ligand discovery using small molecule...

JPET#76943 PiP

1

Ligand discovery using small molecule microarrays

Xuezhong He, Guillermo Gerona-Navarro, Samie R. Jaffrey*

Department of Pharmacology

Weill Medical College

Cornell University

1300 York Avenue, Box 70

New York, NY 10021

(212) 746-6243 (phone)

(212) 746-5596 (fax)

JPET Fast Forward. Published on November 10, 2004 as DOI:10.1124/jpet.104.076943

Copyright 2004 by the American Society for Pharmacology and Experimental Therapeutics.

This article has not been copyedited and formatted. The final version may differ from this version.JPET Fast Forward. Published on November 10, 2004 as DOI: 10.1124/jpet.104.076943

at ASPE

T Journals on M

ay 21, 2018jpet.aspetjournals.org

Dow

nloaded from

http://jpet.aspetjournals.org/

JPET#76943 PiP

2

Running Title

Applications of small molecule microarrays

Samie R. Jaffrey

Department of Pharmacology

Weill Medical College

Cornell University

1300 York Avenue, Box 70

New York, NY 10021

(212) 746-6243 (phone)

(212) 746-5596 (fax)

[email protected]

Text pages (not including references section): 16

Tables: 0

Figures: 0

References: 57

Words in Abstract: 59

Words in Introduction: 222

Words in Discussion: 5202

Abbreviations: AP, alkaline phosphatase; BSA, bovine serum albumin; GST, glutathione S-transferase;

MALDI, matrix-assisted laser desorption/ionization; PNA, protein nucleic acid; RT-PCR, reverse

transcription-polymerase chin reaction; SELEX, Selective enrichment of ligands by exponential

enrichment.


at ASPE

T Journals on M


Dow

nloaded from


JPET#76943 PiP

3

Abstract

Small molecule microarrays have recently been used to identify ligands for several proteins, and several

themes regarding screening strategies and limitations have emerged. In this review, some of the technical

issues related to the manufacture and screening of small molecule microarrays, as well as prospects for

small molecule microarrays in several areas of drug discovery and chemistry are discussed.


at ASPE

T Journals on M


Dow

nloaded from


JPET#76943 PiP

4

Introduction

A major advance in drug discovery has been the development of techniques to generate large

libraries of chemical compounds using combinatorial chemistry. Several combinatorial chemistry

techniques exist to generate large libraries, including the split-pool (or “one bead one compound”)

approach. These techniques have increased the number of compounds that are available for screening

purposes. However, with the large number of different types of libraries available, methods to rapidly

screen these libraries for molecules that have biologically-relevant activity are needed. In this review, we

will discuss one new approach for improving library screening—the small molecule microarray. In this

review, we will use the term “small molecule” to refer to non-DNA molecules that are prepared using

synthetic chemistry, rather than molecules that are obtained from cells or biochemical reactions, such as

proteins. The small molecule microarray is a library of compounds that is immobilized on a two-

dimensional surface, such as a glass slide, and is spatially addressable and miniaturized. Small molecule

microarrays have recently been used to identify ligands for several proteins, and several themes regarding

screening strategies and limitations have emerged. In this review, we will discuss the manufacture and

use of microarrays as well as its potential for contributing to the lead identification process. We will also

discuss prospects for small molecule microarrays in other areas of drug discovery and chemistry.

Historical Foundation of Small Molecule Microarrays

On-bead screening

Small molecule microarrays merge the relatively mature technology of on-bead library screening

with microarray technology, the latter of which has become an especially powerful tool used primarily by

molecular biologists in the context of DNA microarrays.

Screening immobilized libraries of small molecules on beads was described in one of the original

reports that described the technique of split-pool library synthesis (Lam et al., 1991). The split-pool


at ASPE

T Journals on M


Dow

nloaded from


JPET#76943 PiP

5

approach (also known as “portioning-mixing” and related names) involves dividing polymeric supports

(i.e., the “beads”) into portions for coupling to modular units (i.e. “diversity elements”), and then

recombining the beads (Furka et al., 1991; Houghten et al., 1991; Lam et al., 1991). Once combined, the

beads are washed, deprotected, etc., and the process of splitting the beads and pooling the beads is

repeated for several cycles to make a large collection of molecules. Thus, each bead will contain

molecules whose structure is a consequence of the history of the diversity element coupling steps to

which the bead was subjected. This method is also known as the “one bead-one compound” method

(Lam et al., 1991) since each bead contains one type of molecule, although each bead will contain > 1013

copies of each molecule.

To screen the molecules on the bead, a labeled protein is incubated with the beads. Beads that

contain a molecule that interacts with the labeled protein are then detected via the label on the protein.

For example, the protein may be labeled with biotin, and protein-bound beads may be detected by

incubating the beads with streptavidin-alkaline phosphatase (AP). Incubation of the AP-bead complex

with a chromogenic AP substrate results in the generation of a purple dye that partitions into the

hydrophobic interior of the bead, resulting in the beads turning purple (Lam et al., 1991). The beads can

then be detected visually and isolated to determine the identity of the molecule on the bead.

This and related approaches have been used successfully to find peptide ligands for several

proteins such as streptavidin (Lam et al., 1991) and the src SH3 domain (Morken et al., 1998), ligands for

RNAs (Hwang et al., 2003), and peptoid ligands for mdm2 and GST (Alluri et al., 2003). Up to 1-2

million compounds per day can be queried on beads. Once a bead has been identified as possessing a

molecule that binds to a specific protein probe, the structure of the molecule on the bead has to be

determined. For peptide and peptoid libraries, the molecule on the bead can be eluted and determined by

Edman degradation. However, for other libraries, determining the identity of the molecule on the bead

can be very difficult—typically not enough is available for NMR and MS can be ambiguous. A popular

approach is to use an encoding scheme during the synthesis of the libraries. Thus, during the steps where

diversity elements are coupled, a chemical tag that reflects the identity of the diversity element is


at ASPE

T Journals on M


Dow

nloaded from


JPET#76943 PiP

6

covalently coupled to the bead. At the end of the synthesis, the history of the coupling steps can be

determined by liberating the tags and detecting them chromatographically (Ohlmeyer et al., 1993). If

encoding can be readily performed, on-bead screening followed by deconvolution of the encoding tag can

be an exceptionally high-throughput method for ligand discovery.

Microarray-based screens

A second emerging technology that helped to bring about small molecule microarrays was the

DNA microarray technology that was popularized in the 1990s. Microarray technologies involving

several types of macromolecules have existed since the mid 1980s (Ekins et al., 1996; Ekins and Chu,

1999), but have recently become highly popularized and accessible with the advent of DNA microarrays.

These microarrays consist of small spots (“microspots”) of a few micrometers in diameter, each

containing specific molecules, such as DNA, proteins, or small molecules. These spots are arrayed on a

surface, most commonly silica or glass, due to the ease with which it can be functionalized, its chemical

stability, and due to its low intrinsic fluorescence. Although glass has been the principle support for

small molecule microarrays, molecules have been immobilized on a wide variety of other materials, such

as gold, ceramics, polyacrylamide and nitrocellulose, and these can also be used for the generation of

microarrays.

Microarrays have specific benefits that have made them useful for analytical studies. These

include the ability to mass produce large numbers of arrays, the ability to perform massively parallel

binding assays, the rapid kinetics of binding that occurs upon miniaturization, and high sensitivity of

detection of binding to microarrays that results from concentrating analyte-binding molecules to a

microspot.

Experiments with DNA microarrays exemplify the beneficial features of microarray technology.

DNA microarrays permit the quantification of up to tens of thousands polynucleotides in a sample in a

single experiment. Thus, the DNA microarray contains thousands of microspots, each containing a

specific DNA sequence that can detect a complementary analyte. These microspots are arranged in a grid


at ASPE

T Journals on M


Dow

nloaded from


JPET#76943 PiP

7

such each DNA is spatially addressable. The binding of fluorescently-labeled polynucleotides, prepared

enzymatically from cellular lysates, allows quantification of the abundance of specific polynucleotides by

quantifying the fluorescence at each microspot. This quantification can be assessed by microarray-

scanning instruments that rapidly quantitate the fluorescence at each microspot.

Although microarray screening is massively parallel, a key property of this type of screen is that

they permit a higher degree of sensitivity than can be obtained by other binding assays. The high degree

of sensitivity derives from the high density of receptor on the glass surface in the microspot. Thus, the

labeled analyte is detected in a small area, in which the background nonspecific signal is minimized.

Furthermore, the degree of binding is predictably related to the concentration of labeled probe. As the

size of the spot shrinks, the fractional occupancy, F, of the receptors in the microspot is related to the KD

of the analyte, such that F is equal to [A]/(KD + [A]), where [A] is the analyte concentration (Ekins,

1998).

Manufacture of Small Molecule Microarrays

Small molecule microarrays from combinatorial libraries

Small molecule microarrays can be prepared by either first synthesizing the molecules and then

immobilizing them on the surface or by synthesizing them directly on the surface. In the first approach,

stock solutions of library members are arrayed using a commercial microarraying robot such as those

used for preparing DNA microarrays. High capacity synthetic beads can be used to generate microliter

volume stock solutions with concentrations > 1 mM (Tallarico et al., 2001). Subnanoliter volumes are

dispensed to generate spots with diameters typically between 50 and 500 µm. These molecules can be

covalently coupled to the glass surface if the small molecule contains chemical moieties that react with

functional groups that have been incorporated onto the glass surface. For example, Schreiber and

colleagues (Hergenrother et al., 2000) describe the generation of alcohol-reactive glass slides that were

activated with silyl chloride. Alcohol-containing small molecules were generated on ~500 µm

polystyrene macrobeads, eluted and resuspended at a concentration of 5 mM in 5 µl DMF and then


at ASPE

T Journals on M


Dow

nloaded from


JPET#76943 PiP

8

immobilized on the glass slide using a microarraying robot that dispensed 1 nl per spot (Hergenrother et

al., 2000). Thus, each stock solution is sufficient to produce a spot on several thousand microarrays. In

another study, thiol-containing molecules were immobilized on maleimide-functionalized glass surfaces

(MacBeath et al., 1999). If the library member has more than one hydroxyl, in the first case, or thiol in the

second case, the library member may be displayed heterogeneously throughout the spot, limiting the

number of hydroxyls that can be present in library members.

The strategy taken by Lam and colleagues (Falsey et al., 2001) seems to reduce these limitations

on library construction by employing chemoselective ligation to covalently couple the small molecule to

the glass surface. In this case, the authors used the selective reaction of an N-terminal cysteine with

immobilized glyoxal to link peptides to glass via a thiazolidine. Small molecule libraries that contain

cysteine like moieties would also be expected to couple to these surfaces. A variety of other

chemoselective ligation reactions exist that could be used for coupling small molecules to solid supports

(Kohn et al., 2003).

Once the microarray is prepared, the microspots can be screened using a labeled probe. In most

cases so far, this has been a fluorescently-tagged protein. Microspots that become fluorescent after

incubation with the probe indicate a small molecule-probe interaction. When the microarray is prepared

using a split-pool library, the identity of the small molecule spotted at any position in the microarray can

be determined by recovering the specific bead that generated the molecule at that position, and then by

eluting the encoding tags that were used during the library synthesis (Kuruvilla et al., 2002; Barnes-

Seeman et al., 2003; Koehler et al., 2003).

Screening these microarrays can be problematic due to steric hinderance from the surface. One

factor that could affect the binding is the length of the linker used to separate the glass surface from the

small molecule (Barnes-Seeman et al., 2003). Alternatively, the chemical properties of the linker may

affect binding. In one study of linkers for on-bead screening, it was found that cationic linkers had a

substantially beneficial effect on ligand accessibility (Thorpe and Walle, 2000).


at ASPE

T Journals on M


Dow

nloaded from


JPET#76943 PiP

9

Ligand identification Using Small Molecule Microarrays

Screening microarrays for hits using binding assays

The product of a small molecule microarray screen is a molecule that interacts with the protein

that is used as the probe. However, these molecules will not be useful if they do not bind to functional

sites on the protein. Indeed, proteins have large surface areas, and the relevant active site is a small

fraction of this area. Thus the likelihood that a ligand would bind this particular area on the surface of a

protein is presumably very small. Despite this, data from several varied approaches suggest that ligands

identified by binding may preferentially target functionally important sites.

One example is SELEX, Selective Enrichment of Ligands by EXponential enrichment. This is a

highly effective approach for generating RNA ligands for target proteins using binding to select for

ligands (Famulok et al., 2000). In this procedure, 1015 individual RNAs are generated and RNAs that

bind a protein target are recovered and amplified by RT-PCR. Frequently, these ligands bind with

nanomolar or subnanomolar affinities (Gold et al., 1995). When RNA ligands are generated against an

enzyme, they frequently bind to the active site (Faria and Ulrich, 2002). For example, several RNA

ligands were generated that bind to HIV reverse transcriptase, and despite the fact that they lacked

sequence similarity to each other, each was found to be a potent inhibitor of the enzyme, several with

IC50s in the midnanomolar range (Tuerk et al., 1992; Allen et al., 1996). The crystal structure of an

aptamer-HIV reverse transcriptase complex confirmed that the RNA binds to the active site (Jaeger et al.,

1998).

Another ligand-generating approach that uses binding assays to select for ligands is phage display

(Cwirla et al., 1990; Devlin et al., 1990; Scott and Smith, 1990). In this procedure, peptides are presented

on the surface of filamentous bacteriophage. To identify peptide ligands that bind a specific protein, a

phage-binding assay is used to recover phage that interact with an immobilized protein target. One study

systematically explored the ability of phage display-derived peptides to target active sites in enzymes


at ASPE

T Journals on M


Dow

nloaded from


JPET#76943 PiP

10

(Hyde-DeRuyscher et al., 2000). In this study, several diverse enzymes were queried: alcohol

dehydrogenase, tyrosyl- and prolyl-tRNA synthetases, carboxypeptidase, beta-glucosidase, hexokinase,

and glycogen phosphorylase. In all cases, except for carboxypeptidase, phage display peptides were

recovered that inhibited enzyme activity, each in the high nanomolar to mid micromolar range (Hyde-

DeRuyscher et al., 2000).

In addition to these approaches, ligands generated from screening small molecule microarrays

directly demonstrate that ligands discovered by screening with protein probes bind to functionally

important sites on proteins. In a study to find ligands for Ure2p, a yeast protein that binds certain

transcription factors, a screen of 3,780 compounds based on a functionalized 1,3-dioxane skeleton

resulted in the identification of eight ligands for Ure2p, one of which inhibited Ure2p’s repressive activity

in yeast cells (Kuruvilla et al., 2002). This ligand, uretupamine A, bound to Ure2p with a KD of 18.1 µM.

The other seven ligands failed to inhibit Ure2p, but this may have been due to poor cell permeability.

Alternatively, Ure2p may have multiple functionally important sites, only one of which was assayed in

the derepression assay, and the other ligands bind to one of these sites. In another study, a microarray of

12,396 compounds was used to find ligands for Hap3p, a yeast transcription factor (Koehler et al., 2003).

In this study, a single ligand was identified as binding the Hap3p fusion protein, and this compound

inhibited Hap3p-dependent transcription in cellular assays. In both these cases, the ligand presumably

blocked a protein-protein interaction, which is considered exceptionally difficult to target (Spencer,

1998). These data further suggest that binding assays used by small molecule microarrays are capable of

generating ligands that target functionally important sites, although each ligand needs to be individually

tested to determine whether it targets the desired site on the protein.

These studies raise the intriguing question: Why do screens that utilize binding to recover ligands

result in ligands that target functional sites? Binding sites typically are depressions on the surface of

proteins, have a higher degree than average of exposure of hydrophobic groups on the surface; have

residues that are highly mobile compared to residues elsewhere on the surface of the protein; and contain

water molecules that are not tightly bound to the protein (Ringe, 1995; Ringe and Mattos, 1999). Ringe


at ASPE

T Journals on M


Dow

nloaded from


JPET#76943 PiP

11

and associates have proposed that these displaceable water molecules provide the gain in entropy that

propels the ligand binding reaction (Ringe and Mattos, 1999). Thus, the unique properties of binding

sites may account for the tendency of binding experiments to yield ligands that interact with protein

binding sites.

This concept is supported by structural and thermodynamic studies of ligand-protein interactions

(Arkin and Wells, 2004). In these studies, the binding energy for a ligand-protein is not distributed

evenly across the interface, but is localized to small regions of the protein. These regions are

characterized by charged and hydrophobic residues, often less than five amino acids, that represent

energetic focal points of the interaction (Clackson and Wells, 1995; Clackson et al., 1998; DeLano et al.,

2000). Importantly, phage display-derived ligands bind to hot spots for protein-protein interactions

(Fairbrother et al., 1998; DeLano et al., 2000). Furthermore, small molecules that bind IL-2, bind to the

hot spot that mediates its interaction with the IL-2 receptor (Thanos et al., 2003). Thus, small molecules

and peptides target hot spots. In addition to having hot spots that have a propensity to interact with

ligands, proteins may have undergone evolutionary selection to reduce the “stickiness” of non-binding

portions of protein surfaces in order to reduce irrelevant interactions that may occur due to the high

concentration of intracellular molecules in the cytosol.

In order to “force” a small molecule microarray screen to yield ligands that target a particular site,

one could imagine simple adjustments to the screening protocol. For example, to confirm that ligands are

binding to an active site, the screen can be performed with two proteins, one containing a mutation or

deletion at the active site. Each of the proteins could be labeled with a different fluorescent tag, e.g. Cy3

and Cy5. Thus, microspots that show binding of both proteins would represent ligands that do not bind to

the region of interest.

Affinity limits in microarray screening

A fundamental difference between microarray binding assays and activity assays is that low

affinity binding cannot be detected in a microarray format. In most library screens using activity assays,


at ASPE

T Journals on M


Dow

nloaded from


JPET#76943 PiP

12

KDs of hits in the range of 0.1 – 1.0 µM would be considered successful. However, in microarray screens,

hits with KDs in this range may be so weak and transient that the labeled protein will be lost in the washes.

This phenomenon is due to the high rate of dissociation for interactions with KDs in this range. The KD is

the ratio of two kinetic parameters, i.e. koff/kon. Although kon varies for different ligands, kon values

generally tend within an order of magnitude of 106 M-1s-1. Thus for a protein-ligand interaction with a KD

of 100 nM, and a kon = 1 x 106 M-1s-1, the value of koff is 1 x 10-1 s-1. The dissociation half-life is 0.693/koff,

meaning that the dissociation half-life is ~6.93 sec. Thus, if a microarray is washed for as little as 3 min,

over 25 half-lives of dissociation will occur. This will completely deplete all the bound protein bound to

a microspot. Thus, certain classes of interactions are selected when using a microarray approach:

nanomolar-affinity interactions, interactions that have large kon, or interactions that are associated with

additional conformational changes that stabilize the initial receptor-ligand complex. Because of the

relationship between koff and KD, high affinity interactions seen on microarrays can be distinguished from

lower affinity interactions, in part, by their stability to washing.

Since a library may be devoid of any ligand that binds a protein probe with the requisite affinity

to survive washings, dimeric probes can be used to detect ligands that bind with lower affinity. As

opposed to monomeric probes, dimeric probes exhibit a higher effective affinity when binding to a

surface. This avidity effect is due to the simultaneous binding of immobilized ligand by the dimeric

probe. For example, in the screen to identify Hap3p ligands, Hap3p was fused to the dimeric GST moiety

for screening (Koehler et al., 2003). Screens using monomeric protein, e.g. His-tagged fusion proteins,

can lead to the identification of ligands that have higher affinity.

Factors affecting the success of the small molecule microarray screens

Library design. Although small molecule microarrays may permit screening tens of thousands of

compounds, it is still far from guaranteed that a library screen will identify a small-molecule that binds to

a target with high affinity. The problem is compounded when little is known about likely structural

features of the protein’s physiologic interacting partners, thus necessitating the use of an unbiased library.


at ASPE

T Journals on M


Dow

nloaded from


JPET#76943 PiP

13

How can the probability of success be improved? Probably the factor that may most affect the success of

a screen is the size of the library. This idea makes sense—if you screen enough molecules, sooner or later

you will come upon a hit. However, the relationship between the size of a library and the number of hits

remains to be established. For example, when a library screen that utilizes 105 members fails, is it because

the specific skeletal structure utilized by the library is wholly incapable of generating a hit, or would the

use of 106, or 107, or even 108 members have resulted in a hit? Another approach is the design of libraries

that have features of natural products (Nicolaou et al., 2000; Breinbauer et al., 2002; Kissau et al., 2003).

The rationale for these approaches is based in the concept of the “privileged structure” or “privileged

substructure” (Evans et al., 1988), which postulates that certain structures (such as benzodiazepines and

benzazepines) have an intrinsic ability to bind multiple, unrelated classes of proteins. Microarrays that

use libraries of privileged structures will help to address if these molecules are more likely to generate

ligands that other types of libraries.

Choice of target. The success of the screen will also depend on the target. In a review of 150

(non-microarray) screens at Pfizer to determine hit rates, it was shown that the hit rate was 1 per 120,000

compounds for enzymes, and significantly less for small molecules that targeted protein-protein

interactions (Spencer, 1998). The difficulty in obtaining ligands that affect protein-protein interactions

may arise from the nature of protein interfaces, which often are flat and large (from 800Å2 to 4660 Å2)

(Clackson and Wells, 1995; Lo Conte et al., 1999). Despite these considerations, small molecule

microarrays have been successful in generating ligands that appear likely to disrupt these types of

interactions (Kuruvilla et al., 2002; Barnes-Seeman et al., 2003; Koehler et al., 2003).

Sensitivity and specificity in solid-phase screening. Another factor affecting the success of the

screen is the sensitivity and specificity of the signal. The sensitivity of the method is determined by the

nonspecific binding of the protein used as the probe to the microarray. Nonspecific binding could be to

the solid support (e.g. the functionalized glass), or could be to the specific chemical entity that is present


at ASPE

T Journals on M


Dow

nloaded from


JPET#76943 PiP

14

on the microspot. In the former case, specific blocking agents could be tested and identified that reduce

nonspecific binding. In the latter case, each microspot can be considered as a chemically different

surface, with its own nonspecific-binding properties. Resolving this nonspecific binding may be much

more difficult to resolve. Nonspecific interactions occur when a protein makes weak contacts with

multiple immobilized individual small molecules. These multiple weak nonspecific interactions produce

an avidity effect leading to tenacious protein binding to the microspot. Methods to reduce nonspecific

binding need to be chemically complex to deal with the varied and idiosyncratic nonspecific binding that

will be encountered at each microspot. In some cases, bovine serum albumin (BSA) was been used, while

in others, highly complex proteinaceous blocking agents such as milk and clarified E. coli lysate (Alluri et

al., 2003) were required to reduce nonspecific binding, as well as incubation buffers and salt washes as

high as 1 M (Alluri et al., 2003). These approaches will presumably also interfere with specific binding,

thus reducing the likelihood of success with this approach. Nevertheless, if the libraries are constructed

with the idea of reducing nonspecific binding, and appropriate blocking techniques are implemented,

blocking techniques can allow sensitive detection of binding.

In cases where background binding can be resolved, sensitivity can be improved by novel

detection methods. Fluorescence tags typically cannot be used for screening on-bead libraries due to the

intrinsic fluorescence of most polystyrene-based synthetic beads. However, newer quantum dot-based

labels seem to overcome this problem (Olivos et al., 2003) and have the potential to increase the

sensitivity of screening on small molecule microarrays as well. Microarray screens that utilize

fluorescence microscopy techniques such as those based on evanescent wave microscopy have the

potential to substantially increase the sensitivity of these screens (Neuschafer et al., 2003).

Although issues regarding library size and features affect the success of any screen—on a

microarray, on beads, or in a conventional solution-phase screening assay, small molecule microarray

screening has a distinct advantage over the other approaches in terms of the ease of screening. Thus,

multiple copies of a microarray can be fabricated and each can be screened against a different protein.

Thus, even if the library on the microarray is too small to be assured of finding a ligand for a given


at ASPE

T Journals on M


Dow

nloaded from


JPET#76943 PiP

15

protein, the likelihood of coming up with a ligand for some protein is improved by the ease of screening

multiple proteins in parallel. In the case of on-bead screening, beads could conceivably be washed after

being screened by individual proteins, but the screening must be accomplished serially, and detecting hits

is more tedious.

Future Prospects

Manufacturing small molecule microarrays by photolithography

An alternative approach for the preparation of small molecule microarrays is in situ synthesis of

the small molecules on the glass surface. This has been performed using photolithographic synthesis. In

the most common implementation of this method, chemical synthesis is achieved by initiating the

removal of photolabile protecting groups by directing ultraviolet light to specific sites using a

chrome/glass mask. After deprotection at specific sites, the entire array is exposed to a reactant, but only

the deprotected sites react. Next, other sites that were not deprotected, are deprotected in a spatially-

specific manner using the mask. This procedure is limited by the need to generate specific masks to

match the pattern of photodeprotection which can be a costly and time-consuming process. Nevertheless,

small molecule microarrays consisting of peptides (Fodor et al., 1991), oligocarbamates (Cho et al.,

1993), nucleotides (Pease et al., 1994), and peptoids (Li et al., 2004) have been described.

Recently the technique of maskless photolithography has been described (Singh-Gasson et al.,

1999). In this technique, a digital micromirror array, such as those used in computer display projection

systems, is used to project ultraviolet light in a specific pattern on the microarray to initiate highly

spatially specific photodeprotection. This method has been used to synthesize oligonucleotide (Nuwaysir

et al., 2002) and peptoid (Li et al., 2004) microarrays. In another approach using a digital micromirror

array, acid labile protecting groups and photogenerated acids were used to achieve photolithography of

oligonucleotides (LeProust et al., 2000).


at ASPE

T Journals on M


Dow

nloaded from


JPET#76943 PiP

16

Although photolithography is clearly becoming more rapid and inexpensive, it remains unclear

about its long-term viability for small molecule microarrays, especially for chemistries less robust than

peptide and oligonucleotide synthesis. In the case of nucleotide microarrays, the fidelity of the synthesis

can be readily determined in situ by the ability of the microarray to recognize specific oligonucleotides;

however, for the synthesis of other types of small molecules, confirmation of the synthesis, yield and

byproducts are especially difficult considering the small amounts of material on the microarray.

Conceivably subjecting the microarray to MALDI-based analysis, assuming MALDI-labile linkers are

used in the synthesis, may assist in validating the synthesis of other types of small molecules. A simpler

approach is the spotting of presynthesized molecules obtained through the more routine split-pool

approach. Spotting of these molecules would presumably be performed once the split-pool chemistry has

been optimized. Furthermore, the in situ synthesis protocol seems exceptionally low-throughput

considering that each microarray needs to be individually synthesized. Synthesizing the library by a split-

pool procedure once and then making many microarray replicas is likely to be more efficient and robust.

Screening chemical libraries on DNA microarrays

An innovative and fundamentally different approach for screening libraries using microarrays

was described by Winssinger and colleagues (Winssinger et al., 2001). Molecules are incubated with

their target proteins in solution, and bound ligands are separated from unbound ligands using gel filtration

chromatography. To couple the small molecule to an array this group used a prefabricated spatially-

addressable DNA microarray. The key to this approach was the bifunctional nature of their library: one

component of the molecule was the protein interacting moiety and the other component was a

fluorescently-tagged protein nucleic acid (PNA) sequence. PNAs are constructed using standard amide

bond forming chemistries, yet interact in a sequence-specific manner with specific DNA sequences. The

bifunctional molecule was constructed from a lysine residue such that the moiety off the ε amine is the

protein-binding component, while the moiety off the α amine is the PNA. The PNA component serves to


at ASPE

T Journals on M


Dow

nloaded from


JPET#76943 PiP

17

both target the small molecule to specific sites on the microarray as well as to encode the identity of the

protein-binding moiety. After recovery of the protein-bound molecules, the molecules were hybridized to

the array. The fluorescence at a specific localization on the array indicates the identity of a specific

molecule in the protein-bound fraction (Winssinger et al., 2001). This approach reduces the nonspecific

surface binding to a surface comprised of a specific small molecule, and reduces the steric hindrance

issues associated with protein binding to immobilized small molecules. Although exceptionally clever,

this approach is limited by the solubility of these fairly massive bifunctional molecules, possible binding

of the target protein to the PNA component, as well as the compatibility of PNA synthesis with the

synthetic steps required for nonpeptidic small molecule libraries.

Living microarrays

Most small molecule microarrays have been screened with protein probes. An alternate

way to screen is with living cells that can produce a detectable response to certain immobilized

small molecules. Lam and colleagues described microarrays of immobilized peptides that were

screened with WEHI231 lymphoma cells (Falsey et al., 2001). Adhesion of these cells to the

microarray was monitored, and resulted in the identification and confirmation of peptide

molecules that adhere to these cells. This technology could be used to generate materials that

promote cell adhesion or cell repulsion. Many devices and materials that are implanted in the

body undergo a fibrotic reaction or endothelialization which impairs their function. For example,

stents that are used to open narrowed coronary arteries are limited by their propensity to undergo

restenosis as cells adhere and occlude the stent. Alternatively, physiologic assays that utilize

green fluoresccent protein (GFP), such as subcellular localization assays or transcription assays,

could be designed, and microarrays could be screened by fluorescence microscopy. An

alternative type of living microarray involves microarrays of noncovalently-bound small

molecules embedded in a high-density matrix. These microarrays would permit diffusion of


at ASPE

T Journals on M


Dow

nloaded from


JPET#76943 PiP

18

small molecules into cells and could be used to study the effects on cells that grow above each

microspot.

Identification of DNA, RNA, and carbohydrate ligands

Although the published work with small molecule microarrays has involved screening with

labeled proteins, other molecules such as DNAs, RNAs, carbohydrates, and peptides could clearly be used

in screens. Indeed, on-bead screening with RNA probes have resulted in the identification of ligands for

the TAR RNA sequence (Hwang et al., 2003). Similarly, small molecule microarrays might be useful to

identify molecules that interact with DNA in a sequence-specific manner. The relative prevalence of

literature with protein probes compared to nonprotein probes may be related to the ease with which

proteins can be labeled with affinity tags, such as biotin, or affinity tags. Newer RNA labeling techniques

(Babendure et al., 2003) may facilitate screens using biosynthetically-prepared polynucleotides.

Identification of catalysts and sensors using small molecule microarrays

Although small molecule microarrays are typically screened against labeled proteins, small

molecule microarrays are likely to have utility in the identification of novel small molecule sensors and

catalysts. Rather than probing small molecule arrays with proteins, microarrays can be screened with

small molecules as the solution phase screening agent. Molecules on the microarray that interact with the

analyte could be identified via tags on the analyte, e.g. fluorescent tags; alternatively, environmentally-

sensitive fluorophores within the microarrayed molecules could be used with unlabeled solution-phase

analyte probes. Screening for microspots with analyte-dependent changes in fluorescence

excitation/emission properties could identify sensors that could serve as novel diagnostics. Suslick and

colleagues describe the use of arrays of metalloporphyrins that change color upon exposure to various

odorants (Rakow and Suslick, 2000). These odorants form complexes with the metal ion in the

macrocycle. These microarrays comprise molecules that analyte-dependent patterns of reactivity. In an

alternative approach, small molecule microarrays have been used to profile enzymatic activities in tissues.


at ASPE

T Journals on M


Dow

nloaded from


JPET#76943 PiP

19

Gao and colleagues have prepared arrays of various chromogenic dye derivatives (Zhu et al., 2003) that

become fluorescent when hydrolyzed upon incubation with solutions containing specific enzymes.

Conceivably these arrays could be used to obtain an "enzymatic profile" of different biological samples.

Similarly, novel catalysts could be detected using a microarray-based approach. Morken and

colleagues (Taylor and Morken, 1998) have described approaches for thermographic selection of catalysts

from an on-bead library. They exploited the fact that most chemical reactions have a measurable heat of

reaction (∆Hro) and thus the progress of a reaction can be measured by monitoring beads that have

undergone a temperature change. Using this principle, the authors searched for bead-bound molecules

that catalyzed the hydrolysis of acetic anhydride with an infrared camera. Conceivably, reactions

catalyzed by small molecules on microarrays may induce highly localized and concentrated temperature

changes permitting screening with thermal imaging.


at ASPE

T Journals on M


Dow

nloaded from


JPET#76943 PiP

20

References

Allen P, Collins B, Brown D, Hostomsky Z and Gold L (1996) A specific RNA structural motif

mediates high affinity binding by the HIV-1 nucleocapsid protein (NCp7). Virology 225:306-315.

Alluri P, Reddy M, Bachhawat-Sikder K, Olivos H and Kodadek T (2003) Isolation of protein

ligands from large peptoid libraries. J. Am. Chem. Soc. 125:13995-14004.

Arkin MR and Wells JA (2004) Small-molecule inhibitors of protein-protein interactions:

progressing towards the dream. Nature Reviews. Drug Discovery 3:301-317.

Babendure JR, Adams SR and Tsien RY (2003) Aptamers switch on fluorescence of

triphenylmethane dyes. Journal of the American Chemical Society 125:14716-14717.

Barnes-Seeman D, Park S, Koehler A and Schreiber S (2003) Expanding the functional group

compatibility of small-molecule microarrays: discovery of novel calmodulin ligands. Angew

Chem Int Ed Engl 42:2376-2379.

Breinbauer R, Manger M, Scheck M and Waldmann H (2002) Natural product guided compound

library development. Current Medicinal Chemistry 9:2129-2145.

Cho CY, Moran EJ, Cherry SR, Stephans JC, Fodor SP, Adams CL, Sundaram A, Jacobs JW and

Schultz PG (1993) An unnatural biopolymer. Science 261:1303-1305.

Clackson T, Ultsch MH, Wells JA and de Vos AM (1998) Structural and functional analysis of

the 1:1 growth hormone:receptor complex reveals the molecular basis for receptor affinity.

Journal of Molecular Biology 277:1111-1128.


at ASPE

T Journals on M


Dow

nloaded from


JPET#76943 PiP

21

Clackson T and Wells JA (1995) A hot spot of binding energy in a hormone-receptor interface.

Science 267:383-386.

Cwirla SE, Peters EA, Barrett RW and Dower WJ (1990) Peptides on phage: a vast library of

peptides for identifying ligands. Proceedings of the National Academy of Sciences of the United

States of America 87:6378-6382.

DeLano WL, Ultsch MH, de Vos AM and Wells JA (2000) Convergent solutions to binding at a

protein-protein interface. Science 287:1279-1283.

Devlin JJ, Panganiban LC and Devlin PE (1990) Random peptide libraries: a source of specific

protein binding molecules. Science 249:404-406.

Ekins R and Chu FW (1999) Microarrays: their origins and applications. Trends in Biotechnology

17:217-218.

Ekins R, Krause F, Chu FW, Berger H and Edwards P (1996) J. Clin. Ligand Assay 19:145-156.

Ekins RP (1998) Ligand Assays: from Electgrophoresis to miniaturized microarrays. Clin. Chem.

44:2015-2030.

Evans BE, Rittle KE, Bock MG, DiPardo RM, Freidinger RM, Whitter WL, Lundell GF, Veber

DF, Anderson PS and Chang RS (1988) Methods for drug discovery: development of potent,

selective, orally effective cholecystokinin antagonists. Journal of Medicinal Chemistry 31:2235-

2246.


at ASPE

T Journals on M


Dow

nloaded from


JPET#76943 PiP

22

Fairbrother WJ, Christinger HW, Cochran AG, Fuh G, Keenan CJ, Quan C, Shriver SK, Tom JY,

Wells JA and Cunningham BC (1998) Novel peptides selected to bind vascular endothelial

growth factor target the receptor-binding site. Biochemistry 37:17754-17764.

Falsey JR, Renil M, Park S, Li S and Lam KS (2001) Peptide and small molecule microarray for

high throughput cell adhesion and functional assays. Bioconjugate Chemistry 12:346-353.

Famulok M, Mayer G and Blind M (2000) Nucleic acid aptamers-from selection in vitro to

applications in vivo. Accounts of Chemical Research 33:591-599.

Faria M and Ulrich H (2002) The use of synthetic oligonucleotides as protein inhibitors and

anticode drugs in cancer therapy: accomplishments and limitations. Current Cancer Drug Targets

2:355-368.

Fodor SP, Read JL, Pirrung MC, Stryer L, Lu AT and Solas D (1991) Light-directed, spatially

addressable parallel chemical synthesis. Science 251:767-773.

Furka A, Sebestyen F, Asgedom M and Dibo G (1991) General method for rapid synthesis of

multicomponent peptide mixtures. International Journal of Peptide & Protein Research 37:487-

493.

Gold L, Polisky B, Uhlenbeck O and Yarus M (1995) Diversity of oligonucleotide functions.

Annual Review of Biochemistry 64:763-797.


at ASPE

T Journals on M


Dow

nloaded from


JPET#76943 PiP

23

Hergenrother P, Depew K and Schreiber S (2000) Small-Molecule Microarrays: Covalent

Attachment and Screening of Alcohol-Containing Small Molecules on Glass Slides. J. Am. Chem.

Soc. 122:7849-7850.

Houghten RA, Pinilla C, Blondelle SE, Appel JR, Dooley CT and Cuervo JH (1991) Generation

and use of synthetic peptide combinatorial libraries for basic research and drug discovery. Nature

354:84-86.

Hwang S, Tamilarasu N, Kibler K, Cao H, Ali A, Ping YH, Jeang KT and Rana TM (2003)

Discovery of a small molecule Tat-trans-activation-responsive RNA antagonist that potently

inhibits human immunodeficiency virus-1 replication. Journal of Biological Chemistry

278:39092-39103.

Hyde-DeRuyscher R, Paige LA, Christensen DJ, Hyde-DeRuyscher N, Lim A, Fredericks ZL,

Kranz J, Gallant P, Zhang J, Rocklage SM, Fowlkes DM, Wendler PA and Hamilton PT (2000)

Detection of small-molecule enzyme inhibitors with peptides isolated from phage-displayed

combinatorial peptide libraries. Chemistry & Biology 7:17-25.

Jaeger J, Restle T and Steitz TA (1998) The structure of HIV-1 reverse transcriptase complexed

with an RNA pseudoknot inhibitor. EMBO Journal 17:4535-4542.

Kissau L, Stahl P, Mazitschek R, Giannis A and Waldmann H (2003) Development of natural

product-derived receptor tyrosine kinase inhibitors based on conservation of protein domain fold.

Journal of Medicinal Chemistry 46:2917-2931.


at ASPE

T Journals on M


Dow

nloaded from


JPET#76943 PiP

24

Koehler AN, Shamji AF and Schreiber SL (2003) Discovery of an inhibitor of a transcription

factor using small molecule microarrays and diversity-oriented synthesis. Journal of the

American Chemical Society 125:8420-8421.

Kohn M, Wacker R, Peters C, Schroder H, Soulere L, Breinbauer R, Niemeyer CM and

Waldmann H (2003) Staudinger ligation: a new immobilization strategy for the preparation of

small-molecule arrays. Angew. Chem. Int. Ed. Engl. 42:5830-5834.

Kuruvilla FG, Shamji AF, Sternson SM, Hergenrother PJ and Schreiber SL (2002) Dissecting

glucose signalling with diversity-oriented synthesis and small-molecule microarrays. Nature

416:653-657.

Lam KS, Salmon SE, Hersh EM, Hruby VJ, Kazmierski WM and Knapp RJ (1991) A new type

of synthetic peptide library for identifying ligand-binding activity. Nature 354:82-84.

LeProust E, Pellois JP, Yu P, Zhang H, Gao X, Srivannavit O, Gulari E and Zhou X (2000)

Digital light-directed synthesis. A microarray platform that permits rapid reaction optimization on

a combinatorial basis. Journal of Combinatorial Chemistry 2:349-354.

Li S, Bowerman D, Marthandan N, Klyza S, Luebke KJ, Garner HR and Kodadek T (2004)

Photolithographic synthesis of peptoids. J. Am. Chem. Soc. 126:4088-4089.

Lo Conte L, Chothia C and Janin J (1999) The atomic structure of protein-protein recognition

sites. Journal of Molecular Biology 285:2177-2198.


at ASPE

T Journals on M


Dow

nloaded from


JPET#76943 PiP

25

MacBeath G, Koehler A and Schreiber S (1999) Printing Small Molecules as Microarrays and

Detecting Protein-Ligand Interactions en Masse. J. Am. Chem. Soc. 121:7967-7968.

Morken J, Kapoor T, Feng S, Fumiyuki S and Schreiber S (1998) Exploring the leucine-proline

binding pocket of the Src SH3 domain using structure-based, split-pool synthesis and affinty-

based selection. J. Am. Chem. Soc. 120.

Neuschafer D, Budach W, Wanke C and Chibout SD (2003) Evanescent resonator chips: a

universal platform with superior sensitivity for fluorescence-based microarrays. Biosensors &

Bioelectronics 18:489-497.

Nicolaou K, Pfefferkorn J, Roecker A, Cao G, Barluenga S and Mitchell H (2000) Natural

product-like combinatorial libraries based on priveleged structures. 1. General principles and

solid-phase synthesis of benzopyrans. J. Am. Chem. Soc. 122:9939-9953.

Nuwaysir EF, Huang W, Albert TJ, Singh J, Nuwaysir K, Pitas A, Richmond T, Gorski T, Berg

JP, Ballin J, McCormick M, Norton J, Pollock T, Sumwalt T, Butcher L, Porter D, Molla M, Hall

C, Blattner F, Sussman MR, Wallace RL, Cerrina F and Green RD (2002) Gene expression

analysis using oligonucleotide arrays produced by maskless photolithography. Genome Research

12:1749-1755.

Ohlmeyer MH, Swanson RN, Dillard LW, Reader JC, Asouline G, Kobayashi R, Wigler M and

Still WC (1993) Complex synthetic chemical libraries indexed with molecular tags. Proceedings

of the National Academy of Sciences of the United States of America 90:10922-10926.


at ASPE

T Journals on M


Dow

nloaded from


JPET#76943 PiP

26

Olivos HJ, Bachhawat-Sikder K and Kodadek T (2003) Quantum dots as a visual aid for

screening bead-bound combinatorial libraries. Chembiochem 4:1242-1245.

Pease AC, Solas D, Sullivan EJ, Cronin MT, Holmes CP and Fodor SP (1994) Light-generated

oligonucleotide arrays for rapid DNA sequence analysis. Proceedings of the National Academy of

Sciences of the United States of America 91:5022-5026.

Rakow NA and Suslick KS (2000) A colorimetric sensor array for odour visualization. Nature

406:710-713.

Ringe D (1995) What makes a binding site a binding site? Current Opinion in Structural Biology

5:825-829.

Ringe D and Mattos C (1999) Analysis of the binding surfaces of proteins. Medicinal Research

Reviews 19:321-331.

Scott JK and Smith GP (1990) Searching for peptide ligands with an epitope library. Science

249:386-390.

Singh-Gasson S, Green RD, Yue Y, Nelson C, Blattner F, Sussman MR and Cerrina F (1999)

Maskless fabrication of light-directed oligonucleotide microarrays using a digital micromirror

array. Nature Biotechnology 17:974-978.

Spencer RW (1998) High-throughput screening of historic collections: observations on file size,

biological targets, and file diversity. Biotechnology & Bioengineering 61:61-67.


at ASPE

T Journals on M


Dow

nloaded from


JPET#76943 PiP

27

Tallarico JA, Depew KM, Pelish HE, Westwood NJ, Lindsley CW, Shair MD, Schreiber SL and

Foley MA (2001) An alkylsilyl-tethered, high-capacity solid support amenable to diversity-

oriented synthesis for one-bead, one-stock solution chemical genetics. Journal of Combinatorial

Chemistry 3:312-318.

Taylor S and Morken J (1998) Thermographic selection of effective catalysts from an encoded

polymer-bound library. Science 280:267-270.

Thanos C, Randal M and Wells J (2003) Potent small-molecule binding to a dynamic hot spot on

IL-2. J. Am. Chem. Soc. 125:15280-15281.

Thorpe DS and Walle S (2000) Combinatorial chemistry defines general properties of linkers for

the optimal display of peptide ligands for binding soluble protein targets to TentaGel microscopic

beads. Biochemical & Biophysical Research Communications 269:591-595.

Tuerk C, MacDougal S and Gold L (1992) RNA pseudoknots that inhibit human

immunodeficiency virus type 1 reverse transcriptase. Proceedings of the National Academy of

Sciences of the United States of America 89:6988-6992.

Winssinger N, Harris JL, Backes BJ and Schultz PG (2001) From Split-Pool Libraries to

Spatially Addressable Microarrays and Its Application to Functional Proteomic Profiling.

Angewandte Chemie International Edition 40:3152-3155.

Zhu Q, Uttamchandani M, Li D, Lesaicherre ML and Yao SQ (2003) Enzymatic profiling system

in a small-molecule microarray. Organic Letters 5:1257-1260.


at ASPE

T Journals on M


Dow

nloaded from


JPET#76943 PiP

28

Footnotes

*to whom correspondence should be addressed

This work was supported by a grant from the New York Academy of Medicine, Speaker’s Fund

for Biomedical Research.


at ASPE

T Journals on M


Dow

nloaded from


ligand discovery using small molecule...

Documents