ISBN: 0-8247-1939-5

This book is printed on acid-free paper.

Headquarters
Marcel Dekker, Inc.
270 Madison Avenue, New York, NY 10016
tel: 212-696-9000; fax: 212-685-4540

Eastern Hemisphere Distribution

Marcel Dekker AG
Hutgasse 4, Postfach 812, CH-4001 Basel, Switzerland
tel: 41-61-261-8482; fax: 41-61-261-8896

World Wide Web

http://www.dekker.com

The publisher offers discounts on this book when ordered in bulk quantities. For more
information, write to Special Sales/Professional Marketing at the headquarters address
above.

Copyright © 2000 by Marcel Dekker, Inc. All Rights Reserved.

Neither this book nor any part may be reproduced or transmitted in any form or by
any means, electronic or mechanical, including photocopying, microfilming, and re-
cording, or by any information storage and retrieval system, without permission in
writing from the publisher.

Current printing (last digit):

10 9 8 7 6 5 4 3 2 1

PRINTED IN THE UNITED STATES OF AMERICA

Foreword: Chemistry Becomes an
Information Science

Nothing in recent years has had as great an impact on the process of drug
discovery as the rise of genomics and combinatorial chemistry. Nothing has
been more urgently needed. In the United States today it takes an average of
13 years and over $300 million to develop a drug. Although regulatory hurdles
account for a good portion of those costs, major difficulties lie in two other
areas.
First, there are not enough drug targets. There are about 6000 known
drugs, half of which hit human targets. However, those 3000 human-directed
drugs hit only about 500 targets, which means that less than 1% of the human
genome (estimated to contain 80,000 to 100,000 genes) has been exploited
pharmaceutically. It’s even worse for antimicrobial drugs. Consider antifungal
agents: almost every marketed antifungal agent hits one of a handful of targets
in the same metabolic pathway. Genomics, the science of identifying and se-
quencing all of the genes in an organism, is changing all this at an astonishing
rate. Soon we will have a plethora of targets, for pathogens and people. How-
ever, this only makes the second difficulty—that there aren’t enough drugs—
more acute.
Those 6000 known drugs fall into only around 300 chemical classes
(the exact figure depends on how one defines a ‘‘class’’). Recently, a ma-
jor pharmaceutical company screened its entire compound inventory—over
400,000 compounds developed over almost 100 years of work—against a new
target identified by genomics. They did not find a single hit. This sounds sur-
prising until one examines that inventory closely: almost half the compounds
in it could be considered to be derived from a single chemical class.
It is this problem that combinatorial chemistry is designed to solve, and
its explosive growth is testimony to both the magnitude of the problem and

© 2000 by Marcel Dekker, Inc.

the early successes of the combinatorial approach. Combinatorial chemistry
has many guises. In its purest form it involves the synthesis of all possible
compounds from a set of modular building blocks. However, it can also mean
high-throughput parallel synthesis of individual pure compounds, simultane-
ous synthesis of mixtures of compounds free in solution or on solid support,
or a number of other variations on these themes. Regardless of the details of
the process, the objective is the same: to produce a large number of chemically
‘‘diverse’’ compounds as rapidly as possible.
And that it does. Combinatorial methods have rewriten the standards
for synthetic productivity. Until about 10 years ago, a good chemist could
make and characterize perhaps 50 compounds per year. A combinatorial chem-
ist aims for more like 50,000. Such numbers define a revolution, one that has
already transformed the pharmaceutical industry and is likely to eventually
make an impact on every other area of industrial chemistry. Analytical chemis-
try is no exception, which brings us to the subject of this splendid book.
It is, of course, one thing to make 50,000 compounds and quite another
to know what one has made. Yet that is essential, because when new molecules
are available in such staggering numbers chemistry has become an information
science. Consider a chemically ‘‘diverse’’ (whatever that means, and nobody
really knows yet) library of 50,000 compounds. Now screen that against, say,
50 different drug targets. Then imagine that you don’t know what any of the
compounds are. The ones that give ‘‘hits’’ in your assays won’t remain un-
known for long: you will purify those and characterize them. Yet in doing so,
you will throw information away. If you knew the structure of every com-
pound, then the ones that failed in your assays would be almost as valuable
to you as the ones that succeeded because they would define the chemical
types that were not likely to work for that particular set of target classes.
Combinatorial chemistry promises to provide structure and activity data on a
scale never before imagined, but it will do so only if one can characterize
what one makes.
The task for the analytical community, then, is to develop methods of
separating and characterizing compounds suitable for the assembly-line scale
of combinatorial chemistry. This book details the methods currently available
and also discusses emerging techniques that could have a major impact. It
covers the gamut from a concise introduction to the various methods of com-
binatorial synthesis (Weller) to a splendidly useful compendium of commer-
cial resources (Brock and Andrews). In between, we are taken through clear
and critical descriptions of mass spectrometry (Vouros and Hauser-Fang),
I-R (Gremlich), and NMR (Shapiro) methods for high-throughout compound
identification, including the difficulties inherent in dealing with mixtures of

© 2000 by Marcel Dekker, Inc.

compounds that most combinatorial synthetic strategies produce. Detailed
consideration of the mixture problem is the basis for subsequent chapters on
chromatography (Swartz) and capillary electrophoresis (Krull, Gendreau, and
Dai). The latter process is of particular interest because it is so open to
miniaturization, and there seems little doubt that the laboratory of the future
will eventually be on a chip. The remaining chapters treat the information-
explosion problem head-on. Nicell discusses techniques for managing the
reams of data that an analytical lab will face when it deals with combinatorial
libraries, and Kyranos and Chipman outline the various ways of screening
libraries in a high-throughput fashion. Although the focus of the book is on
analytical techniques, these latter chapters in particular make it an excellent
introduction to the entire field of combinatorial chemistry.
It’s funny how the wheel comes around for scientific disciplines. At just
about the time that chemistry as a whole, and perhaps analytical chemistry in
particular, was being written off by some as a ‘‘mature field’’ unlikely to
produce the kind of cutting-edge excitement of, say, neurobiology or human
genetics, along comes combinatorial chemistry. The making, and characteriz-
ing, of molecules has once again been thrust into the heart of things, where
it has been for almost 200 years. If we are living in the Information Age,
then surely it is appropriate that the Central Science should become at last an
information science. This book is a manifesto for the next Chemical Revolu-
tion.

Gregory A. Petsko
Gyula and Katica Tauber Professor of Biochemistry
Director, Rosenstiel Basic Medical Sciences Research Center
Brandeis University
Waltham, Massachusetts

© 2000 by Marcel Dekker, Inc.

Preface

Analytical techniques play a critical role in drug discovery by providing valu-

able information to control the identity, purity, and stability of potential drug
candidates. These techniques, commonly in the form of chromatographic and
spectroscopic methods, are routinely used in the pharmaceutical laboratory
from the initial synthesis of a new chemical entity all the way through the
development cycle to the manufacture and sale of a pharmaceutical product.
The use of analytical techniques in the burgeoning field of combinatorial
chemistry is anything but routine. Combinatorial chemistry is a term that de-
scribes a set of tools for generating extensive chemical diversity rapidly and
efficiently. It has been described as having the potential to revolutionize drug
discovery. In the drug discovery process, large numbers of compounds are
screened for potential biological activity. Combinatorial chemistry does not
alter this process, but introduces a new step that greatly increases the molecular
diversity available for screening. The unique way in which this diversity is
produced, however, places new demands on the analytical methods used to
analyze and produce the desired results. In addition, the sheer volume of infor-
mation generated, and the manual labor that would be required, would be
overwhelming if not for the development of information management systems
and related techniques critical to the development of this field.
Following the introduction presented in Chapter 1, this book discusses
the application and use of specific analytical techniques (mass, infrared, and
nuclear magnetic resonance spectrometry, chromatography, and capillary elec-
trophoresis) in the combinatorial chemistry field (Chapters 2–6). It also dis-
cusses how to make sense of the vast amounts of data generated (Chapter 7),
details how the actual libraries of compounds produced are utilized (Chapter
8), and lists some of the vast commercial resources available to researchers
in the field of combinatorial chemistry (Chapter 9).

© 2000 by Marcel Dekker, Inc.

 Analytical Techniques in Combinatorial Chemistry is intended to pro-
vide specific details on how analytical techniques are brought to bear on the
unique challenges presented in the combinatorial chemistry laboratory. It is
aimed primarily at industrial and pharmaceutical chemists faced with the task
of developing methods, analyzing the results, and documenting and/or manag-
ing the discovery process in a combinatorial setting. Since many major phar-
maceutical companies are in the process of staffing combinatorial chemistry
departments, this publication could also serve as a training and reference
source, or perhaps a graduate-level textbook. While the book is not intended
to be an exhaustive literature review, specific citations are examined that high-
light the use of analytical techniques and the way in which they are utilized
to solve the unique problems encountered. It presents a basic introduction to
the field for a novice, while providing detailed information sufficient for an
expert in a particular analytical techique.
In closing, I would like to thank several colleagues who have contributed
to my efforts in the burgeoning field of drug discovery and combinatorial
chemistry. They include Bob Pfeifer, Beverly Kenney, Bob Karol, Ray Crow-
ley, Pat Fowler, Mike Balogh, John Hedon, and Eric Block of Waters Corpora-
tion (Milford, Massachusetts) as well as Steve Preece, Mark McDowall, and
Andrew Brailsford of Micromass Limited (Manchester, United Kingdom).
Special thanks also go to Carol, Kristina, and Robert of MCS (Uxbridge, Mas-
sachusetts) for their help and consideration in the preparation of this text.

Michael E. Swartz

© 2000 by Marcel Dekker, Inc.

Contents

Foreword Gregory A. Petsko

Preface
Contributors

1. An Introduction to Combinatorial Chemistry

Harold N. Weller

2. The Use of Mass Spectrometry

Annette Hauser-Fang and Paul Voúros

3. Infrared and Raman Spectroscopy

Hans-Ulrich Gremlich

4. NMR Methods
Michael J. Shapiro

5. The Role of Liquid Chromatography

Michael E. Swartz

6. Capillary Electrophoresis in Combinatorial Library

Analysis
Ira S. Krull, Christina A. Gendreau, and Hong Jian Dai

7. Finding a Needle in a Haystack: Information Management

for High-Throughput Synthesis of Small Organic Molecules
David Nickell

© 2000 by Marcel Dekker, Inc.

8. Bioanalytical Screening Methodologies for Accelerated Lead
Generation and Optimization in Drug Discovery
James N. Kyranos and Stewart D. Chipman

9. Commercial Resources
Mary Brock and Mark Andrews

© 2000 by Marcel Dekker, Inc.

Contributors

Mark Andrews, M.S. Strategic Planning Department, Waters Corporation,

Milford, Massachusetts

Mary Brock Market Intelligence Department, Waters Corporation, Milford,

Massachusetts

Stewart D. Chipman, Ph.D. Biology Department, ArQule, Inc., Medford,

Massachusetts

Hong Jian Dai, M.S. Department of Chemistry, Shuster Laboratories,

Quincy, Massachusetts

Christina A. Gendreau, M.S. Chemical Chromatography Division, Waters

Corporation, Milford, Massachusetts

Hans-Ulrich Gremlich, Ph.D. CTA/Analytics Department, Novartis

Pharma AG, Basel, Switzerland

Annette Hauser-Fang Department of Chemistry, Barnett Institute, North-

eastern University, Boston, Massachusetts

Ira S. Krull, Ph.D. Department of Chemistry, Northeastern University,

Boston, Massachusetts

James N. Kyranos, Ph.D. Analytical Chemistry Department, ArQule, Inc.,

Medford, Massachusetts

© 2000 by Marcel Dekker, Inc.

David Nickell, Ph.D. Discovery Informatics Department, Parke-Davis Phar-
maceutical Research, Ann Arbor, Michigan

Michael J. Shapiro, Ph.D. Core Technologies Department, Novartis, Sum-

mit, New Jersey

Michael E. Swartz, Ph.D. Market Development Department, Waters Cor-

poration, Milford, Massachusetts

Paul Voúros, Ph.D. Department of Chemistry, Barnett Institute, Northeast-

ern University, Boston, Massachusetts

Harold N. Weller, Ph.D. Combinatorial Drug Discovery Department,

Bristol-Myers Squibb Co., Princeton, New Jersey

© 2000 by Marcel Dekker, Inc.

1
An Introduction to Combinatorial
Chemistry

Harold N. Weller
Bristol-Myers Squibb Co.
Princeton, New Jersey

I. INTRODUCTION AND OVERVIEW

The advent of combinatorial chemistry has led to a revolution in the drug

discovery process, with impact on many related disciplines. Areas such as
analytical chemistry, process development chemistry, and biological assay are
reflecting the effects of combinatorial chemistry. However, at the same time,
practitioners in those areas often have a vague idea about what combinatorial
chemistry really is. Vendors of instruments and consumable supplies recog-
nize the tremendous market potential of this new technology but are unclear
about how best to capitalize on that potential. Part of the reason is that the
definition of combinatorial chemistry differs from organization to organization
and even within organizations depending on the specific drug discovery target.
The purpose of this chapter is to provide a brief introduction to the assortment
of new technologies collectively referred to as ‘‘combinatorial chemistry.’’ It
will become clear that no single definition can clearly and concisely describe
this new field.
Until the advent of combinatorial chemistry in the early 1990s, the drug
discovery process had remained essentially unchanged for many decades.
Chemists synthesized new molecular entities (compounds), biological proper-
ties were determined, and the chemists then incorporated the biological results
into the design of the next-generation compound. This serial iterative process

© 2000 by Marcel Dekker, Inc.

was repeated many times until a molecule with suitable drug-like properties
was discovered. Through the years, improvements in biological assay provided
higher throughput and more precise data on which to make judgments during
the iterative design cycle. Similarly, improvements in computational chemistry
provided additional rationale and tools for the chemist to use during iterative
design. However, one thing remained constant, i.e., serial synthesis of one
compound at a time with biological results providing the insight for iterative
design of the next generation. With the development of high-throughput in
vitro biological assays, the iterative serial synthesis of individual new com-
pounds became the rate-limiting step in the overall drug discovery process.
In order to overcome the synthesis bottleneck in drug discovery, the
concept of preparing many compounds at one time (parallel synthesis) rather
than one compound at a time (serial synthesis) was born. In its simplest form,
this distinction constitutes the definition of combinatorial chemistry. The ori-
gin of the concept has been ascribed (1) to Furka and others as early as 1982.
Early applications of parallel synthesis methods were primarily in the area of
peptide library synthesis and have been extensively reviewed (2,3). In the early
1990s, however, application to small drug-like molecules was reported (4) and
the explosion in combinatorial chemistry activity began.
As more and more organizations have become involved in combinatorial
chemistry, an increasingly large number of variations on the theme have been
developed. To understand this complexity, one must remember that the overall
objective of the field (within the pharmaceutical industry) is to accelerate the
drug discovery process and that different ‘‘combinatorial’’ technologies may
be best suited to different phases of the drug discovery process.
Drug discovery begins with a hypothesis involving a target biomolecule
(receptor, enzyme, etc.), a proposed molecular intervention (receptor antago-
nism, enzyme inhibition, etc.), and a proposed clinical outcome of that inter-
vention. To begin a full drug discovery program requires a reasonable target
hypothesis, a validated biological assay, and (ideally) a structural starting point
for drug design. Once the assay is in place, the first chemistry phase of a drug
discovery process is the search for a structural starting point. That starting
point is often called a ‘‘lead compound’’ and this phase of drug discovery is
often called the ‘‘lead discovery’’ phase. A suitable lead compound is a small
drug-like molecule with measurable and reproducible activity in the primary
assay or assays. Even in programs with good starting templates (e.g., a large
biomolecule with the desired activity or a known substrate structure), testing
of thousands of compounds is often required before a useful lead compound
is found. In the absence of a rational starting point, random screening may
require testing of many tens of thousands of compounds in order to uncover
a lead in the lead discovery phase.

© 2000 by Marcel Dekker, Inc.

 A combinatorial chemistry program designed to support lead discovery
has several well-defined characteristics. First and foremost, it provides very
large numbers of compounds for screening—perhaps into the tens of thou-
sands or even hundreds of thousands of new compounds. Since screening is
done only in a single primary in vitro assay, only a small amount of each
compound is required. Since the objective is merely to find activity and not
to compare activity among compounds, a high level of purity of individual
compounds is not required (providing that impurities do not themselves inter-
fere with the biological assay). Intentional synthesis and testing of mixtures
is common in the lead discovery phase, and indeed early usage of the term
‘‘combinatorial library’’ to describe synthetic compounds generally referred
to synthetic mixtures containing combinations of many compounds (5). Since
the objective is to find a single compound with measurable activity, testing
of mixtures is often appropriate. A mixture with moderate average activity
may contain many individual compounds with moderate activity or a single
very active compound and many inactive compounds. In the lead discovery
phase, where the probability of finding even a single active compound is slim,
the latter case is often the more likely scenario. The practical limit of the
number of compounds that may constitute a mixture depends on the assay
sensitivity and the desired activity of a presumed lead compound, as well as
the degree of control one has on the ratio of compounds in the mixture.
Once a lead compound has been discovered, the drug discovery project
moves into an early ‘‘lead optimization’’ phase. In this phase, the intrinsic
activity of the lead compound is improved by small structural modifications
to the original structure. Design of target compounds for synthesis is guided
by results from testing of previous compounds in the series. Since compounds
are tested in only a single or a limited number of primary in vitro assays, the
amount of each compound required is small. Often it is important to compare
the activity of one analog with that of another; thus purity must be sufficient
to allow such direct comparisons, and interpretation of data from screening of
mixtures becomes cumbersome. It should be clear that combinatorial synthesis
methods that are highly suitable for the lead discovery phase may be unsuitable
for lead optimization.
Several iterative design cycles of early lead optimization may result in
one or more compounds that are highly active in the primary assay. At this
point, the drug discovery project moves into late-stage lead optimization; test-
ing will begin in a variety of secondary in vitro or in vivo assays such as
measurement of physicochemical properties, pharamcokinetics, metabolism,
toxicity, cell permeability, etc. These secondary assays may be less highly
automated than the primary assay and often require larger amounts of each
compound. Secondary assays may have more strict analysis requirements than

© 2000 by Marcel Dekker, Inc.

automated primary assays; thus high-purity compounds are preferred. As the
project moves from early stage lead optimization to late stage lead optimiza-
tion, the combinatorial synthesis focus will shift from making small amounts
of large numbers of compounds of moderate purity to making larger amounts
of fewer compounds of high purity. Once again, synthesis methods that are
highly suitable for early stage lead optimization may be unsuitable for late
stage lead optimization.
The final phase of a drug discovery project is candidate selection. In
this phase, compounds are screened in an in vivo disease model in addition
to the primary and secondary in vitro assays. Once again, the amount of each
compound needed is increased, the purity requirement is higher, and the total
number of compounds generally required is smaller.
From the discussion above, it should be clear that no single combinato-
rial synthesis technology can optimally serve all phases of a drug discovery
project. In the sections that follow, various combinatorial synthesis methods
will be described, as well as where each particular methodology is most useful
in the drug discovery process. It is the proliferation of synthesis methods, each
suitable for a certain phase of drug discovery, that leads to confusion as to
the role and definition of combinatorial chemistry. In truth, combinatorial
chemistry is a collection of different methodologies for the synthesis of multi-
ple compounds, each with its own strengths and weaknesses, and each suitable
for a slightly different phase of drug discovery. Importantly, to be successful
throughout the entire drug discovery process, an organization must have access
to more than a single combinatorial chemistry technique.

II. AUTOMATED SOLUTION PHASE SYNTHESIS:

APPLICATIONS AND ANALYTICAL CHALLENGES

Organic synthesis is traditionally done in solution. That is, starting materials

and reagents are dissolved in a solvent or homogeneous solvent mixture for
the reaction to take place. After each step of a multistep chemical synthesis
the intermediate reaction products are isolated, purified, and characterized be-
fore the next synthetic step is taken. The challenges in automating solution
phase synthesis are to find creative ways to avoid tedious reaction workup
and product purification after each step. Balancing these challenges is the
availability of the full range of standard analytical methods to assay purity
and structure throughout the synthetic sequence.
Solution phase organic reactions are typically ‘‘worked up’’ beginning
with a series of extractions whereby the mixture is partitioned between an

© 2000 by Marcel Dekker, Inc.

organic phase and an aqueous phase of known pH. In this way, products and
reagents are separated from one another by their solubility and ionic nature.
While organic/aqueous extraction procedures have been automated using ro-
botic liquid handlers (6–8), difficulties associated with accurately detecting
the solvent boundary have limited their applicability. Methods utilizing ion
exchange resins to perform separations based on ionic nature are more easily
automated, e.g., using commercially available solid phase extraction robots
(9). Other methods, including use of solid supported reagents (10,11) and in-
troduction of scavenger resins to reaction mixtures (12), also facilitate separa-
tion of products from reagents during automated solution phase synthesis. Be-
cause the various methods used for reagent removal are often compound-
specific, solution phase combinatorial synthesis is generally directed to synthe-
sis of individual discreet compounds (13), though mixture synthesis has also
been reported (14). Solution phase combinatorial synthesis has been used to
prepare libraries of up to 20,000 compounds for lead discovery (15) but may
be most suited to preparation of moderate numbers of individual discreet com-
pounds for the lead optimization phase of drug discovery.
As seen from the above discussion, the analytical challenges of auto-
mated solution phase synthesis lie primarily with product isolation, purifica-
tion, and characterization. Specific equipment for high throughput or parallel
purification of compounds from combinatorial solution phase synthesis is still
largely unavailable; thus creative use of existing equipment is often required.
For example, disposable solid phase extraction cartridges and robotic appara-
tus were invented primarily for concentration of biological and environmental
samples prior to analysis but are now widely used for preparative scale purifi-
cation of samples from combinatorial synthesis (16). High-throughput prepar-
ative HPLC is also a powerful tool for purification of solution phase synthesis
libraries, but systems designed specifically for that purpose have only recently
been reported and are not yet widely available (17). Methods for characteriza-
tion of solution phase libraries are the same as those used for solid phase
libraries and will be discussed later.

III. SOLID PHASE SYNTHESIS: APPLICATIONS AND

ANALYTICAL CHALLENGES

Synthesis of organic compounds on solid support was first reported for peptide
synthesis in the 1960s (18). In solid-supported (or ‘‘solid phase’’) synthesis,
the starting organic scaffold is covalently bonded to an insoluble polymer
support, generally a functionalized polystyrene derivative. Reagents and build-

© 2000 by Marcel Dekker, Inc.

ing blocks are added in solution and the resulting reaction mixture is heteroge-
neous (reagents in solution, starting scaffold bound to insoluble solid support).
As the reaction takes place, the starting scaffold is modified but remains
bonded to the solid support, while reagents are converted to byproducts but
remain in solution. When the reaction is complete, simple filtration removes
soluble reagents and byproducts, leaving the modified substrate still bound to
the insoluble polymer. The elaborate workup required for reagent removal in
solution phase synthesis (extraction, purification, etc.) is reduced to a simple
filtration when synthesis is done on solid phase. In this way, many sequential
chemical synthesis steps can be carried out on a single starting material with
only filtration (rather than elaborate extraction and purification) between steps.
At the end of the multistep chemical synthesis sequence, the desired product
will still be attached to the insoluble resin, and all reactants and byproducts
will have been washed away. At this point, a chemical reaction is performed
that cleaves the covalent linkage between the product and the solid support,
thus bringing the product into solution. Filtration following this ‘‘cleavage’’
step thus delivers the product in solution and separates it from the solid support
to which is was bound during the synthesis sequence. If each step in the solid
phase synthesis sequence has gone to 100% completion and delivered no unde-
sired polymer bound side products, then the product cleaved from the resin
will be 100% pure. In practice, such high levels of purity are rarely achieved
except in the area of biopolymer synthesis.
The obvious advantages for automation of solid phase synthesis over
solution phase synthesis have been exploited for many years in the area of
biopolymer synthesis. Automated equipment for parallel solid phase synthesis
of peptides is commercially available and often delivers products of very high
purity. Peptide synthesis generally proceeds by anchoring a single amino acid
to polystyrene resin, then adding sequential amino acids via amide bond cou-
pling reactions. Thus there is a single type of bond forming reaction (amide
bond formation), a limited set of monomers (20 naturally occurring amino
acids), and chemists have been optimizing the process for over 30 years. It is
little wonder that products of high purity are commonly obtained directly from
resin cleavage. Compare this situation with the broadly based organic synthe-
sis now being attempted on solid phase i.e., many different reaction types,
virtually unlimited monomer availability, and only a few years of collective
reaction optimization by the chemistry community. In the latter case, products
of high purity are not necessarily obtained directly from resin cleavage unless
the chemist has spent considerable time optimizing reaction conditions for
each step prior to synthesis of a library of compounds.
The chemist setting out to synthesize a library of several dozen to several

© 2000 by Marcel Dekker, Inc.

thousand organic compounds via solid phase synthesis will generally begin
by working out synthesis conditions on a small subset of the library. The first
step is to find a suitable chemical linker to tether the starting substrate to an
insoluble support. The linker is generally attached to the substrate molecule
via a labile functional group such as through an ester linkage (which is later
cleaved to a carboxylic acid) (19) or an acetal group (which is cleaved to
liberate a hydroxyl group) (20). Design of functionalized polymers and linkers
for solid phase synthesis is an emerging technology in itself and has been
reviewed (21,22). Once the substrate is linked to the solid phase resin, the
various building blocks must be added one at a time. Each chemical step must
be optimized to determine the best conditions for optimal yield. One advantage
of solid phase synthesis is that a large excess of reagents may be used to drive
reactions to completion because all excess reagents will be easily removed by
filtration following the reaction. The process of developing suitable reaction
conditions for each step of a library synthesis generally takes several months of
tedious work by the chemist. In contrast, using modern automation techniques,
synthesis of the actual library may only require a few days or at most a few
weeks after synthesis conditions have been defined. Developing reaction con-
ditions for solid supported synthesis is particularly tedious because of the lack
of suitable methods for analysis of resin-bound reaction products. The chemist
no longer has direct access to the full range of conventional methods for
assaying purity (e.g., chromatography) or structure (e.g., spectroscopy) unless
the product is first cleaved from resin for analysis. Solid phase nuclear mag-
netic resonance (NMR) (23,24) and Fourier transform infrared spectroscopy
(FTIR) methods now provide the most useful information about structures
bound to resin, and these methods will be reviewed in subsequent chapters in
this volume. Opportunities thus exist for the analytical chemist to have a sig-
nificant impact on a combinatorial chemistry program in two ways: develop-
ment of methods for analysis of reaction intermediates on solid support, and
development of methods for purification of impure products after cleavage
from solid support. Either of these methods will allow the chemist to shorten
his or her chemical development time and proceed directly to library synthesis
with confidence. Many of the chapters in the remainder of this volume deal
directly with these important issues.
Solid phase synthesis of organic molecules is exemplified by the pio-
neering work of Bunin and Ellman on solid phase synthesis of a small benzodi-
azepine library (29). As shown in Fig. 1, a substituted benzophenone was
tethered to a solid support via a substituent (R 1). The solid supported benzo-
phenone was then elaborated through five synthetic steps to the target benzodi-
azepines with four points of diversity. Cleavage from the solid support re-

© 2000 by Marcel Dekker, Inc.

Figure 1 Solid phase benzodiazepine synthesis (29).

leased the products in high yield. This preliminary report was the first widely
recognized solid supported synthesis of nonoligomeric compounds and set
the stage for rapid development of combinatorial synthesis methods for small
organic molecules.

IV. DIFFERENT APPLICATIONS OF COMBINATORIAL

SYNTHESIS TO DRUG DISCOVERY

Combinatorial synthesis in its broadest form involves parallel, possibly auto-

mated, synthesis of many compounds at once using either solution or solid
phase synthesis methods. Beyond that broad definition, though, an impressive
variety of specific applications and methods has been developed to suit the
particular needs of various projects or corporate sponsors. At one extreme is
synthesis of very large mixtures for early phase lead discovery programs, and
at the other extreme is parallel synthesis of pure discrete compounds for late-
stage lead optimization programs. The following sections will attempt to cate-
gorize the main combinatorial synthesis methods into broad groups.

A. Synthesis of Mixtures by Solid Phase Methods

It would seem that the simplest way to prepare a mixture of compounds on
solid phase would be to attach a single starting substrate to the polymer and
then react that polymer bound substrate with a mixture of building blocks or
reagents to produce a mixture of products. The problem with this approach

© 2000 by Marcel Dekker, Inc.

is that the kinetics for reaction of the substrate with each of the diverse building
blocks may not be identical. When reaction kinetics are not identical, the re-
sulting product mixture will not be equimolar, and it is possible that some
of the presumed products will not be present in the product mixture at all.
Furthermore, each individual polymer bead in the mixture will contain a mix-
ture of products rather than a single product, thus complicating analysis and
deconvolution (as will be discussed later).
A better way to prepare mixtures by solid phase synthesis is by using
the ‘‘split-and-pool’’ method. With this method a single substrate is bound to
the resin; then the resin is split into a number of equal size pools, where the
number of pools represents the number of diverse building blocks to be used
in the first step of the synthesis. Each pool is then reacted with its designated
building block in a separate reaction vessel under conditions where the reac-
tions are driven to completion (e.g., by using excess reagents). Following reac-
tion, the pools are filtered to remove excess reagents and washed with fresh
solvent. All of the polymer from all of the pools is then recombined and mixed
thoroughly. The combined resin is then split again into a new set of pools
representing the number of diverse building blocks to be used in the second
step of the synthesis. At this point, each resin pool should contain an equimolar
amount of each of the products created in the first synthesis step. The process
is then repeated through as many cycles as needed to complete synthesis of
the entire library.
The split-and-pool method is illustrated in Fig. 2 for a two-step synthesis
involving three building blocks at each step resulting in three equimolar mix-
tures of three compounds each. The starting functionalized resin is divided
into three equal pools, and each pool is reacted with a specific reagent [R1(a),
R1(b), or R1(c)] in a separate container. At the end of this step, each reaction
container contains a single specific resin–bound product. Reagents are then
removed from the reactions and the solid resins are washed with fresh solvent.
All three resin batches are then combined and mixed thoroughly to provide
an equimolar mixture of the three possible products. Note that, in principle,
one could obtain the same mixture by simply reacting all of the starting resin
with a mixture of the three reagents in a single vessel. In practice, though,
this process will not provide an equimolar mixture of products due to differen-
tial reaction kinetics. It is the desire to obtain equimolar mixtures of products
that necessitates the split-and-pool method.
Returning to the hypothetical split and pool synthesis (Fig. 2), the chem-
ist would next split the equimolar mixture of three products into another three
identical pools. Reaction of each of these pools separately with three different

© 2000 by Marcel Dekker, Inc.

Figure 2 Split-and-pool synthesis.

© 2000 by Marcel Dekker, Inc.

reagents [R2(a), R2(b), and R2(c)] results in three equimolar mixtures of three
products each, for a total of nine new products. The products can be cleaved
from the solid support to provide three mixtures of three compounds each.
The split-and-pool method has several interesting ramifications. First,
since it does not depend on identical reaction kinetics, it should produce an
equimolar mixture of products assuming that all reactions are driven to com-
pletion. Second, since any individual polymer particle follows only one syn-
thetic path, all of the product molecules bound to that particle should be the
same. That is, one compound is prepared per bead. Third, since one compound
is prepared per bead, there must be at least as many polymer beads in the
total synthesis pool as there are theoretical products being formed during the
synthesis. Imagine, for example, if the split-and-pool synthesis outlined in Fig
2 was begun with only six resin particles! In practice a statistical sampling
of the total resin pool is made with each ‘‘split’’ and statistical analysis can
be done to determine the required number of particles to ensure adequate sam-
pling of a given pool (30,31). Finally, split-and-pool methodology results in
significant reduction in the total number of reactions needed to produce the
library. Imagine, for example, a library where building blocks are added in
three sequential chemical steps and where each building block is represented
by 10 different variations. The total library size will be 10 ⫻ 10 ⫻ 10 ⫽ 1000
compounds. To prepare the compounds conventionally would require 1000
individual reactions at each step or 3000 reactions in total. Using split-and-
pool methods, only 10 reactions are required at each step or 30 reactions total
to prepare the library. Counterbalancing the synthesis efficiency is the effort
required to split and pool the resin particles.
At the end of a split-and-pool synthesis, the chemist has the option to
recombine all resin particles into a single pool and then cleave products, or
to keep the final split mixtures separate and cleave each individually. Using
the example in Fig. 2, the chemist could recombine all polymer resin prior to
cleavage to obtain an equimolar mixture of all nine compounds in the library.
Alternately, the chemist could keep the three pools used in the final step sepa-
rate and cleave each individually, resulting in three separate mixtures of three
compounds each as illustrated in the figure. The choice of which path to take
is dictated largely by the target mixture size. If there are too few compounds
in the final mixture the synthetic advantages of mixtures are lost, whereas if
there are too many compounds in the mixture there is significant risk that a
single active compound will be missed during bioassay due to the dilution
effect of other mixture members. In the early days of combinatorial synthesis,
very large mixture libraries were common, with mixtures containing tens of
thousands or even hundreds of thousands of compounds commonly reported.

© 2000 by Marcel Dekker, Inc.

For example, a library containing 19 ⫻ 105 (nearly 2 million) pentapeptides
was prepared from 19 of the 20 naturally occurring amino acids in 1991 (32),
as was a library of over 34 million hexapeptides consisting of 324 mixtures
of 104,976 compounds each (33).
Because of concerns about not detecting low concentrations of active
compounds in such large mixtures, the current trend is toward much smaller
mixtures containing only hundreds or even dozens of compounds. For exam-
ple, the Affymax group prepared a library containing mixtures of only 540
compounds and identified potent cyclooxygenase-1 inhibitors after deconvolu-
tion (34). The issue of optimum mixture size is not yet settled and discussion
will no doubt continue for some time.
When bioassay of a mixture produces apparent activity, one must then
‘‘deconvolute’’ the mixture to identify the active component (or components).
The most straightforward deconvolution method is independent synthesis of
each individual member of the original mixture. For large mixtures, though,
such an approach is not practical. Instead, iterative resynthesis of increasingly
smaller mixtures is used to ultimately identify the active component. For ex-
ample, a mixture of 104,976 hexapeptides was deconvoluted by four iterative
syntheses of 20 compounds or mixtures each (35). In the first iteration, 20
mixtures of 6859 peptides were synthesized. The most active mixture was
then used as the starting point for 20 mixtures of 361 compounds. Next, 20
mixtures of 19 compounds each were synthesized and, finally, the 19 individ-
ual compounds of the most active mixture were synthesized to identify the
single most active compound in the original mixture.
Iterative deconvolution works best when a single compound of the mix-
ture is much more active than all others. Conversely, deconvolution is difficult
when all members of a library have similar activity—as is often the case in
the later stages of a drug discovery process. Because of the difficulties associ-
ated with mixture deconvolution, split-and-pool synthesis is best suited to the
early lead discovery phase of the drug discovery process. Later stages, requir-
ing direct comparison of one compound against another, are best served by
other methods described below.

B. Chemically Encoded Libraries

The synthetic advantages of preparing mixtures instead of individual com-
pounds are offset by the need to deconvolute mixture libraries following bioas-
say. Deconvolution shortcuts, such as synthesis of multiple libraries containing
orthogonal pools (36), are sometimes useful but can only be relied on when
the activity of the mixture results from a single highly active compound. Keep-

© 2000 by Marcel Dekker, Inc.

ing in mind that split-and-pool methodology results in only one compound
per polymer bead, the difficulties associated with deconvolution and the uncer-
tainties of screening mixtures can be eliminated by the screening of individual
compounds from single beads. Methods have been developed for screening
products while still attached to the beads. Alternately, following split-and-
pool synthesis, the individual beads are separated and the products cleaved
individually from the beads. The individual products are then screened in the
bioassay. Active compounds discovered in the bioassay are individual com-
pounds rather than components of mixtures; thus, no deconvolution is neces-
sary. How, though, can one identify the structure of the active compounds
from the small (subnanomole) amount prepared on a single bead? Direct analy-
sis of the products by sensitive analytical methods has been reported. For
example, peptides and oligonucleotides have been analyzed in the 5-pmol
range by Edman degradation or DNA sequencing (37). Molecular weights
have been obtained for small organic molecules from single-bead synthesis
(38), but large libraries may contain more than a single member with the same
mass. The solution to the problem of product identification from single bead
synthesis lies mainly with encoded libraries.
Chemically encoded libraries are synthesized from bifunctionalized
polymers. During each step of the split-and-pool synthesis, two separate chem-
ical reactions take place. The first adds a building block to the polymer-bound
substrate and leads to the target product. The second adds a building block
to a unique chemical tag that will be used to later identify the product associ-
ated with that bead. The method is shown schematically in Fig. 3 using an
example of a two-step synthesis with three variations at each step. Following
synthesis, products from individual beads are cleaved and screened. When
active compounds are found, the original bead is subjected to an orthogonal
cleavage step that releases the chemical tag. The chemical tag is subjected to
a sensitive and precise analytical method that reveals the structure of the tag
and therefore the structure of the product molecule.
Success of an encoded library strategy depends on the availability of a
method for screening products from individual beads, and of a suitably precise
and sensitive analytical method for decoding the chemical tag. Four primary
tagging methods have been reported. As mentioned above, sensitive methods
exist for sequencing of peptides and oligonucleotides; thus, peptides and oligo-
nucleotides make suitable chemical tags for encoded library synthesis. For
example, Needles and coworkers (39) demonstrated the suitability of oligonu-
cleotide tags for the synthesis of a peptide library using an orthogonal pro-
tecting group strategy. The growing peptide chain was protected by the base-
labile 9-Fluorenylmethoxycarbonyl (FMOC) protecting group, while the

© 2000 by Marcel Dekker, Inc.

Figure 3 Chemically encoded split-and-pool synthesis.

© 2000 by Marcel Dekker, Inc.

growing oligonucleotide chain was protected by the acid-labile dimethoxytri-
tyl (DMT) group. A similar strategy should work well for synthesis of small
organic molecules instead of peptides, providing the chemical sequence
needed for product synthesis is compatible with the oligonucleotide tag. Simi-
lar strategies have been reported using peptides as the chemical tag (40). A
binary encoding strategy has been reported using substituted haloaromatic tags
that can be detected at very low levels by electron capture gas chromatography
(41). The tags are attached via a photolabile (42) or oxidatively labile (43)
linker, thus allowing most types of chemistry to be used in building the desired
product molecules. This method has been reported for synthesis of both pep-
tides (44) and nonpeptide molecules (45). A similar coding method utilizing
secondary amines attached via amide linkages has also been reported (46,47).
In this case, the tags are hydrolyzed using 6 N HCl, the resulting secondary
amines derivatized, separated by HPLC, and detected by fluorescence spec-
troscopy.
The power of the encoded library strategy is illustrated by an example
from Borchardt and Still (48). A 50,000-compound encoded acyl tripeptide
amide library was prepared using the split-and-pool method. Beads from the
library were mixed with a dye-linked synthetic receptor. After 24 hours, beads
containing compounds that bind tightly to the synthetic receptor were stained
deep red. Stained beads were mechanically selected, their binary coded tags
were cleaved by photolysis, and the codes were read by electron capture gas
chromatography to determine the structures of the tight-binding substrates. In
this way, 50 compounds with high receptor affinity were selected from the
library of over 50,000 compounds.
Chemically encoded libraries are a very powerful method for preparing
large numbers of individual compounds whose structures can be determined.
Successful application of a chemically encoded library requires access to a
suitably sensitive high-throughput analytical method for reading the chemical
code. Since code reading itself can become time consuming if many members
of a library are found to be active, and the amount of each compound produced
is limited to the amount that can be synthesized on a single bead, encoded
libraries are well suited to lead discovery and early phase lead optimization
but may be less well suited as the drug discovery process approaches candidate
selection.

C. Mechanically Encoded Libraries

The main limitations of chemically encoded libraries are the necessity to per-
form an extra chemical reaction at each step to introduce the coding tag, the

© 2000 by Marcel Dekker, Inc.

limitations introduced by the presence of the tag on the chemistries that may
be performed on the target substrate, and the limited amount of material that
can be obtained from a single polymer bead. These limitations can all be over-
come using mechanically encoded libraries instead of chemically encoded li-
braries. With mechanically encoded libraries, polymer resin is held in a porous
container that allows soluble reagents to pass through but not resin beads. The
container is labeled with a mechanical tag (perhaps as simple as a printed
label) and the containers of resin are treated in the same way that individual
beads are treated in the split-and-pool method. While the split-and-pool
method provides one compound per bead, the mechanically labeled split-and-
pool method provides one compound per container of beads. If the synthetic
path of each container is recorded at each synthetic step, then simply reading
the container label will identify the product contained within. In practice, a
database is generally created prior to execution of the synthesis outlining the
precise synthetic pathway intended for each resin container. After each syn-
thetic step the containers are retrieved, their labels read, and they are pooled
for the next synthetic step according to the predetermined sequence from the
database. This method is referred to as ‘‘deterministic’’ split-and-pool synthe-
sis as opposed to ‘‘statistical’’ split-and-pool synthesis. Because the fate of
each container is determined ahead of time, statistics play no role and the
number of containers required is exactly equal to the number of products being
synthesized.
The first example of a mechanically encoded library was reported by
Houghten in 1985 (49). Resin was contained in a polypropylene mesh packet
resembling a tea bag. The label was mechanically inscribed on the packet and
was read visually with manual sorting of the packets. In this way, 247 analogs
of a 13-amino-acid peptide were individually prepared. Synthesis of a
500,000-compound mixture library was recently reported using this method
(50). The packets used were porous polyethylene tubes with printed labels
that could be automatically read by optical character recognition (OCR) soft-
ware. Each packet contained a mixture of 21 different functionalized resins,
and the effort resulted in synthesis of approximately 26,000 individual contain-
ers each theoretically containing a mixture of 21 compounds for a total library
of 551,070 compounds. The reader will recall, however, that in syntheses of
this type each packet will contain an equimolar mixture of 21 products only
if the reaction kinetics at each step are identical and that this was the reason for
moving away from direct mixture synthesis and toward split-and-pool mixture
synthesis in the first place. Two groups have recently reported use of radiofre-
quency tags in place of visually readable tags (51,52). These machine-readable
tags offer the possibility of inexpensive automated reading and sorting. All
of the above methods require the use of a nonreactive porous resin container.

© 2000 by Marcel Dekker, Inc.

Radiofrequency tags encapsulated in grafted functionalized polymers have re-
cently been reported (53), as have laser optical–encoded ceramics with grafted
polystyrene supports (54). These devices have the properties of ‘‘really big
beads’’ that contain the identifying tag and thus combine the features of chemi-
cally encoded libraries with the advantages of mechanically encoded libraries.
Mechanically encoded libraries are best suited to synthesis of individual
compounds rather than mixtures. The amount of each compound produced
can be small or large and is limited only by the size of the packets and related
physical limitations. Since reactions can be carried out in conventional glass-
ware, mechanically encoded libraries offer an inexpensive entry into large-
library synthesis. The method is applicable throughout most of the drug dis-
covery process with the possible exception of the very earliest discovery phase
and the very latest candidate selection phase.
Mechanically encoded libraries offer their greatest advantage over indi-
vidual compound synthesis when the library consists of a dense symmetrical
array. Consider, for example, synthesis of a 100-compound library with 10
structural variations at each of two positions (a 10 ⫻ 10 array). Individual
compound synthesis would require that all 100 compounds undergo two chem-
ical synthesis steps (not counting cleavage from the resin) or a total of 200
chemical reactions. Synthesis of the same library using mechanical tags would
require only 20 reactions (10 reaction vessels each containing 10 packets for
the first step, and the same for the second step). For a 100-compound library
consisting of a 2 ⫻ 50 array, though, individual synthesis would still require
200 reactions whereas the mechanically encoded library would require 52 re-
actions (two reactions each with 50 packets in the first step, and 50 reactions
each with two packets at the second step). Thus the advantage over individual
synthesis is diminished. The advantage is also reduced for synthesis of
‘‘sparse’’ arrays. Consider again the 10 ⫻ 10 array, but suppose that sophisti-
cated library design software has excluded some of the possible combinations
so that there are ‘‘holes’’ in the array and a total of only 50 targets. Individual
synthesis would require 50 reactions at each of 2 steps or 100 reactions total,
while the encoded library would still require all 10 reactions at each step
(though with less than 10 packets at each step) for a total of 20 reactions. For
these reasons, encoded library synthesis will probably take its place alongside
individual compound synthesis and split-and-pool mixture synthesis (rather
than replace them) in an overall drug discovery program.

D. Automated Parallel Synthesis

The simplest conceptual manifestation of combinatorial chemistry is that of
parallel synthesis, which is simply synthesis of several compounds in distinct

© 2000 by Marcel Dekker, Inc.

reaction vessels at once (in parallel) rather than sequentially (in series). Any
chemist who has had several reaction flasks stirring at once has practiced a
crude form of parallel synthesis. Parallel synthesis results in individual com-
pounds that can be treated in downstream handling steps identically to com-
pounds synthesized conventionally. Thus it is a conceptually easy first entry
into combinatorial chemistry. The tedium of handling many reaction flasks at a
time has quickly led to a wide variety of apparatus for automating, or partially
automating, parallel synthesis. The first equipment available was modeled
after the venerable peptide synthesizer and offered complete automation of a
multistep reaction sequence. Such equipment is now available commercially
from vendors such as Advanced Chemtech (55) and Arognaut Technologies
(56). The advantages of such systems (total automation) are offset by relatively
complex software and relatively low synthesis capacity. The latter is a result
of the fact that the actual synthesis reaction vessels must remain mated to
apparatus for reagent addition throughout the process in order to meet the
needs of total automation. A number of modular workstations have now been
reported, and some are now available commercially from vendors such as
Diversomer Technologies (57), Bohdan Automation (58), and Tecan (59). One
of the first workstations reported was the Diversomer apparatus from Parke-
Davis (60,61). Solid phase resin is contained in glass gas dispersion tubes that
are arrayed in solvent containment wells. Reagents are added via robotic liquid
handler through a septum that seals the top of the apparatus. A number of
other reaction blocks have been reported. For example, an enclosed reaction
block that can be heated and cooled and to which reagents are added roboti-
cally was described by the Ontogen group (62).
Regardless of the source, reactors for automated parallel synthesis have
certain common characteristics. They all offer a group of reaction vessels that
can be handled as a unit. All offer a means of liquid addition either via robotic
liquid handler or via a closed pneumatic or pumping system. All offer a means
of separating soluble reagents from insoluble polymer-supported products, and
all offer a means of collecting soluble products after cleavage from the poly-
mer resin. Systems that offer total automation of a multistep chemical se-
quence are generally referred to as ‘‘synthesizers.’’ With synthesizers, the
reactions generally take place entirely within the confines of the synthesizer
and proceed completely unattended. Throughput per run is limited to the ca-
pacity of the synthesizer. Systems that require manual intervention and move-
ment of the reactions from place to place in assembly line fashion are referred
to as ‘‘workstations.’’ In this case, a group of reaction vessels, perhaps con-
tained in a single ‘‘reaction block,’’ is manually moved from station to station
as the synthesis progresses. For example, the reaction block may be loaded

© 2000 by Marcel Dekker, Inc.

with polymer resin and moved to a liquid handler for reagent addition. It is
then moved to an incubation station where agitation, atmosphere control, and
temperature control are maintained while the chemical reaction takes place.
This process frees up the liquid handler workstation to accept another reaction
block in assembly-line fashion. The reaction block then moves on to various
stations for reagent removal, resin washing, product cleavage, and so on. The
workstation approach has been well described by Cargill and coworkers (63)
who describe synthesis of thousands of individual compounds per month.
While the workstation approach is not totally automated and requires manual
intervention, it also offers higher overall throughput at lower cost than the
synthesizer approach.
A number of novel approaches to parallel synthesis have been reported.
For example, synthesis of peptides on grafted polyethylene pins spatially ar-
rayed in 96-well microtiter plate format has been described (64). In this way,
a carrier containing 96 grafted pins is inserted into a 96-well microtiter plate
to expose the polymer-bound substrate to soluble reagents. The reagents are
removed by simply lifting the pin carrier from the reagent solutions in the
microtiter plate. Application of this method to synthesis of nonpeptides has
also been reported (65).
The Affymax group has reported a novel photolithography method for
preparing spatially dispersed peptide arrays (66). Using this method, the start-
ing substrate is bound to a functionalized glass substrate, and the reactive
terminus of the starting peptide is protected with a photolabile protecting
group. Using photolithography, the peptide is deprotected only on selected
sections of the sheet. Reagents are then introduced to add a protected amino
acid to the deprotected peptide. The sequence is repeated again and again until
all positions on the sheet have been reacted. This system is very powerful,
but it has the drawbacks that (1) only substrates that can be suitably protected
can be used and (2) the synthesis is essentially a serial rather than a parallel
process because only one type of amino acid can be introduced at a time.
Another new approach to very-high-throughput parallel synthesis is the
ChemSheet (67). This is a flat polymer sheet containing 2304 8-µl reaction
wells. Ink jet dispensing technology allows high-speed delivery of reagents
and solvents to the very small reaction wells. While this technology is not yet
proven in actual chemical synthesis, it illustrates the directions that parallel
synthesis may take in the future.
From the discussion above it should be clear that parallel synthesis pro-
vides individual compounds for assay over a wide range of compound classes
(68), numbers (of compounds), and scale (mg of each compound), thus making
it suitable throughout much of the drug discovery process. Split-and-pool

© 2000 by Marcel Dekker, Inc.

methods and encoded synthesis methods, however, provide greater numbers
of compounds with less effort and thus may be more suitable than parallel
synthesis in the earliest phases of drug discovery.

V. POSTSYNTHESIS PROCESSING AND SAMPLE HANDLING

A. Format Changes and Liquid Handling
Much of the discussion and focus of combinatorial chemistry is on the synthe-
sis process itself. However, most practitioners in the field will attest that post-
synthesis processing is far more time consuming than synthesis and that post-
synthesis processing is often the rate-limiting step in an overall combinatorial
chemistry project. Early design of synthesis apparatus paid scant attention to
the format in which the products are delivered. For example, products deliv-
ered into test tubes may have to be individually moved to a concentrator to
remove solvent, then individually moved to a liquid handler for dissolution
and distribution to microtiter plates for bioassay. Each ‘‘format change’’
(change of vessel type or vessel carrier array) may require manual intervention
and thus will slow the overall process.
Transfer of samples from one vessel to another is generally done by
dissolving the sample in an appropriate solvent and then aliquoting a fixed
amount of solution from the first container into the second container. Commer-
cial liquid handlers are available to automate this process, but their effective-
ness is limited by the original format design. For example, transferring liquid
from vessels in an array to another set of identical vessels in an identical
array can be done using a simple multichannel pipetting device. This is most
commonly observed when transferring samples from one 96-well microtiter
plate to another. Transferring samples from an array into another nonidentical
array, however, may require transfer of single samples using a single-channel
pipettor. The situation is further complicated when the number of samples in
one array is not equal to the number of samples in the other. Consider, for
example, transferring samples from purification on a Zymark Benchmate solid
phase extraction apparatus into a 96-well microtiter plate. The Benchmate
delivers samples into 16 ⫻ 100 mm test tubes in a 5 ⫻ 10 array. The mi-
croplate has 96 wells, in an 8 ⫻ 12 array. The liquid handling protocol for
this format change will be complex, will not take optimal advantage of multi-
channel pipettors, and will result in a partially filled microplate. Other common
examples include delivery of samples for analysis into autosampler racks of
commercial analytical equipment—many of which are round or pie-shaped
and not amenable to loading by a multichannel liquid-handling robot. A recent

© 2000 by Marcel Dekker, Inc.

trend toward use of the common 96-well Cartesian format in autosampling
and fraction collection equipment will help to minimize format changes in the
future. In any event, thoughtful design of a postsynthesis work flow with mini-
mal format changes can greatly enhance throughput in any combinatorial
chemistry process. In particular, avoidance of non-Cartesian racks and carriers
facilitates optimal use of commercial multichannel liquid-handling devices.

B. Analysis
Analysis of combinatorial libraries is placing new and unprecedented demands
on analytical equipment and services. Chemically encoded libraries require
high-throughput analysis methods for library deconvolution, whereas parallel
synthesis procedures require analysis of individual library members. There
has been much discussion about whether it is necessary to analyze every mem-
ber of a combinatorial library, to analyze representative members only, or to
analyze active compounds only. It is probably safe to say, though, that if the
tools were available most scientists would prefer to collect as much analytical
data as possible. The problem, of course, is that while compounds are being
synthesized in parallel the most commonly used analytical methods i.e., gas
chromatography (GC), high-performance liquid chromatography (HPLC), and
mass spectrometry (MS) are inherently serial methods. Serial processing meth-
ods can only keep up with parallel synthesis methods if their speed, capacity,
and throughput are fully optimized. Optimization of these processes is the
topic of a later chapter in this volume. Alternately, development of suitable
parallel analysis methods may someday facilitate analysis of large combinato-
rial libraries. Thin-layer chromatography and capillary zone electrophoresis
may be particularly suited to parallelization for large-library analysis, though
reports of automated systems for combinatorial chemistry analysis are sparse.
One of the major future challenges for analytical chemistry in support of com-
binatorial synthesis will be to develop and optimize high-capacity high-speed
analysis methods and to develop the required hardware to support those
methods.

C. Purification
As the drug discovery process proceeds from early lead discovery through
lead optimization and toward candidate selection, the level of purity required
of each compound increases. Late-stage compounds may be tested in multiple
assays, many of which may involve animal models and may be labor-intensive.
Even the best solid phase combinatorial synthesis methods cannot assure that

© 2000 by Marcel Dekker, Inc.

every member of a synthesis library will be delivered in pure form directly
from the synthesis procedure. High-throughput automated purification meth-
ods are thus required to provide purified compounds for late-stage drug discov-
ery programs. Existing chromatographic methods generally fall into two cate-
gories: ‘‘collect-before-detect’’ and ‘‘detect-before-collect.’’ Collect-before-
detect methods are those where product collection is based on a predetermined
scheme and is not guided by real-time product detection. Such methods are
suitable only when the chromatographic properties of the products can be pre-
dicted with confidence. Applications that fall into this category include ion
exchange chromatography using solid phase extraction cartridges and appara-
tus (69). However, similar methods using pure adsorption chromatography
provide much less predictable elution and are less amenable to high-
throughput purification of a diverse combinatorial library. Detect-before-col-
lect methods rely on feedback from a real-time detection system to guide frac-
tion collection. With real-time detection it is possible to purify diverse libraries
with confidence that the products will be collected. A variety of detection
methods have been reported, including ultraviolet absorbance (70), evapora-
tive light scattering (71), and even single mass detection (72,73). While chro-
matographic methods are inherently serial in nature, solid phase extraction
can be done in parallel using common vacuum box technology. A four-channel
parallel preparative HPLC has recently been reported and will be offered com-
mercially in the near future (74). Once again, the challenge for analytical
chemistry in the future will be to provide purification methods and apparatus
with increasingly large capacity and throughput.

VI. DATA MANAGEMENT

One of the challenges in any combinatorial chemistry effort is managing the

large amount of data being generated. Efficient stand alone systems for manag-
ing structural information, analytical information, and so forth exist today, but
efficient combinatorial synthesis requires integration of data from a variety of
sources to provide complete tracking throughout the synthesis process. Com-
pletely integrated solutions do not exist commercially, and it may be impossi-
ble to provide one due to the variety of instruments and procedures that can
exist throughout the industry.
The data management process often begins with creation of a ‘‘virtual
library’’ of all possible structures that could be generated by combining avail-
able reagents with a core structure. The number of possible structures may run
into the millions or billions of compounds, thus requiring efficient structure

© 2000 by Marcel Dekker, Inc.

searching, generation, and display. From the virtual library, the chemist will
select (perhaps using automated methods) the targets that will actually be syn-
thesized and become part of the real library. This target list will then lead to
a list of reagents needed for the synthesis. Reagents will need to be ordered
from outside vendors or acquired internally. Since most automated synthesis
apparatus handle reagents in solution, reagents will then have to be weighed
and diluted with appropriate solvents to produce reagent solutions of standard
concentration. Reagent ordering, tracking, weighing, and dilution alone re-
quire both significant data handling and creation of instrument scripts (e.g.,
for weighing and diluting samples).
Once all materials are in place and the synthesis targets have been de-
fined, instrument control scripts for performing the actual synthesis must be
generated. While the actual form the scripts can take will depend on the instru-
ment(s) being controlled, all such scripts will require information about the
reagents and synthesis targets. Following synthesis, analytical data will be
acquired (either on the synthesis products themselves or on chemical tags used
for encoding) and data will have to pass to and from analytical instruments.
Finally, sample registration (into a corporate database) and distribution will
occur.
The challenge for instrument vendors is to provide sufficiently open
architecture to allow customers to import and export data seamlessly to and
from instruments. The challenge for combinatorial chemists is to develop pro-
tocols for transferring data from instrument to instrument through a seamless
data flow. This generally involves accepting a data ‘‘export’’ from one instru-
ment, performing calculations based on the data, and reformatting the data for
‘‘import’’ into the next instrument in the sequential work flow. The topic of
data management for combinatorial chemistry will be addressed in detail in
a later chapter in this volume.

VII. SUMMARY

The drug discovery process is a complex and dynamic continuum ranging

from early phase lead discovery, through lead optimization, to candidate selec-
tion. The field of combinatorial chemistry provides a continuum of methods
to support drug discovery. These methods range from split-and-pool synthesis
of large libraries, through encoded synthesis of medium size libraries, and to
parallel synthesis of smaller libraries of highly purified compounds. A success-
ful drug discovery program will match the combinatorial chemistry methods
used with the demands of the overall program at any point in time. It is indis-

© 2000 by Marcel Dekker, Inc.

putable that the advent of combinatorial chemistry has irreversibly changed
the drug discovery process and the way medicinal chemists think about mole-
cules (75). The challenge for analytical chemists is to understand the entire
process and provide a family of specific tools to support each phase of drug
discovery and each type of combinatorial chemistry that may be practiced.

REFERENCES

1. Review: F Balkenhoh, C von dem Bussche-Hunnefeld, A Lansky, C Zechel.

Combinatorial Synthesis of Small Organic Molecules. Angew Chem Int Ed Engl.
35:2288–2337, 1996.
2. Review: G Jung, AG Beck-Sickinger. Multiple peptide synthesis methods and
their applications. Angew Chem Int Ed Engl. 31:367–383, 1992.
3. Review: MA Gallop, RW Barrett, WJ Dower, SPA Fodor, EM Gordon. Applica-
tions of combinatorial technologies to drug discovery. 1. Background and peptide
combinatorial libraries. J Med Chem 37:1233–1251, 1994.
4. Review: EM Gordon, RW Barrett, WJ Dower, SPA Fodor, MA Gallop. Applica-
tions of combinatorial technologies to drug discovery. 2. Combinatorial organic
synthesis, library screening strategies, and future directions. J Med Chem 37:
1385–1401, 1994.
5. RA Houghten, C Pinilla, SE Blondelle, JR Appel, CT Dooley, JH Cuervo. Gener-
ation and use of synthetic peptide combinatorial libraries for basic research and
drug discovery. Nature 354:84–86, 1991.
6. CD Garr, JR Peterson, L Schultz, AR Oliver, TL Underiner, RD Cramer, AM
Ferguson, MS Lawless, and DE Patterson. Solution phase synthesis of chemical
libraries for lead discovery. J Biomol Screen 1:179–186, 1996.
7. SH DeWitt. Automated parallel purification methods. Solid and Solution Phase
Combinatorial Synthesis, New Orleans, 1997.
8. JP Whitten, YF Xie, PE Erickson, TR Webb, EB DeSouza, DE Grigoriadia, JR
McCarthy. Rapid microscale synthesis, a new method for lead optimization using
robotics and solution phase chemistry: application to the synthesis and optimiza-
tion of corticotropin-releasing factor receptor antagonists. J Med Chem 39:4354–
4357, 1996.
9. RM Lawrence, OM Fryszman, MA Poss, A Biller, and HN Weller. Automated
Preparation and Purification of Amides. Proceedings of the International Sympo-
sium on Lab Automation and Robotics, Zymark Corp., Hopkinton, MA, 1995
pp. 211–220.
10. NM Yoon, HJ Lee, JH Ahn, J Choi. J Org Chem 59:4687–4688, 1994.
11. JJ Parlow. Simultaneous multistep synthesis using polymeric reagents. Tetrahe-
dron Lett 36:1395–1396, 1995.
12. SW Kaldor, MG Siegel, JE Fritz, BA Dressman, PJ Hahn. Use of solid supported

© 2000 by Marcel Dekker, Inc.

 nucleophiles and electrophiles for the purification of non-peptide small molecule
libraries. Tetrahedron Lett 37:7193–7196, 1996.
13. DL Boger, CM Tarby, PL Myers, LH Caporale. Generalized dipeptidomimetic
template: solution phase parallel synthesis of combinatorial libraries. J Am Chem
Soc 118:2109–2110, 1996.
14. T Carell, EA Wintner, AB Hashemi, J Rebek Jr. A novel procedure for the syn-
thesis of libraries containing small organic molecules. Angew Chem Int Ed Eng
33:2059–2064, 1994.
15. CD Garr, JR Peterson, L Schultz, AR Oliver, TL Underiner, RD Cramer, AM
Ferguson, MS Lawless, DE Patterson. Solution phase synthesis of chemical li-
braries for lead discovery. J Biomol Screen 1:179–186, 1996.
16. RM Lawrence, SA Biller, OM Fryszman, MA Poss. Automated synthesis and
purification of amides: exploitation of automated solid phase extraction in or-
ganic synthesis. Synthesis 553–558, 1997.
17. HN Weller, MG Young, SJ Michalczyk, GH Reitnauer, RS Cooley, PC Rahn,
DJ Loyd, D Fiore, SJ Fischman. High throughput analysis and purification in
support of automated parallel synthesis. Mol Divers 3:61–70, 1997.
18. RB Merrifield. Solid phase peptide synthesis. I. The synthesis of a tetrapeptide.
J Am Chem Soc 85:2149–2154, 1963.
19. GB Fields, RL Noble. Solid phase peptide synthesis utilizing 9-fluorenylmeth-
oxycarbonylamino acids. Int J Peptide Protein Res 35:161–214, 1990.
20. LA Thompson, JA Ellman. Straightforward and general method for coupling
alcohols to solid supports. Tetrahedron Lett 35:9333–9336, 1994.
21. Review: JS Fruchtel and G Jung. Organic chemistry on solid supports. Angew
Chem Int Ed Eng 35:17–42, 1996.
22. Review: PHH Hermkens, HCJ Ottenheijm, D Rees. Solid-phase organic reac-
tions: a review of the recent literature. Tetrahedron 52:4527–4554, 1996.
23. SK Sarkar, RS Garigipati, JL Adams, PA Keifer. An NMR method to identify
nondestructively chemical compounds bound to a single solid-phase-synthesis
bead for combinatorial chemistry applications. J Am Chem Soc 118:2305–2306,
1996.
24. R Garigipati, B Adams, JL Adams, SK Sarkar. Use of spin echo magic angle
spinning 1H NMR in reaction monitoring in combinatorial organic synthesis. J
Org Chem 61:2911–2914, 1996.
25. B Yan, G Kumaravel. Probing solid-phase reactions by monitoring the IR bands
of compounds on a single ‘‘flattened’’ resin bead. Tetrahedron 52:843–848,
1996.
26. B Yan, JB Fell, G Kumaravel. Progression of organic reactions on resin supports
monitored by single bead FTIR microspectroscopy. J Org Chem 61:7467–7472,
1996.
27. TY Chan, R Chen, MJ Sofia, BC Smith, D Glennon. High throughput on-bead
monitoring of solid phase reactions by diffuse reflectance infrared Fourier trans-
form spectroscopy (DRIFTS). Tetrahedron Lett 38:2821–2824, 1997.
28. DE Pivonka, R Russel, T Gero. Tools for combinatorial chemistry: in situ infra-

© 2000 by Marcel Dekker, Inc.

 red analysis of solid-phase organic reactions. Appl Spectros 50:1471–1478,
1996.
29. BA Bunin, JA Ellman. A general and efficient method for the solid phase synthe-
sis of 1,4-benzodiazepine derivatives. J Am Chem Soc 114:10997–10998, 1992.
30. K Burgess, AI Liaw, N Wang. Statistical sampling of resin pools. J Med Chem
37:2985–2987, 1994.
31. P-L Zhao, RB Nachbar, JA Bolognese, KT Chapman. Two new criteria for
choosing sample size in combinatorial chemistry. J Med Chem 39:350–352,
1996.
32. KS Lam, SE Salmon, EM Hersh, VJ Hruby, WM Kazmierski, RJ Knapp. A new
type of synthetic peptide library for identifying ligand-binding activity. Nature
354:82–84, 1991.
33. RA Houghten, C Pinilla, SE Blondelle, JR Appel, CT Dooley, JH Cuervo. Gener-
ation and use of synthetic peptide combinatorial libraries for basic research and
drug discovery. Nature 354:84–86, 1991.
34. GC Look, JR Schullek, CP Holmes, JP Chinn, EM Gordon, MA Gallop. The
identification of cyclooxygenase-1 inhibitors from 4-thiazolidinone combinato-
rial libraries. Bioorg Medicin Chem Lett 6:707–712, 1996.
35. RA Houghten, C Pinilla, SE Blondelle, JR Appel, CT Dooley, JH Cuervo. Gener-
ation and use of synthetic peptide combinatorial libraries for basic research and
drug discovery. Nature 354:84–86, 1991.
36. B Deprez, X Williard, L Bourel, H Coste, F Hyafil, A Tartar. Orthogonal combi-
natorial chemical libraries. J Am Chem Soc 117:5405–5406, 1995.
37. KS Lam, VJ Hruby, M Lebl, RJ Knapp, WM Kazmierski, EM Hersh, SE Salmon.
The chemical synthesis of large random peptide libraries and their use for the
discovery of ligands for macromolecular acceptors. Bioorg Medicin Chem Lett
3:419–424, 1993.
38. C Chen, LA Ahlberg Randall, RB Miller, AD Jones, MJ Kurth. Analogous or-
ganic synthesis of small-compound libraries: validation of combinatorial chemis-
try in small-molecule synthesis. J Am Chem Soc 116: 2661–2662, 1994.
39. MC Needles, DG Jones, EH Tate, GL Heinkel, LM Kochersperger, WJ Dower,
RJ Barrett, MA Gallop. Generation and screening of an oligonucleotide-encoded
synthetic peptide library. Proc Natl Acad Sci USA 90:10700–10704, 1993.
40. JM Kerr, SC Banville, RN Zuckermann. Encoded combinatorial peptide libraries
containing non-natural amino acids. J Am Chem Soc 115:2529–2531, 1993.
41. HP Nestler, PA Bartlett, WC Still. A general method for molecular tagging of
encoded combinatorial libraries. J Org Chem 59:4723–4724, 1994.
42. MHJ Ohlmeyer, RN Swanson, LW Dillard, JC Reader, G Asouline, R Kobay-
ashi, M Wigler, WC Still. Complex synthetic chemical libraries indexed with
molecular tags. Proc Natl Acad Sci USA. 90:10922–10925, 1993.
43. HP Nestler, PA Bartlett, WC Still. A general method for molecular tagging of
encoded combinatorial libraries. J Org Chem 59:4723–4724, 1994.
44. P Eckes. Binary encoding of compound libraries. Angew Chem Int Ed Eng 33:
1573–1575, 1994.

© 2000 by Marcel Dekker, Inc.

45. JJ Baldwin, JJ Burbaum, I Henderson, MHJ Ohlmeyer. Synthesis of a small
molecule combinatorial library encoded with molecular tags. J Am Chem Soc
117:5588–5589, 1995.
46. ZJ Ni, D Maclean, CP Holmes, B Ruhland, MM Murphy, JW Jacobs, EM Gor-
don, MA Gallop. Versatile approach to encoding combinatorial organic syntheses
using chemically robust secondary amine tags. J Med Chem 39:1601–1608,
1996.
47. D Maclean, JR Schullek, MM Murphy, ZJ Ni, EM Gordon, MA Gallop. Encoded
combinatorial chemistry: synthesis and screening of a library of highly function-
alized pyrrolidines. Proc Natl Acad Sci USA 94:2805–2810, 1997.
48. A Borchardt, WC Still. Synthetic receptor binding elucidated with an encoded
combinatorial library. J Am Chem Soc 116:373–374, 1994.
49. RA Houghten. General method for the rapid solid-phase synthesis of large num-
bers of peptides: specificity of antigen–antibody interaction at the level of indi-
vidual amino acids. Proc Natl Acad Sci USA 82:5131–5135, 1985.
50. EJ Roskamp. The Design and Synthesis of 500,000 Compounds. Molecular Di-
versity and Combinatorial Chemistry. San Diego, October, 1996.
51. EJ Moran, S Sarshar, JF Cargill, MM Shahbaz, A Lio, AMM Mjalli, RW Arm-
strong. Radio frequency tag encoded combinatorial library method for the dis-
covery of tripeptide-substituted cinnamic acid inhibitors of the protein tyrosine
phosphatase PTP1B. J Am Chem Soc 117:10787–10788, 1995.
52. KC Nicolaou, X-Y Xiao, Z Parandoosh, A Senyei, MP Nova. Radiofrequency
encoded combinatorial chemistry. Angew Chem Int Ed Eng 34:2289–2291,
1995.
53. AW Czarnik. No static at all: using radiofrequency memory tubes without (hu-
man) interference. Laboratory Automation ’97. San Diego, January, 1997.
54. X-Y Xiao, C Zhao, H Potash, MP Nova. Combinatorial chemistry with laser
optical encoding. Angew Chem Int Ed Eng 36:780–782, 1997.
55. Advanced Chem Tech, 5609 Fern Valley Road, Louisville, KY 40228.
56. Argonaut Technologies, Inc., 887 Industrial Rd., Suite G, San Carlos, CA 94070.
57. Diversomer Technologies, 2800 Plymouth Road, Ann Arbor, MI 48105.
58. Bohdan Automation, Inc., 1500 McCormick Blvd., Mundelein, IL 60060.
59. Tecan U.S., SLT Lab Instruments, P.O. Box 13953, Research Triangle Park, NC
27709.
60. SH DeWitt, JS Kiely, CJ Stankovis, MC Schroeder, DW Reynolds-Cody, MR
Pavia. ‘‘Diversomers’’: an approach to nonpeptide, nonoligomeric chemical di-
versity. Proc Natl Acad Sci USA. 90:6909–6913, 1993.
61. SH DeWitt and AW Czarnik. Combinatorial organic synthesis using Parke-
Davis’s DIVERSOMER method. Acc Chem Res 29:114–122, 1996.
62. J Cargill and RR Maiefski. Methods and apparatus for the generation of chemical
libraries. U.S. Patent 5,609,826, 1997.
63. JF Cargill and RR Maiefski. Automated combinatorial chemistry on solid phase.
Lab Robot Autom 8:139–148, 1996.
64. HM Geyson, H Meloen, SJ Barteling. Use of peptide synthesis to probe viral

© 2000 by Marcel Dekker, Inc.

 antigen for epitopes to a resolution of a single amino acid. Proc Natl Acad Sci
USA 81:3998–4002, 1984.
65. BA Bunin, MJ Plunkett, JA Ellman. The combinatorial synthesis and chemical
and biological evaluation of a 1,4-benzodiazepine library. Proc Natl Acad Sci
USA. 91:4708–4712, 1994.
66. SPA Fodor, JL Read, MC Pirrung, L Stryer, AT Lu, D Solas. Light-directed,
spatially addressable parallel chemical synthesis. Science 251:767–773, 1991.
67. AV Lemmo, JT Fisher, HM Geysen, DJ Rose. Characterization of an inkjet
chemical microdispenser for combinatorial library synthesis. Anal Chem 69:
543–551, 1997.
68. JA Ellman. Design, synthesis, and evaluation of small-molecule libraries. Acc
Chem Res 29:132–143, 1996.
69. RM Lawrence, SA Biller, OM Fryszman, MA Poss. Automated synthesis and
purification of amides: exploitation of automated solid phase extraction in or-
ganic synthesis. Synthesis 553–558, 1997.
70. HN Weller, MG Young, SJ Michalczyk, GH Reitnauer, RS Cooley, PC Rahn,
DJ Loyd, D Fiore, SJ Fischman. High throughput analysis and purification in
support of automated parallel synthesis. Mol Divers 3:61–70, 1997.
71. CE Kibbey. Quantitation of combinatorial libraries of small organic molecules
by normal-phase HPLC with evaporative light scattering detection. Mol Divers
1:247–258, 1995.
72. DB Kassel. A fully automated mass spectrometry based system for the rapid
analysis and purification of combinatorial libraries. Solid and Solution Phase
Combinatorial Synthesis, New Orleans, 1997.
73. L Zeng, L Burton, K Yung, B Shushan, DB Kassel. An automated analytical/
preparative HPLC/MS system for the rapid characterization and purification of
compound libraries. J Chromatogr 794:3–13, 1998.
74. P Coffey. Parallel purification for combinatorial chemistry. Lab Autom News 2:
7–13, 1997.
75. J Ellman, B Stoddard, J Wells. Combinatorial thinking in chemistry and biology.
Proc Natl Acad Sci USA. 94:2779–2782, 1997.

© 2000 by Marcel Dekker, Inc.

2
The Use of Mass Spectrometry

Annette Hauser-Fang and Paul Voúros

Northeastern University
Boston, Massachusetts

Combinatorial chemistry has evolved as a major area of interest for the phar-
maceutical industry and has created a need for methods of characterization
that are distinctly different from the classical tools of analytical organic chem-
istry like nuclear magnetic resonance (NMR) and C, H, and N analysis (1,2).
There are currently several approaches for combinatorial chemistry that differ
from each other in many ways (3–14). One can generate large solvated mix-
tures of compounds with as many as 10 6 different components in one sample,
the so-called true libraries (15), or one can use the microtiter plate approach
that is used to synthesize similar numbers of compounds individually on mi-
crotiter plates within small wells, so that each well contains only one or very
few components. Clearly, for the second approach, synthesis and analysis must
be automated to be efficient. Alternatively, many combinatorial libraries have
been synthesized on solid supports (16–20) using polymeric beads. Advan-
tages for solid phase chemistry include more efficient and simplified sample
cleanup because washing of the beads usually removes all excess reagents and
byproducts. This is especially true for the so-called mix-and-split synthesis (21)
where each bead carries one component and which has been used with success
by synthetic chemists. Single-bead analysis has been a good match for matrix-
assisted laser desorption ionization (MALDI) and several publications have
shown that this approach can provide excellent data as is shown below.
Within the context of these new developments in drug discovery, mass

© 2000 by Marcel Dekker, Inc.

spectrometry (MS) has begun to assume a leading role (22,23) that, in addition
to the traditional analysis of synthetic samples, appears to enter into the exami-
nation of molecular recognition and screening for lead drug candidates. This
chapter focuses on a review of the use of MS techniques for the characteriza-
tion and the biological screening of combinatorial libraries, often in combina-
tion with on-line separation methods.

I. CHARACTERIZATION OF COMBINATORIAL LIBRARIES BY

MASS SPECTROMETRY

With the development of combinatorial chemistry it has become necessary

to develop methods of analysis that are applicable to the characterization of
components out of mixtures of molecules. It is often not sufficient to rely on
biological screening procedures and later identification of active compounds,
but it may also be necessary to identify synthetic products and verify the gener-
ation of a certain diversity within the combinatorial library. Establishing the
integrity of a library can be important in order to avoid false positives and/
or false negatives that may result from the presence of impurities (24) or the
absence of products anticipated in the synthesis. When opting for certain ap-
proaches, the overall effort, the cost of purchasing and maintaining expensive
equipment, and the superior contribution of certain methods toward the analy-
sis have to be considered. The following sections give an overview of the
applicability of some of the most popular techniques.

A. Low-Resolution CE-MS AND CE-MS/MS Methods Using

Triple Quadrupole Mass Spectrometers or Quadrupole Ion
Traps
A number of publications have focused on the characterization of combinato-
rial libraries by direct infusion, liquid chromatography (LC), or capillary elec-
trophoresis (CE) (25–29) coupled to electrospray ionization-mass spectrome-
try (ESI-MS), electrospray ionization-tandem mass spectrometry (ESI-MS/
MS) (30–37), nanoelectrospray-MS (38,39), and atmospheric pressure chemi-
cal ionization-mass spectrometry (APCI-MS) (40–42) using triple quadrupole
instruments. Results obtained by direct infusion of a library into an ESI-MS
system may be fraught with uncertainty even when dealing with small libraries
because of the low resolving power of the triple-quadrupole instrument. It is
not uncommon to find peaks at virtually every single mass unit depending on
the library size. In fact, even relatively small libraries with, for example, 30
components give rise to at least 60 peaks due to isotopes. Counting in a few

© 2000 by Marcel Dekker, Inc.

impurities one can easily confuse the spectrum and generate many false posi-
tives. Moreover, the ionization efficiencies of individual library components
may vary widely whether in the positive or negative ionization mode. Estab-
lishing a ‘‘hit’’ from a given ion mass signal may thus be a questionable
proposition and even more questionable can be the assessment of molecular
ratios of library constituents. Accordingly, determination of ionization effi-
ciencies of library members is necessary before the presence of nonisobaric
compounds can be fully established. The combined use of positive and nega-
tive ion detection may often provide considerable flexibility in the character-
ization of diverse libraries. Figure 1 shows an example by Dunayevskiy et al.
(27,30) where switching between positive and negative ionization can help
distinguish real library components from impurities. Ideally, under low resolu-
tion MS conditions prior to separation of the library mixture by CE or LC is
advisable along with the need to aquire MS/MS spectra of every peak. But
even CE-MS and LC-MS only help separate a mixture for the identification
of isobaric components. They do not provide a fingerprint of a compound and,
unless the retention time of the compound of interest is known, it is still possi-
ble that an isobaric impurity has been identified as a library component. In
low-resolution MS, fragmentation spectra are always the most definitive proof
of a compound’s existence. Figure 2 shows an application where the three
dimensional presentation of LC-MS and LC-MS/MS data is used to detect
structural trends within the eluted peaks, therefore permitting, for example,
the easy identification of library components belonging to certain subgroups
(43). On the other hand, it is frequently desirable to only determine the repro-
ducibility of the synthesis of a specific library. In such a case, the ESI-MS
profile may be deemed sufficient for general library characterization.
The new development of the quadrupole ion trap technology with its
possibility of automated recognition of peaks and subsequent fragmentation
using MS n has simplified the process of fingerprinting of components by frag-
mentation. It is possible to get completely automated MS n information from
every HPLC peak just on the basis of programming the instrument to fragment
the n th most intense ion in the spectrum until the MS n step. This is helpful for
unattended analysis runs (e.g., overnight) and improves the chances of com-
plete identification of components from their fragmentation spectra (44).
Theoretical and experimental results by Blom (45) have focused on the
precision requirements for the mass spectrometric analysis of combinatorial
library mixtures. In his work, Blom has used discrete mass filters in combina-
tion with mass-MS/MS, M ⫹ 1/M, and M ⫹ 2/M isotope ratio filters to
determine library components specifically from peptide libraries. As pointed
out below, while nonpeptide libraries have mass distributions that are com-
pletely random, peptide library building blocks are limited in their diversity,

© 2000 by Marcel Dekker, Inc.

Figure 1 Mass spectrometric analysis of combinatorial libraries L1–L6: gray, de-
tected in positive ion ESI; black, detected in negative ion ESI; vertical stripes, detected
in both modes; X, detected in MS/MS experiment. (Reprinted from Ref. 30.)

© 2000 by Marcel Dekker, Inc.

Figure 2 (Top) 3-D surface rendering of a full scan LC-MS dataset obtained on
a 625-combinatorial-peptide library. (Middle) 3-D extracted ion surface of m/z 496,
indicating three major regions. (Bottom) 3-D software neutral loss (NL-264) surface,
indicating components that contain substructure V-F-COOH. (Reprinted from Ref. 43.)

and this leads to clustering of the library components around certain m/z
values. The libraries used in this example were complex and in the best case
only 6% of their components could be characterized by their integer masses.
Use of a mass filter alone clearly produced too poor a fingerprint for unambigu-
ous characterization of all components. Therefore all available mass spectro-
metric filters were made to work in a complementary fashion in order to recog-
nize impurities and distinguish peptide components from one another. For the

© 2000 by Marcel Dekker, Inc.

experimental part of the work by Blom, a software-modified triple-quadrupole
mass spectrometer was employed that allowed automated switching between
MS and MS/MS modes whenever the peak intensity of an expected candidate
peak was of such a level as to trigger fragmentation.
Automation (37,41,42,46–53) has not only enabled researchers to ana-
lyze their library components more efficiently but has also helped in the syn-
thetic process. Many companies (50,54) have adopted the approach of synthe-
sizing one or two components of a library separately in a small well using
microtiter plates as reaction vessels and robots for the addition of reagents and
mixing. The microtiter plates with the individual library components are then
automatically analyzed by low resolution MS (54) or high-resolution Fourier
transform ion cyclotron-mass spectrometry (FTICR-MS) (55). Most setups
include HPLC equipment with an autosampler to do continuous flow injections
and molecular weight analysis of the contents of each well on the microtiter
plate. No MS/MS is needed because one mass filter is usually enough for the
confirmation of the one expected molecular weight and maybe a few impurity
peaks. Samples can be analyzed overnight for maximum efficiency. A typical
setup is shown in Fig. 3 (56). However, it should be pointed out that the
absence of additional peaks in the mass spectrum does not guarantee 100%
sample purity in the titer well. The possibility of false negatives always needs
to be addressed by an alternative method of detection because some compo-
nents (or impurities) may be transparent to ESI ionization.

Figure 3 High-throughput QA schematic. (Reprinted from Ref. 56.)

© 2000 by Marcel Dekker, Inc.

B. Matrix-Assisted Laser Desorption Ionization and Other
Desorption Ionization Techniques for the Identification of
Combinatorial Libraries
There have been many publications focusing on matrix assisted laser desorp-
tion ionization (MALDI) (49,57–59, 60–72), secondary ion mass spectrome-
try (SIMS) (73–76), and 252 Cf Plasma desorption mass spectrometry (77–79)
as a tool for the analysis of combinatorial libraries using different kinds of
mass spectrometers. MALDI is generally not an ionization technique that can
be easily coupled to any separation method but, since many libraries today
consist of physically separated compounds on beads or within wells on micro-
titer plates, separation is often not necessary. It is important though, to under-
stand the limitations and advantages of the MALDI approach because only
libraries with relatively high molecular weight components (m/z ⬎ 500) are
suitable for analysis due to matrix interferences at the lower mass ranges in
the spectrum. That may be one of the reasons why a significant part of the
research using MALDI has focused on the analysis of peptide libraries (58–
63, 80–83). MALDI has been used with both time-of-flight (TOF) (65,75) and
Fourier transform ion cyclotron resonance mass spectrometry (49, 84) for the
analysis of split pool and encoded combinatorial libraries (66–68,70,85,86).
It has been shown by Egner et al. (64,65) that it is possible to analyze
compounds directly from prepared polystyrene beads. This approach has been
useful to verify and improve the synthetic method for the generation of combi-
natorial libraries. Assuming that there are approximately 10 6 beads in 1 gram
of polystyrene resin, and the substitution factor for the solid phase chemistry
is 0.4 mmol/g or more, there is about 400 pmol of compound attached to
each single bead. MALDI-TOF analysis has detection limits in the femtomole
region, which makes it compatible with single-bead analysis (76).
In preparation for the experiments, the library components were re-
moved from the bead by exposing it to trifluoroacetic acid vapor for 30 min.
After the cleavage reaction was complete an internal standard was added to-
gether with the matrix solution for the laser desorption. The matrix was formed
around the bead within 15–30 min and MALDI-TOF analysis was performed
directly from the sample well. Results are shown in Fig. 4 where a variety of
peptides was monitored in addition to bradykinin, which acted as the internal
calibrant. As for all MALDI experiments, it was critical that the right matrix
be chosen for the analysis. After initial examinations of different matrices
under a stereo microscope followed by the MALDI-TOF experiment, dihy-
droxy-benzoic acid (DHP) was found to produce the largest crystals and the
best results.

© 2000 by Marcel Dekker, Inc.

Figure 4 (a) Analysis of dansyl-Ile-Thr(O Bu t)-Pro-Gln-Trp-Lys(Boc)-Wang-
linker-resin, giving peptide dansyl-Ile-Thr-Pro-Gln-Trp-Lys (MH ⫹, 1006.0). The peak
at 772.6 represents a small amount of undansylated material. (b) Analysis of Fmoc-
Cys(Trt)-Lys(Boc)-Ile-HMPB-linker-resin, giving Fmoc-Cys-Lys-Ile (MH ⫹, 585.5),
Fmoc-Cys(Trt)-Lys-Ile (MH ⫹, 849.8), and Fmoc-Cys-Lys-Ile disulfide [(M-S-S-
M)H ⫹, 1168.3]. (c)Analysis of the disulfide of Fmoc-Cys-Asn-Cys-Lys(Boc)-Ile-
HMPB-linker-resin giving the disulfide of Fmoc-Cys-Asn-Cys-Lys-Ile (MH ⫹, 801.0).
(Reprinted from Ref. 64.)

© 2000 by Marcel Dekker, Inc.

 Youngquist et al. (62,63) went one step further with their resin bead
analysis by not only getting molecular weight information on their peptide
based library components but sequencing information as well in the same ex-
periment. Their strategy employed capping reagents at each step of the library
synthesis so that beads would carry not only the library component but also
a small percentage of peptides of different chain lengths. Sequencing informa-
tion was obtained from the mass differences between termination products.
Figure 5 shows the principle of the generation of the peptide libraries and the
use of the capping reagents at every step. Termination products ideally differed
by one amino acid (87).

Figure 5 Termination synthesis method to produce sequence-specific families of

peptides on each resin bead in the library. After screening, active peptides are isolated
and the peptide products are released from the bead (—●) by CNBr digestion. (Re-
printed from Ref. 62.)

© 2000 by Marcel Dekker, Inc.

 Other important factors for a successful analysis have included the use
of a linker between the bead and the synthesized peptides. This ensured effi-
cient release from the bead and increased the molecular weight of the analytes
to m/z values greater than 500, so that there was no interference with the low
molecular weight noise that was produced by the desorption of the MALDI
matrix. With the linker in place, the method has been sensitive enough to use
only 1% of the material from an 88-µm bead for sequencing. Decreasing the
bead diameter makes automated library sorting possible. Results were obtained
from beads as small as 17 µm in diameter. The sequencing speed that was
obtained using this method was 25–30 residues per hour and 77 of 80 beads
were sequenced successfully with only one instance where the order of two
amino acids could not be determined.
For both of the above examples it was important that the method could
identify byproducts. When the reaction did not go to completion, so called
deletion peptides were generated that were identified from their mass differ-

Figure 6 MALDI mass spectrum of an unknown peptide. A 5% portion of the prod-

ucts isolated from a single bead was analyzed. (Reprinted from Ref. 62.)

© 2000 by Marcel Dekker, Inc.

Figure 7 Direct analysis of reaction products from beads without the use of prior
cleavage reactions. (Reprinted from Ref. 71.)

ences compared to other known peaks. Those specific mass differences also
pointed towards modifications that happened during the synthesis, enabling
the synthetic chemist to optimize conditions. A MALDI mass spectrum of an
unknown peptide isolated from a single bead is shown in Fig. 6. Only 5% of
the generated products were used for the MALDI analysis.
Future routine uses of MALDI mass spectrometry for the detection of
combinatorial library components could include techniques that enable the
analytical chemist to directly analyze reaction products from beads without
using prior cleavage reactions, as is shown in Fig. 7. This means that standard
linker molecules would have to be designed in such a way that cleavage from
the bead is obtained by the laser irradiation used for the ionization process as
has been employed by Oda et al. and others (71,78) for the identification of
peptides bound to a resin (72).

C. Fourier Transform Ion Cyclotron Resonance-Mass

Spectrometry
Fourier transform ion cyclotron resonance mass spectrometry (FTICR-MS)
has been used for years by the petroleum industry to characterize large mix-
tures of compounds. Both the high mass accuracy and resolution (m/∆m of

© 2000 by Marcel Dekker, Inc.

20,000–40,000) that is routinely achieved on these instruments can help iden-
tify components of a mixture simply by comparing their measured mass to
their theoretical m/z value (89). An ideal base for the analysis of combinatorial
libraries, this technique has been applied by several research groups
(49,55,84,90–96). In one recent example by Fang et al. (90,91), combinatorial
libraries consisting of mixtures of 36, 78, and 120 components were identified
using ESI-FTICR-MS. The three libraries were analyzed in both positive and
negative ionization modes. Depending on the acidic or basic character of the
library constituents they were ionized preferentially either in the positive or
the negative ionization mode, although, a number of components were seen
in both ionization modes. Figure 8a shows a negative ionization ESI spectrum
of the 36-component library with some of its assigned peaks (X stands for the
core molecule). Compounds were identified with mass errors of 5 ppm or less.
The spectrum of the 120 component library in Fig. 8b shows three isobaric
components that could be resolved easily with this high resolution technique.
It was possible to not only distinguish the library constituents from each other
but also to identify isotope and impurity peaks. The method is very powerful
but has limitations with respect to the number of ions that can be injected into
the ICR cell at a given time. Overloading of the ICR cell gave rise to space–
charge effects that compromised resolution and mass accuracy. Even though
it should be possible to theoretically inject several thousand components at
the same time, it was difficult to find conditions that would allow every library
component to be ionized efficiently without overloading the ion trap. The solu-
tion to this problem may be found in using a segmented approach where certain
mass ranges are analyzed while impurity peaks and peaks from other mass
ranges are ejected from the trap.
In a publication by Winger and Campana (94), ESI-FTICR was applied
to selectively identify several components from an octapeptide mixture. All
of the 17 expected peaks were observed by direct infusion using ESI with an
average resolving power of m/∆m ⫽ 25,000. It was possible to distinguish an
isobaric impurity from one of the peptide components by performing accurate
mass measurements. Comparisons to some low-resolution LC-MS data gave
clear indications of the advantages of FTICR.
Other approaches by Nawrocki et al. to the use of FTICR-MS for the
identification of combinatorial libraries have focused on comparing the theo-
retical and observed mass distributions of a synthetic mixture to provide the
possibility for a quick check on the progress of a reaction. If the envelope of
masses was identical to the theoretical distribution, the assumption was made
that the synthesis was successful. It has become increasingly important to
verify the existence of expected combinatorial chemistry products by compar-

© 2000 by Marcel Dekker, Inc.

Figure 8 (a) Negative ion ESI spectrum of the 36-component library sample show-
ing the ppm differences between theoretical and found m/z values for the assigned
peaks. (b) Negative ion ESI spectrum of the 120-component library sample: resolving
nominally isobaric peaks. (Reprinted from Ref. 91.)

© 2000 by Marcel Dekker, Inc.

ing the results of the analysis with a computer-simulated spectrum. The match
between the theoretical and the experimental values is a measure of the com-
pletion of the synthesis. Figure 9 shows the obviously very different outcome
of a synthesis believed to generate a peptide library that included a phosphory-
lated tyrosine and the simulated spectrum. The mass envelope of the simulated
spectrum is shifted to m/z values that are about 300 Da higher than those of
the experimental spectrum. Computer simulation with automated recognition
and interpretation of results is becoming an essential tool for the analytical
chemistry community and is especially useful when dealing with the very
complex direct infusion spectra of combinatorial libraries.

Figure 9 (a) ESI-FTICR broad-band mass spectrum of a library thought to be H-

Gly-pTyr-Xxx-Xxx-Xxx-Cys-OH where Xxx can be any one of the naturally occurring
amino acids with the exception of Cys and Trp. (b) A simulated mass spectrum of H-
Gly-pTyr-Xxx-Xxx-Xxx-Cys-OH. (c) A computer-simulated spectrum of library
H-Xxx-Xxx-Xxx-Cys-OH where Xxx can be any one of the naturally occurring amino
acids with the exception of Cys and Trp. (Reprinted from Ref. 92.)

© 2000 by Marcel Dekker, Inc.

D. Predicting Mass Spectra of Combinatorial Libraries
Steinbeck et al. (97) have used a computer generated system to predict the
diversity of combinatorial libraries. Using their MASP program, they were
able to create libraries with ideal peak overlap in their mass spectra. The pro-
gram can be employed to optimize and adjust the synthetic process for the
creation of true libraries by choosing certain building blocks for the generation
of an ideal mass envelope. By minimizing peak overlap, the simultaneous
monitoring of reaction products by direct infusion MS or MALDI becomes
more realistic, especially for larger libraries. MASP creates theoretical mass
distributions calculated from molecular weights of the components and the
isotope abundances of the elements contained in the combinatorial library
compounds. In addition, the program can adjust its mass envelope according
to the input of parameters such as the desorption factor, which is determined
experimentally for MALDI. In order to minimize peak overlap and generate
libraries that are well suited for mass spectrometric library screening, a ranking
system is established whereby those building blocks that produce the least
peak overlap together are displayed in a list of suggested reactants (48,98,99).
Other theoretical considerations by Zubarev et al. (100) involving pep-
tide mixtures as model compounds have led to a proposed accuracy require-
ment of ⫾1 ppm for the positive identification of certain peptides that have
the same nominal masses. Monoisotopic masses of peptides are not randomly
placed throughout a spectrum but rather concentrated around specific points
that can be theoretically predicted. An equation by Mann (101) gives the posi-
tions of the monoisotopic masses of peptides as Mp ⫽ [M] ⫹ 0.00048M (Da)
where [M] is the lower integer value of the molecular mass M. Figure 10
shows the theoretical distribution of peptides at a monoisotopic molecular
mass of [M] ⫽ 1000 Da. The spectrum contains about 50,000 peptides. Of
the components, 95% are included within a range of 0.33 ⫾ 0.01 Da, which
can be calculated from Wp ⫽ 0.19 ⫹ 0.0001M (Da), where Wp is the width
that includes 95% of all amino acid compositions. Modified amino acids with
known modifications can be included in such simulations and are also applica-
ble to the analysis of mixtures of other biopolymers or combinatorial libraries.

II. SCREENING FOR BIOLOGICAL ACTIVITY

Screening combinatorial library mixtures for binding affinities to receptors

involves analytical methods that are closely related to mass spectrometric epi-

© 2000 by Marcel Dekker, Inc.

Figure 10 Number of possible amino acid compositions (peptide combinations) as
a function of the peptide monoisotopic molecular mass for [M] ⫽ 1000 Da ([M] is
the nominal molecular mass, i.e., lower integer mass value). The histogram is built
with a 10–mDa step. The top density of the distribution is ⬃230 peptide compositions
per mDa or per ppm. (Reprinted from Ref. 100.)

tope mapping (102–104) and other receptor–ligand binding studies. The li-
brary mixture is incubated with a protein or other molecule of interest to form
complexes with a binding ligand. The complexes are then separated from the
nonbinding small molecules by size exclusion chromatography on a gel filtra-
tion column, by ultrafiltration, or by some other method. The purified com-
plexes are dissociated and the released ligands are analyzed by CE-MS or LC-
MS (79). Depending on what needs to be accomplished in the screening step,
it is possible to simply measure the molecular weight of ligands that bind to
a receptor or to get further sequencing information from MS/MS. There are
a number of research groups that have screened combinatorial library compo-
nents against vancomycin, benzodiazepine antibodies, and other receptors
(12,24,31,68,79,103–131). Others have identified enzyme substrates among

© 2000 by Marcel Dekker, Inc.

combinatorial library components (20,88,132–134). Some examples are given
below.

A. Affinity Chromatography–Mass Spectrometry

Results based on HPLC-MS involving benzodiazepine libraries (135) of 19
and 20 components have been published by Nedved et al. (105). A protein
G column was employed for immobilizing benzodiazepine antibodies. After
injecting the benzodiazepine mixture, the benzodiazepine–antibody com-
plexes that were formed were eluted from the protein G column onto two
different reverse phase columns. Selected benzodiazepines were then analyzed
using an on-line mass spectrometer. Comparison of fragmentation spectra of
the known and unknown benzodiazepines resulted in the identification of an
analog of chlordiazepoxide as an active constituent of the unknown 20-compo-
nent library. This technique of immunoaffinity chromatography/LC/LC/MS
(IAC/LC/LC/MS) together with IAC/LC/LC/MS/MS has been very useful
for developing automated screening procedures. Kelly et al. have used compa-
rable setups to screen for SH2 –ligand interactions (120,127).
Similar libraries have been used by Wiebolt et al. in their studies of
immunoaffinity ultrafiltration coupled on-line to mass spectrometry. Overcom-
ing certain limitations of IAC that may result from the possible alteration of
binding characteristics of a receptor after the immobilization process, Wiebolt
et al. incubated receptor and ligands in solution prior to the analysis. Ideally,
any receptor–ligand binding assay should take place in solution in order to
generate an environment that is similar to in vivo conditions. Noncovalent
immunoaffinity complexes were formed between antibodies and benzodiaze-
pine libraries and were separated from nonbinding components by centrifugal
ultrafiltration using a 50,000 Da molecular weight cutoff filter. The complexes
were then dissociated by lowering the pH and the ligands were analyzed by
HPLC–ion spray MS. This general approach was extended for screening other
small-molecule library/receptor preparations in solution.
Even though closer to in vivo conditions than IAC and a good approach
for screening receptor–ligand interactions from solutions, immunoaffinity ul-
trafiltration methods needed to be modified for situations where the availability
of the receptor molecule is extremely limited and/or the price is prohibitive
because at this point there is no possibility of receptor recovery after the bind-
ing studies are completed. Van Breemen et al. (136) have used a similar experi-
mental setup with their pulsed ultrafiltration MS design using a chamber with
a 10,000-Da molecular weight cutoff membrane that enabled them to avoid
the problem of receptor loss after each screening analysis. In some of their

© 2000 by Marcel Dekker, Inc.

Figure 11 Scheme showing the use of pulsed ultrafiltration–MS for screening a
combinatorial library for compounds that bind to a macromolecular receptor. The re-
ceptor is trapped in solution by an ultrafiltration membrane, which allows low molecu-
lar weight solution-phase compounds in a ‘‘pulse’’ of a combinatorial library to pass
through. After unbound compounds are washed away, the ‘‘hits’’ in the library are
eluted from the chamber by destabilizing the ligand–receptor complex using methanol,
pH change, etc. (Reprinted from Ref. 136.)

experiments, receptor and ligands were directly injected into the ultrafiltration
chamber at low concentrations and binding molecules were extracted from
solution by the receptor that was present. Figure 11 shows the general principle
of this approach. Their studies have shown that with the model receptor adeno-
sine deaminase it was possible to repeat the binding and release process of
their model ligand erythro-9-(2-hydroxy-3-nonyl)adenine (EHNA) three times
without adding additional receptor to the chamber. Experiments with human
serum albumin (HSA) as a receptor and warfarin as a ligand reproduced this
result. The HSA in the ultrafiltration chamber was used for three consecutive
binding experiments. The technique was used to identify the highest affinity
ligand from a combinatorial library of 20 components.

B. Affinity Capillary Electrophoresis–Mass Spectrometry

The principle of affinity capillary electrophoresis (ACE) is based on the differ-
ences in electrophoretic mobilities of the ligand(s) and the receptor–ligand
complex. Assuming that the receptor is neutral and not migrating at the chosen
buffer pH, the electrophoretic mobility of the ligand would be ε ⫽ z/m 2/3,
where z is the charge and m the molecular mass of the ligand. After complex-
ation with the receptor the electrophoretic mobility of the complex changes

© 2000 by Marcel Dekker, Inc.

to ε ⫽ z/(M ⫹ m) 2/3, where M is the molecular weight of the receptor. Since
the electrophoretic mobility of a ligand will always depend on its participation
within the receptor–ligand complex, an exact estimate of a possible migration
time cannot be made except that non–binding ligand molecules are not re-
tained and migrate as fast as without receptor. Ligands with strong binding
affinities are retained according to their interactions with the receptor.
Specific examples for ACE-MS by Chu et al. (108–110,137–140) have
shown that peptide libraries of 100 components can be screened for receptor–
ligand binding within a CE capillary coupled on-line to MS. Vancomycin,
which was used as the receptor in this case, was ‘‘immobilized’’ in a fused
silica capillary that was coated with a neutral hydrophilic polymer coating to
prevent electroosmotic flow during the analysis. The buffer pH was chosen
in such a way that the vancomycin remained neutral and did not migrate within
the capillary and into the mass spectrometer. The 100-peptide Fmoc DDXX
library was injected, and unretained peptides emerged at migration times of
3–4 min, as can be seen in Fig. 12. ACE-UV confirmed the presence of two
peaks with much longer retention times of 6–7 min, a good indication of strong
binding affinities by the corresponding peptides, but only on-line MS revealed
that the peak at 7 min contained two coeluting peptides. Compared to the
known ligand of vancomycin that had a migration time of 5 min in this experi-
ment, the two identified binding ligands in the peak at 7 min had higher bind-
ing affinities to vancomycin (26).
Additional information on the use of CE can be found elsewhere in this
volume (chap. 6).

C. Automated Reaction Screening Using Flow Injection

ESI-MS
An approach that employs both high sample throughput and screening without
separation is pointed out by Wu et al. (118) in their publication involving the
rapid screening of combinatorial libraries of inhibitors of enzymatic reactions.
By coupling an autosampler to ESI-MS they were able to develop a screening
method that was capable of identifying one potential enzyme inhibitor every
2 min. The experimental setup was as follows: reaction solutions containing
reactants, the corresponding enzyme for the reactant, a suspected inhibitor
from the library of inhibitor components, and an internal standard were loaded
individually into an autosampler tray. The autosampler was connected on-line
to a single-quadrupole MS using ESI and was programmed to do continuous
flow injections every 2 min. In order to determine enzyme inhibition, product
ions were analyzed and their intensity was compared in to the internal stan-

© 2000 by Marcel Dekker, Inc.

Figure 12 Affinity capillary electrophoresis–mass spectrometry (ACE-MS) of a
synthetic all-D, Fmoc-DDXX library of 100 tetrapeptides using vancomycin as the
receptor. (A–D) Selected ion electropherograms for the masses indicated. (E) Recon-
structed ion electropherogram for runs without (left) and with (right) vancomycin
in the electrophoresis buffer. ACE conditions: capillary, 360 µm o.d. ⫻ 50 µm i.d. ⫻ 38
cm long, coated with a neutral hydrophilic polymer; buffer, 20 mM Tris acetate (pH
8.1) containing no receptor (left) or 70 µM vancomycin (right); electric field, 500 V/
cm, 5 µamp; sample, hydrodynamically loaded, 10 cm/8 s/⬃10 nl at the negative end
of the capillary. MS conditions: instrument, Finnigan TSQ-700 with Finnigan API
interface operated in positive electrospray ionization (ESI) mode; ESI, ⫹ 4.2 kV; gas
sheath: 840 cm/min; liquid sheath (2 µl/min), 10 mM Tris acetate (pH 8.1) in H 2 O/
MeOH (25 :75, v/v); MS, scan range, 525–925 amu scan in 2 s. (Reprinted from Ref.
108.)

© 2000 by Marcel Dekker, Inc.

dard. The least amount of product was generated in three vials that contained
the best inhibitory components.

D. Surface Enhanced Affinity Capture and Probe Affinity

Mass Spectrometry
One of the advantages of matrix-assisted laser desorption ionization (MALDI)
is its sensitivity. As described above, MALDI can be extremely useful for
the identification of solid phase combinatorial library components and even
sequencing of peptide libraries using a single bead. Therefore, it is also a
useful technique for screening combinatorial libraries where sample quantities
do not permit the application of immunoaffinity HPLC-MS or ultrafiltration
techniques with high flow rates, limited sensitivity, and significant consump-
tion of receptor. Specific examples of applying MALDI to the identification
of high-affinity ligands have been published by Hutchens and Yip (125) with
their use of surface enhanced affinity capture (SEAC) and others. (128,129).
In this approach, a receptor molecule is immobilized on agarose beads, and
the analyte is captured directly out of solution. For this first published example,
lactoferrin was detected from preterm infant urine by the addition of prepared
beads to the urine solution. After removing and washing, the beads were
placed on the MALDI target and the ligands detected. Other examples have
shown that it is possible to directly detect an anti-monoclonal antibody to
cytochrome c from cytochrome c immunoaffinity column (IAC) material.
These experiments are useful for concentrating the analyte and providing af-
finity screening but some interference with the capture medium e.g., the IAC
material or the agarose beads, has been observed.
In an alternative approach, Brockman et al. (121) have worked around
these problems by immobilizing their analytes directly onto the surface of the
MALDI probe (141). Using disposable MALDI probe tips coated with gold,
they generated self-assembled monolayers that were then used to immobilize
antibodies directly onto the tips. Figure 13 shows how the dextran based probe
affinity mass spectrometry (PAMS) surface can screen for binding ligands.
The binding molecules in this case were anti-γINF antibodies that were
attached to the dextran. It was possible to overcome the problem of nonspecific
electrostatic interactions between γINF and the immobilized anti-γINF anti-
bodies by increasing the number of binding sites through oxidation of the
dextran molecules and subsequent immobilization of the anti-γINF antibodies,
thus avoiding generation of carboxyl groups. This approach may not be appli-
cable to all kinds of analytes but, if both the immobilization chemistry and the

© 2000 by Marcel Dekker, Inc.

Figure 13 Mechanistic illustration of the probe-based affinity separation. (Reprinted
from Ref. 121.)

analyte recognition work, it can be very useful for trace analysis and analyte
enrichment on the MALDI target.
Youngquist et al. (63) have used color indicators for the identification
of beads carrying active components. Bead-bound combinatorial libraries were
subjected to a selection procedure that left dark blue stains on those peptide-

© 2000 by Marcel Dekker, Inc.

containing beads that were binding to the selected antibody. Eight stained
and five nonstained beads were chosen from the affinity screening experi-
ment; then the bound peptides were digested and analyzed by MALDI as de-
scribed in Sec. II.B. The known recognition sequence of the antibody was
found in six of the peptide samples isolated from the eight stained beads.
Fewer matches were obtained for the two samples derived from the other two
stained beads and no matches were found for samples originating from un-
stained beads.

E. Bioaffinity Characterization–Mass Spectrometry

A method for screening receptor–ligand interactions has been introduced by
Smith et al. and Ganem et al. (107,142–146) with their approach of bioaffinity
characterization–mass spectrometry (BAC-MS). The method relies on the gas
phase interactions present between the receptor and the binding ligands when
using ESI. Both ESI and MALDI are soft ionization techniques that produce
stable gas phase ions of large molecules without fragmentation. It is also possi-
ble to see noncovalent complexes in the mass spectrometer using these tech-
niques (147), but recent research has focused on the specificity of the moni-
tored noncovalent or electrostatic interactions that occur in the gas phase and
many questions have remained open as to whether solution and gas phase
complexes should be compared. Nevertheless, in an interesting experiment, a
receptor and components from a combinatorial library were incubated in solu-
tion and after equilibration the liquid was electrosprayed directly into an
FTICR high-resolution mass spectrometer. The FTICR instrument allowed the
accumulation of selected ions or ion complexes in its ion trap and thereby
provided a separation step. The noncovalently complexed molecules were then
dissociated by applying additional energy. The remaining ligands were re-
tained in the ion trap and were subjected to further MS/MS studies for charac-
terization. Figure 14 shows the principle of BAC-MS. Bradykinin and bovine
ubiquitin were used as model components.
Additional information on library screening can be found elsewhere in
this volume (Chap. 8).

III. FUTURE DIRECTIONS FOR THE APPLICATION OF MASS

SPECTROMETRY IN COMBINATORIAL CHEMISTRY

The most important reason for the success of combinatorial chemistry is the
much greater efficiency when synthesizing, analyzing, and screening compo-

© 2000 by Marcel Dekker, Inc.

Figure 14 A conceptual representation of the BAC-MS technique. (a) The mixture
solution is ionized by electrospray ionization and the complex of interest is selectively
accumulated in the FTICR trap. (b) The noncovalent complex is then ‘‘heated’’ to
liberate the affinity ligand species. (c) The ligand ions are retained for further CAD
studies to present structural information. ‘‘T’’ represents the target biomolecules, and
‘‘L’’ represents ligands that form high-binding-affinity-specific noncovalent complexes
with the target molecules. Circles (solid and empty) represent ions from other mole-
cules present in the mixture. (Reprinted from Ref. 142.)

© 2000 by Marcel Dekker, Inc.

nents. Combinatorial chemistry with its possibility of automation has been
found to increase sample throughput per person by a factor of 8 or more since
the early 1990s (7) and robotic high-throughput systems are mainly responsi-
ble for this dramatic increase (148–150). For even higher sample throughput
it is desirable to use a setup for automated synthesis directly together with MS.
The analysis can be extended to include other characterization and screening
methods like ultraviolet (UV) and nuclear magnetic resonance (NMR) spec-
troscopy (56,151). Zuckerman et al. (79) used an equimolar peptide mixtures
(EPM) synthesizer to generate libraries of up to 36 individual peptides or
peptide mixtures. Each well was screened in an enzyme-linked immunosor-
bent assay (ELISA) and the identified binding ligands were then characterized
in a glycerol matrix by secondary ion MS. Boutin et al. (78) have constructed
a Zymark-based robot to synthesize their combinatorial libraries with subse-
quent analysis by MS, CE, and NMR.
For higher resolving power and smaller sample quantities it may be
desirable to use a capillary electrochromatography (CEC) system as has been
done by Zweigenbaum et al. (44) for their separation of a mixture of steroids.
The CEC system was coupled on-line to an ion trap mass spectrometer. CEC
is especially useful for the separation of isomers.
Other examples of future developments in combinatorial chemistry re-
lated to MS include the use of animals within pharmacokinetic studies
(152,153). The coadministration of several components to one animal could
dramatically speed up the discovery process for metabolites (154,155). Also,
automated combinatorial degradation profiling has been shown to add another
dimension to the automated processes of mixture synthesis and analysis
(98,156,157).

REFERENCES

1. AM Thayer. Special report: Combinatorial chemistry becoming core technol-

ogy at drug discovery companies. C&EN (February 12):57–64, 1996.
2. JH Krieger. Special report: Combinatorial chemistry spawns new software sys-
tems to manage flood of information. C&EN (February 12):67–73, 1996.
3. SH DeWitt. Molecular diversity strategies. Pharmaceut News 1(4):11–14,
1994.
4. E Gordon, M Gallop, D Patel. Strategy and tactics in combinatorial organic
synthesis. Applications to drug discovery. Acc Chem Res 29:144–154, 1996.
5. L Thompson, J Ellman. Synthesis and applications of small molecule libraries.
Chem Rev 96:555, 1996.

© 2000 by Marcel Dekker, Inc.

 6. NK Terrett, M Gardner, DW Gordon, RJ Kobylecki, J Steele. Combinatorial
synthesis: the design of compound libaries and their application to drug discov-
ery. Tetrahedron 51(30):8135–8173, 1995.
7. Intelligent drug design. Nature 384(Suppl 6604):1–26, 1996.
8. A Czarnik. Special issue on combinatorial chemistry (March). Acc Chem Res
29, 1996.
9. AL Burlingame, RK Boyd, SJ Gaskell. Mass spectrometry. Anal Chem 68:
599R–651R, 1996.
10. F Balkenhohl, Cvd Bussche-Hunnefeld, A Lansky, C Sechel. Combinatorial
synthesis of small organic molecules. Angew Chem Int Ed Eng 35:2288–2337,
1996.
11. CY Cho, EJ Moran, SR Cherry, JC Stephaus, SPA Fodor, CL Adams,
A Sundaram, JW Jacobs, PG Schultz. An unnatural biopolymer. Science 261:
1303–1305, 1993.
12. KD Janda. Tagged versus untagged libraries: methods for the generation and
screening of combinatorial chemical libraries. Proc Natl Acad Sci USA 91:
10779–10785, 1994.
13. G Lowe. Combinatorial chemistry. Chem Soc Rev p 309, 1995.
14. M Lebl, V Krchnak, NF Sepetov, V Nikolaev, M Stankova, P Kocis, M Patek,
Z Flegelova, R Ferguson, KS Lam. Synthetic combinatorial libraries: a new
tool for drug design. In: M Atassi, E Appella, eds. Methods in Protein Structure
Analysis. New York: Plenum Press, 1995, pp 335–342.
15. T Carell, EA Wintner, A Bashir-hashemi, J Rebek Jr. A novel procedure for
the synthesis of libraries containing small organic molecules. Angew Chem Int
Ed Eng 33:2059–2061, 1994.
16. PHH Hermkens, HCJ Ottenheijm, D Rees. Solid-phase organic reactions: a re-
view of the recent literature. Tetrahedron 52(13):4527–4554, 1996.
17. J Nielson, S Brennar, K Janda. Synthetic methods for the implementation of
encoded combinatorial chemistry. J Am Chem Soc 115:9812–9813, 1993.
18. SH DeWitt, JS Kiely, CJ Stancovic, MC Schroeder, DM Reynolds-Cody, MR
Pavia. Diversomers-an approach to nonpeptide, nonoligomeric chemical diver-
sity. Proc Natl Acad Sci USA 90:6909–6913, 1993.
19. SH DeWitt, MC Schroeder, CJ Stankovic, JE Strode, AW CZarnik. DI-
VERSOMER Technology: solid phase synthesis, automation, and integration
for the generation of chemical diversity. Drug Dev Res 33:116–124, 1994.
20. M Medal, I Svendsen, K Breddam, F-I Auzanneau. Portion-mixing peptide li-
braries of quenched fluorogenic substrates for complete subsite mapping of en-
doprotease specificity. Proc Natl Acad Sci USA 91:3314–3318, 1994.
21. B Deprez, X Willard, L Bourel, H Coste, F Hyafil, A Tarter. Orthogonal combi-
natorial chemical libraries. J Am Chem Soc 117:5405–5406, 1995.
22. JA Loo. Mass Spectrometry in the combinatorial chemistry revolution. Eur
Mass Spectrom. In press.
23. JA Loo. Bioanalytical mass spectromety: many flavors to choose. Bioconj
Chem 6:644–665, 1995.

© 2000 by Marcel Dekker, Inc.

24. DL Flynn, JZ Crich, RV Devraj, SL Hockerman, JJ Parlow, MS South,
S Woodard. Chemical library purification strategies based on principles of com-
plementary molecular reactivity and molecular recognition. J Am Chem Soc
119(21):4874–4881, 1997.
25. ED Lee, W Muck, JD Henion, TR Covey. On-line capillary zone electrophore-
sis–ion spray tandem mass spectrometry for the determination of dynorphins.
J Chromatography 458:313–321, 1988.
26. RD Smith, JH Wahl, DR Goodlett, SA Hofstadler. Capillary electrophoresis/
mass spectrometry. Anal Chem 65:A574, 1993.
27. Y Dunayevskiy, P Vouros, E Wintner, G Schipps, T Carrell, J Rebek Jr. Appli-
cation of capillary electrophoresis–electrospray ionization mass spectrometry
in the determination of molecular diversity. Proc Natl Acad Sci USA 93:6152–
6157, 1996.
28. P Thibault, C Paris, S Pleasance. Analysis of Peptided and proteins by capillary
electrophoresis/mass spectrometry using acidic buffers and coated capillaries.
Rapid Commun Mass Spectrom 5:484–490, 1991.
29. GA Valadovic, NL Kellher, FW McLafferty. Attomole protein characteriza-
tionby capillary electrophoresis–mass spectrometry. Science 273:1199–1201,
1996.
30. Y Dunayevskiy, P Vouros, T Carell, E Wintner, JJ Rebek Jr. Characterization
of the complexity of small molecule libraries by electrospray ionization mass
spectrometry. Anal Chem 67(17):2906–2915, 1995.
31. T Carrell, EA Wintner, AJ Sutherland, J Rebek, Y Dunayevskiy, P Vouros.
New promise in combinatorial chemistry: symthesis, characterization, and
screening of small-molecule libraries in solution. Chem Biol 3:171–183, 1995.
32. SC Pomerantz, JA McCloskey, TM Tarasow, BE Eaton. Deconvolution of com-
binatorial oligonucleotide libraries by electrospray ionization tandem mass
spectrometry. J Am Chem Soc 119(17):3861–3867, 1997.
33. JW Metzger, K-H Wiesmüller, V Gnau, J Grünges, G Jung. Mass specto-
metry of synthetic pepitide libraries. Angew Chem Int Ed Eng 32:894–896,
1993.
34. JW Metzger, K-H Wiesmüller, V Gnau, J Brünjes, G Jung. Ion spray mass
spectrometry and multiple sequence analysis of synthetic peptide libraries.
Angew Chem Int. Ed Eng 32(6):894–896, 1993.
35. JW Metzger, C Kempter, K-H Wiesm̈uller, G Jung. Electrospray mass spec-
trometry and tandem mass spectrometry of synthetic multicomponent peptide
mixtures: determination of composition and purity. Anal Biochem 219:261–
277, 1994.
36. JW Metzger, S Stenanovic, J Brunjes, K-H Wiesmueller, G Jung. Electrospray
mass spectrometry and multiple sequence analysis of synthetic peptide libraries.
Methods 6:425–431, 1994.
37. WL Fitch, A Lu, K Tsutsui, N Shah. Single bead mass spectrometry for charac-
terization of combinatorial libraries. Proceedings of the 44th ASMS Conference
on Mass Spectrometry and Allied Topics, Portlan, OR, 1996, p. 1043.

© 2000 by Marcel Dekker, Inc.

38. F Bitsch, S Herter, K Mueller, F Gombert. Nonoelectrospray ionization mass
spectometry as a tool for mass measuring compounds synthesized on beads.
Proceedings of the 45th ASMS Conference on Mass Spectrometry and Allied
Topics, Palm Springs, CA, 1997, p. 1263.
39. M Wilm, M Mann. Analytical properties of the nanoelectrospray ion source.
Anal Chem 68:1–8, 1996.
40. RG Smith, S Woodard, K Leschinsky. Mixture analysis using LC/APCI/MS.
Proceedings of the 44th ASMS Conference on Mass Spectometry and Allied
Topics, Portland, OR, 1996, p. 1036.
41. RG McKay, BS Larsen, RS Livingston, JG Doughty. Combinatorial li-
brary automation for rapid APCI analysis. Proceedings of the 44th ASMS
Conference on Mass Spectometry and Allied Topics, Portland, OR, 1996,
p. 1037.
42. LYT Li, JN Kyranos. Automated multi-dimensional HPLC/UV/MS forquanti-
tative method development. Proceedings of the 44th ASMS Conference on
Mass Spectrometry and Allied Topics, Portland, OR, 1996, p. 1041.
43. J Whitney, M Hall, R Rourick, K Volk, E Kerns, M Lee. Using scientific visual-
ization software to analyze structural trends in LC/MS profile datasets. Pro-
ceedings of the 44th ASMS Conference on Mass Spectrometry and Allied Top-
ics, Portland, OR, 1996, p. 1046.
44. J Zweigenbaum, J Henion. High-performance capillary electrochromato-
graphy with ion trap MS n detection. Proceedings of the 44th ASMS Confer-
ence on Mass Spectrometry and Allied Topics, Palm Springs, CA, 1997,
p. 1260.
45. KF Blom. Strategies and data precision requirments for the mass spectrometric
determination of structures form combinatorial mixtures. Anal Chem 69(21):
4354–4362, 1997.
46. H Muenster, R Pesch, AK Ziberna. Automated MS/MS for combinatorial
chemistry library analysis with a hybrid tandem mass spectrometer. Proceed-
ings of the 45th ASMS Conference on Mass Spectrometry and Allied Topics,
Palm Springs, CA, 1997, p. 1250.
47. MF Bean, ME Hemling, KK Sonenson, SA Carr. Automated analysis and data
processing for combinatorial chemistry. Proceedings of the 45th ASMS Confer-
ence on Mass Spectrometry and Allied Topics, Palm Springs, CA, 1997,
p. 1254.
48. R Wilgus, M Geysen, D Wagner, F Schoenen, C Wagner, W Bodnar. Auto-
mated mass spectral analysis for combinatorial chemistry: The Capture Pro-
gram. Proceedings of the 45th ASMS Conference on Mass Spectrometry and
Allied Topics, Palm Springs, CA, 1997, p. 1255.
49. DC Tutko, KD Henry, BE Winger. Sequential MS and MS n analysis of combi-
natorial libraries by using automated MALDI-FT/MS. Proceedings of the 45th
ASMS Conference on Mass Spectrometry and Allied Topics, Palm Springs,
CA, 1997, P. 1259.
50. V Nicolaev, O Issakova, S Wade, N Sepetov. Automatic LC/MS Analysis of

© 2000 by Marcel Dekker, Inc.

 combinatorial libraries—a rational approach. Proceedings of the 45th ASMS
Conference on Mass Spectrometry and Allied Topics, Palm Springs, CA, 1997,
p. 1262.
51. J Batt, PA Bott, MA McDowall, SW Preece, JA Rontree. Automated LC-MS
for the characterization of combinatorial libraries. Proceedings of the 44th
ASMS Conference on Mass Spectrometry and Allied Topics, Portland, OR,
1996, p. 1033.
52. D Wagner, W Bodnar, H Geysen. Automated mass spectral data analysis for
combinatorial chemistry. Proceedings of the 44th ASMS Conference on Mass
Spectrometry and Allied Topics, Portland, OR, 1996, p. 1034.
53. ME Hail, BM Warrack, AS Arroyo, GC DiDonato, MS Lee. Automated high
throughput LC/MS for the rapid profiling of pharmaceutical compounds. Pro-
ceedings of the 44th ASMS Conference on Mass Spectrometry and Allied Top-
ics, Portland, OR, 1996, p. 1038.
54. KL Morand, TM Burt, RLM Dobson, LJ Wilson. Development of a mass spec-
trometry laboratory for support of pharmaceutical combinatorial chemistry pro-
grams. Proceedings of the 45th ASMS Conference on Mass Spectrometry and
Allied Topics, Palm Springs, CA, 1997, p. 1261.
55. CW Ross, MB Young, DR Patrick, HG Ramjit. the simultaneous characteriza-
tion of multi-component libraries from single reaction wells by use of FT/ICR/
MS. Proceedings of the 45th ASMS Conference on Mass Spectrometry and
Allied Topics, Palm Springs, CA, 1997, p. 1257.
56. L Taylor, C Garr, M Kobel, L Burton, L Cameron, W McFee. Quality control
of the Optiverse library by mass spectrometry. Proceedings of the 45th ASMS
Conference on Mass Spectrometry and Allied Topics, Palm Springs, CA, 1997,
p. 1248.
57. R Zambias, DA Boulton, PR Griffin. Microchemical structural determination
of a peptoid covalently bound to a polymeric bead by matrix assisted laser
desorption ionization time-of-flight mass spectrometry. Tetrahedron Lett
35(25):4283–4286, 1994.
58. MA Gallop, RW Barrett, WJ Dower, SPA Fodor, EM Gordon. Application
of combinatorial technologies to drug discovery. 1. Background and peptide
combinatorial libraries. J Med Chem 37:1233–1251, 1994.
59. EM Gordon, RW Barrett, WJ Dower, SPA Fodor, MA Gallop. Applications
of combinatorial technologies to drug discovery. 2. Combinatorial organic syn-
thesis, library screening strategies, and future directions. J Med Chem 37:1385–
1401, 1994.
60. M Bodsnszky, A Bodsnszky. Principle of Peptide Synthesis. Heidelberg:
Springer-Verlag, 1984.
61. Á Furka, F Sebestyén, M Asgedom, G Dibó. A general method for the rapid
synthesis of multicomponent peptide mixtures. Int J Peptide Protein Res 37:
487–493, 1991.
62. R Youngquist, GR Fuentes, M Lacey, T Keough. Matrix-assisted laser desorp-
tion ionization for rapid determination of the sequences of biologically active

© 2000 by Marcel Dekker, Inc.

 peptides isolated from support-bound combinatorial peptide libraries. Rapid
Commun Mass Spectrom 8:77–81, 1994.
63. RS Youngquist, GR Fuentes, MP Lacey, T Keough. Generation and screening
of combinatorial peptide libraries designed for rapid sequencing by mass spec-
trometry. J Am Chem Soc 117:3900–3906, 1995.
64. BJ Egner, M Cardno, M Bradley. Linkers for combinatorial chemistry and reac-
tion analysis using solid phase in situ mass spectrometry. J Chem Soc Chem
Commun pp 2163–2164, 1995.
65. BJ Egner, GJ Langley, M Bradley. Solid phase chemistry: direct monitoring
by matrix-assisted laser desorption/ionization time of flight mass spectrometry.
A tool for combinatorial chemistry. J Org Chem 60:2652–2653, 1995.
66. JM Kerr, SC Banville, RN Zuckermann. Encoded combinatorial peptide li-
braries containing non-natural amino acids. J Am Chem Soc 115:2529–2531,
1993.
67. KC Lewis, WL Fitch, D Maclean. Characterization of split/pool combinatorial
libraries. Proceedings of the 45th ASMS Conference on Mass Spectrometry
and Allied Topics, Palm Springs, CA, 1997, p. 1265.
68. A Borchardt, WC Still. Synthetic receptor binding elucidated with an encoded
combinatorial library. J Am Chem Soc 116:373–374, 1994.
69. S Brenner, RA Lerner. Encoded combinatorial chemistry. Proc Natl Acad Sci
USA 89:5381–5383, 1992.
70. HP Nestler, PA Barlett, WC Still. A general method for molecular tagging of
encoded combinatorial chemical libraries. J Org Chem 59:4723–4724, 1994.
71. MC Fitzgerald, K Harris, CG Shevlin, G Siuzdak. Direct characterization of
solid phase resin-bound molecules by mass spectometry. Bioorg Med Chem
lett 6(8):979–982, 1996.
72. BB Brown, DS Wagner, HM Geysen. A single-bead decode strategy using elec-
trospray ionization mass spectrometry and a new photolabile linker: 3-amino-
3-92-nitrophenyl)propionic acid. Mol Diver 1:4–12, 1995.
73. CL Brummel, INW Lee, Y Zhous, SJ Benkovic, N Winograd. A mass specto-
metric solution to the address problem of combinatorial chemistry. Science 264:
399–402, 1994.
74. SJ Benkovic, N Winograd, CL Brummel, INW Lee. Method for identifying
members of combinatorial libraries. Patent Appl PCT Int Appl WO9525737
1995.
75. C Drouot, C Enjalbal, P Fulcrand, J Martinez, J-L Aubagnac, R Combarieu,
Y dePuydt. Step-by-step control by time-of-flight secondary ion mass spectrom-
etry of a peptide synthesis carried out on polymer beads. Rapid Commun Mass
Spectrom 10:1509–1511, 1996.
76. CL Brummel, JC Vickerman, SA Carr, ME Hemling, GD Roberts, W Johnson,
J Weinstock, D Gaitanopoulos, SJ Benkovic, N Winograd. Evaluation of mass
spectrometric methods applicable to the direct analysis of non-peptide bead-
bound combinatorial libraries. Anal Chem 68:237–242, 1996.
77. PA vanVeelan, UR Tjaden, j vanderGreef. Direct molecular weight determina-

© 2000 by Marcel Dekker, Inc.

 tion of resin-bound oligopeptides using 252 Cf plasma-desorption mass spectrom-
etry. Rapid Commun Mass Spectrom 5:565–568, 1991.
78. J Boutin, P Henning, P Lambert, S Berlin, L Petit, J-P Mahieu, B Serkiz, J-P
Volland, JL Fauchere. Combinatorial peptide libraries: robotic synthesis and
analysis by NMR, mass spectometry, tandem mass spectrometry and high-per-
formance capillary electophoresis techniques. Anal Biochem 234:126–141,
1996.
79. RN Zuckermann, JM Kerr, MA Siani, SC Banville, DV Santi. Identification
of highest-affinity ligands by affinity selections from equimolar peptide mix-
tures generated by robotic synthesis. Proc Natl Acad Sci USA 89:4505–4509,
1992.
80. RA Houghten. A general method for the rapid solid phase synthesis of large
numbers of peptides. Proc Natl Acad Sci USA 82:5131–5135, 1985.
81. RA Houghten, ST DeGraw, MK Bray, SR Hoffmann, ND Frizzell. Simultane-
ous multiple peptide synthesis. Biotechniques 4:522–528, 1986.
82. RA Houghten, C Pinilla, SE Blondelle, JR Apple, CT Dooley, JH Cuervo. Gen-
eration and use of synthetic peptide combinatorial libraries for basic research
and drug discovery. Nature 354:84–86, 1991.
83. M Davies, M Bradley. C-Terminally modified peptides and peptide libraries:
another end to peptide synthesis. Angew Chem Int Ed Engl 36(10):1097–1099,
1997.
84. T Solouki, L Pas-a-Toliš, GS Jackson, S Guan, AG Marshall. High resolution
multistage MS, MS 2 and MS 3 matrix-assisted laser desorption/ionization FT-
ICR mass spectra of peptides from a single laser shot. Anal Chem 68(21):3718–
3725, 1996.
85. H Geysen, CD Wagner, WM Bodnar, DJ Markworth, GJ Parke, FJ Schoenen,
DS Wagner, DS Kinder. Isotope or mass encoding of combinatorial libraries.
Chem Biol 3(8):679–688, 1996.
86. Z Ni, D Maclean, CP Holmes, MM Murphy, B Ruhland, JW Jacobs, EM Gor-
don, MA Gallop. A versatile approach to encoding combinatorial organic syn-
thesis using chemically robust secondary amine tags. J Med Chem 39:1601–
1608, 1996.
87. BT Chait, R Wong, RC Beavis, SBH Kent. Protein ladder sequencing. Science
262:89–92, 1993.
88. Y Oda, MR Carrasco, MC Fitzgerald. Screening resin-bound peptides to study
the substrate specificity of a proteolytic enzyme. Proceedings of the 45th ASMS
Conference on Mass Spectrometry and Allied Topics, Palm Springs, CA, 1997,
p. 1251.
89. S Guan, A Marshall, S Scheppele. Resolution and chemical formula identifica-
tion of aromatic hydrocarbons and aromatic compunds containing sulfur, nitro-
gen, or oxygen in petroleum distillates and refinery streams. Anal Chem 68:
46–71, 1996.
90. AS Fang, P Vouros, C Stacey, GH Kruppa, FH Laukien, T Carell, EA Wintner,
J Rebek. Rapid screening of combinatorial libraries using electospray ionization

© 2000 by Marcel Dekker, Inc.

 Fourier transform ion cyclotron resonance mass spectrometry. Proceedings of
the 44th ASMS Conference on Mass Spectrometry and Allied Topics, Portland,
OR, 1996, p. 1047.
91. AS Fang, P Vouros, CC Stacey, GH Kruppa, FH Laukien, EA Wintner, T Car-
ell, J Rebek. Characterization of combinatorial libraries using electrospray ion-
ization Fourier transform ion cyclotron resonance mass spectrometry. Comb
Chem High Throughput Screen 1(1):22–33, 1998.
92. JP Nawrocki, M Wigger, CH Watson, TW Hayes, MW Senko, SA Benner, JR
Eyler. Analysis of combinatorial libraries using electrospray ionization Fourier
transform ion cyclotron resonance mass spectrometry. Rapid Commun Mass
Spectrom 10:1860–1864, 1996.
93. JP Nawrocki, M Wigger, CH Watson, SA Benner, JR Eyler. Characterization
of combinatorial libraries using ESI-FTICR mass spectrometry. Proceedings of
the 45th ASMS Conference on Mass Spectrometry and Allied Topics, Palm
Springs, CA, 1997, p. 1256.
94. BE Winger, JE Campana. Characterization of combinatorial peptide libraries
by electrospray ionization Fourier transform mass spectrometry. Rapid
Commun Mass Spectrom 10:1811–1813, 1996.
95. Q Wu, O Issakova, M Stankova, L Zhao. Structural determination using
multistage accurate mass spectrometry: an application to combinatorial chemis-
try. Proceedings of the 45th ASMS Conference on Mass Spectrometry and Al-
lied Topics, Palm Springs, CA, 1997, p. 1258.
96. P Speir, CW Ross III, H Ramjiit, G Kruppa, F Laukien. The potential of FTMS
in addressing structural diversity of combinatorial libraries. Proceedings of the
44th ASMS Conference on Mass Spectrometry and Allied Topics, Portland,
OR, 1996, p. 1035.
97. C Steinbeck, K Berlin, C Richert. MASP: a program predicting mass spectra
of combinatorial libraries. J Chem Inf Comput Sci 37:449–457, 1997.
98. K Volk, R Rourick, J Whitney, E Kerns, M Lee. Combinatorial degradation
profiling. Proceedings of the 45th ASMS Conference on Mass Spectrometry
and Allied Topics, Palm Springs, CA, 1997, p. 1252.
99. K Volk, R Rourik, J Whitney, T Spears, E Kerns, M Lee. Predictive combinato-
rial stability profiling. Proceedings of the 44th ASMS Conference on Mass
Spectrometry and Allied Topics, Portland, OR, 1996, p. 1040.
100. RA Zubarev, P Håkansson, B Sundqvist. Accuracy requirements for peptide
characterization by monoisotopic molecular mass measurements. Anal Chem
68:4060–4063, 1996.
101. M Mann. Useful tables of possible and probable peptide masses. Proceedings
of the 43th ASMS Conference on Mass Spectrometry and Allied Topics,
Atlanta, GA, 1995, p 639.
102. D Suckau, J Köhl, G Karwath, K Schneider, M Casaretto, D Bitter-Sauermann,
M Przybylski. Molecular epitope identification by limited proteolysis of a im-
mobilized antigen–anitbody complex and mass spectrometric peptide mapping.
Proc Natl Acad Sci USA 87:9848:9852, 1990.

© 2000 by Marcel Dekker, Inc.

103. YV Lubarskaya, YM Duanyevskiy, P Vouros, BL Karger. Microscale epitope
mapping by affinity capillary electophoresis–mass spectrometry. Anal Chem
69:3008–3014, 1997.
104. M Macht, W Fiedler, K Kürzinger, M Przybylski. Mass spectrometric mapping
of protein epitope structures of mycardial infarct markers myoglobin and tropo-
nin T. Biochemistry 35:15633–15639, 1996.
105. M Nedved, S habibi-Goudarzi, B Ganem, J Henion. Characterization of
benzodiaxepine ‘‘combinatorial’’ chemical libraries by on-line immunoaffinity
extraction, coupled column HPLC–ion spray mass sprectrometry–tandem mass
spectrometry. Anal Chem 68:4228–4236, 1996.
106. T Carell, EA Wintner, J Rebek Jr. A solution phase screening procedure for
the isolation of active compounds from a library of molecules. Angew Chem
Int Ed Eng 33:2061–2064, 1994.
107. X Cheng, R Chen, JE Bruce, BL Schwartz, GA Anderson, SC Gale, RD Smith,J
Gao, GB Sigal, M Mammen. Using electrospray ionization to study competitive
binding of inhibitors to carbonic anhydrase. J Am Chem Soc 117:8859–8860, 1995.
108. Y Chu, DP Kirby, BL Karger. Free Solution identification of candidate peptides
from combinatorial libraries by affinity capillary electrophoresis/mass spec-
trometry. J Am Chem Soc 117:5419–5420, 1995.
109. Y-H Chu, LZ Avila, J Gao, GM Whitesides. Affinity capillary electrophoresis.
Acc Chem Res 28:461–468, 1995.
110. Y-H Chu, YM Dunayevskiy, DP Kirby, P Vouros, BL Karger. Affinity capillary
electrophoresis–mass spectometry for screening combinatorial libraries. J Am
Chem Soc 118:7827–7835, 1996.
111. S Kaur, L McGuire, D Tang, G Dollinger, V Huebner. Affinity selection and
mass spectrometry–based strategies to identify lead compounds in combinato-
rial libraries. J Prot Chem 16(5):505–511, 1997.
112. KS Lam, SE Salmon, EM Hersh, VJ Hruby, WM Kazmierski, RJ Knapp. A
new type of synthetic peptide library for the identification of ligand binding
activity. Nature 354:82–84, 1991.
113. JA Loo, DE De-John, RRO Loo, PC Andrews. Application of mass spectrome-
try for characterizing and identifying ligands from combinatorial libraries. Ann
Rep Medic Chem 31:319–325, 1996.
114. DI Papac, J Hoyes, KB Tomer. Direct analysis of affinity-bound analytes by
MALDI/TOF MS. Anal Chem 6:2609–2613, 1994.
115. C Pinilla, JR Appel, P Blanc, RA Houghten. Rapid identification of high affinity
pepide ligands using positional scanning synthetic peptide combinatorial librar-
ies. Biotechniques 13:901–905, 1992.
116. SE Salmon, KS Lam, M Lebl, A Kandola, PS Khattri, S Wade, M Patek,
P Kocis, V Krchnak, D Thorpe. Discovery of biologically acitve peptides in
random libraries. Proc Natl Acad Sci USA 90:11708–11712, 1993.
117. R Wieboldt, J Zweigenbaum, J Henion. Immunoaffinity ultrafiltration with ion
spray HPLC/MS for screening small-molecule libraries. Anal Chem 69:1683–
1691, 1997.

© 2000 by Marcel Dekker, Inc.

118. J Wu, S Takayama, C-H Wong, G Suizdak. Quantitive electrospray mass spec-
trometry for the rapid assay of enzyme inhibitors. Chem Biol 4:653–657, 1997.
119. RN Zuckermann, EJ Martin, DC Spellmeyer, GB Stauber, KR Shoemaker, JM
Kerr, GM Figliozzi, DA Goff, MA Siani, RJ Simon. Discovery of nanomolar
ligands for seven transmembrane G-protein coupled receptors from a diverse
peptoid library. J Med Chem 37:2678–2685, 1994.
120. LC Bock, LC Griffin, JA Latham, EH Vermass, JJ Toole. Selection single
stranded DNA molecules that bind and inhibit human thrombin. Nature 355:
564–566, 1992.
121. AH Brockman, R Orlando. New immobilization chemistry for probe affinity
mass spectrometry. Rapid Commun Mass Spectrom 10:1688–1692, 1996.
122. YM Dunayevskiy, J-J Lai, C Quinn, F Talley, P Vouros. Mass spectrometric
identification of ligands selected from combinatorial libraries using gel filtra-
tion. Rapid Commun Mass Spectrom 11:1178–1184, 1997.
123. J Eichler, RA Houghten. Identification of substrate–analog trypsin inhibitors
through the screening of synthetic peptide combintorial libraries. Biochemistry
32:11035–11041, 1993.
124. NJ Haskins, DJ Hunter, AJ Organ, SS Rahman, C Thom. Combinatorial chem-
istry: direct analysis of bead surface associated materials. Rapid Commun Mass
Spectrom 9:1437–1440, 1995.
125. TW Hutchens, T Yip. New desorption strategies for the mass spectrometric
analysis of macromolecules. Rapid Commun Mass Spectrom 7:576–580, 1993.
126. M Laskowski Jr, I Kato. Protein inhibitors of proteinases. Annu Rev Biochem
49:593–626, 1990.
127. MA Kelly, H Liang, I Sytwu, I Vlattas, NL Lyons, BR Bowen, LP Wennogle.
Characterization of SH2–ligand interactions via library affinity selection with
mass spectrometric detection. Biochemistry 35:11747–11755, 1996.
128. JR Krone, RW Nelson, D Dogruel, P Williams. Interfacing BIA with MALDI
mass spectrometry. Proceedings of the 44th ASMS Conference on Mass Spec-
trometry and Allied Topics, Portland, OR, 1996, p. 753.
129. RW Nelson, JR Krone, AL Bieber, P Williams. Mass spectrometric immunoas-
say. Anal Chem 67:1153, 1996.
130. GD Dollinger, VD Huebner. Affinity selection of ligands by mass spectrometry.
Patent Appl PCT Int Appl WO 9622530 1996.
131. NJ Hales, E Clayton. Affinity chromatographic method for screening combina-
torial libraries. Patent Appl PCT Int App. WO9632642 1995.
132. DB Kassel, MD Green, RS Wehbie, R Swanstrom, J Berman. HIV-1 Protease
Specificity Derived from a Complex Mixture of Synthetic Substrates. Anal Bio-
chem 228:259–266, 1995.
133. T Kupke, C Kempter, G Jung, F Gotz. Oxidative decarboxylation of peptides
catalyzed by flavoprotein Epi D. J Biol Chem 270:11282–11289, 1995.
134. JH Till, RS Annan, SA Carr, WT Miller. Use of synthetic peptide libraries and
phosphopeptide-selective mass spectrometry to probe protein kinase substrate
specificity. J Biol Chem 269:7423–7428, 1994.

© 2000 by Marcel Dekker, Inc.

135. BA Bunin, JA Ellman. A general and expedient method for the solid phase
synthesis of 1,4-benzodiazepine derivatives. J Am Chem Soc 114:10997–
10998, 1992.
136. RB VanBreemen, CR Haung, D Nikolic, CP Woodbury, YZ Zhao, DL Venton.
Pulsed ultrafiltration mass spectrometry: a new method for screening combina-
torial libraries. Anal Chem 69:2159–2164, 1997.
137. Y-H Chu, GM Whitesides. Affinity capillary electrophoresis can simulta-
neously measure binding constants of multiple peptides to vancomycin. J Org
Chem 57:3524–3525, 1992.
138. Y-H Chu, LZ Avila, HA Biebuyck, GM Whitesides. Use of affinity capillary
electrophoresis to meausure binding constants of ligands to proteins. J Med
Chem 35:2915–2917, 1992.
139. FA Gomez, LZ Avila, Y-H Chu, GM Whitesides. Determination of binding
constants of ligands to proteins by affinity capillary electrophoresis: compensa-
tion for electroosmotic flow. Anal Chem 66:1785–1791, 1994.
140. Y-H Che, WJ Lees, A Stassinopoulos, CT Walsh. Using affinity capillary elec-
trophoresis to determine binding stochiometreis of protein–ligand interactions.
Biochemistry 33:19616–19621, 1994.
141. C Bain, J Evall, G Whitesides. Formation of monolayers by the coadsorption
of thiols on gold: variation in the head group, and solvent. J Am Chem Soc
111:7155–7164, 1989.
142. JE Bruce, GA Anderson, R Chen, X Chen, DC Gale, SA Hofstadler, BL
Schwartz, RD Smith. Bio-affinity characterization mass spectrometry. Rapid
Commun Mass Spectrom 9:64, 1995.
143. JE Bruce, L Pasa-Tolic, P Lei, RD Smith, SA Carr, D Dunnington, W Prichett,
D Yamashita, J Yen, E Applebaum. Bioaffinity characterization mass spectrom-
etry (BACMS) of the human scr SH2 domain protein. Proceedings of the 44th
ASMS Conference on Mass Spectrometry and Allied Topics, Portland, OR,
1996, p. 1403.
144. B Ganem, JK Henion. Detecting noncovalent complexes of biological macro-
molucules: new applicants of ion-spray mass spectrometry. Chemtracts: Org
Chem 6:1–22, 1993.
145. RD Smith, KJ Light-Wahl. The observation of non-covalent interactions in so-
lution by electrospray ionization mass spectrometry: promise pitfalls and prog-
nosis. Biol Mass Spectrom 22:493–501, 1993.
146. J Gao, X Cheng, R Chen, GB Sigal, JE Bruce, BL Schwartz, SA Hofstadler,
GA Anderson, RD Smith, GM Whitesides. Screening derivatized peptide librar-
ies for tight binding inhibitors to carbonic anhydrase II by electrospray ioniza-
tion–mass spectrometry. J Med Chem 39:1949–1955, 1996.
147. D Lafitte, V Benezech, J Bompart, F Laurent, P Bonnet, JP Chapat, G Grassy,
B Calas. Characterization of low affinity complexes between calmodulin and
pyrazine derivatives by electrospray ionization mass spectrometry. J Mass
Spectrom 32:87–93, 1997.
148. FS Pullen, GL Perkins, KL Burton, RS Ware, MS Teague, JP Kiplinger. Putting

© 2000 by Marcel Dekker, Inc.

 mass spectrometry in the hands of the end user. J Am Soc Mass Spectrom 6:
394–399, 1995.
149. R Spreen, et al. Open access MS: a walk-up MS service. Anal Chem News
Features pp 414A–419A, 1996.
150. LCE Taylor, et al. Open access atmospheric pressure chemical ionization mass
spectrometry for routine sample analysis. J Am Soc Mass Spectrom 6:387–
393, 1995.
151. RM Holt, MJ Newman, FS Pullen, DS Richards, AG Swanson. High-perfor-
mance liquid chromatography/NMR spectroscopy/mass spectrometry: further
advances in hyphenated technology. J Mass spectrom 32:64–70, 1997.
152. J Bergman, K Halm, K Adkison, J Shaffer. Simultaneous Pharmacokinetic
screening of a mixture of compounds in the dog using API LC/MS/MS analysis
for increased throughput. J Med Chem 40:827–829, 1997.
153. TV Olah, DA McLoughlin, JD Gilbert. The simultaneous determination of mix-
tures of drug candidates by LC APCI MS as in in vivo drug screening procedure.
Rapid Commun Mass Spectrom 11:17–23, 1997.
154. DA McLoughlin, TV Olah, JD Gilbert. A direct technique for the simultaneous
determination of 10 drug candidates in plasma by liquid chromatography atmo-
spheric pressure chemical ionization mass spectrometry interfaced to a Prospekt
solid-phase extraction system. Pharm Biomed Anal 15(12):1893–1901, 1997.
155. E Taylor, et al. Intestinal absorption screening of mixtures from combinatorial
libraries in the Caco-2 model. Pharm Res 14(5): 572–577, 1997.
156. RA Rourick, JL Whitney, SW Fink, EH Kerns, MS Lee. Applications of auto-
mation to combinatorial degradation profiling. Proceedings of the 45th ASMS
Conference on Mass Spectrometry and Allied Topics, Palm Springs, CA, 1997,
p. 1264.
157. WM Bodnar, DS Wagner, M Geysen. Pharmacokinetic characterization of com-
binatorial libraries. Proceedings of the 44th ASMS Conference on Mass Spec-
trometry and Allied Topics, Portland, OR, 1996, p. 1044.

© 2000 by Marcel Dekker, Inc.

3
Infrared and Raman Spectroscopy

Hans-Ulrich Gremlich
Novartis Pharma AG
Basel, Switzerland

I. INTRODUCTION

Since 1905, when William W. Coblentz obtained the first infrared spectrum
(1), vibrational spectroscopy has become an important analytical tool in re-
search and in technical fields. In the late 1960s, infrared spectrometry was
generally believed to be an instrumental technique of declining popularity that
was gradually being superseded by nuclear magnetic resonance (NMR) and
mass spectrometry (MS) for structural determinations and by gas and liquid
chromatography for quantitative analysis.
However, the appearance of the first research grade Fourier transform
infrared (FTIR) spectrometers in the early 1970s initiated a renaissance of
infrared spectrometry. After analytical instruments (in the late 1970s) and rou-
tine instruments (in the mid-1980s) became available, dedicated instruments
have now become available at reasonable prices. With its fundamental multi-
plex or Fellgett’s advantage and throughput or Jacquinot’s advantage (2),
FTIR offers a versatile approach to measurement problems that is often supe-
rior to other techniques. Furthermore, FTIR is capable of extracting from sam-
ples information that is difficult to obtain or even inaccessible for NMR and
MS.
Raman and IR spectra give images of molecular vibrations that comple-
ment each other, i.e., the combined evaluation of both spectra yields more
information about molecular structure than their separate evaluation. Raman
spectroscopy (2) offers a high degree of versatility, and sampling is easy be-

© 2000 by Marcel Dekker, Inc.

cause no special preparation is required. As it is a scattering technique, sam-
ples are simply placed in the laser beam and the backscattered radiation is
analyzed. While in mid-IR spectroscopy Fourier transform instruments are
used almost exclusively, both conventional dispersive and Fourier transform
techniques have their applications in Raman spectroscopy (2).
By the early 1990s, combinatorial chemistry had revived solid phase
organic synthesis. For this technology, IR spectroscopy has always been an
important analytical tool. It allows spectral analysis directly on the solid phase,
thus obviating the tedious cleavage of compounds from the solid support. In
contrast, thin-layer chromatography (TLC), high-performance liquid chroma-
tography (HPLC), gas chromatography (GC), and MS techniques are clearly
not feasible unless the compound of interest is first cleaved from the polymeric
support. Cleaving the compound from the resin for analysis renders these
methods inherently destructive, to say nothing of the uncertainty involved in
the cleavage chemistry.
As described in the following sections, IR and Raman spectroscopy,
using modern Fourier transform techniques such as IR microspectroscopy,
offer excellent analytical tools for the burgeoning field of combinatorial chem-
istry.

II. INFRARED TRANSMISSION SPECTROSCOPY

Transmission spectroscopy (2) is the simplest sampling technique in IR spec-

troscopy and is used for routine spectral measurements on diverse samples.
Resin samples such as polystyrene or TentaGel (3) beads are usually prepared
as a potassium bromide disc (pellet). A small amount, usually 1–3 mg, of
finely ground solid sample is mixed with approximately 400 mg powdered
potassium bromide and then pressed in an evacuated die under high pressure.
The resulting discs are transparent and yield good spectra.
With diagnostic absorption bands in starting materials or products, IR
spectroscopy in general is efficient in monitoring each reaction step directly
on the solid phase, i.e., without cleaving the substance from the solid support.
The direct functional group analysis provided by IR spectroscopy permits
qualitative analysis of each reaction step throughout the synthesis. An example
is shown in Fig. 1, the success of the addition of adipinic acid monoamide
to a polystyrene resin is immediately seen by overlaying the corresponding
normalized spectra.
Furthermore, IR spectra from KBr pellets of polymer samples can also
be used to study reaction mechanisms in combinatorial synthesis. For instance,

© 2000 by Marcel Dekker, Inc.

Figure 1 KBr transmission spectrum of polystyrene resin (lower curve) and of poly-
styrene resin loaded with adipinic acid monoamide (upper curve). The carbonyl bands
of the acid (at 1720 cm⫺1) and of the amide (at 1650 cm⫺1) are clearly seen in the
overlay plot. The spectra were obtained, in the 4000–400 cm⫺1 range with 32 scans
at 2 cm⫺1 resolution, by means of a Bruker IFS 66 spectrophotometer.

the mechanism of the solid phase synthesis of 1,4-dihydropyridines was eluci-

dated by observing the IR bands of the ester carbonyls (4).

III. INFRARED INTERNAL REFLECTION SPECTROSCOPY

Internal reflection spectroscopy (2), also known as attenuated total reflectance

(ATR), is a versatile, nondestructive technique for obtaining the IR spectrum
of the surface of a material or the spectrum of materials either too thick or
too strongly absorbing to be analyzed by standard transmission spectroscopy.
The technique goes back to Newton who, in studies of the total reflection light
at the interface between two media of different refractive indices, discovered
that an evanescent wave in the less dense medium extends beyond the re-
flecting interface. Infrared spectra can conveniently be obtained by measuring
the interaction of the evanescent wave with the external less dense medium.

© 2000 by Marcel Dekker, Inc.

Figure 2 Internal reflection spectra of Mega Crowns with linker, Fmoc-protected
(upper curve), and Fmoc-deprotected (lower curve). The bands due to the Fmoc pro-
tecting group are marked ‘‘F.’’ The spectra were obtained, in the 4000–650 cm⫺1 range
with 32 scans at 4 cm⫺1 resolution, by means of a Bruker IFS 28 spectrophotometer
equipped with a SplitPea accessory.

The sample is placed in contact with the internal reflection element, the light
is totally reflected, and the sample interacts with the evanescent wave resulting
in the absorption of radiation by the sample. The internal reflection element
is made from a material with a high refractive index, e.g., zinc selenide (ZnSe)
or silicon (Si).
As to combinatorial chemistry, internal reflection spectroscopy is espe-
cially suited for the analysis of the surface of solid polymer substrates, known
as ‘‘pins’’ or ‘‘crowns’’ (5). With FTIR accessories like the SplitPea (6–8)
or the Golden Gate (9), which combine internal reflection and microspectros-
copy, direct observation of the chemical synthesis on pins is possible (10).
This is shown in Fig. 2 where the spectra of a Mega Crown with linker, Fmoc1-
protected (upper curve), and of the same crown after the removal of the pro-

1
9-Fluorenylmethoxycarbonyl.

© 2000 by Marcel Dekker, Inc.

tecting group (lower curve) are overlaid. In the lower curve it is clearly seen
that the bands due to the Fmoc protecting group, marked ‘‘F,’’ have com-
pletely disappeared.
This spectroscopic method is very fast because no further sample prepa-
ration is required, and it is absolutely nondestructive. Thus each step of the
synthesis can be spectroscopically checked with the use of the same pin.
Furthermore, the internal reflection spectra of these reactions are much
less distorted by the absorption bands of the solid support than are the trans-
mission spectra recorded from KBr pellets made of polystyrene or TentaGel
beads. In Fig. 3 the spectra of polystyrene resin (upper curve) and of a polysty-
rene Mega Crown (lower curve), both loaded with adipinic acid monoamide,
are shown. The disturbing polystyrene bands are marked ‘‘P.’’
With all of these advantages, micro internal reflectance spectroscopy
with the SplitPea accessory is an excellent tool for checking successive steps
of a combinatorial synthesis on pins. By overlaying the corresponding spectra,
it is also extremely helpful for the elaboration of a synthesis when reaction

Figure 3 KBr transmission spectrum of polystyrene resin (upper curve) and internal
reflection spectrum of a polystyrene Mega Crown (lower curve), both loaded with
adipinic acid monamide. The polystyrene bands are marked ‘‘P.’’ The spectra were
obtained, in the 4000–650 cm⫺1 range with 32 scans at 4 cm⫺1 resolution, by means
of a Bruker IFS 28 spectrophotometer equipped with a SplitPea accessory.

© 2000 by Marcel Dekker, Inc.

conditions are to be optimized regarding reagents, reaction temperature, or
reaction time. Furthermore, if biological testing of compounds is to be directly
done with the pins, i.e., without cleaving the substance from the support, it
is very important to have an analytical method like internal reflectance spec-
troscopy that is directly applicable to the pins.
Since it can be applied to both pins and beads, micro internal reflectance
spectroscopy with accessories like the SplitPea or Golden Gate provides a fast
and reliable analytical tool at a reasonable price for the rapidly growing field
of combinatorial chemistry.

IV. INFRARED MICROSPECTROSCOPY

By coupling an optical microscope to an FTIR spectrometer, infrared micro-

spectroscopy (2) provides unsurpassed sensitivity, permitting analysis of sam-
ples in the picogram range. Infrared microscopes use reflecting optics such as
Cassegrain objectives instead of IR lenses, which are poor in performance.
As the visible optical train is collinear with the infrared light path, it is possible
to position the sample and to isolate and aperture the area for analysis visually.
Thus IR microscopy combines in an ideal way the capabilities to visually
inspect a sample, e.g., using a video camera (11), and at the same time to
acquire information about the sample at the molecular level. Being a nonde-
structive method, IR microspectroscopy requires virtually no sample prepara-
tion.
In the combinatorial chemistry laboratory, IR microspectroscopy lends
itself to monitoring reaction products and reaction kinetics directly on a single
resin bead (12) without having to cleave the molecule from the bead.
The IR microspectra of single beads can be obtained either in transmis-
sion mode (12,13) or with use of an ATR objective (11,14). The ATR objective
utilizes a small ATR crystal that is placed on the sample with a contact surface
of less than 150 µm in diameter. With a penetration depth of just a few mi-
crometers useful spectra can be obtained from the surface of the resin.
Micro-IR spectra were used for real-time monitoring of solid phase reac-
tions (12,15,16) and, utilizing deuterium isotope containing protecting groups,
for quantitative infrared analysis of solid phase, resin-bound chemical reac-
tions (13).

V. PHOTOACOUSTIC SPECTROSCOPY

Complementing other established techniques such as ATR, photoacoustic

spectroscopy (PAS) (2) is a quick, nondestructive method for analyzing vari-

© 2000 by Marcel Dekker, Inc.

ous materials in the gas, liquid, or solid state with no sample preparation. For
photoacoustic measurements, the sample is sealed in a small-volume cell with
windows for optical transmission. The cell contains a nonabsorbing gas such
as helium. The sample is illuminated with modulated radiation from the inter-
ferometer of a FTIR spectrometer. At wavelengths where the sample absorbs
a fraction of the incident radiation, a modulated temperature fluctuation at the
same frequency as that of the incident radiation is generated in the sample.
This modulation is not necessarily with the same phase as the incident radia-
tion. The surrounding inert gas produces periodic pressure waves in the sealed
cell. These fluctuations in pressure can be detected by a microphone because
the modulation frequency of the incident beam usually lies in the acoustic
range, e.g., between 100 and 1600 Hz. FTIR spectrometers are ideal for photo-
acoustic measurements in the mid-IR, where mirror velocities on the order of
0.05–0.2 cm s⫺1 provide modulation frequencies in the acoustic range (2).
Due to the fact that thermal diffusion length (the distance a thermal wave
travels in the sample before its magnitude is reduced by 1/e, where e is the
base for natural logarithms) and hence intensity of the photoacoustic signal are
increased by lowering the modulation frequency, photoacoustic spectroscopy
lends itself to depth profiling (2). This is especially so when step-scan interfer-
ometers, accommodating any desired modulation frequency, constant over the
entire wavelength range, are used, e.g., in the depth profiling of beads.
Photoacoustic spectra of resin samples have also been reported (16,17).
The photoacoustic spectrum of TentaGel S beads coupled via a trityl linker
with Fmoc-protected tryptophan is shown in Fig. 4. Compared with the photo-
acoustic spectrum of native aminoethyl TentaGel S beads, the appearance of
diverse carbonyl bands is clearly seen: the ester band of the linker at 1750
cm⫺1, the carbamate band of Fmoc protecting group at 1723 cm⫺1, and the
amide I band at 1660 cm⫺1.
Because no sample preparation is required, photoacoustic spectroscopy
can be used to examine a sequence of reactions on the same resin sample
without product loss (18). Concerning sensitive spectroscopic analysis of solid
phase organic chemistry, photoacoustic spectroscopy offers a convenient, non-
destructive alternative method to other IR techniques.

VI. LIQUID CHROMATOGRAPHY–INFRARED SPECTROSCOPY

As in other organic synthesis laboratories, high-performance liquid chroma-

tography (HPLC) is also intensively used for separations in combinatorial
chemistry laboratories. Using standard detectors such as ultraviolet (UV),

© 2000 by Marcel Dekker, Inc.

Figure 4 Photoacoustic spectrum of native aminoethyl TentaGel S beads (lower
curve) and of TentaGel S beads coupled via a trityl linker with Fmoc-protected trypto-
phan (upper curve). The carbonyl bands of the ester (at 1750 cm⫺1), of the carbamate
(at 1723 cm⫺1), and of the amide (at 1660 cm⫺1) are clearly seen in the overlay plot.
The spectra were obtained with 50 scans at 8 cm⫺1 resolution by means of a Bruker
IFS 55 spectrophotometer equipped with a MTEC photoacoustic detector.

chromatography is useful in providing quantitative data, when the identities

of the mixture components are known. Often the provisional identity of a
chromatographically eluting compound can be established through the combi-
nation of its retention time and the corresponding UV spectrum.
Due to the unique fingerprinting capability of IR spectroscopy, the com-
bination of chromatographic separation with spectroscopy significantly im-
proves the analysis of complex mixtures. The crucial point in LC-IR methods,
however, is the interferences caused by the liquid phase. This is overcome by
the revolutionary LC-Transform system (19), which utilizes a high-perfor-
mance ultrasonic nebulizer or a pneumatic linear capillary nozzle to remove
the mobile phase from samples as they elute from the chromatograph. The
capillary is surrounded by hot, flowing sheath gas that provides sufficient ther-
mal energy to evaporate the mobile phase and to contain the nebulized spray
in a tight, focused cone as the spray emerges from the nozzle. The sample is

© 2000 by Marcel Dekker, Inc.

deposited on a germanium sample collection disk as spots with diameters be-
tween 0.5 and 1 mm. Good-quality IR spectra can be obtained from these
spots, which contain microgram to submicrogram mg amounts, by the use of
a specially designed optics module, which basically is a beam condenser.
HPLC-FTIR has been successfully applied to the rather difficult identi-
fication of anabolic steroids (20); difficult, because steroids encompass a large
number of natural and synthetic compounds with minor variations in molecu-
lar configuration. Despite the structural similarities of fluoxymesterone, testos-
terone, methyl testosterone, and epitestosterone, the distinct spectral signatures
of these four compounds, which are all 17-hydroxysteriods with identical
backbone structure, were recognizable.
This methodology easily can be extended to the analysis of combinato-
rial chemistry samples. In combinatorial chemistry, the combination of chro-
matography and MS has proved a powerful method for substance identification
based primarily on molecular weight. Analogously, HPLC-FTIR provides
powerful identification possibilities, but based on molecular structure. As iden-
tifying a substance by MS requires knowledge about its possible molecular
structure, the two techniques form a highly complementary system of identifi-
cation.
If—as it is usually the case in combinatorial chemistry—analysts have
an idea about what an eluted compound is, MS can confirm the component’s
identity precisely. On the other hand, if confirmation is not possible for some
reason, IR spectroscopy generally will provide more valuable insight. A high
degree of confidence can be placed on substance identification that simulta-
neously matches IR spectral characteristics and chromatography elution time.
This is especially so when molecules that have various conformational isomers
are to be distinguished.
Since HPLC-FTIR is a high-throughput automatable hyphenated tech-
nique, it will be a powerful analytical tool in the burgeoning field of combina-
torial chemistry.

VII. RAMAN SPECTROSCOPY

Complementary to infrared, Raman spectroscopy (2) provides unique informa-

tion about molecular structure. Whereas polar groups and antisymmetrical vi-
brations of molecular fragments are better detected by IR spectroscopy, Raman
spectroscopy is more suitable for the identification of unpolar groups and sym-
metrical vibrations of molecular fragments. However, dispersive Raman spec-
troscopy did not find general acceptance in the analytical laboratory during

© 2000 by Marcel Dekker, Inc.

the last 10–20 years because Raman spectra excited with visible lasers are
often masked by fluorescence of the sample or of impurities.
With the advent of Fourier transform–Raman (FT-Raman) spectros-
copy, this situation has changed drastically. FT-Raman systems virtually elim-
inate fluorescence by using near-IR laser sources with 1064 or 785 nm excita-
tion (21). Although near-IR lasers produce much weaker Raman signals than
visible lasers, FT-Raman technology provides the signal-to-noise advantage
necessary to overcome this low signal level, with the consequence that most
laboratory samples give useful FT-Raman spectra.
A general advantage of Raman spectroscopy is the extended spectral
range, and the Stokes shift (2) in FT-Raman spectra is usually recorded from
3500 to 50 cm⫺1. Like IR internal reflection spectroscopy (see Sec. III) and
IR microspectroscopy (see Sec. IV), Raman spectroscopy is a nondestructive
technique requiring little or no sample preparation.
For combinatorial chemistry applications, high-quality FT-Raman spec-
tra can be obtained directly from resin beads, i.e., no cleavage of the molecules
from the polymeric support is necessary. This is shown in Fig. 5, where the
spectra of TentaGel S beads coupled via a trityl linker with Fmoc-protected
tryptophan and the native aminoethyl TentaGel S beads are overlaid. As ex-
pected, significant differences in the spectra occur in the spectral region be-
tween 1620 and 1500 cm⫺1 where aromatic rings show pronounced Raman
activity.
With all of these properties, FT-Raman spectroscopy—similar to usual
vibrational spectroscopy—may be a valuable supplementary technique to IR
spectroscopy in the field of combinatorial chemistry.

VIII. CONCLUSIONS

Infrared and Raman spectroscopy allow direct spectral analysis of the solid
phase, thus avoiding the tedious cleavage of compounds from the solid sup-
port. With diagnostic bands in starting materials or products, IR and Raman
spectroscopy in general are efficient in monitoring each reaction step directly
on the solid phase.
While IR transmission spectroscopy is a general analytical method for
resin samples, internal reflection spectroscopy is especially suited for solid
polymer substrates known as ‘‘pins’’ or ‘‘crowns.’’ Single-bead analysis is
best done by IR microspectroscopy, whereas photoacoustic spectroscopy
allows totally nondestructive analysis of resin samples.

© 2000 by Marcel Dekker, Inc.

Figure 5 FT-Raman spectrum of native aminoethyl TentaGel S beads (lower curve)
and of TentaGel S beads coupled via a trityl linker with Fmoc-protected tryptophan
(upper curve). The spectra were obtained with 300 scans at 4 cm⫺1 resolution by means
of a Bruker IFS 55/S spectrophotometer equipped with a Bruker FRA 106 FT-Raman
accessory. For excitation, a Nd:YAG laser working at 1064 nm was used with a power
of 390 mW.

Providing identification based on molecular structure, HPLC-FTIR is

therefore complementary to LC-MS.
Additionally, Raman spectroscopy as a complement to IR spectroscopy
can be applied to resin samples and—using a Raman microscope—to single
beads.

REFERENCES

1. WW Coblentz. Investigations of infrared spectra. Washington: Carnegie Institu-

tion, 1905. Republished Norwalk: The Coblentz Society, 1962.
2. HU Gremlich. Infrared and Raman spectroscopy. In: Ullmann’s Encyclopedia
of Industrial Chemistry. Weinheim: VCH, 1994, B5, pp. 429–469.

© 2000 by Marcel Dekker, Inc.

 3. Available from RAPP Polymere GmbH, Ernst-Simon-Strasse 9, D-72072
Tuebingen, Germany.
4. MF Gordeev, DV Patel, EM Gordon. Approaches to combinatorial synthesis of
heterocycles: a solid-phase synthesis of 1,4-dihydropyridines. J Org Chem 61:
924–928, 1996.
5. NJ Maeji, AM Bray, RM Valerio, W Wang. Larger scale multipin peptide synthe-
sis. Peptide Res 8:33–38, 1995.
6. Available from Harrick Scientific Corporation, 88 Broadway, Ossining, NY
10562, U.S.A.
7. NJ Harrick, M Milosevic. US Patent 5,210,418, 1993; M Milosevic, NJ Harrick,
CR Wisch. US Patent 5,308,983, 1994.
8. NJ Harrick, M Milosevic, SL Berets. Advances in optical spectroscopy: the ultra-
small sample analyzer. Appl Spectrosc 45:944–948, 1991.
9. Available from Graseby Specac Inc., 301 Commerce Drive, Fairfield, CT 06432,
U.S.A.
10. HU Gremlich, SL Berets. Use of FT-IR internal reflection spectroscopy in combi-
natorial chemistry. Appl Spectrosc 50:532–536, 1996.
11. Bruker Report 143:14–17, 1996.
12. B Yan, G Kumaravel, H Anjaria, A Wu, RC Petter, CF Jewell Jr, JR Wareing.
Infrared spectrum of a single resin bead for real-time monitoring of solid-phase
reactions. J Org Chem 60:5736–5738, 1995.
13. K Russell, DC Cole, FM McLaren, DE Pivonka. Analytical techniques for com-
binatorial chemistry: quantitative infrared spectroscopic measurements of deute-
rium-labeled protecting groups. J Am Chem Soc 118:7941–7945, 1996.
14. PA Martoglio, DW Schiering, MJ Smith, DT Smith. Nicolet Application Note
9693, 1996.
15. B Yan, JB Fell, G Kumaravel. Progression of organic reactions on resin supports
monitored by single bead FTIR microspectroscopy. J Org Chem 61:7467–7472,
1996.
16. B Yan, H Gstach. An indazole synthesis on solid support monitored by single
bead FTIR microspectroscopy. Tetrahedron Lett 37:8325–8328, 1996.
17. W Rapp, J Metzger. Synthesis and identification of peptide and nonpeptide librar-
ies by hyphenated techniques. Proceedings of the 24th European Peptide Sympo-
sium, Edinburgh, pp 743–744, 1996.
18. F Gosselin, M DiRenzo, TH Ellis, WD Lubell. Photoacoustic FTIR spectroscopy,
a nondestructive method for sensitive analysis of solid-phase organic chemistry.
J Org Chem 61:7980–7981, 1996.
19. Available from Lab Connections Inc., 201 Forest Street, Marlborough, MA
01752, U.S.A.
20. JL Dwyer, AE Chapman, X Liu. Analysis of steroids by combined chromatogra-
phy-infrared spectroscopy. LC-GC 13:240–250, 1995.
21. C Lehner, J Sawatzki, NT Kwai. FT-Raman spectroscopy in the industrial envi-
ronment. Proceedings of the 15th International Conference on Raman Spectros-
copy, Pittsburgh, 1996, pp. 1094–1095.

© 2000 by Marcel Dekker, Inc.

4
NMR Methods

Michael J. Shapiro
Novartis
Summit, New Jersey

I. DIRECT ANALYSIS ON SOLID PHASE

The realm of combinatorial chemistry is covered by a wide umbrella of tech-

niques that include both solution phase and solid phase synthesis. The revival
of solid phase synthesis has led to great interest in the development of analyti-
cal methods to monitor the reaction directly on the polymer support during
the course of combinatorial syntheses. These techniques have a significant
advantage compared to cleave-and-analyze characterization, particularly for
optimizing reaction conditions.
Nuclear magnetic resonance (NMR) spectroscopy has been shown to be
useful in this regard. The two procedures available to obtain NMR data on
resin are gel phase NMR and a variant of this methodology using magic angle
spinning (MAS) (1). While Fourier transform infrared (FTIR) (2,3) and mass
spectrometry (MS) (4–6) are faster and more sensitive techniques than NMR,
only NMR can give information concerning the detailed structure of a
molecule.

II. GEL PHASE NMR

Gel phase 13 C NMR is an established technique that was first performed in

1971 (7). Gel phase NMR involves taking an NMR spectrum using a standard

© 2000 by Marcel Dekker, Inc.

high resolution NMR spectrometer. It is nondestructive and the sample can
be readily recovered. Spectra are generally obtained in a standard 5-mm NMR
tube by swelling a resin with an appropriate solvent. Due to signal broadening,
gel phase NMR is generally limited to heteronuclei such as 13 C, 19 F and 31 P
where there is a great deal of chemical shift dispersion.
One of the first applications of gel phase 13 C NMR showed that the build
up of a peptide chain and deprotection of Boc groups could be readily followed
(8). An example of the gel phase 13 C NMR of a peptide is shown in Fig. 1.
Similarly 13 C gel phase NMR was used to evaluate and develop new side
chain deprotection cocktails (9). Liebfritz and Geralt, describing the use of
polyethylene and polystyrene supports, showed the general application of 13 C
gel phase NMR methodology in the solid phase synthesis of peptides (10,11).
Application of gel phase NMR to organic solid phase synthesis was first
reported by Manatt in 1980 (12). Jones and Leznoff extended this work to
show that 13 C NMR data could be obtained from polystyrene-supported phero-
mones and a wide variety of organic substrates (13). The application of 13 C
gel phase NMR to the study of functional group interconversion of polymer-
bound steroids has also been reported (14). High-resolution 1 H gel phase NMR
has been reported for an octapeptide (15). The NMR data were obtained by
use of a deconvolution method to enhance the spectral resolution.
13
C gel phase NMR is generally impractical for the monitoring of reac-
tions in real time. The typical 13 C gel phase NMR experiment takes thousands
of transients (several hours) to get sufficient signal-to-noise. This is especially
true for obtaining resonances in the carbonyl region of the spectrum. In order
to obviate the use of gel phase 13 C NMR as a more practical method, a strategy
exploiting the use of 13 C labeling to monitor the reaction and to further enhance
the experiment sensitivity was developed (16,17). Optimizing the location of
a sample by using an insert in the NMR tube, a sample could be prepared
with only 20 mg of resin. 13 C gel phase NMR data could be obtained in this
manner in only 15–30 min.
19
F makes an excellent probe to monitor solid phase reactions; most solid
supports do not contain fluorine, 19 F NMR is almost as sensitive as proton
NMR, the range of 19 F chemical shifts is large, and the resonance can be
monitored even when the fluorine atom is quite remote from the site of reac-
tion. The buildup of the peptide chain could be monitored by incorporating
19
F into a peptide-protecting group (18).
In the case of the SNAr nucleophilic displacement reaction shown in
Fig. 2, 19 F was found to be an useful probe nucleus (19). Since fluorine is the
leaving group, the disappearance of the 19 F signal offers a window through
which to observe reaction progress. Shown in Fig. 2 is a typical spectrum

© 2000 by Marcel Dekker, Inc.

 Figure 1 Typical peptide 13 C Gel phase NMR spectrum in DMSO-d 6. Product shown is obtained after deprotec-
tion reaction by BF 3 .

© 2000 by Marcel Dekker, Inc.

Figure 2 19 F gel phase NMR spectrum for macrocyclization SNAr reaction for an
oxytocin analog on Wang resin. Bottom spectrum is starting material and top spectrum
is after a reaction time of 1 h. The broad 19 F signal arises from the probe components
and can be used as an internal standard.

© 2000 by Marcel Dekker, Inc.

obtained during the course of the reaction. The broad peak in the spectrum
arises from fluorine-containing components of the Bruker quadranuclear probe
and was conveniently used as an external standard. Using TentaGel resin
linked compounds, 19 F gel phase NMR data could be obtained in less than 10
min for the reactions shown in Fig. 3 (20). Since a wide variety of fluorine
containing compounds are readily available, 19 F gel phase NMR is expected
to increase in utility.

Figure 3 375 MHz 19 F gel phase NMR spectrum for resin linked butyl amide. Con-
version to product is monitored (a) on TentaGel resin and (b) in solution.

© 2000 by Marcel Dekker, Inc.

Figure 4 Horner-Wadsworth-Emmons reaction that was monitored by 31 P gel phase
NMR.

As with 19 F NMR gel phase, 31 P NMR is useful because the resins cur-
rently used do not interfere with the signals produced. Spectra can be obtained
in just 10 min with 50 mg of resin. The first use of 31 P gel phase data was
described by Geralt et al. in conjunction with oligonucleotide synthesis (21).
An especially attractive application of 31 P gel phase NMR was applied to the
Horner-Wadsworth-Emmons olefination reaction depicted in Fig. 4 (22). The
use of 31 P NMR proved to be a highly sensitive method to follow the reaction
on solid phase and the NMR spectra were used to rapidly identify a competing
side reaction that would have been difficult to show using other analytical
methodologies. Van Etten used gel phase 31 P NMR to follow protection and
deprotection of a phosphorylated benzyl group in peptide synthesis (23). Spec-
tral resolution was sufficient to distinguish dibenzyl, monobenzyl, and nonben-
zylated forms of a tyrosine phosphate group.

III. MAGIC ANGLE SPINNING NMR

1
A. H Magic Angle Spinning NMR
It would be of great utility if a better methodology could be provided to monitor
polymer-bound reactions than gel phase heteronuclear NMR. However, as previ-
ously mentioned, the utility of gel phase proton NMR in solid phase synthesis
monitoring is of little value, as the spectra obtained are generally quite broad.

© 2000 by Marcel Dekker, Inc.

 Spinning the sample at the magic angle (54.7°), MAS overcomes the
limitations of using 1 H NMR for solvent-swollen resin samples. The broaden-
ing due to magnetic susceptibility and dipolar coupling are substantially re-
moved. 1 H MAS NMR have been applied to study properties of crosslinked
polystyrene gels for some time (24). However, the application of MAS 1 H
NMR in combinatorial chemistry appeared only recently, where it has been
demonstrated that ‘‘high-resolution’’ NMR data for resin-supported molecules
can be obtained (25,26).
MAS NMR requires special equipment to obtain the data. High quality
MAS NMR spectra can be obtained using specially designed probes that are
now commercially available. While any standard MAS probe can also be used
to collect NMR data, the resolution obtained will probably suffer due to mis-
matches in the magnetic susceptibility in the probes (27).
The quality of the NMR data is also dependent on the conditions by
which the sample was prepared. All solvent/resin combinations do not yield
high-quality 1 H MAS NMR spectra (28,29). The most important factor in ob-
taining high-quality 1 H MAS NMR data is the choice of resin, with TentaGel
resins (TGT) generally giving rise to the best-quality data. The choice of swell-
ing solvent can also play a critical role. For the most part, good solvents in-
clude CD 2 Cl 2 , CDCl 3 , C 6 D 6 , DMF-d 7 , and DMSO-d 6. For example, spectra
obtained from Fmoc-Asp(OtBu)-NovaSyn TGT gave good line widths, less
than 6 Hz, for CD 2 Cl 2 , DMF-d 7 , and DMSO-d 6 where as using benzene-d 6
yielded 19.4 Hz. MAS data obtained on Fmoc-Asp(OtBu)-NovaSyn Wang
yielded similar line widths for all the solvents with line widths ranging from
9 to 14 Hz. The compound attached to the resin also plays a role on the ‘‘high-
resolution-like’’ appearance of the data (30).
The quality of the spectra can be enhanced by using presaturation of
the resin signal to reduce the intensity of the aromatic resonances from the
polystyrene or polyethylene glycol resonances in TentaGel-like resins. The
spin-echo experiment distinguishes between the narrow lines of the substrates
and the much broader lines of the polymer and can be used to remove the
peaks from the polymer as well (31–33).

13
B. C MAS NMR
In the late 1980s Frechet showed that high quality 13 C NMR data of swollen
gels could be readily obtained by MAS using a standard MAS probe (34–36).
The first use of 13 C MAS NMR data for combinatorial chemistry demonstrated
that a potential reaction complication, the production of two similar com-
pounds, could be monitored (37). The reaction of norbornane carboxylic acid

© 2000 by Marcel Dekker, Inc.

Figure 5 Utilization of MAS 13 C NMR for comparison of esterification reaction
products for exo and endo norbornane-2-carboxylic acid mixture on Merrifield and
Wang resins.

© 2000 by Marcel Dekker, Inc.

on Merrifield and Wang resins to produce the ester were compared to evaluate
the potential usability of the particular resin in a library synthesis (Fig. 5). It
was found that the stereochemistry of addition was different for the two resins.
In the case of the Wang resin a 60 :40 ratio of exo/endo product was obtained
while the Merrifield resin yielded an 80 :20 ratio. The spectral quality of the
13
C MAS data was virtually indistinguishable from that of the solution 13 C NMR
data. The MAS 13 C NMR data can be collected in a much shorter time frame (20
min) than for gel phase 13 C NMR data which generally take several hours.

C. Other Nuclei
MAS 19 F NMR can also be used to follow appropriate reactions (19). The
MAS data were seen to be an order of magnitude faster to collect than gel
phase NMR for the same sample. While no reports of using MAS 31 P NMR
have appeared at this writing, there is no reason why this nucleus and others
could not prove valuable in following solid phase chemical reactions.

Figure 6 Two-dimensional CH-correlated gel phase NMR spectrum for chloroform-

swollen Boc-Cys(Acm)-OCH 2-Pab-copoly (styrene 1% divinylbenzene).

© 2000 by Marcel Dekker, Inc.

IV. TWO-DIMENSIONAL NMR

Two-dimensional NMR spectroscopy has proven of great value for product

analysis in solution phase chemistry and by its nature should provide a plat-
form by which enhanced data quality and ease of interpretation is obtained.
Geralt demonstrated that the use of 2D CH correlated NMR spectra (Fig. 6)
enhances the utility of gel phase NMR even though the proton NMR spectrum
was broad (36). Since MAS NMR spectra of the swollen resins are of higher
quality than gel phase NMR, it is possible to treat these samples as if they

Figure 7 Two-dimensional CH-correlated NMR spectrum for a mixture of endo (N)

and exo (X) norbornane epimers. (Top) High-resolution solution data for the mixture of
norbornane-2-carboxylic acid. (Bottom) MAS NMR on benzene-swollen Wang resin.

© 2000 by Marcel Dekker, Inc.

were solutions and therefore use the same techniques that would be used in
‘‘high-resolution’’ NMR studies. The application of 2D NMR techniques to
determine the structure on resin exemplifies this notion (26). The direct-ob-
serve MAS CH-correlated data for the exo/endo norbornane mixture on resin
shown in Fig. 7 illustrates the resolution obtainable. Assignments of both pro-
ton and carbon resonances in both major and minor isomers can be readily
made. The combination of heteronuclear multiple quantum correlation
(HMQC) and total correlation spectroscopy (TOCSY) data, as shown in Fig.
8 for Wang-Fmoc-lys-Boc, allows complete assignment of the proton and car-
bon-13 spectra confirming the structure on resin (37). The step-by-step analy-
sis of a multistep solid phase synthesis involving a Heck reaction was conve-
niently followed using MAS 2D NMR techniques (38). The use of 2D NMR

Figure 8 MAS HMQC and TOCSY NMR spectra for Fmoc-lys-tboc on benzene-
swollen Wang resin.

© 2000 by Marcel Dekker, Inc.

TOCSY, HMQC, and nuclear Overhouser enhanced spectroscopy (NOESY)
(39) data as shown in Fig. 9 was instrumental in the evaluation of products.
The main problem for NMR analysis of on-resin products, with the ex-
ception of TentaGel, is that the 1 H NMR spectra are generally broad with
featureless line widths around 10 Hz or more (28). The additional complication
of large polymer resonances can be attenuated by using a spin-echo sequence
but the residual peaks can still be problematic. Peak assignment can be difficult
due to the loss of coupling information relegating the interpretation of the

Figure 9 MAS (A) TOCSY and (B) NOESY NMR spectra for Fmoc-lys-tboc on
benzene-swollen Wang resin.

© 2000 by Marcel Dekker, Inc.

Figure 10 Schematic representation of nontilted 2D J-resolved experiment. Projec-
tion onto the chemical shift axis recovers the high resolution 1D NMR spectrum.

molecular structure to chemical shift information criteria only. Proton-proton

spin coupling values can be obtained by performing a 2D J-resolved experi-
ment (40,41). It was found that by using a nontilted projection from the 2D
J-resolved experiment high quality as indicated in Figure 10, 1D proton NMR
spectra of resin-bound molecules can be obtained. A comparison of the 1 H
MAS NMR spectrum, obtained under spin echo and by the projection of the
nontilted 2-D J-resolved data, for the two methyl groups of DMF-swollen
isoleucine on Wang resin is shown in Fig. 11. The methyl resonances can be
assigned by observation of the coupling patterns. This increase in resolution
is further exemplified for Alloc-Asp derivatized oxazolidinone attached via
its side chain carboxyl to SCAL-linked aminomethylpolystyrene as seen in
Fig. 12 (42).
In the J-resolved projection, the aromatic rings of the SCAL are clearly
present where as the polymer resonances have ‘‘dropped out’’ of the spectrum.
In addition, the proton coupling constants arising from alloc group are readily
measured as seen in the expanded spectrum (Fig. 13). If a more detailed mea-
sure of the couplings is desired, then the full 2D J-resolved spectrum can be
evaluated in the normal manner as exemplified in Fig. 14. A similar method
to obtain accurate proton-proton coupling constants based on E.COSY spectra
has also appeared recently (43).
It would be advantageous if we could retain the enhanced resolution
afforded by the projection of the 2D J-resolved data and also obtain spin sys-
tem connectivities. MAS spin-echo correlated spectroscopy (SECSY) (44)
allows both spin connectivities and enhanced resolution to be obtained. The

© 2000 by Marcel Dekker, Inc.

Figure 11 Projection of nontilted 400 MHz MAS J-resolved NMR spectrum for
methyl region for Ile on DMF-d 6-swollen Wang resin.

© 2000 by Marcel Dekker, Inc.

Figure 12 400 MHz MAS 1 H NMR spectrum for Alloc-Asp-derivatized oxazolidi-
none on SCAL-linked aminomethylpolystyrene. (a) Spin-echo 1 H NMR spectrum. (b)
Nontilted projection from 2D J-resolved spectrum.

projection of the MAS SECSY data for isoleucine on Wang-1 resin shown in
Fig. 15 illustrates the apparent increase in resolution. The apparent coupling
constants observed in the 1D spectrum cannot be used (45).
The recent development of a gradient high-resolution MAS probe will
extend the utility of 2D experiments by removing artifacts that generally ac-
company MAS 2D NMR data on resin samples (46). The lack of artifacts is
illustrated by the high quality SECSY spectrum shown in Fig. 16. SECSY
data contain the same information as a COSY spectrum, but the appearance
of the spectrum is different. The diagonal lies along the F1 ⫽ 0 and the off-

© 2000 by Marcel Dekker, Inc.

Figure 13 Partial NMR spectrum of projection of nontilted J-resolved NMR spec-
trum for Alloc-Asp-derivatized oxazolidinone attached via its side chain carboxyl to
SCAL-linked aminomethyl polystyrene swollen with DMF-d 7.

diagonal peaks occur in pairs along lines that make an angle of 135° with the
diagonal (47). The ability to obtain structural data from a single resin bead
represents the ultimate level of sensitivity for MAS NMR (48). Reasonable
1
H NMR data can now be realized in about 1 h. This result, in principle,
affords identification of the molecule on the resin directly without resorting
to tagging methodologies (49). The incorporation of site specific 13 C labels
aids the ability to obtain heteronuclear NMR data.
Robotic sample changing and automated data collection is an increas-
ingly important aspect in the efficient utilization of MAS NMR for combinato-

© 2000 by Marcel Dekker, Inc.

Figure 14 Partial 2D J-resolved NMR spectrum for Alloc-Asp-derivatized oxazoli-
dinone attached via its side chain carboxyl to SCAL-linked aminomethylpolystyrene
swollen with DMF-d 7.

rial chemistry. Sample changers for both MAS rotors and for nanoprobe sam-
ples are currently available from the instrument manufacturer.

V. MULTIPIN CROWNS

Combinatorial chemistry has developed using resin beads as a primary format

for the synthesis of nonpeptide targets. An alternate format is the crown/pin
system where resin is grafted onto a base polymer (50). FTIR, which is very
useful for analysis of bead resin chemistry, as a general tool for analysis of
crown chemistry is somewhat limited (51). The IR spectrum is hampered by
interfering peaks that arise from the base plastic, the grafted resin, as well as
from the linker molecule.

© 2000 by Marcel Dekker, Inc.

(a)

(b)

(c)

Figure 15 (a) Spin echo (b) projection of the SECSY data MAS 1 H NMR spectrum
for DMF-d 7-swollen Fmoc-isoleucine on Wang resin 1 and (c) solution NMR spectrum
for Fmoc-isoleucine methyl ester.

© 2000 by Marcel Dekker, Inc.

Figure 16 Gradient 2D SECSY NMR spectrum for Alloc-Asp-derivatized oxazoli-
dinone on SCAL-linked aminomethylpolystyrene swollen with DMF-d 7 obtained with
a 4-mm high-resolution gradient MAS probe.

One dimensional MAS NMR is not affected to the same extent as is IR.
Reasonable NMR data were obtained using a 7-mm high-resolution MAS
probe utilizing the spin-echo pulse sequence, allowing the removal of the
broader resonances due to the plastic as well as the signals from the slower
moving components of the resin. Use of MAS NMR allows the same crown
to be sequentially monitored and returned to the reaction conditions until the
desired transformation is complete (Scheme 1). The ability to follow the trans-
formation of alcohol to aldehyde and then elaboration to olefin on the same
crown exemplifies this technique (Fig. 17). The stereochemistry of the olefin
was found to be trans as determined from the measured coupling constant of
16 Hz. Confirmation of the olefinic peaks was obtained by COSY (52). MAS
NMR, following cleavage, on the same crown, showed complete absence of
product peaks.

© 2000 by Marcel Dekker, Inc.

 Scheme 1

Figure 17 (A) 400 MHz spin-echo MAS 1 H NMR spectrum of a crown with alde-
hyde product attached to the grafted polymer and (B) with olefin attached, swelled in
DMF-d 7.

© 2000 by Marcel Dekker, Inc.

VI. SOLUTION TECHNIQUES
A. HPLC/NMR
While the majority of attention in combinatorial syntheses has been on solid
phase analysis, the use of traditional solution phase organic chemistry to form
compound collections should not be disregarded. MS techniques are widely
used for evaluation of mixtures produced by combinatorial chemistry (4).
However, a potential problem with the MS methodology involves a situation
when isomolecular weight compounds are present. The compounds can be
stereoisomers, positional isomers or by chance identical molecular weight ma-
terials. In these cases, the identity of the substance as determined by MS can
be ambiguous.
The recent redevelopment of HPLC-NMR allows for complete identifi-
cation of individual compounds in complex mixtures (53). While the majority
of reports using HPLC-NMR have been in the drug metabolism area, the utility
for organic compounds and for peptides has been recently demonstrated (54).
It is presently possible to obtain routine high-quality NMR data using this
technique with as little as 5 mg of compound in the chromatogram peak, with
the detection limit using this technology presently being on the order of 100
ng (55).
The utility of HPLC-NMR for analysis of isomolecular weight com-
pounds is illustrated by the analysis of a mixture of dimethoxybenzoylglycines
prepared by split-mix synthesis having aromatic ring positional isomers. The
results from the dimethoxybenzoylglycines was analyzed by stop-flow HPLC-
NMR where individual components are evaluated as each peak enters the
NMR probe while the separation is paused. The assignment of structure from
the separated components is straightforward by consideration of the shifts and
the coupling patterns for the region of interest shown in the stacked plot NMR
spectrum shown in Fig. 18.
On-flow HPLC-NMR analysis can also be performed when sufficient
material is available. It involves collecting the NMR data continuously as the
sample passes through the probe. This is the most efficient method for structure
evaluation by HPLC-NMR. The NMR data are represented in a 2-D plot where
the x direction contains chemical shift information and the y direction is repre-
sentative of the LC retention time. The individual spectra can be extracted
from the 1D slices along the x axis if so desired. The resolutions in the individ-
ual spectra are of somewhat lower quality than in the stop-flow method; how-
ever, the introduction of the second dimension allows for easy structure assign-
ment even for overlapping peaks in the LC separation. As seen in Fig. 19, the
on-flow HPLC-NMR characterization shows four distinct sets of resonances.
The HPLC/NMR for peptides is illustrated by the data obtained for pen-

© 2000 by Marcel Dekker, Inc.

Figure 18 500 MHz 1 H HPLC/NMR stack plot of chemical shift vs. retention time
for a mixture of isomolecular weight dimethoxybenzoylglycines.

tapeptide mixtures FNXEF-OH where X ⫽ D, Q, I, K, or T. In each of these

five component mixtures there are two compounds having identical molecular
weights; therefore, it would be difficult to unambiguously assign structures
by MS without resorting to special techniques. Stop-flow HPLC-NMR al-
lowed all of the compounds to be unambiguously assigned utilizing 2D
TOCSY data as exemplified in Fig. 20 for the compound having a retention
time of 10.8 min. The TOCSY spectrum showed resonances at δ ⫽ 1.3, 1.6,
2.9, and 4.1, indicative of a lysine residue. The additional resonances are con-
sistent with the amino acids; phenylalanine δ ⫽ 4.2 and 3.1, and 4.6, 3.18,
and 2.98; asparagine δ ⫽ 4.7, 2.7, 2.55; and glutamate δ ⫽ 4.27, 2.34–2.05.
This clearly identifies the peptide as FNKEF-OH. The isomolecular weight
complication of lysine vs. glutamine was readily resolved. In another example,
using standard HPLC conditions, assignment of a majority of compounds from
a library of 27 tripeptides (AYM) was achieved (56). This task, which was
accomplished in a single pass using on-flow HPLC-NMR, would have been
difficult if not impossible by any other technique.

© 2000 by Marcel Dekker, Inc.

Figure 19 On-flow 500 MHz HPLC-NMR contour plot spectrum of chemical shift
vs. retention time for a mixture of isomolecular weight dimethoxybenzoylglycines.

The simultaneous determination of structure by the combination of

HPLC-NMR-MS should prove to be an extremely powerful tool for combina-
torial chemistry.

B. Flow NMR
The ability to obtain routine NMR data in a reasonable time frame is greatly
enhanced by the use of robotic sample changers. Currently it is possible to
obtain data from over 100 samples in an unattended manner. The inefficiency
in this procedure is that the samples generally are made by manual methods
restricting the potential throughput and capacity. Mass spectrometry tech-
niques have been developed to obtain spectra in about a minute or so each,
using an autosampler that can take samples directly from 96 welled plates.
These same plates are also used in the high-throughput screening efforts,
thereby minimizing the effort to prepare samples. It would be convenient if

© 2000 by Marcel Dekker, Inc.

Figure 20 2D TOCSY spectrum for FNKEF-OH pentapeptide derivative obtained
using stop-flow HPLC-NMR. Amino acid assignments are indicated by the letter code.

the same format could be used to obtain NMR spectra. Flow NMR has been
recently introduced using a similar format MS and is commercially available.
NMR most likely won’t be able to obtain data in the same time scale as MS;
however, high-quality spectra can most probably be obtained in around 3 min.
The robotics used to obtain these spectra will allow samples to be spot-checked
or to permit one to obtain data on the identified active well as suggested in
the cartoon in Fig. 21.
Since the solvent used in these wells is generally DMSO/water, this
makes the flow NMR experiment that much more difficult. Techniques have
been described to efficiently suppress the protons from multiple solvents and
obtain NMR data on small samples in a relatively short time (57,58). High-
quality NMR data on a 26-µg sample can be obtained in just over 2 min using

© 2000 by Marcel Dekker, Inc.

Figure 21 Schematic representation of flow NMR using 96 welled plates.

either a microprobe or an HPLC-NMR probe. The HPLC probe in principle

can be adapted to handle streams of samples using a modified stop-flow proce-
dure. The perfection of this technique will replace the automation systems as
we presently know them and may render NMR tubes superfluous.
One aspect of combinatorial chemistry is the synthesis of mixtures of
structurally related compounds. This facilitates a high throughput in both syn-
thesis and screening. The split-and-mix synthesis technique produces a mix-
ture of compounds as the final product (59,60). In developing the synthesis
of large libraries, smaller test systems are studied to optimize synthetic strate-
gies. The utility of NMR for studying intact mixtures has not been extensively
demonstrated. 2D NMR methods such as TOCSY have been used in relatively
simple mixture analysis (61). However, for compounds that have their spin
systems insulated by groups such as esters or ethers, TOCSY methodologies
are not sufficient for complete analysis.

© 2000 by Marcel Dekker, Inc.

 Recently, the use of pulsed-field gradient (PFG) technology to obtain
diffusion coefficients of molecules has been demonstrated as a useful tech-
nique for mixture analysis (53). Unlike any other 2D experiment, size-resolved
or diffusion-resolved NMR assigns the resonances based on the diffusion coef-
ficient for each proton (or other spin) in the molecule and therefore can be
used to distinguish resonances arising from different molecules (63–70) (Fig.
22). A method that involves the use of PFG and TOCSY, called diffusion-
encoded spectroscopy (DECODES), simplifies mixture analysis by NMR (71).
The combination of PFG and TOCSY ‘‘decodes’’ the spin systems, allowing
individual components in complicated mixtures to be assigned. A typical DE-
CODES spectrum obtained in this manner is shown in Fig. 23. The use of
TOCSY aids the calculation of the diffusion coefficient and determination of
molecular identity.
The synthesis and screening of mixtures of compounds offers increased
efficiency and throughput compared to making and testing individual com-
pounds. However, utilization of mixtures of compounds requires a method to
determine which molecule in the mixture is responsible for the desired effect.
Typically, mixtures of compounds are prepared by design and these mixtures
are tested without separation. When there is evidence of sufficient activity,

Figure 22 Schematic representation of molecular translation in solution. The DOSY

experiment spatially encodes the molecule in 1 and then DECODES the movement in
2. The spectrum indicates the translational movement by a decreased observed signal
intensity.

© 2000 by Marcel Dekker, Inc.

Figure 23 Selected region of TOCSY spectrum obtained via DECODES sequence
of an ester mixture in benzene. Peaks arise from propyl acetate, ethyl butyryl acetate,
and butyl levulinate.

the mixture is deconvoluted to identify the active component. Several ap-

proaches to identify interesting components in a mixture have been described
(72–77). Methods that identify active components of mixtures without the
need for deconvolution could eliminate ‘‘false positives’’ and greatly reduce
the effort required to analyze mixtures. One such method under investigation
is affinity MS (78–80).
A somewhat similar method, termed affinity NMR, has shown that the
diffusion coefficient of a small molecule binding with a ‘‘receptor’’ in solution
is significantly different from the small compound alone (81). Thus the inter-

© 2000 by Marcel Dekker, Inc.

Figure 24 (a) 1D 400 MHz 1 H NMR spectrum of the nine-component mixture in
CDCl 3. (1) dl-isocitric lactone, (2) (S)-(⫹)-O-acetylmandelic acid, (3) dl-N-acetylho-
mocysteine thiolactone, (4) (⫹)-sec-butyl acetate, (5) propyl acetate; (6) isopropyl bu-
tyrate, (7) ethyl butyryl acetate, (8) butyl levulinate, (9) hydroquinine-9-phenanthryl
ether. (b) PFG 1D 1 H NMR spectrum of the mixture without hydroquinine-9-phe-
nanthryl ether. (c) PFG 1D 1 H NMR spectrum of the nine component mixture using
LED sequence. Chemical shifts arising from compounds 1 and 2 are shown. All other
shifts are from compound 9. The PFG conditions were the same as in (b).

© 2000 by Marcel Dekker, Inc.

acting molecules can be distinguished from inert molecules in a manner remi-
niscent of physical separation of molecules by affinity chromatography. Affin-
ity NMR, using hydroquinine 9-phenanthryl ether as a model receptor, was
successfully applied to a nine-component mixture containing seven inert mate-
rials and the two carboxylic acids shown.
Figure 24 shows the normal 1D 1 H NMR spectrum for the nine compo-
nent mixture; without PFG, a control experiment performed on the mixture
in the absence of hydroquinine 9-phenanthryl ether under identical PFG condi-
tions and the 1D 1H NMR spectrum of the same mixture under the PFG condi-
tions. Only signals from hydoquinine 9-phenanthryl ether and compounds 1
and 2 are observed in the bottom spectrum. As expected no NMR signals are
present in the absence of molecular interactions. The structures of the com-
pounds that interacted with hydroquinine 9-phenanthryl ether were identified
directly in the mixture using the DECODES method.
Since the relatively high concentration of each component required
by NMR adds up to a high total concentration of compounds for the mixture,
the application of this methodology to screen combinatorial chemistry mix-
tures for biological activity will likely be limited by the total compound con-
centration tolerated by the biological target. Nevertheless, this NMR method,
when applied to suitable systems, should add a powerful tool for mixture anal-
ysis.

C. SAR by NMR
A method for identifying active compounds from a library of low molecular
weight ligands using 15 N-labeled proteins has been recently reported (82). The
binding is determined by the 15 N or 1 H chemical shift changes in the protein
upon the addition of the ligand. The method, which at present is limited to
small biomolecular receptors, promises to play an integral part of the drug
discovery process.

VII. FUTURE

‘‘I am not aware of any other field of science outside of NMR that offers so
much creative freedom and opportunity for a creative mind to invent and ex-
plore new experimental schemes that can be fruitfully applied in a variety of
disciplines.’’—Richard Ernst, Nobel Prize, 1993

© 2000 by Marcel Dekker, Inc.

REFERENCES

1. CA Fyfe. Solid State NMR for Chemists; CFC Press: Guelph, Ontario, 1983
2. B Yan, G Kumaravel, H Anjaria, A Wu, R Petter, C Jewell, J Wareing. Infrared
spectrum of a single resin bead for real-time monitoring of solid-phase reactions.
J Org Chem 60:5736–5738, 1995.
3. B Yan, G Kumaravel. Probing solid-phase reactions by monitoring the IR bands
of compounds on a single ‘‘flattened’’ resin bead. Tetrahedron 52:843–848,
1996.
4. JA Boutin, P Hennig, PS Bertin, L Petit, J-P Mahieu, B Serkiz, J-P Volland,
J-L Fauchere. Combinatorial peptide libraries: robotic synthesis and analysis by
nuclear magnetic resonance, mass spectrometry, tandem mass spectrometry, and
high performance capillary electrophoresis techniques. Anal Biochem 234:126–
141, 1996.
5. BJ Egner, GJ Langley, M. Bradley. Solid phase chemistry: direct monitoring by
matrix-assisted laser desorption/ionization time of flight mass spectrometry. A
Tool for Combinatorial Chemistry. J Org Chem 60:2652–2653, 1995.
6. K McMellop, W Davidson, G Hansen, D Freeman, N Pallai. The characterization
of crude products from solid-phase peptide synthesis by ν-HPLC/fast atom bom-
bardment mass spectrometry. Peptide research 4:40–46, 1991.
7. H Sterlicht, GL Kenyon, EL Packer, J Sinclair. J Am Chem Soc 93:199–208,
1971.
8. R Epton, P Goddard, KJ Ivin. Gel phase 13C NMR spectroscopy as an analyti-
cal method in solid (gel) phase peptide synthesis. Polymer 21:1367–1371,
1980.
9. FGW Butwell, R Epton, EJ Mole, N Muzaffar, S Phillips. Deprotection studies
in ultra-high load solid (gel) phase peptide synthesis. 13C NMR investigation
of the efficacy of boron trifluoride-based side-chain deprotection cocktails. Innov
Persp Solid Phase Synth Collect Pap Int Symp 121–132, 1990.
10. E Geralt, J Rizo, E Pedroso. Application of gel-phase 13C-NMR to solid-phase
peptide Synthesis. Tetrahedron 40:4141–4152, 1984.
11. D Liebfritz, W Mayr, R Oekonomopulos, J Jung. 13C NMR spectroscopic stud-
ies on the conformation during stepwise synthesis of peptides bound to solubiliz-
ing polymer supports. Tetrahedron 34:2045-2050 1978.
12. SL Mannatt, D Horowitz, R Horowitz, RP Pinnell. Solvent swelling for enhance-
ment of carbon-13 nuclear magnetic resonance spectral information from insolu-
ble polystyrenes. Anal Chem 52:1529–1534, 1980.
13. AJ Jones, CC Leznoff, PI Svirskaya. Characterization of organic substrates
bound to cross-linked polystyrenes by 13C NMR spectroscopy. Org Magn Reson
18:236–240, 1982.
14. BE C Lossey, RG Cannon, WT Ford, M Periyasamy, S Mohanraj. Synthesis,
reactions, and 13C FT NMR spectroscopy of polymer-bound steroids. J Org
Chem 55:4664–4668, 1990.

© 2000 by Marcel Dekker, Inc.

15. MM Azure, B Calas, A Cave, J Parello. Synthese peptidique en phase solide.
Caracterisation par RMN du proton d’un peptide immobilise sur un support poly-
acrylique. C R Acad Sc Paris 303:553–556, 1986.
16. GC Look, CP Holmes, JP Chinn, MA Gallop. Methods for combinatorial organic
synthesis: the use of fast 13C NMR analysis for gel phase reaction monitoring.
J Org Chem 59:7588–7590, 1994.
17. GC Look, MM Murphy, DA Campbell, MA Gallop. Trimethylorthoformate: a
mild and effective dehydrating reagent for solution and solid phase imine forma-
tion. Tetrahedron Lett 36:2937–2940, 1995.
18. SL Manatt, SF Amsden, CA Bettison, WT Frazer, JT Gudman, BE Lenk, JF
Lubetich, EA McNelly, SC Smith, DJ Templeton, RP Pinnell. A fluorine-19
NMR approach for studying Merrifield solid-phase peptide synthesis. Tetrahe-
dron Lett 21:1397–1400, 1980.
19. MJ Shapiro, G Kumaravel, RC Petter, R Beveridge. 19F NMR monitoring of a
SNAr reaction on solid support. Tetrahedron Lett 37:4671–4674, 1996.
20. A Svensson, T Fex, J Kihlberg. Use of 19F NMR spectroscopy to evaluate reac-
tions in solid phase organic synthesis. Tetrahedron Lett 37:7649–7652, 1996.
21. F Bardella, R Eritja, E Pedroso, E Geralt. Gel-phase 31P-NMR. A new analytical
tool to evaluate solid phase oligonucleotide synthesis. Bioorg Med Chem Lett
3:2793–2796, 1993.
22. CR Johnson, B Zhang. Solid phase synthesis of alkenes using the Horner-Wad-
sworth-Emmons reaction and monitoring by gel phase 31P NMR. Tetrahedron
Lett 36:9253–9256, 1995.
23. Z Tian, C Gu, RRW Roeske, M Zhou, RL Van Etten. Synthesis of phosphotyro-
sine-containing peptides by the solid-phase method. Int J Peptide Protein Res
42:155–158, 1993.
24. B Schneider, D Doskocilova, JJ Dybal. Motional restrictions and chain confor-
mation in various swollen crosslinked polystyrene gels from 1H NMR line-shape
analysis. J Polymer 26:253–259, 1985.
25. WL Fitch, G Detre, CP Holmes, JN Schoolery, P Keifer. High resolutions 1H
NMR in solid phase organic synthesis. J Org Chem 59:7955–7956, 1994.
26. RC Anderson, MA Jarema, MJ Shapiro, JP Stokes, M Ziliox. Analytical tech-
niques in combinatorial chemistry: MAS CH correlation in solvent-swollen resin.
J Org Chem 60:2560–2651, 1995.
27. PA Keifer, L Baltosis, DM Rice, AA Tymiak and JN Shoolery. A comparison of
NMR spectra obtained for solid-phase-synthesis resins using conventional high-
resolution, magic-angle-spinning, and high-resolution magic-angle-spinning
probes. J Magn Reson A 119:65–75, 1996.
28. P Keifer. Influence of resin structure, tether length, and solvent upon the high-
resolution 1H NMR spectra of solid-phase-synthesis resins. J Org Chem 61:
1558–1559, 1996.
29. IE Pop, CF Dhalluin, BP Deprez, PC Melnyk, GM Lippens, AL Tartar. Monitor-
ing of a three-step solid phase synthesis involving a heck reaction using magic
angle spinning NMR spectroscopy. Tetrahedron 52: 12209–12222, 1996.

© 2000 by Marcel Dekker, Inc.

30. MJ Shapiro, unpublished work.
31. EL Hahn, Spin echoes Phys Rev 80:580 1950.
32. DL Rabinstein, KK Mills, EJ Strauss. Proton NMR spectroscopy of human blood
plasma and red blood cells. Anal Chem 60:1380A–1391A, 1988.
33. T Wehler, J Westman. Magic angle spinning NMR: a valuable tool for monitor-
ing the progress of reactions in solid phase synthesis. Tetrahedron Lett 37:4771–
4774, 1996.
34. HDH Stover, JM Frechet. Direct polarization 13C and 1H magic angle spinning
NMR in the characterization of solvent-swollen gels. J Macromolecules 22:
1574–1576, 1989.
35. HDH Stover, JM Frechet. NMR characterization of cross-linked polystyrene
gels. J Macromolecules 24:883–888, 1991.
36. E Giralt, F Alberico, F Bardella, R Eritja, M Feliz, E Pedroso, M Pons, J Rizo.
Gel-phase NMR spectroscopy as a useful tool in solid phase synthesis. Innov
Persp Solid Phase Synthesis Collect. Pap., Int Symp. R. Epton (ed.) 111–120,
1990.
37. RC Anderson, MJ Shapiro, JP Stokes. Structure determination in combinatorial
chemistry: utilization of magic angle spinning HMQC and TOCSY NMR spectra
in the structure determination of Wang-bound lysine. Tetrahedron Lett 36:5311–
5314, 1995.
38. IE Pop, CF Dhalluin, BP Deprez, PC Melnyk, GM Lippens, AL Tartar Monitor-
ing of a three-step solid phase synthesis involving a heck reaction using magic
angle spinning NMR spectroscopy. Tetrahedron 52:12209–12222, 1996.
39. A Kumar, RR Ernst, K Wuthrich. A two-dimensional nuclear overhauser en-
hancement (2D NOE) experiment for the elucidation of complete proton-proton
cross-relaxation networks in biological macromolecules. Biochem Biophys Res
Commun 95:1–6, 1980.
40. WP Aue, E Bartholdi, RR Ernst. Two-dimensional spectroscopy. Application to
nuclear magnetic Resonance. J Chem Phys 64:2229–2246, 1976.
41. MJ Shapiro, J Chin, RE Marti, MA Jarosinski. Enhanced resolution in MAS
NMR for combinatorial chemistry. Tetrahedron Lett 38:1333–1336, 1997.
42. M Patek, M Lehl. Safety-catch anchoring linkage for synthesis of peptide amides
by Boc/Fmoc Strategy. Tetrahedron Lett 32:3891–3894, 1991.
43. A Meissner, P Bloch, E Humpfer, M Spraul, OW Sorensen. Reduction of inho-
mogeneous line broadening in two dimensional high-resolution MAS NMR spec-
tra of molecules attached to swelled resins in solid-phase synthesis. J Am Chem
Soc 119:1787–1788, 1997.
44. K Nagayama, K Wuthrich, RR Ernst. two-dimensional spin echo correlated spec-
troscopy (SECSY) for 1H NMR studies of biological macrcomolecules. Biochem
Biophys Res Commun 90:305–311, 1979.
45. Y Kim, JH Prestegard. Measurement of vicinal couplings from cross peaks in
COSY spectra, J Magn Reson 84:9–13, 1989.
46. WE Mass, FH Laukien and DG Cory. Gradient, high resolution, magic angle
sample spinning NMR. J Am Chem Soc 118:13085–13086, 1996.

© 2000 by Marcel Dekker, Inc.

47. The diagonal peaks lie along the F2 at F1 ⫽ 0 axis and represent the chemical
shift as in a COSY spectrum. The F1 axis represents one-half of the chemical
shift difference between the coupled spins.
48. S Sarkar, RS Garigipati, JL Adams, PA Keifer. An NMR method to identify
nondestructively chemical compounds bound to a single solid-phase-synthesis
bead for combinatorial chemistry applications. J Am Chem Soc. 118:2305–2306,
1996.
49. A Borchard, WC Still. Synthetic receptor binding elucidated with an encoded
combinatorial library. J Am Chem Soc 116:373–374, 1994.
50. RM Valerio, AM Bray, RA Campbell, A Dipasquale, C Margellis, S. J. Rodda,
HM Geysen, NJ Maeji. Multipin peptide synthesis at the micromole scale using
2-hydroxyethyl methacrylate grafted polyethylene supports. Int J Peptide. Protein
Res 42:1–9, 1993.
51. HU Gremlich, SL Berets. Use of FT-IR internal reflection spectroscopy in combi-
natorial chemistry applied spectroscopy 50:532–536, 1996.
52. WP Aue, E Bartholdi, RR Ernst. Two dimensional spectroscopy. Application to
nuclear magnetic resonance. J Chem Phys 64:2229, 1976.
53. M Spraul, M Hoffman, P Dvortsak, JK Nicholson, ID Wilson. High-performance
liquid chromatography coupled to high-field proton nuclear magnetic resonance
spectroscopy: application to the urinary metabolites of ibuprofen. Anal Chem
65:327–330, 1993.
54. K Albert Angew. Direct on-line coupling of capillary electrophoresis and 1H
NMR spectroscopy. Chem Int Ed Engl 34:641–642, 1995.
55. DL Olson, TL Peck, AG Webb and JV Sweedler. On-line NMR detection for
capillary electrophoresis applied to peptide analysis. Peptides: Chem Struct Biol
PTP Pravin, RS Hedges, (eds.). Mayflower Scientific, Ltd. 730–731, 1996.
56. JC Lindon, RD Farrant, PN Sanderson, PM Doyle, SL Goughg, M Spraul,
M Hoffman, JK Nicholson. Separation and characterization of components of
peptide libraries using on-flow coupled HPLC-NMR spectroscopy. Magn Reson
Chem 33:857–863, 1995.
57. S Smallcombe, SL Patt, PA Keifer. WET solvent suppression and its applications
to LC NMR and high-resolution NMR spectroscopy. J Magn Reson A 117:295–
303, 1995.
58. C Dalvit, JM Bohlen. Multiple-solvent suppression in double-quantum NMR ex-
periments with magic angle pulsed field gradients. Magn Reson Chem 34:829–
833. 1996.
59. A Furka, F Sebestyan, M Asgedom, G Dibo. General method for rapid synthesis
of multicomponent peptide mixtures. Int J Peptide Protein Res 37:487–493,
1991.
60. F Sebestyen, G Dibo, A Kovacs, A Furka. Chemical synthesis of peptide librar-
ies. Bioorg Med Chem Lett 3:413–418, 1993.
61. K Johnson, LG Barrientos, PPN Murthy. Application of two-dimensional total
correlation spectroscopy for structure determination of individual inositol phos-
phates in a Mixture. Anal Biochem 231:421–431, 1995.

© 2000 by Marcel Dekker, Inc.

62. P Stilbs, K Paulsen, PC Griffiths. Global least squares analysis of large, corre-
lated spectra data sets. Application to component-resolved FT-PSGE NMR spec-
troscopy. J Chem Phys 100:8180–8189, 1996.
63. SJ Gibbs, CS Johnson Jr. A PFG NMR experiment for accurate diffusion and
flow studies in the presence of eddy currents. J Magn Reson 93:395–402, 1991.
64. KF Morris, CS Johnson Jr. Diffusion-ordered two-dimensional nuclear magnetic
resonance spectroscopy. J Am Chem Soc 114:3139–3141, 1992.
65. KF Morris, CS Johnson Jr. Resolution of discrete and continuous molecular size
distributions by means of diffusion-ordered 2D NMR spectroscopy. J Am Chem
Soc 115:4291–4299, 1993.
66. KF Morris, P Stilbs, CS Johnson Jr. Analysis of mixtures based on molecular
size and hydrophobicity by means of diffusion-ordered 2D NMR. Anal Chem
66:211–215, 1994.
67. M Lin, DA Jayawickrama, RA Rose, JA Delvisio, CK Larive. Nuclear magnetic
resonance spectroscopy analysis of the selective complexation of the cis and
trans isomers of phenylalanylproline by B-cyclodextrin. Anal Chim Acta 307:
449–457, 1995.
68. M Liu, JK Nicholson, JC Lindon. High-resolution diffusion and relaxation edited
one- and two-dimensional 1H NMR spectroscopy of biological fluids. Anal
Chem 68 3370–3376, 1996.
69. N Birlirakis, E Guittet. A new approach in the use of gradients for size-resolved
2D-NMR experiments. J Am Chem Soc 118:13083–13084, 1996.
70. H Barjat, GA Morris, S Smart, AG Swanson, SC Williams. High-resolution dif-
fusion-ordered 2D spectroscopy (HR-DOSY)—A new tool for the analysis of
complex mixtures. J Magn Reson Series B 108:170–172, 1995.
71. M Shapiro, M Lin. Mixture analysis in combinatorial chemistry. Application of
diffusion-resolved NMR spectroscopy. J Org Chem 61:7617–7619 1996.
72. CT Dooey, NN Chung, BC Wilkes, PW Schiller, JM Bidlack, GW Pasternak,
RA Houghten. An all d-amino acid opioid peptide with central analgesic activity
from a combinatorial library. Science 266:2019–2022, 1994.
73. K Burgess. Combinatorial technologies involving reiterative division/coupling/
recombinations: statistical considerations. J Med Chem 37:2985–2987, 1994.
74. E Erb, KD Janda, S Brenner. Recursive deconvolution of combinatorial chemical
libraries. Proc Nat Acad Sci USA 91:11422–11426, 1994.
75. DJ Ecker, TA Vickers, R Hanecak, V Driver, K Anderson. Rational screening
of oligonucleotide combinatorial libraries for drug discovery. Nucleic Acids Res
21:1853–1856, 1993.
76. R Fathi, MJ Rudolph, RG Gentles, R Patel, EW MacMillan, MS Reitman,
D Pelham, AF Cook. Synthesis and properties of combinatorial libraries of phos-
phoramidates. J Org Chem 61:5600–5609, 1996.
77. S Borman, C&E N, 29–54, 1996.
78. RS Youngquist, GR Fuentes, MP Lacey, T Keough. Generation and screening
of combinatorial peptide libraries designed for rapid sequencing by mass spec-
trometry. J Am Chem Soc 117:3900–3906, 1995.

© 2000 by Marcel Dekker, Inc.

79. X Cheng, R Chen, JE Bruce, BL Schwartz, GA Anderson, HSA Ofstadler, DC
Gale, RD Smith, J Gao, GB Sigal, M Mammen, GM Whitesides. Using electro-
spray ionization FTICR mass spectrometry to study competitive binding of inhib-
itors to carbonic anhydrase. J Am Chem Soc 117:8859–8860, 1995.
80. Y-H Chu, YM Dunayevskiy, DP Kirby, P Vouros, BL Karger. Affinity capillary
electrophoresis–mass spectrometry for screening combinatorial libraries. J Am
Chem Soc 118:7827–7835, 1996.
81. M Lin, MJ Shapiro, JR Wareing. Diffusion-edited NMR—affinity NMR for di-
rect observation of molecular interactions. J Am Chem Soc 119:5249–5250,
1997.
82. SB Shuker, PJ Hajduk, RP Meadows, SW Fesik. Discovering high-affinity li-
gands for proteins: SAR by NMR. Science 274:1531–1534, 1996.

© 2000 by Marcel Dekker, Inc.

5
The Role of Liquid Chromatography

Michael E. Swartz
Waters Corporation
Milford, Massachusetts

I. INTRODUCTION

Chromatography alone, or in combination with other analytical techniques,

has been used for a number of years in the drug discovery process in support
of traditional organic synthesis for compound identification, compound purity
and stability determinations, from lead discovery to final lead optimization,
testing, and candidate selection. However, in response to increasing demands
in the pharmaceutical industry to accelerate the drug discovery process and
identify lead compounds in increasing numbers, new avenues of approach,
such as combinatorial chemistry, must be investigated.
Combinatorial chemistry synthesis techniques have presented new chal-
lenges to the analytical chemist. During lead discovery, libraries of large num-
bers of compounds, numbering from 10–20, to hundreds, thousands, ten of
thousands, or even millions of compounds are generated. Therefore, due to
the sheer numbers of compounds, assays must be rapid, as well as capable of
determining quantity, purity, and whether or not the proper compound was
synthesized. Further along the drug discovery path, during lead optimization
and testing leading to candidate selection, compounds are required in larger
quantities. Analytical techniques used in lead discovery now give way to pre-
parative, mass-directed autopurification techniques, capable of isolating and
purifying 10–20 mg of the compound of interest during a single chromato-
graphic analysis. Furthermore, all of the assays, from the analytical to prepara-

© 2000 by Marcel Dekker, Inc.

tive scale, must be accessible to everyone in the drug discovery process, in
what has become to be known as an ‘‘open access’’ environment. In response
to these challenges, chromatographers have had to rethink their strategy in
order to provide timely and complete information and feedback.
In order to be successful, a proper method, like any other type of assay,
run on the appropriate instrument, must be employed. In addition, software,
both for systems operation/integration and data analysis, also plays a central
role. This chapter will address how chemists, using chromatography, have
adapted in answering the challenges presented by a combinatorial chemistry
program of drug discovery. While not intended to be an exhaustive review,
this chapter focuses predominantly on liquid chromatography (LC), discussing
various applications and techniques that highlight its use in the drug discovery
process.
Since application of fast, generic LC methods with mass spectrometry
(LC/MS) is emerging as the technique of choice for assessing the progress
and final quality of large combinatorial arrays in drug discovery, it will be
discussed in some detail, along with other detection techniques. Mass-directed
purification and characterization on the preparative scale will also be ad-
dressed.

II. BASIC HIGH-THROUGHPUT ANALYSIS AND

CHARACTERIZATION
A. Chromatographic Optimization and Injection Overhead
Reduction
The first step in drug discovery is lead identification before high-throughput
screening. This step is characterized by use of combinatorial libraries varying
in size from a few hundred to tens of thousands (or more) compounds. Assays
should provide simple, basic information such as compound identification on
a molecular weight basis, and the synthesis yield (purity). In order to answer
these questions, methods must satisfy certain requirements. Gradient reverse
phase liquid chromatography (RP-LC) with various detection modes can sat-
isfy these requirements and is currently the method of choice for the analysis
of combinatorial libraries and synthesis-related products (1–5).
The sheer number of samples, their diversity, and a lack of suitable
standards for quantitation can at first seem an insurmountable challenge. Since
method development for individual samples is not feasible due to time and
economic factors, a broadly applicable ‘‘generic’’ method must be developed,
without sacrificing information content. Also, an inject-to-inject cycle time of

© 2000 by Marcel Dekker, Inc.

2–5 min (or less!) is desired to accomplish the required sample throughput.
Methods and instruments also need to be compatible with various detection
methods in addition to MS, such as photodiode array (PDA), evaporative light
scattering, and nitrogen- and sulfur-specific chemiluminescence detection. Fi-
nally, it is important to keep in mind that once a lead has been identified, it
may be necessary to scale up the separation (with instrument and chemistry
implications) to isolate larger amounts of pure material for additional studies.
Chromatographic instruments used in support of combinatorial chemis-
try require features somewhat different from those of traditional systems (5–
8). The basic components of a system are the same: a solvent manager (pump),
a sample manager (autosampler or injector), a detector(s), a data system, and,
in some instances, a fraction collector. However, some important differences
exist. The most significant difference is the sample manager. Samples can be
presented in standard vials, tubes, or microtiter plates of various sizes and
capacities, either singly or in various combinations. Therefore some type of
‘‘XYZ’’ sample management device is dictated. Other important differences
include software instrument access and control, and data reduction and re-
porting.
Let’s examine a typical hypothetical situation. Assume an analyst has
just received a combinatorial library on six high-density 384-well microtiter
plates, for a total of 2304 samples of which all are unknown, all are different,
and all differ in the degree of purity, for analysis in the lab. The traditional
approach would be to use a gradient RP-LC method similar to that presented
in Fig. 1. Using a long column and a shallow gradient is typically the first
step in developing and optimizing a method. However, given the number of
compounds involved, individual method development is highly impractical.
Under the conditions used in Fig. 1, including 20 min of postrun reequilibra-
tion time, analysis of the 2304 samples would take more than 96 days! Al-
though multiple systems could be used to analyze the samples, additional steps
must be taken to improve sample throughput and to maintain final method
detector compatibility in a full-time combinatorial chemistry support labora-
tory.
Several options, either alone or in combination, can be employed to
improve sample throughput. Obvious options include using shorter columns,
higher flow rates and temperatures, and/or sacrificing resolution. Figure 2A
shows a separation of the same sample illustrated in Fig. 1, but with a smaller
column at higher flow rates. The chromatographic test sample used spans a
wide elution range; however, the peak capacity of the method remains high.
Under these conditions, the total time for the analysis of the 2304 samples
was decreased by an order of magnitude to less than 8 days. The column used

© 2000 by Marcel Dekker, Inc.

Figure 1 Separation of a chromatographic test mixture by a traditional gradient
method. Separation was performed on Waters Alliance® HPLC System (Waters Cor-
poration, Milford, MA) and a 3.9 by 150mm 5 micron particle size Symmetry® C18
column at 30°C. The mobile phase consisted of 0.1% phosphoric acid as the A solvent,
and acetonitrile as the B solvent, run as a linear gradient from 0–80% B over 40 minutes
at 1.0 mL/min. Total analysis time does not include twenty minutes of post run re-
equilibration of the column and system. UV detection at 254nm, and a 20 µL injection
was used. Peaks 1–12 (0.1mg/mL each in 50/50 methanol/water) are uracil, theophy-
line, acetylfuran, acetanilide, acetyl-, propio-, butyro-, benzo-, valero-, hexano-, hep-
tano-, and octano-phenone, respectively.

to generate the example chromatogram shown in Fig. 2A had a smaller particle

size (3.5 µm) and an increased internal diameter was used (4.6 mm) to accom-
modate the increased flow rate of 4 mL/min. A formic acid instead of a phos-
phate-buffered mobile phase was used for better MS compatibility.
Besides using a lot of mobile phase, analyses run at these high flow
rates are not compatible with an MS detector without some sort of split to
divert flow. For many LC analyses, flow splitters are a fact of life when using
MS detection. However, Fig. 2B illustrates that separation of the chromato-
graphic test sample can be further scaled to a 2.1-mm i.d. column, now running
at an equivalent linear velocity (1.0 mL/min). Under the conditions used in

© 2000 by Marcel Dekker, Inc.

Figure 2 (A) Generic method optimized for high-throughput. Separation was per-
formed on a 4.6 by 50mm 3.5 micron particle size Symmetry® C18 column (Waters
Corporation, Milford, MA) at 30°C. The mobile phase consisted of 0.1% formic acid
as the A solvent, and acetonitrile as the B solvent, run as a linear gradient from 0–
100% B over 3 minutes at 4.0 mL/min. All other conditions (injection, detection, and
sample) were identical to those listed in Figure 1. (B) Chromatography optimized by
APCI/MS detection. Separation was performed on a 2.1 by 50mm 3.5 micron particle
size Symmetry® C18 column (Waters Corporation, Milford, MA) at 1.0 mL/min. and
at 30°C. Mobile phase and gradient conditions were identical to Figure 2A. All other
conditions (injection, detection, and sample) were identical to those listed in Figure 1.

Fig. 2B, atmospheric pressure ionization MS can be used directly without

splitting the flow. However, depending on the compound class or type in the
library being analyzed, electrospray ionization might be preferred and in some
cases might still necessitate a split flow, depending on the flow rate and mobile
phase composition.
Figures 3 and 4 illustrate the diversity of the method. In Fig. 3, six
penicillin-type antibiotics representing a somewhat more realistic sample are
separated. Peak 1 is the synthetic precursor for the rest of the compounds in the

© 2000 by Marcel Dekker, Inc.

Figure 3 Separation of penicillin type antibiotic homologous series. Separation con-
ditions are identical to those reported in Figure 2B. Peaks 1–6 are: 6-aminipenicillanic
acid, amoxicillin, ampicillin, oxicillin, cloxicillin, and dicloxicillin, respectively.

mixture, and peaks 4–6 differ only by a chlorine atom, representing possible
synthesis byproducts. The generic method readily resolves all of the components
in this series of homologous compounds, in spite of the wide polarity range.
Figure 4 shows the separation of another test mix of ‘‘drug-like’’ compounds
reported in the literature for use as a chromatographic test mixture (8).
Although the improvements illustrated in Figs. 2–4 are significant, it is
possible to improve sample throughput even more dramatically. When de-
termining sample throughput capabilities for short run times, additional system
timing issues must be taken into account. That is, to determine actual sample
throughput, cycle-to-cycle inject times must be determined that take into ac-
count all other aspects of each individual sample analysis. Run time, while
important, is only one factor in determining overall cycle-to-cycle inject times.
The amount of time it takes to position the injector, aspirate a sample, load
the sample loop, inject the sample, and rinse the injector apparatus is referred
to as ‘‘injection overhead.’’ While injection overhead varies from instrument
to instrument, times from 1 to 2 min per injection cycle are not uncommon. In

© 2000 by Marcel Dekker, Inc.

Figure 4 Separation of chromatographic test mix showing generic method diversity.
Separation conditions are identical to those reported in Figure 2B. Peaks 1–5 are uracil,
1-hydroxy-7-azabenzotriazole, methoxybenzenesulfonamide, methyl-3-amino-2-thio-
phenecarboxylate, and 4-aminobenzophenone, respectively.

a 40-min run, with 15–20 min of post run time, 1–2 min is hardly significant.
However, with run times as short as 5 min, the injection overhead can add
20% or more to the cycle-to-cycle inject time. Another parameter that must
be considered in the cycle-to-cycle inject time is the time it takes to reequili-
brate the column and the system following the gradient. Each of these func-
tions is illustrated in Fig. 5 for an LC system operated in the traditional ‘‘se-
quential’’ mode. Obviously, anything that can be done to reduce the time it
takes to accomplish these tasks will also improve sample throughput. As
shown in Fig. 6, a separation of an 11-component mix in the sequential mode
with a 5-min run time can consume an additional 1.2 min due to injection
overhead, for a total cycle to cycle inject time of 6.2 min. If, however, some
of these functions can be performed in a ‘‘parallel’’ mode, as illustrated in
Fig. 7 (p. 122), additional time savings could be realized. In the parallel mode
of operation, sample-manager needle and inject port washes, and aspiration
of the next sample takes place during the actual run, saving considerable time.

© 2000 by Marcel Dekker, Inc.

Figure 5 Schematic of traditional sequential mode injection to injection cycle time
(injection overhead) operation of an LC system.

Figure 6 Traditional sequential mode injection showing cycle to cycle inject time.
Separation conditions and peak identification are identical to those reported in Figure
2B. Data system run time was extended to 20 minutes to capture LC instrument timing.

© 2000 by Marcel Dekker, Inc.

 Equilibrating the system separately from the column can also save addi-
tional time. Equilibration time is an essential part of gradient chromatography.
Both the LC system and the column must be returned to initial mobile phase
conditions prior to the next run to ensure repeatability. Traditionally, equilibra-
tion times of 5–10 column volumes have been used (9). However, with the use
of smaller diameter and shorter LC columns, a 10-column volume is not always
enough to equilibrate both the LC system and the column. To compensate for
this situation, the reequilibration volume has been divided into two parts: the
volume required for system equilibration and the volume required for column
equilibration (10). Total equilibration time is then given by the formula:
Tr ⫽ (3VT ⫹ 5VC )/F
where Tr is the equilibration time (min), VT is the total system volume, VC is
the column volume (mL), and F is the flow rate (mL/min). As the column
gets smaller (with a correspondingly lower flow rate), more equilibration time
is taken up by returning the system volume of the LC to initial conditions. In
advanced LC systems with integrated fluidics and control, a purge step can
be added post run at high flow rates (5–7 mL/min) with the column off-line
to significantly reduce system equilibration times. By starting the gradient but
holding off on the injection for the amount of time proportional to the system
volume, the volume of the system can also be used to aid in column equilibra-
tion—a technique referred to as a ‘‘just-in-time gradient.’’ Loading the sample
loop, as illustrated in Fig. 8 (p. 123), during the equilibration saves additional
time by further reducing injection overhead. Comparing Figs. 6 and 8, the
parallel mode with rapid equilibration increases throughput by about 30%
without sacrificing chromatographic information or integrity.
Rapid equilibration techniques are particularly helpful to save time as
the flow rate decreases, in, for example, microbore applications. Figure 9
(p. 124) shows a separation on a 1 ⫻ 50 mm column, at 0.3 mL/min, with
a cycle-to-cycle inject time of 5.5 min using the parallel mode with rapid
equilibration. Run in the traditional sequential manner, the corresponding cy-
cle-to-cycle inject time would be in excess of 7.5 min. At lower flows, the
time saving during reequilibration would be even more significant.
Some general notes on operating LCs in a high-throughput mode are as
follows:
With the trend toward smaller columns, extra care should be taken to
minimize extra column band broadening by using reduced diame-
ter tubing, smaller volume detector cells, and properly fitting con-
nections (9).

© 2000 by Marcel Dekker, Inc.

Figure 7 Schematic of parallel mode operation with rapid equilibration.

Photolabile or temperature-sensitive compounds may require special

handling. LC systems are available that maintain samples in the
dark and/or chilled if required.
The small peak volumes and resolution demands from high-throughput
analyses require faster data collection rates for accurate results.
Adequate detector time constants should also be used.

III. ADVANCED HIGH-THROUGHPUT ANALYSIS

TECHNIQUES
A. Sample Pooling
Sample pooling can also be used to reduce analysis times. Pooling is com-
monly used in drug discovery and early development for pharmacokinetic
screening in animals, intestinal permeability screening using Caco-2 cell–
based assays, and combinatorial library analyses. Sample pooling is an injec-
tion technique that uses multiple sample aspirations and then coinjects the
samples, decreasing analysis times by reducing the number of chromato-

© 2000 by Marcel Dekker, Inc.

Figure 8 Parallel mode injection showing cycle to cycle inject time. Separation con-
ditions and peak identification are identical to those reported in Figure 2B. Data system
run time was extended to 20 minutes to capture LC instrument timing. Note differences
in cycle to cycle timing relative to Figure 6.

graphic runs. It is performed as diagrammed in Fig. 10 (p. 126) either, for

example, within a microtiter plate (intraplate) or between plate (interplate)
pooling, with mixing or air or solvent segmentation. Figure 11 (p. 128) illus-
trates the results of an interplate pooling experiment using two compounds
and two internal standards, run parallel with rapid equilibration mode condi-
tions used in Fig. 6. The technique is quantitative (good accuracy and preci-
sion), as summarized by the data shown in Table 1. In this simple experiment,
sample throughput is doubled without compromising inject-to-inject cycle
time or injection overhead.

B. Two-Column Regeneration
A technique referred to as two-column regeneration has also been used to
increase sample throughput. In this technique, while one LC column is running
the gradient method, a second isocratic pump is used to equlibrate a second

© 2000 by Marcel Dekker, Inc.

Figure 9 A µbore separation cycle time comparison. Separation was performed on
a Waters Alliance® HT HPLC System (Waters Corporation, Milford, MA) using a
1.0 by 50mm 3.5 micron particle size Symmetry® C18 column (Waters Corporation,
Milford, MA) at 60°C. The mobile phase consisted of 0.1% phosphoric acid as the A
solvent, and acetonitrile as the B solvent, run as a linear gradient from 10–95% B over
2 minutes at 300 µL/min. UV detection at 220nm, and a 2 µL injection was used.
Peaks 1–6 (0.1mg/mL each in 50/50 methanol/water) are uracil, caffeine, primidone,
phenacetin, benzophenone and biphenyl, respectively.

Table 1 Quantitative Results of Interplate Pooling Experiment from Figure 9

(n ⫽ 6)
A/IS B/IS
A/IS B/IS Pooled Pooled
Avg. ratio 1.129 2.632 1.123 2.618
%RSD 0.55 0.52 0.21 0.11
Avg. cycle 4.68 4.69 4.69
time (min.)
%RSD 0.30 0.08 0.10

© 2000 by Marcel Dekker, Inc.

column to initial gradient conditions. After the analysis is complete, the col-
umns are switched, and the first column is reequilibrated while the second
column is run. Postrun equilibration time can be decreased using this technique
on a traditional LC system; however, when modern LC systems are used that
operate in the parallel rapid equilibration mode outlined above, any throughput
advantages from the two-column regeneration technique may be minimal on
the analytical scale. Throughput can, however, be increased by as much as
25% on the preparative scale (11).

C. Automated Multichannel Chromatography

Another way to save time is to use multiple separation and analysis channels
on the same instrument. Recently, new technology was reported that allows
multiple samples to be analyzed in parallel as illustrated in Fig. 12 (p. 129)
(12). This technique allows for four LC columns running identical gradients
in parallel on a single LC system to be multiplexed to one MS detector. In
the LC/MS interface the traditional electrospray probe and outer source assem-
bly are replaced by an array of four miniaturized, pneumatically assisted
electrosprays. The interface has a sampling rotor that is monitored in real time
enabling the four liquid inlets to be indexed. Four separate chromatograms
are collected into four separate data files, allowing conventional data analysis,
as presented in Fig. 13 (p. 130). Depending on LC conditions, analysis time
for an entire 96-well microtiter plate can be reduced to as little as 1.1 h, re-
sulting in sample throughput approaching 2000 samples per day.

D. Auxiliary Detection
As noted previously, fast, generic LC/MS methods are the techniques of
choice for assessing the progress and final quality of large combinatorial arrays
in drug discovery. However, ultraviolet (UV) or MS detection alone cannot
always provide the type of quantitative data required when assessing com-
pound purity. To obtain accurate quantitative data, UV detection requires the
use of well-characterized reference standards due to the differences in molar
absorptivity that may exist between members of a combinatorial library. Since
reference standards may not be available for even a fraction of the members
of any combinatorial library, detectors that respond to physical characteristics
independent of compound-to-compound variations must be used, i.e., the
‘‘universal’’ detector. While no truly universal detectors exist, detectors do
exist that provide similar responses to compounds in a particular class. These
detectors include those used in evaporative light scattering detection (ELSD),

© 2000 by Marcel Dekker, Inc.

Figure 10 Injection pooling schematic.

chemiluminescent nitrogen detection (CLND) and its cousin, chemilumines-

cent sulfur detection (CLSD). Used alone or in combination with UV and
MS, use of these auxiliary techniques allows a more accurate and complete
assessment of purity to be obtained.
The principles of ELSD date back a number of years (13–15). More
recently, ELSD has been applied specifically to pharmaceutical analyses and
combinatorial uses (2,16–18). Detection by light scattering is based on the
available mass and not absorptivity (UV) or ionization efficiency (MS), mak-
ing it more accurate in some applications. In ELSD, nebulized column effluent
enters a heated drift tube where rapid evaporation of the LC mobile phase
takes place. A stream of nitrogen gas sweeps any nonvolatile solutes toward
a detection region. Detection is accomplished by a laser and photodiode at an
angle of 90°, perpendicular to the central axis of the drift tube. As the solute
particles pass through the laser beam, the source is scattered. The intensity of
the scattered light measured by the photodiode is proportional to the amount
of solute in the column effluent.

© 2000 by Marcel Dekker, Inc.

 Although most compounds respond well to ELSD, the volatility of some
low molecular weight compounds may cause problems during the evaporation
process. Varying response in different stages of the gradient can also cause
problems. Use of ELSD in combination with UV detection minimizes these
potential limitations.
CLND and, more recently, CLSD play a role similar to that of ELSD
in quantitative assays of true unknowns. The CLND and CLSD detectors re-
spond to the nitrogen and sulfur content of a compound, respectively. Both
detectors operate under relatively the same principles and have been used with
considerable success in drug discovery (19–21). In CLND, the analyte is oxi-
dized at 1050°C, converting nitrogen-containing compounds to nitric oxide.
The nitric oxide reacts with ozone to produce nitrogen dioxide in the excited
state, which releases a photon when decaying to the ground state. The photons
are measured by a photomultiplier tube and converted to an analog signal
dependent only on the total mass injected.
Since a vast majority of drug compounds contain nitrogen, CLND is
very useful for pharmaceutical analyses. The CLND response has been shown
to be independent of gradient composition and, again, is often used in combi-
nation with other (UV and MS) detection systems (18). However, the presence
of nitrogen-containing impurities in the sample or solvent will bias results.
Mobile phases must of course be nitrogen-free, dictating the use of methanol
or another alcohol rather than acetonitrile as mobile phase modifier.

IV. LC/MS INSTRUMENTS IN ROUTINE HIGH-THROUGHPUT

USE
A. Open Access Instrument Operation
Traditionally, MS analyses have been performed in a centralized facility, often
on highly specialized instruments that required constant operator intervention
and maintenance. This situation is highly impractical when supporting a com-
binatorial program because it inhibits high throughput and general access to
instrumentation and data. In response, instruments are now often operated in
an ‘‘open access’’ environment. In such an environment, people not trained
in LC or MS can submit samples on a continuous basis and get rapid turn-
around. The use of MS and its wealth of information is promoted, and spec-
trometrists are freed from the mundane, tedious task of repetitive sample anal-
ysis of perhaps thousands of samples.
An open access setup usually consists of a workstation at the point of
need. The walkup user has access to the workstation as well as to the sample

© 2000 by Marcel Dekker, Inc.

Figure 11 Interplate pooling with internal standards. Separation conditions are iden-
tical to those reported in Figure 2B. Peaks A and B are acetylfuran and valerophenone.
The internal standards for peaks A and B are acetanilide and benzophenone, respec-
tively. Data system run time was extended to 15 minutes to capture LC instrument
timing. Cycle to cycle injection time is 4.7 minutes including interplate pooling.

manager or autosampler used for injection. A system administrator is responsi-

ble for setting up the system and has access to and control of all components.
The walkup user, perhaps a medicinal or synthetic organic chemist, logs in
sample(s) (or an entire plate) at the workstation, places them into the directed
location, and, depending on the queue and system setup, gets a postanalysis
report at the point of use or e-mailed to his desk. Results are typically displayed
in an integrated browser format, similar to that illustrated in Fig. 13. The
browser provides a graphical display for review of the results, a confirmation
of molecular weight, and displays corresponding chromatograms and spectra.

B. Mass-Directed Autopurification
As mentioned previously, during lead optimization and testing leading to can-
didate drug selection, compounds are often required in larger quantities. Ana-
lytical techniques used in lead discovery eventually give way to semiprepara-
tive or preparative mass-directed autopurification techniques, capable of
isolating and purifying 10–20 mg or more of the compound of interest during
a single chromatographic analysis. Short, wide-diameter columns operated at

© 2000 by Marcel Dekker, Inc.

10–50 mL/min are the norm. Isolation and purification on the preparative
scale involves fraction collection and reanalysis. Fractions are reanalyzed on
the analytical scale to see how efficient the purification was. The analytical
scale separation is also run on the system to check to develop the preparative
method prior to actual use. Generic methods exactly like those outlined previ-
ously and scalable column chemistries are used.
Automated purification has evolved over the years from ‘‘collect every-
thing’’ on a time basis to UV-based collection. In both instances, time-con-
suming secondary analyses must be carried out to correctly identify the correct
or desired fraction. More recently, techniques have evolved using mass-di-
rected and intelligent automated purification. Mass-directed fraction collection
is defined as collecting a fraction of a certain specified mass only. Intelligent
automated fraction collection takes fraction collection one step further by
allowing fraction collection based on masses, substructures/fragments, multi-
ple masses, adducts, etc. This results in higher quality data, single fractions,
less time, fewer steps, and less chance for errors. A typical high-throughput
LC/MS mass-directed autopurification system is highlighted in Fig. 14.
The system as diagrammed in Fig. 14 used two-column regeneration to
improve throughput. The system has both preparative and analytical capability
so that samples can be run on the same system for rapid screening of original
samples, or an initial purity assessment, as well as fraction collection capabil-
ity. In addition to the two 6-port column-switching valves, two flow splitters

Figure 12 Schematic of an automated multichannel LC/MS system.

© 2000 by Marcel Dekker, Inc.

Figure 13 MassLynx™ (Micromass UK Limited, Manchester, UK) OpenLynx™
open access browser data report. Plate configuration is shown in upper left. MS and
UV data is displayed for the selected well position. If the requested mass is found, the
well position is highlighted in green. If it is not, it is highlighted in red. Browser’s
of this type are also used to track fractions in mass directed auto-purification
systems.

are also used in this configuration. The upstream splitter divides the prepara-
tive flow and is an integral part of the system. The second splitter splits the
analytical flow for parallel MS and PDA detection.
The use of an upstream splitter as outlined in Fig. 14 has several distinct
advantages. Besides being easy to use and reproducible, this type of splitter
provides constant delay times across a wide flow range, without the need for
plumbing different tubing lengths or diameters. The use of a makeup flow
allows high column loading without saturating the detector (by dilution) and
allows flow to be split to multiple detectors without back streaming (explained
below). The splitter essentially has two sides: a high-pressure side (column
to collector) and a low-pressure side (makeup to detectors). As long as this
balance is maintained the system works correctly. In operation, however, as
the gradient composition changes from aqueous to organic, the back pressure

© 2000 by Marcel Dekker, Inc.

Figure 14 High-throughput mass-directed auto purification system schematic. Sys-
tem as shown is configured for preparative two-column regeneration and an analytical
column for method development or to reanalyze fractions.

in the system decreases. Eventually the gradient back pressure can fall below
the constant pressure of the makeup pump, resulting in splitter back streaming
which prevents sample from getting to the MS probe and prevents late eluting
peaks from being seen or collected. By using a lower viscosity organic solvent
(e.g., 100% MeOH) as the makeup solvent, the back pressure is reduced to
below that of the lowest point in the gradient, and back streaming is prevented.
Subtle variations in split ratio delay times and band broadening are nominal
across all gradient compositions.
Mass-directed autopurification systems are designed so that a peak is
detected at the MS prior to reaching the fraction collector. Different flow rate–
dependent delay times must be taken into account, and the software controls
the synchronization between detection and collector trigger times. Figures 15
and 16 highlight the results of a mass-directed autopurification experiment
(22). In this experiment, a three-component synthetic mixture was fractionated
on the preparative scale and the fractions reanalyzed on the same system.
Figure 15 shows the total ion chromatogram, UV signal, and fraction collec-
tion signal for the 20 mg/mL preparative run. As can be seen from the fraction
collection signal, as the chromatography changes, the intelligent fraction col-
lection compensates. The MS response directed the fraction collection based
on the molecular weight of the three components in each instance. Figure 16
depicts the reanalysis of the second fraction on the analytical scale, on the
same system. In spite of the two closely eluting peaks before and after, the
data show a clean spectrum free of contamination.

© 2000 by Marcel Dekker, Inc.

Figure 15 Mass directed fraction collection. Separation was performed on Waters
Auto-purification System with FractionLynx™ software (Waters Corporation, Milford,
MA) on a 19 by 50mm 7 micron particle size Symmetry® C18 column at ambient
temperature. The mobile phase consisted of 0.1% phosphoric acid as the A solvent,
and acetonitrile as the B solvent, run as a linear gradient from 0–100% B over 8 minutes
at 20.0 mL/min. UV detection at 254nm, and a 10 mL injection was used. Peaks (20
mg/mL each in 50/50 methanol/water) in order of elution are terfenadine, diphenhy-
dramine, and oxybutynin chloride. The two lower traces show the MS total ion chro-
matogram (TIC-lower left in figure-generated with an electrospray interface), and the
UV trace (lower right, an extracted PDA channel) for five separate overlaid runs. A
separate mass directed fraction collection signal for each of the five runs is shown in
the upper right. Note the timing differences—as the preparative chromatography
changes, the intelligent fraction collection compensates.

V. THE FUTURE OF LIQUID CHROMATOGRAPHY IN

COMBINATORIAL CHEMISTRY

As stated earlier, LC has emerged as the technique of choice for assessing the
quality of combinatorial libraries. While this trend will certainly continue, LC,
either alone, or in combination with other techniques, will continue to evolve.
The trend towards shorter, smaller internal diameter columns will continue,
however hardware limitations such as pressure limitations, system volume,
injection overhead times and data storage must be overcome. A balance be-

© 2000 by Marcel Dekker, Inc.

Figure 16 Analytical scale reanalysis of fraction two from Figure 14. The presence
of a single compound (oxybutynin, m/z 357.3) on the TIC (top) is confirmed by the
UV data (bottom), and the MS on the right lacks evidence of contamination from peaks
one and three in the preparative run. Separation conditions that closely match those
reported in Figure 2B were used.

tween cycle to cycle analysis times and peak capacity, whether obtained by
an actual separation or by instrument specificity or deconvolution must be
maintained. Recent reports of the use of capillary LC may also play a signifi-
cant role in new developments in the field of combinatorial chemistry (23).
Capillary LC is currently used mainly for sensitivity enhancements. Chro-
matographic theory predicts that reducing column diameter results in a mass
sensitivity increase that is inversely proportional to the square of the column
diameter ratios. However, as mentioned previously, but especially on the capil-
lary scale, extra care should be taken to minimize extra column band broaden-
ing by using reduced diameter tubing, smaller volume detector cells, and
proper fitting connections (9). In addition, capillary LC has a distinct advan-
tage over analytical scale LC when interfaced to MS. Using analytical systems,
sensitivity is sacrificed either by introducing the entire sample into the MS
interface and losing much of the sample by high capacity pumping or by flow
splitting just before the MS inlet. Low microliter per minute capillary flows
are appropriate for direct interfacing, so virtually the entire sample is intro-
duced into the MS ion source. Additional high throughput capillary LC tech-
niques and methods currently in development may eventually extend the use

© 2000 by Marcel Dekker, Inc.

of this technique to combinatorial applications (24). Miniaturization of differ-
ent sorts is also under way that may eventually find LC applications in combi-
natorial chemistry. Microelectromechanical systems, or MEMS, or other
‘‘chip-based’’ technology may eventually be optimized for use in screening
applications in the drug discovery process.
Detection schemes will also continue to evolve. While NMR is covered
in another chapter in this volume, advances in NMR used as an LC detec-
tor have also been reported (25–28) that may eventually increase the utility
of this tool for combinatorial chemistry. Used in both stopped flow and on
line modes, LC/NMR can be extremely useful for structure elucidation
provided the proper mobile phases or solvent suppression techniques are
used (27).
Affinity chromatographic techniques are also finding applications. For
example, peptide ligands can be immobilized on chromatographic media that
will selectively bind and release target molecules under optimized conditions
(29, 30). Stable, selective, high affinity ligands used in this manner can purify
a target molecule from a complex mixture, even in the presence of closely
related impurities. It is possible to increase or modulate the strength of binding
and incorporate chromatographic selectivity against specific contaminants,
which can accelerate process development and increase product recovery in
drug discovery.

VI. CONCLUSION

LC characterization of combinatorial libraries is a vital tool in drug discovery,

and, in one format or another, will remain so for a long time. High throughput
methods, open access environments, mass directed auto-purification, and total
system integration have lead to increased utilization of LC in ways that a few
short years ago were unimaginable. This trend can only continue, as research-
ers strive for higher throughput, and wider applicability in solving challenges
in drug discovery.

ACKNOWLEDGMENTS

The author would like to acknowledge the contributions of Chris Chumsae,

Andrew Brailsford, Jeffrey Holyoke, and Beverly Kenney of Waters Corpora-
tion who collaborated in various ways in the data generated for use in this
chapter.

© 2000 by Marcel Dekker, Inc.

REFERENCES

1. IM Mutton. Chromatogr 708(47):1–8, 1998.

2. JN Kyranos and JC Hogan, Jr. Anal Chem 70(11):389A–395A, 1998.
3. WK Goetzinger and JN Kyranos. Am Lab April:27–37, 1998.
4. R Cole, KA Laws, DL Hiller, JP Kiplinger, and RS Ware. Am Lab July:15–20,
1998.
5. ME Swartz, M Balogh, and B Kenney. Poster presented at the Eastern Analytical
Symposium, Somerset NJ, November 1998, and Waters Corporation (Milford,
MA) Literature Code M29.
6. JP Kiplinger, RO Cole, S Robinson, EJ Roskamp, RS Ware, HJ O’Connell, A
Brailsford, and J Batt. Structure controlled automated purification of parallel syn-
thesis products in drug discovery. Rapid Commun Mass Spectrom 12:658–664,
1998.
7. L Zeng, L Burton, K Yung, B Shushan, and DB Kassel. Automated analytical/
preparative HPLC-MS system for the rapid characterization and purification of
compound libraries. J Chromatogr A 794:3–13, 1998.
8. HN Weller, MG Young, SJ Michalczyk, GH Reitnauer, RS Cooley, PC Rahn,
DJ Loyd, D Fiore, and SJ Fischman. Mol Div 3:1–24, 1997.
9. LR Snyder, JJ Kirkland and JL Glajch. Practical HPLC Method Development.
Second Edition. New York: John Wiley and Sons, Inc., 1997.
10. JL Li and J Morawski. LC/GC Magazine 16(5):468–476, 1998, and Waters Cor-
poration (Milford, MA) Literature Code Number T158.
11. C Chumsae and A Brailsford. Poster presented at ASMS, Houston, TX. June
1999.
12. Brochure No. BR25/DAM, Micromass Corporation, Manchester UK, April
1999.
13. A Stolyhwho, H Colin, and G Guichon. J Chromatogr 265:1–18, 1983.
14. A Stolyhwho, H Colin, M Martin, and G Guichon. J Chromatogr 288:253–275,
1984.
15. TH Mourey and LE Oppenheimer. Anal Chem 56:2427–2434, 1984.
16. M Lafosse, C Elfakar, L Morin-Allory, and M Dreux. J High Res Chromatogr
15:312–318, 1992.
17. CE Kibbey. Mol Div 1:247–258, 1995.
18. A Brailsford and C Chumsae. Poster presented at ASMS, Houston TX. June
1999.
19. EW Taylor, MG Qian, and GD Dollinger. Anal Chem 70:3339–3347, 1998.
20. R Bizanek, JD Manes, and EM Fujinari. Peptide Res 8:40–44, 1996.
21. WL Fitch, AK Szardenings, and EM Fujinari. Terahedron Lett 38:1689–1692,
1997.
22. ME Swartz and BF Kenney. Poster presented at the Eastern Analytical Sympo-
sium, Somerset, NJ, November 1998.
23. SA Cohen, BF Kenney, J Holyoke, TA Dourdeville, and D Della Rovere. LC/
GC, 17(4S):S9–S16, 1999.

© 2000 by Marcel Dekker, Inc.

24. ME Swartz. Unpublished results, 1999.
25. PA Keifer. DDT, 2(11):468–478, 1997.
26. FS Pullen, AG Swanson, MJ Newman, and DS Richards. Rapid Comm Mass
Spectr 9:1003–1006, 1995.
27. SH Smallcombe, SL Patt, and PA Keeifer. J Mag Res, Series A 117:295–303,
1995.
28. H Barjat, GA Morris, MJ Newman, and AG Swanson. J Mag Res, Series A 119:
115–119, 1996.

© 2000 by Marcel Dekker, Inc.

6
Capillary Electrophoresis in
Combinatorial Library Analysis

Ira S. Krull
Northeastern University
Boston, Massachusetts
Christina A. Gendreau
Waters Corporation
Milford, Massachusetts
Hong Jian Dai
Shuster Laboratories
Quincy, Massachusetts

I. INTRODUCTION AND OVERVIEW: WHY HPCE FOR

COMBINATORIAL MAPPING ANALYSIS?

Although capillary electrophoresis (CE),1 also known as high-performance CE

or HPCE, has been known and described in the literature for almost two de-
cades, its use for combinatorial mapping is much more recent (1–14). There
does not appear to be a previous review, other than that in Analytical Chemis-
try (an American Chemical Society journal), that describes the general applica-
bility and applications of CE for these purposes (15). There are relatively few
actual publications in the refereed literature that have utilized various CE
modes to perform analysis of combinatorial maps. At the same time, there are

1
All acronyms and abbreviations are defined in the glossary preceding the references.

© 2000 by Marcel Dekker, Inc.

many more CE papers that have used affinity recognition to identify an active
antigen or antibody or receptor molecule, but not as part of a larger library
of similarly structured compounds. Most papers in the literature deal with the
synthesis and characterization of libraries in terms of chemical structures, not
biological activity (15–17). Because sections of this book are devoted to ana-
lytical approaches to screen compounds in combinatorial libraries for activity,
it represents the first major effort to describe how analytical chemistry can be
applied to identify individual, active members of a chemical library. In this
chapter, the focus is on describing how CE approaches (methods, instrumenta-
tion, protocols, techniques) can be applied to a combinatorial library and iso-
late, as well as characterize, only those active (lead) compounds in a given
library. This is quite different from characterizing all of the members of a
given library in terms of their individual structures or saying that a given
structure is present in a particular library.
In general, CE is perhaps ideally situated for combinatorial map search-
ing, as well as for providing structural information about lead/target com-
pounds in that library. Any useful analytical approach, be it mass spectrometry
(MS), nuclear magnetic resonance (NMR), Fourier transform infrared (IR),
high-performance liquid chromatography (HPLC), or HPCE, must accomplish
several ideal goals: (a) recognize those compounds that shall prove of biologi-
cal interest and opportunity, using some type of molecular/biological recogni-
tion (antibody–antigen, receptor binding protein–antigen, drug–protein, and
so forth); (b) separate the recognized members of the library from those show-
ing little or no biological interactions with biopolymers or target drugs; (c)
separate active, lead compounds from one another on some reasonable time
scale; (d) indicate general level of molecular recognition (high vs. low); and
(e) provide some type of structural information about the active compounds,
using techniques such as NMR, MS, and FTIR. Quantitation of lead or active
compounds is not really necessary for any library, since presumably one can
make more of those interesting compounds at a later date, once they are shown
to be recognizable and structurally determined. The successful analytical tech-
nique is more qualitative than quantitative, but ideally it will provide an indica-
tion of biological activity and structural information. One without the other
is less than ideal for a useful, practical, and valuable analytical method of
mapping a combinatorial library.
We use the word mapping here not so much to identity all of the struc-
tures present but rather to identity the structures of just those active (target,
potential leads) compounds in that library. It has become difficult to determine
the structures of individual members of a large, complex library; it is much
easier to pinpoint which members of that library are indeed biologically active

© 2000 by Marcel Dekker, Inc.

and then determine their individual structures. Although analytical techniques
have been used mostly to identify potential pharmaceutical agents from a com-
binatorial library, it should be clear that these methods could be applied for
virtually any purpose, so long as one has a recognition element or compound
available. That is, libraries can be applied for the identification of agricultural
chemicals, flavors, perfumes, fragrances, insecticides, insect pheromones, and
so forth. What is crucial for the success of these methods really depends on the
recognition element, and the process by which a compound or compounds will
indicate biological or chemical activity for the stated purposes.
CE is potentially an ideal approach for mapping of combinatorial librar-
ies because it is a purely liquid phase technique, as opposed to HPLC, which
requires some sort of solid–liquid interaction, partitioning, diffusion, etc.
There are arguments against using solid phase techniques, such as immunoaf-
finity chromatography (ICA) or high-performance affinity chromatography
(HPAC) to isolate active, lead compounds, rather than a purely liquid solution
approach. As Dunayevskiy has aptly described and discussed, solid phase
binding assays, such as using immobilized antibodies to identify active anti-
gens in HPAC or immunodetection (ID), can be influenced by the site or nature
of the attachment of the ligand to the solid support (18).
This is a classic problem in all immunorecognition or immunodetection
or affinity isolation techniques in HPLC or, in the future, in capillary electro-
chromatography (CEC). The support (bead) used to hold the ligand or receptor
molecules in any affinity recognition framework can present conformational
limitations that can at times prevent any or all binding to the ligand or receptor.
Unfortunately, there can be no firm rule for the optimal offering of a compound
to a receptor or vice versa. There are a large number of ways to immobilize
a receptor protein or its ligand, but none of these will guarantee that the bound
species can now be adequately recognized and captured by its partner (ligand
receptor) (19–35). The area of immobilization of compounds to a solid support
is immense, with a large number of linkage agents commercially available. It
is never fully clear which of these will provide a bound species offering itself
in the optimal way to its partner, now in a solution passing through that bed.
It is always possible to use more than one method of immobilization of the
same receptor protein in the hope that this will improve the chances of captur-
ing any and all possible partners/ligands (19,21). However, these methods
only approximate complete freedom from steric effects, and one is left with
limits on 100%, true recognition interactions that occur in a purely solution
phase recognition approach. For these very reasons, more and more applica-
tions of combinatorial library searching are being described using purely solu-
tion phase approaches.

© 2000 by Marcel Dekker, Inc.

 Vouros and Dunayevskiy and others have utilized free solution or affinity
CE methods combined (interfaced) with MS detection for searching libraries,
as described below (18). However, CE or affinity CE is nothing but a separation
technique, and it cannot and never will identify structures, especially when one
is unsure of the contents of a given library. CE is a solution-based approach,
though capillary gel electrophoresis (CGE) does use a polymeric gel in the capil-
lary, which separates on the basis of differences in mass-to-charge or charge-
to-mass ratios. This almost sounds like the basis of separations in MS. The
reader must be referred to the available texts or review articles that describe
how CE works, instrumental requirements, and the foundations or fundamen-
tals of its separating abilities (36).
There are numerous variations on free solution CE (FSCE), such as mi-
cellar electrokinetic capillary chromatography (MECC or MEKC), where a
moving, pseudostationary phase is added to the CE buffer, and secondary
chemical equilibria or interactions ensue that effect separations of even neutral
compounds, as well as ionic analytes. However, in general, CE utilizes truly
homogeneous, solution phase separation approaches, without a stationary (per-
manent, fixed) phase, making it perhaps ideally suited for molecular recogni-
tion in searching combinatorial libraries.
In affinity CE (ACE), one adds to the running buffer or to the sample,
the recognition element or ligand, and then looks for changes in mobility pat-
terns of individual members of the sample (mobility shifts, as in flat-bed elec-
trophoresis). This has been described in many publications, without searching
combinatorial libraries, as a method to identify antibody–antigen partners,
active antibodies, and active antigens. Thus, ACE is really a variation on CE,
by virtue of adding something to the running buffer or, more often, to the
sample that will then provide different mobility tendencies and migration pat-
terns (mobility shifts). The two different electropherograms generated, with
and without added ligand or receptor, provide differences that can be utilized
to identify active partners, analogous to difference chromatography, used for
many years in HPLC (19). A variation of affinity CE, if you will, permits the
detection of low levels of enzymes by placing the enzyme substrate into the
running CE buffer, something termed enzyme-modulated microanalysis
(EMMA).
At the same time, there are some limitations in utilizing CE for library
search routines. CE utilizes very small amounts of sample, nanoliter volumes,
and thus very low levels of possible recognition elements. Because detection
is usually through the capillary, except for MS, detection limits are very high
(poor), making this a generally insensitive technique for trace analysis. There
are some recent improvements in detector cell design, bubble cells, Z cells,

© 2000 by Marcel Dekker, Inc.

rectangular capillaries, and so forth that have lowered detection limits for
many analytes in CE (13,14). Also, since ultraviolet detection (UV)/fluores-
cence detection (FL)/electrochemical detection (EC) methods provide little,
if any, actual structural information about a specific CE peak, applications
of CE for combinatorial library searching really require interfacing with MS
detection (37). MS will often provide much lower detection limits for library
components, and (most importantly) it can often provide structural information
and even absolute identification (38).
CE also has the ability of providing improved resolutions for similar
structures, such as peptides, which may be difficult or impossible to separate
by any and all HPLC methods (39). Because it moves analytes and the buffer
by electroosmotic flow (EOF) rather than by pressurized flow delivery as in
HPLC, individual peaks tend to be much narrower, sharper, symmetrical, and
of higher plate counts (efficiency, N (number of theoretical plates or plate
count for peak)). This also leads to improved resolutions of nearly identical
compounds, assuming that they have some differences in mass-to-charge ra-
tios or can interact with buffer components differently, as with their ligands
or receptors.
It would therefore appear at the outset that CE may provide the best
approaches for identifying lead, active compounds from large libraries of pos-
sible targets. There are, of course, some limitations in using current, solution
phase methodologies. Primary among these possible or real limitations is the
idea that activity of a given pool is dependent on the cumulative activity of
all compounds present. It is not always guaranteed that one will locate the
true, optimally active compound. One may have missed the best candidate of
all, depending on how that library was synthesized and its final contents. It
is also quite possible that in using a library of many compounds, recognition
may be a result of synergistic or summary effects, rather than the true effect
of just a single, individual compound. This is very common, and often the
larger the library, the more likely synergism may play a role. It is also possible,
at times, that finding synergistic compounds, more potent when administered
together, is a desirable outcome. Though perhaps more difficult to sell to the
Food and Drug Administration (FDA) for final, investigational new drug
(IND) approval, it has been done many times.
This fault runs counter to the one just mentioned, which suggests that
to find the true optimum compound one should be using larger and larger
libraries, which only makes the possibility of synergism more and more likely.
What a conundrum. These conflicting demands are not necessarily easily re-
solved, other than to utilize individual compounds for testing. However, that
only defeats the goal of being able to screen larger and larger combination

© 2000 by Marcel Dekker, Inc.

libraries in shorter and shorter times with less and less money and manpower.
Thus, while using purely solution phase screening strategies possible in CE-
MS seems ideal, clearly certain loopholes and pitfalls remain. Nothing in life
or science is perfect, and we are surrounded by gray areas, where we would
prefer black or white. There may be no simple way to resolve this lingering
problem.

II. AFFINITY CAPILLARY ELECTROPHORESIS

This section is devoted to a description and summary of those literature reports

wherein some form of affinity recognition has been described. None of these
reports actually utilized a combinatorial library for such recognition; those
applications follow. However, in order to appreciate how and why ACE plays
such a large role in using CE-MS for combinatorial library searching, one
must begin with the more general and simpler use of affinity recognition in
CE. We describe here the use of some type of biorecognition element, be that
antigen, antibody, receptor proteins, binding drug, or ligand, that interacts with
a partner to provide altered electrophoretic migration tendencies and thus mo-
bility differences in CE. As a result of temporary, noncovalent association or
binding of the ligand to its receptor molecule, it experiences a real difference
in its normal, free migration tendency and appearance in the final electrophero-
gram. One can then imagine generating two difference electropherograms: that
for the ligand alone in the absence of its receptor, and that in the presence of
the receptor, where some degree of affinity or binding will occur. Depending
on the migration tendency of the receptor molecule relative to that of the
ligand, which is usually different, one should observe a mobility shift of the
ligand. The degree of this mobility shift is directly related to the affinity con-
stant of the ligand for its receptor; the dissociation constant of the complex;
the nature of the buffer medium and its effect on complex mobility; differences
in mass-to-charge ratios of the ligand, receptor, and ligand–receptor complex;
and perhaps other CE conditions. In order to make use of affinity recognition
in CE or ACE, one must have real, discernible differences in the mobilities
of the free ligand vs. its ligand–receptor situation.
We know of no specific review articles that discuss only affinity CE,
though there is a very recent review that describes the use of CE for antibody
analysis, often with applications that utilize antibody–antigen recognition
called immunoaffinity or affinity CE, as above (40). At times, recent texts also
contain sections that discuss affinity recognition in CE, such as that by Righetti
(10). Because most applications of CE for combinatorial library searching

© 2000 by Marcel Dekker, Inc.

have involved some type of biological recognition, as opposed to molecular
imprinting (recognition) methods, becoming more familiar with ACE and its
variations can only help us to understand and appreciate why ACE-MS tech-
niques may be quite successful for future library searching.
We will therefore start by describing some recent applications of affinity
recognition in CE and what these teach us about utilizing their principles and
practical approaches for eventual application to combinatorial libraries. There
are far more papers that utilize some form of ACE, but far less that then apply
those methods for combinatorial library searching. Although there are some
reports on the use of CE for general, group separations of compounds in larger
combinatorial libraries, it does not appear that CE alone will provide signifi-
cant advantages in discerning the contents and structures of individual mem-
bers of a larger library. Thus, we have entirely emphasized here only those
reports that utilized CE for some type of affinity or immunoaffinity recognition
of a specific antigenic species, whether that was a small or large molecule.
While most such reports have utilized small antigens in ACE, it is entirely
possible that larger protein libraries will also be successfully searched using
ACE methods.
In recent years, ACE has been used to study ligand–receptor interactions
and determine the binding constants of formed complexes (41–43). Some typi-
cal applications include enzyme–inhibitor, DNA–peptide, protein–sugar, pep-
tide–drug, and antibody–antigen (Ab–Ag), etc. We will focus on Ab–Ag in-
teractions.
The adsorption of proteins to the inner wall of the capillary is a major
consideration in ACE assays (44,45). The interaction can lead to band broad-
ening, irreproducible migration times, low resolution, and low recovery of the
protein. Various approaches to control the problem include chemical modifi-
cation to the inner surface of the capillary, choosing a proper buffer species
and pH, and the use of buffer additives to reduce the protein interaction with
the capillary wall. The buffer additive has to be cautiously chosen so as not
to participate in the interaction between Ag and Ab.
Ab–Ag interactions and complexation kinetics vary for different sys-
tems (42). Some possible CE patterns are shown in Fig. 1 for a receptor–
ligand system. For a fast-interaction rate system, each protein molecule spends
the same time forming a complex(es) with the Ab, and although the migration
time changes, the peak shape does not. Of course, it is also possible that several
Ab–Ag complexes may form in a system-dependent way which can only com-
plicate the final CE analysis. For a slowly interacting system, some protein
molecules spend more time forming a complex with the Ab than the others,
which leads to broad peaks, or a disappearance of the peak altogether, or even

© 2000 by Marcel Dekker, Inc.

Figure 1 Schematic diagram of some of the possible interaction patterns in affinity
CE with hypothetical, homogeneous receptor–ligand systems characterized by differ-
ent reaction kinetics. Ref. is a noninteracting component, whereas the other peak repre-
sents a molecule interacting quickly (second panel from top) or more slowly (lower
panels) with ligands of lesser electrophoretic mobility, which are present in the electro-
phoresis buffer (42). (Reproduced with permission from the copyright holder, Elsevier
Science Publishers and Journal of Chromatography.)

separate peaks for complexed and uncomplexed proteins. As the assay proto-
cols are different for slow and fast kinetic systems, some initial experimenta-
tion needs to be performed to determine the exact kinetics prevalent in a given
situation. Experiments similar to those in Fig. 1 can be performed, and the
kinetics can be determined based on the observed peak shape(s) (42).

© 2000 by Marcel Dekker, Inc.

 In a system with fast kinetics, buffers containing different concentrations
of Ag or Ab are prepared, and the sample having a fixed concentration of Ab
or Ag is injected into the capillary. Scatchard analysis of the change in migra-
tion time of the Ab or Ag as a function of the concentrations of Ag or Ab in
the buffer can then determine the binding constant.
In a system with slow kinetics, the Ag and Ab need to be preincubated
before injection into the capillary (45,46). Fixed concentrations of Ag or Ab
are then incubated with different concentrations of Ab or Ag. The quantity of
free Ab or Ag can be determined by the peak area using a calibration plot.
Scatchard analysis is made by plotting the total amount of Ab or Ag vs. bound
Ab or Ag. For the intermediate kinetics system, separation conditions, such
as applied voltage, length of the capillary, pH, and other factors, can be
changed so that the system can be analyzed using one of the final experimental
protocols.

III. SEPARATION AND DETECTION OF ANTIBODIES,

ANTIGENS, AND ANTIBODY–ANTIGEN COMPLEXES BY
HPCE METHODS

Reports on the separation of Abs by HPCE have mainly included the use of
capillary zone electrophoresis (CZE) and capillary isoelectric focusing (CIEF),
with much less being reported by CGE methods (40). The separation mecha-
nism of CZE is based on electrophoretic mobilities of the sample components,
which are affected by solvent characteristics, including pH, ionic strength, and
viscosity. The separation in CIEF is based on differences in isoelectric points
(pI ). Most work on Ab separations by CE has been used to separate a mono-
clonal Ab because its homogeneity when compared with a polyclonal Ab often
leads to a much higher specificity. This will also result in a higher sensitivity
for the final immunoaffinity process as there will be a single peak or species
detected, along with an improved possibility to develop simple and sensitive
assay methods. Monoclonal Abs produced by hybridoma or genetic engi-
neering techniques have often displayed a certain degree of microheterogen-
eity when analyzed by CE (47–56). This microheterogeneity may be caused by
differences in glycosylation, variations in protein sequences, posttranslational
modifications, improper folding, and other factors (57).
An Ab will interact with its Ag and form the usual Ab–Ag complexes.
It is easy to see that if a polyclonal Ab were used for recognition with its Ag,
the final electropherogram of the mixture could become quite complicated.
For the above reasons, Nielsen et al. used the monoclonal Ab specific for

© 2000 by Marcel Dekker, Inc.

human growth hormone (hGH) to study the CZE separation of Ag, Ab, and
the various possible complexes (58). Theoretically, if the monoclonal Ab were
used, only two types of complexes would be formed—those corresponding
to reaction at the one or two Ag binding sites of each Ab molecule. Figure 2
illustrates a series of CZE electropherograms for immunoglobulin (IgG), hGH,
and various mixtures containing an excess of IgG (Ab). This work demon-
strated the ability of CZE to separate the Ag and Ab complexes from excess
Ab and/or free Ag. In Fig. 2, it is apparent that incomplete resolution of all
possible Ab–Ag complexes has occurred, for various reasons. If there are a
large number of complexes formed (⬎2), then the resolution window between
free Ab and Ag may be too small to resolve these completely. Alternatively,
if the complexes are rapidly interconverting with one another, i.e., the kinetics
of formation and dissociation are rapid, perhaps too fast for the resolution
time scale, then these species would always produce broad, unresolved peaks
that are never really separable.

Figure 2 Electropherograms of IgG, hGH, and mixtures containing an excess of

IgG. Experimental conditions used a fused silica capillary 100 cm long (80 cm to the
detector) with 50 µm i.d. and 360 cm o.d. The buffer was 0.1 M tricine, pH 8.0, applied
voltage was 30 kV, injection volumes were ⬃9 nl, detection was at 200 nm (52).
(Reproduced with permission from the copyright holder, Elsevier Science Publishers
and Journal of Chromatography.)

© 2000 by Marcel Dekker, Inc.

 We have used bovine serum albumin (BSA) as an Ag and its mono-
clonal, anti-BSA, Ab to perform a separation study of the complexes by
CZE(20,59). In this example (Fig. 3), an AccuPure Z1 reagent was used to
suppress any unwanted protein and uncoated, silica wall interactions. In the
four sequences illustrated, the first two (a, b) illustrate injection of purified
(affinity chromatography) anti-BSA (a) and then BSA (b), whereas panel (c)
illustrates a mixture of Ag and Ab with anti-BSA Ab in excess. Peaks 1 and
2, it is assumed, represent the two possible Ag–Ab complexes formed in solu-
tion. We are assuming that the earlier eluting complex is due to a 1:1 complex
and peak 2 is perhaps due to a 1:2 complex of Ab–Ag. In Fig. 3, panel (d),
an excess of BSA was present, leading to residual BSA peak, and now what
appears to be only a single Ab–Ag complex, probably the 1:2 (peak 2), having
two BSA molecules complexed with each Ab species because there is no resid-
ual Ab present. The BSA–anti-BSA system is perhaps an ideal example to
study by CZE methods. These complexes appear to be quite stable under the
migration conditions and times. There are fairly narrow, well-resolved, com-
plex peaks, suggesting a lack of interconversion on this time scale.
However, in the absence of true identification of these complexes, it is
not 100% possible to identify each of the complex peaks (20,59). Indeed, in
none of the existing CE immunoaffinity studies reported have any Ab–Ag
complexes been identified by light scattering or mass spectrometric methods
(60a,b). This would require, for example, a size exclusion chromatographic
(SEC) separation of a particular complex and then on-line characterization
with isolation and reinjection under CZE conditions (60c).
In the case of BSA–anti-BSA, the two complexes were well separated.
These separation conditions may not, however, differentiate any microhetero-
geneity, if present, of the Ab. Alternatively, the microheterogeneity of the Ab
may be so little that it cannot be distinguished by these CZE conditions.
In Shimura and Karger’s work (61) that dealt with the immunoassay
(ACE or APCE) of hGH, they used CIEF to separate the complexes from Ag
and Ab. To avoid any microheterogeneity of the Ab that might complicate
their immunoassay, a Fab′ fragment from the Ab was used to interact with
the Ag. The complexes thus formed and the excess tagged Fab′ fragments
were all separated by CIEF. Figure 4 illustrates a set of typical CIEF electroph-
erograms for this system, with conditions indicated. In panel (a), using a buffer
without the Ag, only the two FL-tagged Fab′ species are present. In panel (b),
now with met-rhGH first complexed with the TR-Fab′ species, two separate
Fab′–Ag complexes appear, with perhaps some residual TR-Fab′ species/
peaks still present. The presence of two TR-Fab′ species, panel (a), illustrates
a generic problem in using immunoaffinity (Ab) recognition in all of CE:

© 2000 by Marcel Dekker, Inc.

Figure 3 A series of CZE electropherograms for bovine serum albumin (BSA), anti-
BSA, and complexes of BSA–anti-BSA. Operating conditions: uncoated capillary, 50
µm ⫻ 70 cm, 40 cm to detector, 15 kV applied voltage, UV detection at 214 nm,
vacuum injection 30 kpa-s, buffer 60 mM phosphate, pH 7.8, 1 M AccuPure Z1-methyl
reagent (Waters), sample dissolved in phosphate buffer, pH 7.0. (a) Injection of anti-
BSA purified on protein G column first. (b) Injection of BSA, monomer purified by
SEC first. (c) Mixture of BSA monomer and anti-BSA, with anti-BSA in excess. (d)
Mixture of BSA monomer and anti-BSA, with BSA in excess (peak assignments: peak
1 ⫽ 1:1 complex; peak 2 ⫽ 1:2 complex, other peaks as indicated) (59).

© 2000 by Marcel Dekker, Inc.

Figure 4 Example of an approach to immunoaffinity recognition in CZE using an
FL probe Fab fragment of the Ab (61). CIEF analysis of crude TR-Fab′ and its com-
plexes with met-rhGH. APCE was done with Pharmalyte 3–10 and using TR-Fab′
before purification as an affinity probe with (a) TTA-BSA buffer and (b) met-rhGH
(1 µg/ml) as samples. The free TR-Fab′ and the complexes are marked by * and **,
respectively. (Reproduced with permission from the copyright holder, American
Chemical Society and Analytical Chemistry.)

© 2000 by Marcel Dekker, Inc.

the formation and presence of several complexes. Multiple complexes reduce
sensitivity and specificity of the final assay, and only complicate the identifi-
cation of a single, individual Ag, possibly present together with cross-reacting
analogs that might also form Ag–Ab complexes with the same FL-tagged Ab
(Fab′) species. Thus, though use of Fab′ species for the immunorecognition
step does reduce microheterogeneity problems by removing the carbohydrate
region of the Ab, solution tagging with common FL reagents, such as tetra-
methylrhodamine (TR), often leads to multiple products, which again compli-
cate the final CE patterns (62,63).
There are several reports in the literature that have utilized CE-based
techniques for recognition of a specific antigen and then performed an immu-
noassay on the now-isolated species. These have all involved a form of ACE,
usually immuno-ACE, and thus could be applicable for combinatorial library
searching in the future.
Immunoassays are based on the bioaffinity between an Ag and its corre-
sponding Ab. Quantitation of Ag requires the ability to detect and discriminate
between the Ag–Ab complex and either the free Ab or Ag. The high separating
power of CE makes this a logical candidate for the development of immunoas-
says. In comparison with conventional immunoassays, a CE-based immunoas-
say has the following advantages: (a) It is fast and easy to automate. Fast
complexation rates for a homogeneous solution system (preinjection) require
less incubation time than for a heterogeneous, solution-solid phase system
(conventional immunoassays on a plate/tube). (b) Multiple washing steps
are eliminated. (c) Less sample is needed. (d) The high separation power
of CE can discriminate between specific and nonspecific binding, protein
variants, metabolites, and so forth. This last advantage is potentially signifi-
cant when compared with current methods for performing immunoassays on
a plate or in a tube or even by immunodetection in a flow injection mode (64).
That is, in CE, there is the potential for baseline resolution and thus identifica-
tion among the various possible antigens that could recognize a given Ab,
such as protein variants, metabolites, decomposition products, and deamida-
tion products. This is not possible in any of the current batch-type immunoas-
says.
For CE-based immunoassays, Fab fragments are usually preferred to the
intact Abs, due to the multiple Ag binding sites on Abs and microheterogeneity
of the Fc (crystalline or carbohydrate) portion of the Ab caused by the varia-
tions in glycosylation. That is why in almost all of the reported CZE electroph-
erograms for intact Abs, such peaks are usually quite broad, no matter the
specific separation conditions. Fab or Fab′ fragments are often, under very

© 2000 by Marcel Dekker, Inc.

similar HPCE conditions, much sharper and better resolved, especially in CZE
and CIEF. Papain and pepsin are the most commonly used reagents for Ab
digestion. Even after digestion, it is possible that more than a single Fab or
Fab′ species will be obtained. In the work of Schmalzing et al., three Fab
fragments were observed, due to the fact that papain cleavage is not very
specific, or derived from the microheterogeneity in the Fab fragments (65).
Cation exchange separation was therefore performed to achieve a single Fab
species.
Since Ag–Ab complex formation is a reversible reaction, a certain de-
gree of complex dissociation will take place inside of the capillary. Once the
complex dissociates, the Ag and Ab will rapidly move apart due to the differ-
ences in their electrophoretic mobilities. The level of the complex detected
may actually be lower than that formed in the original mixture. The amount
of decrease depends on the dissociation kinetics and the time required for
separation. To obtain maximal complex signal, a short separation time is nor-
mally recommended (61,66).
At present, most CE-based immunoassays can be divided into two cate-
gories: competitive and direct immunoassays. For competitive immunoassays,
the Ag molecules are first labeled with FL tags. Labeled Ag and a limited
amount of Ab are then added to the sample to be analyzed. The labeled Ag
will compete with the original/sample Ag in the sample for the limited binding
sites on the Ab. After incubation, a small amount of the Ab-Ag mixture is
injected into the capillary. Upon separation, using fluorescence detection, two
peaks will appear in the electropherogram, one coming from the labeled Ag
and the other from the Ab–Ag complex. The free Ag may elute together with
the labeled Ag if the FL tag is a small molecule, or it may be separated from
the labeled Ag, but under both conditions, no free Ag can be observed by FL
detection. The amount of Ag in the sample is directly proportional to the free,
FL-labeled Ag signal, and inversely proportional to the complex Ab-Ag (FL)
signal. In principle, both signals can be used for quantitation. Most of the
competitive immunoassays described are carried out on small molecules with
an FL tag.
For example, Schultz et al. demonstrated a competitive immunoassay
of insulin with a sensitivity around 10⫺9 M (66,67). Using this assay, the insu-
lin content of single islets of Langerhans was determined. Fluorescein Isothio-
cyanate (FITC) tagging of insulin yielded at least three distinct products, all
of which were separated by CZE, as illustrated in Fig. 5 (66). When these
FL-tagged insulin species were then complexed with their Fab, at least two
complexes were formed, appearing together with uncomplexed, FL-tagged in-

© 2000 by Marcel Dekker, Inc.

Figure 5 (Top) Electropherogram of 100 nM FITC-insulin under HPCE conditions
that employed: uncoated capillaries, 25 µm i.d. and 150 µm o.d., total lengths of 25–
30 cm, length to detector 12–15 cm, buffer of 0.05 M sodium phosphate with 0.025
M K2SO4 at pH 7.5, applied voltage 1000 V/cm, hydrostatic injection. (Bottom) Elec-
tropherogram of 100 nM FITC-insulin and 50 nM Fab. Peaks 2, 3, and 5 are FITC-
insulin, while peaks 1 and 4 are due to the formation of the complex of Fab with FITC-
insulin in peaks 2 and 5, respectively. An He-Cd laser was used as the excitation source
(66). (Reproduced with permission from the copyright holder, American Chemical So-
ciety and Analytical Chemistry.)

sulin (FL detection) (Fig. 5). Perhaps these results highlight some of the linger-
ing problems in using immunoassays and/or immunorecognition methods in
HPCE, whereby contrary to batch-type immunoassays, the presence of several
FL- or otherwise tagged Ag or Ab species in CE can lead to multiple, complex
peaks, complicating the final results. These limitations, especially with larger
Ags, also with tagged Abs, have not been resolved (62,63).
Schmalzing et al. developed a similar method to that above, but now
for cortisol in serum (57). In this case, it was fairly easy to FL-tag the small
cortisol molecule for this FL based CE-immunoassay, whereby a single final
tagged antigen formed. This is an excellent example of how the combination

© 2000 by Marcel Dekker, Inc.

of immunorecognition and immunoassay-based methods can be routinely
applied with CZE separations for improved, perhaps absolute, analyte identi-
fication and full quantitation, using excellent calibration plots and validated
methods.
Chen and Pentoney have reported a competitive immunoassay method
for digoxin in serum with sensitivity in the low 10⫺11 M range (68). The
high-resolution power of CE makes the simultaneous analysis of two, or
more than two, species feasible, whereas this may not be the case in most
other separation modes (e.g., HPLC). Chen and Evangelista also reported a
simultaneous competitive immunoassay of morphine and phencyclidine in
urine (69). The immunoassay could be performed routinely and reproducibly
in less than 5 min with detection limits of 4 nM for phencyclidine and 40 nM
for morphine.
The problem with labeling big molecules, such as proteins and antibod-
ies, is that multiple products are typically obtained (as above) because most
of the labeling reactions utilize the primary amino groups on the proteins and
antibodies. The tags can bind in different amounts and at different locations
to the proteins (62,70–73). In one illustrative example, this problem was
avoided by using thiol groups at the hinge region of the Fab’ fragment to
react with the label (61), although several chemical reactions were involved
to achieve the monothiol group on Fab’. Even here, two separate, tagged Fab’
species were produced (Fig. 4), still complicating the overall CE assay and
specificity.
Multiple labeled Ags (or Abs) lead to multiple peaks or, at times, a
broad peak in the CE electropherograms. These species can then form more
than one Ab–Ag complex, which complicates the analysis, due to perhaps
insufficient resolution of the labeled Ags and the Ab–Ag complexes. Some-
times chemical modifications to the Ag (or Ab) are necessary to facilitate the
separation by changing the final mass/charge ratio (74–76).
The principle of direct immunoassays is that Abs are first tagged with
FL labels and then mixed with the Ag sample of interest. The Ab should be
in excess. After incubation, the mixture is injected into the capillary, and the
amount of Ag can be determined by the Ab–Ag complex signal. Competitive
immunoassays are usually preferred for small Ags because the separation of
the Ab–Ag complex from the Ab can be very difficult for small Ags. As
above, Shimura and Karger reported a direct immunoassay of hGH (61). Due
to the focusing effect of CIEF (Fig. 4) and the high sensitivity of CE–laser-
induced fluorescence (LIF), a detection limit of 0.1 ng/ml was achieved. Simi-
larly, Chen demonstrated the direct immunoassay of IgG with a detection limit
of 6 ⫻10⫺10 (74).

© 2000 by Marcel Dekker, Inc.

 A more universal immunoassay method in CE was developed by Reif
et al. (76). In this approach, generic protein G was first tagged at several
locations with solution FITC reagent. The heavily FITC-tagged protein G was
then combined with the Ab, due to the affinity between protein G and the Fc
portion of the antibody. This FITC, dual-protein reagent was then complexed
with the Ag (analyte) in a CE-based assay. In this fashion, FITC-tagged protein
G behaved as an FL label. The Ab did not have to be labeled for detection.
This may have been the only sandwich immunoassay reported in CE. The
term sandwich here refers to an approach that utilizes two antibodies sur-
rounding the antigen, for improved specificity and identification of the correct
analyte/antigen.
All of the work described above used an Ab-Ag reaction in solution.
The same reaction can occur when Ab is immobilized onto the inner walls of
the capillary. For example, Phillips and Chmielinska reported the analysis of
cyclosporin in tears using this CE-based method (77). One third of the capil-
lary was immobilized with the purified Ab. The sample to be analyzed was
introduced into the capillary and incubated for 10 min. After rinsing the capil-
lary with the buffer to remove the unbound materials, the buffer in the reser-
voirs was changed to an acidic solution. Under these conditions, the Ab–Ag
immobilized complexes were broken. The active Ab remained immobilized
on the CE capillary column, whereas the Ag was brought to the detector by
EOF. Different from conventional, batch-type immunoassays, cyclosporin A
(Cys A) and its metabolites could be differentiated due to the differences in
their relative electrophoretic mobilities. As shown in Fig. 6, this is perhaps
an ideal illustration of the power of combining Ab-based immunoassays with
CE separations, and the inherent ability to resolve very similar species after
a first immunorecognition step. It is apparent that this entire approach could
be readily utilized for combinatorial library searching. In this particular in-
stance, the assay did not use En or FL enhancement for detection, but rather
represents a form of immunoaffinity extraction and preconcentration. This was
then followed by the high resolving power of CE for whatever species (un-
tagged) were first recognized by the immobilized Ab. It would, of course, be
possible to first tag the Ag analytes in a sample and then introduce these into
the immobilized Ab-CE system for improved isolation, separation, and detec-
tion. This approach, in general, is quite similar to the combination of immu-
noaffinity sample preparation followed by HPLC separation-detection meth-
ods (20,78–84).
Like conventional immunoassays, immunoassays in CE can be per-
formed in different modes, such as sandwich, double sandwich, competitive,
or direct on/off. However, the more species (analytes) that are possibly pres-

© 2000 by Marcel Dekker, Inc.

Figure 6 Immunoaffinity or immunoassay CE profiles of tear fluid obtained from
(a) a patient with no clinical signs of CyA toxicity and (b) a patient during an episode
of systemic toxicity. Relevant peaks: CyA, cyclosporin A; peaks 1–4 represent cyclo-
sporin metabolites: 1, AM1, 2, AM9, 3, AM1c, 4, AM4N (77). (Reproduced with
permission from the copyright holder, John Wiley & Sons, Ltd. and Biomedical Chro-
matography.)

© 2000 by Marcel Dekker, Inc.

ent, the more challenging and difficult the needed separation may become.
Other than FL tags, CL, EC, and enzyme labels can also be employed. Much
less has been described using such enhanced (signal-amplified) approaches in
immunorecognition or immunoassays combined with CZE. Because of the
amplification effect of enzyme–substrate reactions, enzyme labels should pro-
vide lower (better) detection limits (85).

IV. SPECIFIC APPLICATIONS OF HPCE TO COMBINATORIAL

CHEMISTRY ANALYSIS

A search of the recent literature on the use of HPCE for the analysis of com-
binatorial libraries shows that as of today not a great deal of work has
been done in this area. In fact, only seven published papers could be found,
four of which are from the same group (86–88). However, as three of the
seven were published in 1996 (88–90), it can be suggested that perhaps
this research area is now picking up and will show more results in the near
future.
The first published report on the use of HPCE to analyze a combinatorial
library came in 1993 by Chu et al. (86). They evaluated using affinity capillary
electrophoresis (ACE) for the identification, based on competitive binding, of
tight-binding ligand(s) for a receptor in mixtures of equimolar ligand libraries.
This has become the general approach in HPCE for determining binding con-
stants, biologically recognized and presumably active, interacting partners,
and as a general, potentially automatable, on-line screen of combinatorial li-
braries for active lead/target compounds. In this first example of using ACE
to find active target compounds, vancomycin was used as the receptor and a
small library was created consisting of 32 peptides to examine feasibility. This
approach of library searching has now become the basis of a general method
to perform drug screening and has since been commercialized (91). Of course,
since this generalized approach appears to work for potential, lead drug candi-
dates, it can be readily extended in the search for active agricultural chemicals,
fragrances, perfumes, insect pheromones, affinity ligands, and so forth. The
method is not limited to identifying only drug candidates. That is, it could just
as readily be utilized for seeking agricultural compounds, perfumes, veterinary
products, or any bioactive chemical compound.
Using this approach, vancomycin was first run in the electrophoresis
buffer with a probe ligand, Fmoc-Gly-D-Ala-D-Ala (L), that showed high
binding affinity for the target substrate, vancomycin. It was shown that the
concentration of vancomycin affected the electrophoretic mobility of L (Fig.

© 2000 by Marcel Dekker, Inc.

Figure 7 Illustration of CE-UV for identification of peptides binding to vancomycin
by using affinity recognition in CE. How the concentration of vancomycin in the elec-
trophoresis buffer (20 mM phosphate, pH 7.4) affects the electrophoretic mobility of
Fmoc-Gly-D-Ala-D-Ala (L, black circles) but not Fmoc-Gly-L-Ala-L-Ala (open cir-
cles). Specific conditions indicated elsewhere (86a). (Reproduced with permission of
the copyright holder, publisher and Journal of Organic Chemistry.)

7a, b), as can be seen in the shift of the L peak. When the peptide library was
added, the electrophoretic mobility of L again was changed due to one or
more of the peptides (L′) competing with L for vancomycin. Through the use
of subsets of the library, it was possible to then specify (narrow possibilities)
which of the peptide(s) was L′ (Fig. 8). Figure 9 shows how the mixture of
32 peptides was split into 2 groups of 16 peptides each and analyzed by CE.
One set of 16 could then be eliminated as no shift was seen. The remaining
16 were split into 2 groups of 8 peptides each and the above was repeated.
Eleven experiments were necessary to determine the one peptide that was a

© 2000 by Marcel Dekker, Inc.

Figure 8 Stepwise elimination of noninteracting peptides from a mixture of 32 pep-
tides and identification of one tight-binding ligand for vancomycin. Interpretation of
each electropherogram is described in the text. Specific experimental conditions are
described elsewhere (86a). (Reproduced with permission of the copyright holder, pub-
lisher and Journal of Organic Chemistry.)

tight binding receptor. In this way it was determined that L′ was a,e-Ac2-L-
Lys-D-Ala-D-Ala (Fig. 7c, d). Thus, it was shown that HPCE could be used
to identify ligands from small libraries that bind most tightly to a receptor,
such as vancomycin (86).
In 1995, a report using ACE-MS was published (87). Chu et al. reported
on the development of a simple, one-step procedure for the on-line separation
and identification of ligands that again bind most tightly to a receptor. Vanco-
mycin was again chosen as the receptor, and it was used in the electrophoretic
buffer to completely fill the capillary. In order to prevent vancomycin from
flowing into the MS, the electrophoretic buffer pH was chosen to prevent the
vancomycin from migrating. A neutral, hydrophilic, polymer-coated capillary
was used to minimize EOF in the capillary and to therefore reduce the vanco-

© 2000 by Marcel Dekker, Inc.

Figure 9 Affinity CE-MS (ACE) of a synthetic, all-D, Fmoc-DDXX library of 100
tetrapeptides using vancomycin as the receptor (A–D). Selected ion electropherograms
for the masses are indicated; (E) reconstructed ion electropherogram for runs without
(left) and with (right) vancomycin in the electrophoresis buffer. Specific ACE and MS
conditions are indicated elsewhere (87). (Reproduced with permission of the copyright
holder, the publisher and the Journal of the American Chemical Society.)

mycin flow to the MS. The idea of the research was that ligands that bound
tightly to the receptor would be retained in the capillary for a longer period.
The later eluting peptides could then be identified by MS. From a known 100-
tetrapeptide all-D Fmoc-DDXX library, three peptides (Fmoc-DDYA, Fmoc-
DDFA, and Fmoc-DDHA) were identified that bound tightly to the vancomy-

© 2000 by Marcel Dekker, Inc.

cin. Figure 9a–d illustrates ion electropherograms for these three, both without
(left) and with (right) receptor. Figure 9e shows the reconstructed ion elec-
tropherogram, again both with and without receptor. It was also shown that
the binding was sequence-specific, as Fmoc-DDAY, Fmoc-DDAF, and Fmoc-
DDAH did not bind to vancomycin in these experiments (87).
Chu et al. have also published more extensive results of using ACE-MS
to successfully screen combinatorial libraries using larger numbers (500–1000
compounds) and a larger variety of amino acid residues (88). They felt that
larger libraries would be difficult to analyze because the sensitivity of the MS
instrument was not sufficient to handle many more compounds. They sug-
gested that a more sensitive MS detection method, such as ion trap-MS, might
increase the size of the library suitable for successful screening. Another way
to increase the possible library size, they explained, was to remove the non-
binding peptides prior to introduction of the more interesting, strongly binding
compounds, into ACE-MS. In order to remove many of the nonbinding ligands
and thus possibly allow a larger initial library to be used, an affinity extraction
step was developed in which the receptor was immobilized to a solid support.
The peptides that did not bind were discarded, whereas the bound peptides
were eluted and then analyzed with the ACE-MS method. This caused the
binding peptides to be preselected and preconcentrated, which improved the
capabilities of the ACE-MS method for screening combinatorial libraries (88).
Boutin et al. have also used HPCE [as well as NMR, MS, and tandem
MS (MS/MS)] to analyze combinatorial libraries (89). Their goal was to pro-
vide a complete set of analytical data on a tetrapeptide library synthesized
from 24 amino acids. The purpose of the work was to prove that all of the
amino acids used in the synthesis were present in the resulting library, in
the amounts theoretically calculated. Using HPCE, the library was split into
different classes of compounds based on their net charges and the pH of the
electrophoretic buffer. Integration of the areas under each curve, while taking
into account the effect of the number of charges on the migration rate (amount
⬇ area/time), allowed an estimation of the number of compounds in each
class. Correlation coefficients of 0.98 and higher (n ⫽ 10) were found between
the theoretically calculated number of compounds in each class and the values
obtained by integration. HPCE was also used to show that the side chains of
the amino acids were fully deprotected (89).
Dunayevskiy et al. showed the ability of HPCE-MS to determine the
purity and composition of a library theoretically composed of 171 disubsti-
tuted xanthene derivatives, with the possibility of analyzing libraries of up to
1000 components (92). Previously, the ability of MS alone to analyze a library
of up to 55 components was shown (93), but it was suggested that for more

© 2000 by Marcel Dekker, Inc.

complex mixtures a second separation method needed to be added to better
resolve the species. HPCE-MS was attempted. The investigators found that
124 of the possible 171 compounds overlapped in molecular weight, which
would have made MS identification alone difficult if not impossible. When
HPCE was used prior to MS, most of these 124 compounds were separated,
with only eight MS/MS experiments needed to identify the 19 unresolved
molecules with overlapping molecular weights. Figure 10 shows how the addi-
tion of an organic modifier, 40% methanol in this case, to the electrophoretic
buffer improved the resolution. Figure 10a shows the comigrations without
the organic modifier and 10b shows how the resolution was improved so that
all six peaks could be distinguished when the modifier was added. This addi-
tion of an organic modifier caused only four MS/MS experiments to be needed
for nine compounds (92).
Finally, Jung et al. published a communication showing that cyclohexa-
peptide libraries could be used as chiral selectors in HPCE (90). They ex-
plained that a lot of time and effort could be saved by using a library of
possible chiral selector compounds, rather than attempting to test single com-
pounds one by one. Furthermore, they suggested that some libraries might
show cooperative effects between the components that could affect enantiose-

Figure 10 CE-MS electropherogram of different mixtures of xanthene derivatives

in Tris-acetate buffer at pH 7.9. (a) Mixture 2, Ile/X/Ile(1), Ile/X/Pro (2), Ile/X/Ala
(3), Pro/X/Pro (4), Pro/X/Ala (5), and Ala/X/Ala (6). (b) Mixture 2 dissolved in 20
mM Tris-acetate buffer at pH 7.9 containing 40% (vol/vol) MeOH. E ⫹ 07 ⫽ 107 (92).
(Reproduced with permission of the copyright holder, publisher and the Proceedings of
the National Academy of Science USA).

© 2000 by Marcel Dekker, Inc.

Figure 11 Enantiomeric resolution of (a) DNP-D, L-glutamic acid with c(DFXXXa)
and (b) DNP-D, L-glutamic acid with c(RKXXXa). Specific conditions indicated else-
where (90). (Reproduced with permission of the copyright holder, publisher and
Angew. Chem. Int. Ed. Engl.).

lectivity. They successfully separated 2,4-dinitrophenyl-D,L-glutamic acid,

when they ran a cyclopeptide library composed of c(DFXXXa) in the running
buffer at 20 kV (Fig. 11a). They improved the resolution and selectivity of
this separation when they changed the library to c(RMXXXa) and used inverse
polarity at ⫺10 kV (Fig. 11b). Work is now in progress on the use of their
method to screen for the most effective chiral selector and to then identify
such component(s) (90).
From these examples, it can be seen that the use of HPCE to separate,
identify, and analyze compounds in combinatorial libraries is on the increase.
The questions are how to further improve HPCE approaches for combinatorial
library searching, how to further improve the crude preseparation of uninter-
esting compounds in the library from more interesting and potentially binding
ligands, and then how to improve the overall HPCE separations prior to use of
MS or MS/MS routines to identify each individual binding library component.

V. FUTURE ROLE OF HPCE IN COMBINATORIAL LIBRARY

ANALYSIS AND INTERPRETATION

CE appears to offer some very exciting and scientifically significant advan-

tages insofar as being able to select (recognize) bioactive drug candidates and

© 2000 by Marcel Dekker, Inc.

resolve these from less interesting subsets of a combinatorial library popula-
tion. There are some lingering problems in using CE-based techniques, such
as requiring a major instrumental commitment of CE and MS components.
These ACE-MS techniques also require sophisticated operators, i.e., trained
personnel familiar with both CE and MS operations and sample requirements.
There is the need to develop good separation conditions for the library in the
presence of the receptor molecule(s), so that just those compounds actually
interacting with the receptor will be retarded or retained and separated from
less interesting library components. There is of course also the need for puri-
fied receptor molecules, but that is true in any library search method, be that
LC-, MS-, or CE-based. There are alternative techniques that might work, at
times as well or even better than ACE-MS, such as affinity filtration or affinity
membrane separation methods, prior to HPLC-MS or CE-MS, where the af-
finity step is performed apart from the separation instrument (precolumn). Be-
cause there are numerous separation conditions already available for many
library mixtures, these can be readily utilized for ACE-MS library search
methods/conditions, perhaps with little further method optimization. There is
little question but that ACE-MS methods do work, though the maximum size
of the library that can be successfully searched remains a question. There
are also very few research groups that are routinely utilizing ACE-MS for
combinatorial library searching, although this appears from the available liter-
ature to be a completely viable and successful technique, perhaps for a wide
variety of compound (drug) types. There appears to be a growing role for
ACE and ACE-MS in combinatorial library searching, and one would expect
that its application will only continue to grow in the future.

VI. CONCLUSIONS

This chapter reviews some of the basic principles of CE operations, and how
CE can be used to separate individual library members on the basis of their
interaction and recognition by a receptor molecule. Also reviewed are some
of the basic needs or requirements for a successful library searching approach
in CE, such as having the receptor molecule in the sample before injection or
in the CE buffer during separation. Both are totally viable approaches and have
been described in the literature. The literature involving affinity approaches in
CE, antigen–antibody recognition, antibody–drug interactions, resolution of
ligand–receptor complexes from other components of the sample, and then
the use of ACE and ACE-MS to resolve active, receptor-binding species from
nonactive components of the original library mixture, has also been reviewed.

© 2000 by Marcel Dekker, Inc.

In addition, this chapter describes how resolved library components can then
be structurally identified by various MS approaches, so that active, lead com-
pounds cannot only be shown active against a particular target receptor, but
their actual structures can then be determined with almost 100% success using
modern MS approaches and structure software. These techniques are clearly
very new, but they draw on older, more established CE and MS methods, and
therefore the hyphenated methods of ACE-MS also appear to be perfectly
usable and reliable. These analyses, now possible by ACE-MS, are nothing
more than separation/detection hyphenated methods, which are now combin-
ing an affinity recognition step in the ACE portion with structural determina-
tions via the MS portion of the ACE-MS system. However, the affinity recog-
nition step in the ACE part simplifies or isolates interesting, perhaps target,
compounds from all other library components, making the job of the MS much
simpler and more straightforward. The affinity recognition step and the less
demanding MS analysis are perhaps the real attributes of using CE and ACE-
MS for combinatorial library searching to isolate active drug candidates
against specific receptors (or ligands).

GLOSSARY

Ab ⫽ antibody
Abs ⫽ antibodies
ACE ⫽ affinity capillary electrophoresis
Ag ⫽ antigen
Ab–Ag ⫽ antibody-antigen complex
Ab–En ⫽ antibody–enzyme conjugate
anti-BSA ⫽ antibody to BSA
anti-Ab ⫽ antibody of Ab
APCE ⫽ affinity probe capillary electrophoresis
BSA ⫽ bovine serum albumin
CE ⫽ HPCE ⫽ capillary electrophoresis
CEC ⫽ capillary electrochromatography
CGE ⫽ capillary gel electrophoresis
CIEF ⫽ capillary isoelectric focusing
CL ⫽ chemiluminescence detection
CZE ⫽ capillary zone electrophoresis
Cys A ⫽ cyclosporin A
EC ⫽ electrochemical detection
ELISA ⫽ enzyme-linked immunosorbent assay

© 2000 by Marcel Dekker, Inc.

EMMA ⫽ enzyme-modulated microanalysis
En ⫽ enzyme
EOF ⫽ electroosmotic flow
Fab ⫽ fragment of intact antibody containing recognition region (epitope)
Fab’ ⫽ tow Fab fragments held together by at least one disulfide bridge
Fc ⫽ crystalline fragment of intact antibody containing carbohydrate regions
FL ⫽ fluorescence detection
FITC ⫽ fluorescein isothiocyanate
FMOC-Cl ⫽ 9-fluorenyl methyl chloroformate
FMOC ⫽ 9-fluorenyl methyl formyl (grouping)
FSCE ⫽ free solution capillary electrophoresis
FTIR ⫽ Fourier transform infrared
hGH ⫽ human growth hormone
HPAC ⫽ high-performance affinity chromatography
HPIAC ⫽ higher-performance immunoaffinity chromatography
HPLC ⫽ high-performance liquid chromatography
HRP ⫽ horseradish peroxidase (enzyme)
IACE ⫽ immunoaffinity capillary electrophoresis
ICE ⫽ immunoassay CE
ICA ⫽ immunochromatographic analysis
ID ⫽ immuno-detection
IgG ⫽ immunoglobulin
immuno-ACE ⫽ immunoaffinity capillary electrophoresis
LIF ⫽ laser-induced fluorescence (detection)
MECC ⫽ micellar electrokinetic capillary chromatography
MEKC ⫽ micellar electrokinetic chromatography
MS ⫽ mass spectrometry or spectrometer
MS/MS ⫽ tandem mass spectrometry
NMR ⫽ nuclear magnetic resonance
RIA ⫽ radioimmunoassay detection
SEC ⫽ size exclusion chromatography
TR ⫽ tetramethylrhodamine FL tag (probe)
UV ⫽ ultraviolet detection

ACKNOWLEDGMENTS

Our knowledge and appreciation of affinity recognition in HPLC and HPCE

has been attributable to several firms and individuals. For example, we must
indicate our appreciation to numerous individuals within PerSeptive Biosys-

© 2000 by Marcel Dekker, Inc.

tems, Inc., especially M. Meyes, R. Mhatre, T. Naylor, M. Vanderlaan, S.
Martin, F. Regnier, and others, who over the years have provided us with
guidance, suggestions, encouragement, and materials. Professor H. Zou is ac-
knowledged for his early collaborations in the areas of HPIAC and ID. D.
Fisher also collaborated on some of the early developments and optimization
in utilizing ID in a postcolumn, HPLC format. Professor G. Li also collabo-
rated on some of the early developments in utilizing HPIAC and ID, prior
to interfacing with HPLC and then to our own development of affinity and
immunoaffinity recognition studies in HPCE areas. Several graduate students
and postdoctoral fellows or visiting scientists worked with us on the develop-
ment of affinity CE applications, such as R.-L. Qian, R. Strong, B.-Y. Cho,
H. Zou, and X. Liu. Certain early drafts of sections of this review were pre-
pared by R.-L. Qian and X. Liu, for which we are grateful.
Finally, financial support and technical collaborations have been pro-
vided NU in antibody areas by Pharmacia and Upjohn Pharmaceutical Com-
pany, through the Animal Health and Drug Metabolism Division (J. Nappier
and G. Fate). Additional antibody analysis collaborations have been possible
through SmithKline Beecham Pharmaceuticals (D. Nesta and J. Baldoni).
These contracts have allowed us to become involved in affinity and immuno-
affinity CE areas. Isco Corporation, Thermo Separation Products (Thermo
Quest), and Waters Corporation have all donated major instrumentation, mate-
rials, and supplies to our efforts in the areas of affinity CE. Colleagues at
Supelco, Phase Separations, Ltd., J&W Scientific, and Unimicro Technologies
have all donated coated or packed capillaries for studies in CE and CEC.
We are very appreciative of all these collaborations and technical/financial
assistance in developing CE, CIEF, and, most recently, ACE approaches for
proteins and antibodies.

REFERENCES

1. JW Jorgenson, M Phillips, eds. New Directions in Electrophoretic Methods. ACS

Symposium Series, Volume 335. Washington, D.C.: American Chemical Soci-
ety, 1987.
2. CS Horvath, JG Nikelly, eds. Analytical Biotechnology: Capillary Electrophore-
sis and Chromatography. ACS Symposium Series, Volume 434. Washington,
D.C.: American Chemical Society, 1990.
3. SFY Li. Capillary Electrophoresis: Principles, Practice and Applications, Am-
sterdam: Elsevier, 1992.

© 2000 by Marcel Dekker, Inc.

 4. PD Grossman, JC Colburn, Eds. Capillary Electrophoresis: Theory and Practice.
San Diego, CA: Academic Press, 1992.
5. N Guzman, ed. Capillary Electrophoresis: Technology. New York: Marcel Dek-
ker, 1993.
6. R Weinberger. Practical Capillary Electrophoresis. San Diego, CA: Academic
Press, 1993.
7. JP Landers, ed. CRC Handbook of Capillary Electrophoresis: Principles, Meth-
ods, and Applications. Boca Raton, FL: CRC Press, 1994.
8. DN Heiger. High Performance Capillary Electrophoresis: An Introduction, A
Primer. 2nd ed. Waldbronn, Germany: Hewlett-Packard, 1992.
9. DR Baker. Capillary Electrophoresis. Techniques in Analytical Chemistry Se-
ries. New York: John Wiley and Sons, Inc., 1995.
10. PG Righetti, ed. Capillary Electrophoresis in Analytical Biotechnology. CRC
Series in Analytical Biotechnology. Boca Raton, FL: CRC Press, 1996.
11. P Camillari, ed. Capillary Electrophoresis: Theory and Practice. Boca Raton, FL:
CRC Press, 1993.
12. RA Mosher, W Thormann. The Dynamics of Electrophoresis. Weinheim, FRG:
VCH, 1992.
13. KD Altria, MM Rogan. Introduction to Quantitative Applications of Capillary
Electrophoresis in Pharmaceutical Analysis: A Primer. Fullerton, CA: Beckman
Instruments, 1995.
14. KD Altria, Ed. Capillary Electrophoresis Guidebook: Principles, Operation, and
Applications. Methods in Molecular Biology Vol. 52. Totowa, NJ: Humana
Press, 1996.
15. LA Thompson, JA Ellman. Synthesis and applications of small molecule librar-
ies. Chem Rev 96:555–600, 1996.
16. JC Hogan, Jr. Combinatorial chemistry in drug discovery. Nature Biotech 15:
328–330 1997.
17. S Borman. Combinatorial chemistry, Special Report. C&EN, 43-62 (February
24, 1997).
18. YM Dunayevskiy. Applications of mass spectrometry in combinatorial chemistry
research. Northeastern University, Ph.D. dissertation, December 1996.
19. IS Krull, BY Cho, R Strong, M Vanderlaan. Principles and practice of immuno-
detection in FIA and HPLC. LC/GC Int Magazine 10:278–292, 1997.
20. H Zou, Y Zhang, P Lu, IS Krull. Characterization of immunochemical reaction
for human growth hormone with its monoclonal antibody by perfusion protein
G affinity chromatography and capillary zone electrophoresis. Biomed Chro-
matogr 10:78–82, 1996.
21. BY Cho, R Strong, H Zou, DH Fisher, J Nappier, IS Krull. Immunochromato-
graphic analysis of bovine growth hormone releasing factor involving RP-HPLC-
immunodetection. J Chromatography: 743:181–194, 1996.
22. H Zou, Y Zhang, P Lu, and IS Krull. Perfusion immunoaffinity chromatography
and its application in analysis and purification of biomolecules. Biomed Chroma-
tography 10:122–126, 1996.

© 2000 by Marcel Dekker, Inc.

23. RE Strong, BY Cho, D Fisher, JL Nappier, IS Krull. Immunodetection ap-
proaches and high performance immunoaffinity chromatography for an analog
of bovine growth hormone releasing factor at trace levels. Biomed Chromatogra-
phy, 10:337–345, 1996.
24. RL Lundblad. Chemical Reagents for Protein Modifications. 2nd ed. Boca Raton,
FL: CRC Press, 1991.
25. Pierce Chemical Company, a. Pierce Catalog and Handbook of Life Science and
Analytical Research Products. (Pierce Chemical Company, Rockford, IL, 1994–
96); b. Pierce Molecular Biology Catalog of Molecular Biology Research Prod-
ucts. (Rockford, IL, 1994-97).
26. MD Savage, G Mattson, S Desai, GW Nielander, S Morgensen, EJ Conklin.
Avidin-Biotin Chemistry: A Handbook. Rockford, IL: Pierce Chemical Com-
pany, 1992, pp. 175–190.
27. PF Dimond. ImmunoDetection I: Perfusion immunoassays for biomolecular de-
tection. BioConcepts Technical Newsletter. 1(2), Cambridge, MA: PerSeptive
Biosystems, 1993.
28. ID, Immunodetection, Real-Time Immunoassay. Cambridge, MA: PerSeptive
Biosystems Technical Literature, 1994.
29. Integral, Micro-Analytical Workstation Applications Guide. Cambridge, MA:
PerSeptive Biosystems, Inc., 1994.
30. PerSeptive Biosystems Technical Operating Instructions, ImmunoDetection Sen-
sor Cartridge. Cambridge, MA: PerSeptive Biosystems, 1992.
31. CB Quern. Immunodetection technology II: Enzyme-labeled perfusion immuno-
assays. BioConcepts Technical Newsletter. 2:1–4, Cambridge, MA: PerSeptive
Biosystems, 1994.
32. UB Sleytr, P Messner, D Pum, M Sara, eds. Immobilized Macromolecules: Ap-
plication Potentials. Heidelberg: Springer-Verlag, 1993.
33. RF Taylor, Ed. Protein Immobilization: Fundamentals and Applications. New
York: Marcel Dekker, 1991.
34. M Husain, C Bieniarz. FC site-specific labeling of immunoglobulins with calf
intestinal alkaline phosphatase. Bioconj Chem 5:482–490, 1994.
35. GT Hermanson, AK Mallia, PK Smith. Immobilized Affinity Ligand Techniques.
2nd ed. San Diego, CA: Academic Press, 1997.
36. IS Krull, J Mazzeo. Capillary electrophoresis: the promise and the practice. Na-
ture 357:92–94, 1992.
37. JR Chapman, ed. Protein and Peptide Analysis by Mass Spectrometry. Methods
in Molecular Biology. Volume 61. Totowa, NJ: Humana Press, 1996.
38. JT Watson. Introduction to Mass Spectrometry. 3rd ed. Hagerstown, MD: Lip-
pincott-Raven Publishers, 1997.
39. BY Cho, R Strong, IS Krull. HPCE of a fermentation derived, cyclic peptide
analog, animal growth promotor. J Chromatography. B, 697:163–174, 1997.
40. IS Krull, X Liu, J Dai, C Gendreau, G. Li. HPCE methods for the identification
and quantitation of antibodies. J Pharm Biomed Anal. 16:377–393, 1997.

© 2000 by Marcel Dekker, Inc.

41. Y Chu, WJ Lees, A Stassinopoulos, CT Walsh. Using affinity capillary electro-
phoresis to determine binding stoichiometries of protein-ligand interactions. Bio-
chemistry 33:10616–10621, 1994.
42. NHH Heegaard. Determination of antigen–antibody affinity by immuno-capil-
lary electrophoresis. J Chromatography 680:405–412, 1994.
43. J Liu, KJ Volk, MS Lee, EH Kers, IE Rosenberg. Affinity capillary electro-
phoresis applied to the studies of interactions of a member of heat shock pro-
tein family with an immunosuppressant. J Chromatography 680:395–403,
1994.
44. AM Arentoft, H FroKiaer, S Michaelsen, H Sorensen, S Sorensen. High-perfor-
mance capillary electrophoresis for the determination of trypsin and chymotryp-
sin inhibitors and their association with trypsin, chymotrypsin and monoclonal
antibodies. J Chromatography 652:189–198, 1993.
45. NHH Heegaard, FA Robey. Use of capillary zone electrophoresis to evaluate
the binding of anionic carbohydrates to synthetic peptides derived from human
serum amyloid P component. Anal Chem 64:2479–2482, 1992.
46. NHH Heegaard, FA Robey. Use of capillary zone electrophoresis for the analysis
of DNA-binding to a peptide derived from amyloid P component. J Liq Chro-
matogr 16:1923–1939, 1993.
47. NA Guzman, L Hernandez, BG Hoebel. Capillary electrophoresis, a new era in
microseparations. BioPharm 2:22–37, 1989.
48. NA Guzman, MA Trebilcock, JP Advis. Capillary electrophoresis for the analyti-
cal separation and semi-preparative collection of monoclonal antibodies. Anal
Chim Acta 249:247–255, 1991.
49. SK Hjerten, F Kilar, JL Liao, A Chen, C Siebert, M Zhu. Carrier-free zone
electrophoresis, displacement electrophoresis and isoelectric focusing in a
high-performance electrophoresis apparatus. J Chromatography 403:47–61,
1987.
50. T Wehr, M Zhu, R Rodriguez, D Burke, K Duncan. High performance isoelectric
focusing using capillary electrophoresis instrumentation. Am Biotech Lab. 8:22–
29 1990.
51. M Zhu, R Rodriguez, T Wehr. Optimizing separation parameters in capillary
isoelectric focusing. J Chromatography 559:479–488, 1991.
52. BJ Compton. Electrophoretic mobility modeling of proteins in free zone capillary
electrophoresis and its application to monoclonal antibody microheterogeneity
analysis. J Chromatography 559:357–366, 1991.
53. SJ Harrington, R Varro, TM Li. High-performance capillary electrophoresis as
a fast in-process control method for enzyme-labelled monoclonal antibody conju-
gates. J Chromatography 559:385–390, 1991.
54. MA Costello, C Woititz, J DeFeo, D Stremlo, LFL Wen, DJ Palling, K Iqbal,
NA Guzman. Characterization of humanized anti-TAC monoclonal antibody by
traditional separation techniques and capillary electrophoresis. J Liq Chromatog-
raphy 15:1081–1097, 1992.

© 2000 by Marcel Dekker, Inc.

55. NA Guzman, J Moschera, K Iqbal, AW Malick. Effect of buffer constituents on
the determination of therapeutic proteins by capillary electrophoresis. J Chroma-
tography 608:197–204, 1992.
56. (a) TL Huang, PCH Shieh, N Cooke. Isoelectric focusing of proteins in capillary
electrophoresis with pressure-driven mobilization. Chromatographia 39:543–
548 1994; (b) TL Huang, PCH Shieh, N Cooke. The separation of hemoglobin
variants by capillary zone electrophoresis. JHRCCC 17:676–678, 1994.
57. D Schmalzing, W Nashabeh, XW Yao, R Mhatre, FE Regnier, NB Afeyan, M
Fuchs. Capillary electrophoresis-based immunoassay for cortisol in serum. Anal
Chem 67:606–612, 1995.
58. RG Nielsen, EC Rickard, PF Santa, DA Sharknas, GS Sittampalam. Separation
of antibody-antigen complexes by capillary zone electrophoresis, isoelectric fo-
cusing and high-performance size-exclusion chromatography. J Chromatography
539:177–185, 1991.
59. R Qian, I Krull. Unpublished results, 1994–1996.
60. (a) IS Krull, R Mhatre, J Cunniff. LC/GC Magazine. 12:914, 1994; (b) IS Krull,
R Mhatre, J Cunniff. LC/GC Magazine. 13:30, 1995; (c) RL Qian, R Mhatre,
IS Krull. Characterization of antigen–antibody complexes by size exclusion
chromatography (SEC) coupled with low angle light scattering photometry
(LALLS) and viscometry. J Chromatogr. A, 787:101–119, 1997.
61. K Shimura, B Karger. Affinity probe capillary electrophoresis: analysis of recom-
binant human growth hormone with a fluorescent labeled antibody fragment.
Anal Chem 66:9–15, 1994.
62. (a) G Li, J Yu, IS Krull, S Cohen. An improvement for the synthesis of a styrene-
divinylbenzene, copolymer based, 6-aminoquinoline carbamate reagent. Appli-
cation for derivatization of amino acids, peptides, and proteins. J Liquid Chro-
matogr 18:3889–3918, 1995; (b) IS Krull, R Strong, Z Sosic, BY Cho, S Beale,
S Cohen. Labeling reactions applicable to chromatography and electrophoresis
of minute amounts of proteins. J Chromatography B, Biomed Applics. B, 699:
173–208, 1997.
63. PR Banks, DM Paquette. Monitoring of a conjugation reaction between fluores-
cein isothiocyanate and myoglobin by capillary zone electrophoresis. J Chroma-
tography 693:145–154, 1995.
64. IS Krull, BY Cho, R Strong, M Vanderlaan. Principles and practice of immuno-
detection in FIA and HPLC. LC/GC Int Magazine 10:278–292, 1997.
65. (a) E McCarthy, G Vella, R Mhatre, YP Lim. Rapid purification and monitoring
of immunoglobulin from ascites by perfusion ion-exchange chromatography.
J Chromatography 743:163–170, 1996; (b) R Mhatre, W Nashabeh, D
Schmalzing, X Yao, M Fuchs, D Whitney, F Regnier. Purification of antibody
Fab fragments by cation-exchange chromatography and pH gradient elution. J
Chromatography 707:225–231, 1995.
66. NM Schultz, RT Kennedy. Rapid immunoassays using capillary electrophoresis
with fluorescence detection. Anal Chem 65:3161–3165, 1993.
67. NM Schultz, L Huang, RT Kennedy. Capillary electrophoresis-based immunoas-

© 2000 by Marcel Dekker, Inc.

 say to determine insulin content and insulin secretion from single Islets of Lang-
erhans. Anal Chem 67:924–929, 1995.
68. FA Chen, SL Pentoney, Jr. Characterization of digoxigenin-labeled beta-phy-
coerythrin by capillary electrophoresis with laser-induced fluorescence applica-
tion to homogeneous digoxin immunoassay. J Chromatography 680:425–430,
1994.
69. FA Chen, RA Evangelista. Feasibility studies for simultaneous immunochemical
multianalyte drug assay by capillary electrophoresis with laser-induced fluores-
cence. Clin Chem 40:1819–1822, 1994.
70. UB Sleytr, P Messner, D Pum, M Sara, eds. Immobilised Macromolecules: Ap-
plication Potentials. Heidelberg: Springer-Verlag, 1993.
71. RF Taylor, ed. Protein Immobilization: Fundamentals and Applications. New
York: Marcel Dekker, 1991.
72. J Turkova, I Vins, MJ Benes, Z Kucerova. Review: Oriented immobilization of
enzymes and other biologically active proteins through their antibodies. Int J
Biol Chem 1:1–15, 1994.
73. SS Wong. Chemistry of Protein Conjugation and Cross-Linking. Boca Raton,
FL:CRC Press, 1993.
74. FA Chen. Characterization of charge-modified and fluorescein- labeled antibody
by capillary electrophoresis using laser-induced fluorescence. Application to im-
munoassay of low level immunoglobulin A. J Chromatography 680:419–423,
1994.
75. FA Chen, JC Sternberg. Characterization of proteins by capillary electrophoresis
in fused-silica columns: review on serum protein analysis and application to im-
munoassays. Electrophoresis 15:13–21, 1994.
76. O Reif, R Lausch, T Scheper, R Freitag. Fluorescein isothiocyanate-labeled Pro-
tein G as an affinity ligand in affinity/immunocapillary electrophoresis with flu-
orescence detection. Anal Chem 66:4027–4033, 1994.
77. TM Phillips, JJ Chmielinska. Immunoaffinity capillary electrophoretic analysis
of cyclosporin in tears. Biomed Chromatography 8:242–246, 1994.
78. BY Cho, R Strong, H Zou, DH Fisher, J Nappier, IS Krull. Immunochromato-
graphic analysis of bovine growth hormone releasing factor involving RP-HPLC-
immunodetection. J Chromatography 743:181–194, 1996.
79. (a) LJ Janis, FE Regnier. Immunological-chromatographic analysis. J Chroma-
tography 444:1–11, 1988; (b) LJ Janis, FE Regnier. Dual-column immunoas-
says using protein G affinity chromatography. Anal Chem 61:1901–1906,
1989.
80. CL Flurer, M Novotny. Dual microcolumn immunoaffinity liquid chromatogra-
phy: an analytical application to human plasma proteins. Anal Chem 65:817–
821, 1993.
81. M de Frutos, FE Regnier. Tandem chromatographic-immunological analyses.
Anal Chem 65:17A–25A, 1993.
82. M de Frutos. Chromatography-immunology coupling, a powerful tool for envi-
ronmental analysis. Trends Anal Chem 14:133–140, 1995.

© 2000 by Marcel Dekker, Inc.

83. A Farjam, GJ De Jong, RW Frei, UA Brinkman, W Haasnoot, ARM Hamers,
R Schilt, FA Huf. Immunoaffinity pre-column for selective on-line sample pre-
treatment in high-performance liquid chromatography determination of 9-nortes-
tosterone. J Chromatography 452:419–433, 1988.
84. A Farjam, AE Brugman, A Soldaat, P Timmerman, H Lingeman, GJ De Jong,
RW Frei, UAT Brinkman. Immunoaffinity precolumn for selective sample pre-
treatment in column liquid chromatography: immunoselective desorption. Chro-
matographia 31:469–477, 1991.
85. RE Strong, BY Cho, D Fisher, JL Nappier, IS Krull. Immunodetection ap-
proaches and high performance immunoaffinity chromatography for an analog
of bovine growth hormone releasing factor at trace levels. Biomed Chromatogra-
phy 10:337–345, 1996.
86. (a) YH Chu, LZ Avila, HA Biebuyck, GM Whitesides. Using affinity capillary
electrophoresis to identify the peptide in a peptide library that binds most tightly
to vancomycin. J Org Chem 58:648–652, 1993; (b) LZ Avila, YH Chu, EC
Blossey, GM Whitesides. Use of affinity capillary electrophoresis to determine
kinetic and equilibrium constants for binding of arylsulfonamides to bovine car-
bonic anhydrase. J Med Chem 36:126–133, 1993.
87. YH Chu, DP Kirby, BL Karger. Free solution identification of candidate peptides
from combinatorial libraries by affinity capillary electrophoresis/mass spectrom-
etry. J Am Chem Soc 117:5419–5420, 1995.
88. YH Chu, YM Dunayevskiy, DP Kirby, P Vouros, BL Karger. Affinity capillary
electrophoresis-mass spectrometry for screening combinatorial libraries. J Am
Chem Soc 118:7827–7835, 1996.
89. JA Boutin, P Hennig, PH Lambert, S Bertin, L Petit, JP Mahieu, B Serkiz, JP
Volland, JL Fauchere. Combinatorial peptide libraries: robotic synthesis and
analysis by nuclear magnetic resonance, mass spectrometry, tandem mass spec-
trometry, and high-performance capillary electrophoresis techniques. Anal Bio-
chem 234:126–141, 1996.
90. G Jung, H Hofstetter, S Feiertag, D Stoll, O Hofstetter, KH Wiesmuller, V
Schurig. Cyclopeptide libraries as new chiral selectors in capillary electrophore-
sis. Angew Chem Int Ed Eng, 35:2148–2150, 1996.
91. HK Associates, Inc., Cambridge, MA, 1996–1997.
92. YM Dunayevskiy, P Vouros, EA Winter, GW Shipps, T Carell, J Rebek Jr. Ap-
plication of capillary electrophoresis-electrospray ionization mass spectrometry
in the determination of molecular diversity. Proc Natl Acad Sci USA 93:6152–
6157, 1996.
93. YM Dunayevskiy, P Vouros, T Carell, EA Wintner, J Rebek Jr. Characterization
of the complexity of small-molecule libraries by electrospray ionization mass
spectrometry. Anal Chem 67:1100–1104, 1995.

© 2000 by Marcel Dekker, Inc.

7
Finding a Needle in a Haystack:
Information Management for
High-Throughput Synthesis of Small
Organic Molecules

David Nickell
Parke-Davis Pharmaceutical Research
Ann Arbor, Michigan

Unless it produces action, information is overhead

from an interview with F. Dressler in the Wall Street Journal,
May 9, 1997

I. INTRODUCTION: WHAT IS A HAYSTACK?

The process of drug discovery is like searching for a needle in a haystack.

Many compounds must be tested before a single marketable drug is identified.
Sorting through the information generated by the discovery process is also
analogous to finding a needle in a haystack. With the appropriate tools, we
can improve our chance of finding the needle. It is the purpose of this paper to
enhance the reader’s knowledge regarding the issues surrounding information
management for high-throughput organic synthesis and to describe a hypothet-
ical information management system.
Information and the generation of knowledge based on it are integral to
the research and development process in the pharmaceutical industry. Without
effective management of our information resources, these assets will have

© 2000 by Marcel Dekker, Inc.

little more value than the background noise of everyday life. Because of the
introduction of new technology (e.g., high-throughput screening and organic
synthesis) to the drug discovery process, we must adapt our current informa-
tion management systems.
New drug development in the pharmaceutical industry is undergoing
rapid evolution. Recent data indicate that the average time from discovery to
marketing approval has increased to 13–15 years (1). Because of the dropout
rate for new drug candidates, many thousands of compounds must be evalu-
ated before approval is given to market one new chemical entity. New technol-
ogies are being introduced in the pharmaceutical industry to bring new drugs
to market in a more cost-effective and rapid manner.
The introduction of high-throughput screening (HTS) technology in the
early 1990s (2) forced the pharmaceutical industry to reevaluate much of its
standard practices. Researchers could now evaluate thousands or even hun-
dreds of thousands of compounds per month. Because of the large amounts
of data produced by HTS, in-place information management systems were not
adequate for storage or retrieval of the data for analysis. A cottage industry
grew from the variety of systems developed to manipulate and store the data
from HTS.
However, the high capacity of HTS uncovered a second limiting process.
Medicinal chemists were not able to generate compounds at a pace that could
utilize the capacity of HTS. The development of combinatorial chemistry tech-
nology and its introduction into the modern medicinal chemistry laboratory
gave the practicing medicinal chemist the opportunity to meet this challenge.
Combinatorial chemistry is a term that has come to mean the utilization
of novel chemistry, reaction equipment, automation, and advanced informa-
tion systems to increase productivity and efficiency of synthetic chemistry (3).
It is also generically referred to as high-throughput organic synthesis (HTOS).
Both terms have been used to describe the disciplines of ‘‘parallel synthesis’’
whereby libraries of individual compounds are synthesized (e.g., see Refs. 4
and 5) and that of ‘‘combinatorial synthesis’’ whereby libraries consisting of
mixtures or pools of compounds are produced by the combinatorial mixing
of chemical building blocks (e.g., see Refs. 6 and 7). A number of reviews
on combinatorial chemistry are available in the chemical literature (8–19).
Edwards has reviewed the combinatorial chemistry alliances in the 1990s (20).
A compendium of solid phase chemistry publications has been assembled by
James (21).
High-throughput organic synthesis is a marriage of science and technol-
ogy. It requires the collaboration of many disciplines including chemistry,
robotics, and information management technology. Automation of HTOS

© 2000 by Marcel Dekker, Inc.

must be approached from a high-level view. Like retrosynthetic analysis in
which synthetic pathways to a target molecule are identified by a series of
steps ‘‘in which a target structure can be transformed in a succession of steps
through simpler structures into a starting point for a synthesis’’ (22), analysis
of an information system for HTOS can be deconvoluted to component parts.
In the current environment, many pharmaceutical companies are approaching
the development of an information system from the reverse direction (e.g.,
buying automated synthesizers, then designing the information system). This
approach is of the ‘‘penny wise, pound foolish’’ school of thought because it
sacrifices long term system integration for short-term compound generation.
Companies who develop automated synthesis systems from the bottom up
often find themselves awash in data after running the automated synthesizers
for a few months.
Often the starting point in these analyses is how to automate existing
processes. This may not be practical and may involve a great deal of time and
resources. It is often better to rethink a traditional process from an automation
point of view. A classic example of this is the effort that many companies
have invested in the automation of a traditional aqueous workup. Mixing of
biphasic mixtures has led to many problems. A better solution, in this case
developed by rethinking the problem, is to use solid phase extraction (SPE)
technology.
As described by Peccoud, ‘‘automation of molecular biology spans at
least three levels—automation of instruments, automation of experiments and
automation of the laboratory’’ (23). The same statement can be made about
HTOS. Many of the systems in development today address only the automa-
tion of the instruments. Only a few companies are taking the next step to
automate chemical experiments. The integration of all instruments used in the
experiments with centralized data and control systems will form the basis for
future automated HTOS laboratories.
HTOS experiments are unique in that they are automation-limited. Be-
cause the reactions cannot proceed more quickly than the robotics can deliver
reagents or transfer reagents to reaction stations, the limiting factor becomes
the automation. This is an unexpected attribute of the automated process as
common wisdom dictates that automation will facilitate a process rather than
be a rate-limiting step. It should be noted that automation will not necessarily
increase the speed of the synthetic process. A well-trained chemist can often
exceed the rate at which a liquid-handling robot can prepare reagents or reac-
tions. The primary advantages of the use of robotics in organic synthesis are
improved accuracy and liberating a chemist from repetitive procedures.
The pharmaceutical industry is a knowledge-based endeavor. In fact, it

© 2000 by Marcel Dekker, Inc.

can be said the product of pharmaceutical research and development is infor-
mation and the knowledge gained from it. Information derived from different
types of bioassays and chemical entities is used to make decisions about which
compounds will be advanced along the drug development pathway. However,
information is only as valuable as the increase in the knowledge it produces.
Knowledge, being defined as information and the understanding inferred from
the relationships between pieces of information, is gained from the use of
information (e.g., interpretation). The amount of knowledge gained can be
described as a function of the available hardware, the software used to store
and interpret the data, and data accessibility (in other words, how available
is the information when it is needed?). Because the value of information is
time-dependent (i.e., it has its maximum value in a finite window of opportu-
nity) it must be readily accessible. In recent years, a significant amount of
resources have been directed in the industry to ‘‘manage’’ this information
as it has been recognized that the knowledge generated by individual pharma-
ceutical companies is an important corporate resource.
Like many aspects of the drug discovery process, HTOS is information-
intensive. The focus of HTOS is the conversion of a set of reagents (building
blocks) to a set of products. A traditional chemical synthesis, in which only
one compound is synthesized at a time in a serial manner, also produces and
utilizes information. A chemist would record the list of reagents to be used
and the amounts required, how the reagents are to be combined, and a list of
the desired products. Along with information about performing the synthesis,
the chemist will record data about the chemical analysis and structure of the
products. For several hundred years the traditional method of recording this
information was to painstakingly describe each experiment in a laboratory
notebook. While this process required time and energy on the part of the chem-
ist, it was not an impossible burden. With the introduction of HTOS, the man-
ual documentation of a chemical synthesis is a very daunting if not impossible
task. For example, let us consider a five-step synthesis performed using tradi-
tional techniques. Each step of the synthetic process would typically be re-
corded on an individual page in a laboratory notebook and would require four
to six manual calculations. If we now perform an HTOS experiment in which
96 compounds are to be synthesized the volume of information to be recorded
balloons rapidly. A total of 480 pages would be required to record this infor-
mation along with 384–576 calculations! As one can see from this simple
example, a chemist performing HTOS experiments would soon be over-
whelmed by the amount of information necessary for each HTOS experiment.
To develop an automated system to relieve the chemist of this information

© 2000 by Marcel Dekker, Inc.

management burden, we must first examine the synthetic process in general
as any HTOS information management system must model the work flow for
the synthetic process.

II. BUILDING A HAYSTACK: THE SYNTHETIC PROCESS

Any chemical synthesis can be resolved into three basic processes as shown
in Table 1. The information associated with each process must be captured
and stored in an automated procedure.
The design process is a melding of target selection with synthetic meth-
odology. For a traditional synthesis, the potential targets are only limited by
the chemist’s skill. Because the glassware used for traditional syntheses is of
modular design and can be built into a large number of configurations, the
reactor design generally does not impose any restrictions on the reactions that
can be run. Thus the chemist can select any method for synthesis of the desired
targets.
HTOS, however, introduces a new set of restrictions on synthetic meth-
odology. The commercial availability of chemical building blocks becomes a
major factor in library design. Due to the long lead times for custom-synthe-
sized building blocks, the synthesis of compound libraries for the drug discov-
ery are often restricted to commercially available starting materials. Availabil-
ity of building blocks also impact the selection of the chemistry used to
synthesize the final products. The synthetic methodologies are restricted to
those chemical reactions that use the available building blocks. The synthetic
transformations must also be selected with regard to compatibility of the inter-
mediate and final products with the conditions used in the reaction sequence.
In general, regardless of a manual or automated operation, chemical transfor-
mations should be selected that provide the best chance of success (e.g., high

Table 1 Basic Process of a Chemical Synthesis

Process Activities
Design Selecting compounds to make and how to synthe-
size them
Creation Synthesizing and purifying the products
Validation Confirming that you actually made what you in-
tended

© 2000 by Marcel Dekker, Inc.

yield, limited reaction steps, solution or solid phase techniques, easy to handle
reagents, common solvents, etc.).
Library design is also influenced by the purpose of the library to be
synthesized. For lead generation studies a diverse set of compounds is desired,
whereas for lead optimization studies, the library should represent compounds
that will provide a representational structure–activity relationship (SAR) for
the structural scaffold under investigation. For compounds to be used in HTS,
members of the library should be chosen that will provide the maximum infor-
mation content when evaluated in the screening system.
During the HTOS design process, it is important to capture the informa-
tion used to make the choices of compounds to be included in the library.
These decision points can often be used to influence the selection of the com-
pounds to be included in the next-generation library. Along with the informa-
tion about the decision making process, a representation of the compounds to
be included in the library must be generated. This may be a fully enumerated
structure set (useful for small libraries) or generic representations of libraries
such as Markush structures. When properly stored, these virtual libraries can
be used for compound searching and registration.
The creation of the target compounds is the next process in a chemical
synthesis. In a traditional synthesis, a chemist would assemble the synthesis
apparatus, charge a round-bottomed flask with the necessary reagents, adjust
the temperature of the reaction then allow it to proceed for an appropriate
period of time. When the reaction is complete, the chemist would work up
the reaction mixture and purify the crude reaction product if necessary. The
procedure followed would be recorded in a laboratory notebook along with
any purification strategy.
While an HTOS experiment and a traditional synthesis have much in
common in the creation phase, there are some features that are unique to
HTOS. For example, because HTOS libraries are spatially oriented (i.e., each
compound or discrete mixture of compounds has a specific location in the
reaction block), locations for each member of the library must be determined
and tracked throughout the process. Pirrung and Chen (24) and Deprez et al.
(25) have used ‘‘indexed’’ combinatorial libraries, in which product mixtures
are tested and their activities used as indices to the rows or columns of a two-
dimensional matrix reflecting the activities of individual compounds. Unless
special tagging techniques are utilized, e.g., radiofrequency encoding (26,27),
chemical tagging (28), etc. (29), the locations of the final products in the li-
brary array must be accurately tracked by an information management system.
Tracking reagents to be used as well as the products to be made in each step

© 2000 by Marcel Dekker, Inc.

of an HTOS experiment is often the primary objective of the software used
to drive automated synthesizers.
The physical format of the compound library must also be determined
before starting an HTOS experiment. The decision as to whether to synthesize
mixtures of compounds or discrete chemical entities is related to the purpose
for synthesizing the library. Libraries of mixtures allow for the screening of
more compounds than discrete libraries, but large numbers of compounds per
pool can adversely affect the detection of active compounds. The testing of
mixtures also introduces the possibility of signal-to-noise ratio deterioration
or the introduction of false positives (30). Within the pharmaceutical industry
today, the trend is to produce smaller libraries of well-defined (discrete) com-
pounds using parallel synthesis techniques (31).
Validation of the products of a synthetic sequence has historically been
the crown jewel in the synthetic process. Traditionally, a chemist would rigor-
ously attempt to produce the purest compounds in the highest yield before
submitting them for biological testing. Validation technologies have advanced
from the use of melting points as an indicator of purity to sophisticated spectro-
graphic techniques in use today. All of these techniques had one aspect in
common, i.e., they all relied on examining the compounds as neat compounds
or solutions of free compounds. Advances in analytical technology not only
allowed analyses of smaller amounts of compounds but also produced greater
amounts of information. Even for the medicinal chemist using traditional
chemical techniques, information overload is becoming a problem. It is not
uncommon to find in the offices of today’s chemist stacks of notebooks of
analytical data corresponding to the compounds he or she has synthesized.
The adoption of the solid phase synthesis technology by practicing me-
dicinal chemists has forced a reevaluation of reaction and compound valida-
tion. While traditional analytical techniques can be used for solution phase
chemistry (32), there is a lack of analytical support for solid phase chemistry
(33). Techniques for analyzing small molecules produced on solid supports
while still attached to the resin include microanalysis (34), infrared (IR) spec-
troscopy (35–38), nuclear magnetic resonance (NMR) spectroscopy (39–66),
and color detection reagents for reactive functional groups (67,68). Solid phase
syntheses have also been analyzed by cleavage of a small sample from the
support followed by high-performance liquid chromatography (HPLC), gas
chromatography (GC), thin-layer chromatography (TLC), or quantitative
NMR (69) analysis or monitoring the reaction solution for decrease in reagents
or generation of byproducts (38). Minimal amounts of sample cleaved from
the resin can be analyzed by matrix-assisted laser description ionization

© 2000 by Marcel Dekker, Inc.

(MALDI)–time-of-flight (TOF) MS (70), although the volume of data gener-
ated for a library by this technique can require gigabytes of disk storage space.
Validation of the compounds in HTOS libraries has become an area that
is receiving more attention. Because of the movement away from pools of
compounds (mixtures) to discrete compounds for SAR development, quality
control and quality assurance of the samples has become an issue. The issue
of quality vs. quantity in HTOS has been discussed in a recent paper by Mac-
Donald et al. (71). Bauer has described an information management system
that incorporates several quality checks on the data generated by HTOS sys-
tems (72).

III. FINDING THE NEEDLE: A MODEL HTOS INFORMATION

MANAGEMENT SYSTEM

The work flow of the HTOS process must be the guiding principle when de-
signing a system to manage HTOS information (73). The system must be
designed for maximum flexibility to allow for integration of new functionality.
Maximum flexibility is provided by incorporating a system architecture that
is both open and modular. An open system provides ‘‘hooks’’ to the outside
world. An HTOS information management system must be able to communi-
cate to programs outside its scope using standard protocols. Providing connec-
tivity to the outside world also provides a foundation for the second corner-
stone of an information management system: modularity.
An open system with modular architecture is the model followed by
desktop computer hardware and software designers. It is now possible to as-
semble a PC entirely from components manufactured by different vendors.
Object-oriented programming techniques also allow for the use of modularized
‘‘components,’’ thus saving the programmer many hours of labor. When mod-
ularity is designed into an HTOS information management system, it can be
easily expanded to accommodate new processes and equipment. It is a safe
assumption that chemists will want to assemble an HTOS system based on
the best available technology. This means that the design module may be pur-
chased from vendor A while the robot is purchased from vendor B. If these
two components cannot be tightly integrated, one of the suppliers will lose a
sale. Open and modular systems will also allow the system to grow as needed.
Much like the traditional laboratory glassware commonly used today, the
chemists will assemble the components that are needed, replacing those that
are obsolete. As with desktop computers, proprietary closed HTOS systems
are only an intermediary design phase on the road to open modular systems.

© 2000 by Marcel Dekker, Inc.

 An HTOS information management system must incorporate modules
that can be used to design products and synthetic routes, manage the data
necessary to perform the actual synthesis, and interface with analytical instru-
ments to provide bidirectional information transfer. Although not part of the
synthetic process, the information management system must also provide an
automatic mechanism to register the new synthetic targets into compound reg-
istration systems. Manual registration will not be possible due the increase in
the number of compounds per chemist that will result from the widespread
adoption of HTOS techniques.
Figure 1 represents a schematic diagram for a model HTOS information
management system. It comprises a series of replaceable modules that reflect

Figure 1 Schematic of HTOS information management system.

© 2000 by Marcel Dekker, Inc.

the work flow of the synthetic process. The reaction workbook is the central
‘‘hub’’ in which data about the experiments are collected. It is also the primary
interface to the system for the chemist. Each module will be described in the
following sections.
A design module will provide the capability to select the compounds
and the method by which they will be synthesized. In order to perform this
task, the design module must be linked to external databases that can provide
data about the availability of chemical reagents and information about syn-
thetic methods. Once the products are selected and the synthetic method se-
lected, this information will be collated in the reaction workbook. The function
of the design module is complementary to diversity assessment tools.
The best analogy for a reaction workbook is the typical laboratory note-
book but modified for HTOS so that it can easily manipulate the data for
multiple reactions. A reaction workbook should provide, at a minimum, the
functions shown in Table 2.
A reaction workbook module should function in either a standalone
desktop environment or connected to an enterprise-wide information system.
It must be an ‘‘intelligent’’ system to prevent the chemists from making mis-
takes that because of the repetitive use of data can cause cascading errors in
reagent delivery and process timing. Using the information input into the sys-
tem, it will automatically perform the necessary calculations such as the vol-
ume of reagents to be transferred and the total amounts of neat reagents needed
to prepare these solutions.
Product and reagent location tracking is one of the most important func-
tions of a reaction workbook due to the spatial orientation of the product arrays
generated by HTOS techniques. The locations of reagents together with the
volumes to be transferred and the sequence in which the reagents are to be
delivered will be used to automatically create synthetic procedures for a liquid-

Table 2 Reaction Workbook Functionality

Perform necessary synthetic calculations

Track the locations of the products and the reagents
Associate the reagents with the products
Enumerate the products based on the available reagents
Store the synthetic procedures
Provide operational instructions for the associated
automation
Collect analytical data on the products
Correlate screening data with products

© 2000 by Marcel Dekker, Inc.

handling robot. An ‘‘outline’’ for the synthetic procedure must be entered
into the system by the chemist or downloaded from a database of synthetic
procedures. Typically, the outline of the procedure will be entered by the
chemist in the form of commands like ‘‘add the generic building block A.’’
The reaction workbook module will first calculate the volumes of reagents to
be delivered and determine to which locations they should be dispensed. It
will then create the necessary liquid-handling robot commands based on the
locations of each of the building blocks. When complete, the list of action
steps in each individual procedure will correspond to the physical steps a
chemist would perform if he carried out the synthesis manually. These ‘‘reci-
pes’’ can be stored for future reference and use. The reaction workbook mod-
ule will use the recipes to generate robot-specific commands for the automa-
tion associated with the information management system. These synthetic
procedures for the automated synthesizers may also become the ‘‘documenta-
tion’’ for the HTOS libraries. These procedures document what a chemist was
attempting to make and how he or she tried to make it. For larger libraries,
this method may be the only practical way to document what was made as
enumeration of every chemical structure in a large library would require the
dedication of significant computer resources.
A reaction workbook module will also become the primary vehicle for
compound registration because of its links to corporate compound registration
systems. To facilitate the registration of chemical structures, a reaction work-
book must be ‘‘chemically aware.’’ In other words, it should recognize chemi-
cal structures and be able to derive information (e.g., molecular weight,
R-group tables, etc.) from them. Ideally the reaction workbook should be ‘‘re-
action’’-based. The chemist should be able to specify the desired chemical
transformation using chemical reaction notation, specify a list of reagents to
be used, and then automatically enumerate the individual products. A reaction
workbook performs a yeoman’s service for the chemist performing an HTOS
experiment. It dutifully records and tracks the information used in the synthe-
sis. While performing many behind-the-scenes calculations and data manipu-
lations, it functions on the very simple assumption that the chemist has previ-
ously selected the reagents to be used in the design module and thus all
possible compounds will be prepared from the list of selected building blocks.
No facility need be incorporated to allow for selection of individual molecules
or sublibraries if the libraries are small. If, on the other hand, the libraries are
large, a design module should be used to create sublibraries that are of a size
that can be produced on an automated synthesizer.
The control of the system automation is the function of the robot control
module. This module will not only control the liquid-handling robots but will

© 2000 by Marcel Dekker, Inc.

also schedule operations for multiple workstations (e.g., weighing, multiple
reaction stations, etc.). It will also interface with reaction history databases.
These databases consist of two components: (a) robot commands to perform
the desired synthesis; (b) a transaction log. The log database will be useful
in determining if the process was successfully completed or for diagnosing
the cause of any system failures. Because there is not a standard communica-
tion protocol that can be used to issue commands and receive feedback from
robotic peripherals, it is difficult to use a standard control system to control
the automation. Some companies that are developing automation systems are
investigating the integration of LabVIEW (74) as robot control software. How-
ever, within the manufacturing industries there are a group of real-time ma-
chine and process control vendors (75) whose software might be adaptable
for HTOS automation control.
After the library has been synthesized, the product analysis must be
completed to validate the library members. During the development of HTOS
technologies, two philosophies have developed regarding the characterization
of products. One view is that the components of a library need not be analyzed.
The compounds are synthesized, then screened for activity in a bioassay. If
a sample shows activity, it can be resynthesized and fully characterized using
traditional techniques. This methodology provides the most effective way to
screen large numbers of compounds.
With the movement toward smaller libraries of discrete compounds, li-
brary validation has become a more important issue. To maximize the informa-
tion content of every HTOS experiment, each compound in a library must be
analyzed for chemical structure and purity. Many of the analytical instruments
available today are designed to function in a serial mode. Thus the instrument
must acquire information about a single compound and then move on to the
next, one at a time. The analysis of HTOS libraries is perhaps the weak link
in the high-throughput discovery process. While screening and chemical syn-
thesis have adopted parallel paradigms to enhance the speed of the processes,
purification and analysis techniques are still serially oriented. This is an area
where future parallel automated technology will have a significant impact.
To move data effectively, bidirectional interface between the analytical
instruments and the reaction workbook module is needed to provide a mecha-
nism to submit the libraries for analysis and to integrate the resulting data
with the synthesis data. The data could be transferred and collected in the
reaction workbook module where a chemist can review the data for integrity
and forward it to corporate data repositories. A fully automated system that can
review the analytical data and identify the library members that pass selection
criteria would greatly enhance the throughput of the validation system.

© 2000 by Marcel Dekker, Inc.

Product registration has also become an issue in HTOS. For those companies
who wish to register each sample produced in an HTOS experiment, a chemi-
cal structure along with a unique sample identifier must be generated for each
sample. The samples must be associated with a particular plate or synthesizer
as well as the sample’s location within the array. The entire HTOS experiment
must also be given a unique identifier. The compounds must also be identified
as having been generated from an HTOS experiment. This is important be-
cause the samples are registered with proposed structures as the chemist truly
has no idea if the synthesis produced the desired compounds. All of these
‘‘tags’’ are necessary so that an individual sample can be uniquely identified.
For large libraries, this approach may not be practical.
Many groups register only the compounds from an HTOS experiment
that are bioactive. After the samples are fully characterized, they are registered
in a traditional manner. Because the ‘‘hit’’ rate from HTS is typically less
than 1%, this method does not seriously overload in-place registration systems.
The disadvantage to this method is that the information that a compound has
been synthesized is lost. Not having access to that information can result in
the repetition of experiments.
Whichever method is selected for registering HTOS samples, the pro-
cess should be as automated as possible. Because the reaction workbook mod-
ule will contain all the information necessary to register a library, a compound
registration mechanism should be incorporated within it.

IV. BETTER METHODS FOR FINDING THE NEEDLE: THE

FUTURE

For the near future, fully automated systems are not practical. Although work
is proceeding on fully integrated high-throughput drug discovery systems (76),
none has yet reached the marketplace because the decision-making algorithms
and software are not yet in place. Effective integration between information
systems and automation also needs to be developed. For the time being, the
medicinal chemist will provide the decision making ‘‘brains’’ to analyze
the data generated by HTOS and HTS systems and the integration between
the components. This human–machine interface will provide the ultimate in
modularity as different chemists can interact with the system in different ways
depending on their scientific background and knowledge of the drug discovery
process.
Although the volume of data generated by HTOS experiments can easily
overwhelm a poorly designed information management system, it is not the

© 2000 by Marcel Dekker, Inc.

only problem one needs to resolve. As mentioned earlier in this chapter, the
value of information is derived from the knowledge gained from it. To improve
the knowledge gained from the HTOS experiments, appropriate tools must
be available to analyze the experimental data. To date, these tools are not
available. For this reason, HTOS technology is not being used to its fullest
potential.
Well-designed data analysis tools must incorporate three principle com-
ponents: (a) the tools must have the ability to handle large data-sets; (b) the
tools must be capable of representing the data using a visual paradigm; (c)
the tools must incorporate algorithms that can derive relationships between
different types of data. Typically, scientists are still using the tools that were
developed 20 years ago. These tools (e.g., spread sheets) were adequate for
the interpreting of small numbers of data but are not up to the job when analyz-
ing billions of data points. Such tools must be able to visualize (77) and com-
pare the MS data or the NMR data for all of the components of a library in
an easy-to-comprehend manner. Analysis tools must also be able to visually
represent different types of data in a way that will facilitate its interpretation.
For example, how can the calculated log P’s of a series of compounds pro-
duced by parallel synthetic techniques be represented? How can the HTS re-
sults be represented and be compared to the substitution pattern at a particular
variable site of a synthetic scaffold? These are the sort of problems with which
medicinal chemists are currently struggling by manual methods. Although
commercial software developers are beginning to address this issue (78), there
is no solution available that is fully integrated with an HTOS information
management system.
HTOS technologies have forced the practicing chemist to reevaluate
how he or she does business. While a number of interesting puzzle pieces
have been identified, it is not clear as to how they fit together. This is one of
the most exciting challenges facing us today and in the near future.

REFERENCES

1. JA DiMasi. Trends in drug development costs, times, and risks. Drug Info 29:
375–384, 1995.
2. M Reichman, AL Harris. Practical high-throughput screening. In: W Moos,
M Pavia, B Kay, A Ellington, eds. Annual Reports in Combinatorial Chemistry
and Molecular Diversity, Vol. 1. Leiden, The Netherlands: ESCOM, 1997, pp
273–286.

© 2000 by Marcel Dekker, Inc.

 3. SH DeWitt, AW Czarnik. Automated synthesis and combinatorial chemistry.
Curr Op Biotech 6:640–644, 1995.
4. SH DeWitt, JS Kiely, CJ Stankovic, MC Schroeder, DMR Cody, MR Pavia.
DIVERSOMERS: an approach to non-peptide non-oligomeric chemical diver-
sity. Proc Natl Acad Sci USA 90:6909–6913, 1993.
5. BA Bunin, JA Ellman. A general and expedient method for the solid phase syn-
thesis of 1,4-benzodiazepine deviates. J Am Chem Soc 114:10997–10998,
1992.
6. M Pirrung, J Chau, J Chen. Discovery of a novel tetrahydroacridine acetylcholn-
esterase inhibitor through an indexed combinatorial library. Chem Bio 621–626,
1995.
7. T Carell, E Wintner, A Sutherland, J Rebek, Y Dunayevskiy, P Vouros. New
promise in combinatorial chemistry: synthesis, characterization, and screening
of small-molecule libraries in solution. Cur Bio 5:171–183, 1995.
8. A number of relevant articles can be found in AW Czarnik, SH DeWitt, eds. A
Practical Guide to Combinatorial Chemistry. Washington, DC: ACS Books,
1997.
9. WA Warr. Combinatorial chemistry and molecular diversity. An overview. J
Chem Inf Comp Sci 37:134–140, 1997.
10. F Balkenhol, C von dem Dussche-Hunnefeld, A Lansky, C Zechel. Combinato-
rial synthesis of small organic molecules. Angew Chem Int Ed Eng 35:2288–
2337, 1996.
11. A number of excellent references are located in the March 1996 issue of Acc
Chem Res, which is dedicated to combinatorial chemistry.
12. A tabular collection of reactions used in solid phase organic synthesis can be
found in P Hermkens, H Ollenheijm, D Rees. Solid-phase organic reactions: a
review of the current literature. Tetrahedron 52:4257–4554, 1996.
13. Thompson, JA Ellman. Synthesis and applications of small molecule libraries.
Chem Rev 96(1):555–600, 1996.
14. G Lowe. Combinatorial chemistry. Chem Soc Rev 245:309–382, 1995.
15. NK Terrett, M Gardner, DW Gordon, RJ Kobylecki, J Steele. Combinatorial
synthesis—the design of compound libraries and their application to drug discov-
ery. Tetrahedron 51(30):8135–8173, 1995.
16. MA Gallop, RW Barrett, WJ Dower, SPA Fodor, AM Gordon. Applications of
combinatorial technologies to drug discovery 1. J Med Chem 37:1233–1251,
1994.
17. EM Gordon, RW Barrett, WJ Dower, SPA Fodor, MA Gallop. Applications of
combinatorial technologies to drug discovery 2. J Med Chem 37:1385–1401,
1994.
18. JS Fruchtel, G Jung. Organic chemistry on solid supports. Angew Chem Int Ed
Eng 35:17–42, 1996.
19. X Will, I, Pop, L Bourel, D Horvath, R Baudelle, P Melnyk, B Deprez, A Tartar.
Combinatorial chemistry: a rational approach to chemical diversity. Eur J Med
Chem 31:87–98, 1996.

© 2000 by Marcel Dekker, Inc.

20. MG Edwards. Combinatorial chemistry alliances in the 1990s: review of deal
structures. In W Moos, M Pavia, B Kay, A Ellington, eds. Annual Reports in
Combinatorial Chemistry and Molecular Diversity, Vol. 1. Leiden, The Nether-
lands: ESCOM, 1997, pp 314–320.
21. IW James. A Compendium of solid phase chemistry publications. In: W Moos,
M Pavia, B Kay, A Ellington, eds. Annual Reports in Combinatorial Chemistry
and Molecular Diversity, Vol. 1. Leiden, The Netherlands: ESCOM, 1997, pp
326–344.
22. EJ Corey, AK Long, SD Rubenstein. Computer-assisted analysis in organic syn-
thesis. Science 228:408–418, 1985.
23. J Peccoud. Automating molecular biology: a question of communication. Bio-
technology 13:741–745, 1995.
24. MC Pirrung, J Chen. Preparation and screening against acetylcholinesterase of a
non-peptide ‘‘indexed’’ combinatorial library. J Am Chem Soc 117:1240–1245,
1995.
25. B Deprez, WL Bourel, H Coste, F Hyafil, A Tartar. Orthogonal combinatorial
chemistry libraries. J Am Chem Soc 116:5405–5406, 1995.
26. KC Nicolaou, XY Xiao, Z Parandoosh, A Senyei, M Nova. Radiofrequency en-
coded combinatorial chemistry. Angew Chem Int Ed Eng 34:2289–2291, 1995.
27. EJ Moran, S Sarshar, JF Cargill, MM Shahbaz, A Lio, AMM Mjalli, RW Arm-
strong. Radio frequency tag encoded combinatorial library method for discovery
of tripeptide-substituted cinnamic acid inhibitors of the protein tyrosine phospha-
tase PTP1B. J Am Chem Soc 117:10787–10788, 1995.
28. MHJ Ohlmeyer, RN Swanson, LW Dillard, JC Reader, G Asouline, R Koba-
yashi, M Wigler, W C Still. Complex synthetic chemical libraries indexed with
molecular tags. Proc Natl Acad Sci USA 90:10922–10926, 1993.
29. For a review of current encoding technologies in use for combinatorial chemistry
see AW Czarnik. New combinatorial methods—encoding. Curr Op Chem Biol
1:60–66, 1997.
30. NK Terrett, D Bojanic, D. Brown, PJ Bungay, M Gardner, DW Gordon,
CJ Mayers, J Steele. Bioorgan Med Chem Lett 5:917–922, 1995.
31. P Meyers. Combinatorial chemistry: the road to hitsville? Pharmaceut News 3:
16–18, 1996.
32. CE Kibbey, Analytical tools for solution-phase synthesis. In: AW Czarnik,
SH DeWitt, eds. A Practical Guide to Combinatorial Chemistry. Washington,
DC: ACS Books, 1997, pp 199–248.
33. For a recent review of analytical methods used for the quality control of combina-
torial libraries, see WL Fitch. Analytical methods for the quality control of com-
binatorial libraries. In: W Moos, M Pavia, B Kay, A Ellington, eds. Annual Re-
ports in Combinatorial Chemistry and Molecular Diversity, Vol. 1. Leiden, The
Netherlands: ESCOM, 1997, pp 59–68.
34. Although a number of groups have investigated microanalysis of resin-bound
small molecules, the technique has not provided reliable results. SH DeWitt,
personal communication, 1997.

© 2000 by Marcel Dekker, Inc.

35. JR Hauske, P Dorff. A solid phase CBZ chloride equivalent—a new matrix spe-
cific linker. Tetrahedron Lett 36:1589–1592, 1995.
36. B Yan, G Kumaravel, H Anjaria, A We, RC Petter, CR Jewell Jr, JR Wareing.
Infrared spectrum of a single resin bead for real-time monitoring of solid phase
reactions. J Org Chem 60:5736–5738, 1995.
37. B Yan, G Kumaravel. Probing solid phase reactions by monitoring the IR bands
of compounds on a single ‘‘flattened’’ resin bead. Tetrahedron 55: 843–848,
1996.
38. K Russell, DC Cole, FM McLaren, DE Pivonka. Tools for combinatorial chemis-
try. In: IM Chaiken, KD Janda, eds. Molecular Diversity and Combinatorial
Chemistry: Libraries and Drug Discovery, Washington, DC: ACS Books, 1996,
pp 246–254.
39. EC Blossey, RG Cannon. Synthesis, reactions, and 13C FT NMR spectroscopy
of polymer-bound steroids. J Org Chem 55: 4664–4668, 1990.
40. C Boojamra, K Burow, L Thompson, J. Ellman. Solid-phase synthesis of 1,4-
benzodiazepine-2,5-diones. Library preparation and demonstration of synthesis
generality. J Org Chem 62:1240–1256, 1997.
41. JA Boutin, P Hennig, P-H Lambert, S Bertin, L Petit, J-P Mahieu, B Serkiz,
J-P Volland, J-L Fauchere. Combinatorial peptide libraries: robotic synthesis and
analysis by nuclear magnetic resonance, mass spectrometry, and high-perfor-
mance capillary electrophoresis techniques. Anal Biochem 234:126–141, 1996.
42. FGW Butwell, R Epton, EJ Mole, N Mazaffar, S Phillips. Deprotection studies
in ultra-high load solid gel phase peptide synthesis. 1. Carbon-13 NMR investiga-
tion of the efficacy of boron trifluoride based side-chain deprotection cocktails.
Innovation Perspect Solid Phase Synth., 1st Int Symp, SPCC UK, Birmingham,
U.K., 1990.
43. JK Chen, SL Schreiber. Combinatorial synthesis and multidimensional NMR
spectroscopy: an approach to understanding protein-ligand interactions. Angew
Chem Int Ed Eng 34:953–969, 1995.
44. L Dapremont, AM Valerio, AM Bray, KM Stewart, TJ Mason, AW Wang,
NJ Maeji. Multiple synthesis using the multipin method. Physiol Chem and Phys
Med NMR 339–343, 1995.
45. J Ellman, D Mendel, P Schultz, D Wemmer, B Volkman. Site-specific isotopic
labeling of proteins for NMR studies. J Am Chem Soc 114:7959–7961, 1992.
46. R Epton, P Goddard, KJ Ivin. Gel phase carbon-13 NMR spectroscopy as an
analytical method in solid (gel) phase peptide synthesis. Polymer 21:1367–1371,
1980.
47. W Ford, S Yacoub. A 13C NMR method to determine the origin of cross-linked
chloromethyl polystrenes used in polymer-supported synthesis. J Org Chem 46:
819–821, 1981.
48. W Ford, S Mohanraj, EC Blossey. Carbon-13 NMR spectra of polymer-support
reagents–methods and procedures. Macromol Synth 10:91–96, 1990.
49. E Giralt, J Rizo, E Pedroso. Application of gel-phase 13C-NMR to monitor solid
phase peptide synthesis. Tetrahedron 40:4141–4152, 1984.

© 2000 by Marcel Dekker, Inc.

50. MF Gordeev, DV Patel, J Wu, EM Gordon. Approaches to combinatorial synthe-
sis of heterocycles: solid phase synthesis of pyridines and pyrido [2,3-d] pyrimi-
dines. Tetrahedron Lett 37:4643–4646, 1996.
51. C Johnson, B Zhang. Solid phase synthesis of alkenes using the Horner-Wads-
worth-Emmons reaction and monitoring by gel phase P NMR. Tetrahedron Lett
36:9253–9256, 1995.
52. AJ Jones, CC Leznoff, PI Svirskaya. Characterization of organic substrates
bound to cross-linked polystyrenes by 13C NMR spectroscopy. Org Magn Reson
18:236–240, 1982.
53. WM Kazmierski, RD Ferguson, RJ Knapp, GK Lui, HI Yamamura, VJ Hruby.
Reduced peptide bond cyclic somatosatin based opioid octapeptides. Synthesis,
conformational properties and pharmacological characterization. Int J Pept Pro-
tein Res 39:401–414, 1992.
54. T Keating, R Armstrong. A remarkable two-step synthesis of diverse 1,4-benzo-
diazepine-2,5-diones using the Ugi four-component condensation. J Org Chem
61:8935–8939, 1996.
55. PA Keifer. Influence of resin structure, tether length and solvent upon the high-
resolution 1H NMR spectra of solid phase synthesis resin. J Org Chem 61:1558–
1559, 1996.
56. GC Look, CP Holmes, JP Chin, MA Gallop. Methods for combinatorial organic
synthesis: the use of fast 13C NMR analysis for gel phase reaction monitoring.
J Org Chem 59:7588–7590, 1994.
57. C Mapelli, MD Swerdloff. Monitoring of conformational and reaction events in
resin-bound peptides by 13C NMR spectroscopy in various solvents. In: Pep-
tides. Leiden, The Netherlands: ESCOM, 1995, pp 316–319.
58. S Garigipati, B Adams, L Adams, SK Sarkar. Use of spin echo magic angle
spinning 1H NMR in reaction monitoring in combinatorial organic synthesis. J
Org Chem 61:2911–2914, 1996.
59. A Meissner, P Bloch, E Humpfer, M Spraul, OW Sorensen. Reduction of inho-
mogeneous line broadening in two-dimensional high-resolution MAS NMR
spectra of molecules attached to swelled resins in solid phase synthesis. J Am
Chem Soc 119:1787–1788, 1997.
60. MM Murphy, JR Schullek, EM Gordon, MA Gallop. Combinatorial organic syn-
thesis of highly functionalized pyrrolidines: identification of a potent angiotensin
converting enzyme inhibitor from a mercaptoacyl proline library. J Am Chem
Soc 117:7029–7030, 1995.
61. GC Look, MM Murphy, DA Campbell, MA Gallop. Trimethylorthoformate: a
mild and effective dehydrating reagent for solution and solid phase imine forma-
tion. Tetredron Lett 36:2937–2940, 1995.
62. RC Anderson, JP Stokes, MJ Shapiro. Structure determination in combinatorial
chemistry: utilization of magic angle spinning HMQC and TOCSY NMR spectra
in the structure determination of Wang-bound lysine. Tetredron Lett 36: 5311–
5314, 1995.
63. RC Anderson, MA Jarema, MJ Shapiro, JP Stokes, M Ziliox. Analytical tech-

© 2000 by Marcel Dekker, Inc.

 niques in combinatorial chemistry: MAS CH correlation in solvent-swollen resin.
J Org Chem 60:2650–2651, 1995.
64. SK Sarkar, RS Garigipati, JL Adams, PA Keiffer. An NMR method to identify
nondestructively chemical compounds bound to a single solid phase synthesis
bead for combinatorial chemistry applications. J Am Chem Soc 118:2305–2306,
1996.
65. D Sherrington, P Dennison, C Snape, R Law. Quantitative solid-state 13C NMR
analysis of the functional groups present in divinylbenzene networks. Polymer
Prepr 34:558–559, 1993.
66. C Dhalluin, I Pop, B Depreze, P Melnyk, A Tartar, G Lippens. Magic-angle
spinning NMR spectroscopy of polystyrene-bound organic molecules. In:
M Chaiken, KD Janda, eds. Molecular Diversity and Combinatorial Chemistry:
Libraries and Drug Discovery. Washington, DC: ACS Books, 1996, pp 255–
272.
67. AA Virgilio, JA Ellman. Simultaneous solid phase synthesis of β-turn mimetic
incorporating side-chain functionality. J Am Chem Soc 116:11580–11581, 1994.
68. SS Chu, SH Reich. NPIT: a new reagent for quantitatively monitoring reactions
of amines in combinatorial synthesis. Bioorg Med Chem Lett 5:1053–1058,
1995.
69. SH DeWitt, MC Schroeder, CJ Stankovic, JE Strode, AW Czarnik.
DIVERSOMER technology: solid phase synthesis, automation, and integration
for the generation of chemical diversity. Drug Dev Res 33:116–124, 1994.
70. BJ Egner, GD Langley, M Bradley. Solid phase chemistry: direct monitoring
by monitoring by matrix-assisted laser desorption/ionization time of flight mass
spectrometry. A tool for combinatorial chemistry. J Org Chem 60:2652–2653,
1995.
71. AA MacDonald, DG Nickell, SH DeWitt. Combinatorial chemistry: quality vs.
quantity. Pharmaceut News 3:19–21, 1996.
72. BE Barr. Automation issues at the interface between combinatorial chemistry
and high-throughput screening. In: IM Chaiken, K D Janda, eds. Molecular Di-
versity and Combinatorial Chemistry: Libraries and Drug Discovery. Washing-
ton, DC: ACS Books, 1996, pp 233–243.
73. For an overview of the information management issues associated with combina-
torial chemistry, see SM Muskal. Information management. In: AW Czarnik,
SH DeWitt, eds. A Practical Guide to Combinatorial Chemistry. Washington,
DC: ACS Books, 1997, pp 357–397.
74. National Instruments (6504 Bridge Point Parkway, Austin, TX 78730, http:/ /
www.natinst.com).
75. For examples, see the software packages from Think and Do Software, Inc. (4750
Venture, Ann Arbor, MI 48108, http:/ /www.ThinkandDo.com) or Nematron
(5840 Interface Dr., Ann Arbor, MI 48103, http:/ /www.nematron.com.).
76. DK Agrafiotis, RF Bone, FR Salemme, RM Soll, US Patent 5,463,564, 1995.
77. New paradigms for visualization of large amounts of data are being developed by
several companies. For examples, see InXight’s series of VizControls (InXight

© 2000 by Marcel Dekker, Inc.

 Software, Inc., 3400 Hillview Ave., Palo Alto, CA 94304, http:/ /www.InXight.
com) or Perspecta’s Smart Content System (Perspecta, Inc., 600 Townsend St.,
Suite 170E, San Francisco, CA 94103, http:/ /www.perspecta.com).
78. Oxford Molecular Group Oxford, U.K. recently introduced DIVA (a visualiza-
tion and analysis tool for chemical and biological data) at the 210th ACS national
meeting in San Francisco.

© 2000 by Marcel Dekker, Inc.

8
Bioanalytical Screening
Methodologies for Accelerated Lead
Generation and Optimization in
Drug Discovery

James N. Kyranos and Stewart D. Chipman

ArQule, Inc.
Medford, Massachusetts

I. HISTORICAL PERSPECTIVE

The testing of small, synthetic organic molecules for their ability to modify
the biological activity of enzymes, receptor, etc., for use as human therapeutics
is the process typically referred to as drug discovery. This process has evolved
significantly over the last several decades with the advent of increasingly so-
phisticated biological and chemical methodologies, as well as laboratory scale
automation. In the first half of the twentieth century, drug discovery was pri-
marily performed by chemists and pharmacologists. The pharmacologist’s tool
of choice was in vivo bioassays to test mixtures of compounds isolated from
natural product sources by natural product chemists or small organic com-
pounds that had been synthesized individually by organic chemists. The real-
ization that protein–protein and protein–small molecule interactions transmit-
ted information in cellular biochemical pathways was the seminal observation
that led to the ‘‘lock-and-key’’ hypothesis of biomolecular interactions. This
concept that the interaction of specific shapes, charges, etc., on biological mol-
ecules can serve to control cellular and organism metabolism is now well

© 2000 by Marcel Dekker, Inc.

established. The discovery of small organic molecules that antagonize or ago-
nize biochemical interactions relevant to disease processes is the goal of scien-
tists engaged in modern drug discovery. Further advances in protein purifica-
tion and structural analysis has led to an even greater understanding of these
molecular recognition elements which, along with advances in computer-aided
design software/hardware, has lead to the development of rational design of
small organic molecules for synthesis and testing.
The first biochemical assays were developed using isolated cells, cellular
extracts or purified proteins and were conducted in glass test tubes. The contin-
uous incremental improvement of the test tube format for bioassay has lead
to the 96-well-plate standard currently favored by most scientists. In this chap-
ter we will review the current status of bioassays used in modern drug discov-
ery and how these assays have been automated to accelerate the pace of drug
discovery. We will also discuss how high-throughput screening (HTS) has
driven the need for varying types of high-throughput organic synthesis
(HTOS) and how HTS and HTOS are being integrated into integrated indus-
trial processes to accelerate the discovery of human therapeutics. Finally, we
will discuss new methodologies that address the question of how scientists
will test the large quantity of chemical compounds being generated by combi-
natorial chemistry approaches against the enormous volume of therapeutically
relevant targets being discovered by the discipline of genomics and proteo-
mics.

II. INTRODUCTION

Drugs are chemical compounds that modulate the activity of proteins and other
targets associated with a disease state to achieve a desired therapeutic re-
sponse. The discovery and development of drugs has traditionally been an
inefficient and expensive process. Recent studies have shown that time frames
of 10–15 years from the discovery of a validated molecular target to the market
introduction of a Food and Drug Administration (FDA)–approved drug are
typical, with the average cost estimated to be in excess of $300 million (1).
The traditional drug discovery process is set in motion by the realization
of an unmet therapeutic need (Fig. 1). The first major step in the process is
the identification of one or more molecular targets whose role in the patho-
physiology of the disease process has been established. The discovery of novel
genes in the human genome has been greatly accelerated by gene sequencing
and sequence analysis. However, the role of many of these genes remains
ambiguous, limiting their utility as drug discovery targets. Differential expres-
sion analysis of diseases versus normal tissues has provided more insight into

© 2000 by Marcel Dekker, Inc.

Figure 1 Traditional drug discovery process.

the role of novel molecular targets in human disease (2), but finding validated
targets will remain one of the rate-limiting steps in the process for some time.
Once a suitable drug discovery target (i.e., enzyme, receptor, etc.) has
been identified, the second step is to establish an efficient biological test or
‘‘screen’’ against which natural product extracts or small molecules can be
assayed for agonism or antagonism. Typically these bioassays can take many
forms, mirroring the diversity of the function of the biological targets them-
selves. In Sec. III we will describe several representative types of assays that
have become popular in recent years. Once a suitable bioassay has been devel-
oped natural product extracts and small organic molecules are screened for
activity against the target in order to generate biologically active structures
or ‘‘hits.’’
Many automated robotic devices have been developed to accelerate the
screening process, frequently referred to as high-throughput screening, or
HTS. A discussion of automated robotic assay systems available will be found
in Sec. IV. The advent of HTS systems with capacities approaching 10 5 com-
pounds per day has driven the need for high-throughput organic synthesis
(HTOS) and analysis. Further, the development of combinatorial chemistry
and automated synthesis/analysis has necessitated strategies for integration of
HTOS and HTS to maximize efficiency. The integration of these discovery
technologies, and others, for accelerated lead generation and optimization will
be discussed in Sec. V, with equal emphasis on technical developments and
resource management.

© 2000 by Marcel Dekker, Inc.

 The next major step in the process is lead optimization. Hits are opti-
mized for potency against the primary therapeutic target, specificity against
related and unrelated biological targets, and cellular toxicity/permeability.
Those compounds with suitable characteristics for a potential drug are referred
to as lead compounds. Typically during optimization a wider range of assay
types are utilized to evaluate a smaller set of chemical compounds; this is an
iterative design, synthesis, and testing cycle involving the biologist and the
computational and medicinal chemists. Leads with appropriate properties are
advanced to surrogate and functional animal models of the human disease to
evaluate toxicity and bioavailability (including adsorption, distribution, me-
tabolism, excretion), as well as disease modification or efficacy (3). Com-
pounds that successfully meet the criteria of disease modification (efficacy)
and safety are ready for Investigational New Drug (IND) application and fur-
ther human clinical testing prior to market approval by the FDA.
In the early 1980s, high throughput biological screening using auto-
mated workstations and robotic systems increased the productivity of the pri-
mary screening biological component to the point where synthesis of new
chemical entities became the rate-limiting step to the whole process. More-
over, as more biological assays were developed, optimized, and routinely per-
formed on the robotic platforms, the limitation of the current synthetic ap-
proach became more obvious. In an effort to increase the productivity of
synthetic chemistry, alternative modes of compound synthesis were investi-
gated.
By the early 1990s split and pool combinatorial techniques had been
developed for synthesis of a large collection of mixtures of peptides and small
molecular weight compounds that could be used to identify initial leads against
biological targets (4,5). Although combinatorial chemistry successfully pro-
vided large numbers of initial hits from the mixtures that were screened, the
follow-up necessary to identify the active component from a complicated mix-
ture was and still is quite challenging. Moreover, the screening of large mix-
tures is prone to a higher probability of false positive results due to incremen-
tal, additive effects of a large number of minor interactions. False negative
results can be observed due to antagonist and agonist effects in the same mix-
ture or due to a single compound that potently binds being offset by a larger
number of compounds that bind weakly (6). Because of the above chal-
lenges, the pharmaceutical industry continues to prefer acquisition of chemical
libraries of relatively large quantities (⬎10 mg) that contain single, well-
characterized compounds having relatively high purity of the component of
interest.
The requirements of single-component-per-well, highly characterized,

© 2000 by Marcel Dekker, Inc.

pure compounds led to the development of high-throughput, parallel, combina-
torial, organic synthesis borrowing many of the concepts developed for HTS
(7). The ability to synthesize several thousand analogs simultaneously also
increased the need to develop high-throughput analytical characterization as
well as information management techniques and software to deal with the
copious volumes of data. The development of HTOS, analytical chemistry,
and chemical information management as drug discovery tools continues
to narrow the numbers gap between chemical synthesis and biological test-
ing.
Within an organization, the appropriate sizing of each of the four main
disciplines (genomics, organic synthesis, biological evaluation, and preclinical
testing) is critical for maximally efficient high-throughput drug discovery (Fig.
1). Appropriate management of a team of scientists who possess diverse skills,
along with the numerous discovery tools they utilize, becomes a critical deter-
minant of success in any drug discovery organization (8).
Finally, recent advances in genomics and synthetic chemistry that have
yielded hundreds of new biological targets and thousands of new chemical
entities focus attention back to the biological screening component as the start-
ing point for the next major advancement in the drug discovery field. The
conventional methods utilized for HTS need to be modified and/or replaced
with economical, faster, high-resolution techniques that increase the produc-
tivity and efficiency of the drug discovery process. Many of the rapid, high-
resolution methods presently under development use highly sensitive detection
techniques such as mass spectrometry or fluorescence. These in turn allow an
increase in the physical density of discrete samples per unit space (i.e., 1564
wells per plate) and limit the consumption of precious biological target and
generation of chemical, biological, and radiological waste. We will review the
past, present, and future of biological testing methodologies.

III. ASSAY DEVELOPMENT

The binding of a protein to its ligand is driven by a number of highly specific

interactions, i.e., ionic interactions, hydrogen bonding, etc. These highly spe-
cific interactions serve to transmit biological ‘‘information.’’ This signal trans-
duction occurs by several differing mechanisms including protein conforma-
tional change, substrate cleavage, or substrate modification. Therefore, it is
of little surprise that the development of biological assays to discovery inhibi-
tors of bimolecular interactions is a process almost as heterogeneous as the
function of the proteins themselves. Two types of biological targets for which

© 2000 by Marcel Dekker, Inc.

drug discovery assays are commonly developed are receptors and enzymes.
We will present several examples of typical assays that have been developed
for these classes of biological targets. However, the list is by no means exhaus-
tive and each biological target will overlay unique considerations.
The simplest form of bioassay would detect any binding event between
a receptor/enzyme and a ligand/substrate. For certain applications such a
‘‘nonfunctional’’ assay will provide useful information, and later in this chap-
ter we will discuss extensively this approach to screening. However, only bind-
ing events that directly or indirectly interfere with the functionally relevant,
ligand-binding surface (e.g., active site) of the protein will interfere with bio-
logical activity and therefore be of potential therapeutic interest. Whether the
highest affinity binding ligands detected in these nonfunctional assay are those
that interfere with binding at the protein’s active site (since this is the region
of the protein surface with the most binding interactions) is a matter of unre-
solved speculation. However, molecular bioassays that directly measure bind-
ing of a substrate to its enzyme’s active site have been employed as the primary
lead discovery assay in drug discovery. Molecular bioassays are based on the
measurement of direct, selective binding of a labeled ligand to the protein’s
active site or the competitive displacement of the direct binding of a labeled
ligand.
Enzymes are proteins that catalyze the formation or cleavage of chemi-
cal bonds. A wide range of enzymatic assays are utilized in modern drug
discovery; for proteases the most simple measure the release of a quenched
chromophore, fluorophore, or radioisotope from a peptide substrate. These
styles of homogeneous assays are simple, with excellent signal-to-noise pro-
files. Fluorescence-quenched substrates (e.g., aminomethylcoumarin-AA 1-
AA 2-AA 3-dinitrophenol-AA 4-AA 5) have gained immense popularity because
the emission spectra of the aminomethylcoumarin in the intact substrate is
entirely quenched by the dinitrophenol moiety (9). Substrate cleavage of
amino acid 1–3 will result in the release of the fluorescence quench with a
greater than 1000 times increase in the emission signal of the fluorophore.
Such a homogeneous assay requires no separation or washing steps, yields a
high signal-to-noise ratio, and thus is very amenable to robotic HTS automa-
tion of protease inhibition assays.
Another significant class of enzymes—the phosphotransferases (10)—
catalyze the addition or cleavage of phosphate from amino acids and nucleo-
tides. There has been an explosion in the identification of protein kinases, i.e.,
enzymes that frequently perform an intercellular signaling role to transmit
information generated by ligands binding to the cell membrane to the cyto-
plasm or nucleus. Many investigators have described heterogeneous assays

© 2000 by Marcel Dekker, Inc.

where they monitored the inhibition of activity of therapeutically interesting
kinases by the incorporation of radioactively tagged 32 P from the substrate,
requiring a subsequent separation of the free and bound 32 P by centrifugation
or filtration (11). More recently, homogeneous-style kinase assays have been
developed that require no separation step; enzyme activity is measured by
fluorescence energy transfer (12). In this assay format a fluorescent molecule
is attached to a substrate that can be phosphorylated. Upon addition of phos-
phate by the kinase, the substrate is recognized by a antibody with a second
fluorescent molecule, which when in close proximity to the first fluorescent
molecule causes a shift in its emission wavelength.
Receptors are proteins that bind ligands and transmit information via
changes on other portions of the protein surface. The specific binding of a
ligand is favored by a surface-to-surface fit that maximizes the ionic interac-
tions, hydrogen binding, and other molecular interactions. Once a ligand has
been identified that exhibits reasonably specific binding characteristics, then
a competitive assay can be assembled (13). Homogeneous-style assay formats
(i.e., scintillation proximity assay, or SPA) have been described for receptor
binding that offer the advantages of the homogeneous format and the clean
signal-to-noise ratio of radiochemical assays (14). For example, in the SPA
assay, a receptor is bound to a polyvinyltoluene-based scintillation bead via
one of several specific, high-affinity interactions such as protein A-IgG or
avidin-biotin. These SPA bead–receptor complexes are then incubated with
a control ligand that is radiolabeled and of known affinity. Once the radioli-
gand is bound to the receptor and thus is in close proximity to the SPA bead,
the energy of the decaying radionuclide is transferred to the SPA bead and
the energy is converted to a light signal. This light signal is quantitated in a
scintillation counter. A decrease or increase in the assay signal associated with
the addition of a chemical entity is correlated with antagonism or agonism of
the bead, respectively. Other similar formats are available; for instance, the
scintillant can be embedded into the floor of the 96-well-plate (15) or, alterna-
tively, the scintillant can be replaced with a fluorescence molecule to enable
a fluorescence resonance energy transfer (FRET) system (12). These assay
formats offer both advantages and disadvantages depending on the exact assay
conditions.

A. The Test Tube?

The desire to test increasingly larger numbers of chemical entities, combined
with the attendant high cost of reagents needed, has resulted in the evolution
in the size of the container in which bioassays are conducted. A migration

© 2000 by Marcel Dekker, Inc.

toward smaller and smaller tube sizes has also been facilitated by technological
advances in liquid-handling devices and detection devices. As multiple test
tube (8-ml working volume) assays became more common, vendors began to
develop 6-well (2-ml working volume), 24-well (0.5-ml working volume), and
finally 96-well (250-µl working volume) well plates. Scientists have adopted
the standard 96-well plate because at this point liquid handling in the 1- to
100-µl range can be easily performed with manual and robotic liquid-handling
devices (16). In addition, the advent of microprocessor-controlled fiberoptic
detection devices with precise positional control has enabled the manufacture
of detection devices capable of obtaining excellent visible, ultraviolet, or fluo-
rescent signals from assays run in the 96-well-plate format (17).
It now appears that with current conventional liquid-handling and detec-
tion devices it is possible to utilize the 384-well (70-µl working volume) for-
mat (18). Further miniaturization of the plate format has yielded the 1536-
well plate (15-µl working volume) that is currently being utilized by several
investigators (19). However, utilization of this format requires nonconven-
tional ‘‘ink-jet’’-style liquid handlers and imaging-type detection systems
(20,21). These types of systems are currently available to lab scientists but at
premium prices compared to 96-well technology. Several fundamental issues
also remain as roadblocks for this technology, including sample evaporation,
solvent compatibility with the ‘‘biojet’’ dispenser, and positional accuracy of
the dispenser for each well. The decision to migrate to denser well formats
may be driven in large degree by one’s assay format, reagent costs, and re-
search budget.
The next generation of assay technology will likely be a move away
from plate-based screening altogether. Although massively parallel plate tech-
nology has afforded extraordinary throughputs, the inherent limitations of per-
forming assays in wells will push many researchers to examine methodologies
that do not require the use of tubes. We will discuss some of these advanced
screening methodologies below.

IV. AUTOMATION OF BIOASSAYS

A. Manual Assays
As manual, multichannel pipeting devices and plate-based detection systems
became staple tools in the modern discovery laboratory, bioassays for mod-
erate throughput have been developed in a wide variety of formats. The flexi-
bility of these detection devices allowed assays that utilize visible-,
fluorescence-, luminescence-, and radioactive-based readouts. Manual pi-

© 2000 by Marcel Dekker, Inc.

peting allows for fine adjustment in the pipeting parameters for detergent,
organic solvents, and other additives to the aqueous solution. These pipeting
parameters include aspirate speed, delay time, ejection speed, and blowout
speed and volume. Adjusting a mechanical pipeting device to accommodate
the large variety of variables that it will encounter demonstrates the utility
and flexibility of the human interface.

B. Liquid-Handling Robotics
Repetitive pipeting by humans is an error-prone activity. In order to eliminate
the errors inherent in human high-throughput liquid handling, many manufac-
tures have developed a wide variety of robotic devices. The most simplistic
of these devices is the plate-based liquid-handling robot. Several examples of
the more popular version of these devices include Beckman BioMek, Gilson
Model 215, Packard Multiprobe, and Tecan Genesis (22). These liquid-
handling workstations offer single-probe, 8 fixed probes, 4 or 8 variable span
probes, and 96 fixed probes pipeting capability. Other options include fixed
stainless steel probes, disposable pipet tips (20, 200, or 1000 µl size), a wide
variety of liquid reservoirs with temperature control and mixing, system fluid-
ics versus mechanical piston pipeting control, etc. All of these liquid handlers
have computer-controlled functionality, with some of the better software offer-
ing GUI interfaces to facilitate programming of the instrument. The size of
the pipeting deck varies greatly for these instruments. Some of the instruments
are designed with the flexibility to adapt to many different microtiter plates,
test tubes, and reservoirs, whereas others are dedicated to only the 96-well-
plate format. Capacity will vary greatly, some liquid handlers being designed
for only a few plates before a change of the deck is required and others
allowing for longer periods of unattended operation. Which particular instru-
ment will best fit your appropriate needs is probaly based on your specific
needs.

C. Integrated HTS Robotics

Further bioassay automation requires the assembly of integrated robotic sys-
tems, such as those supplied by Zymark, Beckman/Sagian, or Robocon (23).
These custom or semicustom robotic systems increase the bioassay throughput
capability of the standalone liquid-handling robots by incorporating pipeting
stations, light/temperature/atmosphere–controlled incubators, wash stations,
detection devices, plate storage carousels, and plate transfer systems onto one
system. Computerized control of the various subsystem routines allows for

© 2000 by Marcel Dekker, Inc.

optimum scheduling and thus maximum throughput of the assay. These HTS
robots are capable of operating for up to a 24-hour unattended duty cycle.
Throughputs for a typical assay on these systems can range from several thou-
sand per day for a ELISA-style assay to 15,000 per day (and up) for a homoge-
neous assay. These systems range from the highly customized component ones
whereby a team of hardware engineers and software programmers are required
for appropriate support, to the more ‘‘off-the-shelf ’’ systems that can be main-
tained and operated by biological scientists with minimal mechanical skills.

V. INTEGRATION OF HTOS AND HTS FOR ACCELERATED

LEAD GENERATION AND OPTIMIZATION

A number of research teams, including scientists at ArQule (7,24), have devel-

oped and reduced to practice the automated, parallel, solution phase synthesis
of large (10 3 –10 4) compound sets, or arrays, of small organic molecules. In
this chemical synthesis scheme, each array of molecules is defined by a unique
chemical scaffold or ‘‘molecular core’’ to which are attached diverse pharma-
cophoric side groups. These arrays are created by combinatorially reacting a
series of functionally identical but structurally diverse building blocks to pro-
duce a single compound. The compound is then chemically analyzed by high-
performance liquid chromatography (HPLC) and mass spectrometry (MS),
and can be purified if necessary. The compounds are logically arranged in
spatially addressable 96-well microtiter plates with a single compound per
well (Fig. 2). This spatially addressable array format yields an ‘‘information-
rich’’ array whereby chemical structure, mass, and other properties are stored
in a chemical database that is searchable via a unique identifier. Many of these
chemical arrays, each based on a unique molecular core, have been assembled
into large (typically more than 250,000 compounds) chemical compound sets.
This large compound set has been utilized to discover novel lead structures for
pharmaceutical drug discovery that can be further optimized using traditional
medicinal chemistry approaches.
Spatially addressable chemical arrays are inherently information-rich;
this property results from utilizing reagents, arranged in the x, y, and z axis
of the combinatorial cross, that have a high degree of structural relatedness
(i.e., methyl, ethyl, isopropyl, etc.). As a result, the neighbors of a given prod-
uct in this combinatorial array will also have a high degree of structural relat-
edness. This property of spatially addressable arrays provides an enormous
potential for gaining structure–activity relationship (SAR) data from the pri-
mary bioassay. A very simple example of this principle can be demonstrated
where a combinatorial array is created by crossing 8 related reagents on the

© 2000 by Marcel Dekker, Inc.

Figure 2 Layout of a spatially addressable chemical compound array.

y axis and 10 related reagents on the x axis to yield 80 compounds. These

chemical compounds were assayed versus a cysteine protease and the resultant
data can be observed in Fig. 3. The SAR data obtained from this primary
screen were used to rapidly accelerate the subsequent lead optimization efforts
that resulted in a further 10-fold or higher gain in potency against the target
protease.

A. Design and Synthesis

Several considerations drive the design and synthesis of spatially addressable,
combinatorial arrays (25). The selection of building blocks and backbone

© 2000 by Marcel Dekker, Inc.

Figure 3 Percentage inhibition of cysteine protease activity by array X at 10 µM.

chemistries is driven by a combination of factors. These include years of accu-

mulated medicinal chemistry knowledge, structure-guided design approaches
based on high-resolution data of the target protein (when available), and bio-
logical assay data for inhibition of activity of the target protein (or related
proteins) with closely related compounds (Table 1). In addition, there are prac-
tical criteria for synthesis that also drive the selection of a particular chemistry

Table 1 Factors Guiding the Design and Automated

Synthesis of a Spatially Addressable Chemical Array
Chemical Array Design
• Medicinal chemistry experience
• Structure-guided design
• Biological data
Chemical Array Synthesis
• Synthetic feasibility
• Compatibility of chemistry with automated hardware
• Availability of building block reagents

© 2000 by Marcel Dekker, Inc.

and the inclusion of particular building blocks. These criteria include the gen-
eral synthetic feasibility of the chemistry as regards the current portfolio of
automated chemical synthesis hardware and the range of chemical transforma-
tions feasible on that platform. Finally, one must consider the availability of
novel building block reagents and their influence on the chemical diversity of
the array.
The work flow in synthetic organic chemistry can be broken down to a
series of typical unit operations. Chemical synthesis, purification, and analysis
can be leveraged by the application of laboratory scale automation to these
unit operations to enhance their efficiency. Finally, one can apply the basic
principle of computerized information management to enhance the level of
process control. The integration of these technological approaches results in
what can be referred to as industrialized HTOS. This approach to the chemical
synthesis process has enabled the automated parallel synthesis, purification,
and analysis of more than 2 ⫻ 10 5 compounds per year. Similar approaches
in the electronics industry and biological testing disciplines have resulted in
enormous productivity gains.

B. Automated High-Throughput Organic Synthesis

The discipline of combinatorial chemistry has automated a number of the unit
operations in synthetic organic chemistry using a workstation approach (Table
2). Heavy emphasis has been placed on the adaptation of ‘‘off-the-shelf,’’
laboratory scale automation for the specific needs of synthetic chemistry to
further simplify this approach. However, where necessary extensive customi-
zation or ground-up engineering of modules has been undertaken by research
teams, as well as various automation vendors (26). Further addition of the

Table 2 Automated Molecular Assembly Process

Workstations
Chemical inventory management
Weighing and dissolution
Open-well chemical synthesis
Thermal control and agitation
Centrifugal solvent evaporation
Liquid-liquid extraction
High-throughput preparative chromatography
Analytical chemistry
Array replication, shipping, and tracking

© 2000 by Marcel Dekker, Inc.

chemical design and synthesis principles outlined above and computerized
process control has resulted in the integration of these workstations into a
HTOS process. One possible iteration of the elements of this process and their
interrelationships is represented in Fig. 4. In the figure they are represented
as a linear process, but in actuality the order of the units operations in the
process is infinitely flexible to enable the real-time adapting of the chemistry
process to observations made by operators and analysts. Thus we are able to
adopt commonly accepted industrial production practices to gain efficiency
while essentially performing chemical research. Clearly, other chemical unit
operations remain to be automated and integrated into the process. Automation
of some of these unit operations will be a highly complex task and will involve
both hardware and instrument control software engineering of the robotic unit
operation. However, the integration of these modules into an HTOS process
will be facilitated by the flexible architecture of the computerized information
management software. It has been developed for maximal process control,
along with standardized reaction block hardware designed to allow the physi-
cal reaction module to be quickly and easily moved from one station to an-
other.

Figure 4 Industrialized high-throughput organic synthesis.

© 2000 by Marcel Dekker, Inc.

C. Information Management
In order to control the industrialized HTOS process scientists have developed
software applications that facilitate the design and capture of the intellectual
process that a chemist goes through when designing combinatorial chemical
arrays (these have been termed array information management software, or
AIMS). These software applications operate on conventional computer op-
erating systems and computer/LAN hardware and offer a degree of process
control with all of the attendant advantages. This software allows the chemist
to develop an array production module (APM) or ‘‘recipe’’ that captures the
set of mechanical instructions that the robotic instrument control software will
require in order to execute the chemical synthesis on the automated platform.
The AIMS facilitates and institutionalizes the APM knowledge base in a
highly automated HTOS environment.
Once APMs have been created in the AIMS, a process control and moni-
toring software (PCMS) module can be constructed that provides for two-way
communication between a database that contains the APMs and the instrument
control software. This software allows for the retrieval of APM data from the
AIMS by a production scientist and transfer to the instrument control software.
Electronic transfer of APMs between unit operations in this manner allows
for (a) charting of APM progress on the production floor and (b) collection
of APM audit data. The PCMS will facilitate maximal process control for the
automated chemical synthesis platform. Functionalities for the array informa-
tion management and process control management software are summarized
in Table 3.
The guidance system for the design, synthesis, and analysis of chemis-

Table 3 Summary of AIMS/PCMS Functionality

Array Information Management System

• Retrieval of chemical reagent data from chemical information database.
• Creation of array production models (APMs) by chemists to capture array
layout and mechanical instructions.
• Ability to monitor the status of APMs on the production floor.
• Retrieval and reporting of array production history.
Process Control and Monitoring System
• Retrieval of APM details by instrument control applications from array
information management software.
• Charting of array production progress through production and QC.
• Persistence of audit data from instrument control applications tracking APM
natural history.

© 2000 by Marcel Dekker, Inc.

tries via HTOS is a series of interrelated information management systems.
In addition to the AIMS/PCMS described above, organizations employing
HTOS laboratories as one component of an integrated drug discovery platform
have established fairly traditional chemical, analytical, biological, and corpo-
rate business information databases. Once chemicals have been synthesized
and analyzed they are automatically registered in a chemical information man-
agement database, such as an ISIS-searchable Oracle relational database. An
essential feature of such a database is the assignment of a unique identifier
for each compound in the chemical compound set. This unique identifier pro-
vides links to information about the chemistry backbone, size of the array,
and other chemical, physical and biological properties, as well as business
data contained in the chemical and inter-linked databases. In drug discovery-
based organizations, SAR data that are stored in an easily searchable (i.e., via
a web browser) database can be utilized by medicinal chemists, biologists,
and computational chemists to enhance the design of future arrays. Readily
shared SAR data also facilitate the acceleration of lead optimization and pre-
clinical studies by a large and diverse group of scientists.

D. Integration of HTOS and MHTS

The advent of ever increasing HTS methodologies has driven the need for
high-throughput organic chemistry via combinatorial and parallel synthesis
approaches. However, even with HTS technologies, the testing of 10 5 –10 6
chemical compounds against multiple biological targets poses several signifi-
cant hurdles, primarily the cost and availability of key reagents. Several strate-
gies have been typically employed to manage the biological testing of large
chemical compound sets against multiple biological targets. Single compound
per bioassay per well is the most straightforward. The advantages are that
no deconvolution is required and the potential for ‘‘masking’’ of bioactivity
is minimized. Single compound per bioassay fits particularly well with the
information-rich nature of spatially addressable chemical arrays. Essentially
the entire primary bioassay provides extensive SAR data as was shown in Fig.
2, with the negative bioassay data also adding value for the subsequent lead
optimization activities. However, the cost of this approach is significantly
higher than that of the alternative strategy of compound pooling. Pooling of
between 3 and 10 compounds per bioassay has been utilized to quickly and
efficiently assay large compound sets (27,28). The major disadvantages are
the need for subsequent deconvolution of positive readouts, potential for mask-
ing of one compound’s activity by others and, specifically for spatially ad-

© 2000 by Marcel Dekker, Inc.

dressable chemical arrays, the information content of the compound set is
partially lost.
In order to efficiently bioassay large compound arrays while preserving
the integrity of the SAR information content of the spatial array, we have
found that multiple, parallel, high-throughput bioassays are an attractive alter-
native. We will review several case examples of multiplexed bioassays with
spatially-addressable chemical arrays and discuss issues related to data man-
agement, prevention of readout crosstalk and multiplexing of primary and sec-
ondary assays.

E. Multiplexed High-Throughput Screening (MHTS)

The concept of multiplexed bioassays to accelerate assays has been utilized
previously. One of the best examples is DNA sequencing whereby an assay
is performed using the method developed by Sanger with four different
fluorescence-labeled reporters for each A, T, C, or G nucleotide (29). The reac-
tions are loaded onto a 6% denaturing polyacrylamide gel and electrophoresed.
As the fluorescence-labeled DNA moves past the laser in an Applied Biosys-
tems DNA Sequencer, each dye is excited and emits a characteristic wave-
length enabling nucleotide base assignment. The analogous use of multiplexed
bioassays for HTS of spatially addressable chemical arrays preserves the
information-rich nature of the array while allowing pooling of the assays to
save costs.
Although MHTS bioassays can be multiplexed (2–4 per well) to acceler-
ate testing of spatially addressable combinatorial libraries (⬎10 5 compounds)
versus multiple biological targets, there are some practical limitations. For
instance, the approach is more amenable to technically straightforward bioas-
says (i.e., SPA, colorimetric, or time resolved fluorescence (TRF)) since the
assay conditions are more likely to be similar. Multiplexing is best employed
when combining primary bioassays that utilize the same or different reporter
systems (see below). In theory, multiplexing of first- and second-degree assays
is possible, but again assay conditions must be similar, which is less often the
case. Some of the inherent advantages of MHTS assay format include reduced
consumable and personnel costs versus single compound/single bioassay, and
an alternative to compound pooling that maximally preserves the information-
rich nature of spatially addressable style combinatorial arrays. Single-com-
pound-per-well testing prevents masking of activity and eliminates the need
for deconvolution.
In the first example we have multiplexed two primary bioassays that
utilize different reporter systems, i.e., visible and/or fluorescence readout. In

© 2000 by Marcel Dekker, Inc.

vitro serine and metalloprotease assays were run in 96-well plates either sepa-
rately and read in the appropriate detector, or multiplexed and read first in the
spectrophotometer and then in a spectrofluorometer. Common assay condi-
tions were developed and optimized for both enzymes. The serine protease
assay substrate (e.g., AA 1-AA 2-AA 3 ⬃ pNA) had a visible readout at A 405;
the metalloprotease assay substrate (i.e., AMC-AA 1-AA 2-AA 3-DNP-AA4-
AA 5 ) had a fluorescence readout (Ex 320, Em 405). ArQule compounds were dis-
solved in dimethylsulfoxide (DMSO), and tested at a final concentration of
10 µM (final solvent: 2% DMSO in aqueous pH 7.5 buffer). Serine- and metal-
loprotease-positive control inhibitors were spiked into a plate containing
DMSO at a range of concentrations that yielded from ⬃20 to ⬃90% inhibition.
The result from the separate and multiplexed assays is shown in Fig. 5; clearly,
the observed inhibition in the multiplexed assay is quantitatively similar to
the individual assays. In this particular example, the signal for the multiplexed
fluorescence assay is reduced somewhat due to absorption by the chromogenic
substrate; however the signal-to-noise ratio (S/N ) remained acceptable.
In the second example, two primary, in vitro serine protease assays with
identical p-nitroaniline-based chromogenic (A 405) readouts were either assayed
separately or multiplexed. Positive control inhibitors specific to each enzyme
were added to the assay at an appropriate concentration to determine assay

Figure 5 MHTS of first-degree bioassays: different reporter systems.

© 2000 by Marcel Dekker, Inc.

Table 4 Multiplexed High-Throughput Screening of First-Degree
Bioassays—Same Reporter Systems

Multiplexed assays
Assay 1 Assay 2 (1 and 2) a

1 µM Inhibitor A 79 ⫾ 2 0 43 ⫾ 2
10 µM Inhibitor B 0 48 ⫾ 3 20 ⫾ 1
(S/N ⫽ 11) (S/N ⫽ 11) (S/N ⫽ 9)
a
Cross-talk between both enzymes vs. both substrates ⫽ 4%.

cross-talk and readout specificity. This represents a more difficult data inter-
pretation scenario because any positive inhibition data will require separate
follow-up testing for confirmation. In this case of identical readouts, the per-
centage inhibition determined for a given compound is mathematically halved.
Experimental data for this assay (Table 4) demonstrated that when assayed
separately the positive control inhibitor yielded the expected percent inhibition
with a very high S/N value of 11. When the two assays were multiplexed
inhibition by either positive control inhibitor was detectable (i.e., signal was
well over the noise) S/N was only slightly compromised. Cross-talk was only
4% in this assay format.

F. Considerations for Establishing Multiplexed Assays

Several considerations for multiplexing assays can be determined from the
above data. Clearly, multiplexing different assay readouts is optimal (e.g.,
chromogenic with fluorogenic or radiolabeled), since the possibility for cross-
talk is minimized and the readout is unambiguous. Usually, to meet this criteria
one would choose different assay classes (e.g., metalloprotease with serine
protease), since the possibility for cross-talk would be minimized. Several
conditions are necessary to determine if two or more assays are appropriate
for multiplexing. These include the following: (a) cross-talk between enzymes
and substrates must be minimal (⬍5%); (b) enzymes must not cleave or
modify other enzymes or receptors in the assay during storage on the robot;
(c) cross-talk between readouts (e.g., excitation/emission) must be minimal;
(d) multiplexed ELISAs must confirm antibody specificity between antigens;
and (e) multiplexed assays must utilize similar assay conditions (i.e., pH, time,
buffers, etc.).

© 2000 by Marcel Dekker, Inc.

VI. ADVANCED TECHNOLOGIES FOR THE FUTURE OF
BIOANALYTICAL TESTING

Many of the high-resolution screening methods presently under development

employ high-sensitivity techniques, such as mass spectrometry or fluorescence
detection. These sensitive techniques limit the use of precious biological target
and/or increase the physical density of discrete samples per unit space. The
latter consideration limits the use of solvents and generation of waste. Because
of its inherent high sensitivity and direct mass measurement capabilities, mass
spectrometry has been found at the forefront of new screening methodology
development. The following examples represent the different approaches that
have been taken with mass spectrometry.

A. Direct Protein-Ligand Binding

Many traditional bioassays rely on a competitive binding process whereby a
ligand with known affinity for the target protein is incubated with a single
chemical compound or a mixture. If the compound(s) of interest are interacting
with the target protein with greater affinity, then the ligand will be excluded
from binding. By measuring the difference in signal between the control sam-
ple (containing only the ligand of known affinity) and the sample with ligand
and novel compound(s), any reduction in the interaction of the ligand is attrib-
uted to interaction by the chemical entity of interest. Although this indirect
approach works fairly well with single-compound systems, it is more difficult
with mixtures because of the necessary deconvolution of the mixtures. More-
over, this competitive binding process is not an option when a known binder
for a target is not known.
However, by infusing a preincubated mixture of the receptor and li-
gand(s) into the mass spectrometer via electrospray ionization, the receptor–
ligand complex can be seen directly (30,31). Typically, electrospray ionization
of a large macromolecular protein will yield a mass-to-charge envelope of the
molecular weight of the protein divided by the number of possible charges
(between 2 and 30) that can be stabilized on that molecule. When the receptor–
ligand complex is formed and survives the electrospray process, the molecular
weight of the complex will be greater than the free receptor by the weight of
the interacting ligand. A new peak, representative of the complex, will be seen
in the receptor–ligand mass spectrum if the effective mass increase of the
ligand divided by the number of charges on the complex is greater than the
mass difference between two adjacent charge states of the free receptor. Using
this approach, one can directly observe the receptor–ligand complex and deter-

© 2000 by Marcel Dekker, Inc.

mine which chemical entity is interacting with the receptor. Since the mass
and the charge states for a given receptor are known and the masses of the
combinatorial chemical entities are also known, then all possible combinations
of receptor–ligand masses can be determined and monitored.
This direct receptor–ligand binding approach has been used to identify
compounds with affinity for the receptor; however, by using infusion tech-
niques that are very labor-intensive the overall throughput is rather limited.
Recently, Li et. al. (32) automated the procedure by incorporating flow injec-
tion analysis using a conventional autosampler to increase the throughput. By
injecting a small aliquot of the receptor–ligand solution in the flowing stream
to the mass spectrometer, the analysis of each sample was conducted in under
3 min, thus gaining a significant increase in throughput.

B. Immunoaffinity Extraction LC/MS

The immunoaffinity extraction LC/MS technique (33) is a variation of affinity
chromatography. The receptor of interest is loaded on the column, followed
by injection of the combinatorial library of small molecular weight com-
pounds. The compounds that interact with the receptor are captured on the
affinity column while the rest of the library is removed from the column.
Following the loading step the pH of the mobile phase is changed, which strips
the affinity column of the receptor–ligand complexes and loads the complexes
onto a second column containing a restricted access media (RAM) stationary
phase (34). The RAM column separates the receptor from the ligands and
the ligands are then flushed onto a reverse phase column for separation and
identification by MS. Using this approach a large number of potential ligands
can be rapidly screened against a variety of receptors. Moreover, by incorpo-
rating a number of switching valves and computerized control the whole pro-
cess can be automated to provide a relatively high-throughput screen.

C. Multidimensional Chromatography/Mass Spectrometry

This technique uses size exclusion and reverse phase chromatography in tan-
dem to separate the receptor bound ligand followed by mass spectrometry to
identify the particular ligand from a combinatorial library (35). The protein
target is incubated with a single compound or mixtures under conditions that
optimize protein–ligand binding. An aliquot of the mixture is then loaded onto
a size exclusion column that separates the protein and protein–ligand complex
from the smaller molecular weight unbound ligand. The protein and the pro-
tein–ligand complex eluding from the size exclusion column in the void vol-

© 2000 by Marcel Dekker, Inc.

ume is directed onto a reverse phase column, whereas the retained low molecu-
lar weight compounds are directed to waste. Once the protein–ligand complex
is directed to the reverse phase column with a high organic mobile phase, the
complex is disrupted and the ligands that were originally bound to the protein
are released. The reverse phase column then separates the different ligands
from each other as well as from the protein and the column flow is directed
into the mass spectrometer for identification of the ligands.

D. Bioaffinity Characterization Mass Spectrometry

In bioaffinity characterization mass spectrometry (BACMS) the combinatorial
library is incubated with the protein under suitable conditions and the mixture
is then injected into the mass spectrometer under electrospray mode (36,37).
The protein–ligand complex can be isolated and then energized to dissociate
the complex. Because of the relatively low concentrations of complexes that
are formed, a Fourier transform ion cyclotron resonance (FTICR) instrument
is used to accumulate the protein–ligand complex of interest in the FTICR
cell. Once enough signal is accumulated in the cell, the complex is dissociated
and the ligand is identified by high-resolution MS and MS/MS experiments.
The benefit of this methodology is that, like all other solution-based character-
izations, there is no stearic constraint on complexation due to a solid support
tethering the protein.

E. Affinity Capillary Electrophoresis/MS

Once again, the strength of this technique is that the protein–ligand complex
is allowed to form in solution, thereby removing any steric interference that
may by possible in solid phase affinity chromatography (38,39). In this tech-
nique, an electrophoresis capillary is filled with an aliquot of the receptor
solution, which has been pH-adjusted so that the receptor is neutral. The com-
binatorial library is then injected as a small plug in the capillary and the elec-
trophoresis is started. As the library components move through the electropho-
resis capillary, the ones that interact with the receptor are slowed down and
come off the column later than those that do not interact with the receptor.
By using a mass spectrometer as the final detector, the identity of the ligands
that interacted can be determined for subsequent synthesis and additional
study. Moreover, by appropriately adjusting the mobilities of both the ligands
and the receptor, a variation known as the ‘‘plug-plug’’ approach can be per-
formed. In this mode, both the ligands and the receptor are introduced in the

© 2000 by Marcel Dekker, Inc.

capillary and allowed to interact at certain points in the capillary. By using
this approach, weaker interacting ligands can be identified.

F. Immunoaffinity Ultrafiltration LC/MS

In this procedure the receptor is incubated with the combinatorial library under
the appropriate conditions that will encourage receptor–ligand binding. Fol-
lowing incubation, the solution is transferred to centrifugal filters and centri-
fuged at 10000g for several minutes. The filter allows small molecular weight
components—including the unbound ligands—to pass through the filter and
retains compounds greater than the molecular weight cutoff of the filter, in-
cluding free receptor and receptor–ligand complexes (40). Because of the in-
herent dead volume of the filter several aliquots of appropriate buffer are
passed though the filter to remove any residual free ligand(s) solution. The
affinity complex is then disrupted with an appropriate solvent and the solution
is again centrifuged. This time the ligands that were originally bound to the
receptor are allowed to pass in to a clean receptor tube and the receptor is
still maintained on the filter. The solution containing the ligands extracted
from the receptor–ligand complexes is then analyzed by HPLC and identified
by MS using electrospray ionization. This procedure tends to select library
components with the greatest affinity for the target receptor.

G. Pulsed Ultrafiltration Mass Spectrometry

The pulsed ultrafiltration MS technique is similar to the immunoaffinity ultra-
filtration LC/MS process in that both techniques use a membrane with molecu-
lar weight cutoffs to separate free small molecular weight ligands from recep-
tors or receptor–ligand complexes. The pulsed ultrafiltration process utilizes
a typical HPLC/MS system, except that the chromatographic column is re-
placed by a pulsed ultrafiltration chamber. This chamber consists of either a
1.0-in.-diameter in-line solvent filtration unit that uses a high molecular weight
cutoff filter membrane, or a home-built chamber that incorporates mechanical
agitation (41). The mixture of receptor and a small molecular weight combina-
torial library are first incubated and the solution is loaded into the pulsed
ultrafiltration chamber. After the chamber is washed with an appropriate sol-
vent for several minutes to remove any unbound ligands, the mobile phase
solution is changed to one that will disrupt the receptor–ligand complex and
release the free ligands. The free ligands are then transported to the mass
spectrometer and identified by MS or MS/MS analysis. Alternatively, the re-
ceptor can be loaded into the pulsed ultrafiltration chamber and an aliquot of

© 2000 by Marcel Dekker, Inc.

a combinatorial library can be passed through the chamber. The ligands that
have a high affinity for the receptor will bind, whereas the others will pass
through the chamber. Then by changing the mobile phase and releasing the
bound ligands to be identified by MS, the receptor is now available for an
additional aliquot of the same or a different combinatorial library to be
screened. This is valid as long as the conditions used to disrupt the receptor–
ligand complex does not permanently denature the receptor.

ACKNOWLEDGMENTS

The authors acknowledge their colleagues at ArQule, Inc., many of whom

have made significant contributions to this body of scientific work.

REFERENCES

1. P Banerjee, M Rokofsky. Re-inventing drug discovery: the quest for innovation

and productivity. (1997) www.ac.com/services/pharma/phar dd.html
2. www.lifespan.com
3. www.navicyte.com
4. RA Houghten, C Pinilla, SE Blondelle, JR Appel, CT Dooley, JH Cuervo. Gener-
ation and use of synthetic peptide combinatorial libraries for basic research and
drug discovery. Nature 7;354(6348):84–86, 1991.
5. http:/ /www.pcop.com/chemistry.html, http:/ /www.combichem.com/home.html,
http:/ /www.trega.com/technology.htm
6. S Kaur, L McGuire, D Tang, G Dollinger, V Huebner. Affinity selection and
mass spectrometry-based strategies to identify lead compounds in combinatorial
libraries. J Protein Chem 16(5):505–511, 1997.
7. JC Hogan. Combinatorial chemistry in drug discovery. Nat Biotechnol. 15(4):
328–330, 1997; JC Hogan. Directed combinatorial chemistry. Nature 384(6604
Suppl):17–19, 1996.
8. P Banerjee. Re-Inventing Drug Discovery: The Quest for Innovation and Produc-
tivity Presentation at Discovery ’98: Emerging Technologies for Drug Discovery.
NMHCC Biotechnology Conferences. May 18, San Diego.
9. CG Knight. Fluorimetric assays of proteolytic enzymes. Meth Enzymol 248:18–
34, 1995.
10. T Hunter. The Croonian Lecture 1997. The phosphorylation of proteins on tyro-
sine: its role in cell growth and disease. Philos Trans R Soc Lond B Biol Sci
353(1368):583–605.
11. Y Miyata, B Chambraud, C Radanyi, J Leclerc, MC Lebeau, JM Renoir, R Shirai,
MG Catelli, I Yahara, EE Baulieu. Phosphorylation of the immunosuppressant

© 2000 by Marcel Dekker, Inc.

 FK506-binding protein FKBP52 by casein kinase II: regulation of HSP90-bind-
ing activity of FKBP52. Proc Natl Acad Sci USA 94(26):14500–14505, 1997.
12. http:/ /www.packardinst.com/offices/home pbbv.htm
13. G Deng, G Royle, S Wang, K Crain, DJ Loskutoff. Structural and functional
analysis of the plasminogen activator inhibitor-1 binding motif in the soma-
tomedin B domain of vitronectin. J Biol Chem 271(22):12716–12723, 1996.
14. http://www.apbiotech.com/prod/html/apb prod.html
15. http://www.apbiotech.com/prod/html/apb prod.html,
http://www.nenlifesci.com/p srv 13.htm#74
16. http://www.corningcostar.com/costar/sub.html?sg⫽20,
http://www.polyfiltronics.com/overview.html,
http:/ /www.greineramerica.com/products/96well.htm
17. http:/ /www.wallac.fi/mag2/p6.html, http:/ /www.moldev.com/,
http:/ /www.bmglabtechnologies.com/fluo.htm
18. http:/ /www.corningcostar.com/launch384well.html,
http:/ /www.greineramerica.com/products/384well.htm
19. http:/ /www.greineramerica.com/products/1536well.htm
20. http:/ /www.cartesiantech.com/prod01.htm
21. http:/ /www.fujimed.com/science/index.html, http:/ /www.mdyn.com/
22. http:/ /www.packardinst.com/cgibin/hazel.exe?action⫽SERVE&
item⫽prod serv/mprobe.htm, http:/ /www.tecan.com/index tecan.shtml,
http:/ /134.217.3.35/beckman/biorsrch/prodinfo/biomek/biowork.asp
23. http://134.217.3.35/beckman/biorsrch/prodinfo/biomek/biomek.asp,
http:/ /www.zymark.com/MARKETIN/drugd/allegr.htm
24. www.arqule.com
25. CM Baldino, DS Casebier, J Caserta, G Slobodkin, C Tu, DL Coffen. Convergent
Parallel Synthesis. Synlett 488–490, 1997.
26. http://www.zymark.com/MARKETIN/drugd/Prsps.htm,
http://www.gilson.com/gpss.htm, http://www.tecan.com/index tecan.shtml
27. DA Konings, JR Wyatt, DJ Ecker, SM Freier. Deconvolution of combinatorial
libraries for drug discovery: theoretical comparison of pooling strategies. J Med
Chem 39(14):2710–2719, 1996.
28. L Wilson-Lingardo, PW Davis, DJ Ecker, N Hebert, O Acevedo, K Sprankle,
T Brennan, L Schwarcz, SM Freier, JR Wyatt. Deconvolution of combinatorial
libraries for drug discovery: experimental comparison of pooling strategies. J
Med Chem 39(14):2720–2726, 1996.
29. http:/ /www2.perkin-elmer.com/ab/about/dna/dna seq/bdterm.html
30. YT Li, JD Henion. J Am Chem Soc 113:6294–6296, 1991.
31. JAE Kraunsoe, RT Aplin, B Green, G Lowe. FEBS Lett 396:108–112, 1996.
32. LYT Li, CH Vestal, JN Kyranos. Proceedings of the 45th ASMS Conference
on Mass Spectrometry and Allied Topics, Palm Springs, California, June 1–5,
1997, p. 897.
33. ML Nedved, S Habibi-Goudarzi, B Ganem, JD Henion. Anal Chem 68:4228–
4236, 1996.

© 2000 by Marcel Dekker, Inc.

34. TCJ Pinkerton. J Chromatogr 544:13–23, 1991.
35. Y Hsieh, N Gordon, F Regnier, N Afeyan, SA Martin, GJ Vella. Mol Divers 2:
189–196, 1996.
36. JE Bruce, GA Anderson, R Chen, X Cheng, DC Gale, SA Hofstadler, BL
Schwartz, RD Smith. Rapid Commun Mass Spectrom 9:644–650, 1995.
37. J Gao, X Cheng, R Chen, GB Sigal, JE Bruce, BL Schwartz, SA Hofstadler,
GA Anderson, RD Smith, GM Whitesides. J Med Chem 39:1949–1956, 1996.
38. YH Chu, YM Dunaevsky, DP Kirby, P Vouros, BL Karger. J Am Chem Soc
118:7827–7835, 1996.
39. YH Chu, DP Kirby, BL Karger. J Am Chem Soc 117:5419–5420, 1995.
40. R Wieboldt, J Zwiegenbaum, JD Henion. Anal Chem 69:1683–1691, 1997.
41. RB Van Breemen, CR Huang, D Nikolic, CP Woodbury, YZ Zhao, DL Venton.
Anal Chem 69:2159–2164, 1997.

© 2000 by Marcel Dekker, Inc.

9
Commercial Resources

Mary Brock and Mark Andrews

Waters Corporation
Milford, Massachusetts

Combinatorial chemistry and high-throughput screening (HTS) have become

the fastest growth areas in pharmaceutical development. Combinatorial chem-
istry is a collection of technologies and disciplines. The ultimate objectives
are to prepare compound libraries so that active leads can be identified more
quickly, in greater numbers, and at lower costs. Over the past few years there
has been an explosion in the number of companies developing, utilizing, and/
or offering the various techniques for use in the lab. Press releases announce
activities in these areas on a daily basis. It is a challenge to keep up with the
current players, whether they are developing or licensing proprietary technolo-
gies, establishing collaborative relationships, commercializing products, ac-
quiring or selling business operations, or any of a number of other activities.
Initially the intent of this chapter was to provide the reader with a com-
prehensive list of commercial suppliers of products (instrumentation and sup-
plies) relating to combinatorial chemistry and HTS. While conducting research
into these vendors, it became obvious that a lab would likely consider access
to a particular technology, or to an awareness of the potential for collaboration
with another company, to be of as much value as a listing of companies that
provide specific instrumentation and related products. It also became obvious
that a ‘‘comprehensive’’ list would be impossible because of the ongoing high
level of activity among old and new participants in this area.
Therefore, this chapter is a sampling of companies that could potentially
be a resource for your lab. It will provide a snapshot of some of the activities
and companies that are currently involved in offering products and services

© 2000 by Marcel Dekker, Inc.

relating to combinatorial chemistry. We offer our apologies to those compa-
nies or products that were inadvertently omitted from this listing.
The information provided was obtained from various public sources,
including company press releases, company internet web sites, articles ap-
pearing in many of the industry trade journals, magazines, product advertising,
product brochures, and other company sources. The entries contain informa-
tion available at the time the chapter was written and are as accurate as the
original sources. A listing of a company or its products does not constitute
an endorsement.

Abbott Laboratories
100 Abbott Park Road, Abbott Park, IL 60064-3500, USA
Phone: (708) 937-6100; Fax: (708) 937-2013
Toll-free in U.S.: (800) 323-9100
Internet: http:/ /www.abbott.com
Stock: ABT (NYSE)

Profile: One of the world’s leading health care companies, Abbott is dedicated
to improving people’s lives through the discovery, development, manufacture,
and marketing of health care products and services. Its strategy is to provide
innovative technologies that improve the quality of health care while helping
customers lower their costs.

Products/technologies: A team of scientists developed what they consider a

revolutionary technique for discovering new drugs that could potentially have
far-reaching implications for disease management. Called SAR by NMR
(Structure–Activity Relationships by Nuclear Magnetic Resonance), Abbott
claims the technique can significantly speed the difficult and time-consuming
process for identifying molecules that bind to important protein drug targets.

Abbott compares SAR by NMR to combinatorial chemistry as follows: Both

methods use a building block approach in the construction of molecules; how-
ever, in combinatorial chemistry, thousands to millions of molecules are typi-
cally synthesized and tested for activity. In SAR by NMR, less than 10 mole-
cules is required because the chemistry is highly focused on linking molecules
demonstrated to bind to the protein target using structural information on how
they interact. Patent applications have been filed on the method, but the com-
pany plans to make the technology available to other drug developers.

© 2000 by Marcel Dekker, Inc.

Contact: Ellen Molleston Walvoord, V.P., Investor Relations and Public Af-
fairs

Acacia Bioscience
4136 Lakeside Drive, Richmond, CA 94806, USA
Phone: (510) 669-2330; Fax: (510) 669-2334
E-mail: info@acaciabio.com
Internet: http://www.acaciabio.com

Profile: Acacia Bioscience is a functional genomics and drug discovery com-

pany using advances in genome science and combinatorial chemistry to iden-
tify potential lead compounds with higher degrees of therapeutic value.

Products/technologies: Acacia’s drug discovery technology, called the Ge-

nome Reporter Matrix™, screens molecules against multiple intracellular dis-
ease pathways, revealing the in vivo targets of compounds with therapeutic
value and increasing the efficiency of the drug discovery process. The Matrix
can be used to enhance the screening of combinatorial libraries, detect the
potential side effects of a drug candidate, resurrect failed drug candidates, and
accelerate the discovery of new drug leads.

Contact: Bruce Cohen, President and CEO

ACADIA Pharmaceuticals
3911 Sorrento Valley Blvd., San Diego, CA 92121, USA
Phone: (619) 558-2871; Fax: (619) 558-2872
E-mail: receptor@together.net
Internet: http://www.acadia-pharm.com

Profile: ACADIA Pharmaceuticals is a biotechnology company engaged in

development and use of high-throughput solutions for drug discovery.
Founded in 1993, the company has developed a platform of proprietary break-
through technologies for the functional characterization of genes encoding
potential drug targets. ACADIA pursues drug discovery alliances with major
pharmaceutical firms, as well as with biotechnology companies with expertise
in genomics and combinatorial chemistry. It continues to develop and expand
its technology platform and in-house discovery efforts on novel targets. Re-
search facilities are maintained in both San Diego and Copenhagen, Denmark.

© 2000 by Marcel Dekker, Inc.

The company changed its name in August 1997 from Receptor Technologies
Inc., and moved from Winooski, Vermont, to its California location.

Products/technologies: It has developed a proprietary technology that is

referred to as Receptor Selection and Amplification Technology (R-SAT™).
R-SAT permits the assaying of drug activity using recombinant receptors. The
R-SAT assay is a high throughput assay that can be performed with a wide
range of receptors that use a diversity of signal transduction mechanisms. The
company claims that precise quantitative data for agonist, partial agonists, as
well as neutral antagonists and negative antagonists (inverse agonists) can be
obtained. Company offers contract assays, mass screening, reagents, and a
novel method for expression cloning based on the R-SAT assay.

Contact: Mark R. Brann, Ph.D., CEO and Chief Scientific Officer

Advanced ChemTech
5609 Fern Valley Road, Louisville, KY 40228-1075, USA
Phone: (502) 969-0000; Fax: (502) 968-1000
Toll-free in U.S.: (800) 456-1403
E-mail: info@peptide.com
Internet: http:/ /www.peptide.com

Profile: A manufacturer of synthesis instrumentation, biochemicals, and life

science products, the company is a pioneer in the field of multiple synthesis
technology. It manufactures a diverse line of instrumentation with applications
in molecular diversity and rational design for drug discovery and development,
including commercial models for high throughput organic small molecule and
mixture combinatorial synthesis, as well as numerous proprietary synthesizer
models. In addition to its headquarters in Kentucky, subsidiaries and distribu-
tors are located in Europe, Japan, Singapore/Southeast Asia, Korea, China/
Hong Kong, India, and Taiwan.

Products/technologies: These include the BenchMark Synthesis series of

combinatorial and high-throughput organic synthesis systems (384 HTS, 496
MOS, 440 MOS); ReacTech organic synthesizer; plus other instrumentation
for combinatorial peptide and organic synthesis; catalog of peptide libraries,
custom peptides, and peptide library synthesis services; extensive selection of
amino acid derivatives, solid supports reagents, and other starting materials
for solid phase organic, peptide, and combinatorial synthesis. The ACT Model
496 Multiple Organic Synthesizer™ is a fully automated dual-arm robotics

© 2000 by Marcel Dekker, Inc.

synthesizer that can make up to 96 different compounds in a single run; reac-
tion blocks with 40, 16, and 8 wells are also available. The ACT Model 357
FBS can be used in automated single or multiple organic and bioorganic mo-
lecular synthesis. It can synthesize up to 36 different compounds and can serve
as a large-scale, single-product synthesizer with scales of up to 36 g or more
of starting resin. Combinatorial libraries can also be prepared by the split-pool
method.

Contact: Mark L. Peterson, Ph.D.

Afferent Systems, Inc.

440 Collingwood St., San Francisco, CA 94114, USA
Phone: (415) 647-6659; Fax: (415) 647-6551
E-mail: info@afferent.com
Internet: http://www.afferent.com

Profile: The company develops software for combinatorial chemistry, includ-

ing instrument control, and product data generation, storage, and access. It
also offers solution phase and solid phase combinatorial synthesis instruments
that are low cost and offer a high throughput. Software runs on PCs under
Windows 95 and NT 4.0 and on SGI workstations. Free short-term evaluation
licenses are offered for those interested in evaluating the software.

Products/technologies: Myriad™ integrated system for combinatorial chemis-

try informatics combines multiple functional modules under a single graphical
user interface. The first three modules are a combinatorial database generator,
combinatorial database engine and browser, and a synthetic instrument con-
troller. The Afferent synthesis station is based on Myriad, the Gilson 215
liquid handler, and optionally the Charybdis Calypso reaction block.

Affymax
4001 Miranda Ave., Palo Alto, CA 94304, USA
Phone: (415) 812-8700; Fax: (415) 424-0832
E-mail: webmaster@affymax.com
Internet: http://www.affymax.com

Profile: Affymax is a wholly owned subsidiary of Glaxo Wellcome.

Products/technologies: ESL (Encoded Synthetic Library) technology enables

the identification of individual organic compounds from combinatorial librar-

© 2000 by Marcel Dekker, Inc.

ies containing hundreds to hundreds of thousands of such compounds. Using
ESL, a scientist can synthesize a large library of compounds in combinatorial
mixtures, attaching tags at each reaction step. When an active compound is
found, the tags allow the scientist to readily identify the specific active struc-
ture.

Contact: Gordon Ringold, Ph.D., CEO and Scientific Director, Affymax Re-
search Institute

Affymetrix
3380 Central Expressway, Santa Clara, CA 95051, USA
Phone: (408) 731-5000; Fax: (408) 481-0422
E-mail: sales@affymetrix.com
Internet: http:/ /www.affymetrix.com
Stock: AFFX (NASDAQ)

Profile: Affymetrix started operations in 1991 as a division of Affymax, N.V.

and began operating independently as a wholly owned subsidiary of Affymax
in February 1993. In March 1995, Glaxo Wellcome purchased Affymax, in-
cluding its then 65% interest in Affymetrix. As a result of subsequent financ-
ings, Glaxo Wellcome’s ownership was approximately 34% of Affymetrix as
of August 1997. The company has approximately 150 employees.

Products/technologies: The company’s GeneChip technology has potential

for genomic applications—including linkage, association studies, and DNA
fingerprinting. A collaboration agreement with Whitehead Institute in Cam-
bridge, Massachusetts is focused on the development of a set of single-nucleo-
tide polymorphic (SNP) markers that can be scored on a microchip—this
could have uses in studies on gene expression or as a diagnostic screening
tool to identify even single-point gene mutations.

Alanex Corporation
3550 General Atomics Ct., San Diego, CA 92121, USA
Phone: (619) 455-3200; Fax: (619) 455-3201

Profile: Alanex began operations as a limited partnership in May 1991. In

1993 the partnership was converted to a California corporation. Alanex was
acquired in May 1997 and operates as a privately held subsidiary of Agouron

© 2000 by Marcel Dekker, Inc.

Pharmaceuticals (traded as AGPH on NASDAQ). The company is involved
in the discovery and optimization of small-molecule drugs using its proprietary
combinatorial chemistry, computational chemistry, high-throughput screen-
ing, and integrated information management system. An important element
of the company’s strategy is to enter into collaborations with pharmaceutical
companies.

Products/technologies: Alanex has developed and integrated its proprietary

ChemInformatics system with combinatorial chemistry and high throughput
screening. This combined approach is designed to significantly shorten the
time required to find lead compounds and optimize them into viable drug
candidates. Its proprietary core drug discovery technology, Pharmacophore
Directed Parallel Synthesis (PDPS), combines combinatorial chemistry with
computational and medicinal chemistries that, when used in conjunction with
high-throughput screening and pharmacology, form an integrated drug discov-
ery platform that can be broadly applied to a wide array of biological targets.
Its proprietary library design software, LiBrain, is used to maximize the diver-
sity of its exploratory library by selecting for synthesis compounds from Ala-
nex’s virtual library of chemical structures.

Contact: Ed Baracchini, Ph.D.

Amersham Pharmacia Biotech

Bjorkgatan 30, S-751 82 Uppsala, Sweden
Phone: 46 (0)18 16 50 00; Fax: 46 (0)18 16 64 58
Internet: http://www.apbiotech.com
Stock: Parent companies listed as PH&U on Stockholm Stock Exchange; and
PNU (NYSE)

Profile: In 1997 Amersham International plc and Pharmacia & Upjohn, Inc.
merged their life science businesses, Amersham Life Science and Pharmacia
Biotech, creating the world’s largest research-based biotechnology supplier.
With over $700 million in combined annual sales and more than 3600 employ-
ees, the company will focus on lab and industrial chromatography, industrial
DNA synthesis, high throughput drug screening, custom radiochemical syn-
thesis, molecular biology reagents, DNA sequencing and mapping, and elec-
trophoresis. The new company will focus R&D efforts on technologies for
matrix/polymer chemistry, high throughput sequencing, drug screening, and
microarrays.

© 2000 by Marcel Dekker, Inc.

Products/technologies: The SMART System is optimized for micropurifica-
tion and microbore chromatography. It recovers and purifies biologically ac-
tive material present in subnanogram, nanogram, and picogram amounts. By
scaling down column dimensions and optimizing system components, it de-
livers the low volumes and high concentrations of pure biologically active
substances that life science researchers need for further investigation by mi-
croanalytical techniques. It is also used in peptide sequencing and protein
structure–function applications. The company, along with Molecular Dynam-
ics, developed a microarray system that permits researchers to make and ana-
lyze high-density microarrays with increased speed, efficiency, and sensitivity.
Data from a single microarray experiment provide researchers with the ability
to accurately measure gene expression levels in thousands of samples. In late
1996 the two companies launched the Microarray Technology Access Program
to make available preferential, precommercial access to their microarray tech-
nology.
Contact: Mike Evans, Vice President

Argonaut Technologies
887 Industrial Road, Suite G, San Carlos, CA 94070, USA
Phone: (415) 598-1350; Fax: (415) 598-1359
E-mail: info@argotech.com
Internet: http:/ /www.argotech.com
Profile: Argonaut was incorporated in November 1994 and completed a first-
round venture financing of $4.5 million in January 1995. A second-round ven-
ture financing of $9.5 million was completed in May 1996. In addition to
the research and development facility the company occupies in San Carlos,
California, it also has facilities in Tucson, Arizona, and Basel, Switzerland,
as well as a distributor in Japan.
Products/technologies: The company designs and develops a full range of
instruments, reagents, consumables, and software that apply solid phase tech-
nologies to the synthesis and purification of small organic molecules. Products
are aimed at improving the productivity of the medicinal chemistry laboratory
by exploiting the vast productivity gains derived from high-speed parallel
synthesis and combinatorial chemistry. Products include ArgoGel resins and
ArgoPore resins for solid phase synthesis. The Nautilus 2400 Organic Synthe-
sizer handles complex reactions, has the flexibility to synthesize small mole-
cules using a broad range of chemistries, and handles a wide range of reagents,
with capability for temperature control and use of inert atmospheres. The Quest

© 2000 by Marcel Dekker, Inc.

210 manual synthesizer can run up to 20 parallel reactions and can be used for
solid or solution phase chemistries in a wide range of synthesis applications.
Contact: David P. Binkley, Ph.D., President and CEO

ArQule, Inc.
200 Boston Avenue, Suite 3600, Medford, MA 02155, USA
Phone: (781) 395-4100; Fax: (781) 395-1225
E-mail: jsorvillo@arqule.com
Internet: http://www.arqule.com
Stock: ARQL (NASDAQ)
Profile: ArQule was started in 1993 and completed its IPO in October 1996.
The company created a technology platform for the discovery and production
of chemical compounds with commercial potential, and provides novel com-
pounds to the pharmaceutical and biotechnology industries.
Products/technologies: Proprietary modular building block technology inte-
grates structure-guided drug design, high-speed parallel chemical synthesis;
and information technology to accelerate the identification and optimization
of drug development candidates. The company’s Mapping Array™ products
are proprietary libraries of novel, diverse, small-organic-molecule compounds
used for screening against biological targets in lead generation. Directed
Array™ products are customized libraries of closely related compounds for
lead optimization. Products are based on a technology platform composed of
structure-guided drug design, modular building block chemistry, combinato-
rial chemistry, informatics, and the company’s Automated Molecular Assem-
bly Plant (AMAP™). Scalability of products enables the development process
to accelerate directly into the lead optimization phase.
Contact: Eric B. Gordon, M.D., President and CEO

Aurora Biosciences Corporation

11149 N. Torrey Pines Road, La Jolla, CA 92037, USA
Phone: (619) 452-5000; Fax: (619) 452-5723
E-mail: info@aurorabio.com
Internet: http://www.auroabio.com
Stock: ABSC (NASDAQ)

Profile: Founded in 1995, Aurora’s primary mission is to advance drug discov-

ery by developing miniaturized, automated systems for ultrahigh-throughput

© 2000 by Marcel Dekker, Inc.

screening based on proprietary, versatile, fluorescence-based assays. The com-
pany was incorporated in California in May 1995 and reincorporated in Dela-
ware in January 1996. The company announced its initial public offering of
common stock in June 1997.

Products/technologies: Designs and develops proprietary drug discovery sys-

tems, services, and technologies to accelerate and enhance the discovery of
new medicines. These systems enable very rapid screening of multiple geno-
mic and other molecular targets to quickly and economically identify lead
compounds with novel therapeutic potential. Aurora is developing an inte-
grated technology platform comprising a portfolio of proprietary fluorescent
assay technologies and an ultrahigh-throughput screening system designed to
allow assay miniaturization. According to the company, their ultrahigh-
throughput screening system (UHTSS™) can handle hundreds of thousands
of compounds every 24 hours.

Contact: Paul Grayson, Vice President, Corporate Development

Axiom Biotechnologies Inc.

3550 General Atomics Court, San Diego, CA 92121-1194, USA
Phone: (619) 455-4500; Fax: (619) 455-4501
E-mail: jlinton@axiombio.com
Internet: http:/ /www.axiombio.com

Profile: A privately held company established in August 1995, Axiom is a

drug discovery company that provides pharmaceutical companies with novel
high-throughput pharmacology systems to accelerate the process of drug
discovery. Axiom serves the pharmaceutical industry by offering access to
its novel systems through research collaborations. In such collaborations
Axiom provides research and development expertise, and technology transfer
of its proprietary high-throughput pharmacology systems, including instru-
mentation, cell-based assays, laboratory information management systems,
and analytical pharmacoinformatics tools, to accelerate in-house discovery
efforts.

Products/technologies: These include the High Throughput Pharmacology

System (HT-PS) for drug discovery that uses natural cell lines, which more
closely simulate the body’s environment, to test compounds for a desired
drug activity. The system consists of novel high-throughput pharmacology

© 2000 by Marcel Dekker, Inc.

instrumentation, high-sensitivity cell-based assays, and novel pharmacoinfor-
matics tools for managing and mining high-throughput pharmacology experi-
mental data. Complementing Axiom’s HT-PS system is its LIMS system, built
on an ORACLE platform that can be customized to meet individual labs’
needs.

Contact: James P. Linton, Senior Director, Business Development

AXYS Pharmaceuticals, Inc.

180 Kimball Way, South San Francisco, CA 94080, USA
Phone: (650) 829-1000; Fax: (650) 829-1001
Internet: http://www.axyspharm.com
Stock: AXPH (NASDAQ)

Profile: AXYS is involved in the integration of drug discovery technologies

from gene identification through clinical development and is focused on the
discovery of small-molecule therapeutics.

Products/technologies: Formerly Arris Pharmaceutical Corp., the company

announced in January 1998 its plans to commercialize its capabilities in com-
binatorial chemistry and pharmacogenomics as well as its patented technology.
The company licenses its combinatorial chemistry technology to pharmaceuti-
cal and biotech companies, and collaborates with companies wishing to screen
its libraries.

Contact: Daniel Petree

Beckman Coulter, Inc.

2500 Harbor Blvd. Fullerton, CA 92834-3100, USA
Phone: (714) 871-4848; Fax: (714) 773-8898
Internet: http://www.beckman.com
Stock: BEC (NYSE)

Profile: Founded in 1935, the company develops, manufactures, and markets

automated systems and supplies for life science research and clinical diagnos-
tic laboratories. Through its acquisition of Sagian, Beckman gained access to
high-throughput screening technology and liquid-handling robotics instrumen-

© 2000 by Marcel Dekker, Inc.

tation. In late 1996 the company formed an alliance with MDL Information
Systems to develop an integrated hardware and software system for high-
throughput screening used in the drug discovery process. In October1997,
Beckman acquired Coulter Corp. Beckman Coulter, Inc. will have combined
worldwide sales of more than $1.7 billion.

Products/technologies: These include lab robotics, including ORCA, and

other products from the purchase of the lab robotics division of Sagian, and the
Biomek 2000 biorobotics workstation; 96-channel pipettor; SAMI software,
robotic peripherals, and custom engineering services. New products intro-
duced at the 1997 Pittsburgh Conference (Pittcon) were the Biomek integrated
laboratory automation system for high-throughput drug screening using
ELISA system, a receptor binding assay system, and a cell-based assay system.

Contact: Jay Steffenhagen, Vice President, Planning/Development

Belmont Research Inc.

84 Sherman Street, Cambridge, MA 02140, USA
Phone: (617) 868-6878; Fax: (617) 868-2654
E-mail: software-chan@belmont.com
Internet: http:/ /www.belmont.com

Profile: Belmont Research develops, markets, and supports data visualization,

data management, and rapid application development software. The com-
pany’s worldwide customer base includes leading scientific, engineering, and
business organizations in information-intensive industries such as pharmaceu-
ticals, microelectronics, biotechnology, and government. Belmont helps phar-
maceutical companies solve key database problems such as standardization
and reuse of clinical database definitions and components, global tracking and
reporting of drug adverse events to national regulatory authorities, tracking
of data discrepancies in high-integrity databases, and visualization of chemical
activity databases for new drug discovery.

Products/technologies: These include multidimensional data visualization and

graphical reporting for applications relating to clinical trials, quality control,
and high-throughput screening. The company also provides software con-
sulting services.

Contact: Jim Ong

© 2000 by Marcel Dekker, Inc.

BioDiversity
Brunel Science Park, Uxbridge, Middlesex, UK
Phone: 44 1895 812020; Fax: 44 0895 811051
E-mail: biodiv@dial.pipex.com

Profile: A small UK contract research organization, the company was estab-

lished to address the problems commonly encountered when natural samples
are used for drug screening. The three founding members have expertise in
the field isolation of fungi and actinomycetes, and are developing new methods
for field and laboratory cultivation of unusual microbial groups.

Products: Prescreening analysis and characterization of microbial samples.

Plans to extend and develop its library of organisms and characterized samples
so that a database associated with chemical data can be made available for
industry.

Contact: Neil Porter, Managing Director

BioFocus
130 Abbott Drive, Sittingbourne Research Centre, Sittingbourne, Kent, ME9
8AZ, UK
Tel: 44 1795 412300; Fax: 44 1795 471123
E-mail: admin1@biofocus.co.uk
Internet: http://www.biofocus.com
Stock: Traded on the OFEX market

Profile: BioFocus is a European pharmaceutical company formed in early

1997 to specialize in combinatorial chemistry by applying sophisticated tech-
nologies to the lead discovery and optimization process. Its scientific team
gained its experience in drug discovery and development in one of the world’s
largest pharmaceutical companies. In August 1997 the company had a success-
ful flotation on the OFEX market.

Products/technologies: The company tailors its combinatorial chemistry

methodologies to meet specific client needs. It offers three main levels of
products and services, which include a fully integrated lead discovery, expan-
sion, and optimization service; design and synthesis of high-purity, diverse or
focused representational arrays; and supply of custom-designed combinatorial
synthesis sets.

Contact: Alan Clabon, Marketing and Customer Relations Director

© 2000 by Marcel Dekker, Inc.

Biotage, Inc.
1500 Avon Street Ext., Charlottesville, VA 22902, USA
Phone: (804) 979-2319; Fax: (804) 979-4743
Internet: http:/ /www.dyax.com

Profile: Biotage is a division of the privately held Dyax. In 1995 Biotage, Inc.
and Protein Engineering Corporation were merged to form Dyax for the pur-
pose of developing novel proteins and peptides for use in separations, thera-
peutics, and diagnostics based on proprietary phage display technology. (See
also listing for Dyax.)

Products/technologies: An important part of their overall business plan is par-

allel purification—synthesis and purification; its proprietary technology in-
cludes prepacked flash cartridges. Products include the PFP640™ purification
system, FLASH 40 compression modules, SGM641™ step gradient module,
AFC642™ step gradient module and AFC642 automated fraction collector.
The company also offers the Parallex™ HPLC system for high-throughput
parallel purification, designed to purify libraries, not just single compounds. It
is collaborating with CombiChem in the development of an automated parallel
purification system, the CombiSynx PWS-20x, for combinatorial libraries.

Contact: Robert A. Dishman, Ph.D., President and CEO

Bohdan Automation, Inc.

1500 McCormick Blvd., Mundelein, IL 60060, USA
Phone: (847) 680-3939; Fax (847) 680-1199
E-mail: sales@bohdaninc.com
Internet: http:/ /www.bohdaninc.com

Profile: The company has over 25 years of experience in the field of automa-
tion. After many years of working in the industrial automation area, Bohdan
began automating laboratory activities for the pharmaceutical, chemical, petro-
chemical, food, environmental, and other laboratory areas around 1985.

Products/technologies: Pioneered the development of the automated labora-

tory workstation. Its concept employs a miniaturized version of a gantry style
robot that Bohdan developed in 1978. The benchtop chassis, which can be
customized to a lab’s needs, is used to automate activities in an application-
specific or functionally dedicated fashion. Its workstations are designed to

© 2000 by Marcel Dekker, Inc.

automate all aspects of solution phase and solid phase, small molecule organic
synthesis. Bohdan offers a complete family of automated laboratory work-
stations, such as the Automated RAM™ Synthesizer Workstation, that can
support most weighing applications, has a capacity for hundreds of sample
containers, and can process samples at the rate of 120 per hour. It also markets
a combinatorial chemistry reaction block that accommodates a wide variety
of organic solvents and handles both solid phase and solution phase chemistry.

Bruker Instruments Inc.

19 Fortune Drive, Billerica, MA 01821, USA
Phone: (978) 667-9580; Fax: (978) 667-3954
Internet: http://www.bruker.com

Profile: Bruker Instruments was incorporated in 1960 and is headquartered in

Germany. Since its founding, the company has diversified into a broad spec-
trum of analytical techniques and methods in both research and QC/QA appli-
cations. These include NMR, FTIR, MRI, MS, and EPR.

Products/technologies: Robotic sample introduction for automated MS and

LC-MS includes the HP 1100 HPLC unit (automated LC-FTMS) and the Gil-
son 215 for 96/384 microtiter plates.

Calbiochem-Novabiochem
Boulevard Industrial Park, Padge Road, Beeston, Notts NG9 2JK, UK
Phone: 44 (115) 9430-840; Fax: 44 (115) 943-0951
E-mail: technical@calbiochem.com
Internet: http://www.calbiochem.com

Profile: Calbiochem, Novabiochem, and Oncogene Research Products are

brand names of CN Biosciences (trades as CNBI on NASDAQ). CNBI devel-
ops, produces, markets, and distributes a broad array of products used world-
wide in disease-related life sciences research at pharmaceutical and biotech-
nology companies, academic institutions, and government laboratories.

Products: CNBI’s product lines include biochemical and biological reagents,

antibodies, assays, and research kits that it sells principally through its general
and specialty catalogs. Resins and linker products are used in combinatorial
chemistry.

© 2000 by Marcel Dekker, Inc.

Calipher Technologies Corp.
1275 California Avenue, Palo Alto, CA 94304 USA
Phone: (415) 842-1960; Fax: (415) 841-1970
E-mail: info@calipertech.com
Internet: http:/ /www.calipertech.com

Profile: Founded in June 1995, the company was formed to advance laboratory
techniques in step with genomics and combinatorial chemistry. It was among
the last companies formed by Avalon Ventures before it closed out its venture
capital activities. Caliper’s president and CEO was previously a general part-
ner in Avalon.

Products/technologies: LabChips microminiature analytical system technol-

ogy—their Lab-on-a-Chip technology miniaturizes and automates experi-
ments, such as drug screening and DNA analysis, on microchips the size of a
dime. Chips can screen compound libraries against genomic targets, diagnose
diseases, perform separation procedures, and conduct bioanalysis of FDA ma-
terial. The company also has microfluidics technology.

Contact: Lawrence Bock, President and CEO

Cambridge Combinatorial Ltd.

Cambridge, England
Phone: 44 123 462244

Profile: The company was launched in February 1997 to capitalize on the

increasing trend to outsource elements of the drug discovery process by large
pharmaceutical and emerging biotechnology companies. Its mission is to spe-
cialize in the design, production, and supply of chemical structures for the drug
discovery industry. Oxford Molecular Group PLC (OMG) was fundamental
in the establishment of Cambridge Combinatorial. OMG’s CEO is Dr. Tony
Marchington, brother of Cambridge Combinatorial CEO Dr. Allan March-
ington. Principal shareholders are management, the founding scientists, and
the University of Cambridge. In August 1997, Oxford Molecular Group an-
nounced that it had taken an option to buy Cambridge Combinatorial. The
move is intended to turn OMG from a company that designs software into one
that designs new drugs. In February 1998 the company acquired the exclusive

© 2000 by Marcel Dekker, Inc.

worldwide patent rights to laminar combinatorial chemistry technology from
Pfizer.

Products/technologies: Initial products will be moderately sized libraries of

up to 20,000 compounds in a pure, well-characterized reproducible form. Mil-
ligram batches of each component in a library will be produced at the same
time to provide the end-user with sufficient material for every stage of testing.
Combinatorial libraries include those designed by Cambridge Combinatorial’s
founding shareholder, the Oxford Molecular Group. Laminar combinatory
chemistry technology, acquired from Pfizer, allows the production of thou-
sands of pure chemical compounds in milligram quantities, as single entities
of a known structure, synthesized to the customer’s individual needs.

Contact: Dr. Allan Marchington, CEO

Cellomics Inc.
635 William Pitt Way, Pittsburgh, PA 15238, USA
Phone: (412) 826-3600; Fax: (412) 826-3850
E-mail: taylor@cellomics.com
Internet: http://www.cellomics.com

Profile: Cellomics Inc., formerly BioDx Inc., is a privately held corporation

founded in October 1996. Cellomics’ mission is to improve the efficiency of
the drug discovery process by delivering a cell-based screening platform that
automates target validation and lead optimization using fluorescence-based
assays.

Products/technologies: ArrayScan™ system is the platform for high-content

screening (HCS) approaches for screening combinatorial libraries. Integrated
platform consists of proprietary, fluorescence reagents and assays; a novel,
proprietary, cell-based HCS system; and bioinformatics software. The plat-
form provides deep biological information (time, space, and activity) about a
drug candidate’s physiological impact on specific cellular targets in living
cells. Cellomics comarkets the Carl Zeiss Ultra High Throughput Screening
(UHTS) system and can be combined directly with the Cellomics HCS system
through an exclusive, worldwide alliance.

Contact: D. Lansing Taylor, Ph,D.

© 2000 by Marcel Dekker, Inc.

Charybdis Technologies, Inc.
2131 Palomar Airport Rd., Suite 300, Carlsbad, CA 92009, USA
Phone: (619) 431-5160; Fax: (619) 431-5163
E-mail: srbush@charybtech.com
Internet: http:/ /www.carybtech.com

Profile: According to company statements, Charybdis was started in 1996 to

design, develop, and implement the new standard in combinatorial technology
and informatics, and to apply this paradigm towards the discovery of novel,
potent, and effective chemical entities.

Products/technologies: The Calypso System™, designed for both lead genera-

tion and lead optimization, is an integrated system for high-throughput and
combinatorial organic synthesis, and is based around the Calypso 96/48/24
reaction block. It is capable of integration with a variety of automation plat-
forms including the Myriad software written by Afferent Systems. The Ca-
lypso System gas manifold is a replacement top coverplate for the Calypso
reaction block.

Contact: Thomas J. Baiga, President; Steven R. Bush, Director of Information

Technologies

ChemBridge Corporation
16981 Via Tazon, Suite G, San Diego, CA 92127, USA
Phone: (619) 451-7400; Fax: (619) 451-7401
Toll-free in U.S.: (800) 964-6143
E-mail: chem@chembridg.com
Internet: http:/ /www.chembridge.com

Profile: The company is an international provider of chemical tools for drug

discovery and development. With main offices in San Diego, California, it
also has a research, development, and production infrastructure in Russia.
Members of ChemBridge’s management team have senior executive and se-
nior research experience with leading U.S., European, and Japanese pharma-
ceutical and biotech companies.

Products/technologies: Hand-synthesized small molecule screening libraries

for lead generation offer DIVERSet™96 Program, SCREEN-Set™ Program,
CHERRY-Pick™ Program, and EXPRESS-Pick™ Program; COMBI-

© 2000 by Marcel Dekker, Inc.

TOOLS™ Program for combinatorial chemistry and parallel synthesis; and
custom chemical synthesis and manufacturing.

ChemGenics Pharmaceuticals, Inc.

One Kendall Square, Building 300, Cambridge, MA 02139, USA
Phone: (617) 374-9090; Fax: (617) 225-2997

Profile: ChemGenics is a drug discovery company that applies its two technol-
ogy platforms, Drug Discovery Genomics™ and Advanced Drug Selection
Technologies, to key rate-limiting steps to gene-based drug discovery; it gen-
erates novel targets from disease genes and lead structures from molecular
diversity. The company was established through an alliance that combined
Myco Pharmaceuticals’ leadership in gene technologies and its expertise in
drug discovery with the advanced separations/analysis tools and drug discov-
ery technologies of PerSeptive Biosystems. In January 1997 it was jointly
announced that Millennium Pharmaceuticals would acquire the privately held
ChemGenics Pharmaceuticals as part of their mutual strategy to become a
world leader in drug discovery. Following the merger, the combined company
will employ over 345 people. PerSeptive Biosystems is a principal ChemGen-
ics shareholder. In August 1997, Perkin-Elmer announced its plans to acquire
PerSeptive.

Products/technologies: ChemGenics has state-of-the-art, high throughput

screening capabilities; it has validated its screening technology in combinato-
rial, synthetic chemical, and natural products libraries. The company has a
proprietary natural products drug source and access to compound libraries
through corporate alliances; these complement the combinatorial and pharma-
ceutical compound libraries that Millennium has accessed through its collabo-
rations with Eli Lilly and Wyeth-Ayerst. ChemGenics also has microbial ge-
netics technologies and has developed biocombinatorial methods involving
genetic engineering to enhance chemical diversity.

Contact: Alan L. Crane, Vice President, Business Development

Chemical Computing Group Inc.

1255 University St., Suite 1600, Montreal, Quebec, Canada H3B 3X3
Phone: (514) 393-1055; Fax: (514) 874-9538
E-mail: hayden@chemcomp.com

© 2000 by Marcel Dekker, Inc.

Profile: CCG develops and markets scientific software for protein modeling,
3D bioinformatics, cheminformatics, molecular modeling, combinatorial
chemistry, and high-throughput screening.

Products/technologies: QuaSAR-Binary™, a proprietary technology for the

analysis of high-throughput screening experimental data. The technology can
analyze the results of HTS experiments and the structure of molecules in order
to make predictions regarding the biological activity of chemical compounds.
It can make timely predictions about the biological activity of chemical com-
pounds from large amounts of binary HTS data; it is based on fundamentals
that enable the direct modeling of experimental error and uncertainty. CCG
is the supplier of the Molecular Operating Environment (MOE), which the
company says is the next generation molecular computing system and cross-
platform software system for chemical visualization, simulation, and analysis.

Contact: Bill Hayden, Vice President, Marketing

Chemical Design Ltd.

Roundway House, Cromwell Park, Chipping Norton, Oxfordshire, OX7 5SR,
UK
Phone: 44 1608 6444000; Fax: 44 1608 642244
E-mail: sales@chemdesign.co.uk
Internet: http:/ /www.chemdesn.com

Profile: Chemical Design is one of the oldest molecular modeling and database
companies. Founded in 1983, the company reported over 700 installations of
its software worldwide in 1997. Hardware sales were a strong component of
its business in its early years; however, in 1990, the company moved into
database software development and phased out hardware sales. In addition to
the UK location, offices are also located in the United States and Germany.

Products/technologies: Produces Chem-X combinatorial chemistry software

that is used to facilitate the discovery and design of novel compounds that
exhibit a particular therapeutic activity. ChemDiverse software is a module
for Chem-X. Chem-X is a complete clientserver system for R-group selection,
library design, library registration and enumeration, integrated with biological
data and robotics systems. There are also options for reaction, 2D and 3D
searching. The HTS chemicals collection on CD-ROM provides a source of
diverse molecules for high-throughput screening programs and reagents

© 2000 by Marcel Dekker, Inc.

for combinatorial chemistry; can be used on a wide range of platforms. An
entry level Chem-X/BASE system for MS-Windows (including Windows 95)
or Apple Macintosh (including Power Macintosh) is contained on the CD-
ROM. Additional products include Chem-X/INVENTORY and ChemIdea
module.

ChemStar, Ltd.
Geroev Panfilovtsev str. 20, Bldg. 1, Office 300, Moscow 123514, Russia
Phone: (7 095) 948 5468; Fax: (7 095) 977 5919
E-mail: chemstar@ict.msk.ru
Internet: http://www.glasnet.ru/chemstar/

Profile: ChemStar offers scientific and business cooperation in screening or-

ganic substances for the development of new products to be used in pharma-
ceutical, agricultural, and other fields. Company information indicates that
it is associated with leading chemists of the Commonwealth of Independent
States.

Products/technologies: Offers an extensive database of organic compounds

for screening, as well as samples of organic compounds for high-throughput
screening in quantities ranging from 1 to 1000 mg.

Contact: Geroev Panfilovtsev

Chiron Corp.
4560 Horton Street, Emeryville, CA 94608-2916, USA
Phone: (510) 655-8730; Fax: (510) 655-9910
Toll-free in U.S.: (800) 524-4766
E-mail: Corpcom@cc.chiron.com
Internet: http://www.chiron.com
Stock: CHIR (NASDAQ)

Profile: Founded in May 1981, Chiron is a $1⫹ billion science-driven health-

care company that combines diagnostic, vaccine, and therapeutic strategies
for controlling disease. The company has research programs in gene therapy,
combinatorial chemistry, cancer, infectious and cardiovascular disease, and
critical care through its Chiron Technologies business unit.

© 2000 by Marcel Dekker, Inc.

Products/technologies: Proprietary small molecule combinatorial chemistry
technology, small molecule libraries, and proprietary automated chemical syn-
thesis systems. The Multipin system technology (involving parallel synthesis
of individual peptides on a solid phase support of 96 polyethylene pins arrayed
in a format similar to microtiter plates) is now also being applied to libraries
of small organic molecules. Multipin SPOC kits (solid phase organic chemis-
try) are sold through Chiron Mimotopes.

Contact: Kimberly Kraemer, Manager, Corporate Communications

Chiroscience Group plc

Cambridge Science Park, Milton Road, Cambridge CB4 4WE, UK
Phone: 44 (0)1223 420430; Fax: 44 (0)1223 420440
E-mail: info@chiroscience.com
Internet: http:/ /www.chiroscience.com
Stock: CRO (London Stock Exchange)

Profile: Founded in 1992 to capitalize on its founders’ expertise in chiral tech-

nologies and the anticipated market growth in chiral drugs, the company ac-
quired Darwin Molecular Corp. in December 1996. The move paved the way
for Chiroscience’s entry into genomics, based on Darwin’s integrated gene-
based discovery platform that includes advanced bioinformatics, combinato-
rial chemistry, and cell-based screening capabilities. It continues to expand its
technological knowledge in chiral and medicinal chemistry, drug discovery, and
disease mechanisms, and has used the knowledge to build a broad drug discovery
program. ChiroTech, the company’s technology services business, has devel-
oped relationships with companies in the pharmaceutical industry by solving
complex chiral problems through various partnerships and collaborations.

Products/technologies: Rational drug design capabilities; chiral technology;

mass spectrometry Tag technology. ChiroTech offers the ChiroChem Collec-
tion for application in combinatorial and medicinal chemistry. The collection
of molecules, derived from specifically selected classes of compounds, is be-
ing launched using a phased approach. The regular supply of these structures,
both novel and technically challenging in their synthesis, will complement the
customer drive to establish greater structural diversity. Each new series in the
collection will be accompanied by a diagram illustrating the extent of diversity
achieved.

Contact: Dr. Brian Gennery, Director of Development

© 2000 by Marcel Dekker, Inc.

ChiroTech (See Chiroscience Group)

CombiChem
9050 Camino Santa Fe, San Diego, CA 92121, USA
Phone: (619) 530-0484; Fax: (619) 530-9998
E-mail: cci@combichem.com
Internet: http://www.combichem.com

Profile: Founded in 1994, this privately held company develops and provides
integrated discovery chemistry technologies and services in collaboration with
pharmaceutical, biotechnology, and fine-chemical companies to accelerate the
drug discovery process. In these partnerships, the companies provide the bio-
logical capabilities and Combichem provides the chemical expertise of the
discovery process. Its mission is to provide a highly efficient and successful
approach to the discovery of new drug molecules.

Products/technologies: The company has proprietary technology and products

in the areas of instrumentation for automated multiple-parallel synthesis, ad-
vanced chemical design software for diversity assessment and optimized
chemistries. Products include the Drug Discovery Engine, which accelerates
drug discovery through the integration of medicinal chemistry, small library
design technology, and automation tools. Lead Generation is a process em-
ployed for novel targets with no known leads. Its proprietary Universal In-
former Library generates information and launches the drug discovery cycle.
With Lead Evolution, hypotheses are generated based on available data; it then
matches the hypotheses against its proprietary virtual library and ultimately
synthesizes alternate drug templates. Lead Optimization is a proprietary design
methodology that constructs libraries around a collaborative partner’s lead to
improve it before identifying it as a drug development candidate. CombiChem
is collaborating with Biotage in the development of an automated parallel
purification system, the CombiSynx PWS-20x, for combinatorial libraries.

Contact: Vincent Anido, Jr., Ph.D., President and CEO

ComGenex, Inc.
Hollan Erno u.5, Budapest, H-1136, Hungary
Phone: 36-1-1124-874; Fax: 36-1-214-2310
E-mail: www@cdk-cgx.hu
Internet: http://www.comgenex.com

© 2000 by Marcel Dekker, Inc.

Profile: The company was established in Budapest, Hungary, in November
1992 after an incubatory period of 5 years under CompuDrug Chemistry Ltd.
ComGenex supplies substances for classical and high-throughput screening.
The company is said to have a strong chemical background in solid and solu-
tion phase combinatorial chemistry and adapts its products and services to the
ever changing and growing needs of the market. ComGenex is a sister com-
pany of RoboSynthon, Inc., which markets the MultiReactor™ combinatorial
chemistry workstation for parallel organic synthesis manufactured by Com-
Genex. Offices are also located in Europe, Japan, and the United States.

Products/technologies: The company has more than 15,000 compounds in

stock for high-throughput screening; this stock is dynamically changing by
about 10–30% of the compounds each month. The supply of compounds is
split between external suppliers and in-house synthesis, which is carried out
using its Matrix Technology, a proprietary solution phase combinatorial chem-
istry technology. ComGenex also offers lab services; its labs are located in the
Central Chemical Research Institute of the Hungarian Academy of Sciences.
Additionally, it provides plant extracts, offering a collection of more than 390
extracts from the geographic region of Hungary.

Contact: Dr. Ferenc Darvas, President and CEO

Daylight Chemical Information Systems, Inc.

27401 Los Altos, Suite 370, Mission Viejo, CA, USA
Phone: (714) 367-9990; Fax: (714) 367-0990
E-mail: info@daylight.com
Internet: http:/ /www.daylight.com

Profile: Daylight Chemical Information Systems, Inc. was incorporated in

1987 and grew from the MedChem Project at Pomona College. The invention
of the SMILES language, by Dave Weininger, first at the U.S. Environmental
Protection Agency in the early 1980s and then Pomona, laid the groundwork
for the creation of a new chemical information system. Daylight’s mission
has been to provide high-performance chemical information processing tools
to chemists. New software has been developed continually and the list of avail-
able and supported software continues to grow. Emphasis has been placed on
the Daylight Toolkit, a set of programming libraries constituting a chemical
information infrastructure on which custom applications can be built. The
Toolkit has been used both by Daylight developers to build supported applica-

© 2000 by Marcel Dekker, Inc.

tions and by customers to build custom in-house applications. In addition to
the corporate office in Mission Viejo, the company has a research office in
Santa Fe, New Mexico.

Products/technologies: Daylight Software Release 4.5 is an integrated set of

programs and libraries providing high performance chemical information pro-
cessing, including structural entry, display, distributed data storage, retrieval,
and searching. Release 4.5 runs under Unix with user access via X-Windows
(e.g., Macs, PCs). The company also provides server programs providing net-
work services, and various toolkits (object-oriented programming libraries).
Daylight’s combinatorial library toolkits are used to register, store, and search
designed libraries.

Dyax Corp.
765 Concord Avenue, Cambridge, MA 02138-1044, USA
Phone: (617) 868-0868; Fax: (617) 868-0898
E-mail: dyax@dyax.com
Internet: http://www.dyax.com

Profile: A privately held company, Dyax was formed in 1996 with the merger
of Biotage, Inc., a provider of preparative chromatography products and lo-
cated in Charlottesville, VA, and Protein Engineering Corp., a Cambridge,
MA, biotechnology company. Its ‘‘purification by design’’ philosophy incor-
porates the design and creation of a broad range of systems to effectively meet
any purification challenge.

Products/technologies: The company’s proprietary Phage Display technology

is used for discovery of high-affinity ligands in drug discovery, for develop-
ment of custom affinity purification ligands, and for development of highly
specific diagnostics. Phage display is a method for generating novel molecules
with high binding affinity for virtually any target. As a part of its existing
separations business, the company is exploiting this technology to develop
high-value affinity separation products designed to increase the efficiency and
reduce the cost of pharmaceutical purification. Dyax is also applying phage
display to the discovery and development of a new generation of diagnostic
imaging agents and novel pharmaceutical compounds that bind tightly and
specifically to known diagnostic markers and therapeutic targets. Other prod-
ucts include chromatographic systems, columns, and media for the purification
process; Proprep sanitary protein gradient LC systems have capacities ranging

© 2000 by Marcel Dekker, Inc.

from 0.05 to 50 liters per minute; Parallex parallel purification systems, in-
clude the PFP 640 and the Parallex HPLC system, a high-throughput prepara-
tive HPLC system. See also Biotage listing.

Contact: Pamela Hay, Director of Corporate Development

Dynex Technologies
14340 Sullyfield Circle, Chantilly, VA 20151-1683, USA
Phone: (703) 631-7800; Fax: (703) 631-7816
Toll-free in U.S.: (800) 336-4543
E-mail: webmaster@dynextechnologies.com
Internet: http:/ /www.dynextechnologies.com

Profile: Formerly Dynatech Laboratories, the company has supplied microtiter

technology products and services to research and clinical laboratories in the
healthcare market for more than 30 years. Its products are marketed to scien-
tists and researchers within universities, hospitals, and industry involved in
research assays and clinical diagnostics. Dynex Technologies is a subsidiary
of Thermo BioAnalysis Corporation, a Thermo Electron company.

Products/technologies: These include Microtiter® Plastics (plates, strips, ac-

cessories); Microtiter plate coating systems; Microtiter detection reader sys-
tems, automated Microtiter plate immunoassay processing systems (such as
the Dias™ immunoassay system); Microtiter plate washers; and Revelation
software.

Contact: Monica Durrant, Director

EVOTEC BioSystems GmbH

Grandweg 64, D-22529 Hamburg, Germany
Phone: 49 40 560810; Fax: 49 40 56081222
E-mail: evotec@evotec.de
Internet: http:/ /www.evotec.de

Profile: EVOTEC was cofounded in 1993 by Nobel laureate Manfred Eigen,

Ph.D., of the Max Planck Society, a pioneer in the research of molecular evolu-
tion and one of the first researchers to address molecular diversity through
parallelization and miniaturization. The company’s technology originates from

© 2000 by Marcel Dekker, Inc.

Dr. Eigen’s work and from other Max Planck Institutes. It is also in collabora-
tion with other researchers at academic institutions throughout Europe. Nearly
half of its staff of 70 hold doctorate degrees. The company has been funded
through private capital as well as government grants and early revenues from
research contracts with international companies. EVOTEC makes its technolo-
gies available to third parties under product development or technology and
transfer agreements.

Products/technologies: These include instrumentation and optical technology

supporting ultrahigh-throughput screening. Its optical system provides accu-
racy and flexibility in the detection of molecular interaction. The EVOscreen
instrument incorporates Evotec’s proprietary fluorescence detection technol-
ogy—a technology sensitive enough to detect single molecules, even within
cells, and yield important information about their interactions—with an auto-
mated, miniaturized platform designed for rapid screen configuration. Single-
cell/single-bead selection techniques combine conical fluorescence scanning
and single-bead/living cell retrieval systems, directed molecular evolution
methods, and artificial intelligence–guided chemical synthesis of functional
compounds.

Contact: Christen Hence, Ph.D., CEO

Genevac Ltd.
The Sovereign Centre, Farthing Road, Ipswich IP1 5AP, UK
Phone: 44 1473 240000; Fax: 44 1473 461176
E-mail: sales@genevac.co.uk
Internet: http://www.genevac.co.uk

Products/technologies: Atlas high-throughput evaporators are designed to

meet the demands of combinatorial chemistry, parallel synthesis, natural
products research, and HPLC prep. The new technology and large sample
capacity allows rapid evaporation of up to 1 liter or more of even the most
difficult solvents in a few hours at low or moderate sample temperatures; suit-
able for use with most sample formats, including microtiter plates, test tubes,
and vials.

Contact: Harry Cole, Sales Manager

© 2000 by Marcel Dekker, Inc.

Genelabs Technologies Inc.
505 Penobscot Drive, Redwood City, CA 94063, USA
Phone: (650) 369-9500
Internet: http:/ /www.genelabs.com
Stock: GNLB (Nasdaq)

Profile: Genelabs is a biopharmaceutical company with research focused on

the discovery of small-molecule drugs that act by binding to DNA or RNA
to regulate gene expression or inactivate pathogens. In February 1998 the com-
pany began acquisition of directed combinatorial chemistry compounds
through agreements with Tripos Inc., MDS Panlabs, and SRI International.

Products/technologies: Merlin and Viria assay systems screen combinatorial

chemistry libraries for molecules and subunits of molecules that bind to seg-
ments of genetic material as small as four base pairs. The database will archive
the subunits and the tiny genetic segments they disable. The database created
to archive the subunits and the tiny genetic segments they disable will initially
enable the rapid design of drugs to counteract pathogens employed in biologi-
cal warfare; it will also have application as a database for traditional drug
discovery.

Contact: Cynthia Edwards, Vice President, Research

Genome Systems Inc.

4633 World Parkway Circle, St. Louis, MO 63114, USA
Phone: (314) 427-3222; Fax: (314) 427-3324
Toll-free in U.S.: (800) 430-0030
E-mail: info@GenomeSystems.Com
Internet: http:/ /www.genomesystems.com

Profile: For more than 4 years, Genome Systems has supplied researchers with
genomic clones and technical support. The company makes the libraries it
screens.

Products/technologies: ‘‘Do-it-yourself library screening’’ — perform li-

brary screening with ‘‘Down-to-the-Well’’™ PCRable DNA pools or by using
high density colony filters.

© 2000 by Marcel Dekker, Inc.

Genome Therapeutics Corp.
100 Beaver Street, Waltham, MA 02154, USA
Phone: (718) 893-5007; Fax: (718) 893-8277
E-mail: webmaster@genomecorp.com
Internet: http://www.cric.com
Stock: GENE (NASDAQ)

Profile: The company was founded in 1961; initial research was conducted
under an SBIR grant from the National Science Foundation. The company
develops proprietary gene databases and uses genomic research technologies
to accelerate the development of novel therapeutics, vaccines, and diagnostics.
Through alliances, the company is attempting to bridge the gap between gene
discovery and drug discovery.

Products/technologies: These include proprietary high-throughput ‘‘multi-

plex’’ DNA sequencing; positional cloning; bioinformatics; bacterial geno-
mics; PathoGenome sequence database.

Contact: John Richard, Vice President, Business Development

Gilson, Inc.
3000 W. Beltline Hwy., Middleton, WI 53562-0027, USA
Phone: (608) 836-1551; Fax (608) 831-4451
Toll-free in U.S.: (800) 445-7661
E-mail: sales@gilson.com
Internet: http://www.gilson.com

Profile: Gilson is a manufacturer of specialized analytical instrumentation for

scientific research and industrial markets since the early 1950s. The company
has developed software and instrumentation for HPLC, LC, and sample prepa-
ration technologies. It also produces high-precision pipettes and tips and auto-
mated instruments and systems for increasing throughput and productivity in
the laboratory.

Products/technologies: Products include automated purification and analysis

systems for combinatorial chromatography, including its high-throughput
Combinatorial Chromatography System for HPLC purification and analysis.
Additional product offerings are the UniPoint System software; automated

© 2000 by Marcel Dekker, Inc.

sample preparation products; autosamplers that accept up to 10 deep well or
standard microplates; dual-function autosamplers for small-scale preparative
applications; and graphic sample tracking system to eliminate errors in HPLC
purification of combinatorial products. Its ASPEC XL4 off-line, high capacity
(108 samples/batch) SPE systems are used for purification applications using
GC and LC-MS and combinatorial chemistry techniques. Other products are
a chemical synthesis workstation and a high-throughput personal synthesizer
system for lead generation and lead optimization libraries. Its high-throughput
LC-MS Interface System is for sample preparation, injection and fraction col-
lection.

Hewlett-Packard Co.
Centerville Road, Wilmington, DE 19808, USA
Phone: (302) 633-8696; Fax: (302) 633-8916
Toll-free in U.S.: (800) 227-9770
Internet: http:/ /www.hp.com/go/chem
Stock: HWP (NYSE)

Profile: A major manufacturer of chemical-analysis instrumentation since

1965, H-P’s core competencies are in measurement, computers, and communi-
cations. The company combines a full range of data-handling and standard
networking products with HP and other vendors’ analytical instruments to
provide chemists with integrated information to increase productivity. In No-
vember 1997, HP and IRORI announced an agreement to develop an inte-
grated drug discovery platform that combines IRORI’s microchip synthesis
and screening technologies with HP’s measurement and information portfolio
for chemical analysis.

Products/technologies: Offers an automated combinatorial chemistry analysis

system built around the HP 1100 Series HPLC, with the Gilson 233 XL sam-
pling injector. Its GeneArray Scanner focuses a laser onto a 3-µm section of
a 20-µm probe array on the Affymetrix chip and scans the chip in as little as
5 min. The company claims the GeneArray can read next-generation Gene-
Chip probe arrays with up to 400,000 DNA probe sequences.

Houghten Pharmaceutials (see Trega Biosciences)

© 2000 by Marcel Dekker, Inc.

ICAgen, Inc.
4222 Emperor Boulevard, Suite 460, Durham, NC 27703, USA
Phone: (919) 941-5206; Fax: (919) 941-0813
E-mail: info@icagen.com
Internet: http://www.icagen.com

Profile: Founded in 1992, this privately held company is involved in the re-
search and development of drugs that work by opening and closing ion chan-
nels. The company is currently in partnerships for the discovery of novel ion
channel modulating drugs, and is the first biopharmaceutical company to inte-
grate combinatorial biology with combinatorial chemistry focused exclusively
on ion channel drug discovery and development. ICAgen has discovery pro-
grams in cardiovascular, central nervous system, and immunological disorders.

Products/technologies: These include proprietary biological targets and chem-

ical libraries.

Contact: Kay Wagoner, Ph.D., Chief Executive Officer

IGEN International Inc.

16020 Industrial Drive, Gaithersburg, MD 20877, USA
Phone: (301) 984-8000; Fax: (301) 230-0158
E-mail: gurewitz@igen.com
Internet: http://www.igen.com
Stock: IGEN (NASDAQ)

Profile: Founded in 1982, the company develops, manufactures, and markets

diagnostic systems utilizing its patented ORIGEN technology. The technology
is used in products developed by IGEN as well as its licensees. ORIGEN is
based on the principle of electrochemiluminescence, which uses labels that,
when attached to a biological substance and then electrochemically stimulated,
emit light at a particular wavelength.

Products/technologies: These include the Origen high-throughput screening

system, built around IGEN’s ECL modules (ECLMs). ECLM is expected to
provide the advantages of rapid assay kinetics and improved assay range,
while reducing the volume of critical reagent.

Contact: Herman Spolders, V.P., Business Development and Planning

© 2000 by Marcel Dekker, Inc.

Incyte Pharmaceuticals, Inc.
3174 Porter Drive, Palo Alto, CA 94304, USA
Phone: (650) 855-0555; Fax: (650) 855-0572
E-mail: Webmaster@incyte.com
Internet: http:/ /www.incyte.com
Stock: INCY (NASDAQ)

Profile: The company was incorporated in 1991 by a founding team of scien-

tists and venture capitalists; in 1996, Genome Systems Inc. and Combion Inc.
were acquired. Incyte designs, develops, and markets genomic databases and
services to pharmaceutical companies on a nonexclusive basis. A wholly
owned subsidiary, Genome Systems, Inc., is located in St. Louis, Missouri.
The combined companies employ more than 500 people.

Products/technologies: Incycte designs, develops, and markets genomic data-

bases and services to pharmaceutical companies. The company has proprietary
DNA sequence, gene expression, and microbial genome databases and high-
throughput full-length cloning technology. Its LifeSeq® database is claimed
by the company to be one of the world’s largest sources of genomic data,
containing gene sequence and expression information from normal as well as
diseased cells and tissues from most of the major tissues of the human body.

Contact: Lisa L. Peterson, Director, Business Development

Institute National de la Propriete Industrielle (INPI)

26 bis, rue de Saint-Petersbourg, 75800 Paris Cedex 08, France
Phone: 01 53 04 53 04; Fax: 01 42 93 59 30
Internet: http:/ /www.inpi.fr

Profile: INPI is the French Patent and Trademark Office and the French Regis-
ter of Commerce & Trade. INPI registers patents, trademarks, industrial de-
signs and companies. One of its missions is to make this information more
readily available by providing worldwide access to more than 35 million rec-
ords of technical and business information. In December 1997, INPI and Der-
went Information announced an agreement in principle to create an extensive
structure searchable patent information resource for the pharmaceutical and
chemical communities. Their new venture is aimed at focusing on the identifi-
cation of lead compounds by shortening the time scale during the drug and
chemical development life cycles. It is particularly relevant to those organiza-

© 2000 by Marcel Dekker, Inc.

tions incorporating combinatorial chemistry methodology into their develop-
ment process.

Products/technologies: It is intended that a new file will be available to cus-

tomers by mid-1998 as a fast-alerting, structure-searchable on-line file. INPI
and Derwent plan to extend the backfile coverage in parallel with the current
coverage. Within a couple of years, they anticipate that retrospective coverage
will be a full 20 years, focusing initially on pharmaceutical patents and ex-
tending coverage to all technologies later.

Institute for Scientific Information

3501 Market Street, Philadelphia, PA 19104, USA
Phone: (215) 336-4474; Fax: (215) 386-2911
Toll-free in U.S.: (800) 336-4474
E-mail: custserv@isinet.com
Internet: http://www.isinet.com

Profile: ISI was founded in 1958 and is a database publisher that indexes
bibliographic data, cited references, and author abstracts from scientific, tech-
nical, and medical sources. The company publishes Current Contents® and
the Science Citation Index®. Its objective is to provide immediate desktop
access to the most significant scientific literature, in its entirety, to anyone,
anytime, anywhere.

Products/technologies: Index Chemicus® database allows access to important

reaction diagrams, full bibliographic information, full-length author abstracts,
and a variety of searchable indices; published in one substructure- and text-
searchable database.

Irori Quantum Microchemistry

11025 North Torrey Pines Road, La Jolla, CA 92037, USA
Phone: (619) 546-1300; Fax: (619) 546-3083
E-mail: info@irori.com
Internet: http://www.irori.com

Profile: Established in 1995, the company is venture-funded. It is an early-

stage development microelectronic and microchemistry company focusing on
applying memory technology to solid-state matrices. Irori applies its SMART

© 2000 by Marcel Dekker, Inc.

technology to the development of small-molecule combinatorial libraries. In
November 1997, IRORI and Hewlett-Packard announced an agreement to de-
velop an integrated drug discovery platform that combines IRORI’s microchip
synthesis and screening technologies with HP’s measurement and information
portfolio for chemical analysis.

Products/technologies: SMART system (single or multiple, addressable, ra-

diofrequency tags) and MEMs (memory enhanced matrices); SMART Micro-
spheres, which are used to record information about individual chemical reac-
tions as members of a combinatorial library are synthesized. Current
microchip-based products from IRORI include a combinatorial chemistry
module, a robotic sorter (AutoSortTM-10K), and a chemistry cleavage station
(AccuCleaveTM-96). Initial bioassay and high-throughput screening systems
are scheduled for launch during late 1998 with diagnostic/sensory industry to
follow.

Contact: Michael P. Nova, M.D., President and CEO

Isco, Inc.
4700 Superior Street, Lincoln, NE 68505, USA
Phone: (402) 464-0231; Fax: (402) 464-4543
E-mail: info.sid@isco.com
Internet: http:/ /www.isco.com
Stock: ISKO (NASDAQ/NMS)

Profile: Founded in 1958, the company designs, manufactures, and markets

instruments used by engineers, technicians, and scientists in the laboratory
and in the field to address concerns such as research and development in chem-
istry and biotechnology, water pollution, process monitoring, quality control,
and environmental testing.

Products/technologies: Automated flash systems for combinatorial chemistry

are designed for either high-throughput or high productivity. Both high-
throughput systems, the 10-channel Parallel System and the 5-channel Parallel
System can be used with either a single solvent or multiple solvents. The initial
5- and 10-channel systems offer the ability to collect fractions based on time.
The high productivity systems can run up to 16 or 32 columns with Isco’s
sequential automated flash systems. One version is engineered for single-use
columns, the other for reusable columns or cartridges. Additional products

© 2000 by Marcel Dekker, Inc.

include LC, HPLC, optical detectors, fraction collectors, pumps, electrophore-
sis apparatus, SFE systems and syringe pumps.

Contact: Douglas M. Grant, President and CEO

LC Packings
80 Carolina Street, San Francisco, CA 94103, USA
Phone: (415) 552-1855; Fax: (415) 552-1859
Toll-free in U.S.: (800) 621-2625
E-mail: info@lcpackings.com
Internet: http://www.lcpackings.com

Profile: LC Packings is a privately held company based in San Francisco,

Amsterdam and Zurich. It was founded in Zurich in 1987 with the goal to
develop, manufacture, and commercialize packed microcolumns for use in
HPLC. The company offers a complete range of products for use in microsep-
aration techniques such as micro, capillary and nano LC, including electro-
chromatography, capillary electrophoresis, and LC-MS.

Products/technologies: The mixing routine of the FAMOS™ microsampling

HPLC workstation enables automated method development for sample han-
dling techniques of single beads to produce combinatorial libraries; high-
throughput is achieved by multiple analysis per vial.

LEAP Technologies
P. O. Box 969, Carrboro, NC 27410, USA
Phone: (919) 929-8814; Fax: (919) 929-8956
Toll-free in U.S.: (800) 229-8814
E-mail: info@leaptec.com
Internet: http://www.leaptec.com

Profile: LEAP Technologies has been in the liquid sample loading business
since 1989. The company provides front-end automation for chromatography
(LC and GC), mass spectroscopy, elemental analysis, dissolution testing, and
other analytical techniques. It specializes in applications that demand reliabil-
ity, flexibility, precision, and high-throughput. LEAP works with major chro-
matography and mass spectroscopy companies to provide total integrated solu-
tions for critical automation applications.

© 2000 by Marcel Dekker, Inc.

Products/technologies: LC products include the CTC HTS PALS high
throughput screening LC-MS sample loader; CTC A200LC autosampler for
column-switching LC-MS; FLUX Rheos 4000 (true low-flow low-dead-
volume-LC pump); FLUX compact degassers. For GC, the company offers
the CTC COMBI PAL GC sampler for headspace and liquid injections, CTC
A200SE GC autosampler. Products for dissolution testing include the HIDRA
high-yield fiberoptic UV-Vis dissolution system; the flow-through dissolution
systems (USP Apparatus #4), and the PAL automated sampling system.

Eli Lilly and Company

Lilly Corporate Center, Indianapolis, IN 46285, USA
Phone: (317) 276-2000; Fax: (317) 276-2095
Internet: http:/ /www.lilly.com
Stock: LLY (NYSE)

Profile: Founded in 1876 by Colonel Eli Lilly, the company was incorporated
in 1901. It is a global research–based pharmaceutical corporation dedicated
to creating and delivering innovative pharmaceutical-based healthcare solu-
tions that enable people to live longer, healthier, and more active lives.

Products/technologies: In 1994, Eli Lilly felt they had secured a solid position
in the field of combinatorial chemistry when they acquired Sphinx Pharmaceu-
ticals, which Lilly viewed as having prominent achievements in this technol-
ogy. In March 1997 the company signed a contract with Taisho Pharmaceuti-
cal Co. to introduce technology in the field of combinatorial chemistry,
including the construction of a large-scale, diversified library of chemical com-
pounds, and to efficiently synthesize derivatives.

Contact: Stephen Stitle, Vice President, Corporate Affairs

LJL BioSystems, Inc.

404 Tasman Drive, Sunnyvale, CA 94043, USA
Phone: (408) 541-8755; Fax: (408) 541-8786
Toll-free in U.S.: (888) 611-4555
E-mail: Htinfo@ljlbio.com
Internet: http:/ /www.ljlbio.com

Profile: A privately held company founded in 1988, LJL develops and manu-
factures automated assay systems for high-demand, high-throughput applica-

© 2000 by Marcel Dekker, Inc.

tions in clinical, research, and drug discovery applications. Its plan is to de-
velop and market systems to accelerate drug discovery and to supply
instrumentation, high-value, application-specific reagents and related consum-
ables, applications software for database management, and informatics for the
acceleration of high-throughput screening (HTS). The company is a registered
medical device manufacturer, operates under GMP, and is ISO 9001 certified.
In 1993, LJL was ranked forty-first in the Inc. 500 list of fastest growing
private companies.

Products/technologies: These include high-throughput screening systems de-

signed to alleviate the screening bottlenecks resulting from recent advances
in genomics and combinatorial chemistry; proprietary, high performance re-
agents, and instrumentation for a large majority of the biosassays used in the
HTS market. Through a licensing agreement LJL will incorporate FluorRx’s
fluorescence lifetime (FLT) sensing technology into LJL’s HTS product sys-
tems. FLT technology measures the excited state lifetime, rather than the inten-
sity, of proprietary fluorophores by means of a robust and sensitive method
of fluorescence lifetime measurement, termed phase modulation. In September
1997 the company announced its first HTS product system, Criterion™. The
system consists of Analyst™, an automated, multimode assay detection instru-
ment designed to seamlessly connect to existing robotics systems; a spectrum
of applications-specific reagents and assay kits; assay development and appli-
cation support services. Analyst is capable of performing all four major types
of optical detection assays currently in use (i.e., fluorescence, fluorescence
polarization, time resolved fluorescence, and luminescence assays). It reads
assays in 96- or smaller assay volume, 384-well microtiter plates with no loss
of sensitivity or dynamic range.

Contact: William G. Burton, Ph.D., V.P. of Technology and Business Devel-

opment

Lynx Therapeutics
3832 Bay Center Pl., Hayward, CA 94545, USA
Phone: (510) 670-9300; Fax: (510) 670-9302
E-mail: postmaster@lynxcalif.com

Profile: The company was founded in 1992 by the founder and former CEO
and Chairman of Applied Biosystems, now a division of Perkin-Elmer. Its
objective was to acquire, develop, integrate, and deploy novel molecular,

© 2000 by Marcel Dekker, Inc.

chemical, and information technologies that provide a strong competitive ad-
vantage; and to leverage existing and future technology platforms through
corporate collaborations and contracts in order to fund limited internal product
development programs. Recently formed a Germany biotech startup company
in Heidelberg with pharmaceutical giant BASF. According to the company,
the joint venture, BASF-Lynx Bioscience AG, is a first for Germany and has
the goal of serving as the launching pad for a biotech industry in Germany.
It is also meant to serve as a genomics drug discovery arm for BASF and to
finance the further development of Lynx’s technologies.

Products/technologies: Massively parallel signature sequencing technology

has the power to examine the genetic activity of a single cell.

Contact: Sam Eletr, PhD., Chairman and CEO

MDL Information Systems, Inc.

14600 Catalina St., San Leandro, CA 94577, USA
Phone: (510) 895-1313; Fax: (510) 614-3608
E-mail: webmaster@mdli.com
Internet: http:/ /www.mdli.com
Stock: MDLI (NASDAQ)

Profile: Founded in 1978, MDL supplies integrated scientific information

management systems, databases, and services that are used worldwide in phar-
maceutical and chemical companies and in industries that use chemical prod-
ucts. The company’s name was changed from Molecular Design Limited in
1993, the year it had its IPO. It is an information management software firm
that has a combinatorial chemistry line of products. In late 1996 the company
formed an alliance with Beckman Instruments to develop an integrated hard-
ware and software system for high-throughput screening used in the drug dis-
covery process. MDL is a wholly owned subsidiary of Elsevier Science.

Products/technologies: These include software programs, such as the Avail-

able Chemicals Director, which is a collection of chemical products offered
by more than 200 chemical suppliers and totaling over 400,000 compounds.
Other products include a line of systems for high-throughput discovery (incor-
porates Molecular Informatics’ BioMerge software, a genomic data manage-
ment product); ISIS, a client/server system that helps researchers select di-
verse chemicals for discovery; MDL SCREEN, a client/server product that

© 2000 by Marcel Dekker, Inc.

manages the process of high-throughput screening; and MDL Central Library,
a client/server system for managing combinatorial chemistry libraries.

Contact: Thomas D. Jones, President and COO

MDS Panlabs
11804 North Creek Parkway South, Bothell, WA 98011-8805, USA
Phone: (206) 487-8200; Fax: (206) 487-3878
E-mail: info@mdsintl.com
Internet: http://www.mdsintl.com

Profile: MDS is a health and life sciences company whose products and ser-
vices are used in the prevention, diagnosis, and management of illness and
disease. The company has six operating divisions: MDS Laboratory Services,
MDS Ingram & Bell, MDS Nordion, MDS Communicare, MDS Sciex, and
MDS Pharmaceutical Services. MDS Pharmaceutical Services provides drug
discovery and development services to help the pharmaceutical and biotech-
nology industries to speed new drugs to market. MDS Panlabs is part of this
division and provides discovery of diversified R&D support services. The
Pharmaceutical Services division also offers clinical trial services, contract
research, and specialized research in quality control testing and analytical
R&D.

Products/technologies: For product discovery assistance, the company offers

high-throughput bioactivity screening for rapid identification of new drug
leads; chemical and natural product libraries for drug discovery; and human
DNA products from cell culture for use as disease models. In product support,
MDS has PharmaScreen™ pharmacology profiles for early evaluation of new
drug candidates; and rapid analoging for new lead molecules. Bioprocess de-
velopment services include recombinant DNA microbes for fermentation man-
ufacturing; strain, media, and process optimization studies to reduce produc-
tion costs; high-yielding cultures for production of penicillin, cephalosporin,
and erythromycin antibiotics; and process safety assessment and validation.

Micromass Limited
Floats Road, Wythenshawe, Manchester M23 9LZ, UK
Phone: 44 161 945 4170; Fax: 44 161 998 8915
E-mail: mark.mcdowall@micromass.co.uk
Internet: http://www.micromass.co.uk

© 2000 by Marcel Dekker, Inc.

Profile: Micromass is a worldwide developer, manufacturer, and distributor
of mass spectrometry (MS) instruments. In business since 1969, the company
is a technology leader in the areas of magnetic sector, time-of-flight and quad-
rupole geometry mass spectrometers. When MS is combined with HPLC, LC-
MS gives scientists a powerful tool for analyzing newly synthesized com-
pounds and measuring the metabolites of drugs under investigation. Waters
Corporation, the leader in HPLC technology, acquired Micromass in Septem-
ber 1997. Waters trades as WAT on the NYSE.

Products/technologies: Some product offerings include systems for rapid

analysis of combinatorial libraries and purification of new drug candidates.
These systems are based on Waters HPLC components, Micromass Platform
LC or Platform LCZ benchtop API mass spectrometer and MassLynx™ soft-
ware with special options for combinatorial chemistry, such as OpenLynx™
Diversity and FractionLynx™. The Waters LC-MS Solution for Drug Discov-
ery provides a single-quadrupole solution for combinatorial analysis. Addi-
tionally, Micromass offers triple quadrupole, time-of-flight, magnetic sector,
and hybrid geometry analyzers.

Contact: Mark McDowall, International Marketing Manager

Millennium Pharmaceuticals, Inc.

640 Memorial Drive, Cambridge, MA 02139-4815, USA
Phone: (617) 679-7000; Fax: (617) 374-9379
E-mail: heller@mpi.com
Internet: http:/ /www.bio.com/companies/Millennium.html
Stock: LMNM (NASDAQ)

Profile: The company was founded in 1993. PerSeptive Biosystems is one of

its principal investors. In January 1997 it was jointly announced that Millen-
nium Pharmaceuticals would acquire privately held ChemGenics Pharmaceu-
ticals as part of their mutual strategy to become a world leader in drug discov-
ery. Following the merger, the combined company will employ more than 345
people. PerSeptive Biosystems is also a principal ChemGenics shareholder.
In August 1997, Perkin-Elmer announced its plans to acquire PerSeptive.

Products/technologies: In addition to the technologies that ChemGenics will

bring (as noted earlier in this chapter), Millennium employs large-scale genet-
ics, genomics, and bioinformatics in an integrated, broad-based drug discovery

© 2000 by Marcel Dekker, Inc.

program applicable to all major human diseases both independently, and in
strategic alliances with leading pharmaceutical companies.

Contact: Beverly Holley, Director of Investor and Public Relations

Millipore Corporation
80 Ashby Road, Bedford, MA 01730-2271, USA
Phone: (781) 275-9200; Fax: (781) 533-3301
E-mail: Philip Onigman@millipore.com
Internet: http://www.millipore.com
Stock: MIL (NYSE)

Profile: Millipore makes products utilizing separation technology for the anal-
ysis, identification, and purification of fluids. Products include disc and car-
tridge filters and housings, filter-based test kits, precision pumps, and other
ancillary equipment and supplies. The company’s products are used in the
pharmaceutical and biotechnology industries in sterilization, including virus
reduction and sterility testing of products such as antibiotics and protein solu-
tions; cell harvesting; and isolation of compounds from complex mixtures.
Millipore, which acquired W. R. Grace’s Amicon Separation Sciences, also
makes protein purification tools and semiconductor manufacturing equipment
through its subsidiary, Tylan General.

Products/technologies: MultiScreen® Assay Plate System is a 96-well plate

configuration with membrane filters individually sealed to the bottom of each
well. This allows for incubation and flow-through washes. These plates are
now available with PTFE membranes and a plastic base that is solvent-resis-
tant.

Contact: Philip Onigman

Molecular Devices Corporation

1311 Orleans Drive, Sunnyvale, CA 94089, USA
Phone: (408) 747-1700; Fax: (408) 747-3601
E-mail: info@moldev.com
Internet: http://www.moldev.com
Stock: NDCC (NASDAQ)

© 2000 by Marcel Dekker, Inc.

Profile: Founded in 1983, the company designs, develops, manufactures and
markets proprietary, high-performance, bioanalytical measurement systems,
including consumables, that are designed to accelerate and improve the cost
effectiveness of the drug discovery and development process. Its systems have
applications in many aspects of the therapeutic development process, from
drug discovery and clinical research through manufacturing and quality con-
trol. In 1996 the company acquired NovelTech Systems, Inc., which designs,
develops, manufactures, and markets high-throughput drug screening prod-
ucts.

Products/technologies: Sells bioanalytical systems for research in the life sci-

ences, supporting combinatorial chemistry, genomics, and drug discovery. Its
line of microplate reader systems is aimed at the life science market. The
Fluorescence Imaging Plate Reader (FLIPR) is for high-throughput screening
using live cell targets. SPECTRAmax PLUS is a high-throughput spectropho-
tometer that combines the power and performance of a spectrophotometer with
the flexibility and throughput of a microplate reader in one system. It uses
new technology called PathCheck™.

Contact: Andrew Galligan

Molecular Dynamics, Inc.

928 East Arques Avenue, Sunnyvale, CA 94086-4520, USA
Phone: (408) 773-1222; Fax: (408) 773-1493
Toll-free in U.S.: (800) 333-5703
E-mail: info@mdyn.com
Internet: http:/ /www.mdyn.com
Stock: MDYN (NASDAQ)

Profile: The company is a developer, manufacturer, and international marketer

of systems that accelerate genetic discovery and analysis. The company’s
products increase scientists’ ability to visualize, quantify, and analyze genetic
information. Molecular Dynamics is a founding member of the Genetic Analy-
sis Technology Consortium (GATC), formed to develop and publish global
standards and to provide a unified technology platform to design, process,
read, and analyze DNA arrays. Molecular Dynamics and Amersham Phar-
macia Biotech developed a microarray system that permits researchers to make
and analyze high-density microarrays with increased speed, efficiency, and
sensitivity. Data from a single microarray experiment provides researchers

© 2000 by Marcel Dekker, Inc.

with the ability to accurately measure gene expression levels in thousands of
samples. In late 1996, the two companies launched the Microarray Technology
Access Program to make available preferential, precommercial access to their
microarray technology.

Products/technologies: Combines advanced laser scanning, electro-optical,

and software technologies to produce optical scanners and confocal micro-
scopes. These instrument systems are designed to improve the productivity of
molecular and cell biologists by automating routine procedures, increasing
speed and accuracy of quantitation, and enabling new bioanalytical techniques.
MegaBACE 1000 Genetic Analysis System, a high-throughput system, uses
novel capillary technology for rapid and highly parallel analysis of genetic
samples. The company also has laser fluorescence confocal scanning technol-
ogy for DNA analysis, as well as innovative microfluidic technologies.

Contact: Jay Flatley, President and CEO

Molecular Informatics (See PE Molecular Informatics)

Molecular Simulations Inc.

9685 Scranton Road, San Diego, CA 92121-3752, USA
Phone: (619) 458-9990; Fax: (619) 458-0136
E-mail: solutions@msi.com
Internet: http://www.msi.com

Profile: Founded in 1984, this privately held company provides molecular

modeling and simulation software for both life and materials science research.
The company employs more than 280 people (approximately half of whom
are Ph.D. scientists); it operates sales offices around the world and a research
and development facility in Cambridge, England. MSI’s Combinatorial Chem-
istry Consortium addresses the full scope of the combinatorial chemistry pro-
cess and is focused on maximizing the productivity of library design and anal-
ysis. In February 1998, Molecular Simulations Inc. and Pharmacopeia Inc.
announced a definitive agreement whereby Pharmacopeia will acquire all of
the outstanding stock of MSI. The transaction is expected to be completed in
the second quarter of 1998; upon completion MSI will become a wholly owned
subsidiary.

© 2000 by Marcel Dekker, Inc.

Products/technologies: MSI’s software combines the core atomistic simula-
tion technologies of molecular mechanics and quantum mechanics with molec-
ular visualization, modeling and instrument simulation. Products include C2
Diversity combinatorial chemistry software that provides a guideline for de-
sign and analysis of combinatorial libraries; and SAR information in library
design of large (100,000⫹) enumerated libraries. In October 1997, the com-
pany announced the first commercial release of WebLab ViewerPro, which
brings molecular visualization and analysis to desktop personal computers.
The product runs on Windows 95 or NT, Power Macintosh, and Silicon Graph-
ics platforms, and is used to build and view molecules, proteins, and crystalline
materials, helping scientists to understand molecular structures and properties.
Cerius 3.0 drug design software contains features for drug design, including
the integration of organic and aromatic chemistry for the improved rendering
of pharmaceutical compounds, faster rotation speeds, and a combinatorial
chemistry module. MSI implemented a web-based architecture that will ex-
pand the availability of MSI software products and libraries to any scientist
using a web browser on a desktop computer.

Contact: Brenda Pfeiffer, Corporate Communications

Neugenesis
Manoa Innovation Center, 2800 Woodlawn Dr., Suite 251, Honolulu, HI
96822-1865, USA
Phone: (808) 539-3801; Fax: (808) 539-3804 or 539-3625
E-mail: info@neugenesis.com
Internet: http:/ /www.neugenesis.com

Profile: A private company, Neugenesis was founded in 1992 by several key

individuals who recognized the potential commercial value of Neurospora for
biotechnology applications. It has nine employees and occupies a 1600 square
foot facility for its headquarters and R&D facilities. The company uses propri-
etary combinatorial biology technology to discover and develop new biothera-
peutics, both alone and in partnership with other biopharmaceutical compa-
nies.

Products/technologies: These include a combinatorial biology drug discovery

platform for the development, improvement, and production of therapeutic
proteins; recombinant protein production systems for monomeric, dimeric, and
heteromeric proteins. Neugenesis’ combinatorial biology technology (the

© 2000 by Marcel Dekker, Inc.

CombiKARYON™) is a way to create diversity in complex proteins. Two
important features are directed vs. random libraries to reduce the level of
screening required and provide more efficient libraries overall; and shape/
space enrichment whereby new proteins created through the Neugenesis sys-
tem will exhibit improvements yet maintain the core structure of the protein.

Contact: Lynn M. Uehara, VP, Business Development

NeuralMed, Inc.
2525 Meridian Parkway, Suite 240, Durham, NC 27713, USA
Phone: (919) 549-0270; Fax: (919) 549-0271
E-mail: info@neuralmed.com
Internet: http://www.neuralmed.com

Profile: The company was founded in 1994 in Pittsburgh, Pennsylvania, by

a group of engineers, statisticians, and business executives from several com-
panies specializing in data mining applications. In early 1996, NeuralMed
relocated to Research Triangle Park, North Carolina, to better access the scien-
tific and engineering talent of the area. Its staff has expertise in data mining
technologies, software systems development, system support, and consulting.
Also, it has a broad range of experience in both the private and public sectors,
including international expertise and university teaching.

Products/technologies: Has delivered various applications, including combi-

natorial chemistry predictions, by use of five neural network algorithms for
building predictive models.

Contact: Tom Stallings, President and CEO

Neurogen Corp.
35 N.E. Industrial Road, Branford, CT 06405, USA
Phone: (203) 488-8201; Fax: (203) 481-8683
E-mail: info@nrgn.com
Internet: http://www.bio.com/co/Neurogen.html
Stock: NRGN (NASDAQ)

Profile: Organized around scientists from Yale University and industry, the
company was founded in 1988. It is engaged in the design and development

© 2000 by Marcel Dekker, Inc.

of a new generation of drugs that the company expects will provide improved
treatment for a broad variety of neuropsychiatric disorders, including anxiety,
obesity, schizophrenia, sleep disorders, dementia, depression, stress-related
disorders and epilepsy.

Products/technologies: These include a combinatorial chemistry library. Its

AIDD (Accelerated Intelligent Drug Design) program is a proprietary blend
of combinatorial chemistry with high throughput screening, robotics, and in-
formatics. New drug candidates are developed through the integration of cut-
ting edge neurobiology, medicinal chemistry and molecular biology, and
AIDD.

Contact: Stephen Davis

NeXstar Pharmaceuticals, Inc.

2860 Wilderness Pl., Suite 200, Boulder, CO 80301, USA
Phone: (303) 444-5893; Fax: (303) 444-0672
E-mail: Kdoherty@NeXstar.com
Internet: http:/ /www.nexstar.com
Stock: NXTR (NASDAQ)

Profile: Founded in 1995 with the merger of NeXagen and Vestar, the com-
pany utilizes proprietary compounds to develop therapeutics and diagnostics
to serve unmet medical needs. From NeXagen, NeXstar acquired the SELEX
(Systematic Evolution of Ligands by Exponential Enrichment) combinatorial
chemistry technology.

Products/technologies: Proprietary SELEX combinatorial chemistry technol-

ogy that can produce aptamers, which are modified oligonucleotides that bind
to cancer-specific markers. SELEX is a combinatorial chemistry technology
that can create an extremely large library of aptamers of oligonucleotides (syn-
thetic oligonucleotides) to identify those oligonucleotides with the highest
specificity and affinity for the target. NeXstar’s Blended SELEX technology
converts molecules with low specificity but potentially useful biological activ-
ity into high-specificity drug candidates. A planned extension of SELEX, Par-
allel SELEX will be used to develop orally available small-molecule drug
candidates.

Contact: Michael Burke, Vice President, Business Development

© 2000 by Marcel Dekker, Inc.

Oncogene Science, Inc.
See OSI Pharmaceuticals, Inc.

Oak Samples Trading, Ltd.

5 Sapiernoye pole St., 252042 Kiev, Ukraine
Phone: 380 44 269 3467; Fax: 380 44 269-3467 (local night hours)
E-mail: sdl@public.ua.net
Internet: http://www.osc.edu/ccl/commercial/OST.html

Profile: Oak Samples Trading(OST) company was founded in 1994 to estab-

lish reliable links between Ukrainian chemists and their foreign colleagues.
The main objectives of OST were to facilitate research information exchange
and to encourage investments in Ukranian science. Today the company’s focus
is on providing samples for high-throughput screening. During the course of
their research, Ukrainian chemists synthesize many new compounds that are
potentially active in agricultural and pharmaceutical screens.

Products/technologies: OST’s collection of small molecules was gathered

from about 20 institutions throughout the Ukraine. These molecules are the
results of the creative efforts of nearly 100 Ukrainian chemists. The OST col-
lection of compounds is available as ISIS/Base-formatted diskettes (Windows)
for preview prior to purchase and will be available through FTP and WWW
services.

Ontogen Corp.
2325 Camino Vida Roble, Carlsbad, CA 92009, USA
Phone: (619) 930-0100; Fax: (619) 930-0200
Internet: http://www.ontogen.com

Profile: Ontogen is a pharmaceutical company focusing on advancing tech-

niques for the development of small molecule therapeutics to treat cancer and
diseases of the immune system. The company has approximately 45 em-
ployees.

Products/technologies: For automation of combinatorial chemistry on solid

phase in a modular approach, the OntoBlock central piece of hardware consists
of 96 2-ml polymeric reaction vessels containing beads on which the synthesis
takes place. From there the test components are stored, are characterized, or

© 2000 by Marcel Dekker, Inc.

undergo high-throughput screening. Characterization is done by high speed
electrospray mass spectrometry. The synthesis station uses reagent dispensers
configured with coaxial needles to displace reagents by pressurization with
inert gas.

Contact: Barry E. Toyonaga, Ph.D., President and CEO

OSI Pharmaceuticals, Inc.

106 Charles Lindbergh Blvd., Uniondale, NY 11553-3649, USA
Phone: (516) 222-0023; Fax: (516) 222-0114
E-mail: mhaines@itg.com
Internet: http:/ /www.osip.com
Stock: OSIP (NASDAQ)

Profile: OSI Pharmaceuticals is a drug discovery company utilizing a platform

of proprietary, broadly enabling technologies to facilitate the rapid and cost-
effective discovery and development of novel, small molecule compounds for
the treatment of major human diseases. The company conducts the full range
of drug discovery activities, from target identification to drug candidate. OSI
has a number of fully funded collaborations with various major pharmaceutical
companies. The company was founded in 1983 by nine scientists from NCI.
During 1996 it acquired MYCOsearch, Inc., and Aston Molecules, a UK-based
provider of discovery and pharmaceutical development services to the world-
wide pharmaceutical industry. Effective October 1, 1997, the company’s name
was changed to OSI Pharmaceuticals, Inc. from Oncogene Science, Inc. In
late September 1997, OSI opened its new MYCOsearch Natural Products Drug
Discovery Center in Durham, North Carolina.

Products/technologies: The proprietary gene transcriptional technologies are

used to develop biopharmaceutical products for the diagnosis and treatment
of cancer and other cell control–linked diseases. MYCOsearch, a division of
OSI, added its Natural Products compound library, other chemical libraries,
combinatorial and medicinal chemistry, informatics, molecular modeling, and
pharmaceutical development operations. Additionally, the combinatorial
chemistry technologies developed by Aston Molecules adds to the company’s
ability to optimize lead compounds identified in the company’s live cell–based
high throughput screening systems.

Contact: Matthew D. Haines, Director, Corporate Communications

© 2000 by Marcel Dekker, Inc.

Oxford Asymmetry Limited
151 Milton Park, Abingdon, Oxon OX14 4SD, England
Phone: 44(0)1235 861561; Fax: 44(0)1235 863139
E-mail: sales@oa-od.com
Internet: http://www.oa-od.com

Profile: A privately held company, it comprises Oxford Diversity, Oxford

Asymmetry, and Oxford USA. Oxford Diversity is the company’s combinato-
rial chemistry business; it has drug discovery experience in molecular model-
ing, organic chemistry, and combinatorial techniques leading to the identifica-
tion of active leads. The company specializes in custom manufacturing and
drug development for the pharmaceutical industry. In September 1997, Oxford
Asymmetry expanded its pilot plant capacity and doubled its laboratory space.
In February 1998 the company announced plans to raise funds through a listing
on the London Stock Exchange.

Products/technologies: Offers a range of advanced chemical products and ser-

vices required for rapid drug discovery and development, including everything
from the production of diverse chemical libraries to the supply of multitonnes
of bulk drug substance; and the Prospector series of combinatorial libraries.

Oxford Molecular Group, PLC (OMG)

The Medawar Centre, Oxford Science Park, Oxford OX4 4GA, UK
Phone: 44 1865 784600; Fax: 44 1865 784601
E-mail: products@oxmol.com
Internet: http://www.oxmol.com
Stock: OMG (London Stock Exchange)

Profile: The company, whose name was changed in 1996 from IntelliGenetics,
is a worldwide provider of integrated solutions for drug discovery for compa-
nies and universities. OMG was fundamental in the establishment of a new
company, Cambridge Combinatorial, in February 1997 (see separate listing).
In August 1997, OMG announced that it had taken an option to buy Cambridge
Combinatorial. The move is intended to turn OMG from a company that de-
signs software into one that designs new drugs. Its Collaborative Discovery
division, along with its associate companies, Cambridge Combinatorial Lim-
ited and Cambridge Drug Discovery, provide research services and consulting
in drug design, combinatorial chemistry and high-throughput screening.

© 2000 by Marcel Dekker, Inc.

Products/technologies: OMG offers a wide range of computer-aided molecu-
lar design (CAMD) software, advanced bioinformatics tools, cheminfor-
matics tools, and collaborative research services. Some products now on the
market include PC/GENE® sequence analysis software, InteleGenetics® Suite,
GeneWorks®, GENESEQ™ patent sequence DNA and protein database,
MPSRCH™, BIONET®, MacVector™ sequence analysis software, RS3 Dis-
covery software for corporate chemical information management. DIVA™, a
visualization and analysis tool for chemical and biological data stored in large
databases, reduces the time needed for analyzing large sets of data. OMIGA,
the first in a family of integrated, enterprise-wide software tools and a compre-
hensive sequence analysis tool, is available for Windows 95 and Windows
NT 4 platforms. Through its Collaborative Discovery Division, the company
provides software and service solutions to the drug discovery industry.

Contact: Dr. Tony Marchington, CEO

Packard Instrument Company, Inc.

800 Research Parkway, Meriden, CT 06450, USA
Phone: (203) 238-2351; Fax: (203) 639-2172
Toll-free in U.S.: (800) 323-1891
E-mail: webmaster@packardinst.com
Internet: http:/ /www.packardinst.com

Profile: Packard is a subsidiary of Canberra Industries, Meriden, Connecticut.

Its instruments, reagents, and applications support are sold to organizations
that conduct research and develop products in life science, diagnostics, and
environmental analysis. The company employs more than 550 people world-
wide with 50 employees at its corporate headquarters and 275 at its manufac-
turing facility in Downers Grove, Illinois.

Products/technologies: The company offers a complete line of microplate

reader instrumentation. Discovery, a nonisotopic high-throughput screening
analysis system, is based on a Nobel Prize–winning chemistry designed to
make drug discovery faster, safer and more cost effective. Technology can be
applied to a screening process called Homogeneous Time Resolved Fluores-
cence (HTRF), which can replace older, conventional radioisotopic processes,
reducing exposure risks and waste disposal costs. The patented HTRF chemis-
try is from CIS Biointernational of France. Using Packard’s high capacity
microplate analyzer Discovery, designed specifically for automated drug

© 2000 by Marcel Dekker, Inc.

screening, HTRF can screen 50,000 samples per day, compared to only 10,000
using conventional technology. HTRF incorporates photon-counting electron-
ics, dual photomultiplier tubes, and time-delayed measurements to eliminate
background fluorescence. MultiPROBE® VersaTip™ technology will sense
small sample volumes in microplates.

Contact: Richard McKernan, President

Pangea Systems, Inc.

1999 Harrison Street, Suite 1100; Oakland, CA 94612, USA
Phone: (510) 628-0100
E-mail: info-pr@PangeaSystems.com
Internet: http://www.pangeasystems.com

Profile: A bioinformatics company, Pangea applies information technology

to biological research. Software applications integrate data with analysis and
visualization tools for biological and chemical information. Incorporated in
1993, it is a privately held company.

Products/technologies: EcoCyc 4.0 is a bioinformatics metabolic pathways

dataset for E. coli and is available through the world wide web; links approxi-
mately 2000 more genes than the previously available version. PathoLogic™
creates metabolic pathway datasets of other pathogenic organisms. Gene-
World® is an automated, high throughput application for analysis of DNA and
protein sequences; GeneMill™ is a workflow management application de-
signed for DNA sequencing laboratories. GeneThesaurus™ is a sequence and
annotation data warehouse containing information from multiple sources and
integrates common public gene and protein sequence and protein structure
databases.

Contact: Marie Martin

Panlabs, International
11804 North Creek Parkway, S., Bothell, WA 98011-8805, USA
Phone: (206) 487-8200; Fax: (206) 487-3787
E-mail: panlabs@panlabs.com
Internet: http://www.panlabs.com
Stock (parent company): MHG.A, MHG.B (TSE)

© 2000 by Marcel Dekker, Inc.

Profile: Founded in 1970, this contract research organization (CRO) became
a wholly owned subsidiary of MDS Health Group Ltd., Canada, in 1995. In
mid-1995 Panlabs joined with Tripos in a collaboration to offer pioneering
services in drug discovery to pharmaceutical, biotechnology, and chemical
research organizations.

Product/technologies: Provides discovery services in the form of screening

assays, screening materials from chemical libraries, plants, and microbes with
robotic screening capacity of 100,000 compounds per week and pharmacologi-
cal profiling of lead compounds on a contractual basis. Products and services
include PharmaScreen®, DiscoveryScreen®, ProfilingScreen™, Immuno-
Screen™, pharmacology discovery and profiling, high throughput screening
(HTS). The company has proprietary technologies to synthesize, extract and
deliver the chemical libraries in 96-well plates for screening. Along with Tri-
pos, the company markets Optiverse™ screening libraries for HTS programs.

Contact: Dr. Christopher Ball, Chairman and CEO

Perkin-Elmer Corp. / PE SCIEX / PE Applied Biosystems

761 Main Ave., Norwalk, CT 06859, USA
Phone: (203) 762-1000; Fax: (203) 762-6000
Internet: http:/ /www.perkin-elmer.com
http:/ /www.pesciex.com
Stock: PKN (NYSE)

Profile: P-E is a world leader in the development, manufacture, and marketing

of life science systems and analytical instruments used in markets such as
biotechnology, pharmaceuticals, environmental testing, chemicals, food, agri-
culture, and chemical manufacturing. The company had sales of more than
$1.3 billion in fiscal year 1997 and employs 5700 people worldwide. Perkin-
Elmer Sciex Instruments (PE SCIEX) is a joint venture between MDS SCIEX
and P-E. PE SCIEX is a leader in the development of mass spectrometers for
use in the analytical and life sciences. In 1996, PE signed an exclusive licens-
ing agreement for the development of LXR Biotechnology’s patented scanning
laser digital imaging (SLDI) technology; the Scanning Laser Digital Imager
images cells and tissues. In July 1997, PE Applied Biosystems, which makes
automated DNA analysis systems, signed an agreement with Tecan U.S. to
develop and sell systems for automating combinatorial chemistry. PE Applied
Biosystems will focus on marketing and support of high speed parallel synthe-

© 2000 by Marcel Dekker, Inc.

sis systems developed by both companies. Tropix, an operating unit of PE
Applied Biosystems Division, established an additional laboratory in 1997 to
develop customer screening assays for the pharmaceutical industry. In Novem-
ber 1997, P-E acquired Molecular Information, Inc., a developer of bioinfor-
matics software. P-E acquired an additional 14.5% stake in Tecan AG in De-
cember 1997, increasing its shares to 52%. In January 1998, P-E acquired
PerSeptive Biosystems. (See separate listings for acquired companies).

Products/technologies: P-E offers integrated, high-throughput, and high-

performance systems for DNA amplification and analysis, and LC-MS sys-
tems used by pharmaceutical companies in research and combinatorial chemis-
try. The company has developed a comprehensive set of enabling tools for
pharmaceutical research—tools that compress the time and reduce the cost of
pharmaceutical drug discovery and development. Perkin-Elmer Cetus has a
worldwide marketing agreement through SCIEX with MDS Health Group
Limited, Canada, for its API III MS analyzer. Under development are various
research projects relating to software, software systems, and automated data
management and analysis systems to speed the drug discovery process. Other
products include the PE-SCIEX API 150 MCA (Mass Chromatographic Ana-
lyzer) LC-MS, Open MS™ control; and purpose-designed workstations, such
as drug discovery or combinatorial chemistry built around the MCA. Tropix,
which has developed proprietary drug discovery assay systems on a limited
basis, now makes this service available to the worldwide pharmaceutical in-
dustry.

Contact: Tony L. White, Chairman, President and CEO

PE Molecular Informatics
1800 Old Pecos Trail, Suite M, Santa Fe, NM 87505, USA
Phone: (505) 995-4475; Fax: (505) 982-7690
Internet: http://www.molinfo.com

Profile: Acquired by Perkin-Elmer in November 1997, Molecular Informatics

provides comprehensive bioinformatics software systems to manage, inte-
grate, and analyze the vast amounts of information flowing into the scientific
discovery process. The company is a pioneer in the development of infrastruc-
ture software for pharmaceutical, biotechnology, and agrochemical industries
that perform genome data collection and management, genome mapping, drug

© 2000 by Marcel Dekker, Inc.

discovery, and clinical and diagnostic genetic research. The company is a spin-
off of the National Center for Genome Resources, Santa Fe, New Mexico.

Products/technologies: Currently provides two products for the management

of genomic data. The BioMerge™ system consists of an advanced, object-
oriented relational database and a group of programs that organize public,
proprietary, and third-party data in a single database. The system provides for
editing and annotating sequence data. An optional Auditor database captures
a history of changes made to your database. Customers can purchase automatic
daily updates of public data with the ‘‘Genomes Today’’ option. BioLIMS™,
developed for the Applied Biosystems (ABI) division of Perkin-Elmer Corpo-
ration, manages automated DNA sequencing for ABI’s Prism™ DNA sequenc-
ing instruments. With BioLIMS, sequence analysis results are entered into a
central database, rather than as thousands of separate files.

Contact: Edward Cantrell

PerSeptive Biosystems, Inc.

500 Old Connecticut Path, Framingham, MA 01701, USA
Phone: (508) 383-7700; Fax: (508) 383-7851
Toll-free in U.S.: (800) 899-5858
E-mail: tsupport@pbio.com
Internet: http:/ /www.pbio.com

Profile: The company was founded in 1988 to commercialize Perfusion Chro-

matography® products based on discoveries in particle coatings and design.
Numerous acquisitions followed; the most recent being that of PerSeptive
Technologies II Corporation in March 1996, which strengthened PerSeptive’s
position in drug discovery. Millennium Pharmaceuticals, which is partially
owned by PerSeptive, announced in January 1997 that it would acquire pri-
vately held ChemGenics Pharmaceuticals as part of their strategy to become
a world leader in drug discovery. PerSeptive Biosystems is also a principal
ChemGenics shareholder. In August 1997, Perkin-Elmer (see separate listing)
announced its plans to acquire the company; the transaction was completed
in January 1998.

Products/technologies: These include Perfusion Chromatography® technol-

ogy; advanced workstations, chemistry and other tools for the biotechnology
industry; robotics interface device for preparing biological samples for analy-

© 2000 by Marcel Dekker, Inc.

sis by mass spectrometry. The MARINER™ LC/MS Biospectrometry™
Workstation for combinatorial chemistry is targeted for high-throughput appli-
cations in characterizing proteins and peptides, in metabolic studies, in
combinatorial chemistry strategies, and in high-throughput screening. Se-
lectronics, a proprietary drug discovery technology, combines multistep chro-
matography, affinity selection, and mass spectrometry for high-throughput se-
lection of new drug lead compounds from a variety of chemical sources
including chemical combinatorial libraries, natural product libraries, and syn-
thetic chemical files.

Contact: John F. Smith, President

Pharmacopeia, Inc.
101 College Road E., Princeton, NJ 08540, USA
Phone: (609) 452-3600; Fax: (609) 452-2434
Internet: http://www.pcop.com
Stock: PCOP (NASDAQ)

Profile: Founded in 1993, the company is active in the field of drug discovery
using small-molecule combinatorial chemistry. It went public in December
1995. In February 1998, Pharmacopeia and Molecular Simulations, Inc. an-
nounced a definitive agreement whereby Pharmacopeia will acquire all of the
outstanding stock of MSI.

Products/technologies: These include discovery libraries of combinatorial

chemistry collections of novel small molecules for screening. Using Encoded
Combinatorial Libraries on Polymeric Support (ECLIPS™), its proprietary
tagging technology, the company generates large libraries consisting of mil-
lions of diverse individually labeled small molecules. Using these libraries,
Pharmacopeia is developing three potential profit centers. These include the
licensing of libraries to pharmaceutical companies for evaluation in multiple
drug discovery programs; the identification and optimization of lead com-
pounds for specific targets provided by customers; and the licensing to phar-
maceutical companies of drug development candidates developed in the com-
pany’s internal drug discovery programs.

Contact: Lewis J. Shuster, Executive Vice President, Corporate Development

and Chief Financial Officer

© 2000 by Marcel Dekker, Inc.

Pharm-Eco Laboratories, Inc.
128 Spring Street, Lexington, MA 02173-7800, USA
Phone: (781) 861-9303; Fax: (781) 861-9386
E-mail: main@pharmeco.com
Internet: http:/ /www.pharmeco.com

Profile: This privately held company was founded in 1972. It is a full-service

drug synthesis and chemical services company that performs a variety of labo-
ratory, process scale-up, and manufacturing tasks including development of
processes and synthesis routes for new medicinal products, validating bulk
pharmaceutical processes, and authoring Drug Master Files. The company also
has a pilot plant/small volume manufacturing site in North Andover, Massa-
chusetts.

Products/technologies: This contract research and development group offers

GMP production; Drug Master Files; validation of analytical methods;
multistep organic synthesis development; and consultancy services, such as
during clinical trials.

Contact: Susie Boast

Polymer Laboratories Ltd.

Essex Rd., Church Stretton, Shrophire SY6 6AX, UK
Phone: 44 01694 723581
E-mail: PL@polymerlabs.com
Internet: http:/ /www.polymerlabs.com

Profile: The company was founded in 1976 to develop techniques and instru-
mentation for the characterization of polymer systems, and to develop high-
technology polymer products for chromatography, and for diagnostic and
pharmaceutical applications. The company operates from offices in the USA,
UK, and Europe. Primary products are in the specialized areas for gel
permeation/size exclusion chromatography (GPC/SEC) including columns,
standards, advanced high temperature systems and detectors, and GPC/SEC
software.

Products/technologies: PL sells a range of combinatorial chemistry resins

consisting of polystyrene resins with chloromethyl functionality (PL-CMS)
for use in solid phase synthesis of peptides, small-molecule libraries, and other

© 2000 by Marcel Dekker, Inc.

solid-supported organic reactions. Larger particles with higher loading can
also be manufactured using the same approach. PL-CMS resins are available
in sizes up to 300 µm in diameter with up to 4 meq/g of loading.

Contact: Dr. Frank Warner, President

Porvair Sciences Ltd.

Unit 6, Shepperton Business Park, Govett Avenue, Shepperton, Middlesex
TW17 8BA, UK
Phone: 44 1932 240255; Fax: 44 1932 254393
E-mail: porvair-paul@classic.msn.com
Internet: http://www.porvair-sciences.com

Profile: The company has used its expertise in plastics technologies and mold-
ing processes to engineer the Microlute plate to meet the needs of researchers
and laboratory analysts who require SPE sample preparation.

Products/technologies: Porvair sells the Microlute™ Solid Phase Extraction

in a Microplate system that provides 96 solid phase extractions in one compact
unit (using any brand of sorbent). It can be automated using most standard
liquid handling and robotic systems. Porvair’s 384-well plate is compatible
with most automated liquid handling instruments, readers for EIA, fluores-
cence, luminescence, and scintillation assays as well as robotic-handling de-
vices.

Contact: Tony Castleman

Protogene Laboratories, Inc.

4030 Fabianway, Palo Alto, CA 94303, USA
Phone: (415) 842-1888; Fax: (415) 857-9229
Internet: http://www.protogene.com

Profile: Was founded as a spin off of an earlier company by the same name.
The company’s parallel array synthesis technology was originally developed
for DNA synthesis by one of the company founders and others at Stanford
University’s Human Genome Center under a Department of Energy Human
Genome grant. After commercializing the technology and becoming the
world’s largest manufacturer of DNA, the DNA business was sold to its mar-

© 2000 by Marcel Dekker, Inc.

keting partner, Life Technologies, Inc. The company announced a collabora-
tion with Ribozyme Pharmaceuticals, Inc. in December 1996; one goal is to
create an instrument that will synthesize thousands of RPI’s proprietary
nuclease resistant ribozymes, combinatorial libraries of other compounds, and
methods that allow for low-level quantitative detection of RNA in cells.

Products/technologies: Protogene Laboratories focuses on a world scale ap-

proach to synthesis facilities on systems for synthesizing DNA on surfaces,
and on combinatorial and high-throughput synthesis of drug libraries.

Contact: Dr. Robert Molinari, President and CEO

Receptor Technologies, Inc.

(See ACADIA Pharmaceuticals)

Regis Technologies, Inc.

8210 Austin Avenue, Morton Grove, IL 60053, USA
Phone: (847) 967-6000; Fax: (847) 967-5876
Toll-free in U.S.: (800) 323-8144
E-mail: whelko@aol.com
Internet: http:/ /www.registech.com

Profile: Founded in 1956, this privately held company has served as a partner
in the development, scale-up, and manufacture of complex, multistep, custom
synthetic products for clients in the pharmaceutical, diagnostic, and biotech
industries. Over the years the company has merged its synthetic proficiency
and chromatographic capabilities to offer a preparative separation service de-
signed for the resolution of compounds on a gram to kilogram scale. It devel-
ops proprietary separations technology for the life sciences industry and pro-
vides contract manufacturing for pharmaceuticals.

Products/technologies: IAM.PC Drug Discovery chromatography column

(immobilized artificial membrane, or IAM) is a screening method for the high-
throughput estimation of drug permeability. IAM chromatography phases pre-
pared from PC analogs closely mimic the surface of a biological cell mem-
brane, which makes IAM useful for the study of drug–membrane interactions.

Contact: Louis Glunz, Ph.D., President

© 2000 by Marcel Dekker, Inc.

Ribozyme Pharmaceuticals, Inc.
2950 Wilderness Pl., Boulder, CO 80301, USA
Phone: (303) 449-6500; Fax: (303) 449-6995
E-mail: webmaster@rpi.com
Internet: http://www.rpi.com
Stock: RYZM (NASDAQ)

Profile: Founded in 1992, the company commercializes patented ribozyme

technology in the human therapeutic, agricultural, animal health, and diagnos-
tic fields, through both internal efforts and collaborations with strategic part-
ners. It announced a collaboration with Protogene Laboratories, Inc. in Decem-
ber 1996; one goal is to create an instrument that will synthesize thousands
of RPI’s proprietary nuclease resistant ribozymes, combinatorial libraries of
other compounds, and methods that allow for low-level quantitative detection
of RNA in cells.

Products/technologies: Proprietary technology for the synthesis of ribozymes;

involved in the application of ribozymes to human therapeutics—automated
parallel array technologies.

Contact: Ralph E. Christoffersen, Ph.D., President and CEO

Robosynthon, Inc.
1105 Grandview, S. San Francisco, CA 94080, USA
Phone: (650) 244-0799; Fax: (650) 244-0795
E-mail: info@robosynthon.com
Internet: http://www.robosynthon.com

Profile: The company is a sister company of ComGenex, Inc., which manufac-

turers the MultiReactor™ workstation for parallel organic synthesis marketed
by Robosynthon.

Products/technologies: The MultiReactor is the first in a line of combinatorial

chemistry workstations for parallel organic synthesis. It allows chemists to
synthesize 24 reactions simultaneously with precise temperature and magnetic
stirring in all tubes. The MaxiReactor™ scales syntheses discovered with the
MultiReactor. It can run six synthetic reactions simultaneously, handle up to
500 ml and gram quantities, claims better accuracy and uniformity than oil
baths, has independently controlled mixing and heating zones, and offers data

© 2000 by Marcel Dekker, Inc.

logging for regulatory compliance. Robosynthon also sells a wide range of
classical synthetic glassware.

Rosetta Inpharmatics
12040 115th Avenue, NE, Kirkland, WA 98034, USA
Phone: (425) 823-7300; Fax: (425) 821-5354

Profile: A biotechnology startup, the company raised initial funding in 1997

from U.S. and international investors. It plans to build ‘‘biochips’’ holding as
many as 100,000 strands of synthesized genetic material to be used along with
proprietary analytical techniques to help major pharmaceutical companies
avoid costly false starts in developing new drugs. The two initial target areas
are cancer and immunological diseases.

Products/technologies: Its key pieces of technology are an ink-jet oligon-

ucleotide synthesizer and analytical tools developed by the company’s cofoun-
ders to assess a drug candidate’s desirable and undesirable effects. The synthe-
sizer will be used to build chip-like arrays with thousands of tiny wells. Each
well holds a string of genetic material, built up by directing the device to
sequentially spray the four nucleotides, or building blocks of DNA, much as
a color ink-jet printer sprays four shades of ink. With such arrays, the company
will test drug candidates to assess their effects on various segments of the
genome.

Contact: Stephen Friend, President and Chief Scientific Officer

Shimadzu Scientific Instruments

7102 Riverwood Drive, Columbia, MD 21046, USA
Phone: (410) 381-1227; Fax: (410) 381-1222
Toll-free in U.S.: (800) 477-1227
E-mail: webmaster@shimadzu.com
Internet: http:/ /www.shimadzu.com

Profile: Shimadzu Corporation, headquartered in Kyoto, Japan, was estab-

lished in 1875. It is one of the world’s largest manufacturers of analytical
instrumentation. It entered the North American market in 1975 with the forma-
tion of Shimadzu Scientific Instruments.

© 2000 by Marcel Dekker, Inc.

Products/technologies: Offerings include the MTP-40 autosampler for direct
sampling from wells to vials in combinatorial projects; and the MTP-10A
autosampler for direct sampling from 96-well microplates eliminating the
need to transfer samples from wells to vials in combinatorial chemistry proj-
ects.

SIBIA Neurosciences, Inc.

505 Coast Blvd. South, Suite 300, LaJolla, CA 92037-4641, USA
Phone: (619) 452-5892; Fax: (619) 452-1609
E-mail: mdunn@sbia.com
Internet: http://www.sibia.com
Stock: SIBI (NASDAQ)

Profile: The company is engaged in the discovery and development of novel

small-molecule therapeutics for the treatment of neurodegenerative, neuropsy-
chiatric, and neurological disorders. It is involved in the development of pro-
prietary drug discovery platforms that combine key tools necessary for modern
drug discovery, including genomics, high-throughput screening, advanced
combinatorial chemistry, and pharmacology. The company’s proprietary mo-
lecular targets and drug candidates, together with its drug discovery technolo-
gies and research expertise, have enabled the company to establish several
corporate collaborations.

Products/technologies: Proprietary high-throughput screening technology

consists of automated optical detection systems for primary screening assays
and the profiling of compounds to identify new drug leads.

Contact: Michael J. Dunn, Vice President, Business Development

SIDDCO (See Systems Integration Drug Discovery Co., Inc.)

Soane Biosciences, Inc.

3906 Trust Way, Hayward, CA 94545-3716, USA
Phone: (510) 293-1855
E-mail: mfranklin@soane.com
Internet: http://www.soane.com

© 2000 by Marcel Dekker, Inc.

Profile: In 1997 the company purchased substantially all the assets of privately
held Seurat Analytical Systems, located in Sunnyvale, California.

Products/technologies: The high-throughput sample-handling devices for

drug discovery applications provide a means of transferring vast numbers of
samples in parallel from the macro world of microtiter plates to the micro
world of chips.

Contact: Roger Noesky, CEO

Spark Holland B.V.

P. de Keyserstraat 8, NL-7825 VE Emmen, The Netherlands
Phone: 31 591-631700; Fax: 31 591-630035
E-mail: sales@spark.nl

Profile: The company is a maker of HPLC equipment and specializes in selling

autosamplers and automated SPE instruments.

Products/technologies: Midas autosampler with 96-tray capacity.

SRI International
333 Ravenswood Avenue, Menlo Park, CA 94025, USA
Phone: (415) 859-3000; Fax: (415) 859-2889
Toll-free in U.S.: (800) 982-8655
E-mail: bdd@sri.com
Internet: http:/ /www.sri.com/pharmaceutical

Profile: Founded as Stanford Research Institute in 1946, SRI is a private,

nonprofit contract research organization and employs over 2400 individuals
worldwide. In addition to its California location, the company also has offices
in Europe and Asia.

Products/technologies: SRI has combinatorial chemistry methods for finding

and optimizing advanced lubricants and coatings formulations that claim to
allow scientists to formulate and characterize hundreds of thousands of differ-
ent fluids, identifying those with desired rheological properties. SRI said this
represents a modification of combinatorial technology, including miniature
biosensors, developed over the last several years for drug screening and clini-

© 2000 by Marcel Dekker, Inc.

cal diagnostics. The technology is capable of preparing and analyzing different
variations of mixtures with up to 20 compounds; the picoliter samples are then
simultaneously measured for viscosity and yield stress. SRI also developed a
parallel screening technique based on the emission and absorption of visible
light. The technique produces spectral information on reaction products and
can be used for screening catalytic activity. The company is attempting to
extend the method to use ultraviolet and near-infrared light, which would
broaden the types of compounds and information that can be screened.

Contact: Pieter Bax, Ph.D., V.P., Biopharmaceutical Development Division

Structural Bioinformatics, Inc.

10929 Technology Place, San Diego, CA 92127, USA
Phone: (619) 675-2400; Fax: (619) 451-3828
E-mail: webmaster@strubix.com
Internet: http://www.strubix.com

Profile: Founded in 1996, this privately held company is a developer of leading

edge supercomputer-based technology for rapid conversion of novel gene se-
quence information into protein structural information and drug lead com-
pounds.

Products/technologies: A supercomputational operating system makes possi-

ble the immediate and practical use of genomic (gene sequence) data in a
broad range of structure-based drug discovery and design processes leading
to the rapid design and identification of small-molecule lead compounds. Its
proprietary SBd-Base allows comparisons of pharmacologically important
protein shapes that cannot be made through simple gene sequence comparisons
but are essential for finding highly selective drugs. It plans to expand its 4-D
protein structural database and rapid nonpeptide drug lead molecule generation
technologies. The company provides access to the technology through partner-
ships and subscriptions.

Contact: Susan K. Brugess

Supelco
Supelco Park, Bellefonte, PA 16823, USA
Phone: (814) 359-3441; Fax: (800) 447-3044

© 2000 by Marcel Dekker, Inc.

Toll-free in U.S.: (800) 247-6628
E-mail: supelco@sial.com
Internet: http:/ /www.supelco.sial.com/supelco.html

Profile: Supelco is a member of the Sigma-Aldrich family and has 30 years

experience in providing chromatography products for analysis and purifica-
tion. Aldrich combines its own products with those of Sigma, Fluka, and Su-
pelco to provide a broad range of products for the combinatorial market.

Products/technologies: These include Visiprep™ solid phase extraction vac-

uum manifolds to manipulate combinatorial chemistry reaction mixtures; a
rare chemicals library of over 60,000 products; CombiKits™ to create a custo-
mized kit of preweighed building blocks; SUPELCOSIL ABZ⫹ HPLC col-
umns; and a chemical directory on CD-ROM.

Synopsys Scientific Systems, Ltd.

175 Woodhouse Lane, Leeds LS2 3AR, UK
Phone: 44 (0) 113245 3339; Fax: 44 (0) 113243 8733
E-mail: sales@synopsys.co.uk
Internet: http:/ /www.snopsys.co.uk/

Profile: The company provides scientific information products and services

to the fine-chemical, pharmaceutical, agrochemical, biotechnology, and aca-
demic research communities. Synopsys operates worldwide, with offices and
distributors in Europe, the United States, and Japan. It specializes in chemical
reaction databases that deliver added-value information and refined data.

Products/technologies: These include the Accord range of chemical software.

Architecture is designed to be compatible with existing chemical and non-
chemical software, yet be scalable and focused to meet future needs. Database
products are available for use with chemical database systems such as
REACCS, ISIS, and ORAC. Database products are also available for solid
phase synthesis, methods in organic synthesis (MOS), BioCatalysis, Protecting
Groups, and BIOSTER. The Protecting Groups database comprises over
24,000 carefully selected protecting group reactions. It provides chemists with
ready access to the full range of protecting group chemistry and assists users
in identifying the most appropriate conditions to perform a protection or depro-
tection step. BIOSTER database is a compilation of over 2000 bioisosteric
transformations—including drugs, agrochemicals, and enzyme inhibitors.

© 2000 by Marcel Dekker, Inc.

Systems Integration Drug Discovery Co., Inc. (SIDDCO)
9000 S. Rita Road, Bldg. 40, Tucson, AZ 85747, USA
Phone: (520) 663-4001; Fax: (520) 663-0795
E-mail: reising@siddco.com

Profile: Founded in late 1996, SIDDCO is a privately held company that is

based on a mission to integrate and apply state-of-the-art approaches to drug
discovery. SIDDCO is a combinatorial chemistry consortium that pools the
resources of the consortium partners to fund a critical mass of chemists to
develop the combinatorial chemistry technology database and methodologies.
SIDDCO provides a dedicated team of medicinal chemists to each partner to
exploit the shared technology and satisfy the drug discovery and drug optimi-
zation needs of specific projects.

Products/technologies: Combinatorial chemistry and high-density miniatur-

ized high-throughput assay technologies are integrated and applied in a com-
prehensive drug discovery program to serve the needs of SIDDCO and its
industrial partners. Multiarray plate screening (MAPS) converts genomic se-
quence information directly into screening assays for discovering drugs and
increases the number of biochemical assays that can be efficiently conducted.
Chemical libraries, validated library technology, and automated methods may
be used by consortium partners on a consortium-exclusive basis. SIDDCO
also provides a dedicated team of medicinal chemists (SWOT ⫽ SWift Opti-
mization Team) to each consortium partner pursuing the confidential design
and synthesis of libraries for specific targets and the optimization of leads,
royalty-free.

Contact: Pete Reisinger, Vice President, Operations

TECAN Group
Feldbachstrasse 80, CH-8634 Hombrechtikon, Switzerland
Phone: 41 (0)55 254 81 11; Fax: 41 (0)55 244 38 83
E-mail: tecan@tecan.ch
Internet: http://www.tecan.com
Stock: Traded on Zurich stock exchange TECI (EBS)

Profile: Founded in 1980, the company is involved in the development and

manufacture of robotic liquid-handling systems, automated sample processors,

© 2000 by Marcel Dekker, Inc.

microplate photometry, and major components. Tecan U.S., the North Ameri-
can headquarters in Research Triangle Park, North Carolina, has been devel-
oping new product concepts for pharmaceutical drug discovery and human
genomic markets since 1994. This has led to a broad line of laboratory auto-
mation and to developments in the field of automated parallel synthesis of
medicinal compounds using both solution and solid phase chemistry. Cavro
Scientific Instruments, Inc. is a Tecan Group company. In December 1997,
Perkin Elmer Corporation acquired a 14.5% stake in Tecan AG increasing its
share to 52%.

Products/technologies: The company sells the CombiTec parallel synthesis

system, an organic chemical synthesizer that includes a robotic sample pro-
cessor, and reaction blocks of 8–56 chambers. Other products include the
TRAC system for high-throughput screening with more than 100 microplates;
the GENESIS Series Robotic Sample Processor (RSP); fully automated mi-
croplate-based systems; and the Cavro RSP 9000 Robotic Sample Processor
(XYZ module).

Contact: Hansruedi Merz, Sales Manager

Terrapin Technologies, Inc.

750 Gateway Blvd., South San Francisco, CA 94080, USA
Phone: (650) 244-9303; Fax: (650) 244-9388
E-mail: inquiry@trpntech.com
Internet: http:/ /www.terrapintech.com

Profile: Founded in 1986, Terrapin is a privately held pharmaceutical discov-

ery company that targets unmet medical needs in diabetes, allergy/asthma,
and oncology/hematology. Research programs focus on the discovery of new
intracellular targets as well as novel therapeutic compounds. The company is
engaged in collaborations relating to combinatorial chemistry.

Products/technologies: These include enabling technology and an integrated

approach to drug discovery that allows for the rapid identification and optimi-
zation of therapeutic leads from large chemical libraries. It has a compound
library and a proprietary molecular fingerprinting technology (TRAP™).
Applying computational, combinatorial, and medicinal chemistry techniques,

© 2000 by Marcel Dekker, Inc.

the company refines selected compounds, enhancing properties such as po-
tency and specificity.

Contact: Sharon Tetlow, V.P., Finance and Administration

The Automation Partnership

Melbourn Science Park, Melbourn, Royston, Hertfordshire SG6 6HB UK
Phone: 44 1763 262026; Fax: 44 1763 262613
E-mail: info@autoprt.co.uk
Internet: http://www.autoprt.co.uk

Profile: The Automation Partnership had previously been the Automation Di-
vision of The Technology Partnership until June 1995 when it was spun out
as a wholly owned subsidiary. In 1997, TTP Group was set up as a holding
company with The Automation Partnership, The Technology Partnership plc,
and Signal Computing Ltd. as group members. The company’s Haystack™
system automates the storage, tracking, retrieval, and dispensation of proprie-
tary compounds in support of combinatorial chemistry and for high-throughput
drug discovery.

Products/technologies: Haystack Automated Compound Storage stores dry

compounds in vials; stores microtubes of liquid compounds; dispenses pow-
ders in Haywain™ for dissolving; retrieves samples on demand for biochemi-
cal assays; tracks and schedules all compounds, samples and operations; works
very fast around the clock, is modular and fail-safe. Large systems have held
0.5 million dry compounds and 2.5 million liquid samples. Modules can be
configured to the needs of the customer. The database software, intrinsic to
any configuration, interfaces to customers’ existing systems. Modules that are
available include dry compound store (variable size), liquid sample store
(variable size), microtiter plate store (variable size), bar coded glass bottle
and caps, Haywain dispensing unit, manual weigh station, solubilization and
aliquotting unit. The system can be combined with OPTIMA™ substance man-
agement and customer ordering software from EMAX Solution Partners. The
software provides the user interface for registration, dispatch, and build order
functions, as well as removal orders, inventory changes, inquiries, and alter-
ations.

Contact: Judy Kramer

© 2000 by Marcel Dekker, Inc.

The Technology Partnership plc
Melbourn Science Park, Melbourn, Royston, Herts SG8 6EE, UK
Phone: 44 1763 262626; Fax: 44 1763 261582
E-mail: cgb@techprt.co.uk
Internet: http:/ /www.ttpgroup.co.uk

Profile: The Technology Partnership (TTP) is a product development and en-

gineering company based near Cambridge, UK. TTP develops new products,
improves existing products, and supplies automated manufacturing systems.
TTP operates in a wide range of market sectors including pharmaceuticals,
novel printing technology, office products, and digital mobile phones. TTP is
known for its Cellmate™ automated vaccine manufacturing system and Hay-
stack™ automated compound handling system used in drug discovery. As well
as providing new products for the world’s leading companies, TTP has a policy
of investing a significant percentage of its revenue in technologically advanced
in-house developments. These then create business opportunities that may be
exploited as joint ventures, licensing agreements, or spinoff companies. TTP
employs more than 300 staff, who own 75% of the company. A member of
the TTP Group plc, its first U.S. office was set up as TAP, Inc. in Killingworth,
CT, in early 1997.

Products/technologies: Myriad™ automated synthesis systems are based on

a series of robotic processing modules. These modules split the synthesis pro-
cess into efficient, high-throughput units. The chemist can also develop new
reactions on a Personal Synthesizer™, fully compatible with the Myriad sys-
tem. Synthesizer products include the Chemistry Developer (low-cost system
for reaction development); the Personal Synthesizer (24-vessel system used
for lead optimization and process development); and the Core System (high-
throughput synthesizer designed for production of large libraries.

Contact: Christopher Graeme-Barber

ThermoQuest Corp.
355 River Oaks Parkway, San Jose, CA 95134, USA
Phone: (408) 577-1053
E-mail: webmaster@thermoquest.com
Internet: http:/ /www.thermo.com/subsid/tmq.html
Stock: TMQ (ASE)

© 2000 by Marcel Dekker, Inc.

Profile: The company develops, manufactures, and sells mass spectrometers,
liquid chromatographs, and gas chromatographs for the environmental, phar-
maceutical, and industrial marketplaces. These analytical instruments are used
in the quantitative and qualitative chemical analysis of organic and inorganic
compounds at ultratrace levels of detection. ThermoQuest is a public subsid-
iary of Thermo Instrument Systems, Inc. (AMEX: THI), a Thermo Electron
(NYSE: TMO) company.

Products/technologies: ThermoQuest sells the Navigator LC-MS (Finnigan)

with additional modules for combinatorial and biochemical applications; and
the LCQ or TSQ for LC-MS-MS.

3-Dimensional Pharmaceuticals, Inc.

Eagleview Corporate Center, 655 Stockton Drive, Suite 104, Exton, PA 19341,
USA
Phone: (610) 458-8959; Fax: (610) 458-8258
E-mail: webmaster@3dp.com
Internet: http://www.3dp.com

Profile: This privately held company was founded in 1993 to integrate ad-
vanced technologies in structure-based design, combinatorial chemistry, and
chemi-informatics for the cost-effective discovery of orally active pharmaceu-
ticals. The company has developed a system capable of generating new drugs
through computer-controlled robotic synthesis and analysis of chemical librar-
ies. Current drug discovery efforts are focused on cardiovascular disease and
cancer.

Products/technologies: 3DP’s integrated drug discovery technology combines

the selection power of DirectedDiversity®, a proprietary method of directing
combinatorial chemistry toward specific molecular targets, with the precision
of structure-based design, to discover and optimize new drugs rapidly. Ac-
cording to the company, this technology platform allows it to discover and
refine drugs active against a wide range of therapeutic targets more quickly
than conventional approaches. The process allows parallel and iterative chem-
istry; it is an operating system for gathering information from many parts
of the drug discovery pipeline. The company also has a chemi-informatics
database.

Contact: Thomas P. Stagnaro, President and CEO

© 2000 by Marcel Dekker, Inc.

Torrey Pines Institute for Molecular Studies (TPIMS)
3550 General Atomics Court, San Diego, CA 92121, USA
Phone: (619) 455-3803; Fax: (619) 455-3804
E-mail: Houghten@tpims.org
Internet: http:/ /www.tpims.org

Profile: TPIMS is a nonprofit biomedical research institute focused on the

development of combinatorial chemistry techniques that can be applied to all
compound types. It was founded in 1988 to continue earlier research begun
at The Scripps Research Institute in La Jolla, California, and in 1989 it began
its research activities. Less than one year after beginning its operations, TPIMS
scientists had developed a method for synthesizing and screening combinato-
rial libraries of tens of millions of peptides and other nonpeptide compounds.
As a result of this early research, TPIMS became an internationally recognized
research center in the field of molecular diversity and combinatorial chemistry.
Research is funded by the National Institutes of Health, the National Science
Foundation, the U.S. Army, and by a variety of pharmaceutical companies.

Products/technologies: These include PositionalScan, a patented new technol-

ogy that permits the rapid identification of highly active individual compounds
from compound libraries. It is a screening method that allows highly active
compounds to be identified in a very short time. The technology is licensed
to Houghten Pharmaceuticals, Inc. (HPI) for use in its internal and col-
laborative combinatorial chemistry programs. In 1994 the concept of ‘‘li-
braries from libraries’’ was introduced. Originally the permethylation of a
positional scanning hexapeptide library was described. This technique permit-
ted preexisting combinatorial libraries to be transformed in a single synthetic
step to other compound classes having very different physical/chemical prop-
erties. Peralkylation of other combinatorial libraries using other alkylation re-
agents has also been performed. In 1995 a description of the exhaustive reduc-
tion of a positional scanning hexapeptide library was published. These
techniques have been combined to produce a large number of combinatorial
libraries from preexisting libraries; additionally, these techniques have been
applied to other reaction schemes and pharmacophores. Combinatorial librar-
ies include cyclic peptide libraries and combinatorial libraries spaced around
cyclic peptide templates; peptidomimetics; and heterocyclic combinatorial
libraries.

Contact: Dr. Richard A. Houghten, President and Chief Technical Officer

© 2000 by Marcel Dekker, Inc.

TransCell Technologies, Inc.
8 Cedar Brook Drive, Cranbury, NJ 08512, USA
Phone: (609) 655-6932; Fax: (609) 655-6930
E-mail: george@transcell.com
Internet: http://www.transcell.com

Profile: Founded in 1991, Transcell Technologies is a privately held, majority-

owned subsidiary of Interneuron Pharmaceuticals, Inc. (IPIC on NASDAQ).
Its business strategy is to form strategic alliances with major corporate partners
in the area of new drug discovery, and to license and apply the technology. The
company is researching the production of combinatorial libraries of synthetic
carbohydrates and glycoconjugates for drug discovery. TransCell has broadly
applicable, proprietary technologies that can be applied to corporate collabora-
tions in a number of areas, including infectious diseases, cancer, and metabolic
diseases. It is developing new pharmaceuticals based on carbohydrate combi-
natorial chemistry.

Products/technologies: It has two core technologies: a cell permeation en-

hancer to improve drug delivery, and the rapid synthesis of oligosaccharides
for the creation of libraries of oligosaccharides and glycoconjugates for use
in identifying drug candidates.

Contact: George W. Brodhead, VP, Business Development

Trega Biosciences, Inc.

3550 General Atomics Ct., San Diego, CA 92121, USA
Phone: (619) 455-3814; Fax: (619) 455-2544
Internet: http://www.trega.com
Stock: TRGA (NASDAQ)

Profile: Founded in 1992, the company develops combinatorial chemistry li-

braries for internal R & D programs and for corporate collaborative develop-
ment programs for the discovery of novel small-molecule drug therapies. The
company’s name was changed on May 1, 1997, from Houghten Pharmaceuti-
cals. The company has leveraged its technology platform by entering into
pharmaceutical alliances, enabling partners to access Trega’s technologies in
exchange for licensing fees, potential milestone payments, and royalties, and
by establishing joint discovery alliances with biotechnology companies.

© 2000 by Marcel Dekker, Inc.

Products/technologies: Utilizes its combinatorial chemistry libraries for the
discovery of novel, small-molecule drug therapies in its internal discovery
programs, in collaboration with pharmaceutical companies and in joint discov-
ery alliances with biotechnology firms. Results of their combinatorial research
agreement with Torrey Pines Institute for Molecular Studies (TPIMS) have
included a range of inventions including PositionalScan, a method for identi-
fying active compounds from mixture-based libraries.

Contact: Susan Adler, Senior Director, Business Development

Tripos, Inc.
1699 S. Hanley Road, St. Louis, MO 63144, USA
Phone: (314) 647-1099; Fax: (314) 647-9241
Toll-free in U.S.: (800) 323-2960
E-mail: info@tripos.com
Internet: http:/ /www.tripos.com

Profile: Founded in 1979, the company is a developer of molecular modeling

and computational chemistry software that focuses on pharmaceuticals and
biotechnology. Its Accelerated Discovery Services division provides a range
of research services, including library design, to comprehensive discovery col-
laborations. In November 1997, the company announced the acquisition of
Receptor Research Ltd., a UK-based company specializing in custom synthe-
sis using solid phase chemistry for biological applications. The acquisition will
allow Tripos to directly apply its molecular informatics, design, and analysis
expertise in new compound discovery collaborations in the pharmaceutical,
biotechnology, agrochemical, and related industries.

Products/technologies: Chemical diversity technology is used to design librar-

ies with high structural variation. The company supplies drug discovery
services, chemical compounds, and scientific software to facilitate the dis-
covery of new therapeutic and bioactive compounds in pharmaceutical,
biotechnology, chemical, and agrochemical industries. ChemSpace proprie-
tary technology is used for the design and analysis of combinatorial libraries
and related data; the ChemSpace virtual database of over one trillion syntheti-
cally accessible, small organic chemical structures provides a platform from
which screening and focused chemical libraries can be designed and managed.
Combinatorial chemistry software applications add on to its traditional Sybyl
software for molecular modeling and analysis, along with Unison Windows

© 2000 by Marcel Dekker, Inc.

software for desktop data access and management. Legion is a combinatorial
chemistry application, and Selector provides software tools for managing and
analyzing chemical diversity. Tripos/Panlabs offers Optiverse™, a standard
compound screening library, that includes integrated design and synthesis for
optimal lead discovery and lead refinement.

Contact: Dr. John P. McAlister, President and CEO

Varian Associates, Inc.

3050 Hansen Way, Palo Alto, CA 94304-1000, USA
Phone: (650) 493-4000; Fax: (281) 240-6752
Toll-free in U.S.: (800) 926-3000
Internet: http://www.varian.com
Stock: VAR (NYSE)

Profile: Varian is a billion dollar high-technology company. As a diversified,

international electronics company, it designs, manufactures, and markets high-
tech systems and components for applications in worldwide markets. It is orga-
nized around three core businesses that include semiconductor equipment,
healthcare systems, and instruments. Its instruments group supplies analytical
and research instrumentation and related equipment for chemical composition
studies of myriad substances. Additionally, this operation manufactures vac-
uum products, helium leak detectors, and solid phase extraction sample prepa-
ration products that are used for chemical isolation and purification. The com-
pany acquired the assets of Dynatech Precision Sampling Corporation, a
manufacturer of several autosampling devices for LC and GC. In early 1997
it also acquired the LC line of Rainin Instrument Company, Inc.

Products/technologies: These include solid phase extraction products in

96-well format; and Mercury™, which is a small, low-cost NMR spectrometer.
The acquisition of Rainin allows Varian to compete in the preparative HPLC
market and gives it a lower cost product line to complement its research grade
LC systems.

Versicor Inc.
270 E. Grand Avenue, South San Francisco, CA 94080, USA
Phone: (415) 829-7000; Fax: (415) 635-0973

© 2000 by Marcel Dekker, Inc.

Profile: Versicor Inc., a majority-owned subsidiary of Sepracor Inc. (Nasdaq:
SEPR), is a second-generation company that integrates biology/genomics,
combinatorial chemistry, high-throughput screening, and informatics in a bal-
anced format to discover new anti-infective drugs against resistant pathogens.

Products/technologies: Screening and synthesis of chemical libraries, high-

throughput assay development, and lead validation.

Contact: Jonae R. Barnes, Manager, Corporate Communications

Waters Corporation
34 Maple Street, Milford, MA 01757, USA
Phone: (508) 478-2000; Fax: (508) 872-1990
Toll-free in U.S.: (800) 252-HPLC
E-mail: brian_ j_murphy@waters.com
Internet: http:/ /www.waters.com
Stock: WAT (NYSE)

Profile: Waters is the largest company in the analytical instrument industry

devoted to HPLC (instrumentation, consumables, and associated postsale sup-
port services) and related technologies. The company has held the leadership
position in HPLC for more than 35 years, with the pharmaceutical industry
being Waters’ largest single market. Its HPLC and LC-MS systems are ideally
suited for combinatorial analysis. In addition to leading in the $2.1 billion
HPLC market, the company has also expanded its product and technology
offerings through several recent acquisitions. Acquisitions in 1997 included
Micromass Limited (Manchester, England), a technology leader in the areas
of magnetic sector, time-of-flight, and quadrupole geometry mass spectrome-
ters, and the largest supplier of benchtop MS instruments, including systems
for combinatorial screening and rapid purification of drug candidates. Also
acquired in 1997 was YMC Inc. (Wilmington, NC), the North American dis-
tributor of YMC brand columns and bulk packing materials. Sales to interna-
tional customers account for more than 60% of total revenues.

Products/technologies: The Waters LC-MS solutions for drug discovery con-

sist of a Waters Alliance™ HPLC system (with optional 996-photodiode array
or 2487 UV/Vis detectors), Waters 2700 Sample Manager (for 96-well micro-
titer plates), Micromass Platform LC or Platform LCZ benchtop API mass
spectrometer, and MassLynx™ software with special options for combinatorial

© 2000 by Marcel Dekker, Inc.

chemistry, such as OpenLynx™ Diversity and FractionLynx™. Waters also
provides columns and sample preparation products, such as Symmetry®, Sym-
metryShield®, and Oasis HLB cartridges and extraction plates, that are de-
signed for high-throughput drug assays and preparative applications.

Contact: Brian J. Murphy, Corporate Communications

Xenometrix, Inc.
2425 North 55th Street, Suite 111, Boulder, CO 80301-5700, USA
Phone: (303) 447-1773; Fax: (303) 447-1758
Toll-free in U.S.: (800) 4DNATOX
E-mail: Jwillows@xeno.com
Internet: http://www.xeno.com.

Profile: A biotechnology company with proprietary technology, Xenometrix

helps the pharmaceutical industry improve the effectiveness of drug discovery
and development.

Products/technologies: Unique gene response profiles provide mechanistic in-

formation that can be used to optimize drug leads, as well as automated assays
for screening drug leads for safety. Tools are offered as kits for customers to
perform in their own laboratories or as a service through Xenometrix’s Cus-
tomer Research Laboratory.

Contact: Stephen J. Sullivan, President and CEO

Xenova Group plc

240 Bath Road, Slough, Berkshire SL1 4EF, UK
Phone: 44 1753 692229; Fax: 44 1753 692613
Internet: http://www.xenova.co.uk
Stock: XNVAY (NASDAQ)

Profile: Founded in 1986, Xenova went public in 1994. The company provides
rapid, efficient, and cost-effective routes to discovering novel drugs where
there is no obvious chemical starting point.

Products/technologies: The company has diverse libraries of natural com-

pounds prepared from plant, fungal, and microbial sources; and proprietary

© 2000 by Marcel Dekker, Inc.

processes for the isolation and characterization of individual compounds,
which are compatible with modern high-throughput screening technologies.

Contact: Dr. Louis Nisbet, CEO

YMC, Inc.
3233 Burnt Mill Drive, Wilmington, NC 28403, USA
Phone: (910) 762-7154; Fax: (910) 343-0907
Toll-free in U.S.: (800) YMC-6311
E-mail: ymcinc@ymc-hplc.com
Internet: http:/ /www.ymc-hplc.com

Profile: YMC, Inc. manufactures and supplies YMC brand HPLC columns
and bulk packing materials. It was founded as a joint venture company in
1985 with Yamamura Chemical Laboratory Co. Ltd. (YMC Co. Ltd.) as the
North American distributor. Waters Corporation has held an equity position
in YMC Co. Ltd. since 1987, and in 1997 it acquired YMC, Inc. Waters trades
as WAT on the NYSE.

Products/technologies: The CombiChrom™ columns provide HPLC analysis

and purification of diverse mixtures. Also, CombiScreen™ columns are used
for one-step scale-up for fast analysis and high throughput; and CombiPrep™
columns are designed for one-step scale-up for purification of complex mix-
tures.

Contact: Jim Carroll

Zymark Corporation
Zymark Center, Hopkinton, MA 01748, USA
Phone: (508) 435-9500; Fax (508) 435-3439
E-mail: solutions@zymark.com
Internet: http:/ /www.zymark.com

Profile: Founded in 1981, Zymark is a leader in laboratory automation for the

pharmaceutical, biotechnology, and chemical industries. Having pioneered the
technology, the company serves drug discovery research, bioanalytical devel-
opment, formulations development, and quality control functions. In 1994
Zymark transitioned from a technology-based product business to a laboratory

© 2000 by Marcel Dekker, Inc.

automation solutions business. The focus is on providing solutions to compa-
nies to improve lab productivity by integrating technology, support, consulting
services, and project management. In September 1996, Zymark became a
member of the Berwind Group of companies. The Berwind Group is a highly
diversified, family-owned enterprise, consisting of a number of independently
managed businesses.

Products/technologies: In the area of discovery research, the company offers

microassay technologies including high-throughput screening sstems for a
range of biological assays (the Zymate immunoassay system, cell-based assay
systems, and receptor binding systems); compound preparation and distribu-
tion systems, and automated organic synthesis systems. Zymark’s RapidPlate-
96 workstation simultaneously transfers to and from 96 wells of a microplate.
The company also has designed a software interface, the PCS productivity,
communications, and scheduling software that can be customized to customer
needs. The Allegro ultrahigh-throughput screening technology applies auto-
mation and assay technologies in a production line architecture that allows
screening of 100,000 assay points per day.

Contact: Kenneth N. Rapp, Vice President, Sales

© 2000 by Marcel Dekker, Inc.