Aminodeoxychorismate synthase inhibitors from one-bead one-compound combinatorial libraries: "Staged" inhibitor design

Article abstract of DOI:10.1021/jm0609869

4-Amino-4-deoxychorismate synthase (ADCS) catalyzes the first step in the conversion of chorismate into p-aminobenzoate, which is incorporated into folic acid. We aim to discover compounds that inhibit ADCS and serve as leads for a new class of antimicrobial compounds. This report presents (1) synthesis of a mass-tag encoded library based on a "staged" design, (2) massively parallel fluorescence-based on-bead screening, (3) rapid structural identification of hits, and (4) full kinetic analysis of ADCS. All inhibitors are competitive against chorismate and Mg²⁺. The most potent ADCS inhibitor identified has a K_i of 360 μM. We show that the combinatorial diversity elements add substantial binding affinity by interacting with residues outside of but proximal to the active site. The methods presented here constitute a paradigm for inhibitor discovery through active site targeting, enabled by rapid library synthesis, facile massively parallel screening, and straightforward hit identification.

Full text of DOI:10.1021/jm0609869

J. Med. Chem. 2006, 49, 7413-7426
7413
Aminodeoxychorismate Synthase Inhibitors from One-Bead One-Compound Combinatorial
Libraries: “Staged” Inhibitor Design
Seth Dixon,† Kristin T. Ziebart,† Ze He,‡ Melissa Jeddeloh,† Choong Leol Yoo,† Xiaobing Wang,§ Alan Lehman,§
Kit S. Lam,§ Michael D. Toney,*,† and Mark J. Kurth*,†
Department of Chemistry, UniVersity of California, One Shields AVenue, DaVis, California 95616, Department of Chemistry,
Central Connecticut State UniVersity, 1615 Stanley Street, New Britain, Connecticut 06050, and Department of Internal Medicine,
DiVision of Hematology and Oncology, UC DaVis Cancer Center, 4501 X Street, Sacramento, California 95817
ReceiVed August 15, 2006
4-Amino-4-deoxychorismate synthase (ADCS) catalyzes the first step in the conversion of chorismate into
p-aminobenzoate, which is incorporated into folic acid. We aim to discover compounds that inhibit ADCS
and serve as leads for a new class of antimicrobial compounds. This report presents (1) synthesis of a
mass-tag encoded library based on a “staged” design, (2) massively parallel fluorescence-based on-bead
screening, (3) rapid structural identification of hits, and (4) full kinetic analysis of ADCS. All inhibitors are
competitive against chorismate and Mg2+. The most potent ADCS inhibitor identified has a Ki of 360 µM.
We show that the combinatorial diversity elements add substantial binding affinity by interacting with residues
outside of but proximal to the active site. The methods presented here constitute a paradigm for inhibitor
discovery through active site targeting, enabled by rapid library synthesis, facile massively parallel screening,
and straightforward hit identification.
Introduction
There is an ominous need for the discovery of fundamentally
new classes of anti-infectives to fight the large-scale acquisition
of bacterial and parasitic drug resistance. Historically, anti-
infectives have been based mostly on natural product scaffolds,
although some non-natural product based drugs are employed
clinically (e.g., fluoroquinolones, sulfa drugs). The modern
paradigm for discovery of new classes of non-natural product
based drugs hinges on the meticulous choice of drug target and
judicious attention to the molecular properties of drug candidates
during their development. Intelligent combinatorial chemistry,
wherein libraries are focused using knowledge of the structure
of the target protein and restrictions on molecular properties,
holds great promise for rapid lead discovery in anti-infective
development.
Figure 1. (A) Structure of chorismate and the chorismate-like inhibitor
chosen for use in the test library synthesis. (B) Staged approach to our
library design with the employed PAYLOAD derivative depicted.
Over the past decade, combinatorial chemistry and high-
throughput screening have advanced greatly as tools in drug
discovery.1-3 With one-bead one-compound (OBOC) methods,4-7
researchers have the tools to create vast libraries for discovery
of biologically active compounds. However, difficulties still arise
because of the sensitive nature of the solid support,8 the
challenges of synthesizing chemically unique structure-encoding
tags,9 the pitfalls of on-bead library testing,10-14 and the structure
decoding process.9,15,16 Diverse methods must be brought
together in a cohesive protocol to efficiently discover inhibitors
using OBOC methods. For example, a library should be
optimized by exploiting knowledge of the active/binding site.
Additionally, a structure-encoding method must be implemented
that accommodates the chemistry employed and that allows for
the rapid and reliable identification of surface-bound ligands.
Furthermore, elimination of false positives in on-bead screening
is a necessity.
Here, the target enzyme is 4-amino-4-deoxychorismate syn-
thase (ADCSa), which has a rich history because of its potential
as a target for antimicrobials.17-21 Its substrate is chorismate
(Figure 1A). Our library design incorporates a simple chorismate
analogue (PAYLOAD) to give active site specificity connected
through a SPACER to COMBI moieties. This “staged” design
is inspired by the cholesterol-lowering statin drugs that inhibit
hydroxymethylglutarate-CoA reductase. They contain a moiety
structurally analogous to hydroxymethylglutarate coupled to
ones that are completely unrelated to the cognate substrate.22
Indeed, there is recent literature supporting this basic design
strategy, which is grounded in multisubstrate analogue inhibitor
theory.23,24 Figure 1B illustrates this approach, where libraries
are designed to exploit binding opportunities not only within
* To whom correspondence should be addressed. For M.D.T.: phone,
530-754-5282; fax, 530-752-8995; e-mail, mdtoney@ucdavis.edu. For
M.J.K.: phone, 530-752-8192; fax, 530-752-8995; e-mail, mjkurth@
ucdavis.edu.
a Abbreviations: ADCS, 4-amino-4-deoxychorismate synthase; PLP,
pyridoxal phosphate; ADCL, 4-amino-4-deoxychorismate lyase; LDH,
lactate dehydrogenase; NADH, nicotinamide adenine dinucleotide, reduced
form; IPTG, â-isopropylthiogalactoside; DCM, dichloromethane; DMF,
dimethylformamide; DIEA, N,N-diisopropylethylamine; HOBt, n-hydroxy-
benzotriazole; DIC, diisopropylcarbodiimide; OBOC, one-bead one-
compound.
† University of California.
‡ Central Connecticut State University.
§ UC Davis Cancer Center.
10.1021/jm0609869 CCC: $33.50 © 2006 American Chemical Society
Published on Web 11/10/2006

7414 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 25
Dixon et al.
Figure 2. ADCS Ki values for potential PAYLOAD structures.
but also proximal to the ADCS active site. In the work reported
here, the PAYLOAD (chorismate analogue) and SPACER stages
(lysine) were held constant while the COMBI stage [acid-
(lysine-PAYLOAD)-amino acid] was varied.
inhibitors. Surprisingly, even simple chorismate analogues (as
well as further elaborated inhibitors) are competitive against
both chorismate and Mg2+. This knowledge allowed us to show
that PAYLOAD and PAYLOAD-SPACER inhibitors bind
∼60-fold weaker than chorismate and that the most potent
inhibitor identified here binds ∼1.5-fold more tightly than
chorismate.
With these tools now in place and our inhibitor design
principle validated, focused libraries that include diversity
structures more suited to ADCS inhibition can be produced.
Most importantly, the methods reported here seamlessly bridge
organic synthesis, combinatorial chemistry, and biological
screening in a streamlined and cohesive protocol for the rapid
discovery of biological ligands from highly diverse libraries.
With this library design in mind, cyanogen bromide cleavable
mass spectrum tags were chosen as a structure-encoding method
for rapid identification of surface-presented molecules. Our
previous experience with mass-tag encoded libraries nicely
situated us for this work.9,25 One concern with using encoding
tags with on-bead screening is that they will interact with the
target protein, leading to misidentification of the test compounds
as the ligands. We were confident this would not be a problem
using topologically segregated bifunctional beads because our
recent work16 demonstrated that the resin structure and the
density of surface-bound ligands restrict proteins to the outside
of the beads. Thus, confining the encoding tags to the bead
interior exposes the protein only to surface ligands in on-bead
assays.
Last, a facile method for high-throughput identification of
ADCS-bound (“hit”) beads is needed. On the basis of our
previous experience, we decided to employ commercial, acti-
vated fluorescent dyes to label the surface of ADCS at lysine
side chains.26 This allows identification of ADCS-bound beads
by manual observation under a fluorescence microscope or by
automated COPAS (complex object parametric analyzer and
sorter) sorting. Our original application of this technique was
limited by background autofluorescence of random library beads,
leading to false positives. Autofluorescence is a property
inherent to all libraries we have studied and has been reported
by others in the literature.13,27 Therefore, automated COPAS
sorting of the library prior to screening with fluorescently
labeled ADCS was used to remove beads that autofluoresced.
The COPAS instrument has the additional advantage of
identifying the region of the spectrum that has the lowest amount
of autofluorescence, allowing removal of the fewest beads from
the library and selection of the most appropriate fluorescent dye.
A meaningful kinetic analysis of ADCS inhibitors requires
an understanding of the enzyme’s kinetic mechanism. ADCS
requires Mg2+ for activity, which binds to the C1 carboxylate
of chorismate and several active site glutamates at the solvent-
exposed end of the active site. We therefore expected ADCS
(and other homologous enzymes by extension) to obey an
ordered kinetic mechanism with chorismate binding first and
Mg2+ second. Initial rate kinetic data confirm this, as do
fluorescence-quenching experiments with chorismate and staged
Results
Design and Synthesis of the PAYLOAD and SPACER.
Figure 2 depicts several PAYLOAD compounds that were
considered for library synthesis. P2, P3, P4, and P5 are available
from commercial sources. P6,28 P7,29 and P930 were synthesized
according to literature procedures. P1 was synthesized according
to procedures outlined in Scheme 1. Diels-Alder cycloaddition
of the modified Danishefsky’s diene (P1a) with methyl acrylate
produced a diastereomeric mixture of cycloadducts (P1b).
Exposure of these cycloadducts to a 10% HF-pyridine complex
in acetonitrile led to a mixture of isomeric enones (P1c and
P1c′). Vinylic alkylation of the enone mixture with methyl
acrylate was expected to produce P1d′, which would deliver
P1′, the actual target of this synthesis, upon ester hydrolysis.
Exclusive formation of P1d occurred instead, resulting in
synthesis of the revised PAYLOAD P1.
Scheme 2 describes the synthesis of P8. Reduction of P8a
to P8b enabled N-alkylation with tert-butyl 2-bromoacetate to
produce P8c. Subsequent saponification of P8c produced
PAYLOAD P8. P10 was chosen as the library PAYLOAD
because of its ease in synthesis, simple adaptability to solid-
phase synthesis, and its structural similarity to chorismate (i.e.,
a carboxylic acid containing side chain); its synthesis is
described below and in Scheme 3.
For this proof-of-principle library, semiorthogonally protected
lysine (i.e., Fmoc-Lys(Dde)-OH) was chosen as a framework
on which to tether the PAYLOAD and COMBI stages. The
carboxylic acid of the lysine was used in peptide bond formation
with an amino acid diversity element, and the R-amine was

ADCS Inhibitors from High-Throughput Screening
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 25 7415
Scheme 1. Synthesis of P1a
a Reagents and conditions: (i) methyl acrylate, toluene, room temp, 16 h; (ii) 10% HF/pyridine in CH3CN, room temp, 2 h; (iii) methyl acrylate, DBU,
DMF, reflux, 16 h; (iv) Ba(OH)2, MeOH, room temp, 16 h.
Scheme 2. Synthesis of P8a
a Reagents and conditions: (i) H2, Pd-C, EtOH, 16 h, room temp; (ii) TEA, THF, N2, ∼50-60 °C, 16 h; (iii) TFA/DCM (1:1, v:v).
N-acylated with a carboxylic acid diversity element to deliver
the COMBI stage. The ꢀ-amine incorporated the PAYLOAD
stage through N-acylation. The coupling strategy required that
the aryl carboxylic acid of PAYLOAD P10 be unprotected while
the aryloxy acetic acid moiety is protected as a tert-butyl ester.
As outlined in Scheme 3, the aryl acid moiety was protected as
the methyl ester because it competed in the SN2 displacement
of bromide in the tert-butyl 2-bromoacetacetate alkylation step.
O-Alkylation of this phenol (1) delivered 2, and subsequent
treatment with lithium hydroxide led to a mixture of the desired
product 3 as well as 33% of the diacid P10. A sample of P10
was saved for inhibition studies, and the remainder of P10 was
converted back into 3 by treatment with Boc-anhydride. This
final conversion was carried out in high yield with complete
selectivity for the aliphatic carboxylic acid.
Three other derivatives (4, 5, and 7) were synthesized to
deconvolute the contributions of each inhibitor stage to binding.
Compound 4 was synthesized from 3 using Rink amide resin.
Compound 5 was synthesized to determine the contribution of
the lysine SPACER to binding affinity. Solid-phase synthesis
was again employed, giving 5 in high purity and good yield.
Compound 7 was synthesized to determine if having a positively
charged amino group near the normal position of the Mg2+ ion
would enhance inhibitor binding. A reductive amination was
performed on solid phase with 6, the aldehyde analogue of 3,
to deliver 7 in high purity and good yield after a series of
efficient steps.
solubility and molecular weight, which facilitates MALDI-TOF
analysis by avoiding mass signals due to the matrix. The
remaining transformations outlined in Scheme 4 completed
construction of the coding tags and the target molecule found
on the surface of the bead.
The protecting groups on the coding tag (Fmoc) and the
surface of the bead (Alloc) were removed, and the first COMBI
diversity element (amino acid X1) was introduced. Next, the
semiorthogonally protected lysine was attached to the growing
coding tag and the target molecule. The beads were then swollen
again in water for 24 h, and the outer surface and a fraction of
the coding tag on the interior were protected with Fmoc (0.7
equiv), leaving the center of the bead with a free amine that
was orthogonally protected with Boc. In this way, the innermost
volume of the bead delivers a tag that reports the first COMBI
diversity element (e.g., X1).
The second COMBI diversity element was introduced after
Fmoc deprotection. PAYLOAD 3 was then coupled to the Dde-
deprotected distal amine. The second coding tag contained the
complete target molecule. The entire surface target molecule
could then be identified by MALDI-TOF after CNBr cleavage
of the coding tags from the resin. PAYLOAD 3 coupling and
removal of all protecting groups completed the MS-encoded
library synthesis. The incorporation of 48 amino acid diversity
elements for X1 and 48 carboxylic acid diversity elements for
X2 resulted in 2304 test compounds.
Identification of each diversity element was accomplished
with the following equations:
Library Synthesis. The library used in this work was
prepared as outlined in Scheme 4, using topologically segregated
amino-functionalized resin. By employment of ∼90% of the
total available functional groups for the encoding tag (restricted
to the bead interior), strong MALDI-TOF tag signals were
reliably detected.
X1 ) M1 - 712 - (K coupled with P10)
X2 ) M2 - M1 - 712 - (K coupled with P10)
Synthesis of the encoding tag was performed largely accord-
ing to our published procedures.9 The cleavable linker comprises
four entities: (1) methionine, which enables resin cleavage by
cyanogen bromide (methionine + cyanogen bromide f ho-
moserine lactone);31 (2) â-bromoalanine, which facilitates tag
identification via bromine isotopic patterns in the mass spec-
trum;32 (3) arginine; and (4) a linker. The last two increase tag
In these equations, M1 and M2 are MALDI-TOF mass signals
corresponding to the first and second tag signals that each
display a signature bromine isotopic pattern. As a test, several
beads were removed from the library during synthesis and
sequenced to validate the library synthesis. These beads
delivered the expected MALDI-TOF signals at each test point.
These spectra are supplied in the Supporting Information.

7416 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 25
Scheme 3. Synthesis of P10 and Other Derivativesa
Dixon et al.
a Reagents and conditions: (i) AcCl, MeOH, N2, reflux, 8 h; (ii) K2CO3, THF, N2, 4 h; (iii) LiOH, H2O, THF, 6 h; (iv) (Boc)2O, DMAP, CH3CN, N2,
0 °C, room temp, 16 h; (v) 20% piperidine in DMF, 2 × 10 min; (vi) 3, HOBt, DIC; (vii) Fmoc-Lys(Dde)-OH, HOBt, DIC; (viii) acetic anhydride (10
equiv), DIEA (12 equiv), DMF; (ix) 2% hydrazine in DMF, 2 × 10 min; (x) 2, HOBt, DIC; (xi) 6, NaCNBH3, 1% AcOH, TMOF; (xii) TFA/H20/TIS (95,
2.5, 2.5, v/v/v).
COPAS Sorting of the Library. In our and others’ experi-
ence, combinatorial libraries created using OBOC methods with
aromatic compounds contain beads that autofluoresce. Therefore,
this library was presorted to remove strongly autofluorescent
beads prior to screening with fluorescently labeled ADCS. This
presort established which region of the fluorescence spectrum
exhibited the most autofluorescence and helped define which
fluorescent dye would incur the least intereference in the
fluorescent-based screening. Relative to a defined baseline, the
red and green regions of the spectrum exhibited 15% and 5%
autofluorescence, respectively. Autofluorescence was more
frequent but less intense in the red region and more intense
and less frequent in the green region. Therefore, the sort was
performed in the green region such that the number of eliminated
beads was minimized. The results of the presorting are shown
in Figure 3; the original library contained autofluorescing beads
that the COPAS instrument effectively removed (see Figure S1
in the Supporting Information).
each. In each batch, ∼2% of the beads that strongly fluoresced
were isolated manually with a micropipette; of these, 15 of the
brightest fluorescing beads were selected. The “hit” beads were
washed with 8 M guanidine hydrochloride (to remove ADCS)
and water.
MALDI-TOF Analysis and Resin-Free Resynthesis. Each
of the washed “hit” beads was individually incubated with
cyanogen bromide to cleave the tags. MALDI-TOF analysis
produced signals that allowed surface ligand identification for
11 of the 15 washed and cleaved “hit” beads. This corresponds
to 73% readability, which is in accord with literature values.33
The 11 compounds identified by the MALDI-TOF are shown
in Figure 4. As a representative example, the spectrum used to
identify L2 is shown in Figure 5. Additional examples are given
in the Supporting Information. From these results, two sets of
duplicate ligands (L1 ) L4; L3 ) L8 ) L11) were found,
suggesting that the majority of the library had been screened
and that the ligands identified were among the tightest binding.
These ligands exhibit a distinct structural pattern, with each
containing a hydrophobic and hydrophilic group in the COMBI
stage. The ligands were resynthesized on-bead using Rink amide
resin and released for kinetic analysis. The resynthesized ligands
were purified to >95% by HPLC.
Screening of the Library. The Alexa Fluor 488 labeled
enzyme exhibited 84% of the activity of the unlabeled enzyme,
as determined from the LDH-coupled assay (vide infra). After
a 1 h incubation with labeled enzyme, the library was viewed
under a fluorescence microscope in 15 batches of ∼2500 beads

ADCS Inhibitors from High-Throughput Screening
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 25 7417
Scheme 4. Synthesis of MS-Encoded Librarya
a Reagents and conditions: (i) H2O, 48 h; (ii) Alloc-OSu (0.1 equiv), DIEA, DCM/Et2O (55:45, v:v), 30 min; (iii) Fmoc-Met-OH, HOBT, DIC; (iv) 20%
piperidine in DMF, 2 × 10 min; (v) N-Fmoc-3-(4-bromophenyl)-â-alanine, HOBt, DIC; (vi) Fmoc-Arg(Pmc)-OH, HOBt, DIC; (vii) N-Fmoc-2,2′-
ethylenedioxybis(ethylamine)monosuccinamide, HOBt, DIC; (viii) Pd(PPh3)4 (0.24 equiv), PhSiH3 (20 equiv), DCM; (ix) split beads into 42 columns; (x)
library assembly (Fmoc-X1-OH, HOBt, DIC); (xi) mix beads; (xii) Fmoc-Lys(Dde)-OH, HOBt, DIC; (xiii) H2O, 24 h; (xiv) Fmoc-Osu (0.7 equiv), DIEA,
DCM/Et2O (55:45, v:v), 1 h; (xv) (Boc)2O (1 equiv), DIEA, DCM; (xvi) library assembly X2-CO2H, DIC, HOBt; (xvii) 2% hydrazine in DMF; (xviii)
3-(2-tert-butoxy-2-oxoethoxy)benzoic acid (3), HOBt, DIC; (xix) TFA/phenol/TIS/H2O/ethanedithiol (90/5/2/2/1, v/w/v/v/v).
Kinetic Results. Initial rate data for determining the kinetic
mechanism of ADCS were obtained by measuring the increase
in absorbance at 290 nm due to PABA formation in a coupled
assay with excess ADCL. Figure 6A shows the results of these
experiments; the data fit best to the equation for an ordered
sequential mechanism, with chorismate binding first:
analysis of the data revealed each compound to be a competitive
inhibitor of ADCS. The data obtained for P10 are depicted in
Figure 6B as a representative example. Kinetic analyses were
performed for all identified ligands. All have lower Ki values
relative to P10 (Figure 2).
Surprisingly, compounds P10, 4, 5, and L5 were all found
to be competitive with respect to Mg2+ as well as chorismate.
As a representative example, Figure 6C illustrates the data
obtained for P10. The LDH-coupled assay was used. Inhibitor
and Mg2+ were varied across appropriate concentrations, while
chorismate was held constant at 200 µM.
Vi )
Vmax[chorismate][Mg2+]
KchorismateKMg2+ + KMg2+[chorismate] + [chorismate][Mg2+]
(1)
Discussion
The data in Figure 6A provided initial estimates of the kinetic
constants Kchorismate and KMg . More accurate estimates were
obtained from experiments in which the concentration of one
substrate was varied across a wide range while the other was
The increase in drug-resistant pathogenic microorganisms and
the near-abandonment of antimicrobial drug development by
the pharmaceutical industry make the involvement of academia
in this pursuit a pressing priority. The success of academic
efforts will hinge on creative methods development that enables
discovery with substantially less manpower than that available
to companies. An obvious basis on which to choose drug targets
is their essentiality to microbial survival and their absence in
humans. Herein, the development of facile, powerful methods
for readily automatable massively parallel screening of OBOC
combinatorial libraries has been described in application to the
discovery of ADCS inhibitors, this enzyme and its homologues
being critical to microbial survival as well as being absent in
mammals.
2+
held constant at its estimated KM value. This gave Kchorismate
0.50 ( 0.05 mM and KMg ) 10 ( 1 µM.
)
2+
Each inhibitor was assayed against ADCS by varying the
inhibitor concentrations across an appropriate range and holding
the chorismate concentration constant at 13 µM (the KM value
under the assay conditions). Strong absorbance below 300 nm
due to the inhibitors necessitated the use of a second coupling
enzyme, LDH. Initial rates were measured by following the
decrease in absorbance at 340 nm due to NADH oxidation. The
data were fitted to the following competitive inhibition equation:
The shikimate pathway converts glucose into chorismate, the
branch-point metabolite for carbocyclic aromatic compound
biosynthesis in biological systems. Five primary metabolic
pathways start from chorismate (Phe/Tyr, salicylate/sidero-
phores, PABA/folates, Trp, and electron-transfer quinones),
while several secondary metabolites also begin with choris-
mate.18 These pathways, including the shikimate pathway itself,
are absent in humans; their products are essential components
V
max[chorismate]
Vi )
(2)
[I]
KM
+ [chorismate]
( )
Ki
As justification for this fit, P9, P10, 4, 5, 7, and L5 were
subjected to further kinetic analysis, wherein inhibitor and
chorismate concentrations were each varied. Lineweaver-Burk

7418 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 25
Dixon et al.
Figure 3. COPAS sorts of beads based on autofluorescence: (A) first sort of the target library in which 12 541 beads were sorted in the batch
shown, with 6.0% beyond the user-defined threshold (black line in plot); (B) second sort of the target library in which 7735 beads were sorted in
the batch shown, with 4.0% of the beads above the threshold; (C) third sort of the target library in which 21 118 beads were sorted in the batch
shown, with less than 1.0% above the threshold. Blue, green, and red dots represent the occurrence of 100, 10, and 1 beads, respectively, in the
same location. The red circles are added to focus attention on these regions of the plots, which highlight the removal of green-fluorescing beads.
of our diet. The attractiveness of these pathways as antimicrobial
targets has long been recognized.19,20 Several effective present
day commercial compounds/drugs act on these pathways,
including the commercial herbicide glyphosate, the sulfa drugs,
and others.
make productive binding interactions with the enzyme surface
proximal to the active site.
The first task was to find a stable chorismate analogue that
would be easily amenable to solid-phase synthesis and would
direct the inhibitor library to the active site. On the basis of
previous inhibition studies by others, a side chain containing a
carboxylate group linked through one or two carbons was
essential for tight binding to IS and AS.36,37 To this end, we
surveyed the compounds presented in Figure 2 as possible
candidates. P10 was chosen as the best compromise between
all of these requirements.
The second component in our design is the SPACER. In the
full embodiment of our discovery efforts, this moiety will be
the target of diversification and optimization through combi-
natorial chemistry and screening with labeled enzyme. For this
initial proof-of-principle demonstration, we decided to simplify
matters and use lysine as the SPACER; modeling showed that
it is an excellent first approximation to the length required to
extend out of the active site to the surface cleft, where the
COMBI moiety would have opportunities for serendipitous
interactions with residues near the surface (see Figure S2 in
the Supporting Information). Lysine also contains the R-amino
and R-carboxylate functional groups that are readily amenable
to solid-phase elaboration with the combinatorial elements.
A series of 48 acids were chosen as diversity elements for
acylation of the R-amine and a series of 48 amino acids were
chosen as diversity elements for coupling to the R-carboxylate
of lysine, leading to our proof-of-principle one-bead one-
compound library of 2304 compounds. Encoding of the
structures was achieved with a mass spectrum tagging scheme
using topologically segregated resin beads, described in detail
above, which allowed rapid and facile test compound structure
identification (Figure 5).
With the library in hand, we first presorted it using the
COPAS instrument to increase the signal-to-noise for hit
compounds by removing the autofluorescing beads that interfere
with identification of ADCS-bound beads (Figures 3 and S1).
This presorting had a major impact on our ability to identify
ADCS-bound beads with a low false-positive frequency, since
6% of the beads were removed in the presort and ∼2% of the
presorted beads incubated with labeled ADCS were fluorescent.
ADCS, anthranilate synthase, and isochorismate synthase are
three structurally and mechanistically homologous chorismate-
utilizing enzymes that initiate PABA, Trp, and salicylate
biosynthesis, respectively. Our antimicrobial drug discovery goal
is to find a compound that is a tight-binding inhibitor of one or
more of these three similar enzymes, since each pathway is
critical to bacterial pathogenicity.34,35 A compound that inhibits
more than one will have a radically reduced likelihood that
primary resistance by simultaneous mutation of two or more of
these enzymes will develop. It would be equivalent to a
multidrug cocktail in a single inhibitor molecule. The high
degree of similarity between these enzymes makes this lofty
goal a reasonable pursuit. We report inhibition studies targeting
ADCS here. Efforts targeting anthranilate synthase and isocho-
rismate synthase will be undertaken in due course.
Given this stated goal, we decided that, rather than optimize
a single small substrate-like molecule for tight inhibition of two
or more of these enzymes, a more generalized strategy would
have a higher probability of success. The structures of these
enzymes all show that the active sites are removed from solvent
but not extraordinarily deeply buried in the protein interior.
While we pondered the structures, the popular cholesterol-
lowering statin drugs came to mind because these varied
compounds all have a substrate-like moiety that directs them
to the active site. This substrate-like moiety does not give tight
binding but is connected by a “linker” moiety to a group of
structurally diverse moieties, each of which is unique to a given
drug. This latter component makes serendipitous interactions
with the enzyme proximal to the active site and contributes the
greatest portion of the binding energy of these tight-binding
hydroxymethylglutarate-CoA inhibitors.22 We reasoned that a
similar strategy might be the most productive for finding a single
compound that inhibits one or more of ADCS, AS, and IS.
We have termed the tripartite inhibitor design employed here
a “staged” design in which there is an active site-directing
PAYLOAD stage, a SPACER stage that extends out of the
active site, and a COMBI stage that is the focus of the
combinatorial chemistry employed here and that is expected to
Figure 4 shows 11 structures obtained from the 15 brightest
beads picked in the ADCS screen. These were resynthesized

ADCS Inhibitors from High-Throughput Screening
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 25 7419
Figure 4. Ki values for ligands identified from the Alexa Fluor 488 labeled ADCS screen. L1 and L4 are identical structures, as are L3, L8, and
L11.
on Rink amide resin, and the inhibition constants of those
sufficiently soluble to assay are reported under the structures.
These compounds are not tight binding by any standard, but
they serve the very important role of demonstrating that the
“staged” inhibitor design is a promising line of pursuit. Each
of the staged inhibitors assayed has a substantially lower
inhibition constant than P10, indicating that the SPACER and/
or COMBI moieties provide significant increases in binding
affinity.
The tightest binding inhibitor identified is L5 with a Ki value
of 360 µM. We designed and synthesized the derivatives shown
in Figure 7 to deconvolute the contributions to binding affinity
by the various components. Surprisingly, the conversion of P10
into its simple amide (4) had no significant effect on binding
affinity. This important point is addressed below in relation to
the kinetic mechanism of ADCS. Additional elaboration by
formation of the amide with the ꢀ-amino group of N-acetyl-L-
lysine amide (5) also does not significantly alter the binding
affinity of the P10-based inhibitor, demonstrating that any
potential negative interactions between the lysine and the
enzyme are offset by positive ones and more importantly that
all of the interactions that lead to the ∼100-fold increase in
binding affinity of L5 compared to P10 are entirely due to
interactions made between the COMBI moiety and the enzyme
surface proximal to the active site.
The kinetic mechanism of ADCS has not been characterized
in the literature. It is important for our analyses here to
understand whether chorismate and Mg2+ binding obeys a
random or ordered sequential mechanism so that the effect of
their concentrations on inhibitor binding can be ascertained. Our
results show that ADCS obeys a strictly ordered kinetic
mechanism with chorismate binding first and Mg2+ second:
Kchorismate
KMg
2+
E + chorismate y
z E‚chorismate y
z
Mg2+
kcat
E‚chorismate‚Mg2+ 8 E + ADC + Mg2+
Full inhibition analyses of several inhibitors including P10,
4, 5, and L5 show that they are all competitive with respect to
both chorismate and Mg2+. The results for P10 are shown in
Figure 6B. This was expected for the latter compounds but not
for P10. It is a close analogue of chorismate, whose C1
carboxylate group is a bidentate ligand to Mg2+ in the active
enzyme complex.38
Given that P10 is structurally different enough from choris-
mate to preclude Mg2+ binding, then the correct comparison
for binding is not to chorismate in the ternary complex with
enzyme and Mg2+ but rather to chorismate in the binary complex
with enzyme alone. This value, Kchorismate in the ordered kinetic

7420 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 25
Dixon et al.
Figure 5. Example MALDI-TOF spectra obtained from an ADCS-selected hit bead. Each major peak exhibiting a monobromine splitting pattern
was used to calculate the diversity element present; the diversity structure determined is shown with the cleavable linker. Peak A is the mass signal
of a peptide fragment without the bromine component. Peak B is the mass signal used to calculate tag 1. Peak C is the mass signal used to calculate
tag 2, and peak D is the mass signal associated with a formylated tag 2.
mechanism, was determined here to be 0.5 mM, much larger
(or that the amide group provides for energetically equivalent
interactions).
than the value of 13 µM previously measured for chorismate
in the presence of 5 mM Mg2+ 39
Therefore, P10 binds only
.
The similar Ki for 5 compared to P10 and 4 implies that the
lysine SPACER moiety incurs neither good nor bad interactions
(or they are equal) with the enzyme as it traverses the Mg2+
binding site to the enzyme surface. This positive result
demonstrates that the steric bulk of the linker is readily
accommodated. The lysine SPACER used here was chosen for
synthetic facility. A combinatorial effort aimed at optimizing
the PAYLOAD-SPACER combination has the potential to
generate substantial binding energy from the SPACER moiety
given several glutamates (which form most of the Mg2+ binding
site) and other polar groups in this region. Retrospectively, an
initial effort was made to capture some of this binding energy
by synthesizing 7. It was disappointing to find that conversion
of the amide of 5 to the benzylamine of 7 did not generate
substantial additional binding interactions. This may well be
due to the suboptimal position of the positively charged amino
group that was intended to capture some of the binding energy
Mg2+ enjoys. It is likely that SPACER optimization through a
non-peptide-based combinatorial program will remedy this.
∼60-fold more weakly than chorismate to give the equivalent
enzyme complex. Competitive inhibition with respect to both
chorismate and Mg2+ by simple chorismate analogues with
ADCS but not with anthranilate synthase or isochorismate
synthase may explain previous inhibition studies that showed
that chorismate analogues generally bind more tightly to the
latter two compared to ADCS in the presence of Mg2+ 37,40
.
The competitive inhibition of P10 with respect to both
chorismate and Mg2+ also provides a satisfying explanation for
why conversion of P10 to its amide 4 does not drastically
increase the inhibition constant. Conversion of the carboxylate
group to the amide would be expected to disrupt the energeti-
cally favorable Mg2+ binding if P10 did, in fact, allow Mg2+
to bind. The observation that conversion to the simple amide
has an insignificant energetic effect is additional evidence that
Mg2+ does not bind to the ADCS-P10 complex. Furthermore,
this also suggests that the negative charge of the aryl carboxylate
group of P10 does not contribute significantly to its binding

ADCS Inhibitors from High-Throughput Screening
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 25 7421
carboxylate used for resin attachment. For both L3 and L5 the
inclusion of the linker analogue had only a minor effect on the
affinity of the test compound for ADCS. This gratifying result
demonstrates that covalent attachment of the test compound to
the resin through the linker has negligible deleterious effects
on the affinity of the enzyme for it.
This work has demonstrated the viability of a general
paradigm for drug discovery based on combinatorial chemistry
and a “staged” inhibitor design. A great majority of enzymes
bind their substrates in an active site that is close to the surface
of the protein and have an open channel that leads to solvent.
A substrate-like moiety that directs a molecule to an enzyme
active site, even if it alone has a relatively low affinity, can be
elaborated to take advantage of serendipitous interactions that
can lead to high affinity. The use of fluorescently labeled
enzyme to detect binding to surface-bound test compounds
ultimately enables full automation of this simple screening
procedure by COPAS-based presorting of the library followed
by sorting of the enzyme binding reactions in which beads
become fluorescent because of enzyme association to surface
ligands.
Materials and Methods
Reagent and Instrumentation. All reagents and solvents were
purchased from commercial suppliers and used without further
purification. TentaGel S NH2 resin (90 µm, 0.26 mmol/g) was
purchased from Rapp Polymere GmbH (Tu¨bingen, Germany), and
all the calculations for synthesis were based on a substitution of
0.26 mmol/g. Fmoc-protected amino acids were purchased from
Chem-Impex International, Inc. (Wood Dale, IL) and Advanced
ChemTech (Louisville, KY). N-Fmoc-3-(4-bromophenyl)-â-alanine
was purchased from InnovaChem (Tuscon, AZ). N-Fmoc-2,2′-
ethylenedioxybis(ethylamine)monosuccinamide was synthesized
according to literature procedure.25 HOBt, Fmoc-OSu, and DIC
were purchased from GL Biochem (Shanghai, China). Alexa Fluor
488 succinimidyl ester was purchased from Invitrogen (Molecular
Probes). Fmoc Rink amide MHBA resin (capacity, 0.65 mmol/g)
was purchased from Nova Biochem. Melting points are uncorrected.
All infrared spectra were determined on a Genesis II Mattson FT-
IR. 1H and 13C NMR were measured in either DMSO-d6 or CDCl3
at 400 (or 300) and 100 (or 75) MHz, respectively. Elemental
analyses were determined at MidWest Microlab, Indianapolis, IN.
MALDI-TOF MS analyses were performed with a Bruker Biflex
III MALDI-TOF mass spectrometer (Bruker-Franzen Analytik,
Bremen, Germany) equipped with a pulsed N2 laser (337 nm), a
delayed extraction ion source, and a reflectron. Waters Alliance
LC/MS, equipped with a Waters 2695 HPLC and a Waters PDA
996, was used for HPLC purity analyses employing a gradient
elution of 0-100% B over 30 min, monitored from 200 to 400
nm. A Waters Micromass ZQ mass detector was used in ESI MS
for the identification of various products, with a product concentra-
tion of 1 µg/mL. Further analytical HPLC was performed on a
System Gold 125 solvent module (Beckman) with a C18 column
(Vydac, 5 µM, 4.6 mm i.d. × 25 cm). A gradient elution of 0-80%
solvent B over 25 min followed by 80-100% solvent B over 3
min was used at a flow rate of 1 mL/min (solvent A is H2O/0.1%
TFA; solvent B is acetonitrile/0.1% TFA). Preparative HPLC was
performed on a System Gold 126NMP solvent module (Beckman)
with a C18 column (Vydac, 5 µM, 2.5 cm i.d. × 25 cm). A gradient
elution of 0-60% solvent B over 25 min followed by 60-100%
solvent B over 25 min followed by 100% solvent B for 5 min was
used at a flow rate of 7 mL/min (solvent A is H2O/0.1% TFA;
solvent B is acetonitrile/0.1% TFA). The COPAS sorter was
purchased from Union Biometrica (Somerville, MA).
Figure 6. Kinetic analyses of ADCS. The structures of the compounds
whose concentrations were varied along with those indicated on the
X-axes are shown in the figures. (A) Double-reciprocal plot showing
the effect of Mg2+ upon chorismate binding. This and other data (not
shown) definitely demonstrate that the kinetic mechanism for ADCS
is ordered, with chorismate binding first and Mg2+ second. The
concentrations of chorismate are 50, 100, 250, 375, and 650 µM. (B)
Double-reciprocal plot showing that P10 is competitive against cho-
rismate. The concentrations of P10 are 0, 10, 25, 40 mM. (C) Double-
reciprocal plot showing that P10 is competitive against Mg2+. The
concentrations of P10 are 0, 15, 22.5, 30 mM.
Last, the effect of the resin linker on the affinity of ADCS
for resin-bound test compounds was delineated. Figure 7
presents the structures of L3 and L5 derivatives where a close
structural analogue of the resin linker is attached to the
Synthesis of Methyl 3-hydroxybenzoate (1). m-Hydroxybenzoic
acid (10.0 g, 72.5 mmol) was dissolved in methanol (70 mL), and
the solution was placed under nitrogen and cooled in an ice bath.
Acetyl chloride (7.2 mL, 7.96 g, 101.4 mmol) was added via

7422 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 25
Dixon et al.
Figure 7. Structures and Ki values for inhibitor elaborations and derivatives.
syringe. The reaction mixture was removed from the ice bath and
then heated to reflux in an oil bath for 8 h, after which the solvent
was removed by rotary evaporation. The resulting solid was
dissolved in water (50 mL) and extracted with ethyl acetate (3 ×
50 mL). The combined organic layers were dried with MgSO4 and
concentrated by rotary evaporation. Crude product (1) was used
without further purification; yellow solid (10.65 g, 93.7% crude
yield, purity ∼99%). MS (ESI) m/z: 152.02 [M+]. 1H NMR
(DMSO-d6): δ 9.83 (s, 1H), 7.39 (m, 3H), 7.16 (m, 1H), 3.83
(s, 3H).
purification with 3:1 n-hexane/ethyl acetate with 1% acetic acid to
deliver 3 followed by 1:1 n-hexane/ethyl acetate with 2% acetic
acid to deliver P10 (84% overall yield: 67% yield of 3 and 33%
yield of P10).
3. Mp 131-134 °C. IR (neat, selected peaks) 1743.56, 1675.52
cm-1. MS (ESI) m/z: 252.15 [M+]. 1H NMR (DMSO-d6): δ 7.55
(d, J ) 8.0, 2H), 7.40 (m, 2H), 7.17 (d, J ) 8.4, 1H), 4.72 (s, 2H),
1.42 (s, 9H). 13C NMR (DMSO-d6): δ 167.8, 167.0, 157.7, 132.2,
129.8, 122.1, 119.5, 114.6, 81.6, 65.1, 27.7. Anal. Calcd for
C13H16O5: C, 61.90; H, 6.39; O, 31.71. Found: C, 61.61; H, 6.34;
O, 32.05.
Synthesis of Methyl 3-(2-tert-Butoxy-2-oxoethoxy)benzoate
(2). Compound 1 (5.0 g, 32.7 mmol) was dissolved in THF (70
mL), and to this solution was added K2CO3 (13.5 g, 98.0 mmol).
The mixture was placed under nitrogen and allowed to stir for 1 h
and then cooled in an ice bath. tert-Butyl 2-bromoacetate (3.96 g,
39.2 mmol) was added via syringe. The solution was removed from
the ice bath and warmed to room temperature at which time TLC
indicated the reaction had gone to completion. The solvent was
removed by rotary evaporation. The resulting solid was dissolved
in water (50 mL) and extracted with ethyl acetate (3 × 50 mL).
The combined organic layers were dried with MgSO4 and concen-
trated by rotary evaporation. The crude product (2) was used without
further purification as a yellow solid (8.39 g, 96.7% crude yield,
P10. Mp 203-205 °C. IR (neat, selected peaks) 1704.40 cm-1
1
(two peaks coalescing). MS (ESI) m/z: 196.34 [M+]. H NMR
(DMSO-d6): δ 7.58 (d, J ) 7.2, 1H), 7.39-7.44 (m, 2H), 7.17-
7.19 (d, J ) 8.0, 1H), 4.76 (s, 2H). 13C NMR (DMSO-d6): δ 170.2,
167.1, 157.8, 132.2, 129.8, 122.1, 119.5, 114.6, 64.6. Anal. Calcd
for C9H8O5: C, 55.11; H, 4.11; O, 40.78. Found: C, 54.99; H,
4.22; O, 40.79.
Method B: Synthesis of 3-(2-tert-Butoxy-2-oxoethoxy)benzoic
Acid (3). Compound P10 (3.65 g, 18.6 mmol) was dissolved in
acetonitrile (25 mL), and DMAP (227 mg, 1.86 mmol) was added.
The reaction mixture was placed under nitrogen and cooled in an
ice bath. Di-tert-butyl dicarbonate (6.10 g, 27.9 mmol) dissolved
in acetonitrile (25 mL) was added via syringe, and the reaction
mixture was allowed to stir overnight during which time the ice
bath expired. The reaction mixture was concentrated by rotary
evaporation, and the resulting red-brown oil was purified by silica
gel column purification (3:1 n-hexane/ethyl acetate with 1% acetic
acid) to yield 3 as a white powder (3.75 g, 80% yield; data as listed
above).
Synthesis of Methyl 2-(3-Carbamoylphenoxy)acetic Acid (4).
Rink amide MBHA (2 g, 0.5 mmol/g loading capacity) resin was
weighed into a plastic column and swollen in DMF for 2 h, after
which the DMF was drained, 20% piperidine in DMF was added
to the resin, and the column was rotated for 10 min two times.
After the resin was washed with DMF (2×), water (3×), MeOH
(3×), DCM (5×), and DMF (2×), compound 3 (3 equiv), HOBt
1
purity ∼94%). MS (ESI) m/z: 266.71 [M+]. H NMR (DMSO-
d6): δ 7.59 (d, J ) 2.7, 1H), 7.43 (m, 2H), 7.23 (d, J ) 2.4, 1H),
4.79 (s, 2H), 3.84 (s, 3H), 1.44 (s, 9H).
Method A: Synthesis of 3-(2-tert-Butoxy-2-oxoethoxy)benzoic
Acid (3) and 3-(Carboxymethoxy)benzoic Acid (P10). Compound
2 (17.88 g, 67.2 mmol) was dissolved in THF/water (1:1, 50 mL),
and to this stirring solution was added LiOH (3.10 g, 73.9 mmol).
This solution was stirred at room temperature until TLC showed
the disappearance of 2, at which time the THF was removed by
rotary evaporation and the pH of the remaining aqueous solution
was adjusted to between 2 and 3 with 1 M HCl. This acidic solution
was extracted with ethyl acetate (3 × 50 mL), and the combined
organic layers were dried over MgSO4 and concentrated by rotary
evaporation. The crude product was subjected to silica gel column

ADCS Inhibitors from High-Throughput Screening
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 25 7423
(3.5 equiv), and DIC (3.5 equiv) were added. When the Kaiser test
was negative, a cleavage mixture of TFA, H2O, and TIS (95, 2.5,
2.5, v/v/v) were added to the resin and the column was rotated for
2 h, after which the filtrate was drained and collected. The TFA
was evaporated under a constant stream of nitrogen, and the crude
product was precipitated with ether and cooled in a refrigerator
overnight, after which the ether-crude mixture was centrifuged,
the ether decanted, and the remaining solid dissolved in acetonitrile
and the process was repeated. Once finished, the product was
decanted, dried by vacuum, and used without further purification
(143 mg, 73.3% yield). IR (neat, selected peaks) 3455.94, 3417.96,
1723.07, 1647.11 cm-1. MS (ESI) m/z: 196.10 [M + H+]. 1H NMR
(DMSO-d6): δ 13.04 (s, 1H), 7.97 (s, 1H), 7.33-7.48 (m, 4H),
7.05-7.07 (dd, J ) 2.4, 8.0, 1H), 4.72 (s, 2H). 13C NMR (DMSO-
d6): δ 170.1, 167.4, 157.7, 135.6, 129.3, 120.2, 117.5, 113.2, 64.5.
Anal. Calcd for C9H9NO4: C, 55.39; H, 4.65; N, 7.18; O, 32.79.
Found: C, 55.30; H, 4.65; N, 7.10; O, 32.95.
Synthesis of 2-(3-(5-Acetamido-6-amino-6-oxohexylcarbam-
oyl)phenoxy)acetic Acid (5). Rink amide MBHA (250 mg, 0.5
mmol/g loading capacity) resin was weighed into a plastic column
and swollen in DMF for 2 h, after which the DMF was drained,
20% piperidine in DMF was added to the resin, and the column
was rotated for 10 min two times. After washing the resin with
DMF (2×), water (3×), MeOH (3×), DCM (5×), and DMF (2×),
Fmoc-Lys(Dde)-OH (3 equiv), HOBt (3.5 equiv), and DIC (3.5
equiv) were added and the column was rotated until the Kaiser
test was negative. The resin was washed according to the procedure
outlined above, and the Fmoc group was removed. After the wash,
acetic anhydride (10 equiv) and DIEA (12 equiv) in DMF were
added to the resin. After the Kaiser test was negative, the resin
was washed and compound 3 (3 equiv), HOBt (3.5 equiv), and
DIC (3.5 equiv) were added. When the Kaiser test was negative, a
cleavage mixture of TFA, H2O, and TIS (95, 2.5, 2.5, v/v/v) was
added to the resin and the column was rotated for 2 h, after which
the filtrate was drained and collected. The TFA was evaporated
under a constant stream of nitrogen. The crude product was
precipitated with ether and cooled in a refrigerator overnight, after
which the ether-crude mixture was centrifuged. Once finished, the
product was purified by reversed-phase HPLC (38 mg, 84% yield).
IR (neat, selected peaks) 3460.69, 3323.01, 3308.77, 1723.07,
1613.87 cm-1. MS (ESI) m/z: 366.17 [M + H+]. 1H NMR (DMSO-
d6): δ 8.42 (t, J ) 5.6, 1H), 7.87 (d, J ) 8.0, 1H), 7.31-7.41 (m,
4H), 7.01-7.04 (d, J ) 8.4, 1H), 6.93 (s, 1H), 4.70 (s, 2H), 4.09-
4.14 (m, 1H), 3.16-3.21 (q, J ) 6.4, 2H), 1.80 (s, 3H), 1.57-1.64
(m, 2H), 1.43-1.49 (m, 2H), 1.25-1.31 (m, 2H). 13C NMR
(DMSO-d6): δ 174.0, 170.1, 169.2, 165.6, 157.7, 136.0, 129.4,
119.9, 117.3, 113.0, 64.5, 52.3, 40.1, 31.8, 28.9, 23.0, 22.6. Anal.
Calcd for C17H23N3O6: C, 55.88; H, 6.34; N, 11.50; O, 26.27.
Found: C, 55.81; H, 6.35; N, 11.61; O, 26.23.
Synthesis of Methyl tert-Butyl 2-(3-formylphenoxy)acetate (6).
m-Hydroxybenzaldehyde (10.0 g, 81.9 mmol) was dissolved in THF
(70 mL), and to this solution was added K2CO3 (33.9 g, 245.7
mmol). The mixture was placed under nitrogen, allowed to stir for
1 h, and then cooled in an ice bath, and tert-butyl 2-bromoacetate
(19.1 g, 98.3 mmol) was added via syringe. The solution was
removed from the ice bath and warmed to room temperature, at
which time TLC indicated the reaction had gone to completion.
The solvent was removed by rotary evaporation. The resulting solid
was dissolved in water (50 mL) and extracted with ethyl acetate (3
× 100 mL). The combined organic layers were dried with MgSO4
and concentrated by rotary evaporation. The crude product (2) was
used without further purification as a yellow solid (18.1 g, 93.5%
crude yield, purity ∼98%). IR (neat, selected peaks) 1739.49,
1698.50 cm-1. MS (ESI) m/z: 267.15 [M + H+]. 1H NMR (DMSO-
d6): δ 9.97 (s, 1H), 7.52-7.55 (m, 2H), 7.36-7.37 (d, J ) 7.8,
1H), 7.26-7.27 (m, 1H), 4.76 (s, 2H), 1.42 (s, 9H). 13C NMR
(DMSO-d6): δ 192.8, 167.6, 158.2, 137.5, 130.4, 123.1, 121.4,
113.6, 81.6, 65.1, 27.7. Anal. Calcd for C13H16O4: C, 66.09; H,
6.83; O, 27.09. Found: C, 66.02; H, 6.89; O, 27.09.
0.5 mmol/g loading capacity) resin was weighed into a plastic
column and swollen in DMF for 2 h, after which the DMF was
drained, 20% piperidine in DMF was added to the resin, and the
column was rotated for 10 min two times. After the resin was
washed with DMF (2×), H2O (3×), MeOH (3×), DCM (5×), and
DMF (2×), Fmoc-Lys(Dde)-OH (3 equiv), HOBt (3.5 equiv), and
DIC (3.5 equiv) were added and the column was rotated until the
Kaiser test was negative. The resin was washed according to the
procedure outlined above, and the Fmoc group was removed. After
washing, acetic anhydride (10 equiv) and DIEA (12 equiv) in DMF
were added to the resin. After the Kaiser test was negative, the
resin was washed, compound 4 (10 equiv) dissolved in TMOF was
added, and the column was rotated for 30 min. After this was added
NaCNBH3 (12 equiv) followed by acetic acid (100 µL). This
reaction mixture was rotated in the column for 1 h, after which the
Kaiser test was negative and the chloranil test was positive. A
cleavage mixture of TFA, H2O, and TIS (95, 2.5, 2.5, v/v/v) was
added to the resin and the column was rotated for 2 h, after which
the filtrate was drained and collected. The TFA was evaporated
under a constant stream of nitrogen and the crude product was
precipitated with ether and cooled in a refrigerator overnight, after
which the ether-crude mixture was centrifuged. Once finished, the
product was purified by reversed-phase HPLC (38 mg, 84% yield).
IR (neat, selected peaks) 3318.26, 3304.02, 3194,83, 2928.96,
2857.75, 1661.35 (two peaks coalescing), 1547.41 cm-1. MS (ESI)
1
m/z: 352.03 [M + H+]. H NMR (DMSO-d6): δ 8.73 (s, 1H),
7.93 (s, 2H), 6.95-7.07 (m, 4H), 4.70 (s, 2H), 4.10 (m, 1H), 3.35
(s, 2H), 2.88 (s, 1H), 2.08 (q, J ) 3.2, 2H), 1.84 (s, 3H), 1.60 (m,
2H), 1.48 (m, 2H), 1.23-1.28 (m, 2H). 13C NMR (DMSO-d6): δ
173.7, 170.0, 169.2, 157.8, 133.4, 129.9, 122.3, 116.0, 114.6, 64.3,
51.9, 49.8, 46.5, 31.4, 25.0, 22.6. Anal. Calcd for C17H25N3O5: C,
58.11; H, 7.17; N, 11.96; O, 22.77. Found: C, 58.19; H, 7.21; N,
12.02; O, 22.58.
Mass-Tag Encoded Peptide Library Synthesis. Synthesis of
the encoded peptide library was initiated by swelling Tentagel beads
(5 g, 90 µm capacity, 0.27 mmol/g) in a 60 mL plastic column in
water (50 mL) for 48 h. The water was drained, and the beads
were transferred to a glass reaction vessel (250 mL capacity) with
a sintered glass filter frit. A solution of DCM/ether (45:55, 100
mL) was added to the beads, and Alloc-OSu (0.1 equiv), dissolved
in DCM/ether (45:55, 50 mL), was added to the suspended resin
beads. DIEA (2 equiv) was added to the solution and the glass
container was then shaken vigorously for 30 min, after which the
solution was drained and the beads were transferred back to the 60
mL plastic column. The beads were swollen in DMF for 2 h, and
the cleavable linker was assembled according standard Fmoc peptide
chemistry with the following reagents: Fmoc-Met-OH (3 equiv),
Fmoc-Arg(Pmc)-OH (3 equiv), N-Fmoc-3-(4-bromophenyl)-â-
alanine, N-Fmoc-2,2′-ethylenedioxybis(ethylamine) monosuccina-
mide (3 equiv). All coupling reactions were performed in DMF
(50 mL) with the standard coupling reagents HOBt (3.5 equiv) and
DIC (3.5 equiv). All Fmoc deprotections were performed with 20%
piperidine in DMF (45 mL). After each reaction step, the beads
were washed with DMF, water, methanol, and DCM (5 × 35 mL
each). Following these reactions, the column containing the beads
was subjected to Alloc deprotection using tetrakis(tirphenylphos-
phine)palladium(0) (0.2 equiv) and phenylsilane (20 equiv) in DCM
for 30 min. After this time, the solution was drained, a fresh solution
of the same reagents was added, and the column was rotated for
another 30 min. The Fmoc group on the cleavable linker was
removed next with 20% piperidine in DMF (45 mL) for 20 min.
The resulting resin beads were then equally distributed into 48
plastic columns (2 mL) secured in a Teflon block. One of the Fmoc
protected amino acid diversity elements (3 equiv) and HOBt (3,5
equiv) in DMF (1.5 mL) and DIC (3.5 equiv) were added to each
column. The Teflon block was then placed on a shaker, and the
reactions were allowed to proceed for 3 h. Once all Kaiser tests
were negative for each column, the beads were pooled and the Fmoc
protecting group was removed. Semiorthogonally protected Fmoc-
Lys(Dde)-OH (3 equiv) was coupled using HOBt and DIC (3.5
equiv each). Once the coupling was complete, the Fmoc group was
Synthesis of 2-(3-((5-Acetamido-6-amino-6-oxohexylamino)-
methyl)phenoxy)acetic Acid (7). Rink amide MBHA (250 mg,

7424 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 25
Dixon et al.
removed and the beads were washed with DMF, water, methanol,
and DCM (5 × 35 mL each) and then dried completely under
vacuum for 1 day. When dry, the beads were reswollen in water
for 24 h. The beads were once again transferred to a glass reaction
vessel with a filter frit. An amount of 100 mL of the DCM/ether
solution (45:55) solution was added followed by Fmoc-OSu (0.7
equiv) dissolved in the DCM/ether solution (50 mL). DIEA (2
equiv) was added to the column and it was shaken vigorously for
30 min, after which the solution was drained and the beads were
transferred back to a 60 mL plastic column. (Boc)2O (0.6 equiv)
in DIEA (1.2 equiv) was added, and the reaction was allowed to
proceed until a Kaiser test of the beads was negative. The Fmoc
protecting group was then removed, and the beads were distributed
evenly to 48 plastic columns (2 mL) secured in a Teflon block.
One of the carboxylic acid diversity elements (10 equiv) and HOBt
(15 equiv) in DMF (1.5 mL) along with DIC (15 equiv) were added
to each column. The Teflon block was placed in a shaker, and the
reactions were allowed to proceed for 16 h, after which a Kaiser
test was performed on each column. All results were negative. The
beads were pooled and washed with DMF, water, methanol, DCM,
and DMF (5 × 35 mL). Next, the Dde group was removed using
2% hydrazine in DMF (40 mL. reaction repeated twice) and
compound 3 (5 equiv) was coupled to the beads using HOBt (7.5
equiv) in DMF (40 mL) and DIC (7.5 equiv). When the reaction
was complete, deprotection of all remaining protecting groups was
performed with 45 mL of 82.5% TFA, 5% phenol, 5% thioanisole,
5% H2O, and 2.5% TIS and rotation for 2 h. After the final
deprotection, the beads were neutralized with DIEA in DMF and
then washed thoroughly with DMF, water, methanol, and DCM (5
× 40 mL each), dried, and stored in a vacuum desiccator for future
use.
COPAS Sorting of the Peptide Library. The mass-tag encoded
peptide library was hydrated in water, transferred to the COPAS
instrument, diluted with a solution of PBS containing 0.2% Tween-
20, 0.05% NaN3, and 0.1% BSA (PBSTBN), then sorted with the
lens tuned to emit at 488 and 605 nm, and detected by a green and
red photomultiplier tube, respectively. An initial sort of beads (2300
beads) defined the main cluster of autofluorescence for the library
of beads. On the basis of this established base of autofluorescence,
a threshold was chosen that allowed the COPAS instrument to divert
beads with strong autofluorescence from the majority of the beads
with low autofluorescence. The beads were sorted at this fluores-
cence level several times in batches of 5000-25000 beads per batch.
Library Prescreening, Screening, and Isolation of Hits.
Tentagel beads (approximately 3000-4000) containing the target
peptide library were swollen in 1 mL of buffer A (50 mM bicine
[pH 8.5], 1 mM DTT, 50 mM KCl, 0.2% Tween-20, 0.1% gelatin)
for 60 min. The enzyme solution used for bead screening contained
0.31 nM Alexa Fluor 488 labeled ADCS in buffer A. The beads
were then incubated with 1 mL of abeled ADCS for 60 min at
room temperature. The incubation solutions were visualized under
a fluorescence microscope fitted with a long-band green filter. The
brightest beads were isolated manually with a pipet tip.
with 0.1% TFA on a MALDI-TOF target and allowed to evaporate
to dryness. This target was then inserted into the MALDI-TOF MS
instrument, and the samples were analyzed in reflectron mode.
Resynthesis and Characterization of Peptides. Larger-scale
resynthesis of the peptides identified as ADCS binders was
performed on 100 mg of Fmoc Rink MHBA amide resin (capacity,
0.65 mmol/g) contained in a 1 mL plastic column. The beads were
swollen in DMF for 60 min, drained, and then treated with 3 mL
of 25% piperidine in DMF twice, first for 5 min and next for 15
min. After Fmoc deprotection, the beads were washed with DMF
(5 × 0.8 mL) and the peptide sequence was constructed using the
appropriate amino (3 equiv) and carboxylic acids (10 equiv) along
with HOBt and DIC (13 equiv each). The target peptide was cleaved
from the resin using 0.9 mL of 95% TFA, 2.5% H2O, and 2.5%
TIS for 2 h. The cleavage solution was collected and concentrated
by evaporation, and 5-10 mL ether was added until the product
had completely precipitated. The filtrate was cooled to -80 °C for
14 h and centrifuged. The precipitated product was dried under
vacuum for 24 h and then purified by HPLC. Collected fractions
were lyophilized, and the resulting solids were subjected to HPLC
and ES-MS analysis (see Supporting Information for details).
Enzyme Preparation, Assay, and Labeling. The pabA, pabB,
and pabC genes encoding PabA, ADCS, and ADCL, respectively,
were amplified from Escherichia coli K12 chromosomal DNA. The
pabA gene was cloned into a pET3a (Novagen) vector using NdeI
and BamHI restriction sites. pabB and pabC genes were cloned
into a pET28a vector (Novagen) using NdeI and BamHI restriction
sites for expression as a 6×His-tagged fusion protein.
PabA was overexpressed by E. coli BL21 (DE3) Gold cells
grown in TB medium after induction by 0.5 mM IPTG at OD600
)
0.6. Cells were harvested by centrifugation and resuspended in lysis
buffer (100 mM TEA, pH 7.8, 30 mM mercaptoethanol, 1 mM
EDTA, 0.5 mg/mL lysozyme, 0.2 units/mL DNase I) prior to
disruption by sonication. Cell debris was pelleted by centrifugation
at 14 000 rpm, and a 15-40% ammonium sulfate fractionation was
performed. The 40% precipitate was dissolved in a minimum
volume of starting buffer (50 mM TEA, pH 7.8, 10 mM mercap-
toethanol, 5 mM MgCl2, 1 mM EDTA) and loaded onto a
Q-Sepharose fast flow column. Protein was eluted with a linear
gradient of 500 mL of 0-300 mM KCl in starting buffer. Fractions
were concentrated by ultrafiltration and dialyzed against 20 mM
KPi, pH 7.5, 50 mM KCl, and 1 mM DTT. Purified enzyme was
flash-frozen and stored at -80 °C. Protein concentration was
measured with the Bio-Rad DC assay, using IgG as standard. Yield
was ∼83 mg of purified protein per 21 g of cell paste.
PabB was overexpressed in E. coli BL21 (DE3) Gold cells grown
in TB medium with induction by 0.5 mM IPTG at OD600 ) 0.6.
Cells were harvested by centrifugation and resuspended in lysis
buffer (20 mM Na2HPO4, pH 7.4, 500 mM NaCl, 1 mM mercap-
toethanol, 10 mM imidazole, 0.5 mg/mL lysozyme, 0.2 units/mL
DNase I) prior to disruption by sonication. Cell debris was pelleted
by centrifugation at 14 000 rpm, and the supernatant was incubated
at room temperature for 30 min with chelating Sepharose fast flow
resin (Pharmacia) that was charged with Ni2+. Resin-bound protein
was eluted with a linear gradient of 10-300 mM imidazole in
starting buffer (20 mM Na2HPO4, pH 7.4, 500 mM NaCl, 1 mM
mercaptoethanol, 10 mM imidazole). Concentration, dialysis, and
storage conditions were as described for PabA. Yield was ∼185
mg of purified protein per 5 g of cell paste.
Cleavage of the Resin Beads for MALDI-TOF MS Analysis.
Beads isolated for library synthesis checks were washed with water
and then transferred to a 250 µL polypropylene microcentrifuge
tube in water with the help of a microscope. Beads isolated from
library screenings in the presence of ADCS were washed with water
and 8 M guanidine HCl in succession three times for 5 min in each
cycle. One final washing was done with water, and the individual
beads were transferred to a 250 µL polypropylene microcentrifuge
tube containing water with the help of a microscope. After transfer
to the polypropylene tube, MALDI-TOF MS analysis was per-
formed on each compound according to the following procedure.
Cyanogen bromide (20 mg/mL) in 70% formic acid was added (30
µL) to each bead-containing tube. The tube was sealed and left
overnight at room temperature. The tube was then lyophilized to
dryness. A water/acetonitrile (45:55, 70 µL) solution with 0.1%
TFA was added to each tube, and 0.7 µL of this solution was mixed
with an equal volume of a saturated solution of an R-cyano-4-
hydroxycinnamic acid solution in water/acetonitrile (45:55) solution
PabC preparation was the same as for PabB except that 20 µM
PLP was included in the lysis, starting, and dialysis buffers. Yield
was ∼460 mg of purified protein per 37 g of cell paste.
ADCS and the lysine-reactive green fluorescence dye Alexa Fluor
488 (Molecular Probes) were mixed in a 1:50 ratio in 0.2 M sodium
bicarbonate solution, pH 8.4. This mixture was incubated in the
dark at room temperature for 1.5 h in the presence of 1 mM DTT.
Unreacted dye molecules were separated from ADCS by using
Sephadex G25 spin columns. The extent of labeling was ∼1 dye
per ADCS.
All kinetic assays were performed on a Kontron Uvikon 930
UV-vis spectrophotometer at 25 °C. Reactions of mixtures

ADCS Inhibitors from High-Throughput Screening
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 25 7425
(11) Lam, K. S.; Lehman, A. L.; Song, A.; Doan, N.; Enstrom, A. M.;
Maxwell, J.; Liu, R. Synthesis and screening of “one-bead one-
compound” combinatorial peptide libraries. Methods Enzymol. 2003,
369, 298-322.
(12) Hwang, S.; Tamilarasu, N.; Ryan, K.; Huq, I.; Richter, S.; Still, W.
C.; Rana, T. M. Inhibition of gene expression in human cells through
small molecule-RNA interactions. Proc. Natl. Acad. Sci. U.S.A.
1999, 96 (23), 12997-13002.
(13) Alluri, P. G.; Reddy, M. M.; Bachhawat-Sikder, K.; Olivos, H. J.;
Kodadek, T. Isolation of protein ligands from large peptoid libraries.
J. Am. Chem. Soc. 2003, 125 (46), 13995-14004.
(14) Hu, Y.; Helm, J. S.; Chen, L.; Ginsberg, C.; Gross, B.; Kraybill, B.;
Tiyanont, K.; Fang, X.; Wu, T.; Walker, S. Identification of selective
inhibitors for the glycosyltransferase MurG via high-throughput
screening. Chem. Biol. 2004, 11 (5), 703-711.
(15) Youngquist, R. S.; Fuentes, G. R.; Lacey, M. P.; Keough, T. Matrix-
assisted laser desorption ionization for rapid determination of the
sequences of biologically active peptides isolated from support-bound
combinatorial peptide libraries. Rapid Commun. Mass Spectrom.
1994, 8 (1), 77-81.
(16) Liu, R.; Marik, J.; Lam, K. S. A novel peptide-based encoding system
for “one-bead one-compound” peptidomimetic and small molecule
combinatorial libraries. J. Am. Chem. Soc. 2002, 124 (26), 7678-
7680.
containing inhibitors were followed by the disappearance of NADH
in a coupled assay with excess ADCL and LDH. Each 500 µL of
mixture consisted of 100 mM bicine, pH 8.5, 20 mM L-glutamine,
5 mM MgCl2, 20 µM PLP, 7 µM ADCL, 200 µM NADH, 13 µM
chorismate, 2 units of LDH, 0.5 µM PabA, and 0.5 µM ADCS.
For Ki determinations, inhibitors were dissolved in water and
brought to pH 9 with 25 mM NaOH. Inhibitor concentrations were
varied from 200 to 1000 µM, while the chorismate concentration
was held constant. Reaction mixtures were incubated for 5 min at
25 °C prior to reaction initiation by an equimolar solution of PabA
and ADCS. Ki values were determined by fitting initial rate data
to eq 1.
The ordered binding kinetic data were collected by monitoring
the increase in absorbance at 290 nm, due to formation of PABA,
in a coupled assay with excess ADCL. Each 500 µL of mixture
contained 100 mM bicine, pH 8.5, 20 mM L-glutamine, 20 µM
PLP, 7 µM ADCL, 0.5 µM PabA, and 0.5 µM ADCS. Chorismate
concentration was varied from 50 to 650 µM, and Mg2+ concentra-
tion was varied from 15 to 450 µM. Reaction mixtures were
incubated for 5 min at 25 °C prior to reaction initiation by an
equimolar solution of PabA and ADCS. To determine Kchorismate
(17) Roberts, C. W.; Roberts, F.; Lyons, R. E.; Kirisits, M. J.; Mui, E. J.;
Finnerty, J.; Johnson, J. J.; Ferguson, D. J.; Coggins, J. R.; Krell,
T.; Coombs, G. H.; Milhous, W. K.; Kyle, D. E.; Tzipori, S.;
Barnwell, J.; Dame, J. B.; Carlton, J.; McLeod, R. The shikimate
pathway and its branches in apicomplexan parasites. J. Infect. Dis.
2002, 185 (Suppl. 1), S25-S36.
2+
and KMg , a saturation curve for each substrate was obtained.
Magnesium concentration was held constant at 15 µM, and
chorismate concentration varied from 50 to 825 µM. Chorismate
concentration was held constant at 500 µM, and magnesium
concentration varied from 5 to 750 µM.
(18) Knaggs, A. R. The biosynthesis of shikimate metabolites. Nat. Prod.
Rep. 2003, 20 (1), 119-136.
Acknowledgment. We thank the National Science Founda-
tion (Grant CHE-0313888) for financial support, as well as Scott
Eaton, Wren Yahraus, and Michael Descilletes for helpful
discussions.
(19) Smith, D. A.; Parish, T.; Stoker, N. G.; Bancroft, G. J. Characteriza-
tion of auxotrophic mutants of Mycobacterium tuberculosis and their
potential as vaccine candidates. Infect. Immun. 2001, 69 (2), 1142-
1150.
(20) McArthur, J. D.; West, N. P.; Cole, J. N.; Jungnitz, H.; Guzman, C.
A.; Chin, J.; Lehrbach, P. R.; Djordjevic, S. P.; Walker, M. J. An
aromatic amino acid auxotrophic mutant of Bordetella bronchiseptica
is attenuated and immunogenic in a mouse model of infection. FEMS
Microbiol. Lett. 2003, 221 (1), 7-16.
(21) Trias, J. The role of combichem in antibiotic discovery. Curr. Opin.
Microbiol. 2001, 4 (5), 520-525.
(22) Istvan, E. S.; Deisenhofer, J. Structural mechanism for statin inhibition
of HMG-CoA reductase. Science 2001, 292 (5519), 1160-1164.
(23) Maly, D. J.; Choong, I. C.; Ellman, J. A. Combinatorial target-guided
ligand assembly: identification of potent subtype-selective c-Src
inhibitors. Proc. Natl. Acad. Sci. U.S.A. 2000, 97 (6), 2419-2424.
(24) Szczepankiewicz, B. G.; Liu, G.; Hajduk, P. J.; Abad-Zapatero, C.;
Pei, Z.; Xin, Z.; Lubben, T. H.; Trevillyan, J. M.; Stashko, M. A.;
Ballaron, S. J.; Liang, H.; Huang, F.; Hutchins, C. W.; Fesik, S. W.;
Jirousek, M. R. Discovery of a potent, selective protein tyrosine
phosphatase 1B inhibitor using a linked-fragment strategy. J. Am.
Chem. Soc. 2003, 125 (14), 4087-4096.
(25) Song, A.; Zhang, J.; Lebrilla, C. B.; Lam, K. S. A novel and rapid
encoding method based on mass spectrometry for “one-bead-one-
compound” small molecule combinatorial libraries. J. Am. Chem.
Soc. 2003, 125 (20), 6180-6188.
(26) Dixon, S. M.; Li, P.; Liu, R.; Wolosker, H.; Lam, K. S.; Kurth, M.
J.; Toney, M. D. Slow-binding human serine racemase inhibitors from
high-throughput screening of combinatorial libraries. J. Med. Chem
2006, 49, 2388-2397.
(27) George, J.; Teear, M. L.; Norey, C. G.; Burns, D. D. Evaluation of
an imaging platform during the development of a FRET protease
assay. J. Biomol. Screening 2003, 8 (1), 72-80.
Supporting Information Available: Figures S1 (fluorescence
images of the on-bead library) and S2 (model of 5 bound to ADCS
1
with the PAYLOAD in the active site); H NMR and 13C NMR
spectra for molecules 1-3, P10, and 4-7; experimental procedures
and 1H NMR and 13C NMR data for P1 and P8; structures of amino
acid and carboxylic acid diversity elements used in the library
synthesis; MALDI-TOF MS results of encoding tags cleaved during
the synthesis of the library and selected identified compounds;
MALDI-TOF mass spectra for L1, L3, L5, and L9; ESI-MS data,
MS spectra, purity data, and HPLC traces for L1-L11, L3 linker,
and L5 linker. This material is available free of charge via the
Internet at http://pubs.acs.org.
References
(1) Easson, M. A. M.; Rees, D. C. Combinatorial chemistry: tools for
the medicinal chemist. Med. Chem. (Wiley) 2002, 359-381.
(2) Thompson, L. A.; Ellman, J. A. Synthesis and applications of small
molecule libraries. Chem. ReV. 1996, 96 (1), 555-600.
(3) Lee, A.; Breitenbucher, J. G. The impact of combinatorial chemistry
on drug discovery. Curr. Opin. Drug DiscoVery DeV. 2003, 6 (4),
494-508.
(4) Furka, A.; Sebestyen, F.; Asgedom, M.; Dibo, G. General method
for rapid synthesis of multicomponent peptide mixtures. Int. J. Pept.
Protein Res. 1991, 37 (6), 487-493.
(28) Mehrotra, M. M.; Heath, J. A.; Rose, J. W.; Smyth, M. S.; Seroogy,
J.; Volkots, D. L.; Ruhter, G.; Schotten, T.; Alaimo, L.; Park, G.;
Pandey, A.; Scarborough, R. M. Spirocyclic nonpeptide glycoprotein
IIb-IIIa antagonists. Part 3: synthesis and SAR of potent and specific
2,8-diazaspiro[4.5]decanes. Bioorg. Med. Chem. Lett. 2002, 12 (7),
1103-1107.
(29) DeVries, V. G.; Largis, E. E.; Miner, T. G.; Shepherd, R. G.;
Upeslacis, J. Potential antiatherosclerotic agents. 4. [(Functionalized-
alkyl)amino]benzoic acid analogues of cetaben. J. Med. Chem. 1983,
26 (10), 1411-1421.
(30) Lu, Z.; Tweig, R. J. Copper-catalyzed aryl amination in aqueous
media with 2-dimethylaminoethanol ligand. Tetrahedron Lett. 2005,
46 (17), 2997-3001.
(5) Lam, K. S.; Salmon, S. E.; Hersh, E. M.; Hruby, V. J.; Kazmierski,
W. M.; Knapp, R. J. A new type of synthetic peptide library for
identifying ligand-binding activity. Nature 1991, 354 (6348), 82-
84.
(6) Houghten, R. A.; Pinilla, C.; Blondelle, S. E.; Appel, J. R.; Dooley,
C. T.; Cuervo, J. H. Generation and use of synthetic peptide
combinatorial libraries for basic research and drug discovery. Nature
1991, 354 (6348), 84-86.
(7) Lam, K. S.; Lebl, M.; Krchnak, V. The “one-bead-one-compound”
combinatorial library method. Chem. ReV. 1997, 97 (2), 411-448.
(8) Dixon, S.; Wang, X.; Lam, K. S.; Kurth, M. J. Solid-phase synthesis
of quinoxaline, thiazine, and oxazine analogs through a benzyne
intermediate. Tetrahedron Lett. 2005, 46, 7443-7446.
(9) Wang, X.; Zhang, J.; Song, A.; Lebrilla, C. B.; Lam, K. S. Encoding
method for OBOC small molecule libraries using a biphasic approach
for ladder-synthesis of coding tags. J. Am. Chem. Soc. 2004, 126
(18), 5740-5749.
(31) Zhang, X.; Dillen, L.; Vanhoutte, K.; Van Dongen, W.; Esmans, E.;
Claeys, M. Characterization of unstable intermediates and oxidized
products formed during cyanogen bromide cleavage of peptides and
proteins by electrospray mass spectrometry. Anal. Chem. 1996, 68,
3422-3430.
(10) Lam, K. S.; Lebl, M. ImmunoMethods 1992, 1, 11-15.

7426 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 25
Dixon et al.
(32) Yue, W.; Koen, Y. M.; Williams, T. D.; Hanzlik, R. P. Use of isotopic
signatures for mass spectral detection of protein adduction by
chemically reactive metabolites of bromobenzene: Studies with
model proteins. Chem. Res. Toxicol. 2005, 18, 1748-1754.
P. A. Chorismate-utilizing enzymes isochorismate synthase, anthra-
nilate synthase, and p-aminobenzoate synthase: mechanistic insight
through inhibitor design. J. Am. Chem. Soc. 1995, 117 (8), 2128-
2140.
(33) Tornoe, C. W.; Sanderson, S. J.; Mottram, J. C.; Coombs, G. H.;
Meldal, M. Combinatorial library of peptidotriazoles: identification
of [1,2,3]-triazole inhibitors against a recombinant Leishmania
mexicana cysteine protease. J. Comb. Chem. 2004, 6 (3), 312-324.
(34) Carniel, E. The Yersinia high-pathogenicity island: an iron-uptake
island. Microbes. Infect. 2001, 3 (7), 561-569.
(38) Morollo, A. A.; Eck, M. J. Structure of the cooperative allosteric
anthranilate synthase from Salmonella typhimurium. Nat. Struct. Biol.
2001, 8 (3), 243-247.
(39) He, Z.; Stigers Lavoie, K. D.; Bartlett, P. A.; Toney, M. D.
Conservation of mechanism in three chorismate-utilizing enzymes.
J. Am. Chem. Soc. 2004, 126 (8), 2378-2385.
(40) Walsh, C. T.; Erion, M. D.; Walts, A. E.; Delany, J. J., 3rd; Berchtold,
G. A. Chorismate aminations: partial purification of Escherichia coli
PABA synthase and mechanistic comparison with anthranilate
synthase. Biochemistry 1987, 26 (15), 4734-4745.
(35) Budzikiewicz, H. Siderophores of the human pathogenic fluorescent
pseudomonads. Curr. Top. Med. Chem. 2001, 1 (1), 1-6.
(36) Payne, R. J.; Toscano, M. D.; Bulloch, E. M.; Abell, A. D.; Abell,
C. Design and synthesis of aromatic inhibitors of anthranilate
synthase. Org. Biomol. Chem. 2005, 3 (12), 2271-2281.
(37) Kozlowski, M. C.; Tom, N. J.; Seto, C. T.; Sefler, A. M.; Bartlett,
JM0609869