reduction of the imines with NaBH3CN to the corresponding amines (Figure

1.4).

11

A

B

R' HPLC NaCNBH3 'NH Detection

carbonic anhydrase carbonic anhydrase -imine-

A -reduction of bound imine? H R'

~NH VH~

~SO,NH, mest ampl~ied

HJl.-O.

tr-S03H

~HN-.,/'N Ph 1 II~

CO,H 0 H 0 1 A S03H

o H~CO,H

U H,N'J('N

°H~ y C02H

HN-.,/' Ph""( Il ~

CO,H 0H~ y CO,H

œ~N H 1 ~ ,,&

CO,H

H~ V--S02NH2

Figure 1.4: The first dynamic combinatoriallibrary.151 A) Imines are formed from the aqueous, reversible

combination of amines and aldehydes in aqueous media, and can subsequently bind to the receptor,

carbonic anhydrase. Reduction of the imines with NaCNBH3 to the corresponding amines is necessary to

detect the amplified library members by HPLC. B) 12 library members are created from the combination of

four amines and three aldehydes (one ofwhich contains a receptor-binding aryl sulfonamide).

The reduction step made the results somewhat ambiguous, as it was not clear whether

compounds that were amplified by the receptor, or compounds more able to undergo

reduction (possibly due to reagent accessibility because they were free in solution) were

being reduced. This "lock-in" reduction added a level of complexity undesirable in a

DCL. AU early DCLs required sorne form oflock-in reaction.

12

Although many linking reactions are possible, dynamic combinatorial libraries

most often use formation of hydrazones, imines or disulfides to create the library, with

disulfide exchange being the most common. Disulfides rapidly equilibrate at pHs greater

than 8, but are "locked in" below pH 5. In one early mode! system a library of disulfide-

linked sugar dimers was equilibrated in the presence of the plant lectin, concanavalin A,

which binds mannose-rich oligosaccharides. Equilibration of the disulfides took place in

the presence of the receptor, after which the results were locked in by lowering the pH. [6]

The receptor was immobilized, allowing any unbound compounds to be filtered off,

simplifying the subsequent HPLC analysis. Mannose-containing dimers were the only

ones detected (Figure 1.5).

OH OH

HO~", OH OH ~

HO~cJ

HO~~f6\ HO-S:-?-f6, HO~ HO~ Con A

+ OH OH 0-... ~== ... OHOH cYS

~.\ ~q 1

H~O - ~ HO~oY OH

o 0

sugar-0-o-~~s-s~~-o-o-sugar

Sugars = D-mannose (Man), D-galactose (Gal), D-glucose, L- arablnase (Ara), D-xylase (XyI)

n = 3 for Man and Gal n = 2 for Gal, Gle, Ara and Xyl

OH OH

HO~ï-'6\ HO~

Con A OH OH '"

HO~O S HO 1

J

Figure 1.5: An example of a DeL from O. Ramstrëim and J.-M. Lehn.[6] Top: Disaccharides are attached to

sulfides using various linker lengths, and are equilibrated in the presence of plant lectin concanavalin A.

Bottom: The structure of the linker.

More recent applications of DCLs have found pharmaceutically promlsmg

inhibitors. In a recent example from Therascope Pharmaceuticals (now Alantos

13

Pharmaceuticals), an imine library was created from the condensation of a diamine and

over 50 ketones (Figure 1.6)pl This library was synthesized in the presence and absence

of neuraminidase, a key enzyme in the propagation of the influenza virus. After reduction

of the imines, Le/MS analysis identified several hits. The researchers performed two

control experiments to ensure that amplification of library members was due only to

binding to the neuramidase active site. In one of these controls, the library was

synthesized in the presence of bovine serum albumin (BSA). If amplification of an

apparent hit took place, it was not due to binding to neuraminidase. In another control

experiment, the library was synthesized in the presence of neuraminidase and a known,

potent inhibitor, zanamavir. If amplification of a hit happened in this case, it was not due

to active site binding, and was therefore deemed to be a false positive. One of the hits

(bottom right of Figure 1.6) was eliminated as a possibility by control experiments

because it was amplified in the presence of BSA, and in the presence of neuramidase and

zanamavir.

The DeL experiment gave several hits, however, the relative amplifications were

puzzling. The compound that was most prevalent in the Le/MS trace, with a high level of

amplification, was not a potent inhibitor at aIl. The strongest inhibitor was amplified

approximately three-fold less (Figure 1.6). Explanations ofhow these results are possible

in sorne DeLs will come in section 1.3.3.

14

HO 0

NaCNBH3 9 lC/MS Detection ..

HN" "NH2

~NHAC HO)()

HO 0

,,,.X,,'" IIY 2 R, ......... R2 NHAc

Neuramidase

H20, pH =6 ,~~~" IIY 2 R, ......... R

2 NHAc

Unknown Amplifications

HO 0

"~,Q,~~ ~NHAC

largest amounts detected by lCIMS Amplification Factor> 30 Amplification Factor> 90 (highest) Amplification Factor = 84 KI = 85 nM (strongest binder) ~ = 92 nM KI = 700 nM

HO 0

"",9"" ,-A NHAc HO-1 -

Amplified in DCl but Eliminated by Control Experiments

Figure 1.6: Creation of an imine library against neuraminidase using a diamine and several ketones. (7) The

library gave several hits, identifying potential drug leads against this enzyme. However, the most prevalent,

and the most amplified species were not the strongest binders.

1.3.2 DeL Systems with in situ Detection

Although detecting bound library members us mg HPLC reqU1res a lock-in

reaction, other methods can detect library members in situ. In a recent example,[8] Kubota

et al. used difference NMR to detect amplified members of a receptor library formed

from (en)Pd(N03)2 and several pyridyl-appended bridging ligands around a small guest

mole cule (e.g. CBr4). Although not a drug discovery application, modifications of this

technique might detect drug-like molecules bound to a receptor.

Turner,[9] in the first example of dynamic library synthesis using an enzyme, made

a library of aldols using an aldolase. N-acetylneuraminic acid aldolase (NANA aldolase)

c1eaves sialic acid to N-acetylmannosamine and pyruvate (Figure 1.7). By adding excess

pyruvate, the synthesis equilibrium was pushed towards formation of aldol products.

Amplification of sialic acid resulted when a small (three-membered) library was

15

synthesized and incubated for 160 h in the presence of the plant lectin, wheat germ

agglutinin (WGA). The researchers were able to take aliquots of the reaction mixtures at

various time points, denature both the receptor and the aldolase (which rendered the

library products stable as the aldolase could no longer catalyze the reverse reaction) and

immediate1y analyze them by HPLC.

o

R2 R, ~oNa ~OHOH OH OHOH OH HO~OH 0 3 R ONa WGA Ho .......... ·) ... J~~--ol--ONa

HO 2 2 R 0 ..-===="'" ~ HN=tfV 11 NANA aldolase ' HO 1 0 Aé HO 0

a) R, = NHAc, R2 = CH20H a) R, = NHAc, R2 = CH20H Sialic acid: mes! ampified b) R, = OH, R2 = CH20H b) R, = OH, R2 = CH20H c) R, = OH, R2 = H c) R, = OH, R2= H

Figure 1.7: A DeL made from NANA aldolase catalyzed aldol formation in the presence ofwheat germ

agglutinin (WGA).[9] Amplification of the mM inhibitor, sialic acid (la) occurs at the expense of non

inhibitors 1 band 1 c when W GA is present during library synthesis.

Congreve et al. [10] detected an oxindole bound to cyclin-dependent kinase-2

(CDK-2) by X-ray crystallography (Figure 1.8). The oxindole library formed from

hydrazines and isatins in DMS0I20% H20 in the presence of prote in crystals. Crystal

structures identified which oxindole bound to the protein. An inhibitor with an IC50 of 30

nM was detected directly by observation of its electron density, and a non-inhibitor was

not detected. Although a seemingly amazing result, this research is plagued by several

problems. First, it is not clear how the receptor, present in very low amounts compared to

the amount of products formed, influences the synthesis. The researchers state that a

small amount of products could be diffusing into the crystals, where they undergo

another, separate dynamic equilibrium, however this is speculative. In addition, the

results may be skewed towards detection of compounds that more easily diffuse into the

16

crystals. This seems likely as severallibrary members that were equally as potent as the

one detected were not se en in electron density maps. Finally, the general applicability of

this method is highly questionable as the CDK-2 crystals used in this study are unusually

robust. Most macromolecular crystals would have little chance of standing up to the harsh

conditions required for library synthe sis (80% DMSO in this case).

Direct Detection by X-Ray

Figure 1.8: An oxindole library created in the presence of cyclin-dependent kinase-2 (CDK-2) crystals.1101

Detection of bound library members does not require a lock-in reaction as X-ray crystal structures of the

receptor-inhibitor crystal complex can be analyzed directly. This method, however, may not be widely

applicable as many macromolecular crystals are too fragile to be present during a library synthesis.

A recent modification of thermodynamic receptor-assistance is "tethering" whose

key advantages are first, to focus binding to a particular region on a receptor, and second,

to detect the results of the experiment in situ by running a mass spectrograph of the

equilibrated receptor-inhibitor complex. In this method the thiol of a cysteine residue

(either existing near the site of interest or added by protein engineering) reacted with a

library of disulfides. These disulfides can undergo exchange with the cysteine thiol to

form a receptor-disulfide. The most stable receptor-disulfides were formed from original

disulfides containing fragments that bound the receptor the most tightly. MS analysis of

the receptor-disulfides identified the bound species. [11] By screening 7000 disulfides in

batches of 5-20 compounds, tethering identified a potent inhibitor of interleukin 2, a

17

target in immune-disorder therapy (Figure 1.9).[12] The screening required small batches

to distinguish between species with identical molecular weights, and with similar binding

constants.

1. Introduce Cys Mutations Near Active Site M 10 Mutations. Y31C & l72C Found to be Closest to Binding Site

~-~ 2. DCl Using Disulfide Exchange between Cys-SH and 7000 Disulfides (Tethering)

+ ' ~SH ~

Il-2 l. Disulfide Exchange

S_S-R s-S

~ Binding Stabilizes@'1 Disulfide Bond R

Il-2 Il-2

3. Improve Design of a Known Inhibitorwith Tethering "Hif'

o ~~NH2 H NH N :

N~H i\ o

(Y"o

o ~~NH2 H NH N •

N rH k Fur1her "'1 improvements

IC50 = 3 ~M A('A CI

B 0.2 ~M < IC50 < 3.0 ~M IC50 = 0.06 ~M

Figure 1.9: An example of "tethering" using DCL princip les from Sunesis pharmaceuticalsY2) 1. Crystal

structure-guided mutations introduced a cysteine near the binding site of IL-2. 2. Each mutant is

individually screened against 7000 disulfide-containing fragments (in batches of 5-20) using dynamic

combinatorial interactions. MS identified most stable disulfides. 3. A known inhibitor of IL-2 is improved

50-fold by modifying it with the tethering "hit".

Recent expansions of the tethering approach include extended tethering, which

has been used to find a sub-micromolar inhibitor of cysteine aspartyl protease-3 (caspase-

3). Caspase-3 is a receptor involved in apoptosis and a target in cancer therapy.[13] In

extended tethering an "extender" molecule is used to irreversibly alkylate a cysteine

residue near the enzyme's active site. This extender contains a free sulfhydryl that can

then undergo dynamic interchange with a library of disulfides in a tethering process. The

18

advantage of extended tethering cornes from the variety of extender molecules that can be

added to the receptor. Depending on their size and orientation, these extenders can direct

disulfide exchange towards different binding pockets of a receptor.

Caspase-3 is a protease, and normally binds peptides. It has a series of shallow

binding pockets (Sl-S4) in which individual amino acids normally bind. The extender

molecules used in this study included an aspartyl moiety that allows binding to the SI

pocket. A cysteine residue, known by X-ray structure data to be near this pocket, was

modified with two extenders of different lengths (Figure 1.10). Extender A was shorter,

and was expected to direct binding to the S4 pocket, while extender B would direct

binding to the more distal S2 pocket. Separate dynamic interchange experiments between

caspase-3 with either extender A or Band 7000 disulfides (in batches of between 8-20)

revealed two independent hits by MS, one for each extender. The two compounds, made

from either the extender A disulfide, or extender B disulfide were synthesized in

bioactive form (replacement of the labile disulfide with either methylenes or a sulfone).

They proved to be sub-micromolar inhibitors of caspase-3, and X-ray diffraction showed

that they bound to different pockets of the receptor.

19

1. Modify Receptor (with Native Active Site Cys) with 2 Different "Extender" Molecules

54 Caspase-3B

N CI 0 0

1. ~0~~Y'5.Â.. CI 0 HO C) 8

2 (Extender A)

ÇÇCI 0 ~ 0

OR1. 1 A O~~,~.Â.. I~\ s

CI 0 00 H02C (Extender B)

2. NH30H

o H )lyNY'5H

s~\ on () _ C02H, V <-JI 51 54 52

Caspase-3A

o HO~ )lyN,g ~ ~

h\"O"n -A _ \ ~ V <-JI

51 54 52 Caspase-3B

2. DCl Using Disulfide Exchange between Extender A or B, and 7000 Disulfides (Extended Tethering)

4~'Ç;" n _ 1 \ ~ ~ ~ + ,

51 54 52 ~, Caspase-3A

o HO~ )lyN'g ~ /,

s~ \ éi 5H C02H

51 54 Caspase-3B

R 1 5 + ,

~,

Disulfide ExchangEl.

Disulfide ExchangEl.

3. Replace Bio-Inactive Functionality of Extended Tethering Hits

Figure 1.10: Extended tethering with caspase-3, a protease with several binding pockets (SI_S4).[13] 1.

Irreversible alkylation of a Cys residue in the active site of the receptor with either extender molecule A, or

the longer, B. 2. DCL with 7000 disulfides in sm ail batches with either extender A-modified caspase-3

(Caspase-3A) or Caspase-3B gives independent hits for each extender. 3. Synthesis ofbioactive versions of

the hit molecules.

20

1.3.3 Limitations in DeLs

CUITent research is also addressing sorne of the selectivity and amplification

issues associated with systems under thermodynamic control. The level of selectivity

possible can be a problem in traditional DCLs, as aU compounds made are present to

sorne degree in the equilibrated system. Several compromises have been made to

overcome this problem. As previously described, tethering screens compounds in small

batches and then combines the hits from each batch in a new, enriched experiment. In

another example the selectivity problem was overcome by creating sub-libraries in which

each library left out a particular starting compound. Starting materials whose absence

resulted in no significant amplification of any library members were identified as being

crucial for binding, thus "deconvoluting" the resultsY41 However, large libraries remain

difficult to screen because the level of selectivity possible for even the strongest binding

library member over the next strongest cannot be higher than the ratio of their binding

constants, [15, 161limiting the enhancement level of a DCL to one (see Box 1.1).

The analytical problems caused by this lack of enhancement become more

apparent in large libraries. Although the selectivity between two compounds will remain

the same regardless of the library size, the absolute differences in concentration will

become smaller and smaUer as the library size increases. This aspect of DCLs is

illustrated in the following explanation and derivation.

Dynamic combinatoriallibrary experiments contain two equilibria: an equilibrium

for the synthesis of inhibitors and an equilibrium for binding of the inhibitors to the

receptor (Figure 1.11). The binding equilibrium removes the good inhibitors from the

21

synthe sis equilibrium and, to re-establish equilibrium, the synthesis produces more of the

good inhibitors than it would in the absence of receptor. The following derivation shows

that the ratio of good inhibitor to a poor inhibitor depends linearly on the ratio of the

binding constants.

~D comma" atartlng

materlal.

synthesls

new Inhlbltor. targat - Inhlbltor complexee

Figure 1.11. The dynamic combinatorial library equilibria yield a higher total amount of a good inhibitor.

Binding of the inhibitor to a target removes it from the synthesis equilibrium so the synthesis produces

more of the good inhibitor. However, there are limits to the level ofselectivity a DeL can exhibit.

Given a typical DeL in which library members lA - IN are synthesized under

equilibrium conditions, and then can reversibly interact with a receptor (biological

receptor or target T):

K K SM] SIA

lA a lA IA-T A

SM] K K

SI.

lB a 18 -. IB-T •

SM] K K

SIe

lc ale Ic-T c ..

K K SM] SIN. IN a IN .. I~T

N

and where equations 1-3 define respectively SM]x the starting material used for

synthesizing any inhibitor, Ix, KS]x the equilibrium constant for synthe sis of inhibitor, Ix,

KaJ the equilibrium constant for binding ofinhibitor, Ixto the target, and [Ix ], the total x T

of bound and unbound forms of lx:

22

Equations 1-3 define the concentrations of a particular inhibitor lA :

K = [lA] slA [ SM

IA]

K = [TelA ]

alA [T][IA] (1,2,3)

[IAJ=[IA]+[Te lA]

Then combining equations 1-3 and solving for [lAT] gives equation 4.

(4)

The fraction of [lA] relative to the rest of the components in a library of N members is: T

[lAT] KS/JSM/J{KaIJT] + 1) (5)

[IBT

] + [le T] + ... + [IN T] KsI. [SMI• ](KaIJT] + 1)+ KslJ SMIc ](KadT] + 1)+ ... + KslJ SM/J(KadT] + 1)

Under conditions oftight binding of aH components and an excess oftarget, Kalx [T]» 1

for aH components, equation 5 will simplify to:

The highest selectivity due to receptor binding in DeLs arises when the synthetic

equilibria lie far to the reactant side, and the synthesis equilibria are equal for aH library

members. Under these ideal conditions, the concentration of starting materials is

approximately the same for aHlibrary members and equation 5 simplifies further to:

[lAT] _ K alA

[IBT

] + [leT] + .... + [INT

] - K alB + K alc + .... + KaIN

(7)

Equation 7 shows that the relative concentration of aH lA in the system relative to

aIl other library members at equilibrium is proportional to the ratio of its binding constant

23

to the sum of the binding constants of aH other library members. This would mean that in

a library of 1000 compounds (a practical size for drug discovery), if lA bound ten times

more strongly than aIl other compounds (differences in binding are often much smaller

than this) then lA would comprise only ~ 1 % of the total amount of library members

present. This situation would be aIl but impossible to analyze by conventional methods. It

should be noted that equation 7 depicts an ideal case where either the same starting

material forms aIllibrary members (as in the case of geometrical isomerization), or a case

in which the change in concentration of starting materials is negligible with respect to the

concentration of products. More complex cases are currently being modelled.

Amplification of weak or non-binders can occur in certain types of libraries. In

cases where several different starting materials combine to form potential binders,

products that represent a statistical mixture of these starting materials can be amplified

independently of their binding affinity. For example, it has been shown that in a

theoretical system in which equal amounts of components A, B and C couple to make

trimeric library members (A3, A2B, ABC, AB2 etc), and only component A imparts

receptor-affinity to any member forme d, then a library member with an ABC

composition is amplified 500 fold over one composed of three A units (A3), the

theoretically strongest binder (Figure 1.12).[17] This is a highly undesirable result, as the

amplified compounds should only represent tight binding. [18]

24

~.

A A A R Rillillili ,X. pt-À pt~B .. < < X-X

C B A B / C / 'c A- A-

Theoretical Perturbation: Incorporation of component A gives 300-fold stability (receptor-affinity)

Relative Stability

o Library ~ C omposition

A K-À

900

0.1

B B ' 'c / 'c B- A-

1 300

1.6 45.7

Figure 1.12: Theoretical effect of a DCL in which the combination of three different starting materials (A,

Band C) yields a library oftrimers.[17] If component A (and only A) contained a receptor-binding moiety,

and hence imparted stability to any trimeric product, the amplification of the most stable library member is

actually decreased dramatically. Instead, a product that has only moderate affinity, but represents a

statistical mixture of starting materials is the most amplified compound. This undesirable amplification is

due to entropic effects that favour the formation ofthe ABC trimer over the A3 trimer.

1.4 Receptor Assisted Systems That Add Kinetic Control

Receptor-assisted systems under purely thermodynamic control have not yet been

shown to be able to achieve selectivities above one. Systems that use kinetic factors to

induce the formation of tight-binding compounds can overcome this limitation. Two

methods will be described: receptor-accelerated synthesis and affinity chromatography.

In receptor-accelerated synthesis, a rate of product formation is the key to amplification

and selectivity. Although not yet applied to receptor-assisted combinatorial systems,

affinity chromatography uses the rate of elution of compounds to provide a kinetic-based

resolution.

25

1.4.1 Receptor-Accelerated Synthesis

In receptor-accelerated synthesis (RAS), a library of starting materials

competitively binds to a receptor, and undergoes an irreversible cross-coupling to form a

tight binding ligand. The binding of the starting materials to the receptor brings them

close to one another, speeding up a reaction that would not have occurred in solution.

Thus, only precursors that bind the receptor tightly make products, and the amplification

and selectivity of product formation cornes from both receptor-affinity and the ability of

the receptor to accelerate the coupling.

Early, moderately successful, examples of RAS used either an enzyme's active

site as a template in which ligands were synthesized, or used small molecules as

templates to couple larger molecules. An early example coupled alkyl chlorides to 0.

mercaptotosylamide in the presence of carbonic anhydrase. [19] o.-Mercaptotosylamide has

a sulfonamide group that binds to the active site zinc. Comparisons of HPLC traces with

and without the addition of carbonic anhydrase showed a two-fold increase in selectivity

for binders with a nine-fold difference in binding constants. Thus, in this case the

selectivity was less than the difference in binding constants (enhancement of less than

one).

In a second example, Nicolaou and co-workers dimerized vancomycin analogues

by either disulfide formation or olefin metathesis in the presence of a fragment of the

vancomycin receptor (Figure 1.13). In parallel experiments, one dimer formed two times

faster than a control compound with no receptor affinity.[20] Further parallel experiments

26

showed dimerization rate accelerations directly related to receptor affinity. Had the same

results been obtained in a one-pot system, they would have achieved enhancement levels

approaching one.

1: X=SAc n = 2, 3, 4, 7, 8 R= D-LeuNMe

HO~ ~OH ~N~otb

n:o" :_-~IO:I O~~~ OH 3: N OH 0 H

X = HC=CH2 H N ~N N,R NH H ~ H

n = 1, 2, 3, 7 ~ 0 0 R = H, D-LeuNMe HO 1 0 ~-Ala, L-Asn, y-Abu ~ ...., AH NH2 L-Arg, L-Phe - 1

HO OH

Figure 1.13: Nicolaou's vancomycin monomer scaffold.!20] Dimerizations of various analogues of the

thioacetate 1 using NaOH, H20 and air were faster in the presence of a binding motif of the vancomycin

receptor. Dimerizations of analogues of3 by olefin metathesis were less successful.

Mosbach's group enhanced the rate of a nucleophillic aromatic substitution using

an enzyme active site as a template. The receptor kallikrein induced selection of a new

polyheteroaromatic inhibitor from a five-membered library. This system produced

enhancements approaching 1.5, indicating that the relative amount of a binder to its

closest competing compound was 1.5 times higher than their relative binding constants.

This is higher than the thermodynamically govemed maximum of one possible in a DCL.

Further investigations showed that a molecularly imprinted polymer could increase the

coupling rate using the same starting materials and could provide nearly as high

amplification and selectivity as the natural enzyme. [21]

The most successful example of RAS yielded a femtomolar inhibitor of

acetylcholine esterase (AChE) by optimizing the linker length and orientation between a

27

micromolar and a nanomolar inhibitor of AChE. Two poylaromatic AchE inhibitors

attached to linkers of varying lengths with terminal azides or acetylenes coupled to form

64 new potentially stronger inhibitors. Only linkers of the correct length and in the

correct orientation brought the reactive functionalities of two bound starting materials

close enough for a [3+2] dipolar cycloaddition to occur (Figure 1.14).[22]

"HGI (YNyj ~

HN '(CH2)riN3 AChE

TZ2-6 (n = 2-6)

Figure 1.14: An example of receptor-accelerated synthesis from W. G. Lewis et al.122] Eight azides and

eight acetylenes were combined in the presence of acetylcholine esterase (AChE). Optimizing the linker

length and geometry between a micromolar and a nanomolar inhibitor allowed a [3 + 2] dipolar

cycloaddition to form a femtomolar inhibitor. This was possible because both starting materials could bind

to adjacent sites on the receptor.

1.4.2 Limitations in RAS

The two challenges of RAS are binding the starting materials to the receptor and

significantly speeding up their coupling. For amplification to occur, the receptor must

tightly bind the two starting materials simultaneously. This requirement could eliminate

cases where two weakly binding species link to form a strong binding one. Similarly,

optimization of substituents around a tightly binding mole cule many be difficult with

RAS if the substituent fragments do not bind weIl. RAS is probably best suited to

28

optimize linkers between two molecules that bind weIl at adjacent sites. The need to

significantly speed up product formation is also challenging because the receptor is not

necessarily a catalyst for the desired reaction simply by holding the two reactants close to

one another. To detect this rate acceleration and to minimize the spontaneous reaction in

solution, researchers have not carried out RAS with aIl library components present

simultaneously. Rather, they tested each combination of starting materials in a series of

parallel experiments.

1.4.3 Affinity Chromatography

For decades researchers have used affinity chromatography to isolate tight

binding compounds from a static mixture. Recent advances in affinity chromatography

such as affinity capillary electrophoresis[23] and frontal affinity chromatography coupled

with mass spectrometry (F ACMSi24] continue to extend the usefulness of these methods.

Applying affinity chromatography to a dynamic mixture of compounds requires either a

lock-in reaction, or sorne other form of deconvolution to permit identification of the tight

binding compounds. However, using the elution of the mixture as a kinetic step in

removing unbound species can lead to higher levels of selectivity than is possible in

DCLs.

In an early example of using affinity chromatography to select a tight binder from

an equilibrating mixture, Miller and co-workers combined eight salicylaldimines with

zinc dichloride to form a library of 36 bis(salicylaldiminato )zinc complexes on an affinity

column of cellulose resin on which poly (dA-T) DNA was immobilized (Figure 1.15).[25]

29

After equilibration, the mixture was eluted from the column, and was expected to be rich

in non-binding complexes. The complexes were then hydrolyzed with TF A to the

salicylaldimine "monomers" which were then derivitized with 2-naphthoyl chloride (for

UV detection). Subsequent HPLC analysis of the eluate was expected to show the

monomers that did not lead to DNA-binding Zn complexes. Subsequent control

experiments and analysis revealed that one of the most tightly retained monomers formed

a complex that exhibited an inhibition constant of 1 /-lM. Other monomers that were not

detected by HPLC were eliminated as possible hits by the control experiments, which

showed that they either bound to the column as monomers, rather than as complexes, or

that they bound to cellulose rather than to DNA.

Incubation on Affinity Column

1. Elution of Unbound Complexes 2. Lyophilization

Jrtzn~;'N~ (J-(/' ·"o-<=>

1. TFA (hydrolysis of Zn complex) 2. 2-naphlhoyl chloride (derivitizaton for UV deleclion) 3. Separation/Delection (HPLC)

~N·R ~OH

Monomers that do not form DNA-Binding Complexes

R = (CH2l:!CH20H

(CH2)2CH20Me

H2C~ (R&S) N

H2C

UO 1

(R&S)

H2C~ H2C~F

~1?: v.-.o··~:Zn'r:o:Q.

Sirongest DNA-Binding Complex (Kt = 11JM)

Figure 1.15: The bis(salicylaldiminato)zinc DNA-binding complex affinity column experiment.[25] 36

complexes reversibly form on a column ofpoly (dA-T). Elution of the mixture should be rich in complexes

that do not bind to the column. Detection of the salicylaldimine monomers ofthese complexes shows which

monomers do not lead to productive complexes, and allows for identification of strong binding ones.

30

Eliseev and Nelen used iterative affinity chromatography in conjunction with UV

induced isomerization of dienoic acids to enrich the arginine-binding compounds in a

mixture. The cis and trans alkene isomers of dienoic acids were passed through a column

where they could bind to immobilized arginine (Figure 1.16).[26] The eluted dienoic acids,

enriched in the less-tightly binding trans isomers, were photoisomerized and passed

through the column again. After thirty cycles of Arg binding and UV -induced

isomerization of eluted library members, the amount of the tightest binding, cis, cis

compound was 50% greater than after one cycle. The enhancement could be as high as

~2.7 for the tightest binding, cis,cis compound. This enhancement is higher than that

possible in a thermodynamic system because this experiment included a kinetic

component - the elution of more weakly bound compounds, and used iterative cycles to

increase the enhancement achieved through one binding/elution/isomerization cycle.

Iterative chromatography systems require immobilization of a receptor onto a column,

and a process that can randomize the compounds in the column eluate. This method has

not yet been used in a drug discovery application, but needs only the above two

conditions to be fulfilled before it can be.

31

0- -0 cis.cis: Strongest Binder

Kinetic flow rate

Figure 1.16: The affinity column-UV generator loop by Nelen and Eliseev.[26) A mixture of isomers is

passed through the arginine column. UV-induced isomerization to their most thermodynamically stable

state occurs only to compounds that do not bind. Re-introduction of the mixture to the column adds more

tight binders. Combining the thermodynamic binding with the rate of introduction of the new library

compounds (kinetic) results in amplification and high selectivity for the strongest arginine binder.

References

1. M. A. Rouhi, Chem. Eng. News, 2003, 81 (41), 77-78, 82-83, 86, 88-91.

2. S. Borman, Chem. Eng. News, 2003, 81 (43),45-56 and references therein.

3. 1. T. Goodwin, D. G. Lynn, J Am. Chem. Soc. 1992,114,9197-9198.

4. a) A. V. Eliseev, M. 1. Nelen, J Am Chem Soc. 1997,119, 1147-1148; b) H.

Hioki, W. C. Still, J Org. Chem. 1998,63, 904-905; c) A. V. Eliseev, M. 1.

Nelen, Chem. Eur. J, 1998,4, 825-834; d) V. Berl, 1. Huc, J.-M. Lehn, A.

DeCian,1. Fischer, Eur. J Org. Chem. 1999,3089-3094; e) G. R. L. Cousins, S.-

A. Poulsen, J. K. M. Sanders, Chem. Commun. 1999,16, 1575-1576; t) S. Otto,

R. L. E. Furlan, J. K. M. Sanders, J Am. Chem. Soc. 2000,122, 12063-12064; g)

R. L. E. Furlan, G. R. L. Cousins, 1. K. M. Sanders, Chem. Commun. 2000, 18,

1761-1762; h) R. L. E. Furlan, Y.-F. Ng, S. Otto, J. K. M. Sanders, J Am. Chem.

Soc. 2001,123,8876-8877; i) G. L. Cousins, R. L. E. Furlan, Y.-F. Ng, J. E.

32

Redman, l K. M. Sanders, Angew. Chem. Int. Ed. Engl. 2001, 40, 423-428; j) S.

Otto, R. L. E. Furlan, J. K. M. Sanders, Science 2002, 297, 590-593; k) R. L. E.

Furlan, Y.-F. Ng, G. L. Cousins, J. E. Redman, J. K. M. Sanders, Chem Commun.

2002,58,771-778; 1) S. L. Roberts, R. L. E. Furlan, G. L. Cousins, J. K. M.

Sanders, Tetrahedron 2002, 9, 938-939; m) E. Stulz, S. M. Scott, A. D. Bond, S.

l Teat, J. K. M. Sanders, Chem. Eur. J 2003,9,6039-6048; n) A. L. Kieran, A.

D. Bond, A. M. Be1enguer, l K. M. Sanders, Chem. Commun.2003, 21,2674-

2675; 0) K. D. Johnstone, N. Bampos, l K. M. Sanders, M. l Gunter, Chem.

Commun. 2003,12, 1396-1397; p) S. L. Roberts, R. L. E. Furlan, S. Otto, J. K. M.

Sanders,Org. Biomolec. Chem. 2003,1, 1625-1633; q) L. Ordriozola, N.

Kyritsakas, J.-M. Lehn, Chem. Commun. 2003, 1, 62-63; r) L. H. Uppadine, J.-M.

Lehn, Angew. Chem. Int. Ed. Engl. 2003,240-243; s) V. Patroniak, P. N. W.

Baxter, J.-M. Lehn, M. Kubicki, M. Nissinen, K. Rissanen, Eur. J Inorg. Chem.

2003,22,4001-4009; t) W. G. Skene, E. Couzigne, J.-M. Lehn, Chem. Eur. J

2003, 9, 5506-5566.

5. 1. Huc, J.-M. Lehn, Proc. Nat. Acad. Sei. USA, 1997, 94, 2106-2110.

6. O. Ramstrom, l-M. Lehn, ChemBioChem, 2000, 1,41-48.

7. M. Hochgürtel, R. Biesinger, H. Kroth, D. Piecha, M. W. Hofmann, S. Krause, O.

Schaaf, C. Nicolau, A. V. Eliseev, J Med. Chem. 2003,46, 356-358.

8. Y. Kubota, S. Sakamoto, K. Yamaguchi, M. Fujita, Proc. Nat. Acad. Sci. USA,

2002, 99, 4854-4856.

9. R. J. Lins, S. L. Flitsch, N. J. Turner, E. Irving, S. A. Brown, Angew. Chem. Int.

Ed. Engl. 2002,41,3405-3407.

33

10. M. S. Congreve, D. J. Davis, L. Devine, C. Granata, M. O'Reilly, P. G. Wyatt, H.

Jhoti, Angew. Chem. Int. E.d Engl. 2003,42, 4479-4482; Angew. Chem.2003,

115,4617-4620.

Il. D. A. Erlanson, A C. Braisted, D. R. Raphael, M. Randal, R M. Stroud, E. M.

Gordon, J. A Wells, Proc. Nat/. Acad Sei. USA 2000, 97, 9367-9372.

12. A C. Braisted, J. D. Oslob, W. L. Delano, J. Hyde, R S. McDowell, N. Waal, C.

Yu, M. R Arkin, B. C. Rairnundo, J. Am. Chem. Soc. 2003,125,3714-3715.

13. D. A Erlanson, 1. W. Larn, C. Wiesrnann, T. N. Luong, R. L. Sirnrnons, W. L.

Delano,1. C. Choong, M. T. Burdett, W. M. Flanagan, D. Lee, E. M. Gordon, T.

O'Brien, Nature Biotechno/. 2003,21,308-314.

14. T. Bunyapaiboonsri, O. Rarnstrorn, S. Lohrnan, J.-M. Lehn, L. Peng, M.

Goeldner, ChemBioChem, 2001, 2, 438-444.

15.1. S. Moore, N. W. Zirnrnerrnan, Org. Lett. 2000,2,915-918.

16. A V. Eliseev, M.1. Nelen, J. Am. Chem. Soc. 1997,119, 1147-1148.

17. Z. Grote, R. Scopelliti, K. Severin, Angew. Chem. Int. Ed Engl. 2003,42, 3821-

3825; Angew. Chem. 2003,115,3951-3955.

18. For other reviews, see: a) A. Ganesan Angew. Chem. Int. Ed Engl. 1998,37,

2828-2831; Angew. Chem. 1998, 110, 2989-2992; b) 1. Huc, 1.-M. Lehn Actualité

Chimique 2000,51-54; c) C. Karan, B.L. Miller Drug Discov. Today 2000,2,67-

75; d) R Nguyen, 1. Huc Comb. Chem. High-Throughput Screen. 2001,4,53-74;

e) J.-M. Lehn, AV. Eliseev Science 2001,291,2331-2332; f) O.Rarnstrorn, J.-M.

Lehn Nat. Rev. Drug Disc. 2002, l, 26-36; g) S. Otto, RL.E. Furlan, J.K.M.

Sanders Curr. Op. Chem. Biol., 2002, 6, 321-327. h) O. Rarnstrorn, T.

34

Bunyapaiboonsri, S. Lohman, J.-M. Lehn Biochim. Biophys. Acta 2002, 1572

178-186.

19. R. Nguyen, I. Huc, Angew. Chem. Int. Ed. Engl. 2001,40, 1774-1776; Angew.

Chem. 2001,113,1824-1826.

20. K. C. Nicolaou, R. Hughes, S. Y. Cho, N. Winssinger, C. Smethurst, H.

Labischinski, R. Endermann, Angew. Chem. Int. Ed. Engl. 2000,39, 3823-3828;

Angew. Chem. 2000,112,3981-3986

21. Y. Yu, L. Ye, K. Haupt, K. Mosbach, Angew. Chem. Int. Ed. Engl. 2002,41,

4459-4463; Angew. Chem. 2002, 114,4639-4643.

22. W. G. Lewis, L. G. Green, F. Grynszpan, Z. Radic, P. R. Carlier, P. Taylor, M. G.

Finn, K. B. Sharpless, Angew. Chem. Int. Ed. Engl. 2002,41, 1053-1057; Angew.

Chem. 2002,114, 1095-1099.

23. Y.-H. Chu, L. Z. Avila, J. M. Gao, G. M. Whitesides, Ace. Chem. Res. 1995,28,

461-468.

24. D. C. Schriemer, D. R. Bundle, L. Li, O. Hindgaul, Angew. Chem. Int. Ed. Engl.

1998,37,3383-3387; Angew. Chem. 1998,110,3625-3628.

25. B. Klekota, M. H. Hammond, B. L. Miller Tet. Lett. 1997,38,8639-8642.

26. M. 1. Nelen, A. V. Eliseev, Chem. Eur. J 1998,4,825-833.

35

CHAPTERTWO

AMPLIFICATION OF SCREENING SENSITIVITY THROUGH SELECTIVE

DESTRUCTION: THEORY AND SCREENING OF A LIBRARY OF CARDONIC

ANHYDRASE INHIBITORS

36

Abstract

Application of RACC to larger libraries is chaUenging because of the need to

accurately measure the amount of each inhibitor. In this chapter, we dramatically simplify

this analysis by adding a reaction that destroys the unbound inhibitors. This works similar

to a kinetic resolution, with the best inhibitor being the last one remaining. We

demonstrate this method for a static library of several sulfonamide inhibitors of carbonic

anhydrase. Four sulfonamide containing dipeptides, 4a-d, were prepared and their

inhibition constants measured. These inhibitors migrated to the carbonic anhydrase

compartment of a two-compartment vessel. Although higher concentrations of the better

inhibitors were observed in the carbonic anhydrase compartment, the concentration

differences were smaU (1.83 : 1.71 : 1.54 : 1.46 : 1 for 4a : 4b : 4c : 4d : 5, where 5 is a

non-inhibiting dipeptide EtOC-Phe-Phe). Addition of a protease rapidly cleaved the

weaker inhibitors (4d and 5). Intermediate inhibitor 4c was cleaved at a slower rate and

at the end of the reaction, only 4a and 4b remained. In a separate experiment, the ratio of

4a to 4b was found to increase over time to a final ratio of nearly 4: 1. This is greater than

the ratio of their inhibition constants (approx. 2:1). The theoretical model predicts that

these ratios would increase even further as the destruction proceeds. This removal of

poorer inhibitors simplifies identification of the best inhibitor in a complex mixture.

2.1 Introduction

Synthesis using combinatorial chemistry (CC) aUows testing of hundreds of

thousands of drug candidates using high throughput screening (HTS) techniques.

37

Although this rapid pace has changed drug development, the search for faster and more

efficient testing methods continues due to the lack of drugs found using CC coupled with

HTS,. One promising method is in situ screening of mixtures such as in dynamic

combinatoriallibraries. [1] Dynamic combinatorial libraries are equilibrating mixtures of

organic molecules. Equilibration in the presence of a therapeutic target increases the

equilibrium amounts of those library members that bind tightly to that target. The

difference in library composition with and without a stoichiometric amount of target

identifies the best inhibitors.

To render dynamic combinatorial chemistry practical in drug discovery, methods

must be developed to screen dynamic combinatoriallibraries with thousands of members.

This screening is complicated because it is often difficult to measure the concentration of

each library member in the absence and presence of a target. Further, the libraries will

likely contain not one, but many good inhibitors because many library members have

similar structures and thus similar binding constants. In these cases, adding the target

increases the concentration of many library members, rather than a single member, and

makes analysis very difficult or impossible. Eliseev and Nelen[2] estimated that a dynamic

library combined with an affinity column containing the target would yield one major

compound (>50%) only if KstronglKweak was at least n, where n is the number of

members of the library. Thus, for one member to predominate in a library of 1000

members, that member must would have to bind > 1000 times stronger than the others, an

unlikely possibility. This inability to distinguish between inhibitors of similar binding

constants is a major limitation of the current dynamic combinatoriallibraries.

38

This chapter describes a screening method that enhances the ability to detect the

best inhibitor in a mixture of similar inhibitors. The key to the method is an irreversible

destruction reaction that destroys the unbound and weakly bound inhibitors, similar to a

kinetic resolution. The best inhibitor is the one in greatest concentration after a certain

period of time has been allowed for destruction of more weakly bound library members.

The necessary period of time would presumably be determined by the binding strengths

of the inhibitors. We demonstrate that this method works for a static library and discuss

its potential application to a dynamic system.

Our library targets carbonic anhydrase and consists of dipeptides with an N-

terminal 4'-sulfonamidophenylalanine (1, Phesa)Yl These dipeptides can either bind to

carbonic anhydrase or be cleaved by a protease (Scheme 2.1). This cleavage increases the

ratio of the strongest binder relative to weaker binders. Importantly, the ratio may

increase to values significantly greater than the ratio of the binding constants, thus

overcoming the limitation identified by Eliseev and N el en and making it easy to identify

the best inhibitor in the mixture.

~ 0 83

Ji ..;...... selective

R{' N C02H pressure H ~

1-';::'

R ,Q

, , , , , , ,

!lœE) : carbonic i anhydrase , 1 l hydrolysis of

poorer inhibitors

membrane

ffJ O

H 83

2 + H2NÀ

C02H

1'<:::

R1

,Q

1 R1 = S02NH2 R2 = H

Scheme 2.1: Aryl Sulfonamide-Based Dipeptide Libraries as Inhibitors of Carbonic Anhydrase. Strong

binding inhibitors will be bound to carbonic anhydrase and protected. Weaker inhibitors will be hydrolyzed

by a protease.

39

2.2 Theory: Finding the Best Inhibitor by Destruction of Poorer Inhibitors

Section 1.2.3 (Limitations in DCLs) outlined the theoreticallimits in selectivity of

a traditional DCL. One way to enhance the concentration differences between inhibitors

with similar binding constants is to add an irreversible reaction that removes the

unbound, poorer inhibitor (Scheme 2.2). This situation is similar to a kinetic resolution of

enantiomers. As the destruction reaction winnows away the poorer inhibitors, the relative

concentration of the best inhibitor increases exponentially. The analysis below is similar

to that for kinetic resolutions. [4]

tergal· Inhlbltor complexes

dlSloclatlon D+O destruction DO .. [7 <l

Inhlbltora

Scheme 2.2: Destruction of inhibitors. The free concentration of the poorer inhibitor is higher, thus it is

destroyed more readily. This destruction reaction exponentially increases the relative concentration of the

good inhibitor similar to a kinetic resolution.

Consider two inhibitors, lA and lB, that compete for a target, T, and are also

destroyed by an irreversible reaction to yield P and Q with rates of k dl and k dl A B

40

Kdl k -""--- T-/A --!.a.... T / dIA P (1) +A~

K k T / dlB T 1 dIB Q (2)

-B -- +B-

The rate of disappearance of inhibitor lA is

dr/A] = -k [I 1 (3) dt dIA A

if [lA] is the total concentration ofbound and unbound forms ofinhibitor lA, it can be T

shown that

(4)

Upon solving for lA and substituting into equation 3, the rate of disappearance of lA is

given by

d[IAl kdlA Kdl)IA) --=-

KdI + [Tl A

dt (5)

Under our experimental conditions the concentration of the free target, T, will be much

larger than the dissociation constant, K dlA typical of our library members, so [ 11 » K dIA ,

therefore equation 5 simplifies to

dt [Tl (6)

A similar equation is obtained for inhibitor lB. The ratio oftheir partial reaction rates is

(7)

This equation shows that the relative rate of disappearance of the two inhibitors depends

linearly on their total concentration, their relative binding ability and their relative rate of

destruction. For simplicity, we define S as the product of the relative binding abilities and

41

relative rates of destruction of the two inhibitors. If the rates of destruction of the two

inhibitors are equal, then S is the ratio of the inhibition constants and will be greater than

one if lA is a stronger inhibitor of the target than lB.

Upon integration of equation 7, one finds that the ratio of the total amounts of the

two inhibitors varies exponentially with S (equation 8), where [lA ] represents the initial T 0

total concentration of inhibitor lA. This exponential relationship enhances the ability to

detect small differences as the destruction reaction progresses.

In([JBT 1/[JBT ]0) = S

In([JAT 1/[JA

T ]0)

(8)

By measuring the relative concentration of the two inhibitors during the

destruction reaction, the value of Scan be determined using equation 8. Alternatively,

equation 9 below, which expresses [lA ], [ lB ], [ lA ] , and [ lB] in terms of the T T TO TO

conversion, C, and the ratio of the total concentrations of the two inhibitors, can be used

to determine S.

ln[(l- C)(~)] = S

ln[(l- C)(l :RR)] (9)

These predictions are shown graphically for several values of S in Figure 2.1. If

the rates of destruction of the two inhibitors are equal, then S is the ratio of the inhibition

constants. As the destruction reaction proceeds (conversion increases from zero to one),

the ratio of the total amounts of the two inhibitors, [lA ]/[ lB ], varies when S is not equal T T

to one. When S is large (e.g., 40), the relative concentration of the good inhibitor

increases steeply near 50% conversion. When S is small (e.g., 2), the relative

concentration of the good inhibitor increases steeply near 90% conversion. In either case,

42

the ratio of the total amounts of the two inhibitors, [lA ]/[ lB ], can be much larger than T T

the value of S.

o 0.1 02 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Conversion

Figure 2.1: Predicted ratio of the total (bound and unbound) concentrations oftwo hypothetical inhibitors,

lA and lB as a function of the degree of conversion for given values of S. The degree of conversion is the

fraction of the total amount of inhibitors that have been destroyed. The calculated lines follow equations 8

and 9 where the initial total concentration of the inhibitors is one. This graph shows that the ratio of the two

inhibitors can be much larger than the value of S, even for values of Sas low as 2.

2.3 Results

2.3.1 Synthesis of 4'-Sulfonamidophenylalanine Dipeptides

(S)-4'-Sulfonamidophenylalanine (1 or PhesJ was prepared from (S) -N-

acetylphenylalanine by a modification of the procedure described by Escher et al. [5) Thus,

chlorosulfonylation of N-acetylphenylalanine in chlorosulfonic acid at 60 oC followed by

ammonolysis afforded N-acetyl-4'-sulfonamidophenylalanine. Direct purification of this

intermediate proved difficult. Therefore it was deacetylated using hog kidney acylase 1,[6)

43

and the resulting free amino acid 1 was purified by ion exchange chromatography and

recrystallization. Using this procedure, 1 was prepared as an analytically pure solid in

40% yield from N-acetyl-Phe. The a-amino group was selectively blocked using ethyl

chloroformate under standard Schotten-Baumann conditions. The requisite dipeptides

were then prepared by coupling 2 with tert-butyl amino acid esters using EDCIHOBTpl

followed by trifluoroacetic acid mediated deprotection of the ester function to afford

dipeptides 4a-d (Scheme 2.3). No acylation of the sulfonamide nitrogen was observed

under either the Schotten-Baumann or peptide coupling conditions. Dipeptide Etoc-Phe-

Phe (5), which does not contain a sulfonamide group and serves as a control, was

prepared by standard methods.

1:1 0 RI

Et02CJr~~02R2

H2N's~ d''b 2

TFA CH-"I r 3 R = t·Bu • "'" 2 L-. 4 R2 = H

5

a)RI=Bn b) RI = H

c) RI = i-Bu

Scheme 2.3: Preparation of 4'-Sulfonamidophenylalanine dipeptides.

2.3.2 Inhibition of Carbonic Anhydrase

Sulfonamides 1 and 2 as weIl as sulfonamide dipeptides 4a-d aIl inhibited the

carbonic-anhydrase-catalyzed hydrolysis of 4-nitrophenyl acetate (NP A). The inhibition

was competitive and Lineweaver-Burk plots revealed similar inhibition constants, which

44

varied by only a factor of 10 (Table 1). The parent amino acid 1 (Phesa) was the poorest

sulfonamide inhibitor (KI = 13 /lM), while dipeptides 4a (Etoc-Phesa-Phe) and 4b (Etoc-

Phesa-Gly) were the best sulfonamide inhibitors (KI = 1.2 /lM, 2.5 /lM respectively).

Dipeptides 4c (Etoc-Phesa-Leu) and 4d (Etoc-Phesa-Pro) showed slightly higher inhibition

constants (4.4 /lM, 9.4 /lM respectively). Other simple sulfonamides also inhibit carbonic

anhydrase with similar inhibition constants. [8) As expected, the dipeptide lacking a

sulfonamide group, 5, did not inhibit carbonic anhydrase.

Table 2.1: Inhibition of carbonic anhydrase by sulfonamides 1 and 2, sulfonamide dipeptides 4a-d, and

dipeptide 5.

Compound KI (/lM)a

Phesa (1) 13 ± 1.6

Etoc-Phesa (2) 12 ± 1.4

Etoc-Phesa-Phe (4a) 1.2 ± 0.2

Etoc-Phesa-Gly (4b) 2.5 ± 0.5

Etoc-Phesa-Leu (4c) 4.4 ± 0.7

Etoc-Phesa-Pro (4d) 9.4 ± 1.6

Etoc-Phe-Phe (5) »1000b

a Competitive inhibition constants for the carbonic anhydrase-catalyzed hydrolysis of p-nitrophenyl

acetetate (PNPA) at 25°C in phosphate buffer (10 mM pH 7.5). A typical procedure was to add carbonic

anhydrase solution (lOO.O flL, 0.05 mg/mL) containing inhibitor (0.0-100.0 flM in most cases) to an

acetonitrile solution of pNPA (5.0 flL, 2.0-32 mM) and follow the release of p-nitrophenoxide

spectrophotometrically at 404 nM. b No inhibition detected at an inhibitor concentration of 1 mM.

45

2.3.3 Selective Extraction of Inhibitors By Carbonic Anhydrase

First, we demonstrated that a strongly binding inhibitor concentrates into the

carbonic anhydrase-containing compartment ofa two-compartment vessel (c.f., figure 2.4

without Pronase). The two compartments were created by suspending a dialysis bag

containing a solution of bovine carbonic anhydrase[9] (~0.33 mM, 10 mg/mL) in a

solution of phosphate buffer. The dialysis membrane (12 kDa cutoff) separated the two

compartments so that small molecules such as the sulfonamide dipeptides could diffuse

freely across the membrane, while carbonic anhydrase (30 kDa) could not. Both

compartments initially contained a mixture of 0.16 mM sulfonamide dipeptide 4a and

0.19 mM non-inhibitor dipeptide 5. Over several hours the total sulfonamide

concentration increased in the inside compartment containing carbonic anhydrase and

decreased in the outside compartment as analyzed by HPLC (Figure 2.2). On the other

hand, the concentrations of the non-inhibitor 5 remained similar in both compartments.

This result showed that tight binding to a target could concentrate a good inhibitor into

one compartment of a two-compartment reaction vessel.

46

[dipeptide) (mM)

0.3 Outside Chamber

-e~~------____ 5 0.2

0.1

------4a o+-----~----~----~----~

o 3 6 Time (h)

9 12

Figure 2.2:(10) Selective concentration of the sulfonamide 4a over non-inhibitor 5 into the carbonic-

anhydrase-containing compartment of a two-compartment vesse!. One compartment contained carbonic

anhydrase (0.34 mM), while both compartments (20 mL each) initially contained equal concentrations of

sulfonamide 4a (0.16 mM) and non-inhibitor 5 (0.19 mM). The sulfonamide diffused freely across the

dialysis membrane and concentrated in the carbonic-anhydrase-containing compartment as shown. In

contrast, the concentrations of non-inhibitor 5 remained similar in both compartments. After 12 h, the

concentration of sulfonamide 4a in the outside compartment decreased to 0.04 mM and increased in the

inside compartment to 0.28 mM (total of free and carbonic anhydrase-bound). The final ratio of 4a to 5 in

the carbonic anhydrase chamber was 1.75:1.

In a similar experiment using a mixture of inhibitors, we could further detect

differences in relative inhibition constants. A more tightly binding inhibitor concentrated

in the carbonic anhydrase compartment to a greater extent than a less tightly binding

inhibitor. Starting with an equimolar mixture of sulfonamide dipeptides 4a-d and the

non-inhibitor 5 in both compartments, the sulfonamide dipeptides concentrated into the

carbonic anhydrase compartment, Figure 2.3. The fraction of dipeptide in the carbonic

anhydrase compartment varied: 88% for 4a, 82% for 4b, 74% for 4c, 70% for 4d, and

48% for the non-inhibitor 5 (or a ratio of 1.83 : 1.71 : 1.54 : 1.46: 1 for 4a : 4b : 4c : 4d :

47

5). The order of highest to lowest concentration in the carbonic anhydrase chamber

reflects the order of the binding constants of the inhibitors.

20 a) Outside Chamber

16

12L-.-________ -+ __________ ~

[dipeptide 1 (mM) 8

5

4d 4c

4 4b 4a

o+----.----.---~----~---,

o 10 20 30 40 50

Time(h)

20

16

12 [dipeptide)

(mM) 8

4

0

b) Inside Chamber 4a 4b

~ ___ --::::::64c 4d

r-.----------+-----------s

0 10 20 JO 40 50

Time(h)

Figure 2.3:[101 Selective concentration of the sulfonamides 4a-4d over non-inhibitor 5 into the carbonic-

anhydrase-containing compartment of a two-compartment vessel separated by a dialysis membrane. One

compartment contained carbonic anhydrase (0.485 mM), while both compartments (20 mL each) initially

contained equal concentrations of sulfonamides 4a-4d and non-inhibitor 5 (-0.11 mM each). The

sulfonamides diffused freely across the dialysis membrane and concentrated in the carbonic-anhydrase-

containing compartment. In contrast, the concentration of non-inhibitor 5 increased slightly in the outer

compartment.

These results show that is possible to distinguish between inhibitors, but the

differences in concentration are small, especially among inhibitors of similar strength.

Even comparing the best inhibitor (4a) with a non-inhibitor (5) gives a concentration

differing by less than a factor of two. To enhance this difference in concentration, we

explored the use of proteases to destroy the poorer inhibitors.

48

2.3.4 Selective Protection of Inhibitors by Carbonic Anhydrase from Hydrolysis

2.3.4.1 Screening of Proteases.

We screened twenty-two commercially available proteases to identify those that

could hydrolyze the dipeptide Etoc-Phesa-Phe (4a). All proteases showed sorne activity.

Using 0.1 mg ofprotease and 2 /lmol (2 mM) dipeptide 4a, the five most active proteases

(a-chymotrypsin, protease from Streptomyces casepitosus, proteinase K, Pronase from

Streptomyces grise us, protease from Bacillus thermoproteolyticus rokko) cleaved all of

the dipeptide within twenty four hours, while two moderately active proteases (protease

N "Amano", protease from Bacillus polymyxa) cleaved all of the dipeptide within forty

eight hours. The remaining proteases cleaved less than half of the dipeptide after seventy

two hours. We chose one of the most active yet inexpensive enzymes, Pronase from

Streptomyces griseus, for subsequent experiments. Pronase was found to cleave all five

dipeptides (4a-d and 5), although the glycine and proline dipeptides (4b and 4d) were

cleaved more slowly (80-90% hydrolysis within 24 h). To ensure high cleavage rates,

larger amounts of Pronase were used in the competitive degradation experiments

described below.

2.3.4.2 Selective Protection of Inhibitors

We compared the ability of carbonic anhydrase to prote ct sulfonamide inhibitor

4a from hydrolysis while allowing a non-inhibitor, 5, to be hydrolyzed. An experiment

49

similar to that described above, except with Pro nase added to the outer chamber was set

up (Figure 2.4). In the Pronase containing chamber, both dipeptides were rapidly cleaved

to the constituent pieces within 15 minutes. On the other hand, the inside compartment

showed a steady decrease in the concentration of non-inhibitor 5 over 12 h (Figure 2.5),

while the concentration of sulfonamide 4a remained nearly constant (a decrease of 9%

over 12 h). [11] After even just 6 h, the ratio 4a to 5 in the inside compartment was 3.7 : 1

and continued to increase to greater than 20:1 after 12 h. By comparison, the experiment

that does not contain Pronase had a final ratio of 4a to 5 of 1.75 : 1.

Outer Cham ber Inner Cham ber

l~!!a ~ HO, • H NHEtOC

l 1"'" : R1 R,

jl 1 carbonic lil anhydrase e' .' Il lI:

~l pronase 1

~!!a

Ho, • HEtOC

H • CA 1'" ,

Inhlbltor 1 CA comple.

Figure 2.4: Reaction design for the selective destruction experiments. The dipeptides can diffuse across the

dialysis membrane into either chamber. One chamber contains carbonic anhydrase, the other contains

Pronase. Dipeptides in the Pronase cham ber are rapidly cleaved to their constituent pieces. Carbonic

anhydrase prevents strong binding dipeptides from diffusing across the membrane and thus slows their

hydrolysis.

50

0.2 Inside Chamber

0.16 n-:-----.--.----=----"""4a

0.12 [dipeptide]

(mM) 0.08

0.04

0+-__ .-__ ~ ____ ~== __ ~5

o 3 6

Time(h)

9 12

Figure 2.5:[10] Selective protection from hydrolysis of sulfonamide 4a over non-inhibitor 5 by carbonic

anhydrase. A vessel containing two compartments of equal volume (20 mL each) separated by a dialysis

membrane was filled with a solution of. sulfonamide 4a (0.16 mM) and non-inhibitor 5 (0.19 mM). The

inside compartment contained carbonic anhydrase (0.34 mM), while the outside compartment contained

Pronase. The protease rapidly cleaved the dipeptides in the outside compartment to the corresponding

amino acids (data not shown). The non-inhibitor 5 diffused freely across the dialysis membrane and was

cleaved by the protease. In contrast, the inhibitor 4a bound to carbonic anhydrase in the inside

compartment was not consumed at a significant rate. After 6 h, the concentration of sulfonamide 4a in the

inside compartment decreased by only 6% (0.15 mM), while the concentration ofnon-inhibitor 5 decreased

to 0.041 mM during the same time period (ratio = 3.7:1).

In a similar experiment, dipeptides 4a and 4b, which have very similar binding

constants, were exposed to carbonic anhydrase and Pronase. In this experiment, the

dipeptides were placed only in the carbonic anhydrase chamber and an excess of carbonic

anhydrase was used (1.6:1 ratio of CA to dipeptides) so that the conditions adhered

rigorously to the theory described above. As expected due to the excess of target and tight

binding ofboth dipeptides, the hydrolysis of 4a and 4b was slow. However, as in the first

reaction, the weaker binder, 4b, was consumed at a higher rate (Figure 2.6). After 193 h,

83% of the total dipeptides have been hydrolyzed and the ratio of 4a to 4b was 3.8:1.

51

This final ratio is in excess of the ratio of the independently determined binding constants

of the dipeptides (2.1 :1).

0.3 Inside Cham ber

0.25 • • 0.2 t

• [dipeptide) 0 15

(mM) .

0.1

0.05 4a

4b 0

0 40 80 120 160 200

Time(h)

Figure 2.6:[10] Selective protection from hydrolysis of dipeptide 4a over 4b by carbonic anhydrase. A

reaction vessel was separated into two compartments (20 mL each) by a dialysis membrane. The inside

compartment contained carbonic anhydrase (13.6 f.1mol), dipeptide 4a (4.3 f.1mol) and dipeptide 4b (4.3

f.1mol) in 20 mL ofbuffer. The outer compartment contained Pronase (5 mg) dissolved in 20 mL ofbuffer.

The time dependence of the concentration in the carbonic anhydrase chamber is shown in the figure. At

83% conversion (193 h) the ratio of 4a to 4b was 3.8:1.

In a related experiment, we compared two sulfonamide dipeptides 4a and 4c,

which also have similar inhibition constants (Figure 2.7). In this experiment, both the

inside and outside chambers initially contained equal concentrations of the dipeptides,

and total concentration of dipeptides was in excess (2.1: 1 ratio of dipeptides to CA). The

result was a much faster decrease in concentration of both dipeptides initially present in

the carbonic anhydrase chamber. This faster rate reflects the rapid release of one

equivalent of Phesa (2) from the Pronase chamber. Although 2 is a weaker binder than

52

either 4a or 4c, enough of it was produced such that it could displace a small amount of

4a and 4c from the carbonic anhydrase binding pocket, thus accelerating their hydrolysis

by Pronase. However, the net result was still the same. After 6 h, 93% of 4c was

hydrolyzed after 6 h, but only 58% of 4a was hydrolyzed. Thus, the ratio of

concentrations was 6: 1, which is much larger than the 1.6: 1 ratio observed in a control

experiment which did not contain Pronase and larger than the 3.7: 1 ratio of their binding

constants.

0.2 Inside Chamber

0.16

0.12 [dipeptide]

(mM) 0.08

0.04

4a

0+-______ ~------~----~4c

o 2 4 6

Time (h)

Figure 2.7:[10) Selective protection from hydrolysis of dipeptide 4a over 4c by carbonic anhydrase. A

reaction vessel was separated into two compartments (20 mL each) by a dialysis membrane. The inside

compartment contained carbonic anhydrase (0.34 mM), while the outside compartment contained Pronase

(4 mg). Both compartments initially contained similar concentrations of dipeptide 4a (0.l6 mM) and

dipeptide 4c (0.14 mM). The protease rapidly cleaved the dipeptides in the outside compartment to the

corresponding amino acids (data not shown). The time course of the reaction in the carbonic anhydrase

chamber is shown in the figure. After 6 h, 93% of 4c inside the CA chamber had been hydrolyzed, while

only 58% of 4a had hydrolyzed. A control experiment that did not contain carbonic anhydrase showed an

equal rate ofhydrolysis for the two dipeptides in the chamber not containing Pronase.

53

FinaHy, an experiment containing aH five dipeptides (4a-d and 5) was conducted

using an excess of carbonic anhydrase (ratio of CA to dipeptides is 1.2: 1). The

experiment was consistent both with the theory and with the prior results. Dipeptide 5

was cleaved rapidly while dipeptides 4a-d disappeared at rates that corresponded to their

binding constants (Figure 2.8). Weaker inhibitors rapidly diffuse into the outside

chamber, and may occupy aH the available protease sites, causing the apparent lag-time

for destruction of stronger inhibitors at the beginning of the experiment. More detailed

analysis of this phenomena is required.

0.25 Inside Chamber

0.2

0.15 [di peptide)

(mM) 0.1

40 80 120 160 200

Time (h)

Figure 2.8:[10] Selective protection from hydrolysis of dipeptides by carbonic anhydrase. A reaction vessel

separated into two compartments (20 mL each) by a dialysis membrane was set up. The inside

compartment contained carbonic anhydrase (25.6 I1mol) and dipeptides 4a-d and 5 (4.3 I1mol each) in 20

mL of buffer. The outer compartment contained Pronase (5 mg) dissolved in 20 mL of buffer. The time

dependence of the concentrations in the carbonic anhydrase chamber is shown in the figure.

2.4 Discussion/Conclusions

54

As expected, the four sulfonamide dipeptides 4a-d aIl inhibit carbonic anhydrase

competitively with similar inhibition constants (within a factor of 10 of each other).

Classical kinetics using initial rates easily identified these differences, but these classical

methods are slow and require measuring each inhibitor separately. This becomes

laborious for libraries containing thousands of members.

To rapidly identify the best inhibitor, we used competitive binding to carbonic

anhydrase in one compartment of a two-compartment cell. The inhibitors concentrated

into the carbonic anhydrase compartment of a two-compartment cell. Higher

concentrations of the better inhibitors were observed in the carbonic anhydrase

compartment, but the concentration differences were small (1.83 : 1.71 : 1.54 : 1.46 : 1

for 4a : 4b : 4c : 4d : 5). If the mixture contained a thousand dipeptides, this competitive

experiment would not identify the best inhibitor because it would be hard to separate aU

the dipeptides and the differences in concentration with and without target would be

small.

Although this experiment does not include a dipeptide-synthesis step and thus is

not a dynamic library, the diffusion across the membrane mimics a synthesis step in a

dynamic library in that both are equilibrium processes. For the diffusion process, in the

absence of a target, each compartment should contain equal amounts of each inhibitor. In

the presence of the target, the carbonic anhydrase chamber contains more of the tight

binding inhibitors. Thus, the equilibrium for the diffusion reaction has shifted.

A non-selective destruction of the library members should enhance differences in

the relative concentrations of the members bound to the target. The po or binding

members are destroyed at a higher rate than the strong-binding members and as the

55

degradation progresses, the ratio improves exponentially in favour of the latter. This was

observed in our library, where dipeptide hydrolysis by Pronase was used as the

destruction process. In a competition experiment between a strong and weak binder (4a

vs. 5), the ratio of 4a to 5 in the carbonic anhydrase chamber increased from 1.75:1 in the

absence ofPronase to 3.7:1 in the presence ofPronase (at 45% conversion). Furthermore,

this ratio continued to increase to >20: 1 as the reaction progressed. A second experiment

with two species with very similar KIS (4a vs. 4 b) had a final ratio of 3.8: 1 when the ratio

of the binding constants was 2.1: 1. As shown in Figure 2.9a, these results follow the

theoretical model closely. Similar results were obtained for an experiment containing two

inhibitors (4a and 4c) where an excess of a weaker binder, Phesa (2), was generated in the

reaction mixture. The presence of 2 accelerated the rate of cleavage of 4a and 4c but, as

can be se en in Figure 2.9b, the ratio of dipeptides during the course of the reaction still

followed the theoretical model. At 70% conversion, the ratio of 4a to 4c was 6: 1, which

is much larger than the 1.6: 1 ratio observed in a reaction not containing Pronase. In all

cases, the model indicates that the ratios should continue to increase if the reactions are

carried out for even longer periods. In experiments with a large number of library

members, this increase will be critical in allowing the tightest binding species to be easily

identified. [12]

56

a) 5

4

Ratio 3

4a/4b 2

•

b) 10

8

Ratio 6

4a/4c 4

2

o+---~----~----~--~~--~ o+---~----~----~--~--~

o 0.2 0.4 0.6 0.8 o 0.2 0.4 0.6 0.8

Conversion Conversion

Figure 2.9: The graph shows theoretical and experimental ratios for the screening experiments. Theoretical

lines are shown as smooth lines. The S values correspond to the ratios of the experimentally determined

binding constants. The data points show the experimentally determined ratios at different conversions for a)

4a/4b (c.f. Fig. 6) and b) 4a/4c (c.f. Fig. 7).

One potentiallimitation of this screening method is selectivity in the destruction

reaction. For example, Pronase c1eaves dipeptide 4c at a much slower rate than dipeptide

4a. In such a case, S from equations 10-12 will not be equal to the ratio of the binding

constants and thus the degradation reaction will not follow the theoretical curves of figure

1. To accommodate this situation, we used a large amount of protease and, more

importantly, we employed a dialysis membrane to separate the target-inhibitor complexes

from the protease. In this setup, the rate limiting step in the destruction reaction is not the

protease-catalyzed c1eavage but diffusion across the dialysis membrane. Unlike the

protease-catalyzed c1eavage, the rate of diffusion does not vary significantly with the

structure of the inhibitor and the result is that the destruction reaction follows the

theoretical curve. Although Pronase accepts a wide variety of peptides, substrate

specificity of the enzyme may become problematic if highly diverse libraries are studied.

A dipeptide that is not c1eaved by Pronase would be retained in the reaction mixture, even

57

if it did not bind to carbonic anhydrase. One way to alleviate this problem would be to

use a mixture of enzymes with a wide range of specificities. Altematively, it is important

to note that the de gradation reaction is not limited to enzymatic processes. Other

chemical degradation methods can be envisioned, depending upon the type of library

being studied. For example, a library based on disulfide exchange could be degraded by

adding a reducing agent (e.g. a phosphine) to cleave any unbound disulfides.

Altematively, physical methods for removal of unbound inhibitors (e.g. adsorption to a

solid phase, extraction) should accomplish the same effect as a chemical degradation.

Another potential limitation of this screening method, and indeed for methods

based on the dynamic combinatorial library technique, is the need for stoichiometric

amounts of the target. The initial experiments reported here used large amounts of

carbonic anhydrase (100-500 mg/experiment) as we expect to apply it to a dynamic

library process where the best inhibitor will actually be isolated and characterized.

However, for purely analytical screening purposes, the methods can easily be scaled

down using smaller compartments, assuming that more sensitive analytical tools are used

(e.g., mass spectroscopy). These modifications could reduce the amount oftarget needed

to <0.1 mg/experiment, an amount that is easily accessible for targets that have been

cloned and over-expressed.

In conclusion, we have developed a method for screening mixtures of compounds

against a therapeutic target that readily identifies the best binder in a library. The method

works by selectively degrading the poorer inhibitors with an enzyme. This results in a

significant improvement in the ability to distinguish between inhibitors that have very

close binding constants. We plan to extend this method to dynamic libraries with the goal

58

of improving the enhancement observed in synthesis of good inhibitors in the presence of

a therapeutic target.


Romas J. Kazlauskas developed the theory in section 2.2

Andrew D. Corbett and Jonathan Croteau developed and performed the chemical

synthe sis described in section 2.3.1

Ronghua Shu performed the screening of proteases in section 2.3.4.1

The author (Jeremy D. Cheeseman) developed and performed an other experiments

described in this chapter.

References

1. Reviews: a) A. Ganesan, Angew. Chem. Int. Ed Eng. 1998,37,2828-2831; b) J.

M. Lehn, Chem. Eur. J 1999,5,2455-2463; c) G. R. L. Cousins, S. A. Poulsen, J.

K. M. Sanders, Curr. Opin. Chem. Biol. 2000, 4, 270-279; d) 1. Huc, R. Nguyen,

Comb. Chem. High Throughput Screening 2001, 4, 53-74.

2. a) A. V. Eliseev, M. 1. Nelen, J Am. Chem. Soc. 1997,119, 1147-1148; b) A. V.

Eliseev, M. 1. Nelen, Chem. Eur. J 1998, 4, 825-834.

3. Glaucoma patients often take carbonic anhydrase inhibitors to reduce the pressure

in the eye. An commercial inhibitors contain a sulfonamide moiety. We chose

carbonic anhydrase as a test case for inhibitor design and screening methods.

4. The analysis below follows closely the mathematical treatment for kinetic

resolutions. For examples see: a) V. S. Martin, S. S. Woodard, T. Katsuki, Y.

Yamada, M. Ikeda, K. B. Sharpless, J Am. Chem. Soc. 1981,103,6237-6240; b)

59

C. S. Chen, Y. Fujimoto, G. Girdaukas, C. J. Sih, J. Am. Chem. Soc. 1982, 104,

7294-7299; c) H. B. Kagan, J. C. Fiaud, Top. Stereochem. 1988, 18, 249-330.

5. E. Escher, M. Bernier, P. Parent, Helv. Chim. Acta. 1983,66, 1355-1365.

6. Researchers often use hog kidney acylase to resolve enantiomers of N-acetyl

amino acids. For examples see: a) H. K. Chenault, J. Dahmer, G. M. Whitesides,

J. Am. Chem. Soc. 1989, 111, 6354-6364; b) S. M. Roberts, Ed., Preparative

Biotransformations, Wiley: Chichester 1992-1998, Module 1:14 In our case, this

intermediate was already enantiomericaUy pure. We used hog kidney acylase to

cleave the acetyl group under mil der conditions than those required by chemical

cleavage methods.

7. EDC = 1-(3-dimethylaminopropyl)-3-ethy1carbodiimide; HOBT 1-

hydroxybenzotriazole.

8. For example, Nguyen and Huc investigated a simple sulfonamides with inhibition

constants of ~0.1 to 1 IlM (R. Nguyen, 1. Huc, Angew. Chem. Int. Ed 2001,40,

1774-1776), while Doyon et al. investigated other simple sulfonamides with

inhibition constants of ~0.001 IlM (J. B. Doyon, E. A. M. Hansen, c.-y. Kim, J.

S. Chang, D. W. Christianson, R. D. Madder, J. G. Voet, T. A. Baird Jr., C. A.

Fierke, A. Jain, Org. Lett. 2000,2, 1189-1192).

9. These experiments required stoichiometric amounts of carbonic anhydrase. We

used an inexpensive mixture of isozymes from bovine sources. Although material

was not pure carbonic anhydrase, we calculated the concentrations assuming it

was pure. Thus, the true concentration will be less than the number given.

10. Lines drawn in aU figures (except Figure 2.1 and Figure 2.9) are for illustration

purposes only. They do not represent theoreticallines of any sort.

Il. Both 4a and 5 diffused through the membrane at identical rates with a half-life of

about three hours. (Data not shown.)

12. The reaction mixture will contain the products of the degradation reactions.

However, in most cases, this method will be applied to combinatoriallibraries

and, as such, the degradation products wiU often be the common starting materials

used to make the library members. Thus only a limited number of degradation

products will be produced.

60

CHAPTER THREE

FIRST GENERATION PSEUDO-DYNAMIC COMBINATORIAL LIBRARIES

61

Abstract

This chapter describes the integration of the screening process developed in

chapter two with an in situ library synthesis from solid phase in the first pseudo-dynamic

combinatorial library. The library experiments took place in a three-chambered reaction

vessel that separated the synthesis resin and protease from the receptor, carbonic

anhydrase. Our library consisted of two series of four dipeptides each. Dipeptides 4a-d

carried an inhibitory arylsulfonamide moiety that could bind to the zinc metalloezyme,

carbonic anhydrase. Dipeptides 5a-d served as negative controls. The selectivity for the

strongest binding dipeptide (4a) was greater than 50 : 1 in a library with cycle times of 12

and 24 hours. However, the amplification of 4a was less than 1 %.

3.1: Designing the First Pseudo-Dynamic Combinatorial Library: Description of

the Three Key Processes: Synthesis of the Library, Binding to the Receptor, and

Destruction/Recycling of Unbound library Members

Amplification and selectivity are the two aspects of any receptor-assisted

combinatorial system that need to be optimized. High selectivity for only strong binding

compounds ensures that only compounds that might provide useful drug leads survive

screening. Amplification of the selected compounds is necessary for their detection and

characterization. Although the selective destruction process in chapter two provided

selectivity higher than is normally possible in traditional DeLs, this selectivity was

brought about by a destruction process that left only small amounts of the tight binders

62

left in the system.[I] Ideally, the amount of a tight binder after a highly selective

experiment should be easy to detect and characterize. Pseudo-dynamic libraries were

developed so that not only could the receptor select its tightest binders from a mixture,

but also increase the amounts of these binders over time, facilitating their detection.

Pseudo-dynamic libraries (P-DCLs) combine kinetic and thermodynamic events

to increase amplification and selectivity. They include the irreversible (and hence

kinetically govemed) formation of a library in situ. They then allow members of the

library to interact with a receptor in a thermodynamic process common to DCLs. The

synthesis and binding is coupled to a kinetic destruction process that removes unbound

library members. Finally, re-using some of the species left over from the destruction

process in a new round of library synthesis allows the library to undergo another cycle of

binding and destruction.

This system has several features that can potentially improve its ability to select

and amplify tight binders over that se en in other RACC systems, and over the selective

protection from destruction alone (Figure 3.1). The initial binding to the receptor should

select library members based on their relative binding affinities, similar to a traditional

DCL, but the destruction process should increase the selectivity beyond the levels

possible in a DCL. Recycling starting materials in the re-creation of the library will re

introduce tight binders to the receptor, allowing them to re-compete for binding sites with

other, weaker library members left over from the first cycle of synthesis, binding and

destruction.

63

(,J-(> 0 "'~, s,.~ <>? ~=R=ece=p=tor==", Receptor: 0 ~·O -.. .. ·6~ ~h

D Figure 3.1: Schematic of a hypothetical pseudo-dynamic combinatoriallibrary.

3.1.1 Synthesis of the Library

The tirst process in a p-DCL is an irreversible library synthesis. Based on

previous resultsY] we made a library of dipeptides, sorne with an aryl sulfonamide

moiety able to bind to the receptor, carbonic-anhydrase (CA). Previous studies showed

that TentaGel-supported tetrafluorophenol active esters react cleanly with free amino

acids in water under alkaline (pH 8-10) conditions to form dipeptides. [2] Aqueous peptide

synthesis was necessary to ensure receptor stability, and a solid phase approach would

allow for periodic replenishment of starting materials. The initial library consisted of

eight dipeptides, four ofwhich had a CA-binding aryl sulfonamide moiety (Figure 3.2).

64

x= SOzNHz, H R,=Bn,Rz=H R, = CHzipr, Rz = H R,=H,Rz=H R,RZ = (CHzla

1 ,&: 1 ,&: HZNOZS~ H~

H~ ~",COZH H~ ~",COZH Eloc 0 Eloc 0

Eloc-Phe..G1y (4b) Eloc-PheGly (5b)

Eloc..phePhe (5a)

Figure 3.2: Solid phase, aqueous synthesis of a library of dipeptides. Top: the general scheme of library

synthesis in pseudo-dynamic libraries. Bottom: The solid phase synthesis of the first generation pseudo-

dynamic library. A nucleophile (amino acid) attacks that N-Etoc-protected active ester of an amino acid on

the tetrafluorophenol-TentaGel resin. Sorne dipeptides contain an aryl sulfonamide moiety that can bind to

carbonic anhydrase.

The tirst generation p-DCLs coupled nucleophiles (glycine, proline, leucine and

phenylalanine) to active esters of phenylalanine and 4'-sulfonamidophenylalanine.

Previous studies showed a general trend going from pH 6 to 10 that favoured dipeptide

synthesis over hydrolysis from the resinYl Since 9.0 was the highest pH at which the

receptor was stable for several days, the library synthesis was investigated at pH 9.0

using four different concentrations of nucleophile (10, 5, 4, and 3 mM, Table 3.1). The

65

nuc1eophile was in excess relative to the active ester in every case. The product

distribution did not change significantly with varying nuc1eophile concentration, giving

an average ratio of 4b : 4d : 4a : 4c of 3.0 : 1.5 : 1.2 : 1.0. We now had a versatile,

aqueous, solid phase synthetic strategy for the generation of a dipeptidic library of

potential CA inhibitors.

Table 3.1. Distribution of products from the library described in Figure 3.1.

Concentration 4b 4d 4a 4c

(mM)

10 (16 equiv) 44% 25% 17% 14%

5 (8.0 equiv) 46% 21 % 18% 14%

4 (6.4 equiv) 44% 23% 18% 15%

3 (4.8 equiv) 44% 22% 18% 16%

Average ratio 3.0 1.5 1.2 1.0

3.1.2 Binding to the Receptor and Destruction ofUnbound Library Members

A kinetic synthesis can generate a library of dipeptides. In a p-DCL, the synthesis

is followed by an initial, reversible interaction of library members with the receptor,

providing selectivity based on relative binding affinities (Figure 3.3). At this point, one of

the major drawbacks to DCLs, that being the amplification of non-binding library

members (see section 1.2.3 and Figure 1.10),[3] has already been overcome because the

66

library synthesis is irreversible. This means that the product distribution is based only on

the relative synthetic rates of each library member and their binding affinities, and

entropic factors cannot affect the product distribution, which can be the case in fully

reversible systems.

The selectivity should be similar to that seen in the experiments in which a

receptor selectively extracted inhibitors based on their binding affinities (described in

section 2.3.3, Figures 2.2 and 2.3). However, library members that are synthesized to

greater degrees will be present in larger quantities and will be able to more effectively

compete for binding sites, which would alter the results to favour the most readily formed

dipeptide rather than the tightest binder.

« P Receptor 0 6 t> == Receptor. ...... h ~

Figure 3.3: The dipeptide library can now interact with the receptor, carbonic anhydrase. This is a

thermodynamic process that gives initial selectivity for tight binders, but is not great enough to overcome

synthetic biases without an added destruction reaction.

The kinetic destruction reaction is the same as described in chapter 2, namely a

non-selective protease catalyzed cleavage of the dipeptides to their constituent amino

acids (by Pronase from Streptomyces griseus). This destruction reaction gives greater

67

selectivity than is possible in DCLs by weaning away weakly bound library members.

This process is analogous to the experiments in which a receptor selectively protected

strong binders from a destruction reaction (described in section 2.3.4.2, Figures 2.4-2.8).

The destruction reaction was hoped to be able to overcome any synthetic bias, and

increase the selectivity many fold over that seen from the thermodynamic binding alone.

o <> ? Receptor Recepto9

o r=6~ ~'b o

Figure 3.4: The destruction of unbound library members. Top: Schematic of the process. Bottom: Sorne

aryl-sulfonamide dipeptides bind to CA and are protected from the Pronase catalyzed hydrolysis to their

constituent amino acids. One half of each dipeptide is recycled in a new round of synthesis, while the other

builds up in the system with each successive destruction cycle.

Carbonic anhydrase (as weIl as many other receptors) has surface-exposed amines

that could react with the active esters used to create the library. A dialysis bag that would

allow the dipeptides, but not macromolecules to cross, was therefore employed to

separate the receptor and resin. The isolation of the resin also allowed for the extraction

of used resin from the system and its easy replacement with fresh reagents. We had

previously made a two-chambered system in which the receptor was separated from a

protease. This concept was extended to form a three-chambered vessel consisting of two

68

dialysis bags, one containing the resin (the synthesis chamber, 10 mL), and one

containing the protease (destruction chamber 20 mL), suspended in a solution of carbonic

anhydrase (binding chamber 20 mL) in a 100 mL container.

Chamber 1: Synthesis

Pronase

Figure 3.5: The three-chambered p-DCL experimental set up. A library of dipeptides is synthesized from

active esters on solid support in chamber 1. The dipeptides can then diffuse into the binding chamber

(chamber 2) and interact with the receptor, carbonic anhydrase. If they are not bound to the receptor, they

will diffuse into chamber 3 and be destroyed by Pronase. The non-N-protected amino acids can then diffuse

into the synthesis chamber again, where new active ester is added in a new round of synthesis, binding and

destruction. MWCO = molecular weight cut off.

3.1.3 Recycling Destruction Products and Iteration of Synthesis

Up to this point, that is the end of the first cycle of synthesis, binding and

destruction, the pseudo-dynamic library is very similar to the systems described in

chapter two. In those static libraries, pre-made dipeptides were added to a chamber

containing CA, which kept them from diffusing into a destruction chamber. The main

69

difference in a full p-DCL is the process of iteration, where new active esters are added,

and can react with freed starting materials produced by the protease. By re-synthesizing

the library, aIl the compounds have another chance to bind to the receptor. As in the first

cycle, tighter binders will be able to occupy relatively more binding sites. In this second

round, however, the tight binders will be able to replace weaker binders left over from the

first cycle. This will increase the relative amounts of the best binders in the binding

chamber, effectively increasing the selectivity over that seen in the static library cases.

Rather than continuously being destroyed, re-synthesis also allows compounds to

build up in the system. This gives larger overall quantities of compounds, introduced to

the receptor, and gives more material at the end of an experiment, further facilitating

analysis. The ability to re-use starting materials that are formed is also important, as

without it, aIl the compounds generated in the destruction process would continue to

build up in the system over time. This aspect of the first generation p-DCLs will be

discussed at the end ofthis chapter, and will become important in chapter 4.

3.2 Results of the Integrated Processes in the First Pseudo-Dynamic

Combinatorial Library

The first moderately successfullibraries showed that the selectivity could indeed

be increased not only over that of DCLs, but over that of the static libraries described in

chapter 2. Pseudo-dynamic experiments vary with respect to cycle time and the number

of equivalents (with respect to the receptor) of dipeptides that are synthesized in each

cycle. The first generation experiments coupled amino acids Gly, Pro, Leu and Phe with

70

active esters of phenylalanine and 4'-sulfonamidophenylalanine. The amino acids were

regenerated by hydrolysis catalyzed by Pronase. Control experiments established that the

hydrolysis rate was similar for an compounds, and was at least four times faster than

diffusion across the dialysis membrane separating the binding and destruction chambers.

The competitive inhibition constants of an species that would be present in the library

were measured at pH 9.0 (Table 3.2) so that the results of the library experiments could

be accurately interpreted.

Table 3.2: Inhibition constants for components of the first generation p-DCLs.

Compound KI (J.tM)a

Phesa(l) 8±2

Etoc-Phesa (2) 13±2

Etoc-Phesa-Phe (4a) 0.45±0.09

Etoc-Phesa-Gly (4b) 0.75±0.1

Etoc-Phesa-Leu (4c) 1.26±0.03

Etoc-Phesa-Pro (4d) 6.6±0.2

Etoc-R-Phe (5a-d) »1000b

a Competitive inhibition constants for the carbonic anhydrase catalyzed hydrolysis of p-nitrophenyl acetate

(PNPA) at 25°C in BICINE buffer (lOmM pH 9.0). A typical procedure was to add carbonic anhydrase

solution (100.0 !lL, 0.05 mg/mL) containing inhibitor (0.0-100.0 !lM in most cases) to an acetonitrile

solution of pNPA (5.0 !lL, 2.0-32 mM) and follow the release of p-nitrophenoxide spectrophotometrically

at 404 nM. b No inhibition detected at an inhibitor concentration of 1 mM.

71

The results of a p-DCL using two, 24 h cycles are shown in Figure 3.6. In this

experiment CA (28.0 mmol, 1.0 eq) and Pronase (catalytic) were suspended in 20 mL

each of a 5 mM stock solution of amino acids in BICINE buffer at pH 9.0. The Pronase

solution was placed in a dialysis bag to form the destruction chamber, and was suspended

in the CA solution in a 100 mL plastic cup (the binding chamber). The synthesis chamber

had a 160 mmol (~5. 7 eq) portion of resin composed of equal amounts of the active esters

of phenylalanine and 4'-sulfonamidophenylalanine.

The graph shows only the three strongest dipeptidic inhibitors, 4a (KI = 0.45 ~M),

4b (Kr = 0.75 ~M), and 4c (Kr = 1.3 ~M). AlI eight dipeptides were present at the

beginning of each cycle, but the non-sulfonamides and the weakest sulfonamide, 4d had

all been removed from the binding chamber within four hours. As was observed in the

static library synthesis (section 3.1.1), 4b is synthesized to a greater degree than any other

dipeptide. However, by 10 hours its concentration is lower than that of the strongest

binding dipeptide, 4a. This shows that the destruction process provides at least enough

selective pressure to overcome a two-fold synthetic bias. At the end of two cycles, only

the strongest binding dipeptide can be observed in the binding chamber. This shows that

the iterative process does increase the selectivity over that of the static library

experiments.

However, despite this excellent selectivity, the amount of the best binder at the

end of the experiment represents less than 1 % of the total available receptor binding sites,

and seems to actually decrease over successive cycles. This is the opposite of what was

expected from the initial design of these systems.

72

0.07

0.06

0.05

0.04 mM

0.03

0.02

0.01

0

0 10 20 30 40

Time (h)

Figure 3.6: P-DCL of two cycles, 24 h each. The strongest binder (4a, KI = 0.45 !lM) dominates in each

cycle, giving high selectivity. However, the amounts of the best binder decrease over successive cycles,

rather than increasing as expected. The receptor is present at a concentration of 1.4 mM.

Figure 3.7 shows the results of a very similar experiment to that of Figure 3.6.

The only difference in this case is that the destruction chamber was not added for six

hours after the initiation of a new cycle. The rationale behind this design was to observe

if the destruction process was taking place too rapidly to allow compounds to build up

over time, giving the apparent lack of amplification seen in Figure 3.6. This did not prove

to be the case, as the strongest binding compound, although still present as the only

binder left after two 24 h cycles, seems to go down in concentration with successive

cycles rather than the opposite.

73

0.08

0.07

0.06

0.05

mM 0.04

0.03

0.02

0.01

0

0 10 20 30 40

Time (h)

Figure 3.7: Two 24 h cycles, with a 6 h delay in the addition of the destruction chamber. As in Figure 3.6,

excellent selectivity is observed, but no build up of the best binder occurs over successive cycles. The

receptor is present at a concentration of 1.4 mM.

Finally, in Figure 3.8 the amount of dipeptides formed in each cycle is halved to

80 ~mol (~2.9 eq in each portion of resin), while the cycle time is also halved to 12 h,

and the number of cycles is doubled. The selectivity is not quite as pronounced as in the

24 h cases, but the strongest binder is still the only one observable after two cycles.

Again, the total amount of the strongest binder drops with successive cycles to occupy

less than 1 % of available binding sites at the end of four cycles.

74

0.05

0.04

0.03

mM 0.02

0.01

0

0 10 20 30 40

Time (h)

Figure 3.8: Four 12 h cycles with half the amount of dipeptide formed per cycle as compared to Figures

3.6 and 3.7. The selectivity remains very high, but the amplification decrease over successive cycles is even

more pronounced. The receptor is present at a concentration of 1.4 mM.

3.3 Discussion/Conclusion

Pseudo-dynamic libraries use two coupled, but irreversible processes in place of

the equilibrium reaction used by traditional DeLs. The level of selectivity shown by just

a few cycles of a pseudo-dynamic library is higher than in any receptor-assisted method

to date. The in situ library synthesis was successful in the initial cycle, and in the

subsequent cycles that used material generated from the destruction of unbound

compounds. So long as the equivalents of nucleophiles in the system at aU times are high

with respect to the active esters, there should be no bias from the library as to which

compounds are re-synthesized, hence each cycle should give the strongest binders an

opportunity to replace weaker ones. Additionally, as long as diffusion across the

membrane to the dialysis chamber is the rate-limiting step in destruction, there should be

no bias in the system with respect to differences in the hydrolysis rates of the library

members.

75

The selectivity of these systems can overcome significant (two-fold) synthetic

bias. This is a result that would not be possible in a DCL. In a DCL, aIl the processes are

reversible, which leads to the concentrations of each compound being linearly dependent

on the product of its binding and synthetic constants (see section 1.2.3 for derivations).

The kinetic destruction and iteration of the binding of compounds to the receptor resolve

an important issue in receptor-assisted combinatorial systems in that, as the synthe sis

becomes more diverse and complex, the chance of having identical rates and levels of

synthesis for each library member becomes increasingly difficult. Overcoming synthetic

bias is an important asset of p-DCLs.

Although the selectivity of these initial systems was promising, the lack of any

amplification over successive cycles was puzzling. The cause of this unfavourable result

was quickly identified as being inherent in the synthesis of the library. In each cycle, a

portion of the dipeptide, but not aIl, is recycled in the new round of synthesis. In these

first generation libraries, the amino acid containing the aryl sulfonamide moiety used for

receptor binding was the un-recycled partner (Figure 3.4). Although not a strong

inhibitor, N-EtocPhesa (2) does inhibit CA with a KI of ~ 13 /lM (Table 3.2). Large

amounts of this compound will rapidly build up in the system, first from hydrolysis of the

active ester by water, and then from the destruction of every unbound sulfonamide

containing library member. This compound can easily achieve concentrations at which it

can out-compete even the strongest binding library member. The likelihood of the

recycling partner being of paramount importance in amplification over cycles was the

subject of the next generation of p-DCLs, and is addressed in the libraries created in

Chapter4.

76


Andrew D. Corbett developed and performed the synthe sis described in section 3.1.1

The author (Jeremy D. Cheeseman) developed and performed aIl other experiments


References

1. J. D. Cheeseman, A.D. Corbett, R. Shu, J. Croteau, J. L. Gleason, R. J. Kazlauskas J

Am. Chem. Soc. 2002,124,5692-5701.

2. A. D. Corbett, J. L. Gleason Tetrahedron Lett. 2002,43, 1369-1372.

3. Z. Grote, R. Scopelliti, K. Severin, Angew. Chem. Int. Ed. Engl. 2003,42,3821-3825;

Angew. Chem. 2003,115,3951-3955.

77

CHAPTER FOUR

PSEUDO-DYNAMIC COMBINATORIAL LIBRARIES: A NEW RECEPTOR

ASSISTED APPROACH FOR DRUG DISCOVERY

78

Abstract

This chapter contains the results of a pseudo-dynamic library in which the amino

acid containing the inhibitory arylsulfonamide moiety is recycled in each new round of

synthesis. The library was made of two new series of four dipeptides each, 7a-d

(sulfonamides) and 8a-d (negative controls). In these libraries, the importance of cycle

time on selectivity becomes apparent. Using eight-hour cycles proved insufficient to

overcome a synthetic bias that favoured 7b, the third strongest binder. However,

lengthening the cycle time to 16 hours improved the selectivity profile to give the

strongest binding dipeptide, 7d in greater than 100 : 1 selectivity and 29% yield.

4.1 Introduction

Emerging methods of combinatorial chemistry incorporate receptor assistance to

combine synthesis and screening. [1] Binding to stoichiometric amounts of a receptor alters

either the thermodynamics or kinetics of library synthesis. Dynamic combinatorial

libraries[2] use a thermodynamic approach where binding shifts a synthetic equilibrium to

increase the amounts of the best binding compounds, in accordance with 's principle.

These libraries usually identify the tightest binding library members, but sorne

experimental conditions can give small or misleading changes in concentrationY] An

alternative method, receptor-accelerated synthesis, uses a kinetic approach. [4] Starting

components that bind to the receptor can couple to create a new, tight-binding compound.

The receptor accelerates the coupling of the better fitting starting components due to their

79

close proximity, but requires that both components bind tightly to the receptor. Here we

demonstrate a new method, a pseudo-dynamic library, which adds a kinetic contribution

to traditional dynamic libraries to dramatically increase the selectivity.

A pseudo-dynamic combinatorial library combines an irreversible synthesis of

library members with an irreversible destruction step. Those library members that bind to

the receptor are protected from destruction. Subsequent synthesis reuses fragments from

destroyed library members, thus amplifying the amounts of the better binders at the

expense of the lesser ones. In cases where one of the fragments contains a binding motif,

this should be the fragment that is reused. Otherwise it can build up in the system and

decrease any amplification that could arise from the reintroduction of the library. The

separate irreversible synthesis and destruction steps allow adjustment to optimize both

the amplification and selectivity.

4.4 Experimental Design

We developed a pseudo-dynamic library of eight dipeptides to identify the best

inhibitor of carbonic anhydrase (CA). CA, a zinc metalloenzyme, is a therapeutic target

for glaucoma and is inhibited by aromatic sulfonamides, which coordinate to the zinc ion.

Four of the eight dipeptides in our library contain 4'-sulfonamidophenylalanine (Phesa, 1),

and thus should bind to CA, while the remaining four contain only Phe and serve as

negative controls. The irreversible synthesis of dipeptides used a solid-supported

coupling of activated esters with an amino acid in aqueous solution (Scheme 1).

TentaGel-supported tetrafluorophenol active esters react cleanly with free amino acids in

80

water under alkaline (PH 8-10) conditions to form dipeptides. [5] A non-selective protease

from Streptomyces griseus (Pronase) destroyed these dipeptides by catalyzing their

irreversible hydrolysis. [6] This library differs from that described in chapter 3 of this

thesis in that the nucleophillic, recycled starting material contained the sulfonamide

binding motif. This arrangement prevented the build up of 4'-sulfonamidophenylalanine

in the system, hopefully allowing for greater amplification of inhibitory dipeptides.

H2Ny C02H

~ V--x 1 X: S02NH2 6X:H

Pronase

X~

~ = 0 C02Et

HO C)...N)l .NoR 2 H l 2 R,

7X:S02NH2 8X:H

Series: a R, : CH2Ph, R2: H bR,: H, R2 : H cR, : CH2CH(CH3n, R2 : H d R, R2 = (CH2h

Scheme 4.1: Creation of a pseudo-dynamic library of dipeptides.

The pseudo-dynamic library was prepared in a three-chambered reaction vessel

formed by suspending two dialysis bags in a surrounding solution (Figure 4.1). One

dialysis bag (the synthesis chamber) contained the active esters; the other dialysis bag

(hydrolysis chamber) contained the protease, while the surrounding solution (screening

chamber) contained the carbonic anhydrase. Adding nucleophiles 1 and 6 to the synthesis

chamber generated the dipeptide library. These dipeptides diffused into the surrounding

solution where they could bind to carbonic anhydrase, and further diffused into the

hydrolysis chamber where Pronase cleaved them. This cleavage regenerated 1 and 6,

which could diffuse back into the synthesis chamber to repeat the cycle. This

arrangement prevented Pronase-catalyzed destruction and active ester-mediated

81

modification of the receptor, CA, and also permitted periodic replenishment of the

activated ester to regulate the rate of synthesis.

Synthesis Chamber Screening Chamber Hydrolysis Chamber

Figure 4.1: Schematic of the pseudo-dynamic combinatorial library experiment. Reaction of two free

amino acids (Phesa (1) and Phe (6» with four solid-supported active esters (N-Etoc-Phe, N-Etoc-Gly, N-

Etoc-Leu and N-Etoc-Pro) creates an eight-member library.

The experiments used four active ester resins derived from N-Etoc-Phe, N-Etoc-

Gly, N-Etoc-Leu and N-Etoc-Pro (0.8 equiv each), nucleophiles 1 and 6 (6.4 equiv each),

carbonic anhydrase (28 Ilmo1, 1 equiv) and Pro nase (25 mg/mL). The large amount of

Pronase made diffusion across the dialysis membrane the rate-limiting step for

hydrolysis; hence, aIl dipeptides were cleaved at similar rates in spite of the substrate

selectivity of Pronase. Periodic addition of fresh portions of active ester resin (defined as

the cycle time) regulated the overall rate of library synthesis. We conducted three

experiments with this system using cycle times of 8, 12 and 16 h. HPLC analysis of

aliquots from the screening chamber showed the progress ofthe experiments (Figure 4.3).

82

4.5 Results

Two control experiments established, first, that the synthetic process afforded aU

the expected dipeptides and, second, that the sulfonamide-containing dipeptides inhibited

carbonic anhydrase. Combining equal amounts of the four active esters with Phesa (1) as

the nucleophile produced four dipeptides 7a : 7b : 7c : 7d in a ratio of 18 : 44 : 15 : 23.

Not surprisingly, coupling of 1 with the less hindered glycine ester to produce 7b was

more efficient than with the more hindered phenylalanine, leucine, or proline esters. In

spite of these differences aIl four dipeptides formed in significant amounts. Using

phenylalanine as the nucleophile gave similar results. For the second control experiment,

aIl eight dipeptides were prepared individuaIly and their ability to inhibit the CA

catalyzed hydrolysis of p-nitrophenyl acetate (Figure 4.2) was measured. As expected the

sulfonamide-containing dipeptide competitively inhibited this hydrolysis with inhibition

constants of 1.1-8.7 IlM, while the non-sulfonamide dipeptides showed no detectable

inhibition. Dipeptide 7d was the best inhibitor, with an inhibition constant of 1.1 IlM and

dipeptide 7c was the next best inhibitor with an inhibition constant of2.5 IlM. Compound

1 also inhibits CA (KI = 13 IlM), but approximately ten-fold less effectively than the

tightest binding dipeptide (7d).

83

(7a) EtocPhePhesa KI = 8. 7 ~M

~H

o "" 1 H N OH

Et02C" 0~ ~O V

(8a) EtocPhePhe I<j» 1.0 mM

"...1 o ;$S02NH2

~ Il OH EtOC" ~N

2 H o

(7b) EtocGlyPhesa KI = 5.6 ~M

H o;$""IH

N Il OH EtOC" ~N

2 H o

(8b) EtocGlyPhe KI» 1.0 mM

;$H

o "" 1 H N OH

Et02C .... 0~ y 0

(8c) EtocLeuPhe I<j» 1.0 mM

""""J A!," U ~Ày0H o

(7d) EtocProPhesa KI = 1.1 ~M

Figure 4.2: Structures and competitive inhibition constants of the library members for the CA-catalyzed

hydrolysis of p-nitrophenyl acetate. The non-sulfonamide compounds showed no detectable inhibition at 1

mM.

In the first pseudo-dynamic experiment (8-hour cycle, Figure 4.3a), the cycle time

was too short for the destruction reaction to remove the less effective inhibitors. During

the first four hours of each cycle, the screening chamber contained all eight dipeptides,

indicating that all eight had formed as expected. At the end of each 8 h cycle, prior to the

next addition of active ester, the hydrolysis had removed the four non-sulfonamide

dipeptides, leaving only the four sulfonamide dipeptides. At the end of six cycles of

active ester addition, dipeptide 7b was present in the highest amount (58% yield, relative

to CA), followed by 7d (33%), 7e (27%) and 7a (8%), respectively. These relative

amounts differ from their relative binding constants. Rather, the higher yield of 7b

reflects its more favourable rate of synthesis. In addition, the sum of all sulfonamide

dipeptides at 48 h was greater than the amount of target (126% yield). This high yield

shows that unbound dipeptides remained and that the destruction reaction had not had

enough time to distinguish between the different sulfonamide inhibitors.

84

Lengthening the cycle time from 8 h to 12 h yielded the best three inhibitors with

the relative amounts in the order of their inhibition constants (Figure 4.3b). Although

sulfonamides 7b, 7c and 7d were present in high concentrations early in the experiment,

at the end of four cycles, the concentration of these weaker binding dipeptides had

diminished substantially. The tightest binding dipeptide, 7d, was present in the highest

amount (15% yield relative to CA), followed by 7c (5%) and 7b (1.5%). Notably, the

ratio at the end of the experiment (10.1 : 3.5 : 1) exceeded the ratio of their binding

constants (5.1 : 2.2 : 1). None of the weaker binding 7a or of the non-sulfonamide

dipeptides remained at the end of the experiment.

The selectivity of the dynamic process improved even further upon extending the

cycle time to 16 h (Figure 4.3c). The initial synthesis during the first cycle favoured

dipeptide 7b, the most rapidly synthesized dipeptide, but this dipeptide disappeared in

later cycles where the main competition was between 7 d and 7 c, the tightest binding

dipeptides. After four cycles (64 h), only these two remained and the ratio of their

concentrations (13 : 1) was significantly higher than the ratio of their binding constants

(2.3 : 1). After three more cycles the selectivity increased to > 1 00 : 1 in favour of the

strongest binding dipeptide, 7d. The yield was 29% relative to the amount of CA and

corresponded to 4 mg of dipeptide. Thus, adjusting the relative rate of library synthesis

and destruction optimized the selectivity so that only the best binding dipeptide remained

and in a good overall yield.

85

a) 1.20

1.00

0.80

0.60

[7]/mM 0.40

0.20

0.00 Ji~~k::=:=~=:=:::::==::==~

b) 0.30

0.25

0.20

0.15

[7]/mM 0.10

c)

0.05

0.60

0.50

0040

0.30

[7]/mM 0.20

0.10

o 10

10

20 30 40 50

tlh ----

20 30 40 50

tlh----

0.00 __ --~ __ ~--.......... __ ...... _ __.

o 20 40 60 80 100

tlh----

Figure 4.3: Pseudo-dynamic Iibrary experiments. Concentrations of sulfonamide containing dipeptides 7a

(e), 7b C"), 7c (.) and 7d (+) over the course of experiments a) 8 h per cycle, b) 12 h per cycle and c) 16

h per cycle.

4.4 Discussion/Conclusion

The selectivity in the pseudo-dynamic library is significantly greater than that in

many traditional dynamic libraries. The optimum conditions produced only the single,

86

tightest-binding dipeptide (> 1 00: 1 selectivity), while a traditional approach would yield a

mixture because the binding constants for the two tightest-binding dipeptides differed by

only 2.3-fold. This higher selectivity greatly simplifies the analysis, as only one

compound need be identified and characterized. The optimization of a pseudo-dynamic

library arises through control of the relative rates of synthesis and destruction. We

previously showed that a destruction reaction operating on a static library in the presence

of a receptor distinguishes between library members with very similar binding constants,

selectively removing the weaker binding speciesJ6] However, when selectivity arises

from destruction alone, significant amounts of the best-binding library member must be

destroyed to achieve high ratios of good binder to slightly poorer binder. This leaves only

a small amount of the best binder for analysis. In pseudo-dynamic libraries, the high

selectivity also stems from the competition between binding to the receptor and

destruction.

The iterative nature of the experiment also contributes to the high selectivity.

Toward the end of each cycle, cleavage by Pronase has reduced the amounts of weak

binding dipeptides, leaving dipeptide 7 d as the major species bound to CA. The

subsequent burst of synthesis produces a mixture of all dipeptides, which compete for the

smaller amount of free target. Pronase then rapidly cleaves all unbound species, which

would consist of a higher proportion of weak binders. Continued action of Pronase

further increases the ratio in favour of the bound species, following our static model,

resulting ultimately in high selectivity for the tightest-binding species.

Our static model of pseudo-dynamic combinatorial libraries[6] indicates that

selectivity stems from the relative binding constants of the inhibitors, not their absolute

87

affinity for the target. Thus, we expect that pseudo-dynamic combinatoriallibraries will

also work with even tighter binding inhibitors, but would require longer cycle times to

distinguish between these more tightly binding inhibitors. lndeed, we are currently

expanding to larger pseudo-dynamic libraries to disco ver such tighter binding inhibitors.


Andrew D. Corbett developed and performed the solution-phase synthesis to acquire pure

samples of the library members shown in figure 4.2.

The author (Jeremy D. Cheeseman) developed and performed aH other experiments


References

1. a) A. Ganesan Angew. Chem. Intl. Ed. Engl. 1998, 37, 2828-2831; Angew. Chem.

1998,110,2989-2992; b) I. Huc, I-M. LehnActua/ité Chimique 2000, 51-54; c)

C. Karan, B.L. Miller Drug Discov. Today 2000,2,67-75; d) R. Nguyen, I. Huc

Comb. Chem. High-Throughput Screen. 2001, 4, 53-74; e) J.-M. Lehn, A.V.

Eliseev Science 2001, 291, 2331-2332; f) O.Ramstrom, J.-M. Lehn Nat. Rev.

Drug Disc. 2002, 1, 26-36; g) S. Otto, R.L.E. Furlan, IK.M. Sanders Curr. Op.

Chem. Biol., 2002,6,321-327. h) O. Ramstrom, T. Bunyapaiboonsri, S. Lohman,

I-M. Lehn Biochim. Biophys. Acta 2002, 1572 178-186.

2. a) I. Huc, J.-M. Lehn Proc. Nat!. Acad. Sci. USA 1997,94, 2106-10; b)

88

P.G.Swann, RA. Casanova, A. Desai, M.M. Frauenhoff, M. Urbancic, U.

Slomczynska, A.l Hopfinger, G.C. LeBreton, D.L. Venton Biopolymers 1997,

40,617-625; c) B. Klekota, M.H. Hammond, B.L. Miller Tetrahedron Lett., 1997,

38, 8639-8642; d) B. Klekota, B.L. Miller Tetrahedron, 1999, 55, 11687-11697;

O. Ramstrom, l-M. Lehn ChemBioChem 2000,1,41-48; e) C. Karan, B.L. Miller

J Am. Chem. Soc., 2001, 123, 7455-7456; f) RJ. Lins, S.L. Flitsch, N.J. Turner,

E. Irving, S.A Brown Angew. Chem. Int. Ed Engl. 2002,41,3405-3407; Angew.

Chem. 2002, 114, 3555-3557; g) M. Hochgurtel, H. Kroth, D. Piecha, M.W.

Hofmann, K.C. Nicolaou, S. Krause, O. Schaaf, G. Sonnenmoser, A.V. Eliseev

Proc. Nat!. Acad. Sei. USA, 2002, 99, 3382-3387; h) I. C. Choong, W. Lew, D.

Lee, P. Pham, M.T. Burdett, J.W. Lam, C. Wiesmann, T.N. Luong, B. Fahr, W.L.

DeLano, R.S. McDowell, D.A Allen, D.A Erlanson, E.M. Gordon, T. O'Brien J

Med Chem. 2002, 45, 5005-5022; i) S. Otto, RL.E. Furlan, J.K.M. Sanders,

Science, 2002, 297, 590-593; j) D.A. Erlanson, J.W. Lam, C. Wiesmann, T.N.

Luong, R.L. Simmons, W.L. DeLano, I.C. Choong, M.T. Burdett, W.M.

Flanagan, D. Lee, E.M. Gordon, T. O'Brien Nature Biotechnol. 2003, 21, 308-

314; k) AC. Braisted, J.D. Oslob, W.L. Delano, l Hyde, RS. McDowell, N.

Waal, C. Yu, M.R. Arkin, B.C. Raimundo J Am. Chem. Soc. 2003, 125, 3714-

3715

3. a) A.V. Eliseev, M.I. Nelen J Am Chem Soc. 1997, 199, 1147-1148; b) J.S.

Moore, N.W. Zimmerman Org. Lett. 2000,2,915-918; c) Z. Grote, R. Scopelliti,

K. Severin Angew. Chem. Int. Ed Engl. 2003,42, 3821-3825; Angew. Chem.

2003,115,3951-3955.

89

4. a) K.C. Nieolaou, RHughes, S.Y. Cho, N. Winssinger, C. Smethurst, H.

Labisehinski, R Endermann Angew. Chem. Int. Ed. Engl. 2000, 39, 3823-3828;

Angew. Chem. 2000,112,3981-3986; b) R Nguyen, I. Huc Ang. Chem. Int. Ed.

Engl. 2001,40, 1774-1776; Angew. Chem. 2001,113,1824-1826; e) W.G. Lewis,

L.G. Green, F. Grynszpan, Z. Radie, P.R Carlier, P. Taylor, M.G. Finn, K.B.

SharplessAngew. Chem. Int. Ed. Engl. 2002,41, 1053-1057; Angew. Chem. 2002,

114, 1095-1099.

5. A.D. Corbett, J.L. Gleason Tetrahedron Lett. 2002,43, 1369-1372.

6. lD. Cheeseman, A.D. Corbett, R. Shu, J. Croteau, J.L. Gleason, RJ. Kazlauskas

J Am. Chem. Soc. 2002,124,5692-5701.

90

CHAPTER FIVE

AMPLIFICATION AND SELECTIVITY IN, EXPANSION AND MODELING OF

PSEUDO-DYNAMIC COMBINATORIAL LIBRARIES

91

Abstract

Pseudo-dynamic combinatoriallibraries are different from other receptor-assisted

methods because they use irreversible synthesis and complementary destruction reactions

bridged by binding to a receptor. This chapter describes experiments that show that

neither an impurity in carbonic anhydrase, nor the concentration of nucleophile 1 affect

the yield of the strongest binder, 7d in the eight-membered, 16 hour cycle experiment

described in chapter four (Figure 4.3c). Further experiments show that fewer compounds

in the library can raise the amplification of 7d, but simultaneously lower the selectivity in

this p-DCL. Expanding the library to 30 compounds results in selection of a single library

member but requires a longer cycle time, probably due to that inhibitor's increased

receptor-affinity. Finally, a preliminary mathematical model of p-DCLs shows that the

relative contribution of the kinetic destruction to selectivity is greater than that of

thermodynamic receptor-binding in these systems.

5.1 Origins of Amplification Maxima in the Eight-Membered p-DCL

5.1.1 Introduction

There are two elements that need to be optimized in any receptor-assisted

combinatoriallibrary: amplification and selectivity. By the definition of amplification

given in chapter one (Box 1.1), any amount of library member left at the end of the

experiment counts as amplification in a p-DCL because without the receptor, the amounts

92

of any compound left at the end of a p-DCL experiment will depend on its level of

synthesis, but more importantly, on the amount of time given for destruction. The

equation for amplification ([IA]receptor![IA]no receptor) becomes artificially large given long

cycle times (in which [IA]no receptor becomes very low). Because of this unique feature of p

DCLs, analyzing the final yield of a library member (from chapters three and four,

"yield" refers to the percentage of receptor binding sites occupied by a library member)

gives a more useful measure of the level of its amplification.

The six, 16-hour cycle experiment (Figure 4.3c) gave a selectivity of 7d : 7c of

>50 : 1. Recycling the amino acid that carries the receptor-recognition moiety of the

dipeptide in each new round of synthesis resulted in a 29% yield of the strongest library

member (7d), much higher than the <1 % yield of the strongest inhibitor when this amino

acid was not recycled. This 29% yield (corresponding to ~4 mg of7d in the system) was

easily detectable by HPLC, and could have been characterized by NMR and MS if

necessary (the extra NMR and MS analysis of the library material was not necessary

because a pure sample of the compound had previously been prepared and analyzed by

these methods). Although this amount of material was easily analyzed, the amount of an

inhibitor present in the screening chamber depends on the amount of receptor available to

bind it. Smaller-scale p-DCLs will be needed in the future because sorne receptors, due to

low bacterial expression levels, rarity in nature, and high costs, are not as readily

available as carbonic anhydrase. In a small-scale p-DCL, analysis of a library hit will

require that the yield of the tight-binding compounds be as high as possible. Since we

hope to exp and the pseudo-dynamic library method to other, less readily available

biological targets, we decided to explore ways of improving the yield of 7d in the eight-

93

membered library (depicted in Figure 4.3c) to determine the features of a p-DCL that

govern amplification.

These explorations took two directions. First, we wanted to be sure that our

enzyme was pure. The carbonic anhydrase used for the experiments described in chapters

three and four was an inexpensive mixture of isozymes from bovine erythrocytes. An

impurity in the measured amount of CA would have made our estimate of its molar

amount in the experiments artificially high. The yield is calculated by dividing the molar

amount of 7d by the molar amount of CA in the screening chamber, hence an impurity in

CA would have made the yield of 7 d appear to be lower than it actually was. Thus, we

needed to quantify the amount of CA active sites present in the commercial sample. The

results described in section 5.1.2 show that an impurity was not significantly affecting the

yield of7d.

The second direction we undertook was to lower the systemic concentration of

nucleophiles in the library experiments themselves. In chapter three we discovered that

the amplification of library members suffered because the replenishable activated amino

acid on solid support carried the main zinc-binding element of the dipeptide and was not

recycled in each successive cycle. It therefore built up successive over cycles, allowing it

to effectively compete for receptor binding sites. However, in the improved libraries of

chapter four in which the recycled nucleophillic amino acid carried the zinc-binding unit,

large amounts of this nucleophile were still being used to ensure efficient dipeptide

synthesis. These high amounts could have allowed the nucleophile to again effectively

compete for binding sites, lowering the yield of the true strongest binder. The results that

94

will be described in section 5.2.3 suggest that high nucleophile concentrations were not

the cause of the low yield either.

5.1.2 Testing Enzyme Purity

To quantitate CA active sites, we performed a series of parallel, two-chamber

experiments in which CA occupied a binding chamber and Pronase occupied a

destruction chamber. Various amounts of a known, sub-IlM inhibitor (4a) were added to

the binding chamber (4.0 eq, 2.5 eq and 1.0 eq with respect to CA) in separate, parallel

experiments. The association constant Ka for 4a was ~lx106 M- I and was therefore

expected to bind to nearly aIl available CA active sites based on its potency. In the

experiments where there was more than one equivalent of 4a, the excess amount should

not have been able to bind, and would therefore rapidly diffuse out of the binding

chamber and be destroyed by Pronase. Once the excess 4a had been removed, the amount

of the remaining, bound dipeptide could be estimated by observing the inflection point in

the rate of its disappearance. The amount of inhibitor left at the point at which its

destruction becomes slow should give an estimate of the amount of receptor active sites

present. The decrease in 4a concentration in each parallel experiment that used an excess

of the inhibitor should aIl exhibit an inflection point at the same concentration: that at

which the remainder is bound to CA. The experiments that did not have an excess of 4a

should not exhibit an inflection point. The results of this experiment are summarized in

Figure 5.1.

95

1.8

1.6

<:S 1.4

] 1.2

5 LO t-;;::::--_"':'''"'''II;::-:;;::-:-~q::::;:::;;:::-i 0.8

$ 0.6

0.4

0.2

O.O+--~~-~-~-~-~-~

o 10 12 14

Time(h) 2.1 1.6 l.l 0.6

Equivalents of CA

Figure 5.1:(1) Quantification of CA binding sites by observing the differences in the rate of destruction of

bound and unbound inhibitor 4a. Varying equivalents of 4a were placed in a chamber containing CA in

each experiment (3.9 eq (+), 2.3 eq (.) hatched lines, and 1.0 eq (Â». Diffusing into a second chamber

containing Pronase should rapidly destroy excess amounts of 4a. The point at which the rate of

disappearance of 4a becomes much lower is the point at which the rate-determining step in the destruction

changes from diffusion across the membrane (fast) to dissociation from CA (slow). The molar amount of

4a left at this point should correlate with the amount of CA active sites present. a) The concentration profile

over time of 4a. The y-axis shows the concentration of 4a in terms of equivalents of CA. The solid

horizontal line extrapolated to the y-axis corresponds to the point of the greatest rate of change in 4a

concentration in the 3.9 eq experiment and crosses the axis at -0.98 eq of CA (the x-axis value is taken

from the midpoint between the two equivalents of CA exhibiting the highest rate of change of [4a]. The

large hatched horizontal line extrapolated to the y-axis corresponds to the point of the greatest rate of

change in 4a concentration in the 2.3 eq experiment and crosses the axis at -1.1 eq of CA. There was no

significant rate of change in 4a concentration in the 1.0 eq experiment (as expected). A blank experiment in

which there was no CA showed rapid hydrolysis of aIl 4a (data not shown). b) The % rate of change of 4a

concentration versus amounts of 4a (in terms of CA equivalents) in order to better iIIustrate the inflection

point on the concentration versus time graph. Looking from left to right on the graph, one can see the

decreasing amounts of 4a measured along the x-axis. The y-axis shows the % 4a concentration of the

concentration measured in the previous time point (aIl time points are exactly one hour long). The solid

vertical line corresponds to the point of the greatest rate of change in 4a concentration in the 3.9 eq

experiment and extrapolates to -0.98 eq of CA. and the large hatched verticalline corresponds to the point

of the greatest rate of change in 4a concentration in the 2.3 eq experiment and extrapolates to -1.1 eq of

CA.

96

In the experiments using 4.0 and 2.5 eq of 4a, the inflection point in the rate of its

destruction (at 0.98 eq and 1.1 eq of CA respectively) correlated with approximately

100% of the amount of CA that was supposed to be present in the binding chamber.

Further, in the experiment with 1.0 eq of 4a, no inflection point was observed. A control

experiment with no CA showed rapid destruction of aIl 4a added to the binding chamber

as expected. These results showed that the molar amounts of CA as calculated from its

weight were accurate measures of the molar amounts of binding sites. This shows that an

impurity was certainly not accounting for 70% of the calculated mass of CA, and was not

significantly affecting the calculated yield in the library experiments.

5.1.3 Lowering the Concentration of Nucleophiles

In the libraries of chapter four, high concentrations of nucleophiles 1 and 6 were

used to ensure efficient synthesis of the library. The amount of 1 initially added to the

synthe sis chamber was eight times the amount of any one of the eight dipeptides

synthesized in one cycle. Since 1 is a moderate inhibitor of CA, its high concentrations in

the system may have allowed it to effectively compete for binding sites with more potent

library members. Initial calculations showed that the amounts of 1 in the 16 h library

experiment were high enough that if aIl of it was present in the binding chamber at once,

it could occupy approximately 50% of CA binding sites. We tested whether or not

minimizing the amount of 1 would free CA binding sites and allow for higher levels of

dipeptide binding.

97

First, static libraries were synthesized using different concentrations of 1 to

determine the minimum concentration of nucleophile needed for efficient synthesis. The

results of the se experiments are summarized in Table 5.1.

Table 5.1: Effect on synthetic rate and yield due to changing the concentration ofnucleophile 1.

Maximum ~ield {l:!moQ Time to

Concentration of 1 reach

Experiment (mMt

7a 7b 7c 7d maXlmu m yield

~h2 A 10 2.5 6.5 2.1 3.7 1.0 B 5.0 2.7 6.5 2.4 4.9 2.5 C 4.0 3.2 8.0 2.9 5.1 3.0 D 3.0 3.4 7.8 3.2 4.9 4.0 E 1.0 1.6 4.1 1.8 3.8 48

a In ail cases there was an excess of nucleophile with respect to the total amount of activated amino acids.

In the previous experiment (Figure 4.3c), the amount of nucleophile added was

approximately 9 mM (close to experiment A in Table 5.1). With 10 mM ofnucleophile,

the maximum synthetic yield of aIllibrary members was reached in one hour. The yield

of aIl dipeptides in experiments A-D were approximately the same, but took increasingly

longer to maximize as the concentration of nucleophile was lowered. Experiment E, in

which only 1 mM of nucleophile was used, was the only one to show significantly lower

yield, probably because at this low concentration, water is able to out-compete 1 in the

coupling reaction.

These results showed that a 3 mM concentration of nucleophile (Table 5.1,

Experiment D) was the lowest possible that could be used to ensure a comparable

coupling yield to the 10 mM nucleophile case. However, the rate of library synthe sis was

98

much slower, taking four hours to reach the maximum yield. We did not want significant

attenuation of the synthesis, so we performed a new pseudo-dynamic library experiment

identical to the 16-h/cycle experiment in Figure 4.3c, but with a combined nucleophile

concentration of 1 and 6 of 4 mM (Figure 5.2).

004

0.3

0.3

0.2 (mM)

0.2

0.1

0.1

0.0 0 20 40 60 80 100

Time (h)

Figure 5.2: Six 16-h cycles of an eight-membered p-DCL with [nucleophiles] = 4 mM. Concentrations of

sulfonamide containing dipeptides 7a (.), 7b ( .... ), 7c (e) and 7d (+) over the course of experiment show a

similar pattern to a 16 h experiment with [nucleophiles] = 9 mM (Figure 4.3c). However, the yield of 7d

drops from 29% to 23% and the ratio between 7d and 7c drops from > 1 00 : 1 to 4 : 1.

Rather than increasing the yield, this experiment gave a lower yield of dipeptide

7d (23%). The selectivity also dropped significantly. This was not what we had expected.

We had hoped, based on comparisons of diffusion and library synthe sis rates using 4 mM

ofnucleophiles that there would be no appreciable difference from experiment 4.3c in the

amounts or rates of dipeptides that were entering the screening chamber. However, since

the yield of 7d was lower, and since it appeared as though the concentration of 7d was

increasing at the end of the sixth cycle (Figure 5.2), we hypothesized that the 4 mM level

of nucleophiles must have attenuated the introduction of library members to the receptor

99

to such a degree that they had not had enough time to build up in the screening chamber.

Renee, the experiment was repeated but allowed to undergo more than six cycles (Figure

5.3).

0.45

0.40

0.35

0.30

0.25 (mM)

0.20

0.15

0.10

0.05

0.00

0 50 100

Time (h)

150 200

Figure 5.3: Twelve 16-h cycles of an eight-membered p-DCL with [nucleophiles] = 4 mM. Concentrations

of sulfonamide containing dipeptides 7a (.), 7b (.â.), 7c (e) and 7d (+) over the course of experiment

show a similar pattern to a 16 h experiment with [nucleophiles] = 9 mM (Figure 4.3c). The yield and

selectivity return to those of the experiment shown in Figure 4.3c (30% yield of7d and a ratio of7d : 7c of

> 1 00 : 1), but do not get higher.

The extended cycles brought the yield and selectivity back up to Figure 4.3c

levels, but did not increase them. Additional experiments were conducted in which the

amount of electrophiles (and hence the amounts of the eight dipeptides synthesized each

cycle) were increased (Figure 5.4). These experiments did not improve the yield, and

lowered the selectivity, even over 16 cycles. The results of the investigations into

increasing the yield are summarized in Table 5.2.

100

0.50

0.45

0.40

0.35

0.30

mM 0.25

0.20

0.15

0.10

0.05

0.00

0 50 100 150 200 250

Time (h)

Figure 5.4: Sixteen 16-h cycles of an eight-membered p-DCL with [nucleophiles] = 4 mM, and increased

amounts of electrophiles added each cycle. Concentrations of the strongest binder 7d (+) plateaus at 30%

over the course of the experiment, and sulfonamide containing dipeptides 7a (.), 7b ( ... ), 7c (e) are not

completely removed from the screening chamber.

Table 5.2: Effect on yield and selectivity due to changing nucleophile, electrophile and number of cycles

in the eight-membered p-DCL using a 16 h cycles time.

Equivalents of Equivalents of

# Cycles to activated amino Maximum

Number nucleophile 1 acid added each ratio between

reach Yield of cycles with respect

cycle with respect 7d and 7c maximum of7db

to CAa

toCA ratio

6c 9.6 3.2 > 100:1 4 30 6d 4.5 3.2 4:1 6 23 12e 4.5 3.2 > 100:1 8 31 16f 4.5 9.6 3:1 16 30

a In alllibraries Phe (6) was present in the same amounts as 1, but because 6 is a non-inhibitor, its presence

did not affect the yield or selectivity.

b Yield refers to the percentage of CA binding sites occupied by inhibitor 7d.

c Experiment 4.3c.

d Experiment 5.2.

e Experiment 5.3.

101

fExperiment 5.4.

Neither extending the same library to 16 cycles, nor adding more active esters per

cycle improved the yield of 4d. Thus we concluded that in this particular system, the

concentration of nucleophiles in the system was not affecting the yield of the strongest

binder.[2]

5.1.4 Discussion: A Steady State Concentration of the Strongest Binding Dipeptide

Since aspects of the synthesis were not significantly affecting the end yield of the

tightest binder, and the enzyme was pure, we hypothesized that the maximum yield for

dipeptide 7 d in an eight-membered library using a 16 h cycle time must have been

reached. The only two factors that remain that can affect the final amount of the tightest

binder in libraries with equal numbers of members are the absolute binding strength of

the inhibitor, and the time allowed for its destruction (the effect of the number of

inhibitors on the yield and selectivity will be discussed in section 5.3). If an inhibitor has

a high binding strength, it binds tightly to a receptor and spends less time in the unbound

state. This leaves less of the compound available for destruction, and ensures that its

bound levels remain at a certain level. After X number of cycles of synthesis and

destruction, a steady state concentration of the inhibitor will be reached, as each cycle it

is synthesized to the same degree, has to compete for binding sites to the same degree,

and spends the same time equilibrating with the receptor and being destroyed. The

number of cycles (X) needed to reach this steady state will depend on the amounts of

library members that are synthesized each cycle.

102

A second factor contributing to the final yield is the time allowed for destruction

before the next synthetic burst. The longer the time allowed for destruction, the more of

the compound will unbind and be destroyed. In chapter four we saw that the time allowed

for destruction was crucial in determining the selectivity. Shorter destruction times allow

more synthesis relative to destruction, and hence give higher levels of alllibrary members

in the system, and thus higher amplifications. However, if the destruction time is not long

enough to remove all but the most strongly binding compound the selectivity will

necessarily be lower. The selectivity and amplification must be simultaneously optimized

in each new p-DCL.

It is important to note that the conclusion that decreasing the cycle time leads to

increased yield, although seemingly true for the 16 h cycle, is not directly applicable to

all systems. In chapter four, (Figure 4.3b), the cycle time was shorter (12 h), but the yield

of7d was only 15%. This lower yield with a shorter cycle time seemingly contradicts the

conclusion that shorter cycle times lead to higher yields. However, in the 12 h experiment

inhibitor 7b was still present to sorne degree in the system at the end of four cycles.

Dipeptide 7b is synthesized 1.5-fold more rapidly than 7d per cycle. The effect ofthis

higher level of the compound could have been to make it an artificially better binding

competitor. The effect of the higher levels of 7b on the yield of 7d, and other library

members at this shorter cycle time are not yet fully understood. However, sorne

experiments (section 5.2) help explain the relative effects of having greater or fewer

inhibitors in the library with a 16-h cycle time.

103

5.2 The Effect of the Number of Inhibitors on

Amplification and Selectivity in Pseudo-Dynamic Combinatorial Libraries

5.2.1 Introduction

The effect of increasing the number of library members in a DCL is to dilute the

absolute amplification of library members (chapter one). This results in a decrease in the

absolute concentration differences between inhibitors as the library size increases. We

hypothesized that since a p-DCL had added kinetic control that this would not be the case

in our system. As a preliminary investigation into this question, we performed two, six

16-h cycle experiments, one in which only the two strongest inhibitors (7c and 7d) were

made, and another in which the strongest inhibitor was left out of the 1ibrary, creating a

library of 7a-c.

5.2.2 Two- and Three-Inhibitor 16 h Cycle P-DCLs

We performed an analogous experiment to the one described in Figure 4.3c, but

using a four-membered library consisting of two inhibitors 7c and 7d, and negative

controls 8c and 8d (Figure 5.5). After six-16 h cycles the ratio of7d : 7c was only 10 : 1,

much lower than the> 100: 1 in the analogous eight-membered experiment. The yield of

7d was ~40%, and that of 7c was ~4%, both of which were higher than the ~30% and

~O% respectively, se en in the eight-membered system. These results suggest that the

104

yields of aU library members increase when there are fewer library members in the

system and that as a consequence ofthis the selectivity is lower.

0.70

0.60

0.50

0.40 mM

0.30

0.20

0.10

0.00

0 20 40 60 80 100

Time (h)

Figure 5.5: Results of a four memberedp-DCL, with two inhibitors 7c (e) and 7d (+) over six-16 h

cycles. The ratio of7d : 7c is lower than in the analogous eight-membered library at only 10 : 1. The yield

of each dipeptide is higher than in the eight-membered library at -40% for 7 d and -10% for 7 c.

In a similar experiment to that of Figure 5.4, a six-membered library was created,

this time excluding the strongest binder, 7d (and its corresponding negative control, 8d).

The results (Figure 5.6) showed that 7 c (now the strongest inhibitor) was present in

highest concentration, but with the weaker inhibitors 7a and 7b still present (at less than

1 % yield each) after six cycles. The yield of 7 c was ~9%, which is higher than in the

four-membered library containing the potent 7d where the yield of 7c was only ~4%, and

than in the eight-membered library in which it was ~O%.

105

0.50

0.40

0.30

mM 0.20

0.10

0.00

0 20 40 60 80 100

Time (h)

Figure 5.6: Results of a six membered p-DCL over six 16-hour cycles with three inhibitors 7a (.), 7b ( .... )

and 7c (e). 7c, now the strongest inhibitor in the library is selected as the strongest binder by the p-DCL

system. At 9%, the final yield of7c is higher than it was in the four-membered library. The yields of7a and

7b are both ~1%. Note: The sampling frequency in this experiment was higher than in Figure 5.5. Ali

experiments in which the sampling frequency was high showed the same cyclical behaviour in each

inhibitor's concentration as seen in the above graph (for example, the experiments shown in Figure 4.3a and

b). However, for the sake of comparison of Figure 4.3a and b to 4.3c, sorne of the data points were omitted

for clarity,

5.2.3 Discussion: Fewer Inhibitors Increase Amplification, but Decrease Selectivity

inP-DCLs

The above results demonstrate that more inhibitors in the system lead to more

competition for binding sites. Furthermore, a weak binder will be able to compete more

effectively with a library member of medium strength than with a strong one. A

comparison between the results in Figure 5.4 and Figure 4.3c shows that removing

weaker binders 7 a and 7b from the system leads to less overall competition for binding

sites. This in turn increases the yields of both 7d and 7c, and gives higher amplification

106

for both compounds. This consequently results in a lower selectivity between the two.

The yield of the second strongest binder, 7c increases from 0% to 4% and the yie1d of7d

rises from 30% to 40%. Adding weaker inhibitors should decrease the yields ofboth 7d

and 7c, but decrease that of7c to a greater degree.

Comparison of the results shown in Figures 5.4 and 5.5 show that 7d has almost

double the effect on the levels of 7c than do 7a and 7b combined (7c's concentration rose

to 4% due to the lack of 7a and 7b (Figure 5.7) but rose even more (to 9%) due to the

lack of only 7d. Competition for binding sites has a relative effect on each inhibitor that

favours strong binders, depending on the number of other compounds present, and on the

binding strengths of those compounds.

The number of compounds in the library represents a third factor influencing the

amplification of the strongest binder, and interestingly, fewer inhibitors lead to lower

selectivity. Further experiments in the 16 h system will be needed to determine whether

the converse of this result is true, that is, that more inhibitors leading to higher selectivity.

5.3 Expansion of the Pseudo-Dynamic Combinatorial Library

5.3.1 Introduction: New Library Members and P-DCL Scheme

With several successful experiments, and a greater understanding of the nature of

pseudo-dynamic libraries, the next step was to exp and the size of the library by adding

more electrophilic amino acids, and new nuc1eophilic amino acids. In the experiments

described thus far experiments we had chosen an aryl sulfonamide moiety to impart

107

carbonic anhydrase affinity because this unit is known to display affinity for the zinc ion

in the active site of CA. [3] Studies have indicated that hydroxamic acids can impart even

higher affinity for zinc.[4] Our existing library consisted of aryl sulfonamides (series 7)

and negative controls (series 8). We added three new series of amino acids carrying

potential receptor-binding functionalities: alkyl sulfonamides based on sulfonamido

alanine (series 9) to complement the aryl sulfonamides, sulfamic acids based on

sulfonamido lysine (series 10) out of interest, and hydroxamic acids formed from

y-hydroxyglutamine (series 11) which would hopefully generate stronger inhibitors than

the micromolar library hits from series 7 (Figure 5.7). This brought our total number of

nucleophiles to five.

We used six amino acids as our solid-supported electrophiles: Phe (a), Leu (c),

Pro (d), Val (e), Ala (t) and pipecolic acid (g). Ala served as a replacement for Gly (b),

which was not included because its retention time was too low using the HPLC

conditions necessary to separate aIl the compounds in the new library. Val and pipecolic

acid were added to increase diversity. Pipecolic acid was especially interesting because of

its structural similarity to proline, which had formed the dipeptide 7d (EtocProPhesa), the

strongest library member in our previous experiments. This gave six electrophiles that,

when combined with the five nucleophiles, would give 30 library members (Figure 5.7).

108

Recycled Nucleophiles

o H2NyÂ-OH

R3

+

Replenishable Electophiles

Eto.C 0

Rz'N0oH ~1

Series (7): Phe.a Series (8): Phe

H 0

o H2N0oH

" o,:.S;o NH2

Series (9): Ala.a

o H2N0oH

)h HN,~O

~,s .. o NH2

Series (10): Lys ..

o H2N0oH

~O HN

OH Series (11): GIUNHOH

E102C,Nyt-X H 0 0 H 0 H 0

Et02Cfx

E102C,N yÂ-x E102C'NJ E102C,N0x Et02C,N0x

QY --( \.Jx CH.

(a):Phe (c):Leu (d):pro

Representative dieptide •

~ 0 C02EI . ~N HO C ......... N . 2

2 H R1 R '------', , ,

7-11 a,c-g

A.-

(e):Val (f):Ala (g):Pipecolic acid

Figure 5.7: The expanded library members. 30 dipeptides are synthesized in five series based on the

coupling offive nucleophillic amino acid-derivatives (7: Phe.", 8: Phe, 9: Ala.a, 10: Lys.a, and 11: GlUNHOH)

with six N-Etoc protected electrophillic amino acids (a: Phe, c: Leu, d, Pro, e: Val, f: Ala, g: Pipecolic acid.

(X = TentaGel resin).

We synthesized each of the thirty library members separately by adding each of

the nucleophiles to each of the electrophiles on solid support and analyzed them for

purity by HPLC. Control experiments showed that Pronase hydrolyzed an the dipeptides

at similar rates (aIl faster than diffusion across the dialysis membrane). We hoped to

discover a tighter binder than in previous cases in which we used six 16-h cycles to

achieve apparent absolute selectivity for compound 7d, so we performed three paralIel

library experiments with cycle times of 16 h, 24 h, and 32 h. An components were scaled

down two-fold compared to previous libraries, but otherwise we used the same

experimental procedure and three-chambered vessel (Figure 5.8).

109

(Synthesis Cham ber) F

(TentaGel~F 1'" 1 0 C0

2Et + H02C'(NH2 pH 9.0

F.... O~NR2 R3

F R'

(Screening Cham ber)

carbon le anhydrase

replenishable solid-supported •• recylcling of nucleophilic electrophllc amlno acid amlno acid (inhibltory ••••••••••••••••••••••••••••

coupling partner)

(Destruction Chamber)

Pronase from S.griseus

o C02Et . .Â...-N Innocuous by·products

HO l. 'R2

R'

Figure 5.8: Expanded library scheme. The dashed lines represent dialysis membranes separating the three

chambers (synthesis, screening and destruction) similar to ail previous Iibrary set-ups.

5.3.2 Results

In the first cycle of each of the 16-h, 24-h and 32-h cycle time experiments, the

non-inhibitory series 8 had disappeared after the first 4 h, series 9 had disappeared after 9

h, series 10 had disappeared after 12 h, and after 16 h, even series 7 was gone, leaving

only the six members of series 11 left in the screening chamber. In each successive cycle

(whether 16-h, 24-h or 32-h) series 8-10 were initially observed, but had aH depleted to

unobservable levels by the end of the cycle.

After six, 16-h cycles, aH the members from series 11 were still present in the

screening chamber. It took three 24 h cycles to wean away compounds 11a and 11d (the

Phe- and Ala-based dipeptides). One more 24-h cycle removed 11b (the Val dipeptide),

and a total of six were needed to remove 11f (the Pipecolic acid dipeptide). This left two

compounds present at the end of six 24-h cycles. These two dipeptides, 11e (the Leu

dipeptide) and 11d (the Pro dipeptide) were also the only compounds to survive five

cycles of the 32 h library. After the sixth cycle, 11e was no longer observable, leaving

11d as the sole survivor of a 30-member library after a total of 192 hours (Figure 5.9).

110

6' 2.5E+06

t 2.0E+06

~ of 1.5E+06

Sl J:>.

~ I.OE+06

~ ~ 5.0E+05

8 = ;;J O.OE+OO F---,----,---,-----'lI-r----,

o 50 100 150 200 250

Time (h)

Figure 5.9: Uncorrected UV absorbance of the final two remaining members of a 30 membered p-DCL,

lld (+) and lle (.) over six-32 hour cycles. The final remaining library member is lld, EtocProGluNHOH.

5.3 Discussion

Compound 11d was the sole survivor of six-32 h cycles in the pseudo-dynamic

library. We have not yet determined the inhibition constant ofthis compound, but aH the

evidence to date suggests that its survival is due to its being the strongest binder in the 30

membered library. Resolution of 11d from other series 11 members required a much

longer cycle time than did the members of series 7 in the eight-membered library (Figure

4.3c). Series 7, and most notably dipeptide 7d was able to survive several16 h cycles in

the eight-membered case in which there were no stronger binders in the system.

However, the Phesa series (7) was eliminated from the library in less than 16 h in the first

cycle of each of the 16 h, 24 h and 32 h cycles in the expanded library. This suggests that

the selectivity in p-DCLs does in fact depend on the relative binding strengths of the

library members, and that stronger binders in the system will indeed help eliminate

weaker ones, making the hit compounds easier to identify. Hence, p-DCLs are biased

111

towards the retention of strong binders, which is a fundamental requirement for success

in receptor-assisted combinatorial chemistry. However, the results of this experiment,

including the yield and KI of Bd need to be confirmed with an authentic sample.

5.4 Preliminary Modeling of Pseudo-Dynamic Combinatorial Libraries

5.4.1 Introduction

A fundamental problem in traditional dynamic libraries is that the amplification of

tight binders becomes diluted as the number of binders in the library increases. Alllibrary

members will be present in amounts corresponding to their binding affinities in an

equilibrated system because of the thermodynamic control over both amplification and

selectivity. In large libraries, the relative concentration differences between two binders

will not change as compared to a small library, but the absolute amounts of all library

members will decrease, making the differences in concentrations difficult to observe. We

sought to determine whether the same is the same true for pseudo-dynamic libraries. We

also wanted to know the degree to which kinetics allow for greater selectivity than

thermodynamics alone, what the effects were of having stronger inhibitors in the system,

and how much the cycle time would need to be adjusted to distinguish between inhibitors

of nanomolar strengths. We attempted to develop a mathematical model the pseudo

dynamic library to answer these questions.

In a p-DCL, three distinct processes occur: library synthe sis, binding to the

receptor, and destruction of unbound library members. Once the library is synthesized,

112

the library members must then traverse a dialysis membrane to interact with the receptor

where they can equilibrate based on binding affinity. Another membrane must then be

crossed before the last process, a kinetic destruction of unbound library members, can

occur. Destruction products can then diffuse back through both membranes to initiate

another round of synthesis and begin the cycle again. After the first cycle, aIl these

processes occur simultaneously in a p-DCL experiment. Modeling the entire process

simultaneously with one equation proved difficult, so the mathematical model consists

instead of three functions applied in succession, which are iterated over several cycles.

This is a major assumption of the model that may render it less accurate when analyzing

full p-DCLs. However, this model does allow for comparison of each process (synthesis,

binding and destruction) with experimental results. We found that the three processes in

the model correlate weIl with experiment, and that the integrated model gave similar

results of amplification and selectivity to an analogous experiment (shown in Figure 5.5).

5.4.2 Modeling Synthesis

Library synthe sis consists of coupling between a nucleophile and an electrophile.

The electrophile is an activated amino acid on solid support, so this reaction may exhibit

pseudo-first order kinetics. However, library members must diffuse across a dialysis

membrane before entering the screening chamber. Since diffusion is a first order process,

a first order equation should accurately model the apparent rate of synthesis (due to

synthesis and diffusion) in our p-DCL. Control experiments have shown that alllibrary

members diffuse accross 1 and 12 kDa MWCO membranes at similar rates. Higher (or

113

lower) order rate equations can easily be applied to model synthe sis as required by each

new type of synthesis. [5]

Synthesis of a dipeptide using a solid supported active ester is illustrated in Figure

5.10, where AA is the solid supported electrophile, Nu is the nucleophile, 1 is the

dipeptide inhibitor formed by the coupling reaction, and ks is the first order rate constant.

AA+ Nu ks. 1

Figure 5.10: General Scheme for library synthesis. AA is the solid supported activated amino acid, Nu is

the nucleophile, l is the dipeptide inhibitor formed by the coupling reaction and k. is the tirst order rate

constant for the coupling reaction.

The rate of the reaction of nucleophile to inhibitor, [1] is shown by equation 1.

v = _ d[Nu] = d[I] = k [Nu] dt dt S

The relationship between [1] and [Nu] is shown by equation 2.

[I] = [Nuo ] - [Nu] (2)

The simplest differential equation from Equation 1 is Equation 3.

d[Nu] =-k dt [Nu] S

(3)

The definite integral of 3 is given by Equation 4.

[Nu], In--2 = -k (t - t )

[Nu]'1 S 2 1 (4)

Using [Nu]o as the nucleophile concentration at tl = 0 and rearrangement gives Equation

5.

114

(5)

Substituting 5 into 2 gives Equation 6.

(6)

Equation 6 gives [1] in terrns oftime and initial [Nu]o, with a rate constant (ks) that can be

modified depending on the synthetic rates of individual inhibitors in the library. [2]

5.4.3 Modeling Receptor-Binding

The second event in a p-DCL is binding to the receptor. This is a

thermodynamically governed equilibration that favours tight binders. The relative

amounts of two compounds competing for binding sites were determined in order to

model this event (Figure 5.11).

lA + T KaA

IA-T ,

K'A

IB+ T KaB

IB-T K'B

Figure 5.11: Two inhibitors lA and lB compete for a biological target T (the receptor) with binding affinities

KaA and KaB' and inhibition constants K1A and KIB where Ka = K/l.

The equation for the inhibition constant of any inhibitor 1 is given by:

(7)

115

The total concentration of bound and unbound forms of 1 can be expressed as:

(8)

where [10 ] is the concentration ofinhibitor 1 after synthesis, and [1] is the concentration of

1 that is not bound to the receptor-target T (i.e. excess 1 from the synthesis).

Equation 6 is true for two inhibitors lA and lB if one assumes tight binding, that is,

at equilibrium nearly an the target is bound to an inhibitor (true for sub millimolar

inhibitors under our typical experimental conditions [7]).

(9a)

or equivalently:

(9b)

Taking the ratio of Equation 7 for each inhibitor lA and lB, gives:

(10)

Solving for [lA • T] gives:

(11)

Substituting Equation 9 into 8 in terms of [lB] gives:

(12)

Substituting Equations 12 and 9b into Il gives Equation 13.

(13)

Multiplying through gives the quadratic equation 10:

116

(KIIA - K1IJ[IA e Ty + (KhJ IBo]- K1IJTJ + KIJTJ + KIIJIAo])[IA eT]- K1IJTJ[ lA 0] = 0 (14)

Equation 10 can be solved using the quadratic formula and gives:

Similarly, solving for [lB eT] gives:

Equations 15 and 16 give the concentrations of each of the inhibitors lA and lB

bound to the receptor given that their initial concentration in the binding chamber after

synthesis [lAo] and [IBo], the concentration of the receptor-target T, and their inhibition

constants are known. AU of these parameters are easily accessible in an actual

experiment, and provide wide versatility in the parameters that can be tested in model p-

DCLs. Unfortunately, modeling more than two inhibitors at once becomes very complex.

5.4.4 Modeling Destruction

The last step in ap-DCL is the destruction ofunbound inhibitors (Figure 5.12). To

model this, we assumed that the equilibration process was fast compared to diffusion

(typical kon rates for sulfonamides range between 0.0033-31 x106 M-1s-1, or at least 3

associations of inhibitor to receptor per second at a (typical) 1 mM concentration of

library member, and koffrates range from 0.01-0.05 S-I)JS] After equilibrating, aH inhibitor

117

present in excess of the receptor was assumed to be hydrolyzed before the destruction

equation was applied.

I·T T+I _k_d_ .. ~ Nu + AA

Figure 5.12: The destruction of any unbound inhibitor 1 governed by its inhibition constant KI and its

kinetic rate constant kd. The destruction rate for ail compounds in the library should be the same, so the

removal of compounds should be based only on how far they lie on the unbound side ofthe equilibrium.

From Chapter two the destruction rate of an inhibitor 1 is given by equation 17.

d[IT] kdIKI)IT] --=-

dt [T]

Integrating equation 13 gives:

(18)

where [Ir] is the concentration of free inhibitor, [la] is the concentration of the inhibitor

bound to the receptor-target after the equilibrium of binding, [1'] is the concentration of

free target, t is time, kd is the destruction rate constant and Kd is the inhibition constant 1 1

ofthe inhibitor, I. Solving 18 for [Ir] gives equation 15:

(19)

Equation 19 gives [Ir] in terms of time, with manipulable parameters for the destruction

rate kd, and the inhibition constant KI. The concentration of free receptor-target T is

assumed to be constant. Although strictly not true, at higher conversions of inhibitor to

118

starting materials (at which points, from chapter two the selectivity starts to increase

dramatically), the [11»[Ir], and changes in [1] due to the change in concentration of 1

become negligible.

5.4.5 Discussion: The Integrated Model and its Comparison to Experiment

We applied Equations 6, 15 or 16 (for either lA or lB respectively) and 19 in

succession in a spreadsheet and then iterated the three equations over time using 16 h

cycles. Since the theoretical curves modelled two inhibitors it was directly compared to

the two inhibitor experiment shown in Figure 5.4 to normalize the ks and kd parameters to

give similar amounts of synthesis and destruction each cycle. [10] A curve was generated

showing the ratio of [lA] : [lB] over six, 16 h cycles (Figure 5.13).

6

= ... ~ 5 :S -= 4 0: ... < 3 ... ;§ ..Q

2 :El .s .. 1 '" .. =:

0

0 16 32 48

Time (h)

64 80 96

Figure 5.13:[9) Theoretical curve of the ratio oftwo inhibitors with K1s of 1.0 J.1M (+) and a 2.5J.1M (e)

generated from the above equations. Parameters for the curve were generated to most closely mimic the

amounts of synthesis and destruction observed in a 16 h experiment with two inhibitors (Figure 5.5). The

relative strengths of these inhibitors were chosen to correspond to those of the two strongest library

members, 7d and 7c respectively.

119

In this theoretical experiment, the ratio of the stronger inhibitor to the weaker

starts at one, corresponding to the equal synthesis of each inhibitor. The ratio jumps

during the short equilibration process (short compared to the time needed for synthesis

and diffusion-see reference #8) and then grows quickly over the destruction process. The

ratio drops towards one as a new cycle of synthe sis occurs, and then rises somewhat due

to the thermodynamic equilibration, and even more during the next destruction. Over the

next few cycles, the ratio seems to plateau reaching approximately 5.6 : 1 (lA: lB) after

six cycles.

This result is consistent with the steady state concentrations the inhibitors reach

after six 16-h cycles shown in Figure 5.4. The stronger inhibitor's concentration increases

with each successive cycle, and the weaker inhibitor's concentration decreases, which is

also observed experimentally. The ratio between the two inhibitors is 5.6 : 1 in the model,

compared with a ratio of approximately 10 : 1 seen experimentally.

The ratio of the two inhibitors after the first equilibration process is 1.7 : 1. This

theoretical number models the thermodynamic effect in the first cycle of the p-DCL. It

can be directly compared with an earlier result shown in chapter two, Figure 2.3, in which

inhibitors are allowed to selectively concentrate into a chamber containing carbonic

anhydrase (with no Pronase, therefore only thermodynamic effects could take place). In

that experiment, the ratio between the two strongest inhibitors 4a and 4b (with K1s of 1.2

/lM and 2.5 /lM respectively) was 1.1 : 1. The theoretical model may give a higher ratio

because there is an initial excess of each inhibitor associating with CA, which would

increase the thermodynamic selectivity, and because the modelled inhibitors have a

slightly higher ratio of inhibition constants.

120

After the first round of destruction, the ratio of inhibitors is 4.2 : 1. This higher

ratio is a result of the kinetic destruction influence. This theoretical result can be

compared with those depicted in Figure 2.6 in which 4a is selectively protected from

destruction over 4b. The ratio of 4a : 4b after 183 h in this experiment was 3.8 : 1. These

results cannot be directly compared with the theoretical ones because the theoretical

result from the first cycle did not start with the same concentration of inhibitors (the

equilibration process happened first). However, if one takes the destruction process alone,

and applies it to two inhibitors of the same strengths as 4a and 4b over 183 h, one finds a

final ratio of ~ 10 : 1. In the example shown in Figure 2.6, the starting concentration of the

weaker 4b was slightly higher than that of 4a, which would lower the final ratio.

However, discrepancy between these two numbers cannot as yet be fully explained.

The model seems to correlate with experiment. It suggests that the

thermodynamic contribution common to all p-DCLs plays a smaller role in determining

selectivity than does the kinetic destruction. The model also shows a potential drawback

to p-DCLs. Since the time needed to achieve high selectivity depends only on the relative

binding constants, p-DCLs can distinguish between inhibitors with similar binding

constants. However, the absolute strength ofthe binders will determine the time per cycle

needed by the destruction reaction to resolve them. If the strengths of the inhibitors are

increased ten-fold, the system requires a ten-fold increase in the destruction rate, or a ten

fold increase in the cycle time will be needed to achieve the same selectivity as when the

inhibitors were ten-fold weaker. The instability of many biological targets may render

these lengths of time impractical. However, this model does not take into account the

possible effect of having large library sizes.

121

The theoretical model assumes tight binding. This may not always be true of aIl

libraries and therefore the model will not predict what will happen when there are no

potent inhibitors in the library. This model also treats synthesis, binding and destruction

as completely separate events, which is not the case in an experimental p-DCL in which

diffusion of starting materials causes sorne attenuation of synthesis in aIl cycles but the

first, and thus may alter sorne of the theoretical results from reality. The final, main

limitation is that these theoretical models can only mode! two inhibitors at a time, and

thus cannot address the fundamental issue of increased selectivity in the presence of more

competition.

5.5 Overall Conclusions

The selectivity of a p-DCL is high enough to distinguish between inhibitors with

similar binding constants. Because decreasing the number of library members leads to

lower selectivity, it may follow that increasing the library size can actually improve the

selectivity. However, resolving stronger inhibitors requires longer cycles. The level of

amplification is a result of the strength of the inhibitor and the time given for destruction

per cycle.

In pseudo-dynamic libraries, thermodynamics provide the essential initial

selective binding to the receptor, but, as in dynamic libraries, this selectivity is often low.

The kinetic destruction during the temporary absence of synthesis winnows away non-,

and poor inhibitors and greatly improves the selectivity for the best binders. Iteration of

the synthesis-binding-destruction cycle aIlows betier binders to build up in the system,

122

giving amplification. Iteration also improves selectivity by re-introducing strong binders

to the system, allowing them to replace weaker binders that were not completely removed

in the previous destruction cycle.

The yield of any compound in a p-DCL is dependent on three factors: its absolute

binding strength, the length of time the destruction reaction is allowed to proceed

unabated by new synthesis, and the strength and number of other competitive binders in

the library. It is possible to allow the destruction process to run for too long, which would

decrease the yield of the best binding compound to undetectable levels. Therefore, in

each new p-DCL, because the selectivity increases as the destruction reaction proceeds, it

is necessary to optimize the cycle time to allow the destruction process to run long

enough to achieve good selectivity without decreasing the yield too much.

The selectivity is higher than in DCLs, and the particular receptor type should not

influence the success of these systems as it does in RAS. Further, no lock-in reaction is

needed as the library members are synthesized irreversibly. By combining the inherent

thermodynamic selectivity of a receptor towards a library of inhibitors of various

strengths with a kinetic removal of weak binders, the selectivity of a receptor-assisted

combinatorial system can be vastly increased .. Being able to iterate synthesis and

destruction steps, and to optimize the amount of resolution allowed to occur, gives a level

of control here-to-fore impossible in receptor-assisted methods.

The mathematical model appears to indicate that the destruction process is the

main determinant of selectivity. Since the destruction rate depends on the strength of the

inhibitors in the library (how fast they can unbind the receptor and be destroyed) stronger

inhibitors will require longer cycle times for resolution. The resolution time required by

123

very potent inhibitors (and hence the most interesting ones) might not be compatible with

receptor stability. This raises an important design issue in p-DCLs, in which receptor

stability must be made compatible with the apparatus and length of time needed for

selectivity. This may become a difficult requirement when using unstable receptors such

as membrane-spanning proteins.

Another design issue in p-DCLs is the nature of the kinetic destruction

component. The part of a p-DCL is not limited to peptide hydrolysis, but could include

other chemical reactions that destroy unbound library members, or physical separation

steps that remove them. An optimal destruction method must be designed to complement

each new type of library synthesis.

5.6 Future Endeavours

5.6.1 Improving the p-DCL Model

Currently the model can only handle two inhibitors at a time. This major

limitation must be overcome if the model is to have any predictive value for library

experiments practical size for drug discovery. Existing models in the field of systems

biology may be helpful directions to take in this regard. [11] The model also shows sorne

fundamental weaknesses in its treatment of certain variables, such as the concentration of

free target (assumed to be constant during the destruction process). Other assumptions

(such as the assumption that aH target is bound by inhibitors at the end of the

124

equilibration process) need to be eliminated as well before the model can be trusted to

give accurate predictions of the outcome of unknown p-DCLs.

5.6.2 Fundamental Experimentation in p-DCLs

Sorne future experiments in the eight-membered system will include further

characterization of the origins of amplification and selectivity with specific focus on the

effect of having more inhibitors in the system. A 12 hour cycle p-DCL experiment such

as the one shown in Figure 4.3b but with five or six inhibitors instead of four will directly

determine whether more inhibitors indeed leads to greater selectivity. Other 16 hour cycle

experiments using varying equivalents of each of the weaker inhibitors 7a, 7b, and 7c

will help determine the relative effects of inhibitors of various potency on the

amplification of the strongest inhibitor 7d. A 16-hour cycle experiment with no 7c will

also help determine the relative effects of weaker inhibitors 7a and 7b on 7d, and will be

directly comparable with the knockout experiment described in Figure 5.8.

5.6.3 Expansions and Miniaturizations

Future projects will be focused around the practical application of the p-DCL

method. Synthesis of new potential inhibitors will continue to be needed to create large,

125

diverse libraries suitable for p-DCLs. Since these experiments necessarily use

stoichiometric amounts of the receptor, a practical application would also require that the

experiment take place in a very small reaction vessel, so as not to use copious amounts of

the receptor. To this end, miniaturization of the experimental design is one of the most

important aspects that needs to be developed.

A new reaction vessel has only to prevent synthetic, or proteolytic modification of

the receptor. Dialysis equipment, including Microdialysis Buttons™ that separate JlL

volumes are available from Hampton Research (Hampton Research; Laguna Niguel,

CA)y21 These would be ideal reaction vessels in which to perform small-scale p-DCLs.

However, before this is possible, a new, soluble solid-phase library synthesis needs to be

developed because the small volumes will not be sufficient to swell a resin such as

TentaGel. Derivitized supports based on polyethylene glycol may be suitable for these

purposes.


The author (Jeremy D. Cheeseman) with David Soriano deI Amo developed and

performed the synthesis of the library members shown in Figures 5.9-5.13.

The author (Jeremy D. Cheeseman) deveioped the theoretical model and performed all

other experiments described in this chapter.

References

126

1. Lines drawn in aIl figures (except Figure 5.12) are for illustration purposes only.

They do not represent theoreticallines of any sort.

2. This is not to say that the amount of nucleophile (assuming it carries the

inhibitory moiety) will not affect the yield in other systems. If the relative strength

of the nucleophile to the library members is increased, the nucleophile will

probably have a more notice able effect.

3. C. T. Supuran, A. Scozzafava Expert. Opin. Ther. Patents, 2002,12,217-242 and

references therein.

4. L. R. Scolnick, A. M. Clements, J. Liao, L. Crenshaw, M. HeIlberg, J. May, T. R.

Dean, D. W. ChristiansonJ Am. Chem. Soc., 1997,119,850-851.

5. It is important to note that for the purposes of examining amplification and

selectivity alone, the exact order of the synthesis is not important. For modeling

purposes it is only necessary to have the required manipulable parameters [Nu]

and ks.

6. Adapted from: 1. Tinoco Jr., K. Sauer, J. C. Wang, Physical Chemistry:

Principles and Applications in Biological Sciences 3rd Ed; Prentice Hall: Upper

Saddle River, New Jersey, 1995; pp 331-333.

7. In our experiments a typical receptor concentration [T] is 1.5 mM. Since

K = [1- T] , for Equation 9 to be true, a, [l][T] [I-T]»[l],or Ka [T]~lO.Inour ,

system, this equates to Ka, ~ 1.4xl0-4 M- 1, or a KIt of 0.15 mM or better. AlI our

inhibitors fall well below this limiting KI.

8. B. W. Clare, C. T. Supuran, Eur. J Med Chem. 1997,32,311-319.

127

9. J. D. Cheeseman, A. D. Corbett, R. Shu, J. Croteau, J. L. Gleason, R. J.

Kazlauskas J. Am. Chem. Soc. 2002,124,5692-5701.

10. Table 5.3: Parameters used to generate Figure 5.6

Inhibitor Kj(M) [Nu] (M) ks (sol) [1] (M) kd (soi)

lA 1.0xlO06 4.0xlO03 2.0xlOoS 1.4xlO03 3.0xlO02

lB 2.5xlO06 4.0xlO03 2.0xl00S 1.4xl003 3.0xlO02

Il. Q.-H. Chen, D. B. Bylund, Receptors & Signal Transduction, 1997, 7, 73-84 and

references therein.

12. Hampton Research Tools Catalogue, 2004, 11, 77-80.

128

CHAPTERSIX

EXPERIMENTAL SECTION

129

6.1 Experimental Section for Chapter 2

General Experimental:. p-Nitrophenyl acetate (PNPA), carbonic anhydrase (CA, from

bovine erythrocytes, a mixture of isozymes, C-3934) and proteases were purchased from

Sigma unless otherwise noted and used without further purification. HPLC analyses were

conducted using a Phenomenex-Cs reversed phase HPLC column (10 x 250 mm) with

detection at 220 nm, unless noted. Elemental analyses were obtained from Quantitative

Technologies Inc. Whitehouse, NJ. High resolution mass spectra were obtained from

Université de Sherbrooke, Sherbrooke, QC.

4'-Sulfonamidophenylalanine (1): N-Acetylphenylalanine (37.7 g, 178 mmol, 1 eq) was

added in portions over a 1 h period to neat chlorosulfonic acid (110 mL, 1.65 mol, 9.5 eq)

at -10°C. The resulting yellow solution was stirred at -10°C for 2.5 h, at 25 oC for 2.5

h, and then heated to 60 oC until gas evolution had ceased (approx. 12 h). The resulting

orange solution was cooled to 0 oC and poured carefully onto 750 mL of ice (Caution:

exotherm!). The resulting mixture was extracted with ethyl acetate (3 x 1 L) and the

combined organic layers were dried over Na2S04, filtered and concentrated in vacua to

afford the sulfonyl chloride (45.1 g, 83%) as an orange solid which was used immediately

without further purification .. IH NMR «CD3)2S0) 8.26-8.21 (d, IH, J = 8.5 Hz), 7.55 (d,

2H, J= 6.9 Hz), 7.22 (d, 2H, J= 6.8 Hz), 4.49-4.34 (m, IH), 3.13-3.00 (dd, IH, J= 14.4

and 6.8 Hz), 2.92-2.79 (dd, IH, J= Il.0 and 10.2 Hz), 1.80 (s, 3H).

The sulfonyl chloride was dissolved in 28% N~OH (240 mL) and the resulting

solution was heated at reflux for 3 h. After cooling to 0 oC, the solution was acidified to

130

pH 1 by addition of 3 M H2S04 (ca. 200 mL) and extracted with ethyl acetate (3 x 500

mL). The combined organic extracts were dried over Na2S04, filtered and concentrated in

vacuo to afford the sulfonamide (29.9 g, 71 %) as a white solid. The N-acetyl sulfonamide

could not be purified to homogeneity by either chromatography or recrystallization. IH

NMR «CD3)2S0) Ô 8.29-8.24 (d, IH, J= 8.5 Hz), 7.77 (d, 2H, J= 3.9 Hz), 7.45 (d, 2H, J

= 6.9 Hz), 7.33 (s, 2H), 4.53-4.41 (m, IH), 3.20-3.09 (dd, IH, J= 14.2 and 6.8 Hz), 3.01-

2.87 (dd, IH, J= 11.2 and 10.1 Hz), 1.80 (s, 3H).

A suspension of the sulfonamide (20.0 g, 69.9 mmol, 1 eq) in distilled water (300

mL) was adjusted to pH 5.00 with LiOH (900 mg). A 0.25 M solution of Na2HP04 (85

mL) was used to raise the pH to 7.50. Acylase 1 from hog kidney (200 mg, 17.8 U/mg,

3560 U) was added as an aqueous solution (12 mL) and the resulting solution was stirred

at 21°C for 70 h. The solution was then acidified to pH 1.0 with 3M H2S04 and extracted

with ethyl acetate (3 x 500 mL), the organic layer was then dried with anhydrous sodium

sulphate and concentrated in vacuo to afford 2.28 g (11 %) of the sulfonamide starting

material. The aqueous layer was the neutralized with 2 M NaOH and concentrated. The

solution was then applied to an Amberlite 120(plus) acidic ion exchange column. The

column was rinsed with water until the eluent was at pH 6.0 and the it was rinsed with

0.50M NH40H solution until the eluent became basic. The basic wash was concentrated

in vacuo and recrystallized from water to afford provided 4'-sulfonamidophenylalanine as

a white solid (11.60 g, 68%). IH NMR (D20 / DCI) Ô 7.62 (d, 2H,J= 8.1 Hz), 7.26 (d,

2H, J= 8.1 Hz), 4.14 (t, IH, J= 6.8 Hz), 3.19-3.12 (dd, IH, J= 14.6 and 5.7 Hz),3.08-

3.01 (dd, IH, J = 14.4 and 6.9 Hz). l3e NMR (D20 / DCI)

131

8170.73,140.451,139.49,130.38,126.55,53.49,35.19. FABMS in satNaCI

m/z 267 (M+ Na, C9H12N204SNa requires 267).

N-Ethoxycarbonyl-4'-sulfonamidophenylalanine (2): Ethyl chloroformate (398 ilL,

4.17 mmol, 1.10 eq) was added to a two phase mixture of 4'-sulfonamidophenylalanine

(925 mg, 3.73 mmol, 1 eq) in 1,4-dioxane (25 mL) and sat. NaHC03 solution (25 mL) at

o oC and the resulting solution was stirred for 6 h at 0 oC. The mixture was extracted with

ethyl acetate (100 mL), the aqueous layer was acidified to pH 1 by addition of2 M HCI

(ca. 20 mL) and then extracted with ethyl acetate (3 x 50 mL). Latter organic extracts

were combined, dried over Na2S04, filtered and concentrated in vacuo to afford the ethyl

carbamate (879 mg, 83%) as an analytically pure oil. IH NMR «CD3)2CO) 8 7.85 (d,

2H, J= 7.1 Hz), 7.51 (d, 2H, J= 6.9 Hz), 6.54 (s, 2H), 6.45 (d, 1H, J= 6.7 Hz), 4.62-4.45

(m, 1H), 4.01-3.97 (q, 2H, J= 2.4 Hz), 3.41-3.28 (dd, 1H, J= 11.3 and 4.0 Hz), 3.16-3.05

(dd, 1H, J = 10.4 and 7.8 Hz), 1.13 (t, 3H, J = 6.0 Hz). HR-CIMS (mlz): [MH+]

calculated for C12H17N206S, 317.0807; found, 317.0817.

Et02C-(4'-S02NH2)Phe-Gly-O-t-butyl (3b): EDC-HCI (136 mg, 0.711 mmol, 1.10 eq),

HOBT (87.3 mg, 0.646 mmol, 1.00 eq) and triethylamine (269 mL, 1.94 mmol, 3.00 eq)

were added to a solution of 2 (204 mg, 0.646 mmol, 1 eq) in THF (3 mL) at 0 oC.

Glycine t-butyl ester-HCl (119 mg, 0.711 mmol, 1.10 eq) was added and the resulting

solution was allowed to warm to 21°C while stirred for 13 h, at which point the bulk of

the THF was removed by concentration in vacuo. The residue was dissolved in ethyl

acetate (45 mL) and extracted with 0.1 M HCI (3 x 25 mL) and sat. NaHC03 solution (3

132

x 25 mL). The organic layer was dried over Na2S04, filtered and concentrated in vacuo.

The solid residue was purified by mixed solvent recrystallization (ethyl acetate/hexanes)

to afford 193 mg (70%) of 3b. lH NMR «CD3)2CO) ô 7.82 (d, 2H, J= 7.5 Hz), 7.63 (s,

1H), 7.50 (d, 2H, J= 7.5 Hz), 6.52 (s, 2H), 6.40 (d, 1H, J= 7.5 Hz), 4.50 (m, 1H), 4.00-

3.88 (m, 4H), 3.40 (dd, 1H, J= 14.1 and 4.2 Hz), 3.02 (dd, 1H, J= 13.5 and 9.9 Hz), 1.45

(s, 9H), 1.12 (t, 3H, J= 6.9 Hz). BC NMR «CD3)2CO) ô 171.48, 168.93, 156.33, 142.70,

130.06, 126.18, 81.04, 60.48, 55.92,41.79, 37.83, 27.52, 14.22. Analysis calculated for

ClsH27N307S C, 50.34; H, 6.34; N, 9.78. Found: C, 50.33; H, 6.35; N, 9.73.

Et02C-(4'-S02NH2)Phe-Gly-OH (4b): TFA (7 mL) was added to a solution of3b (175

mg, 0.409 mmol, 1 eq) in CH2Ch (8 mL) and the solution was stirred for 25 min at 21°C

under an atmosphere of argon. The solvents were removed in vacuo, and the residue was

purified by recrystallization from acetone to afford 121 mg (79 %) of 4b. lH NMR

(CD30D) ô 8.55 (s, 1H), 7.83 (d, 2H, J= 7.2 Hz), 7.46 (d, 2H, J= 7.2 Hz), 4.45-4.42 (m,

1H), 4.02-3.98 (q, 2H, J=6.8), 3.95-3.92 (m, 1H), 3.32-3.25 (m, 2H), 2.97-2.89 (dd, 1H,

J= 13.5 and 9.9 Hz), 1.18-1.14 (t, 3H, J= 6.8). l3C NMR (CD3CD) ô 173.0, 171.8,

157.3, 148.7, 142.4, 137.6, 129.8, 126.0,60.9,56.0,37.7, 13.7. HR-CIMS (mlz): [MH+]

calculated for Cl4H20N307S, 374.1022; found, 374.1030.

Et02C-(4'-S02NH2)Phe-Phe-O-t-butyl (3a): Prepared as for 3b to afford to afford 173

mg (72%). lH NMR (CDCh) ô 7.80 (d, 2H, J= 7.8 Hz), 7.33-7.24 (m, 5H), 7.09 (d, 2H, J

= 6.0 Hz), 6.41 (d, 1H, J= 5.7 Hz), 5.19 (d, 1H, J= 6.1 Hz), 4.96 (s, 2H), 4.75-4.61 (m,

1H), 4.52-4.38 (m, 1H), 4.10-4.03 (q, 2H, J= 6.9 Hz), 3.20-2.96 (m, 4H), 1.39 (s, 9H),

133

1.21 (t, 3H, J= 6.8 Hz). l3C NMR (CDCh) Ô 170.34,170.11,142.21,140.84,136.03,

130.38, 129.65, 128.67, 127.31, 126.96, 82.93, 77.44, 61.76, 53.86, 38.40, 38.21,28.16,

14.69. Analysis calculated for C25H33N307S C, 57.79; H, 6.40; N, 8.09. Found: C, 57.71;

H, 6.34; N, 7.96.

Et02C-(4'-S02NH2)Phe-Leu-O-t-butyl (3e): Prepared as for 3 b to afford 227 mg

(78%). IH NMR «CD3)2CO) Ô 7.81 (d, 2H, J= 8.4 Hz), 7.55 (m, 2H), 7.48 (d, 2H, J=

8.1 Hz), 6.51 (m, 1H), 6.31 (m, 1H), 4.50 (m, 1H), 4.39 (m, 1H), 4.00-3.95, (q, 2H, J=

5.7), 3.32-3.26 (dd, 1H, J= 13.8 and 3.9 Hz), 3.05-2.97 (dd, 1H, J= 13.6 and 9.6 Hz),

1.75-1.64 (m, 2H), 1.61-1.56 (m, 2H), 1.45 (s, 9H), 1.12 (t, 3H, J= 7.4 Hz), 0.94-0.90

(m, 7H). l3C NMR «CD3)zCO) Ô 171.8, 171.0, 156.4, 142.5, 130.1, 126.2, 81.0, 60.5,

55.7, 51.6, 41.1, 37.8, 27.5, 24.8, 22.5, 21.3, 19.5, 14.2. Analysis calculated for

C22H35N307S C, 54.42; H, 7.26; N, 8.65. Found: C, 54.28; H, 1.27; N, 8.46.

Et02C-(4'-S02NH2)Phe-Pro-O-t-butyl (3d): Prepared as for 3 b to afford 312 mg

(67%). IH NMR «CD3)2S0) Ô 7.69 (d, 2H, J = 7.6 Hz), 7.48 (d, 2H, J = 7.0 Hz),7.36 (d,

2H, J = 7.2 Hz), 7.28 (s, 2H), 4.36 (m 1H), 4.19 (m, 1H), 3.85 (t, 2H, J = 6.7 Hz), 3,65

(m, 2H), 2.97-2.94 (dd 1H, J= 11.7 and 4.2 Hz), 2.82-2.74 (dd 1H, J= 11.1 and 9.2 Hz),

2.14 (m, IH), 1.90 (m, 2H), 1.77 (m, IH), 1.35 (s, 9H), 1.04 (t, 3H, J= 7.3 Hz). l3C NMR

«CD3)2S0)Ô 171.6, 170.4, 156.8, 143.0, 142.7, 130.5, 126.1,81.0,60.5,60.1,54.7,47.1,

36.4, 29.2, 28.3, 25.3, 15.2. Analysis calculated for ~lh7N307S C, 50.34; H, 6.34; N,

9.78. Found: C, 50.33; H, 6.35; N, 9.73.

134

Et02C-(4'-S02NH2)Phe-Phe-OH (4a): Prepared as for 4b to afford 270 mg (68 %). lH

NMR (CD30D) ô 8.19 (d, 1H, J= 9.3 Hz), 7.80 (d, 2H, J= 8.4 Hz), 7.38 (d, 2H, J= 9.3

Hz), 7.27-7.21 (m, 5H), 7.07 (d, 1H, J= 8.7 Hz), 4.68-4.63 (m, 1H), 4.40-4.35 (q, 2H, J

=6.1 Hz), 3.24-3.18 (dd, 1H, J=13.9 and 5.2 Hz), 3.18-3.11 (dd, 1H, J=14.3 and 5.5 Hz),

3.04-2.97 (dd, 1H, J= 13.9 and 8.2 Hz), 2.88-2.80 (dd, 1H, J= 13.9 and 9.7 Hz), 1.18-

1.14 (t, 3H, J= 6.9 Hz). l3C NMR (CD3CD) ô 173.1, 172.3, 157.2, 142.3, 137.0, 129.8,

129.2, 128.3, 126.6, 126.0, 60.9, 55.9, 53.9, 37.6, 37.2, 13.7. HR-CIMS (m/z): [MH+]

calculated for C2lH26N307S, 464.1491; found, 464.1501.

Et02C-(4'-S02NH2)Phe-Leu-OH (4c): Prepared as for 4b to afford 174 mg (80 %). lH

NMR (CD3CD) ô 7.82 (d, 2H, J= 8.4 Hz), 7.68 (d, 1H, J= 8.1 Hz), 7.45 (d, 2H, J= 8.1

Hz), 4.47-4.42 (m, 2H), 4.01-3.95 (q, 2H, J= 6.4 Hz), 3.31-3.29 (m, 1H), 3.25-3.3.19

(dd, 1H, J = 13.9 and 4.9 Hz), 2.95-2.87 (dd, 1H, J = 13.9 and 9.7 Hz), 1.71-1.62 (m,

2H), 1.18-1.13 (t, 3H, J= 7.1 Hz), 0.97-0.91 (m, 6H). l3C NMR (CD3CD) ô 174.6,

172.7, 157.2, 142.3, 129.8, 126.7, 126.0, 60.9, 55.8, 50.9, 40.4, 37.7, 24.8, 22.2, 20.6,

13.7. HR-CIMS (m/z): [MH+] calculated for ClsH2SN307S, 430.1648; found, 430.1654.

Et02C-(4'-S02NH2)Phe-Pro-OH (4d): Prepared as for 4b except recrystallized from

iso-propanol / hexanes to afford 94.5 mg (55 %). lH NMR (CD30D) ô 7.83 (d, 2H, J =

8.1 Hz), 7.50 (d, 2H, J= 8.4 Hz), 7.25 (d, 1H, J= 3.9 Hz), 4.66-4.61 (dd, 1H, J= 8.8 and

5.5 Hz), 4.47-4.43 (dd, 1H, J= 8.4 and 3.9 Hz), 4.02-3.95 (q, 2H, J= 7.2 Hz), 3.80-3.75

(m, 1H), 3.56-3.51 (m, 1H), 3.32-3.29 (m, 1H), 3.21-3.14 (dd, 1H, J= 13.9 and 5.2 Hz),

135

2.96-2.89 (dd, IH, J= 13.8 and 8.7 Hz), 2.27-2.21 (m, IH), 2.05-1.96 (m, 2H), 1.19-1.16

(t, 3H, J=7.2 Hz. BC NMR (CD3CD) ô 174.0, 171.1, 157.3, 141.9, 130.1, 129.1, 128.2,

126.0, 60.8, 59.4, 53.9, 37.0, 29.0, 24.6, 13.7. HR-CIMS (mlz): [MH+] calculated for

C17H24N307S, 414.1335; found, 414.1325.

Measurement of Inhibition Constants: Kinetic constants for carbonic anhydrase (CA)

were measured according to Pocker and Stone using p-nitrophenyl acetate (PNPA) as the

substrateYl The CA-catalyzed hydrolysis of pNPA was followed spectrophotometrically

at 25 OC in a 96-well microplate spectrophotometer by monitoring the appearance of p

nitrophenolate at 404 nm. The values of Km and V max were determined by measuring

the hydrolysis rate as a function of the pNP A concentration. To determine the inhibition

constants, the values of Km and V max were re-determined in the presence of varying

amounts of inhibitor. Since the values of Km for pNP A increased in the presence of the

inhibitor, but the values of V max remained unchanged, we concluded that the inhibition

is competitive. The concentration of inhibitor that increased the Km for pNP A by a factor

oftwo is the inhibition constant. A typical procedure was to add CA solution (100.0 mL)

with inhibitor to acetonitrile solution of pNPA (5.0 mL). In the assay solution, the

concentration of inhibitor ranges from 0.0 to 6.0 J.lM, while the concentration of pNPA

ranged from 0.2 to 2.5 mM. The microplate was shaken for 5 s before the first reading

and for 3 s between readings.

Selective Concentration of EtOC-Phesa-Phe (4a) Over EtOC-Phe-Phe, (5) Into a

Compartment Containing Carbonic Anhydrase: A solution of 4a (2.9 mg, 6.3 mmol)

136

and 5 (2.9 mg, 7.5 mmol) in 0.01 M aqueous phosphate buffer (pH 7.5, 40 mL) was

divided into two equal portions. Carbonic anhydrase (0.20 g, approx 6.7 mmol) was

dissolved in the first portion and the resulting solution (20.0 mL) was transferred to a

dialysis bag (12,000 MW cutoff, Sigma D-0655). This dialysis bag was suspended in the

second portion and the reaction vessel was shaken gently (200 rpm) at 30 oC. Aliquots

were removed periodically from each compartment, heated to 80 oC until a white

precipitate formed (~5 min), centrifuged and the supematant filtered through a 0.22 /lm

pore filter. The amount of dipeptides was measured by HPLC using a Zorbax C8 column

and 40/60/0.1 water/methanol/trifluoroacetic acid at 0.40 mL/min. After 12 hours 88% of

4a (retention time Il.4 min) had accumulated inside the dialysis bag while only 42% of 5

(retenti on time 25.5 min) was found inside the bag.

Selective Concentration of Etoc-Phesa-Phe (4a) from a Mixture of Etoc-Phesa-Leu

(4c), Etoc-Phesa-Gly (4b), and Etoc-Phe-Phe (5) by Carbonic Anhydrase: Dipeptides

4a (2.0 mg, 4.3 /lmol), 4b (1.6 mg, 4.3 /lmol), 4c (1.9 mg, 4. /lmol) 4d (1.8 mg, 4.3

/lmol) and 5 (1.7 mg, 4.3 /lmol) were dissolved in 40 mL of 10 mM KH2P04 buffer, pH

7.5 containing 0.1 mg/mL penicillin G (to avoid bacterial growth). Carbonic anhydrase

(CA) (0.29 g, 9.7 /lmol, 0.45 eq) was dissolved in 20 mL ofthis solution and placed in a

dialysis bag (the bag was washed in ddH20 for 1 hr, rinsed in EtOH once and then

washed again with ddH20). The bag was suspended in the remaining 20 mL of inhibitor

solution in a lOO-mL container and shaken at 60 rpm on a 3 dimensional orbital shaker at

room temperature for 49 hrs. Samples (1 mL) were taken periodically from inside and

outside the dialysis bag, heated in an 80°C water bath for 5 min and then centrifuged for

137

10 min. The supernatant was filtered through a 0.22 I-tM sterile filter. The supernatant

(700 I-tL) was added to MeOH (300 I-tL) to form the HPLC sample (30 % MeOH, 70%

aqueous). The sample was run on a Phenomenex C8 reverse phase column under the

following conditions: 0-15 min 30% MeOH, 70% H20, 15-60 min 37% MeOH 63%

H20, 60-90 min 62% MeOH, 38% H20. The peak are as were monitored: PhesaGly: 7.9

min, PhesaPro: 17.6 min, PhesaLeu: 54.5 min, PhesaPhe: 60.0 min, PhePhe: 69.5 min. The

percentages are accurate to +/- 2%. AlI non-sterile apparatus used was autoclaved prior to

use to avoid bacterial growth.

Screening of Proteases for the Hydrolysis of Etoc-Phe-Phe Dipeptide (4a): The

protease to be screened (0.1 mg) was added to a solution of 4a (1.0 mg, 2.2 Ilmol) in 0.01

M aqueous phosphate buffer (pH 7.5). The solution was kept at 30 oC and aliquots were

removed periodically worked up as above, and analyzed by HPLC.

Selective Protection of Inhibitors from Hydrolysis by Carbonic Anhydrase: A

solution of 4a (3.0 mg, 6.5 mmol) and 5 (2.8 mg, 7.3 mmol) in 0.01 M aqueous

phosphate buffer (pH 7.5, 40 mL) was divided into two equal portions. Carbonic

anhydrase (0.20 g, approx 6.7 mmol) was dissolved in the first portion and the resulting

solution (20.0 mL) was transferred to a dialysis bag (12,000 MW cutoff, Sigma D-0655).

Pronase from Streptomyces griseus (Sigma P-5147, 4 mg) was dissolved in the second

portion and the dialysis bag was then suspended in the resulting solution. The reaction

vessel was then shaken gently (200 rpm) at 30 oC and aliquots were removed periodically

from each compartment, worked up as above, and analyzed by HPLC. After 30 min,

138

neither substrate was detectable in the solution outside the dialysis bag. Inside the dialysis

bag, 78% of 5 had hydrolyzed while only 6% of 4a had hydrolyzed after 6 h. In a control

experiment containing no carbonic anhydrase, inside the dialysis bag, 76% of 4a and

80% of 5 had hydrolyzed after 6 h.

Selective Binding of Etoc-Phesa-Phe (4a) over Etoc-Phesa-Leu (4c): A solution of 4a

(3.3 mg 7.1 /lmol) and 4c (3.6 mg, 8.4 /lmol) in 0.01 M aqueous phosphate buffer (pH

7.5,40 mL) was divided into two equal portions. Carbonic anhydrase (0.20 g, approx 6.7

mmol) was dissolved in the first portion and the resulting solution (20.0 mL) was

transferred to a dialysis bag (12,000 MW cutoff, Sigma D-0655). This dialysis bag was

suspended in the second portion and reaction vessel was shaken gently (200 rpm) at

30°C. Aliquots were removed periodically from each compartment, worked up as above,

and analyzed by HPLC using a Zorbax C8 column. After 12 hours 98% of 4a had

accumulated inside the dialysis bag while only 60% of 4c was found inside the bag.

Hydrolysis of Etoc-Phesa-Gly (4b) and Etoc-Phesa-Phe (4a) in the Presence of

Carbonic Anhydrase: PhesaPhe 4a (2.0 mg, 4.3 /lmol) and PhesaGly 4b (1.6 mg, 4.3

/lmol), were dissolved in 20 mL of 10 mM KH2P04 buffer, pH 7.5. Carbonic anhydrase

(CA) (0.4090 g, 13.6 /lmol, 1.60 eq) was dissolved in this solution and placed in a

dialysis bag (the bag was washed in ddH20 for 1 h, rinsed in EtOH once and then washed

again with ddH20). The bag was suspended in 20 mL of the phosphate buffer containing

Pro nase from Streptomycese griseus (5.0 mg, 0.01 eq) in a 150 mL beaker and shaken at

150 rpm at 30 oC for 313 hrs. Samples (1 mL) were taken periodically from inside

139

worked up as above and analyzed by HPLC. After 193 h, only 71% of 4a had hydrolyzed

while 93% of 4b had hydrolyzed.

Hydrolysis of Etoc-Phesa-Leu (4c) and Etoc-Phesa-Phe (4a) in the Absence of

Carbonic Anhydrase: A solution of 4a (2.9 mg 6.3 ~mol) and 4c (2.4 mg, 5.6 ~mol) in

0.01 M aqueous phosphate buffer (pH 7.5, 40 mL) was divided into two equal portions.

The first portion was transferred to a dialysis bag (12,000 MW cutoff, Sigma D-0655).

Pronase from Streptomyces griseus (Sigma P-5147, 4 mg) was dissolved in the second

portion and the dialysis bag was then suspended in the resulting solution. The reaction

vessel was then shaken gently (200 rpm) at 30 oc and aliquots were removed periodically

from each compartment, worked up as above, and analyzed by HPLC using a Zorbax C8

column. After 30 min, neither substrate was detectable in the solution outside the dialysis

bag. After 8 h, 86% of 4a and 88% of 4c inside the dialysis bag had hydrolyzed.

Hydrolysis of Etoc-Phesa-Leu (4c) and Etoc-Phesa-Phe (4a) in the Presence of

Carbonic Anhydrase: A solution of 4a (2.9 mg, 6.3 ~mol) and 4c (2.4 mg, 5.6 ~mol) in

0.01 M aqueous phosphate buffer (pH 7.5, 40 mL) was divided into two equal portions.

Carbonic anhydrase (0.14 g, approx 4.7 mmol) was dissolved in the first portion and the

resulting solution (20.0 mL) was transferred to a dialysis bag (12,000 MW cutoff, Sigma

D-0655). Pronase from Streptomyces griseus (Sigma P-5147, 4 mg) was dissolved in the

second portion and the dialysis bag was then suspended in the resulting solution. The

reaction vessel was then shaken gently (200 rpm) at 30 oC and aliquots were removed

140

periodically from each compartment, worked up as above, and analyzed by HPLC. After

6 h, 93% of 4c had hydrolyzed while only 58% of 4a was hydrolyzed.

Hydrolysis of Etoc-Phesa-Phe (4a), Etoc-Phesa-Gly (4b) Etoc-Phesa-Leu (4c), Etoc

Phesa-Pro (4d) and Etoc-Phe-Phe (5), in the Presence of Carbonic Anhydrase:

PhesaPhe 4a (2.0 mg, 4.3 !-lmol), PhesaGly 4b (1.6 mg, 4.3 !-lmol), PhesaLeu 4c (1.9 mg, 4.

!-lmol), PhesaPro 4d (1.8 mg, 4.3 !-lmol) and PhePhe 5 (1.7 mg, 4. !-lmol) were dissolved in

20 mL of 10 mM KH2P04 buffer, pH 7.5. Carbonic anhydrase (CA) (0.7670 g, 25.6

!-lmol, 1.20 eq) was dissolved in this solution and placed in a dialysis bag (the bag was

washed in ddH20 for 1 h, rinsed in EtOH once and then washed again with ddH20). The

bag was suspended in 20 mL of the phosphate buffer containing Pronase from

Streptomycese griseus (4.9 mg, 0.01 eq) in a 150 mL beaker and shaken at 150 rpm at 30

oC for 193 hrs. Samples (l mL) were taken, worked up as above, and analyzed by HPLC.


Resin Preparation[21: Amino-Tentagel resin (2.0g, 0.88 mmol, 1.0 eq, Novabiochem)

was swelled in distilled THF (30 mL) in a 50-mL coarse-fritted peptide synthesis vessel

(Chemglass). 2,3,5,6-Tetrafluoro-4-hydroxy-benzoic acid (925mg, 4.40 mmol, 5.0 eq,

Aldrich) was added to the resin suspension and the reaction vessel was shaken lightly for

5 min. Pyridine (1.52 mL, 17.6 mmol, 20 eq) was added followed by

diisopropylcarbodiimide, (690 I!L, 4.40 mmol, 5 eq). The resulting mixture containing

the resin suspended in a milky white solution was gently rocked for 16 h. During this

141

time the solution becomes clear. The resin was washed with THF (2x50 mL), DMF (2x50

mL), CH2Ch (50 mL), DMF (50 mL),and then THF (50 mL). The resin was then

suspended in THF «30 mL) and again treated with 2,3,5,6-tetrafluoro-4-hydroxy-benzoic

acid, pyridine and DIC in the manner described above. After the final wash, the resin was

swelled in THF (30 mL), tert-butylamine (1.85 mL, 17.6 mmol, 20.0 eq) was added and

the mixture was gently rocked for 16 h. The resin was subsequently washed with THF

(4x50 mL) and was ready for use.

Resin Activation and Quantification: The batch of resin prepared above was swelled in

distilled THF (20 mL). A 0.50 M solution of EtOC-Phe-OH (10.2 mL, 8.80mmol, 10 eq)

in THF was added, followed by diisopropy1carbodiimide (1.38 mL, 8.80 mmol, 10.0 eq)

and the resulting mixture was gently rocked for 16 h. The resin was washed THF (2x50

mL) and DMF (2x50 mL), and then rocked for 2 h in THF (30 mL). The resin was

washed with THF (2x50 mL), DMF (2x50 mL), CH2Ch (1x50 mL), and then THF (2x50

mL).

To quantify the amount of active ester on the resin, it was suspended in THF (30

mL), tert-butylamine (1.85 mL, 17.6 mmol, 20.0 eq) was added, and the resulting mixture

was gently rocked for 16 h. The resin was then washed with THF (3x50 mL). The filtrate

was concentrated in vacuo and the residue was taken up in ethyl acetate and washed with

O.lM HCI (3x30 mL), water (30 mL) and saturated NaHC03 solution (3x 30ml). The

organic layer was dried over anhydrous Na2S04 and the solvent removed in vacuo to

afford 212mg (0.72 mmol) of EtOC-Phe-N-t-Bu as a foamy off-white solid, indicating an

142

82% loading of active ester relative to the initial number of amine termini on the

T enta Gel resin.

General Procedure for Testing Pronase Activity Towards Dipeptide Hydrolysis: A

series ofPronase solutions containing 0.1 mg (~5 nmol), 10 mg (~0.5 /lmol), 100 mg (~5

/lmol) and 0.5 g (~20 /lmol) are prepared in either 30 mM NaH2P04 buffer containing 0.1

mg/mL PEN Na, pH 7.5, or 30 mM BI CINE buffer containing 0.1 mg/mL PEN Na, pH

9.0 in steri1ized glass vials. Approximately 10 /lmol of the dipeptide is added to each vial

containing Pronase, and to one vial with only buffer (control). The vials are then shaken

gently for several days, with samples taken during the first four hours, and then semi

daily. Each sample is heated to 80°C for ~5 min., centrifuged for 10 min and filtered

through a 0.22 um sterile filter to form the HPLC sample. Samples are then run through a

Phenomenex C8 (2) column monitoring at 220 nm under various separation conditions.

Eight Membered p-DCL: Gly, Pro, Leu, Phe as Nucleophiles, EtocPhe and

EtocPhesa as Electrophiles on Solid Support: A stock amino acid solution was prepared

in 500 mL ddH20 by adding Phe (0.413 g, 2.50 nunol), Leu (0.328 g, 2.50 mmol), Pro

(0.288 g, 2.50 nunol), Gly (0.188 g, 2.50 nunol), BI CINE (2.45 g, 30 mM) and PEN-Na

(0.05 g, 0.1 g/mL). Each amino acid was 5 mM. This stock was adjusted to pH 9.0 and

used to dissolve/suspend aIl components of the system. Tentagel resin with activated

ester, either 0.45 g or 0.9 g, (~0.80 E-4 or 1.6E-4 respectively) mol active ester, (~1.5 or

3 eq EtocPhesa (2), 1.5 or 3 eq EtocPhe wrt CA) was dissolved in 10 mL AA stock (this

gave 1.25 eq of AA's wrt resin in the chamber to begin). This suspension was placed in a

143

washed dialysis bag (MWCO 12 kDa) and agitated manually for ~10 sec to promote resin

swelling. It was then placed in a 100 mL container. Pronase (0.5 g) was dissolved in 20

mL AA stock, injected into a dialysis bag (MWCO 1 kDa) and placed beside the resin

bag in the container. CA (0.84 g, 2.8E-5 mol, 1 eq) was dissolved in 20 mL AA stock and

poured over the two bags in the container forming the immersion solution. The eq of each

AA in the whole system wrt CA was 9.75. The system was set to stirring on a TW3

orbital shaker (Rose Scientific) for 48 hrs. The entire synthe sis chamber was replaced

(new resin) after either 12 or 24 hrs. In sorne cases the entire Pronase chamber was

replaced (new Pronase in new AA stock) after 24 hrs. 200 !AL samples were taken every 2

hrs, heated to 80°C for ~ 5 min., centrifuged for 10 min and filtered through a 0.22 um

sterile filter to form the HPLC sample. Samples were mn through a Phenomenex C8 (2)

column monitoring at 220 nm under the following conditions: 0-30 min gradient 100%

H20/O.l%TFA, 0% MeOH to 70% H20/O.1%TFA, 30% MeOH; 30-70 min 63%

H20/O.l% TFA, 37% MeOH; 70-90 min, gradient 63% H20/0.l%TFA, 37% Me OH to

38% H20/O.1 %TFA, 62% MeOH. The main peaks monitored were EtocPhesaGly (4b) at

25.2 min, EtocPhesaLeu (4c)at 57.4 min and EtocPhesaPhe (4a)at 58.6 min.

Variations:

1) In experiment 3.7 Pronase chamber was not added to the system for 6 hrs after each

synthetic cycle was begun.

2) Synthetic cycle times were either 12 hrs (3.8) or 24 hrs (3.6 and 3.7).

3) In (3.7) the Pronase chamber was removed after 24 hrs and replaced after 30 hrs.

4) In (3.6 and 3.7) 0.9 g resin portions were used. In (3.8) 0.45 g resin portions were

used

144

5) AlI experiments had 4 cycles each.


General Experimental Procedures: AU chemicals and enzymes were purchased from

Sigma-Aldrich Canada with the foUowing exceptions. Acylase 1 from hog kidney was

obtained from Fluka. Amino and bromo NovaSyn Tentagel resins were obtained from

Caledon-NovaBiochem. AU chemicals were used without further purification with the

foUowing exceptions. Tetrahydrofuran was distilled from sodium benzophenone ketyl.

Methylene chloride and triethylamine were distilled from calcium hydride. AU amino

acids used were natural L-enantiomers. AU solid phase syntheses, with the exception of

the library experiments, were carried out in coarse-fritted peptide synthesis vessels

obtained from Chemglass. NMR spectra were recorded at 270, 300, or 400 MHz for IH

and 67.5, 75, and 100 MHz for BC. Elemental analyses were performed by Quantatative

Technologies Inc, Whitehouse, NJ, USA. High-resolution mass spectra were performed

by Université de Sherbrooke, Sherbrooke, QC, Canada.

In aU cases where yields were determined by HPLC, molar absorptivity values

were determined with solutions of known concentration.

Procedures for Preparation of Authentic Standards

Et02C-Gly-(4'-S02NH2)Phe-OH (7h). 2 M NaOH (702 ~L, 1.40 mmol, 4.0 eq) was

added to a solution of Et02C-Gly-(4'-S02NH2)Phe-OMe (136 mg, 0.351 mmol, 1.0 eq)

in methanol (8.0 mL) and aUowed to stir at 21°C for 12 h. 2 M HCI (702 ~L, 1.40 mmol,

145

4.0 eq) was added and the methanol was removed in vacuo. The aqueous solution was

then acidified to pH 1 with 2 M HCI (1.0 mL) and then extracted with ethyl acetate (3 x

30 mL). The organic layer was dried over Na2S04, filtered and concentrated in vacuo to

afford 129 mg of3b (99%) as a white solid. IH NMR «CD3)2S0) Ô 7.93 (d, IH, J= 8.0

Hz), 7.71 (d, 2H, J= 8.4 Hz), 7.36 (d, 2H, J= 8.0 Hz), 7.15 (s, 2H), 7.0 (s (broad), IH),

4.52-4.46 (m, IH), 3.98 (q, 2H, J = 6.4 Hz), 3.63-3.51 (m, 2H), 3.15-3.10 (m, IH), 3.01-

2.96 (dd, IH, J = 13.8 and 8.6 Hz), 1.17-1.13 (t, 3H, J = 7.0). 13C NMR «CD3)2S0) Ô

173.1,169.8,157.2,143.0,142.3,130.3,126.2,60.7, 53.9,43.9,37.2,15.5. HR-CIMS

(m/z): [MH+] calculated for C14H19N307S, 373.0944; found, 373.0949.

Et02C-Phe-(4'-S02NH2)Phe-OH (7a). Prepared using the general procedure to afford

162 mg (84%). IH NMR «CD3)2S0) Ô 8.32 (d, IH,J= 7.6 Hz), 7.70 (d, 2H,J= 8.0 Hz),

7.40 (d, 2H, J = 8.0 Hz), 7.30 (s, 2H), 7.27-7.16 (m, 5H), 7.15-7.14 (m, IH), 4.48-4.45

(m, IH), 4.23-4.19 (m, IH), 3.87-3.84 (m, 2H), 3.16-3.11 (dd, IH, J= 13.6 and 5.2 Hz),

3.03-2.97 (dd, IH, J= 13.4 and 9.0 Hz), 2.94-2.91 (m, IH), 2.69-2.63 (m, IH), 1.08-1.05

(t, 3H, J = 6.8 Hz). 13C NMR «CD3)2S0) Ô 173.1, 172.5, 156.6, 142.9, 142.2, 138.7,

130.4, 129.8, 128.7, 126.9, 126.2, 60.6, 56.6, 53.9, 38.0, 37.1, 15.4. HR-CIMS (mlz):

[MH+] calculated for C21H26N307S, 464.1491; found, 464.1483.

Et02C-Leu-(4'-S02NH2)Phe-OH (7c). Prepared following the general procedure to

afford 727 mg (58%). IH NMR «CD3)2S0) Ô 8.10 (d, IH, J= 8.1 Hz), 7.68 (d, 2H, J=

8.4 Hz), 7.37 (d, 2H, J= 7.8 Hz), 7.29 (s, 2H), 7.16 (d, IH, J= 8.9 Hz), 4.18-4.41 (m,

IH), 3.99-3.92 (m, 3H), 3.14-3.09 (m, IH), 3.01-2.94 (m, IH), 1.62-1.52 (m, IH), 1.40-

146

1.32 (m, 2H), 1.16-1.12 (t, 3H, J= 6.9 Hz), 0.86-0.81 (m, 6H). l3C NMR ((CD3)2S0) Ô

. 173.1, 156.6, 142.9, 142.3, 130.3, 126.1, 60.6, 53.7, 37.0, 24.9, 23.9, 22.3, 15.5. HR

CIMS (m/z): [MH+] calculated for C18H28N307S, 430.1648; found, 430.1654.

Et02C-Pro-(4'-S02NH2)Phe-OH (7d): Prepared following the general procedure to

afford 278 mg (91%). IH NMR ((CD3)2S0) Ô 8.19-8.15 (m, 1H), 7.69 (d, 2H, J = 7.6

Hz), 7.41-7.38 (m, 2H), 7.30 (s, 2H), 4.50-4.40 (m, 1H), 4.18-4.06 (m, 1H), 4.01-3.95 (m,

1H), 3.83-3.79 (q, 2H, J= 6.0 Hz), 3.31-3.24 (m, 2H), 3.17-3.09 (m, 2H), 3.01-2.95 (m,

2H), 2.03-1.98 (m, 1H), 1.71-1.62 (m, 2H), 1.15-1.10 (t, 2H, J = 7.0 Hz). l3C NMR

((CD3)zSO) Ô 173.3, 173.1*, 172.7, 172.6*, 154.9*, 154.6, 142.9, 142.7*, 130.4*, 130.1,

126.0,61.2*,61.1,60.3*,60.0,53.9*,53.6,47.6,47.2*, 36.9, 31.7, 30.5*, 24.5*, 23.7,

15.5*, 15.3 (* indicates minor rotamer).HR-CIMS (m/z): [MH+] calculated for

C17H23N307S, 413.1257; found, 413.1266.

Measurement of Diffusion Rates Out of Dialysis Bags: Resorufin (random amounts

similar molecular weight to the dipeptide library members) was dissolved in X a mL of

ddH20 and placed in a dialysis bag of a chosen MWCO. This dialysis bag was suspended

in a 100 mL container with ya mL of ddH20. Samples were taken initially and after every

30 min from inside and outside the dialysis bag. Samples are observed for absorbance at

574 nm. A graph of increasing absorbance versus time in the outside chamber was then

made to estimate the time needed for diffusion.

a The volumes will vary depending on which system is being mimicked. Usually they are

both 20 mL.

147

Measurement of Inhibition Constants: The CA-catalyzed hydrolysis of pNPA at 25°C

was followed spectrophotometrically on a 96-well microplate spectrophotometer by

monitoring the appearance of p-nitrophenolate at 404 nm. A typical procedure was to add

CA solution (100.0 mL, 3-5 mg/ml) with inhibitor (0.0-6.0 /lM) to acetonitrile solution of

pNPA (5.0 IlL, 0.2-2.5 mM). The microplate was shaken for 5 s before the first reading

and for 3 s between readings.

General Procedure for Pseudo-Dynamic Combinatorial Libraries: A buffer solution

was prepared in 100 mL ddH20 with BICINE (0.490 g, 30 mM) and PEN-Na (10 mg, 0.1

g/mL). The buffer was adjusted to pH 9.0 and used to dissolve/suspend aIl components of

the system. Phesa (1) (66.2 mg, 0.27 mmol, 9.6 eq) and Phe (6) (44.6 mg, 0.27 mmol, 9.6

eq) were dissolved in 10 mL ofbuffer, which was then re-adjusted to pH 9.0. EtocGly,

EtocPro, EtocLeu and EtocPhe Tentagel active esters (each 125 mg, 22.5 /lmol, 0.80 eq)

were suspended in this solution. The resulting suspension was placed in a washed dialysis

bag (MWCO 12 kDa) and agitated manually for ~ 10 sec to promote resin swelling. It was

then placed in a 100 mL container. Pronase (0.5 g) was dissolved in 20 mL buffer, the

resulting solution was added to a dialysis bag (MWCO 1 kDa) to form the destruction

chamber, and was placed beside the other resin bag in the container. CA (0.84 g, 28

/lmol, 1 eq) was dissolved in 20 mL buffer and poured over the two bags in the container

forming the immersion solution (screening chamber). The system was set to stirring on an

orbital shaker. New resin was added directly to the synthe sis compartment after every 8,

12 or 16 h, depending upon the experiment. After two cycles, the solution inside the

148

synthesis chamber was filtered and the filtrate was used to suspend the next portion of

resin. 200 ~L samples were taken from the screening chamber every 4 hrs, heated to

80°C for ~5 min., centrifuged for 10 min and filtered through a 0.22 /lm sterile filter to

form the HPLC sample. Samples were run through a Phenomenex C8 (2) column

monitoring at 220 nm under the following conditions: 0-120 min gradient 80%

H20/O.1%TFA, 20% MeOH to 0% H20/0.1%TFA, 100% Me OH. The peaks monitored

were EtocGly Phesa (7b) at Il.3 min, EtocProPhesa (7 d) at 18.5 min, EtocLeuPhesa (7 c) at

39.8 min, EtocGlyPhe (8b) at 41.0 min, EtocPhePhesa (7a) at 43.5 min, EtocProPhe (8d)

at 52.7 min, EtocLeuPhe (8c) at 69.8 min, and EtocPhePhe (8a) at 70.6 min.


Quantification of CA Active Sites:[4] Four solutions of CA (0.3 g, 10 /lmol in each)

were prepared in 20 mL of 30 mM BICINE buffer containing 0.2 mg/mL PEN Na, pH

9.0.23 mg (50 /lmol) of EtocPhesaPhe (4a) was dissolved in one ofthese solutions for the

5.0 eq experiment. Similarly, 12 mg (30 /lmol) of 4a for the 2.5 eq, and 4.5 mg (10 /lmol)

of 4a for the 1.0 eq and blank experiments were prepared. Four 20 mL solutions of

Pronase (0.01 g in each) were prepared in the BICINE buffer, put inside dialysis bags

(MWCO 12 kDa) and suspended in the Ca/4a solutions inside a 150 mL container. The

solutions were agitated gently on a TW3 orbital shaker (Rose Scientific) for two days.

Samp1es were taken every hour for the first 10 hours, and once the next day. Each sample

was heated to 80°C for ~5 min., centrifuged for 10 min and filtered through a 0.22 /lm

sterile filter to form the HPLC sample. Samples were run through a Phenomenex C8 (2)

149


H20/O.1 % TF A, 20% MeOH to 0% H20/O.1 % TF A, 100% MeOH. 4d absorbance was

monitored at 47.5 min.

Static Library Syntheses With Varying Concentrations of Nucleophile: A buffer

solution was prepared in 100 mL ddH20 with BICINE (0.490 g, 30 mM) and PEN-Na

(10 mg, 0.1 g/mL). The buffer was adjusted to pH 9.0 and used to dissolve/suspend aIl

components of the system. Phesa (1 t was dissolved in 50 mL of buffer in the 1 mM, 3

mM, 4 mM and 5 mM experiments and in 25 mL of buffer in the 10 mM experiment.

EtocGly, EtocPro, EtocLeu and EtocPhe Tentagel active esters (a total of 250 mg, 0.45

mmol, 1 eq per experiment) were suspended in this solution inside a peptide synthesis

vessel. The reactions were set to stirring on a TW3 orbital shaker (Rose Scientific) at 60

rpm. 200 I-tL samples were taken from the vessels at various intervals (at least every half

hour for the first 8 hours), and were then centrifuged for 10 min and filtered through a

0.22 /lm sterile filter to form the HPLC sample. Samples were run through a Phenomenex

C8 (2) column monitoring at 220 nm under the following conditions: 0-120 min gradient

80% H20/0.1 % TF A, 20% MeOH to 0% H20/0.1 % TF A, 100% MeOH. The peaks

monitored were EtocGlyPhesa (7b) at Il.3 min, EtocProPhesa (7 d) at 18.5 min,

EtocLeuPhesa (7c) at 39.8 min EtocGlyPhe (8b) at 41.0 min, EtocPhePhesa (7a) at 43.5

min, EtocProPhe (8d) at 52.7 min, EtocLeuPhe (8c) at 69.8 min, and EtocPhePhe (8a) at

70.6 min.

150

a Table 6.1: Amounts of(l) used in static library experiments.

Equivalents of(l) with Experiment Mass of(l) (mg) Amount of (1) (f.lmol) respect to total activated

ImM 3mM 4mM 5mM 10mM

12 37 49 61 61

amino acids 50 1.1 150 3.3 200 4.4 250 5.6 250 5.6

Eight Membered 16 h p-DCL Experiments: A buffer solution was prepared in 100 mL

ddH20 with BICINE (0.490 g, 30 mM) and PEN-Na (10 mg, 0.1 g/mL). The buffer was

adjusted to pH 9.0 and used to dissolve/suspend aIl components of the system. Phesa (lt

and Phe (6)b were dissolved in 10 mL ofbuffer. EtocGly, EtocPro, EtocLeu and EtocPhe

Tentagel active estersa were suspended in this solution. The resulting suspension was

placed in a washed dialysis bag (MWCO 12 kDa) and agitated manually for ~10 sec to

promote resin swelling. It was then placed in a 100 mL container. Pronase (0.5 g) was

dissolved in 20 mL buffer, the resulting solution was added to a dialysis bag (MWCO 1

kDa) to form the destruction chamber, and was placed beside the other resin bag in the

container. CA (0.84 g, 28 /lmol, 1 eq) was dissolved in 20 mL buffer and poured over the

two bags in the container forming the immersion solution (screening chamber). The

system was set to stirring on a TW3 orbital shaker (Rose Scientific) at 60 rpm. New resin

was added directly to the synthe sis compartment after every 16 h. After every two cycles,

the solution inside the synthesis chamber was filtered and the filtrate was used to suspend

the next portion of resin. 200 /lL samples were taken from the screening chamber every 4

hrs, heated to 80°C for ~5 min., centrifuged for 10 min and filtered through a 0.22 /lm

151

sterile filter to form the HPLC sample. Samples were run through a Phenomenex C8 (2)


H20/O.1 % TF A, 20% MeOH to 0% H20/0.1 % TF A, 100% MeOH. The peaks monitored

were EtocGlyPhesa (7b) at 11.3 min, EtocProPhesa (7d) at 18.5 min, EtocLeuPhesa (7c) at

39.8 min EtocGlyPhe (Sb) at 41.0 min, EtocPhePhesa (7a) at 43.5 min, EtocProPhe (Sd)

at 52.7 min, EtocLeuPhe (Sc) at 69.8 min, and EtocPhePhe (Sa) at 70.6 min.

a Table 6.2: Variations of the eight membered, 16 hour library

Equivalents of Equivalents of activated Number of nucleophile 1 amino acid added each

cycles with respect to cycle with respect to CAb CA

6 9.6 3.2 6 4.5 3.2 12 4.5 3.2 16 4.5 9~

b In alllibraries Phe (6) was present in the same amounts as 1, but because 6 is a non-inhibitor, its presence did not affect the yield or selectivity.

Four Membered p-DCL: A buffer solution was prepared in 100 mL ddH20 with

BICINE (0.490 g, 30 mM) and PEN-Na (10 mg, 0.1 g/mL). The buffer was adjusted to

pH 9.0 and used to dissolve/suspend aIl components of the system. Phesa (1) (33.0 mg,

0.135 mmol, 4.8 eq) and Phe (6) (22.3 mg, 0.135 mmol, 4.8 eq) were dissolved in 10 mL

of buffer, which was then re-adjusted to pH 9.0. EtocPro EtocGly and EtocPhe Tentagel

active esters (each 125 mg, 22.5 ~mol, 0.80 eq) were suspended in this solution. The

resulting suspension was placed in a washed dialysis bag (MWCO 12 kDa) and agitated

manually for ~ 10 sec to promote resin swelling. It was then placed in a 100 mL container.

Pronase (0.5 g) was dissolved in 20 mL buffer. The resulting solution was added to a

dialysis bag (MWCO 1 kDa) to form the destruction chamber and was placed beside the

other resin bag in the container. CA (0.84 g, 28 ~mol, 1 eq) was dissolved in 20 mL

152

buffer and poured over the two bags in the container forming the immersion solution

(screening chamber). The system was set to stirring on a TW3 orbital shaker (Rose

Scientific) at 60 rpm. New resin was added directly to the synthesis compartment after

every 16 h. After two cycles, the solution inside the synthesis chamber was filtered and

the filtrate was used to suspend the next portion of resin. 200 !AL samples were taken

from the screening chamber approximately every 8 hrs, heated to 80°C for -5 min.,

centrifuged for 10 min and filtered through a 0.22 /lm sterile filter to form the HPLC

sample. Samples were run through a Phenomenex C8 (2) column monitoring at 220 nm

under the following conditions: 0-120 min gradient 80% H20/0.1%TFA, 20% Me OH to

0% H20/0.1%TFA, 100% MeOH. The peaks monitored were EtocProPhesa (7d) at 18.5

min, EtocLeuPhesa (7c) at 39.8 min, EtocProPhe (8d) at 52.7 min, and EtocLeuPhe (8c)

at 69.8 min.

Six Membered p-DCL: A buffer solution was prepared in 100 mL ddH20 with BICINE

(0.490 g, 30 mM) and PEN-Na (10 mg, 0.1 g/mL). The buffer was adjusted to pH 9.0 and

used to dissolve/suspend all components of the system. Phesa (1) (49.5 mg, 0.202 mmol,

7.9 eq) and Phe (6) (33.4 mg, 0.20 mmol, 7.9 eq) were dissolved in 10 mL of buffer,

which was then re-adjusted to pH 9.0. EtocPro EtocGly and EtocPhe Tentagel active

esters (each 125 mg, 22.5 /lmol, 0.80 eq) were suspended in this solution. The resulting

suspension was placed in a washed dialysis bag (MWCO 12 kDa) and agitated manually

for -10 sec to promote resin swelling. Tt was then placed in a 100 mL container. Pronase

(0.5 g) was dissolved in 20 mL buffer. The resulting solution was added to a dialysis bag

(MWCO 1 kDa) to form the destruction chamber and was placed beside the other resin

153

bag in the container. CA (0.84 g, 28 J.lmol, 1 eq) was dissolved in 20 mL buffer and

poured over the two bags in the container forming the immersion solution (screening

chamber). The system was set to stirring on a TW3 orbital shaker (Rose Scientific) at 60

rpm. New resin was added directly to the synthesis compartment after every 16 h. After

two cycles, the solution inside the synthesis chamber was filtered and the filtrate was

used to suspend the next portion of resin. 200 J.lL samples were taken from the screening

chamber approximately every 8 hrs, heated to 80°C for ~5 min., centrifuged for 10 min

and filtered through a 0.22 J.lm sterile filter to form the HPLC sample. Samples were run

through a Phenomenex C8 (2) column monitoring at 220 nm under the following

conditions: 0-120 min gradient 80% H20/0.1%TFA, 20% MeOH to 0% H20/0.1%TFA,

100% MeOH. The peaks monitored were EtocGlyPhesa (7b) at Il.3 min, EtocProPhesa

(7d) at 18.5 min, EtocGlyPhe (8b) at 41.0 min, EtocPhePhesa (7a) at 43.5 mm,

EtocProPhe (8d) at 52.7 min, and EtocPhePhe (8a) at 70.6 min.

General Procedure for 30 Membered p-DCLs: A buffer solution was prepared in 100

mL ddH20 with BI CINE (0.490 g, 30 mM) and PEN-Na (10 mg, 0.1 g/mL). The buffer

was adjusted to pH 9.0 and used to dissolve/suspend aIl components of the system. Phesa

(1) (38.7 mg, 0.101 mmol, 7.21 eq), Alasa-TFA (30.5 mg, 0.101mmol, 7.21 eq), Lyssa

TFA (36.6 mg, 0.101mmol, 7.21 eq), GlUNHOH (17.5 mg, 0.101 mmol, 7.21 eq) and Phe

(6) (17.5 mg, 0.101 mmol, 7.21 eq) were dissolved in 5 mL ofbuffer. EtocAla, EtocVal,

EtocPro, EtocPipecolic acid, EtocLeu and EtocPhe Tentagel active esters (each 80 mg,

14.4 J.lmol, 1.03 eq) were suspended in this solution. The resulting suspension was placed

in a washed dialysis bag (MWCO 12 kDa) and agitated manually for ~1O sec to promote

154

resin swelling. It was then placed in a 100 mL container. Pronase (0.25 g) was dissolved

in 10 mL buffer, the resulting solution was added to a dialysis bag (MWCO 1 kDa) to

form the destruction chamber, and was placed beside the other resin bag in the container.

CA (0.42 g, 14 Ilmol, 1 eq) was dissolved in 10 mL buffer and poured over the two bags

in the container forming the immersion solution (screening chamber). The system was set

to stirring on a TW3 orbital shaker (Rose Scientific). New resin was added directly to the

synthesis compartment after every 16, 24 or 32 h, depending on the experiment. After

every two cycles, the solution inside the synthesis chamber was filtered and the filtrate

was used to suspend the next portion of resin. 200 ilL samples were taken from the

screening chamber every 4-16 hrs, heated to 80°C for ~5 min., centrifuged for 10 min and

filtered through a 0.22 Ilm sterile filter to form the HPLC sample. Samples were run

through a Phenomenex C8 (2) column monitoring at 220 nm under the following

conditions: 0-180 min gradient 100% H20/0.1 %TFA, 0% MeOH/0.1 %TFA to 0%

H20/O.I%TFA, 100% MeOH/O.1%TFA. The peaks monitored were EtocProPhesa (7d) at

24.5 min, EtocValPhesa (7e) at 25.0 min, EtocAlaPhesa (7t) at 42.5 min, EtocPipPhesa

(7g) at 70.8 min, EtocLeuPhesa (7e) at 78.1 min, EtocPhePhesa (7a) at 82.4 min,

EtocProPhe (8d) at 58.2 min, EtocValPhe (8e) at 61.2 min, EtocAlaPhe (8t) at 82.1 min,

EtocPipPhe (8g) at 90.0 min, EtocLeuPhe (8e) at 100.1 min, EtocPhePhe (8a) at 108.3

min, EtocProAlasa (ge) at 36.4 min, EtocValAlasa (9b) at 47.3 min, EtocAlaAlasa (9d) at

57.5 min, EtocPipAlasa (9t) at 60.1 min, EtocLeuAlasa (ge) at 67.7 min, EtocPheAlasa

(9a) at 70.6 min, EtocProLyssa (lOe) at 43.8 min, EtocValLyssa (lOb) at 55.1 min,

EtocAlaLyssa (lOd) at 57.9 min, EtocPipLyssa (lOt) at 67.6 min, EtocLeuLyssa (lOc) at

71.8 min, EtocPheLyssa (lOa) at 75.4 min, EtocProGluNHoH (Ile) at 35.6 min,

155

EtocValGluNHoH (l1b) at 41.9 min, EtocAlaGluNHoH (l1d) at 57.2 min, EtocPipGluNHoH

(11f) at 59.9 min, EtocLeuGluNHoH (l1c) at 67.3 min, and EtocPheGluNHoH (l1a) at 69.7

min,

References

1. Y. Pocker, J. T. Stone Biochemistry 1968, 7,3021-3031.

2. A. D. Corbett, J. L. Gleason Tetrahedron Lett. 2002,43, 1369-1372

3. J. D. Cheeseman, A. D. Corbett, R. Shu, J. Croteau, J. L. Gleason, R. J.

Kazlauskas J. Am. Chem. Soc. 2002,124,5692-5701.

4. J. B. Chaires, N. Dattagupta, D. M. Crothers Biochemistry, 1982,21, 3933-

3940.

156

CONTRIBUTION TO KNOWLEDGE

The nature of evolution dictates that there will forever be a need for new synthetic

compounds for use as drugs against disease. Combinatorial chemistry has accelerated the

process of discovering drug leads, and receptor assistance in combinatorial systems has

improved upon the se ideas to evolve new, small molecules.

Pseudo-dynamic libraries exhibit selectivity higher than in any other receptor

assisted system, and as such, are promising in their potential application to the drug

discovery process. The first p-DCL system, its optimization, expansion and theoretical

development enable this method to be applied to other biological targets. This is a highly

versatile method in that it does not require any specific type of receptor to be efficient,

and it does not necessarily require high technology in its application.

157

JACS ARTICLES

Published on Web 04/2612002

Amplification of Screening Sensitivity through Selective Destruction: Theory and Screening of a Library of Carbonic

Anhydrase Inhibitors

Jeremy D. Cheeseman, Andrew D. Corbett, Ronghua Shu, Jonathan Croteau, James L. Gleason,* and Romas J. Kazlauskas*

Contribution from the Department of Chemistry, McGill University, 801 Sherbrooke St. West, Montreal, Quebec H3A 2K6, Canada

Received September 17, 2001

Abstract: A new method for identifying enzyme inhibitors is to conduct their synthesis in the presence of the targeted enzyme. Good inhibitors form in larger amounts than poorer ones because the binding either speeds up synthesis (target-accelerated synthesis) or shifts the synthesis equilibrium (dynamic combinatorial libraries). Several groups have successfully demonstrated this approach with simple systems, but application to larger libraries is challenging because of the need to accurately measure the amount of each inhibitor. ln this report, we dramatically simplify this analysis by adding a reaction that destroys the unbound inhibitors. This works similar to a kinetic resolution, with the oost inhibitor being the last one remaining. We demonstrate this method for a static library of several sulfonamide inhibitors of carbonic anhydrase. Four sulfonamidecontaining dipeptides, EtOC-Phesa-Phe (4a), EtOC-Phesa-Gly (4b), EtOC-Phesa-Leu (4c) and EtOC-PhesaPro (4d), were prepared and their inhibition constants measured. These inhibitors migrated to the carbonic anhydrase compartment of a two-compartment vessel. Although higher concentrations of the better inhibitors were observed in the carbonic anhydrase compartment, the concentration differences were small (1.83:1.71 :1.54:1.46:1 for 4a:4b:4c:4d:5, where 5 is a noninhibiting dipeptide EtOC-Phe-Phe). Addition of a protease rapidly cleaved the weaker inhibitors (4d and 5). Intermediate inhibitor 4c was cleaved at a slower rate, and at the end of the reaction, only 4a and 4b remained. In a separate experiment, the ratio of 4a to 4b was found to increase over time to a final ratio of nearly 4:1. This is greater than the ratio of their inhibition constants (approximately 2:1). The theoretical model predicts that these ratios would increase even further as the destruction proceeds. This removal of poorer inhibitors simplifies identification of the best inhibitor in a complex mixture.

Introduction

Synthesis using combinatorial chemistry allows testing of hundreds of thousands of drug candidates using high throughput screening techniques. Although this rapid pace has revolutionized drug development, the search for faster and more efficient testing methods continues. One promising method is in situ screening of mixtures such as in dynarnic combinatorial libraries.! Dynamic combinatorial libraries are equilibrating mixtures of organic molecules. Equilibration in the presence of a therapeutic target increases the equilibrium amounts of those library members that bind tightly to that target. The difference in library composition with and without a stoichiometric amount of target identifies the best inhibitors.

Dynarnic libraries are still in the developmental stage, and only a few examples have been reported.2 For example, Ramstrom and Lehn3 created a dynamic library of 10 di-

* To whom correspondence should be addressed. E-mail: jim.gleason@ mcgill.ca and [email protected]. (1) Reviews: Ganesan, A. Angew. Chem., Int. Ed. 1998,37,2828-2831; Lehn,

J.-M. Chem. Eur. J. 1999, 5, 2455-2463; Cousins, G. R. L.; Poulsen, S.A.; Sanders, J. K. M. Curr. Opin. Chem. Biol. 2000,4,270-279; Huc, 1.; Nguyen, R. Comb. Chem. High Throughpul Screening 2001, 4, 53-74.

5692 • J. AM. CHEM. SOC. 2002, 124,5692-5701

saccharides by disulfide exchange starting from a mixture of monosaccharide tItiols. The library was screened against concanavalin A, which binds mannose-rich oligosaccharides. A mannoside homodimer was the strongest binder in the library. When the disulfide exchange was carried out in the presence of concanavaHn A, the amount of mannoside homodimer present increased by 40%. This increase in the amount of mannoside homodimer identifies it as the best-binding disaccharide. Analysis of a larger 21-member library was more difficult because HPLC did not resolve each member. Nevertheless, the mannoside homodimer was clearly favored in tItis library as weIl.

To make a real impact on drug discovery, methods must be developed to screen dynarnic combinatorial libraries with thousands of members. This screening is complicated because

(2) For examples aimed towards biological targets, see: (a) Huc, 1.; Lehn, J.-M. Proe. Nol/. Aead. Sei. U.SA. 1997, 94, 2106-2110. (h) Nicolaou, K. C.; Hughes, R.; Cho, S. Y.; Winssinger, N.; Smethurst, c.; Labiscbinski, H.; Endermann, R. Angew. Chem., Int. Ed. 2000,39,3823-3828. (c) Karan, c.; Miller, B. L. J. Am. Chem. Soc. 2001, 123, 7455-7456. (d) Bunyapaiboonsri, T.; Ramstriim, O.; Lohmanu, S.; Lehn, J.-M.; Peng, L.; Goeldner, M. ChemBioChem 2001, 2, 438-444. (e) Nguyen, R.; Huc, 1. Angew. Chem., Inl. Ed. 2001,40, 1774-1776.

(3) Rarnstrôm, O.; Lehn, J.-M. ChemBioChem 2000,1,41-48.

10.1021118017099+ CCC: $22.00 02002 America" Chemlcal Society

Screening Sensitivity through Selective Destruction

Scheme 1. Aryl Sulfonamide-Based Dipeptide Libraries as Inhibitors of Carbonic Anhydrasea

7 0 ~3

ff .... N NÂCOH 2 H 2

1'" R h-

selective pressure lœt)

carbonic anhydrase , 1 hydrolysis of

poorer inhibitors

mem rane

HI 0 R

ff2"N OH + H2N~C02H 1'"

R, h-

1 R, = S02NH2 R2 = H

a Strong binding inhibitors will be bound to carbonic anhydrase and protected. Weaker inhibitors will be hydrolyzed by a protease.

it is often difficult to measure the concentration of each library member in the absence and presence of a target. Further, the libraries will likely contain not one but many good inhibitors because many library members have similar structures and thus similar binding constants. In these cases, adding the target increases the concentration of many library members, rather than a single member, and makes analysis very diffieult or impossible. Eliseev and Nelen4 estimated that a dynarnic library combined with an affinity column containing the target would yield one major compound (>50%) only if KstronglKweak was at least n, where n is the number of members of the library. Thus, for one member to predominate in a library of 1000 members, that member would have to bind > 1000 times stronger than the others, an unlikely possibility. This inability to distinguish between inhibitors of similar binding constants is a major limitation of the CUITent dynarnic combinatorial libraries.

In this report, we propose a screening method that enhances the ability to detect the best inhibitor in a mixture of similar inhibitors. The key to the method is an irreversible destruction reaction that destroys the unbound and weakly bound inhibitors, similar to a kinetic resolution. The best inhibitor is the last one remaining. We demonstrate that this method works for a statie library and discuss its potential application to a dynarnic system.

Our library targets carbonie anhydrase and consists of dipeptides with an N-terrninal 4' -sulfonarnidophenylalanine (1, Phesa).5 These dipeptides can either bind to carbonic anhydrase or be c1eaved by a protease, Scheme 1. This c1eavage increases the ratio of the strongest binder relative to weaker binders. Importantly, the ratio may increase to values significandy greater than the ratio of the binding constants, thus overcoming the limitation identified by Eliseev and Nelen and making it easy to identify the best inhibitor in the mixture.

Theory

Finding the Best Inhibitor by Shifting the Equilibrium To Make More of the Better Inhibitors. Most dynarnic combinatorial library experiments contain two equilibria: an

(4) (a) Eliseev, A. V.; Nelen, M. J. J. Am. Chem. Soc. 1997, 119, 1147-1148. (b) Eliseev, A. V.; Nelen, M. J. Chem. Eur. J.1998, 4, 825-834.

(5) G1aucorna patients oflen talce carbonic anhydrase inhibitors to reduce the pressure in the eye. Ali commercial inhibitors contain a sulfonamide moiety. We chose carbonic anhydrase as a test case for inhibitor design and screening methods.

ARTICLES

Scheme 2. Dynamic Combinatorial Library Equilibria Yield a Higher Total Amount of a Good Inhibitora

~D~D.O ~6) ~ newtnhlbltors target. Inhlbhor

compte ...

a Binding of the inhibitor to a target removes it from the synthesis equilibrium so that the synthesis produces more of the good inhibitor.

equilibrium for the synthesis of inhibitors and an equilibrium for binding of the inhibitors to the target, Scheme 2. The binding equilibrium removes the good inhibitors from the synthe sis equilibrium, and to reestablish equilibrium, the synthesis produces more of the good inhibitors than it would in the absence of target. First, we show that the ratio of good inhibitor to a poor inhibitor depends linearly on the ratio of the binding constants.

Consider a common starting material, SM, in equilibrium with two inhibitors, A and B, which can each bind reversibly to a target, T, to form complexes T'A and T'B

KsA KaA ====:!:: A + T =====:!:: (1)

SM KsB KaB

==:!::B+T==:!:: (2)

If [AT] is the total of bound and unbound forms of A ([AT] = [A] + [T'A]), it can be shown that under typical conditions (e.g., tight binding and an excess of target at a high concentration), the equilibrium ratio of the total amounts of the two inhibitors, [AT] and [BT], depends linearly on their relative association constants (eq 3, see Supporting Information for a derivation).6

(3)

For the ideal case where there is one good inhibitor in a pool of noninhibitors, these equilibria indeed will yield the good inhibitor. For example, assurning the rates of synthesis are equal, a mixture of two inhibitors differing in their inhibition constants by a factor of 10 will give a 1:1 mixture in the absence of target (50 mol % of better inhibitor) but gives al: 10 mixture in the presence of target (91 mol % of better inhibitor). Similarly, a mixture of 1000 inhibitors would yield 0.1 mol % of each inhibitor in the absence of target, but in the presence of target, the poorer inhibitors would decrease to 0.09 mol % each, while the one good inhibitor would increase to 0.9 mol %. This very simple example is already a difficult analysis problem. The more common situation where many inhibitors with similar binding constants are present may become difficult or impossible to analyze.

Finding the Best Inhibitor by Destruction of Poorer Inhibitors. One way to enhance the concentration differences between inhibitors with similar binding constants is to add an irreversible reaction that removes the unbound poorer inhibitor.

(6) If the binding is not tight or the target is not in excess at high concentration, then the concentrations of the inhibitors will he mOre similar than those discussed in the texl above, and il will he even harder to distingnish which is the better inhibitor.

J. AM. CHEM. SOC •• VOL. 124. NO. 20. 2002 5693

ARTICLES

Scheme 3. Destruction of Inhibitors·

&\I~ ~ I=:>>> q) dl •• oc~tJon

tergat • Inhibltor complexe.

o + 0 d ••• ,uctoo 8 D Inhlbltors <J

a The free concentration of the poorer inhibitor is higher, thus it is destroyed more readily. This destruction reaction exponentially increases the relative concentration of the good inhibitor similar to a kinetic resolution.

This situation is similar to a kinetic resolution of enantiomers.7

As the destruction reaction winnows away the poorer inhibitors, the relative concentration of the best inhibitor increases exponentially (Scheme 3). The analysis below is similar to that for kinetic resolutions.7

Consider two inhibitors, A and B, that compete for a target, T, and are also destroyed by an irreversible reaction to yield P and Q with rates of k2A and k2B.

KdA kzA T'A=T+A-P

The rate of disappearance of inhibitor A is

d[A] -=-k [A] dt 2A

(4)

(5)

(6)

If [AT] is the total concentration of bound and unbound forms of inhibitor A, it can be shown that

(7)

Upon solving for [A] and substituting into eq 6, the rate of disappearance of A is given by

d[A] --=

dt (8)

When the target is in excess of the inhibitor, the concentration of the free target, T, will be much larger than the dissociation constant, KdA, so [T] » KdA' therefore, eq 8 simplifies to

d[A]

dt (9)

A similar equation is obtained for inhibitor B. The ratio of their partial reaction rates is

KdB k2B where S = - x - (10)

K dA k2A

This equation shows that the relative rate of disappearance of the two inhibitors depends linearly on their total concentration, their relative binding ability, and their relative rate of destruction. For simplicity, we define S as the product of the relative binding abilities and relative rates of destruction of the

5694 J. AM. CHEM. SOC •• VOL. 124, NO. 20, 2002

12

10

8

2

o

Cheeseman et al.

o 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Conversion

Figure 1. Predicted ratio of the total (bound and unbound) concentrations of two hypothetical inhibitors, A and B, as a function of the degree of conversion for given values of S. The degree of conversion is the fraction of the total amount of inhibitors !hat have been destroyed. The calcu1ated Hnes follow eqs 11 and 12 where the initial total concentration of the inhibitors is one. This graph shows that the ratio of the two inhibitors can be much larger !han the value of S, even for values of S as low as 2.

two inhibitors. If the rates of destruction of the two inhibitors are equal, then S is the ratio of the dissociation constants and will be greater than one if A is a stronger inhibitor of the target than B.

Upon integration of eq 10, one finds that the ratio of the total amounts of the two inhibitors varies exponentially with S (eq Il), where [AT]O represents the initial total concentration of inhibitor A. This exponential relationship enhances the ability to detect small differences as the destruction reaction progresses.

(11)

By measuring the relative concentration of the two inhibitors during the destruction reaction, the value of S can be determined using eq 11. Alternatively, eq 12 below, which expresses [AT], [BT] , [AT]O, and [BT]O in terms of the conversion, C, and the ratio of the total concentrations of the two inhibitors, can be used to determine S.

_ln....:;[~(1_-_C).....:....:...(2....;/('-1_+_R..:..:»-=-] = S ln[(l - C)(2R/(1 + R»]

[AT] + [BT] where C = 1 - [AT]o + [BT]o

These predictions are shown graphically for several values of S in Figure 1. If the rates of destruction of the two inhibitors are equal, then S is the ratio of the dissociation constants. As the destruction reaction proceeds (conversion increases from 0 to 1), the ratio of the total amounts of the two inhibitors, [AT]! [BT], varies when S ~ 1. When S is large (e.g., 40), the relative concentration of the good inhibitor increases steeply near 50% conversion. When S is small (e.g., 2), the relative concentration of the good inhibitor increases steeply near 90% conversion.

(7) The analysis below follows closely the mathematical treatment for kinetic resolutions. For examples, see: Martin, V. S.; Woodard, S. S.; Katsuki, T.; Yamada, Y.; Ikeda, M.; Sharpless, K. B. J. Am. Chem. Soc. 1981, 103, 6237-6240; Chen, C. S., Fujimoto, Y., Girdaukas, G., Sih, C. J. J. Am. Chem. Soc. 1982, 104, 7294-7299; Kagan, H. B., Fiaud, J. C. Top. Stereochem. 1988, 18, 249-330.

Screening Sensitivity through Selective Destruction ARTICLES

Scheme 4. Preparation of 4'-Sulfonamidophenylalanine Dipeptides

~ 1. CIS03H, t;, H~02H EtO~CI Et02C,N ........... :. C~H 2 NH OH' NHz-Xaa-OtBu

. 4 ~ --v' NaHC03 --v-3. hog kldney H N 1 H~ 1 dioxane H ~ 1 EDe. HOBT, CH~12 acylase 1 2 ',R~ .& 2 ,R~ .--:;

o 0 0 0 1 2

In either case, the ratio of the total amounts of the two inhibitors, [Ar]/[Br], can be much larger than the value of S.

Results

Preparation of 4'-Sulfonamidophenylalanine Dipeptides. (S)-4'-Sulfonamidophenylalanine (1 or Phe.,.) was prepared from (S)-N-acetylphenylalanine by a modification of the procedure described by Escher et al. 8 Thus, chlorosulfonylation of Nacetylphenylalanine in chlorosulfonic acid at 60 oC followed by ammonolysis afforded N-acetyl-4' -sulfonamidophenylalanine. Direct purification of this intermediate proved difficult. Therefore, it was deacetylated using hog kidney acylase 1,9 and the resulting free amine acid 1 was purified by ion-exchange chromatography and recrystallization. Using this procedure, 1 was prepared as an analytically pure solid in 40% yield from N-acetyl-Phe. The a-amino group was selectively blocked using ethyl chloroformate under standard Schotten-Baumann conditions. The requisite dipeptides were then prepared by coupling 2 with terl-butyl amino acid esters using EDCIHOBT,IO followed by trifluoroacetic acid-mediated deprotection of the ester function to afford dipeptides 4a-d (Scheme 4). No acylation of the sulfonamide nitrogen was observed under either the Schotten-Baumann or peptide-coupling conditions. Dipeptide EtOC-Phe-Phe (5), which does not contain a sulfonamide group and serves as a control, was prepared by standard methods.

Inhibition of Carbonic Anhydrase. Sulfonamides 1 and 2 as weIl as sulfonamide dipeptides 4a-d all inhibited the carbonic anhydrase-catalyzed hydrolysis of 4-nitrophenyl acetate (pNPA). The inhibition was competitive and Lineweaver-Burk plots revealed similar inhibition constants, which varied by only a factor of 10 (Table 1). The parent amino acid 1 (Phes.) was the poorest sulfonamide inhibitor (KI = 13 ,uM), while dipeptides 4a (EtOC-Phesa-Phe) and 4b (EtOC-Phesa-Gly) were the best sulfonamide inhibitors (KI = 1.2 and 2.5 ,uM, respectively). Dipeptides 4c (EtOC-Phesa-Leu) and 4d (EtOC-Phesa-Pro)

(8) Escher, E.; Bernier, M.; Parent, P. Helv. Chim. Acta 1983, 66, 1355-1365.

(9) Researchers oflen use hog kidney acylase to resolve enantiomers of N-acetyl amino acids. Chenault, H. K.; Dahmer, J.; Whitesides, G. M. J. Am. Chem. Soc. 1989, 11 J, 6354-6364; Roberts, S. M., Ed. Prepara/ive Bio/ransformations; Wiley: Chichester 1992-1998; Module 1: 14. In our case, !his intennediate was a1readyenantiomerically pure. We used hog kidney acylase to c1eave the acetyl group under milder conditions than those required by chemical c1eavage methods.

(10) EDC = 1-(3-dimethylaminopropyl)-3-ethylcarbodiimide; HOBT = l-hydroxybenzotriazole.

5

Table 1. Inhibition of Carbonic Anhydrase by Sulfonamides 1 and 2, Sulfonamide Dipeptides 4a-d, and Dipeptide 5

compound K, (uM)'

Phes.(l) EtOC-Phes• (2) EtOC-Phe..-Phe (4a) EtOC-Phe..-Gly (4b) EtOC-Phe..-Leu (4c) EtOC-Phesa-Pro (4d) EtOC-Phe-Phe (5)

13 ± 1.6 12 ± 1.4

1.2 ± 0.2 2.5 ± 0.5 4.4 ± 0.7 9.4 ± 1.6

»100()b

a Competetive inhibition constants for the carbonic anhydrase-catalyzed hydrolysis of p-nitrophenyl actetate (pNPA) at 25 °C.in phosphate bu~fer (l0 mM pH 7.5). A typical procedure was to add carbomc anhydrase solution (100 ilL, 0.05 mg/mL) containing inhibitor (0.0-100 IlM in most cases) to an acetonitrile solution of pNPA (5.0 ilL, 2.0-32 mM) and follow the release of p-nitrophenoxide spectrophotometrically at 404 nM. b No inhibition detected at an inhibitor concentration of 1 mM

showed slightly higher inhibition constants (4.4 and 9.4 ,uM, respectively). Other simple sulfonamides also inhibit carbonic anhydrase with similar inhibition constants. 1 1 As expected, the dipeptide lacking a sulfonamide group, 5, did not inhibit carbonic anhydrase.

Selective Extraction of Inhibitors by Carbonic Anhydrase. First, we demonstrated that a strongly binding inhibitor concentrates into the carbonic anhydrase-containing compartment of a two-compartrnent vessel (cf. Figure 4 without Pronase). The two compartrnents were created by suspending a dialysis bag containing a solution of bovine carbonic anhydrase 12 (0.34 mM) in a solution of phosphate buffer. The dialysis membrane (12-kDa cutoff) separated the two compartments so that small molecules such as the sulfonamide dipeptides could diffuse freely across the membrane, while carbonic anhydrase (30 kDa) could not. Both compartrnents initially contained a mixture of 0.16 mM sulfonamide dipeptide 4a and 0.19 mM noninhibitor dipeptide 5. Over several hours the total sulfonamide concentration increased in the inside compartrnent containing carbonic

(11) For example, N guyen and Huc investigated a simple sulfonarnides with inhibition constants of ~O.I-l ,uM (Nguyen, R.; Huc, I. Angew. Chem., Int. Ed. 2001,40, 1774-1776), while Doyon et al. investigated other simple sulfonarnides with inhibition constants of ~O.OOI ,uM (Doyon, J. B.; Hansen, E. A. M.; Kim, C.-Y.; Chang, J. S.; Christianson, D. W.; Madder, R. D.; Voet, J. G.; Baird, T. A., Jr.; Fierke, C. A.; Jain, A. Org. Leu. 2000, 2, 1189-1192).

(12) These experiments required stoichiometric amounts o~ carbonic anhydrase. We used an inexpensive mixtore of isozymes from bovme sources. Alth~ugh material was not pure carbonic anhydrase, we calcu1ated the concentrations assuming it was pure. Thus, the true concentration will be less than the number given.

J. AM. CHEM. SOC .• VOL. 124, NO. 20, 2002 5695

ARTICLES

[dipeptide) (mM)

0.3 a) Outside Chamber

-...--a-_____ S

0.2

0.1

-------4a

O+-----,------r----~----~

o 3 6 Tlme (h)

9 12

0.3

0.2

[dipeptide) (mM)

0.1

Cheeseman et al.

b) Ioside Chamber _---.--~4a

-!...._e_.....--_---- s

O+-----.------r-----r----~

o 3 6

Time(h)

9 12

Figure 2.13 Selective concentration of the sulfonamide 4a over noninhibitor S into the carbonic-anhydrase-containing compartment of a two-compartment vesse!. One compartment contained carbonic anhydrase (0.34 mM), while both compartments (20 mL each) initially contained equal concentrations of sulfonarnide 4a (0.16 mM) and noninhibitor 5 (0.19 mM). The sulfonamide diffused freely across the dialysis membrane and concentrated in the carbonicanhydrase-containing compartment as shown In contras!, the concentrations of noninhibitor S remained similar in both compartments. After 12 h, the concentration of sulfonarnide 4a in the outside compartment decreased to 0.04 mM and increased in the inside compartment to 0.28 mM (total of free and carbonic anhydrase-bound). The final ratio of 4a to 5 in the carbonic anhydrase chamber was 1.75: 1.

20 a) Outside Chamber

16

12&-.-__________ ~----------__ 5 [dipeptide) I~;::::::-___ ....... __

(mM) 8 1 . 4 _--, ____ -=~ d

4 4c

~::::::=====:::4b 4a O+----,-----r----~--~r---~

o 10 20 30 40 50

TIme(h)

Idipeptidel (mM)

20

16

b) Ioside Chamber 4a 4b

__ ~ __ -----==~4c 4d

12~~~=-------~-~'---------~~---------'S

8

4

O+-----r---~-----r----,---~ o 10 20 30 40 50

Time(h)

Figure 3.13 Selective concentration of the sulfonamides 4a-4d over noninhibitor 5 into the carbonic-anhydrase-containing compartment of a two-compartment vessel separated by a dialysis membrane. One compartment contained carbonic anhydrase (0.485 mM), while both compartments (20 mL each) initially contained equal concentrations of sulfonarnides 4a-4d and noninhibitor 5 (~O.ll mM each). The sulfonamides diffused freely across the dialysis membrane and concentrated in the carbonic-anhydrase-containing compartment. In contrast, the concentration of noninhibitor S increased slightly in the outer compartment.

anhydrase and decreased in the outside compartment (Figure 2). Alternatively, the concentrations of the noninhibitor 5 remained similar in both compartments. This result showed that tight binding to a target could concentrate a good inhibitor into one compartment of a two-compartment reaction vesse!.

In a similar experiment using a mixture of inhibitors, we could further detect differences in relative inhibition constants. A more tightly binding inhibitor concentrated in the carbonic anhydrase compartment to a greater extent than a less tightly binding inhibitor. Starting with an equimolar mixture of sulfonamide dipeptides 4a-d and the noninhibitor 5 in both compartments, the sulfonamide dipeptides concentrated into the carbonic anhydrase compartment, Figure 3. The fraction of dipeptide in the carbonic anhydrase compartment varied: 88, 82, 74, 70, and 48% for 4a, 4b, 4c, 4d, and the noninhibitor 5, respectively (or a ratio of 1.83: 1. 71: 1.54: 1.46: 1 for 4a:4b:4c:4d:5). The order of highest to lowest concentration in the carbonic anhydrase chamber reflects the order of the binding constants of the inhibitors.

These results show that is possible to distinguish between inhibitors, but the differences in concentration are small, especially among inhibitors of similar strength. Even comparing the best inhibitor (4a) with a noninhibitor (5) gives a concentration differing by less than a factor of 2. To enhance this

5696 J. AM. CHEM. SOC •• VOL. 124, NO. 20,2002

difference in concentration, we explored the use of proteases to destroy the poorer inhibitors.

Screening of Proteases. We screened 22 commercially available proteases to identify those that could hydrolyze the dipeptide EtOC-Phesa-Phe (4a). AU proteases showed sorne activity. Using 0.1 mg of protease and 2.umol (2 mM) dipeptide 4a, the five most active proteases (a-chymotypsin, protease from Streptomyces casepitosus, proteinase K, Pronase from Streptomyces grise us, and protease from Bacillus thermoproteolyticus rokko) cleaved all of the dipeptide within 24 h, while two moderately active proteases (protease N "Amano", protease from Bacillus polymyxa) cleaved a11 of the dipeptide within 48 h. The remaining proteases cleaved less than half of the dipeptide after 72 h. We chose one of the most active yet inexpensive enzymes, Pronase from S. grise us, for subsequent experiments. Pronase was found to cleave all five dipeptides (4a-d and 5), although the glycine and proline dipeptides (4b and 4d) were cleaved more slowly (80-90% hydrolysis within 24 hl. To ensure high cleavage rates, larger amounts of Pronase were used in the competitive degradation experiments described below.

Selective Protection of Inhibitors by Carbonic Anhydrase from Hydrolysis. We compared the ability of carbonic anhydrase to protect sulfonamide inhibitor 4a from hydrolysis while allowing a noninhibitor, 5, to be hydrolyzed. An experiment


Outar Chambar Inner Chamber

Ho.z~ H NHEtOC .,~ ::::;:::::b:: HO ~ NHEtOC ; .,~ : 2 H

pronase 1 1" .... ,

H~HEIOC 1"

" "'

! 1 ::

~i "' .si 1 carbonic ml anhydrase

~ ~I

!

~

Ho,c"- NHEtOC R.~ " • CA 1 ... .,

Inhlbltor 1 CA complex

Figure 4. Reaction design for the selective destruction experiments. The dipeptides can diffuse across the dialysis membrane into either chamber. One chamber contains carbonic anhydrase, the other contains Pronase. Dipeptides in the Pronase charnber are rapidly cleaved to their constituent pieces. Carbonic anhydrase prevents strong binding dipeptides from diffusing across the membrane and thus slows their hydrolysis.

0.2 Ioside Chamber

0.16 r-~"'---,--""--=------,, 4a

0.12 [dipeptide)

(mM) 0.08

0.04

+-__ ,-___ -r __ -r~=_~5

3 6 9 12

Time (h)

Figure 5.13 Selective protection from hydrolysis of sulfonamide 4a over noninlùbitor 5 by carbonic anhydrase. A vessel containing two cornpartrnents of equal volume (20 mL each) separated by a dialysis membrane was filled with a solution of. sulfonamide 43 (0.16 mM) and noninlùbitor 5 (0.19 mM). The inside compartment contained carbonic anhydrase (0.34 mM), while the outside compartment contained Pronase. The protease rapidly cleaved the dipeptides in the outside compartment to the corresponding amino acids (data not shown). The noninhibitor 5 diffused freely across the dialysis membrane and was cleaved by the protease. In contrast, the inhibitor 43 bound to carbonic anhydrase in the inside compartrnent and was not consumed at a significant rate. After 6 h, the concentration of sulfonamide 43 in the inside compartment decreased by oruy 6% (0.15 mM), while the concentration of noninhibitor 5 decreased to 0.041 mM during the sarne time period (ratio = 3.7:1).

similar to that described above, except with Pronase added to

the outer chamber was set up (Figure 4). On the one hand, in

the Pronase-containing chamber, both dipeptides were rapidly

cleaved to the constituent pieces within 15 min. On the other

hand, the inside compartment showed a steady decrease in the

concentration of noninhibitor 5 over 12 h (Figure 5), while the

concentration of sulfonarnide 4a remained nearly constant

(a decrease of 9% over 12 h).14 After even just 6 h, the ratio,

4a:5, in the inside compartment was 3.7:1 and continued to

increase to greater than 20:1 after 12 h. By comparison, the

experiment which does not con tain Pronase had a final ratio of

4a to 5 of 1.75:1. In a simlar experiment, dipeptides 4a and 4b, which have

very similar binding constants, were exposed to carbonic

anhydrase and Pronase. In this experiment, the dipeptides were

(13) Lines drawn in ail figures (except Figure 1 and Figure 9) are for illustration purposes only. They do not represent theoretical Iines of any sort.

(14) Both 4a and 5 diffused through the membrane at identical rates with a half-Iife of about 3 h. (Data not shown.)

0.3 Ioside Chamber

0.25 •

0.2.,..,...-__

[dipeptide) 0 15 (mM) •

0.1

ARTICLES

0.05 43

0+-__ ~ ____ '-__ ~r-__ -r __ ~4b

o 40 80 120 160 200

Time(h)

Figure 6.13 Selective protection from hydrolysis of dipeptide 43 over 4b by carbonic anhydrase. A reaction vessel was separated into Iwo compartments (20 mL each) by a dialysis membrane. The inside compartment contained carbonic anhydrase (13.6 mol), dipeptide 43 (4.3 mol) and dipeptide 4b (4.3 mol) in 20 mL ofbuffer. The outer compartrnent contained Pronase (5 mg) dissolved in 20 mL of buffer. The time course of the reaction in the carbonic anhydrase chamber is shown in the figure. At 83% conversion (193 h) the ratio of 4a to 4b was 3.8:1.

0.2

0.16

0.12 [dipeptide)

(mM) 0.08

0.04

Ioside Chamber

4a

o +-----.....,..-----,.----i 4c

o 2 4 6

Time(h)

Figure 7.n Selective protection from hydrolysis of dipeptide 4a over 4c by carbonic anhydrase. A reaction vessel was separated into two compartments (20 mL each) by a dialysis membrane. The inside compartment contained carbonic anhydrase (0.34 mM), while the outside compartment contained Pronase (4 mg). Both compartments initially contained similar concentrations of dipeptide 43 (0.16 mM) and dipeptide 4c (0.14 mM). The protease rapidly cleaved the dipeptides in the outside compartment to the corresponding amino acids (data not shown). The time course of the reaction in the carbonic anhydrase charnber is shown in the figure. After 6 h, 93% of 4c inside the CA chamber had been hydrolyzed, white oruy 58% of 4a had hydrolyzed A control experiment which did not contain carbonic anhydrase showed an equal rate of hydrolysis for the two dipeptides in the chamber not containing Pronase.

placed only in the carbonic anhydrase chamber, and an excess

of carbonic anhydrase was used (1.6: 1 ratio of CA to dipeptides)

so that the conditions adhered rigorously to those of the theory

described above. As expected, due to the excess of target and

tight binding of both dipeptides, the hydrolysis of 4a and 4b

was slow. However, as in the first reaction, the weaker binder,

4b, was consumed at a higher rate (Figure 6). After 193 h, 83% of the total dipeptides had been hydrolyzed, and the ratio of 4a

to 4b was 3.8:1. This final ratio is in excess of the ratio of the

independently determined binding constants of the dipeptides

(2.1:1).

In a related experiment, we compared two sulfonamide

dipeptides 4a and 4c which also have similar inhibition constants

(Figure 7). In this experiment, both the inside and outside

chambers initially contained equal concentrations of the dipep

tides, and the total concentration of dipeptides was in excess

(2.1: 1 ratio of dipeptides to CA). The result was a much faster

hydrolysis of both dipeptides in the carbonic anhydrase chamber.

This faster rate reflects the rapid release of 1 equiv of PheSA

J. AM. CHEM. SOC .• VOL. 124, NO. 20, 2002 5697

ARTICLES

(2) from the Pronase chamber. Although 2 is a weaker binder than either 4a or 4c, enough of it was produced such that it could displace a small amount of 4a and 4c from the carbonic anhydrase binding pocket, thus accelerating their hydrolysis by Pronase. However, the net result was still the same. After 6 h, 93% of 4c was hydrolyzed after 6 h, but only 58% of 4a was hydrolyzed. Thus, the ratio of concentrations was 6:1, which is much larger than the 1.6: 1 ratio observed in a control experirnent which did not contain Pronase and larger than the 3.7:1 ratio of their binding constants.

Fina11y, an experiment containing a11 five dipeptides (4a-d and 5) was conducted using an excess of carbonic anhydrase (ratio of CA to dipeptides is 1.2:1). The experiment was consistent both with the theory and with the prior results. Dipeptide 5 was cleaved rapidly while dipeptides 4a-d disappeared at rates which corresponded to their binding constants (Figure 8).

Discussion

As expected, the four sulfonamide dipeptides 4a-d all inhibit carbonic anhydrase competitively with similar inhibition constants (within a factor of 10 of each other). Classical kinetics using initial rates easily identified these differences, but these classical methods are slow and require the separate measurement of each inhibitor. This becomes laborious for libraries containing thousands of members.

To rapidly identify the best inhibitor, we used competitive binding to carbonic anhydrase in one compartrnent of a twocompartrnent ce11. The inhibitors concentrated into the carbonic anhydrase compartment of a two-compartment ce11. Higher concentrations of the better inhibitors were observed in the carbonic anhydrase compartrnent, but the concentration differences were small (1.83:1.71:1.54:1.46:1 for 4a:4b:4c:4d:5). If the mixture contained 1000 dipeptides, this competitive experiment would not identify the best inhibitor because it would be difficult to separate all the dipeptides, and the differences in concentration with and without target would be small.

Although this experiment does not include a dipeptidesynthesis step and thus is not a dynamic library, the diffusion across the membrane mimics a synthesis step in a dynamic library in that both are equilibrium processes. For the diffusion process, in the absence of a target, each compartment should contain equal amounts of each inhibitor. In the presence of the target, the carbonic anhydrase chamber con tains more of the tight-binding inhibitors. Thus, the equilibrium for the diffusion process has shifted.

A nonselective destruction of the library members should enhance differences in the relative concentrations of the members bound to the target. The poor binding members are destroyed at a rate higher than that for the strong-binding members, and as the degradation progresses, the ratio improves exponentially in favor of the latter. This was observed in our library, where dipeptide hydrolysis by Pronase was used as the destruction process. In a competition experiment between a strong and weak binder (4a vs 5), the ratio of 4a to 5 in the carbonic anhydrase chamber increased from 1.75:1 in the absence of Pronase to 3.7:1 in the presence of Pronase (at 45% conversion). Furthennore, this ratio continued to increase to > 20: 1 as the reaction progressed. A second experiment with two species with very similar Kj's (4a vs 4b) had a final ratio of 3.8: 1 when the ratio of the binding constants was 2.1: 1. As

5698 J. AM. CHEM. SOC .• VOL. 124, NO. 20, 2002

0.2S

0.2

O.lS [dipeptide)

(mM) 0.1

O.OS

o

Inside Chamber

40 80 120 Time (h)

Cheeseman et al.

160

4a 4b

200

Figure 8.13 Selective protection from hydrolysis of dipeptides by carbonic anhydrase. A reaction vessel separated into two compartments (20 mL each) by a dialysis membrane was set up. The inside compartment contained carbonic anhydrase (25.6 mol) and dipeptides 4a-d and 5 (4.3 mol each) in 20 mL of buffer. The outer compartment contained Pronase (5 mg) dissolved in 20 mL of buffer. The time course of the reaction in the carbonic anhydrase charnber is shown in the figure.

shown in Figure 9a, these results fo11ow the theoretical model closely. Similar results were obtained for an experirnent containing two inhibitors (4a and 4c) where an excess of a weaker binder, PheSA (2), was generated in the reaction mixture. The presence of 2 accelerated the rate of cleavage of 4a and 4c, but as can be seen in Figure 9b, the ratio of dipeptides during the course of the reaction still fo11owed the theoretical model. At 70% conversion, the ratio of 4a to 4c was 6: 1, which is much larger than the 1.6:1 ratio observed in a reaction not containing Pro nase. In all cases, the model indicates that the ratios should continue to increase if the reactions are carried out for even longer periods. In experiments with a large number of library members, this increase will be critical in allowing the tightest binding species to be easily identified. 15

One potentiallimitation of this screening method is selectivity in the destruction reaction. For example, dipeptide 4c is cleaved by Pronase at a much slower rate than that for dipeptide 4a. In such a case, S from eqs 10-12 will not be equal to the ratio of the binding constants, and thus the degradation reaction will not fo11ow the theoretical curves of Figure 1. To accommodate this situation, we used a large amount of protease, and more importantly, we employed a dialysis membrane to separate the target-inhibitor complexes from the protease. In this setup, the rate-limiting step in the destruction reaction is not the proteasecatalyzed cleavage but diffusion across the dialysis membrane. Unlike the protease-catalyzed cleavage, the rate of diffusion does not vary significantly with the structure of the inhibitor, and the result is that the destruction reaction fo11ows the theoretical curve. Although Pronase accepts a wide variety of peptides, substrate specificity of the enzyme may become problematic if highly diverse libraries are studied. A dipeptide which is not cleaved by Pronase would be retained in the reaction mixture even if it did not bind to carbonic anhydrase. One way ta alleviate this problem would be to use a mixture of enzymes with a wide range of specificities. Altematively, it is important to note that the degradation reaction is not limited to enzymic processes. Other chernical degradation methods can be envisioned, depending upon the type of library being studied. For

(15) The reaction mixture will contain the products of the degradation reactions. However, in most cases, this method will be applied to combinatorial libraries, and as sucb, the degradation products will often be the common starting materials used to make the library members. Thus, only a limited number of degradation products will be produced.


a) 5

4

Ratio 3

4a/4b 2

• • O+------r------r------r------r-----~

o 0.2 0.4 0.6 0.8

Conversion

ARTICLES

b) 10

8

Ratio 6

4a/4c 4

2

0

0 0.2 0.4 0.6 0.8

Conversion Figure 9. The graph shows theoretical and experimental ratios for the screening experiments. Theoretical lines are shown as smooth \ines. The S values correspond to the ratios of the experimentally determined binding constants. The data points show the experimentally deteITnined ratios at different conversions for a) 4a/4b (cf. Figure 6) and b) 4a/4c (cf. Figure 7).

example, a library based on disulfide exchange could be degraded by adding a reducing agent (e.g., a phosphine) to cleave any unbound disulfides. Altematively, physical methods for removal of unbound inhibitors (e.g., adsorption to a solid phase, extraction) should accomplish the same effect as a chemical degradation.

Another potential limitation of this screening method, and indeed for methods based on the dynamic combinatoriallibrary technique, is the need for stoichiometric amounts of the target. The initial experiments reported here used large amounts of carbonic anhydrase (100-500 mg/experiment) as we expect to apply it to a dynamic library process where the best inhibitor will actually be isolated and characterized. However, for purely analytical screening purposes, the methods can easily be scaled down using smaller compartments, assurning that more sensitive analytical tools are used (e.g., mass spectroscopy). These modifications could reduce the amount of target needed to <0.1 mg/experiment, an amount that is easily accessible for targets that have been cloned and overexpressed.

In conclusion, we have developed a method for screening mixtures of compounds against a therapeutic target which readily identifies the best binder in a library. The method works by selectively degrading the poorer inhibitors with an enzyme. This results in a significant improvement in the ability to distinguish between inhibitors which have very close binding constants. We plan to extend this method to dynamic libraries with the goal of improving the enhancement observed in synthesis of good inhibitors in the presence of a therapeutic target.

Experimental Section

General Experimental p-Nitrophenyl acetate (PNPA), carbonic anhydrase (CA, from bovine erythrocytes, a mixture of isozymes, C-3934) and proteases were purchased from Sigma unless otherwise noted and used without further purification. HPLC analyses were conducted using a Phenomenex-Cs reversed phase HPLC column (10 mm x 250 mm) with detection at 220 mn, unless otherwise noted. Elemental analyses were obtained from Quantitative Technologies Inc., Whitehouse, NJ. High-resolution mass spectra were obtained from Université de Sherbrooke, Sherbrooke, QC.

4'-Sulfonamidophenylalanine (1). N-Acetylphenylalanine (37.7 g, 178 mmol, 1 equiv) was added in portions over a I-h period to neat chlorosulfonic acid (110 mL, 1.65 mol, 9.5 equiv) at -10 OC. The resulting yellow solution was stirred at -10 oC for 2.5 h, at 25 oC for 2.5 h, and then heated to 60 oC until gas evolution had ceased (approximately 12 hl. The resulting orange solution was cooled to 0 oC and poured carefully onto 750 mL of ice (Caution: exotherrn!).

The resulting mixture was extracted with ethyl acetate (3 x 1 L), and the combined organic layers were dried over Na2S04, filtered, and concentrated in vacuo to afford the sulfonyl chloride (45.1 g, 83%) as an orange solid which was used immediately without further purification. IH NMR «CD3hSO) Ô 8.26-8.21 (d, tH, J = 8.5 Hz), 7.55 (d, 2H, J = 6.9 Hz), 7.22 (d, 2H, J = 6.8 Hz), 4.49-4.34 (m, tH),3.13-3.00 (dd, tH, J = 14.4 and 6.8 Hz), 2.92-2.79 (dd, tH, J = 11.0 and 10.2 Hz), 1.80 (s, 3H).

The sulfonyl chloride was dissolved in 28% NH40H (240 mL), and the resulting solution was heated at reflux for 3 h. After cooling to 0 oC, the solution was acidified to pH 1 by addition of 3 M H2S04 (ca. 200 mL) and extracted with ethyl acetate (3 x 500 mL). The combined organic extracts were dried over Na2S04, filtered, and concentrated in vacuo to afford the sulfonamide (29.9 g, 71 %) as a white solid. The N-acetyl sulfonamide could not be purified to homogeneity by either chromatography or recrystallization. IH NMR «CD3hSO) Ô 8.29-8.24 (d, tH, J = 8.5 Hz), 7.77 (d, 2H, J = 3.9 Hz), 7.45 (d, 2H, J = 6.9 Hz), 7.33 (s, 2H), 4.53-4.41 (m, tH), 3.20-3.09 (dd, IH, J = 14.2 and 6.8 Hz), 3.01-2.87 (dd, tH, J = 11.2 and 10.1 Hz), 1.80 (s, 3H).

A suspension of the sulfonamide (20.0 g, 69.9 mmol, 1 equiv) in distilled water (300 mL) was adjusted to pH 5.00 with LiOH (900 mg). A 0.25 M solution of Na2HP04 (85 mL) was uSed to raise the pH to 7.50. Acylase 1 from hog kidney (200 mg, 17.8 U/mg, 3560 U) was added as an aqueous solution (12 mL), and the resulting solution was stirred at 21°C for 70 h. The solution was then acidified to pH 1.0 with 3 M H2S04 and extracted with ethyl acetate (3 x 500 mL); the organic layer was then dried with anhydrous sodium sulfate and concentrated in vacuo to afford 2.28 g (II %) of the sulfonamide starting rnaterial. The aqueous layer was neutralized with 2 M NaOH and concentrated. The solution was then applied to an Amberlite 120(plus) acidic ion-exchange column. The column was rinsed with water until the eluent was at pH 6.0, and then it was rinsed with 0.50 M ~OH solution until the eluent became basic. The basic wash was concentrated in vacuo and recrystallized from water to afford the provided 4'sulfonamidophenylalanine as a white solid (11.60 g, 68%).IH NMR (D20IDCI) Ô 7.62 (d, 2H, J = 8.1 Hz), 7.26 (d, 2H, J = 8.1 Hz), 4.14 (t, tH, J = 6.8 Hz), 3.19-3.12 (dd, tH, J = 14.6 and 5.7 Hz),3.08-3.01 (dd, tH, J = 14.4 and 6.9 Hz). l3C NMR (D20IDCI) Ô 170.73, 140.451, 139.49, 130.38, 126.55,53.49,35.19. FABMS in saturated NaCI mlz 267 (M + Na, C9H12N20 4SNa requires 267.)

N-Ethoxycarbonyl-4'-sulfonamidophenylalanine (2). Ethyl chloroforrnate (398 JlL, 4.17 mmol, 1.10 equiv) was added to a two-phase mixture of 4'-sulfonamidophenylalanine (925 mg, 3.73 mmol, 1 equiv) in 1,4-dioxane (25 mL) and saturated NaHC03 solution (25 mL) at 0 oC, and the resulting solution was stirred for 6 h at 0 oC. The mixture was extracted with ethyl acetate (100 mL), and the aqueous layer was acidified to pH 1 by addition of 2 M HCI (ca. 20 mL) and then extracted with ethyl acetate (3 x 50 mL). Latter organic extracts were combined, dried over Na2S04, filtered, and concentrated in vacuo to afford the

J. AM. CHEM. SOC •• VOL. 124, NO. 20, 2002 5699

ARTICLES

ethyl carbamate (879 mg, 83%) as an analytically pure oil. IH NMR «C03hCQ) à 7.85 (d, 2H, J = 7.1 Hz), 7.51 (d, 2H, J = 6.9 Hz), 6.54 (s, 2H), 6.45 (d, IH, J = 6.7 Hz), 4.62-4.45 (m, lH), 4.01-3.97 (q, 2H, J = 2.4 Hz), 3.41-3.28 (dd, lH, J = 11.3 and 4.0 Hz), 3.16-3.05 (dd, lH, J = 10.4 and 7.8 Hz), 1.13 (t, 3H, J = 6.0 Hz). HR-CIMS (mlz): [MH+] calculated for CI2H17N206S, 317.0807; found, 317.0817.

Et01C.(4'.SOzNH1)Phe-Gly.O-terl.butyl (3b). EDC'HCl (136 mg, 0.711 mmol, 1.10 equiv), HOBT (87.3 mg, 0.646 mmol, 1.00 equiv), and triethylamine (269 pL, 1.94 mmol, 3.00 equiv) were added to a solution of 2 (204 mg, 0.646 mmol, 1 equiv) in THF (3 mL) at 0 oc. Glycine tert-butyl ester'HCI (119 mg, 0.711 mmol, 1.10 equiv) was added, and the resulting solution was allowed to warm to 21°C while stirring for 13 h, at which point the bulk of the THF was removed by concentration in vacuo. The residue was dissolved in ethyl acetate (45 mL) and extracted with 0.1 M HCI (3 x 25 mL) and saturated NaRC03 solution (3 x 25 mL). The organic layer was dried over Na2S04, tiltered, and concentrated in vacuo. The solid residue was purified by mixed solvent recrystallization (ethyl acetatelhexanes) to afford 193 mg (70%) of 3b. IH NMR «CD3hCO) à 7.82 (d, 2H, J = 7.5 Hz), 7.63 (s, IH), 7.50 (d, 2H, J = 7.5 Hz), 6.52 (s, 2H), 6.40 (d, lH, J = 7.5 Hz), 4.50 (m, lH), 4.00-3.88 (m, 4H), 3.40 (dd, lH, J = 14.1 and 4.2 Hz), 3.02 (dd, lH, J = 13.5 and 9.9 Hz), 1.45 (s,9H), 1.12 (t, 3H, J = 6.9 Hz). 13C NMR «CD3hCO) à 171.48, 168.93, 156.33, 142.70, 130.06,126.18,81.04,60.48,55.92,41.79,37.83,27.52, 14.22. Analysis calculated for CI8H27N307S C, 50.34; H, 6.34; N, 9.78. Found: C, 50.33; H, 6.35; N, 9.73.

Et01C·(4'·SOlNHZ)Phe·Gly·OH (4b). TFA (7 mL) was added to a solution of 3b (175 mg, 0.409 mmol, 1 equiv) in CH2Ch (8 mL), and the solution was stirred for 25 min at 21°C under an atmosphere of argon. The solvents were removed in vacuo, and the residue was purified by recrystallization from acetone to afford 121 mg (79%) of 4b. IH NMR (C030D) à 8.55 (s, lH), 7.83 (d, 2H, J = 7.2 Hz), 7.46 (d, 2H, J = 7.2 Hz), 4.45-4.42 (m, lH), 4.02-3.98 (q, 2H, J = 6.8), 3.95-3.92 (m, IH), 3.32-3.25 (m, 2H), 2.97-2.89 (dd, lH, J = 13.5 and 9.9 Hz), 1.18-1.14 (t, 3H, J = 6.8). 13C NMR (CD3CD) à 173.0, 171.8,157.3,148.7,142.4,137.6,129.8,126.0,60.9, 56.0, 37.7,13.7. HR-ClMS (mlz): [MW] calculated for CI4H2oN3Û7S, 374.1022; found, 374.1030.

Measurement of Inhibition Constants. Kinetic constants for carbonic anhydrase (CA) were measured according to Pocker and Stone usingp-nitrophenyl acetate (PNPA) as the substrate.16 The CA-catalyzed hydrolysis of pNPA was followed spectrophotometrically at 25 oC in a 96-well microplate spectrophotometer by monitoring the appearance of p-nitrophenolate at 404 nm. The values of Km and V max were deterrnined by measuring the hydrolysis rate as a function of the pNPA concentration. To deterrnine the inhibition constants, the values of Km and V max were redetermined in the presence of varying amounts of inhibitor. Since the values of Km for pNPA increased in the presence of the inhibitor, but the values of Vrnax remained unchanged, we concluded that the inhibition is competitive. The concentration of inhibitor that increased the Km for pNPA by a factor of 2 is the inhibition constant. A typical procedure was to add CA solution (100.0 pL) with inhibitor to acetonitrile solution of pNPA (5.0 pL). In the assay solution, the concentration of inhibitor ranges from 0.0 to 6.0 pM, while the concentration of pNPA ranged from 0.2 to 2.5 mM. The microplate was shaken for 5 s before the fust reading and for 3 s between readings.

Selective Concentration of EtOC·Phe..·Phe (4a) over EtOC·PhePhe, (5) into a Compartment Containing Carbonic Anhydrase. A solution of 4a (2.9 mg, 6.3 pmol) and 5 (2.9 mg, 7.5 pmol) in 0.01 M aqueous phosphate buffer (pH 7.5, 40 mL) was divided into two equal portions. Carbonic anhydrase (0.20 g, approx. 6.7 pmol) was dissolved in the first portion, and the resulting solution (20.0 mL) was transferred to a dialysis bag (12 OOO-MW cutoff, Sigma 0-0655). This dialysis bag was suspended in the second portion, and the reaction vessel was shaken gently (200 rpm) at 30 oC. Aliquots were removed periodically

(16) Poeker, Y.; Stone, J. T. Biochemistry 1968, 7, 3021-3031.

5700 J. AM. CHEM. SOC •• VOL. 124, NO. 20, 2002

Cheeseman et al.

from each compartment, heated to 80 oC until a white precipitate forrned (~5 min), and centrifuged, and the supematant was filtered through a 0.22-pm pore tilter. The amount of dipeptides was measured by HPLC using a Zorbax C8 colurun and 40160/0.1 water/methanolltrifluoroacetic acid at 0.40 mL/min. After 12 h 88% of 4a (retention time 11.4 min) had accumulated inside the dialysis bag while only 42% of 5 (retention time 25.5 min) was found inside the bag.

Selective Concentration Of EtOC·Phe...Phe (4a) from a Mixture of EtOC·Phe. •• Leu (4c), EtOC·Phe..·Gly (4b), and EtOC·Phe-Phe (5) by Carbonlc Anhydrase. Dipeptides 4a (2.0 mg, 4.3 pmol), 4b (1.6 mg, 4.3 pmol), 4c (1.9 mg, 4. pmol) 4d (1.8 mg, 4.3 pmol), and 5 (1.7 mg, 4.3 pmol) were dissolved in 40 mL of 10 mM KH2P04 buffer, pH 7.5 containing 0.1 mg/mL penicillin G (to avoid bacterial growth). Carbonic anhydrase (CA) (0.29 g, 9.7 pmol, 0.45 equiv) was dissolved in 20 mL of this solution and placed in a dialysis bag (the bag was washed in ddH20 for 1 h, rinsed in EtOH once, and then washed again with ddH20). The bag was suspended in the remaining 20 mL of inhibitor solution in a lOO-mL container and shaken at 60 rpm on a three-dimensional orbital shaker at room temperature for 49 h. Samples (1 mL) were taken periodically from inside and outside the dialysis bag, heated in an 80 oC water bath for 5 min, and then centrifuged for 10 min. The supernatant was tiltered through a 0.22-pM sterile tilter. The supematant (700 pL) was added to MeOH (300 pL) to forrn the HPLC sample (30% MeOH, 70% aqueous). The sample was run on a Phenomenex C8 reverse phase colurun under the following conditions: 0-15 min 30% MeOH, 70% H20, 15-60 min 37% MeOH 63% H20, 60-90 min 62% MeOH, 38% H20. The peak areas were monitored: Phes.Gly: 7.9 min, PhesaPro: 17.6 min, Phe •• Leu: 54.5 min, Phes.Phe: 60.0 min, PhePhe: 69.5 min. The percentages are accurate to ±2%. Ali nonsterile apparatus used was autoclaved prior to use to avoid bacterial growth.

Screening of Proteases for the Hydrolysis of EtOC·Phe·Phe Dipeptide (4a). The protease to be screened (0.1 mg) was added to a solution of 4a (1.0 mg, 2.2 mol) in 0.01 M aqueous phosphate buffer (pH 7.5). The solution was kept at 30 oC, and aliquots were removed periodically, worked up as above, and analyzed by HPLC.

Selective Protection of Inhibitors from Hydrolysis by Carbonic Anhydrase. A solution of 4a (3.0 mg, 6.5 pmol) and 5 (2.8 mg, 7.3 IImol) in 0.01 M aqueous phosphate buffer (pH 7.5, 40 mL) was divided into two equal portions. Carbonic anhydrase (0.20 g, approx 6.7 IImol) was dissolved in the first portion, and the resulting solution (20.0 mL) was transferred to a dialysis bag (12 OOO-MW cutoff, Sigma 0-0655). Pronase from S. griseus (Sigma P-5147, 4 mg) was dissolved in the second portion, and the dialysis bag was then suspended in the resulting solution. The reaction vessel was then shaken gently (200 rpm) at 30 oC, and aliquots were removed periodically from each compartment, worked up as above, and analyzed by HPLC. After 30 min, neither substrate was detectable in the solution outside the dialysis bag. Inside the dialysis bag, 78% of 5 had hydrolyzed, while only 6% of 4a had hydrolyzed after 6 h. In a control experiment containing no carbonic anhydrase, inside the dialysis bag, 76% of 4a and 80% of 5 had hydrolyzed after 6 h.

Selective Binding of EtOC·Phe..·Phe (4a) over EtOC·Phe..·Leu (4c). A solution of 4a (3.3 mg 7.1 mol) and 4c (3.6 mg, 8.4 mol) in 0.01 M aqueous phosphate buffer (pH 7.5, 40 mL) was divided into two equal portions. Carbonic anhydrase (0.20 g, approx 6.7 IImol) was dissolved in the first portion, and the resulting solution (20.0 mL) was transferred to a dialysis bag (12 OOQ-MW cutoff, Sigma 0-0655). This dialysis bag was suspended in the second portion, and the reaction vessel was shaken gently (200 rpm) at 30 oC. Aliquots were removed periodically from each compartment, worked up as above, and analyzed by HPLC using a Zorbax C8 colurun. After 12 h 98% of 4a had accumulated inside the dialysis bag, while only 60% of 4c was found inside the bag.

Hydrolysis of EtOC·Phe..·Gly (4b) and EtOC·Phe..·Phe (4a) in the Presence of Carbonic Anhydrase. PhesaPhe 4a (2.0 mg, 4.3 pmol)


and Phe",Gly 4b (1.6 mg, 4.3 .umol) were dissolved in 20 mL of 10 mM KH2P04 buffer, pH 7.5. Carbonic anhydrase (CA) (0.4090 g, 13.6 .umol, 1.60 equiv) was dissolved in this solution and placed in a dialysis bag (the bag was washed in ddH20 for 1 h, rinsed in EtOH once, and then washed again with ddH20). The bag was suspended in 20 mL of the phosphate buffer containing Pronase from S. griseus (5.0 mg, 0.01 equiv) in a 150-mL beaker and shaken at 150 rpm at 30 oc for 313 h. Samples (1 mL) were taken periodically from inside, worked up as above, and analyzed by HPLC. After 193 h, only 71% of 4a had hydrolyzed, while 93% of 4b had hydrolyzed.

Hydrolysis of EtOC·Phes.·Leu (4c) and EtOC·Phe,...Phe (4a) in the Absence of Carbonic Anhydrase. A solution of 4a (2.9 mg 6.3 mol) and 4c (2.4 mg, 5.6 mol) in 0.01 M aqueous phosphate buffer (pH 7.5, 40 mL) was divided into two equal portions. The first portion was transferred to a dialysis bag (12 OOO-MW cutoff, Sigma D-0655). Pronase from S. griseus (Sigma P-5147, 4 mg) was dissolved in the second portion, and the dialysis bag was then suspended in the resulting solution. The reaction vessel was then shaken gently (200 rpm) at 30 oC, and aliquots were removed periodically from each compartment, worked up as above, and analyzed by HPLC using a Zorbax C8 column. After 30 min, neither substrate was detectable in the solution outside the dialysis bag. After 8 h, 86% of 4a and 88% of 4c inside the dialysis bag had hydrolyzed.

Hydrolysis of EtOC·Phe,...Leu (4c) and EtOC·Phe, •• Phe (4a) in the Presence of Carbonic Anhydrase. A solution of 4a (2.9 mg, 6.3 mol) and 4c (2.4 mg, 5.6 mol) in 0.01 M aqueous phosphate buffer (pH 7.5, 40 mL) was divided into two equal portions. Carbonic anhydrase (0.14 g, approx 4.7 .umol) was dissolved in the first portion, and the resulting solution (20.0 mL) was transferred to a dialysis bag (12000-MW cutoff, Sigma D-0655). Pronase from S. griseus (Sigma P-5147, 4 mg) was dissolved in the second portion, and the dialysis

ARTICLES

bag was then suspended in the resulting solution. The reaction vessel was then shaken gently (200 rpm) at 30 oC, and aliquots were removed periodically from each compartrnent, worked up as above, and analyzed by HPLC. After 6 h, 93% of 4c had hydrolyzed, while only 58% of 4a was hydrolyzed.

Hydrolysis of EtOC·Phe, •• Phe (4a), EtOC·Phe .. ·Gly (4b) EtOC· Phe,.·Leu (4c), EtOC·Phe,.·Pro (4d) and EtOC·Phe-Phe (5), in the Presence of Carbonic Anhydrase. Phes.Phe 4a (2.0 mg, 4.3 .umol), Phes.Gly 4b (1.6 mg, 4.3 .umol), Phe.,.Leu 4c (1.9 mg, 4.3 .umol), Phes.Pro 4d (1.8 mg, 4.3 .umol), and PhePhe 5 (1.7 mg, 4.3 .umol) were dissolved in 20 mL of 10 mM KH2P04 buffer, pH 7.5. Carbonic anhydrase (CA) (0.7670 g, 25.6 .umol, 1.20 equiv) was dissolved in this solution and placed in a dialysis bag (the bag was washed in ddH20 for 1 h, rinsed in EtOH once, and then washed again with ddH20). The bag was suspended in 20 mL of the phosphate buffer containing Pronase from S. griseus (4.9 mg, 0.01 equiv) in a 150-mL beaker and shaken at 150 rpm at 30 oC for 193 h. Samples (1 mL) were taken, worked up as above, and analyzed by HPLC. Data for this experiment is shown in Figure 8.

Acknowledgment. We thank Merek Frosst Ine., BioChem Pharma Ine., Boehringer Ingelheim Ine., AstraZeneea Ine., and FCAR for generous funding of this researeh. J.D.C. thanks NSERC for a postgraduate fellowship. The reviewers are aeknowledged for helpful eomments and suggestions.

Supporting Information Available: A derivation of eq 3 and eharaeterization data for 3a,c,d and 4a,c,d (PDF). This material is available free of eharge via the Internet at http://pubs.aes.org.

JA017099+

J. AM. CHEM. SOC •• VOL. 124, NO. 20, 2002 5701

Supporting Information

Amplification of Screening Sensitivity Through Selective Destruction. Theory

and Screening of a Library of Carbonic Anhydrase Inhibitors.

Ronghua Shu, Andrew D. Corbett, Jeremy D. Cheeseman, Jonathan Croteau, James L. Gleason*

and Romas J. Kazlauskas*

Department of Chemistry, McGill University, 801 Sherbrooke Street West,

Montréal, QC, Canada, H3A 2K6.

Derivation of the equation for an equilibrium ratio of products in adynamie combinatorial

library.

Consider a common starting material, SM, in equilibrium with two inhibitors, A and B,

which can each bind reversibly to a target, T, to form complexes TVA and TVB

KsA KaA A+T T·A (1 )

SM KsB

B+T KaB

T·B (2)

Equations 3-5 define respectively KSA' the equilibrium constant for synthesis of inhibitor A,

KaA' the equilibrium constant binding of inhibitor A to the target, and [AT]' the total ofbound

and unbound forms of A.

K = [A] sA [SM]

(3)

K = [T-A] aA [T][A]

(4)

[Ar]=[A]+[T-A] (5)

Combining these equations and solving for [AT] yields equation 6.

(6)

Dividing this equation by an analogous equation for [BT] yields equation 7.

[Ar] _ KsAKaA[T] + 1) [BT ] - K sB (KaB [T]+l)

(7)

Under typical conditions (tight binding and an excess of target at a high concentration), the

product of the association constant (KaA or KaB) and free target concentration is much greater

than one so the equation simplifies to equation 8, below. l

[Ar] = KsA x KaA [BT ] K sB X K aB

Characterization Data:

(8)

Et02C-(4'-S02NH2)Phe-Phe-O-t-butyl (3a). Prepared as 3b to afford to afford 173 mg (72%).

IH NMR (CDCh) Ô 7.80 (d, 2H, J= 7.8 Hz), 7.33-7.24 (m, 5H), 7.09 (d, 2H, J= 6.0 Hz), 6.41

(d, 1H,J= 5.7 Hz), 5.19 (d, 1H,J= 6.1 Hz), 4.96 (s, 2H), 4.75-4.61 (m, 1H), 4.52-4.38 (m, 1H),

4.10-4.03 (q, 2H, J = 6.9 Hz), 3.20-2.96 (m, 4H), 1.39 (s, 9H), 1.21 (t, 3H, J = 6.8 Hz). 13C NMR

(CDCh) Ô 170.34, 170.11, 142.21, 140.84, 136.03, 130.38, 129.65, 128.67, 127.31, 126.96,

82.93, 77.44, 61.76, 53.86, 38.40, 38.21, 28.16, 14.69. Ana1ysis calculated for C2sH33N307S C,

57.79; H, 6.40; N, 8.09. Found: C, 57.71; H, 6.34; N, 7.96.

Et02C-(4'-SÛ2NHüPhe-Leu-O-t-butyl (3e). Prepared as 3b to afford 227 mg (78%). IH NMR

«CD3)2CO) Ô 7.81 (d, 2H, J = 8.4 Hz), 7.55 (m, 2H), 7.48 (d, 2H, J= 8.1 Hz), 6.51 (m, 1H),

6.31 (m, 1H),4.50(m, 1H),4.39 (m, 1H),4.00-3.95,(q,2H,J=5.7),3.32-3.26 (dd, 1H,J= 13.8

and 3.9 Hz), 3.05-2.97 (dd, 1H, J= 13.6 and 9.6 Hz), 1.75-1.64 (m, 2H), 1.61-1.56 (m, 2H), 1.45

(s,9H), 1.12 (t, 3H, J= 7.4 Hz), 0.94-0.90 (m, 7H). I3C NMR «CD3)2CO) Ô 171.8, 171.0, 156.4,

1 If the binding is not tight or the target is not in excess at high concentration, then the

concentrations of the inhibitors will be more similar than that discussed in the text above and it

will be even harder to distinguish which is the better inhibitor.

142.5, 130.1, 126.2,81.0,60.5,55.7,51.6,41.1,37.8,27.5,24.8,22.5, 21.3, 19.5, 14.2. Analysis

calculated for C22H3sN307S C, 54.42; H, 7.26; N, 8.65. Found: C, 54.28; H, 1.27; N, 8.46.

Et02C-(4'-SÜzNH2)Phe-Pro-O-t-butyl (3d). Prepared as 3b to afford 312 mg (67%). IH NMR

«CD3)2S0) ù 7.69 (d, 2H, J= 7.6 Hz), 7.48 (d, 2H, J= 7.0 Hz),7.36 (d, 2H, J= 7.2 Hz), 7.28 (s,

2H), 4.36 (m IH), 4.19 (m, IH), 3.85 (t, 2H, J = 6.7 Hz), 3,65 (m, 2H), 2.97-2.94 (dd IH, J =

11.7 and 4.2 Hz), 2.82-2.74 (dd IH, J= 11.1 and 9.2 Hz), 2.14 (m, IH), 1.90 (m, 2H), 1.77 (m,

IH), 1.35 (s, 9H), 1.04 (t, 3H, J = 7.3 Hz). 13C NMR «CD3)2S0)

ù 171.6,170.4,156.8,143.0,142.7,130.5,126.1,81.0,60.5, 60.1, 54.7, 47.1, 36.4, 29.2, 28.3,

25.3, 15.2. Analysis calculated for CI8H27N307S C, 50.34; H, 6.34; N, 9.78. Found: C, 50.33;

H, 6.35; N, 9.73.

Et02C-(4'-S02NH2)Phe-Phe-OH (4a). Prepared as 4b to afford 270 mg (68 %). IH NMR

(CD30D) Ù 8.19 (d, IH,J= 9.3 Hz), 7.80 (d, 2H,J= 8.4 Hz), 7.38 (d, 2H,J= 9.3 Hz), 7.27-7.21

(m, 5H), 7.07 (d, IH, J= 8.7 Hz), 4.68-4.63 (m, IH), 4.40-4.35 (q, 2H, J =6.1 Hz), 3.24-3.18

(dd, IH, J=13.9 and 5.2 Hz), 3.18-3.11 (dd, IH, J=14.3 and 5.5 Hz), 3.04-2.97 (dd, IH, J= 13.9

and 8.2 Hz), 2.88-2.80 (dd, IH, J= 13.9 and 9.7 Hz), 1.18-1.14 (t, 3H, J= 6.9 Hz). 13C NMR

(CD3CD) Ù 173.1, 172.3, 157.2, 142.3, 137.0, 129.8, 129.2, 128.3, 126.6, 126.0,60.9,55.9,53.9,

37.6, 37.2, 13.7. HR-CIMS (mlz): [MH+] calculated for C2IH26N307S, 464.1491; found,

464.1501.

Et02C-(4'-S02NH2)Phe-Leu-OH (4c). Prepared as 4b to afford 174 mg (80 %). IH NMR

(CD3CD) Ù 7.82 (d, 2H, J= 8.4 Hz), 7.68 (d, IH, J= 8.1 Hz), 7.45 (d, 2H, J= 8.1 Hz), 4.47-

4.42 (m, 2H), 4.01-3.95 (q, 2H, J= 6.4 Hz), 3.31-3.29 (m, IH), 3.25-3.3.19 (dd, IH, J= 13.9 and

4.9 Hz), 2.95-2.87 (dd, IH, J= 13.9 and 9.7 Hz), 1.71-1.62 (m, 2H), 1.18-1.13 (t, 3H, J= 7.1

Hz), 0.97-0.91 (m,6H). 13C NMR (CD3CD) Ù 174.6,172.7, 157.2, 142.3, 129.8, 126.7, 126.0,

60.9,55.8,50.9,40.4,37.7,24.8,22.2,20.6, 13.7. HR-CIMS (mlz): [MH+] calculated for

CI8H28N307S, 430.1648; found, 430.1654.

Et02C-(4'-S02NH2)Phe-Pro-OH (4d). Prepared as 4b except recrystallized from iso-propanol /

hexanes to afford 94.5 mg (55 %). IHNMR (CD30D) Ù 7.83 (d, 2H,J= 8.1 Hz), 7.50 (d, 2H, J

= 8.4 Hz), 7.25 (d, IH,J= 3.9 Hz), 4.66-4.61 (dd, IH,J= 8.8 and 5.5 Hz), 4.47-4.43 (dd, IH,J

= 8.4 and 3.9 Hz), 4.02-3.95 (q, 2H, J= 7.2 Hz), 3.80-3.75 (m, IH), 3.56-3.51 (m, IH), 3.32-3.29

(m, IH), 3.21-3.14 (dd, IH, J= 13.9 and 5.2 Hz), 2.96-2.89 (dd, IH, J= 13.8 and 8.7 Hz), 2.27-

2.21 (m, IH), 2.05-1.96 (m, 2H), 1.19-1.16 (t, 3H,J=7.2 Hz. 13CNMR(CD3CD) Ô 174.0,171.1,

157.3, 141.9, 130.1, 129.1, 128.2, 126.0,60.8,59.4,53.9,37.0,29.0,24.6, 13.7. HR-CIMS

(m1z): [MH+] ca1culated for C17H24N307S, 414.1335; found, 414.1325.

?t!smmunications

© 2004 Wiley·VCH Verl.g GmbH & Co. KG.A, Weinheim

Selective Binding '~p

Pseudodynamic Combinatorial Libraries: A Receptor-Assisted Approach for Drug Discovery**

Andrew D. Corbett, Jeremy D. Cheeseman, Romas 1 Kazlauskas, * and James L. Gleason*

Emerging methods of combinatorial chemistry incorporate receptor assistance to combine synthesis and screening.[I] Stoichiometric binding to a receptor alters either the thermo-

[*) A. D. Corbett, J. D. Cheeseman, Prof. R. J. Kazlauskas,+ Prof. J. L Gleason Department of Chemistry, McGili University 801 Sherbrooke St. West, Montréal, QC, H3A 2 K6 (Canada) Fax: (+ 1) 514·398·3797 E·mail: [email protected].

[email protected]

ri Current address: University of Minnesota Department of Biochemistry Molecular Biology and Biophysics and The Biotechnology Institute 1479 Gortner Avenue, Saint Paul, MN 55108 (USA) Fax: (+ 1) 612·625·5780

[**) The authors thank FQRNT for support ofthis research through the Soutien aux Equipes and VRQ programs, Prof. Sidney M. Hecht (University ofVirginia) for the discussion that led to this research, and Prof. Sijbren Otto, Prof. Jeremy K. M. Sanders, and Prof. K. Barry Sharpless for reading the manuscript.

• Supporting information for this article is available on the WWW under http://www.angewandte.org or from the author.

Angew. Chem. Inl. Ed. 2004, 43, 2432 -2436

dynamics or kinetics of library synthesis. Dynamic combinatorial libraries[21 use a thermodynamic approach where binding shifts a synthetic equilibrium to increase the amounts of the best-binding compounds, in accordance with LeChâtelier's principle. These libraries usually identify the library members that bind the tightest, but sorne experimental conditions can give small or misleading changes in concentrationYl An alternative method, receptor-accelerated synthesis, uses a kinetic approach.f41 Starting components that bind to the receptor can couple to create a new, tight-binding compound. The receptor accelerates the coupling of the bcttcr-fitting ~tnrting çomponçnt~ n~ n rc~ult of thcir proximity, but requires that both components bind tightly to the receptor. Here we demonstrate a new method, a pseudodynamic library, which adds a kinetic contribution to traditional dynamic libraries to dramatically increase the selectivity.

A pseudodynamic combinatorial library combines an irreversible synthesis of library members with an irreversible destruction step. Those library members that bind to the receptor are protected from destruction. Subsequent synthesis reuses fragments from destroyed library members, thus amplifying the amounts of the better binders at the expense of the lesser ones. The separate irreversible synthesis and destruction steps allow adjustment to optimize both the amplification and selectivity.

We developed a pseudodynamic library of eight dipeptides to identify the best inhibitor of carbonic anhydrase (CA). Carbonic anhydrase, a zinc metalloenzyme, is a therapeutic target for glaucoma and is inhibited by aromatic sulfonamides, which coordinate to the zinc ion. Four of the eight dipeptides in our library contain 4'-sulfonamidophenylalanine (Phesa, 1), and thus should bind to CA, while the remaining four contain only Phe and serve as negative controls. The irreversible synthesis of dipeptides used a

solid-supported coupling of activated esters with an amino acid in aqueous solution (Scheme 1). TentaGel-supported tetrafluorophenyl active esters react cleanly with free amino

Pronase

,x. = so,NW, 4X-H

Series: a R' " CH"ph. R·" H bR' =H,R'=H ç R' .. CH,CH(CH,h. R' '" H d R' R' '" {CH,l,

Scheme 1. Creation of a pseudodynamic library of dipeptides.

acids in water under alkaline (pH 8-10) conditions to form dipeptides.[51 A nonselective protease from Streptomyces griseus (Pronase) destroyed these dipeptides by catalyzing their irreversible hydrolysis.f61

The pseudodynamic library was prepared in a threechambered reaction vessel formed by suspending two dialysis bags in a surrounding solution (Figure 1). One dialysis bag (the synthesis chamber) contained the active esters and the other dialysis bag (hydrolysis chamber) contained the protease, while the surrounding solution (screening chamber) contained the carbonic anhydrase. Adding nucleophiles 1 and 2 to the synthesis chamber generated the dipeptide library. These dipeptides diffused into the surrounding solution where they could bind to carbonic anhydrase and then diffuse into the hydrolysis chamber where Pronase cleaved them. This

Figure 1. Schematic representation of the pseudodynamic combinatoriallibrary experiment. Reaction of two free amino acids (Phe .. (1) and Phe (2» with four solid-supported active esters (N-EtO,C-Phe, N-EtO,C-Gly, N-EtO,C-Leu, and N-EtO,C-Pro) creates an eight-member library. MWCO = molecular-weight eut off.

Angew. Chem. Inl. Ed. 2004, 43, 2432 -2436 www.angewandte.org © 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 2433

cleavage regenerated 1 and 2, which could diffuse back into the synthesis chamber to repeat the cycle. This arrangement prevented Pronase-catalyzed destruction and active-estermediated modification of the receptor (CA) and also permitted periodic replenishment of the activated ester to regulate the rate of synthesis.

The experiments used four active ester resins derived from N-EtOzC-Phe, N-Et02C-Gly, N-Et02C-Leu, and NEtOzC-Pro (0.8 equiv each), nucleophiles 1 and 2 (6.4 equiv each), carbonic anhydrase (28 J.lffiol, 1 equiv), and Pronase (25 mgmL-1

). The large amount of Pronase made diffusion across the dialysis membrane the rate-limiting step for hydrolysis; hence, ail the dipeptides were cleaved at similar rates in spite of the substrate selectivity of Pro nase. Periodic addition of fresh portions of active ester resin (defined as the cycle time) regulated the overall rate of library synthesis. We conducted three experiments with this system using cycle times of 8, 12, and 16 h. HPLC analysis of aliquots from the screening chamber showed the progress of the experiments (Figure 2).

Tho control experiments established, first, that the synthetic process afforded all the expected dipeptides and, second, that the sulfonamide-containing dipeptides inhibited carbonic anhydrase. Combining equal amounts of the four active esters with Phesa (1) as the nucleophile produced four dipeptides 3a-d in a ratio of 18:44:15 :23. Not surprisingly, the coupling of 1 with the less-hindered glycine ester to produce 3b was more efficient than with the more-hindered phenylalanine, leucine, or proline esters. In spite of these differences, ail four dipeptides formed in significant amounts. The use of phenylalanine as the nucleophile gave similar results. For the second control experiment, ail eight dipeptides were prepared individually and their ability to inhibit the CA-catalyzed hydrolysis of p-nitrophenyl acetate (Scheme 2) was measured. As expected, the sulfonamide-containing dipeptide competitively inhibited this hydrolysis, with inhibition constants of 1.1-8.7 J.IM, while the non-sulfonamide dipeptides showed no detectable inhibition. Dipeptide 3d was the best inhibitor,

with an inhibition constant of 1.1 J.IM, and dipeptide 3e the next best, with an inhibition constant of 2.5 J.IM. Compound 1 also inhibits CA (Ki = 13 J.IM), but approximately tenfold less effectively than the tightest binding dipeptide (3d).

In the first pseudo-dynamic experiment (8-hour cycle, Figure 2a), the cycle time was too short for the destruction reaction to rem ove the less-effective inhibitors. During the first four hours of each cycle, the screening chamber contained ail eight dipeptides, thus indicating that ail eight had formed as expected. At the end of each 8 h cycle, prior to the next addition of active ester, the hydrolysis had removed the four non-sulfonamide dipeptides, thus leaving only the four sulfonamide dipeptides. At the end of six cycles of active ester addition, dipeptide 3b was present in the highest amount (58% yield, relative to CA), followed by 3d (33%), 3 e (27 %), and 3 a (8 %). These relative amounts differ from their relative binding constants. The higher yield of 3 b instead reflects its more favorable rate of synthesis. In addition, the sum of ail the sulfonamide dipeptides at 48 h was greater than the amount of target (126 % yield). This high yield shows that unbound dipeptides remained and that the destruction reaction had not had enough time to distinguish between the different sulfonamide inhibitors.

Lengthening the cycle time from 8 h to 12 h yielded the best three inhibitors, in relative amounts in the order of their inhibition constants (Figure 2b). Although sulfonamides 3bd were present in high concentrations early in the experiment, the concentration of these weaker binding dipeptides had diminished substantially at the end of four cycles. The tightest binding dipeptide (3d) was present in the highest amount (15% yield relative to CA), followed by 3e (5%) and 3b (1.5 %). Notably, the ratio at the end of the experiment (10.1:3.5:1) exceeded the ratio of their binding constants (5.1:2.2:1). None of the weaker binding 3a or of the nonsulfonamide dipeptides remained at the end of the experiment.

The selectivity of the dynamic process improved even further upon extending the cycle time to 16 h (Figure 2c). The

;$:

S02NH

2 SO NH 1 SO NH

;$:

2 2 SO NH

;$:

22 1 ;$:22 o '" l '" 1 H Il 0 '" H 0 0 '"

..... N~ OH H Il N Il OH Et02C Il Et02C i ~ ..... N~ OH Eto,C ..... Y'N N-...../"-- OH

0:= 0 Et02C ~ ---,/ H 0 '-) ~

1 0 1 0 ~ 3a: El02CPhePhe.. 3b: Et02CGlyPhesa 3e: Et02CLeuPhe.. 3d: Et02CProPhesa

Ki = 8.7 IJM Ki = 5.61JM Ki = 2.51JM Ki = 1.1 IJM

;$:

1 H ,;? H ;$:1 H ;$:H ..... ~J '" OH H w;$1 ~ W '" OH Et02C, W : 1 Et02C , N N~ OH EtO C ..... ~N N-...../"-- OH

V= H 0 Et02C'" ~ 2 --/ H 0 \.J ~

1 0 1_ 0 ~ 4a: EtO,CPhePhe 4b: Et02CGlyPhe 4e: EtO,CLeuPhe 4d: Et02CProPhe

Ki» 1.0 mM Ki» 1.0 mM Ki» 1.0 mM Ki» 1.0 mM

Scheme 2. Competitive inhibition constants of the library members for the CA·catalyzed hydrolysis of p·nitrophenyl acetate. The non-sulfonamide compounds showed no detectable inhibition at l mM.

2434 © 2004 Wiley·YCH Yerl.g GmbH & Co. KG.A. Weinheim www.angewandte.org Ange",. Chem. Int Ed. 2004. 43. 2432 -2436

a) 1.20

1.00

0.80

0.60 [3]1mM 0.40

0.20

10 20 30 40 SO tlh

b) 1.20

1.00

0.80

0.60

0.40

[31/mM 0.20

0.00 0 10 20 30 40 SO

tlh

c) 0.60

O.SO

0.40

0.30

0.20 (3)/mM

0.10

20 40 60 80 100

tlh----

Figure 2. Pseudodynamic library experiments. Concentrations of sulfo· namide containing dipeptides 3a (el, 3b ("'l, 3c (_l, and 3d (+l over the course of experiments: a) 8 h per cycle, bl 12 h per cycle, and cl 16 h per cycle.

initial synthesis during the first cycle favored dipeptide 3 b, the most rapidly synthesized dipeptide, but this dipeptide disappeared in later cycles where the main competition was between 3 d and 3 c, the tightest binding dipeptides. After four cycles (64 h), only these two remained and the ratio of their concentrations (13:1) was significantly higher than the ratio of their binding constants (2.3:1). The selectivity increased to > 100:1 in favor of the strongest binding dipeptide 3d after three more cycles. The yield was 29 % relative to the amount of CA and corresponded to 4 mg of dipeptide. Thus, adjusting the relative rate of the library synthesis and destruction optimized the selectivity so that only the best-binding dipeptide remained and was present in a good overall yield.

The selectivity in the pseudodynamic library is significantly greater than that in many traditional dynamic libraries. The optimum conditions produced only the single, tightestbinding dipeptide (> 100:1 selectivity), while a traditional approach would yield a mixture because the binding constants for the two tightest-binding dipeptides differed by only 2.3-fold. This higher selectivity greatly simplifies the analysis, as only one compound need be identified and characterized. The optimization of a pseudodynamic library arises through control of the relative rates of synthesis and destruction. We previously showed that a destruction reaction operating on a

An ewahdte g::'Chemie

static library in the presence of a receptor distinguishes between library members with very similar binding constants, selectively removing the weaker-binding species,!6] However, when selectivity arises from destruction alone, significant amounts of the best-binding library member must be destroyed to achieve high ratios of good binder to slightly poorer binder. This situation leaves only a small amount of the best binder for analysis. The high selectivity in pseudodynamic libraries also stems from the competition between binding to the receptor and destruction.

The iterative nature of the experiment also contributes to the high selectivity. Cleavage by Pronase has reduced the amounts of weak-binding dipeptides toward the end of each cycle, which leaves dipeptide 3 d as the major species bound to CA. The subsequent burst of synthesis produces a mixture of ail dipeptides which compete for the smaller amount of free target. Pronase then rapidly c1eaves al1 unbound species, which would consist of a higher proportion of weak binders. Following our static model, continued action of Pronase further increases the ratio in favor of the bound species, which results ultimately in high selectivity for the tightest-binding speices.

Our static model of pseudodynamic combinatorial libraries16] indicates that selectivity stems from the relative binding constants of the inhibitors, not their absolute affinity for the target. Thus, we expect that pseudodynamic combinatorial libraries will also work with even tighter-binding inhibitors, but would require longer cycle times to distinguish between these more tightly binding inhibitors. Indeed, we are currently expanding our studies to larger pseudodynamic libraries to discover such tighter-binding inhibitors.

Received: January 15, 2004 [Z53769] Published Online: March 31, 2004

eywords: combinatorial chemistry . drug design. enzyme inhibitors . kinetics . receptors

[1] al A. Ganesan, Angew. Chem. 1998, 110, 2989-2992; Angew. Chem. Int. Ed. Engl. 1998,37,2828-2831; b) 1. Huc, J.-M. Lehn, Actual. Chim. 2000, 51-54; cl C Karan, B. L Miller, Drug Discovery Today 2000, 5, 67 -75; dl R. Nguyen, 1. Huc, Comb. Chem. High Throughput Screening2001, 4, 53-74; el J.-M. Lehn, A. V. Eliseev, Seience 2001, 291, 2331-2332; fl O. Ramstrôm, J.M. Lehn, Nat. Rev. Drug Discovery 2002, 1, 26-36; gl S. Otto, R. L E. Furlan, J. K. M. Sanders, CurT. Opin. Chem. Biol. 2002, 6, 321- 327; hl O. Ramstrôm, T. Bunyapaiboonsri, S. Lohman, J.-M. Lehn, Biochim. Biophys. Acta 2002,1572,178-186.

[2] al 1. Huc, J.·M. Lehn, Proc. Natl. Acad. Sei. USA 1997, 94, 2106-2110; bl P. G. Swann, R. A. Casanova, A. Desai, M. M. Frauenhoff, M. Urbancic, U. Slomczynska, A. J. Hopfinger, G. C LeBreton, D. L Venton, Biopolymers 1997, 40, 617-625; cl B. Klekota, M. H. Hammond, B. L Miller, Tetrahedron Lett. 1997, 38,8639-8642; dl B. Klekota, B. L. Miller, Tetrahedron 1999, 55, 11687 -11697; O. Ramstrôm, J.-M. Lehn, ChemBioChem 2000, 1, 41-48; el C Karan, B. L Miller, 1. Am. Chem. Soc. 2001, 123, 7455-7456; f) R. J. Lins, S. L Flitsch, N. J. Turner, E. Irving, S. A. Brown, Angew. Chem 2002,114,3555-3557; Angew. Chem. Int. Ed. 2002,41,3405-3407; gl M. Hochgurtel, H. Kroth, D. Piecha, M. W. Hofmann, K. C Nicolaou, S. Krause, O. Schaaf, G. Sonnenmoser, A. V. Eliseev, Proc. Natl. Acad. Sei. USA 2002,

Angew. Chem. Int. Ed. 2004, 43, 2432 -2436 www.angewandte.org © 2004 Wiley·YCH Yerlag GmbH & Co. KGaA, Weinheim 2435

'CSf'mmunications

99,3382-3387; h) I. C. Choong, W. Lew, D. Lee, P. Pham, M. T. Burdet!, 1. W. Lam, C. Wiesmann, T. N. Luong, B. Fahr, W. L. DeLano, R. S. MeDowell, D. A. Allen, D. A. Erlanson, E. M. Gordon, T. O'Brien, J. Med. Chem. 2002, 45, 5005-5022; i) S. Otto, R. L. E. Furlan, 1. K. M. Sanders, Science 2002, 297, 590-593; j) D. A. Erlanson, 1. W. Lam, C. Wiesmann, T. N. Luong, R. L. Simmons, W. L. DeLano, I. C. Choong, M. T. Burdett, W. M. Flanagan, D. Lee, E. M. Gordon, T. O'Brien, Nat. Biotechnol. 2003,21,308-314; k) A. C. Braisted, J. D. Oslob, W. L. Delano, 1. Hyde, R. S. MeDowell, N. Waal, C. Yu, M. R. Arkin, B. C. Raimundo, J. Am. Chem. Soc. 2003,125,3714-3715.

[3] a) A. V. Eliseev, M. I. Nelen, J. Am. Chem. Soc. 1997,119, 1147-1148; b) 1. S. Moore, N. W. Zimmerman, Drg. Lett. 2000,2,915-918; e) Z. Grote, R. Seopelliti, K. Severin, Angew. Chem. 2003, 115,3951-3955; Angew. Chem. lnt. Ed. 2003,42,3821-3825.

[4] a) K. C. Nicolaou, R. Hughes, S. Y. Cho, N. Winssinger, C. Smethurst, H. Labischinski, R. Endermann, Angew. Chem. 2000,112,3981-3986; Angew. Chem. lnt. Ed. 2000,39,3823-3828; b) R. Nguyen, 1. Huc, Angew. Chem. 2001,113,1824-1826; Angew. Chem. lnt. Ed. 2001, 40, 1774-1776; c) W. G. Lewis, L. G. Green, F. Grynszpan, Z. Radie, P. R. Carlier, P. Taylor, M. G. Finn, K. B. Sharpless, Angew. Chem. 2002, 114, 1095-1099; Angew. Chem. lnt. Ed. 2002,41,1053-1057.

[5] A. D. Corbet!, J. L. Gleason, Tetrahedron Lett. 2002, 43, 1369-1372.

[6] 1. D. Cheeseman, A. D. Corbet!, R. Shu, 1. Croteau, J. L. Gleason, R. 1. Kazlauskas, J. Am. Chem. Soc. 2002,124,5692-5701.

2436 © 2004 Wiley·VCH Verlag GmbH & Co. KGaA, Weinheim www.angewandte.org Angow. Chom. 'nI. Ed. 2004, 43, 2432 -2436

pseudo-dynamic combinatorial libraries: a …digitool.library.mcgill.ca/thesisfile85059.pdf ·...

Documents