Development of α-keto-based inhibitors of cruzain, a cysteine protease implicated in Chagas disease

Article abstract of DOI:10.1016/j.bmc.2004.12.053

Trypanosoma cruzi, a protozoan parasite, is the causative agent of Chagas disease, a major cause of cardiovascular disease in many Latin American countries. There is an urgent need to develop an improved therapy due to the toxicity of existing drugs and emerging drug resistance. Cruzain, the primary cysteine protease of T. cruzi, is essential for the survival of the parasite in host cells and therefore is an important target for the development of inhibitors as potential therapeutics. A novel series of α-ketoamide-, α-ketoacid-, α-ketoester-, and aldehyde-based inhibitors of cruzain has been developed. The inhibitors were identified by screening protease targeted small molecule libraries and systematically optimizing the P1, P2, P3, and P1′ residues using specific structure-guided methods. A total of 20 compounds displayed picomolar potency in in vitro assays and three inhibitors representing different α-keto-based inhibitor scaffolds demonstrated anti-trypanosomal activity in cell culture. A 2.3 A crystallographic structure of cruzain bound with one of the α-ketoester analogs is also reported. The structure and kinetic assay data illustrate the covalent binding, reversible inhibition mechanism of the inhibitor. Information on the compounds reported here will be useful in the development of new lead compounds as potential therapeutic agents for the treatment of Chagas disease and as biological probes to study the role that cruzain plays in the pathology. This study also demonstrates the validity of structure-guided approaches to focused library design and lead compound optimization.

Full text of DOI:10.1016/j.bmc.2004.12.053

Bioorganic & Medicinal Chemistry 13 (2005) 2141–2156
Development of a-keto-based inhibitors of cruzain, a
cysteine protease implicated in Chagas disease
Youngchool Choe,a, Linda S. Brinen,b, Mark S. Price,a Juan C. Engel,c Meinolf Lange,d
Corinna Grisostomi,d Scott G. Weston,d Peter V. Pallai,d Hong Cheng,d
Larry W. Hardy,d David S. Hartsough,d Marsha McMakin,d
Robert F. Tilton,d Carmen M. Baldinod and Charles S. Craika,*
aDepartment of Pharmaceutical Chemistry, 600 16th Street, University of California at San Francisco, San Francisco,
CA 94143-2280, USA
bDepartment of Cellular and Molecular Pharmacology, University of California at San Francisco, 513 Parnassus Ave, San Francisco,
CA 94143-0511, USA
cDepartment of Pathology, School of Medicine, 513 Parnassus Avenue, University of California at San Francisco, San Francisco,
CA 94143-0102, USA
dDepartments of Biology, Chemistry, and Computational Design, ArQule, Inc., 19 Presidential Way, Woburn, MA 01801, USA
Received 17 November 2004; accepted 30 December 2004
Available online 2 February 2005
Abstract—Trypanosoma cruzi, a protozoan parasite, is the causative agent of Chagas disease, a major cause of cardiovascular dis-
ease in many Latin American countries. There is an urgent need to develop an improved therapy due to the toxicity of existing drugs
and emerging drug resistance. Cruzain, the primary cysteine protease of T. cruzi, is essential for the survival of the parasite in host
cells and therefore is an important target for the development of inhibitors as potential therapeutics. A novel series of a-ketoamide-,
a-ketoacid-, a-ketoester-, and aldehyde-based inhibitors of cruzain has been developed. The inhibitors were identified by screening
protease targeted small molecule libraries and systematically optimizing the P1, P2, P3, and P10 residues using specific structure-
guided methods. A total of 20 compounds displayed picomolar potency in in vitro assays and three inhibitors representing different
˚
a-keto-based inhibitor scaffolds demonstrated anti-trypanosomal activity in cell culture. A 2.3 A crystallographic structure of cruz-
ain bound with one of the a-ketoester analogs is also reported. The structure and kinetic assay data illustrate the covalent binding,
reversible inhibition mechanism of the inhibitor. Information on the compounds reported here will be useful in the development of
new lead compounds as potential therapeutic agents for the treatment of Chagas disease and as biological probes to study the role
that cruzain plays in the pathology. This study also demonstrates the validity of structure-guided approaches to focused library
design and lead compound optimization.
ꢀ 2005 Elsevier Ltd. All rights reserved.
1. Introduction
thought that another 100 million people are at risk of
infection.1 Due to the toxicity of existing drugs com-
bined with the advent of drug resistance, treatment for
this debilitating, often chronic illness is woefully inade-
quate.1 Trypanosoma cruzi, a parasitic protozoan, is
the causative agent of the disease. Its infectious trypo-
mastigote form is transmitted to human hosts through
the triatomine Ôkissing bugÕ vectors. Trypomastigotes
transform into the intracellular amastigotes after invad-
ing cardiac muscle cells through the bloodstream.
Amastigotes complete the infectious life cycle by multi-
plying in the cell, transforming back to infectious
trypomastigotes and rupturing the cell to release an
Chagas disease or American trypanosomiasis, is one of
the most threatening endemics in Central and South
America. Approximately 16–18 million people are in-
fected, resulting in adverse health events such as heart
failure and more than 50,000 deaths each year.1 It is
Keywords: Chagas disease; Cruzain; Cysteine protease inhibitors;
Structure-based library design.
*
Corresponding author. Tel.: +1 415 476 8146; fax: +1 415 502
8298; e-mail: craik@cgl.ucsf.edu
These authors contributed equally to this work.
0968-0896/$ - see front matter ꢀ 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2004.12.053

2142
Y. Choe et al. / Bioorg. Med. Chem. 13 (2005) 2141–2156
amplified quantity of the infectious form back into the
bloodstream.2
2. Results
2.1. Identification of a-ketoamide-based inhibitors of
cruzain
Cruzain, the primary cysteine protease of T. cruzi, is
expressed throughout the life cycle and is known to
be essential for the survival of the parasite within host
cells.3 In addition to its obvious role in parasite nutri-
tion, cruzain has been proposed to be involved in the
penetration of the parasite into the host cell and in
digestion of immunoglobulins as a defense mechanism.
Some inhibitors of cruzain have been shown to success-
fully treat animal models of Chagas disease by blocking
the parasitic life cycle.4 Thus, cruzain presents itself as
an interesting target for the development of potential
therapeutics for the treatment of the disease. Cruzain
was initially discovered from cell-free extracts,5 and sub-
sequently heterologously expressed in Escherichia coli.6
The X-ray crystallographic structures of a recombinant
truncated form of the enzyme in complex with several
small molecule inhibitors have been determined.7 The
overall folding pattern and the arrangement of the active
site residues are similar to those in papain. As in most
papain family proteases, the S2 subsite of the enzyme
is known to be the predominating specificity determi-
nant. Cruzain contains a unique C-terminal extension
of 130 amino acids, which is retained in native forms.
The enzyme is inhibited by organomercurial reagents,
E-64, Tos-Lys-CH2Cl, leupeptin, a number of peptidyl
chloromethanes, and peptidyl fluoromethane deriva-
tives, vinyl sulfones, thio semicarbazones, cystatins,
stefins, and kininogens.2,4 Covalent inhibitors such as
peptidyl epoxy ketones,8 aldehydes,9 and vinyl sulfones,9
have also been reported to inhibit cruzain. However,
published information on cruzain inhibitory activity
from a-keto based compounds has been very
limited.10,11
The in vitro screening of protease targeted libraries for
inhibitory activity against cruzain identified novel peptid-
yl a-ketoamide-based compounds as covalent inhibitors
of the enzyme (Fig. 1). Further exploration of this struc-
tural motif led to the synthesis of a small set of novel
a-ketoester-, a-ketoacid, and a-ketoamide-based com-
pounds (Table 1). A modified Dakin–West reaction
was used for the preparation of the compounds.14 It
should be noted that this synthetic approach produces
a mixture of diastereomers at the P1 a-carbon atom.
The vinyl-based a-keto ester series was obtained by cou-
pling of the peptide aldehyde (benzyloxycarbonyl (Cbz)-
Phe-Ala-CHO) with triphenylphosphorane (Scheme 1).
An examination of the SAR trends of this set revealed
the need for the presence of an amide hydrogen atom.
Kinetic analysis of these compounds demonstrated a
competitive, reversible inhibition of cruzain by the mem-
bers of this set (Fig. 2). In order to more effectively ex-
plore the P10 region of the enzyme, the a-ketoamides
were selected as the chemical scaffold for further inhibi-
tor optimization efforts. However, aldehyde-based com-
pounds, which are easier to prepare than the analogous
a-ketoamides were often used as the initial probes to
establish inhibitory trends against the enzyme.
2.2. Optimization of the a-ketoamide scaffold at the P1
subunit
The optimization began with an analysis of the P1 sub-
unit using peptidyl aldehydes, which were synthesized
following the Reetz procedure,15 as surrogates for the
a-ketoamide series. Various natural and unnatural ami-
no acids including those lacking a chiral center were sys-
tematically incorporated at the P1 position. The SAR
trends observed for this set clearly point to the need
for a chiral center of the S-configuration (or L-amino
acids) at the P1 and a preference for b-branched alkyl
substituents at this position (Table 2). The structural
reason underlying the observed SAR trends is not clear,
Due to its essential function in the parasiteÕs life cycle
and therefore in the propagation of Chagas disease,
cruzain was elected as a target to develop inhibitors
through in vitro screening of a series of protease specific
small molecule libraries and structure-guided optimiza-
tion. The libraries were designed to initially focus
screening on known pharmacologically active and target
class specific small molecules.10 Additional efficiencies
could also be realized from this approach due to the
comprehensive understanding of the automated process
chemistry.12,13 The in vitro screening of these libraries
for cruzain inhibitory activity led to the identification
of novel a-ketoamide-based cruzain inhibitors. The
screening also facilitated the exploration of closely re-
lated a-ketoamide-, a-ketoacid-, a-ketoester-, and alde-
hyde-based analogs. Identification of an inhibitor in
the arrayed analog libraries provides rapid access to a
database of structure–activity relationships (SAR). This
information is then used to guide the systematic optimi-
zation of each inhibitor scaffold. Since the papain family
cysteine proteases including cruzain (clan CA) share a
common three-dimensional structure and very similar
biochemical characteristics, this approach and the inhib-
itor scaffolds developed for cruzain can be readily
applied to other physiologically important clan CA
proteases.
P2
O
O
H
H
N
N
O
N
H
O
CH
O
3
P1'
P1
P3
AQ665183
Cbz-Phe-Ala-CONH-Bz
Figure 1. Chemical structure of AQ665183, an a-ketoamide identified
from the initial screening. The P3, P2, P1, and P10 portions of the
compound are denoted.

Y. Choe et al. / Bioorg. Med. Chem. 13 (2005) 2141–2156
2143
Table 1. Inhibition of cruzain by the initial set of a-ketoester, a-ketoamide, a-ketoacid compounds
Compound
Cruzain inhibition
IC50 (nM)
Ki (nM)
O
O
AQ616475
AQ665184
1000
ND
H
N
O
O
N
H
O
CH 3 O
O
O
39
13.8
H
H
N
N
O
N
H
O
CH3 O
O
O
AQ616476
AQ665183
AQ713492
AQ581332
7.7
0.83
H
N
OH
O
N
H
O
CH3 O
O
O
5.5
3.4
H
H
N
N
O
N
H
O
CH3
O
O
O
O
>5000
ND
H
N
N
O
N
H
O
CH3
O
O
4.7
1102.9
H
N
O
O
N
H
O
CH3
O
ND = not determined.
O
O
O
PPh3
O
Br
P
Na2CO3
O
O
O
O
H
N
O
H
N
O
O
H
O
H
O
N
O
N
H
O
O
Scheme 1. Synthesis of the vinyl-based a-keto ester.
as there is no clearly defined S1 pocket in published
cruzain crystallographic structures.7,16 Larger R groups
such as the phenylethane side chain of homophenyl-
alanine have also been used at this position in other
published series,9 but structural studies revealed no con-
structive interaction of these groups with the cruzain S1

2144
Y. Choe et al. / Bioorg. Med. Chem. 13 (2005) 2141–2156
Table 2. Optimization of the P1 component in aldehyde series
0.035
0.03
Compound
IC50 (nM)
0.025
0.02
1/Vi
O
O
0.015
0.01
H
N
AQ573188
7.2
O
N
H
H
O
0.005
0
P1
-0.5
0
0.5
1
1.5
-0.005
1/[S] (1/uM)
AQ665184 Ki= 13.8 4.6 nM
O
O
AQ625173
AQ625172
175
H
N
O
N
H
H
0.05
0.04
0.03
0.02
0.01
0
O
O
5000
1/Vi
-0.5
H
N
H
O
N
H
O
O
0
0.5
1
1.5
O
O
O
O
AQ722431
AQ718038
854
H
N
-0.01
1/[S] (1/uM)
O
O
N
H
H
H
O
O
AQ616476 Ki= 0.8 0.11 nM
Figure 2. Lineweaver–Burk plots for the selected compounds.
0.466
H
N
N
H
subsite.7 Based upon the trends observed in the peptidyl
aldehyde series, L-valine was selected as the preferred P1
residue. The corresponding a-ketoamide compound
(AQ797526) was synthesized using WassermannÕs proce-
dure17 and found to be a subnanomolar inhibitor of
cruzain while compound AQ752489 demonstrated the
a-keto moiety is required for the activity (Table 3).
O
O
AQ729376
50
H
N
O
N
H
H
O
2.3. Optimization at the P2 subunit
The optimization of the P2 subunit was performed using
the a-ketoamide series. A new synthetic scheme based
on WassermannÕs chemistry was employed in order to
prepare the Cbz-Phe-Val-a-ketoesters with a desired
chirality (S/L) at the P1 carbon (Scheme 2). All the
compounds from this series have L-valine at the P1 posi-
tion, a Cbz group at the P3 position, and an N-benzyl
group at the P10 position as AQ797526 (Table 4).
The SAR trends observed for this set correlate with
the reported large and predominantly hydrophobic
nature of the S2 pocket in cruzain.7,16 These results
are also consistent with published data obtained using
other series of cruzain inhibitors such as conformation-
ally constrained c-lactams containing an electrophilic
aldehyde, vinyl sulfones, or fluoromethyl ketones.9,16
L-Phenylalanine was identified as the favored residue
for the P2 subunit, followed by smaller hydrophobic
amino acids such as L-leucine, L-valine, and L-methio-
nine. Ultimately, L-phenylalanine was selected as the
preferred P2 residue.
O
O
H
AQ718041
0.49
N
O
N
H
H
O
2.4. Optimization at the P3 and the P10 subunits
Evaluation of the P3 subunit quickly indicated that the
removal of the Cbz-group caused dramatic loss of activ-
ity while the conversion to an acetyl group provided a
similar but less pronounced impact. No further optimi-
zation at the P3 position was performed and Cbz was
selected as the preferred P3 group for the remainder of
the study. To optimize the P10 side chain of the a-keto-
benzamide series, the P1, P2, and P3 positions were
held constant with L-valine, L-phenylalanine, and Cbz,
respectively. Given the hundreds of commercially avail-
able benzylamines, a structure-guided approach was

Y. Choe et al. / Bioorg. Med. Chem. 13 (2005) 2141–2156
2145
Table 3. Inhibition data for the amide-based compounds with L-valine at the P1 position
Compound
Cruzain inhibition
IC50 (nM)
Ki (nM)
O
O
H
H
N
N
AQ797526
0.797
0.878
O
N
H
O
O
P1
O
OH
H
H
N
N
AQ752489
5000
ND
O
N
H
O
O
P1
ND = not determined.
CN
EDCI
O
O
O
H
DMAP
PPh3
O3, -78 oC
CH2Cl2/CH3OH
H
O
N
CN
PPh3
O
N
OH
O
O
Cbz-Val-OH
Cbz-Val-C(CN)PPh3
NH2
CH3OH, RT
O
BCl3
H
H
H
N
O
N
N
O
OCH3
O
O
O
O
Cbz-Val-CONHBn
Cbz-Val-COOMe
O
H
H
O
H
HBTU, Z-AA
N
N
Cbz AA
H2N
N
O
O
Val-CONHBn
Scheme 2. Synthesis of the a-ketobenzamide library with variation at the P2 position.
adopted to select a smaller, focused set of these reagents.
A structural model of a covalently bound inhibitor
(AQ665183) to cruzain was built and used to bias
the reagent selection process using molecular docking
methods (Fig. 3).7 Reagent rankings obtained by
this molecular docking approach suggested that
p-substituted phenyl rings should be favored over
m- and o-substituted analogs (Table 5). Docking studies
also suggested that a-substituted benzylamines would
not fit into the active site as well as the unsubstituted ana-
logs. A Cbz-Phe-Val-a-ketobenzamide library with vari-
ation at the P10 position was prepared (Scheme 3). The
prediction based on structural modeling was consistent
with the experimentally determined activity trends
(Table 6). Both the observed trends and the modeling
studies for the a-ketoamide series suggest that the primary
structural justification for the increased affinity of
compounds containing a benzyl amine group is the aro-
matic/hydrophobic interaction with the indole side-chain
of the nearby Trp177 residue (Fig. 3). This is consistent
with observed SAR trends and structural studies of other
cysteine protease inhibitor series such as vinyl sulfones.18,19
2.5. Cell culture test for anti-trypanosomal activity
The a-ketoacid (AQ616476), a-ketoamide (AQ665183),
and vinyl a-ketoester (AQ581332), all shown to be cova-
lent binding, reversible inhibitors from the previously
described series, were tested in cell culture test using T.
cruzi-infected cells. At 30 lM, all three inhibitors dem-
onstrated trypanostatic activity (Fig. 4). The difference
in concentrations required to show activity in in vitro
(picomolar to nanomolar) and those needed for activity
in cell culture (micromolar) can be explained, at least in

2146
Y. Choe et al. / Bioorg. Med. Chem. 13 (2005) 2141–2156
Table 4. Optimization of the P2 position in a-ketoamide series
P2
O
O
O
H
N
H
N
N
H
O
O
P1
P2
Compound
Cruzain IC50 (nM)
P2
Compound
Cruzain IC50 (nM)
OH
N
H
ML-cruzain-1
10,000
ML-cruzain-10
>1000
O
N
H
O
ML-cruzain-2
ML-cruzain-3
>1000
>1000
ML-cruzain-13
ML-cruzain-16
10
N
H
N
H
O
O
O
N
OH
525
N
H
N
O
O
H
O
O
ML-cruzain-5
ML-cruzain-6
13
ML-cruzain-18
ML-cruzain-19
52
N
H
N
H
O
OH
3
2
O
O
N
H
N
H
O
ML-cruzain-7
ML-cruzain-8
7
ML-cruzain-20
ML-cruzain-21
37
13
N
H
N
H
O
O
S
N
11
N
H
N
H
O
O
N
N
ML-cruzain-9
237
N
H
O
part, by permeability issues arising from the peptidic
nature of these compounds. AQ665183 (a-ketoamide)
with a benzyl amide group at the P10 position displayed
better activity (66.7% longer survival than control) than
AQ616476 (a-ketoacid) and AQ581332 (vinyl a-
ketoester).

Y. Choe et al. / Bioorg. Med. Chem. 13 (2005) 2141–2156
2147
Table 5. Structure-guided p10 substituent selection based on DOCK
prediction
O
O
H
N
H
N
R
O
N
H
O
CH3 O
R groups predicted Parent Benzylamine R groups predicted
to have better
affinity
(AQ665183)
to have worse
affinity
O
S
N
Figure 3. Comparison of the model structure of the AQ665183–
cruzain complex with the crystallographic Cbz-Tyr-Ala-fluorometh-
ylketone (FMK)–cruzain complex from the 1AIM structure. The
solvent-accessible surface of cruzain (1AIM) is shown in green, the
crystallographic structure of the FMK-based cruzain inhibitor from
1AIM is shown in orange, the molecular surface associated with the
catalytic Cys25 residue is shown in cyan, the molecular surface
associated with the nearby Trp177 residue is shown in violet, and the
modeled a-ketoamide (Cbz-Phe-Ala-N-benzyl-a-ketoamide) is shown
in yellow.
O
N
N
O
O
F
F
F
CN
O
O
EDCI
O3, -78 oC
H
O
O
H
N
N
CN
DMAP
PPh3
O
N
CH2Cl2/CH3OH
44%
O
N
OH
H
O
PPh3
41%
H
O
Cbz-Phe-Val-C(CN)PPh3
Cbz-Phe-Val-OH
O
O
H
O
O
H
H
N
N
N
OCH3
H2N-R
CH3OH, RT
O
N
R
O
N
H
H
O
O
O
Cbz-Phe-Val-COOMe
O
Cbz-Phe-Val-CONHR
Scheme 3. Synthetic approach to the L-valine-derived a-ketobenzamide library with variation at the P10 position.
˚
lution of 2.2 A and a complete data set was collected
2.6. Slow-binding kinetic analysis
˚
to 2.3 A. Structure solution was accomplished via
molecular replacement with subsequent refinement to
The kinetic studies with the a-ketoester, AQ581332 indi-
cated that the inhibitor is a slow-binding inhibitor of
cruzain (Fig. 5). Slow-binding behavior of the inhibitor
was evident at 6.25 lM concentration of the inhibitor
(kobs = 10.6 · 10ꢀ3 sꢀ1), 5 nM concentration of cruzain,
and 50 lM concentration of substrate, Cbz-Phe-Arg-
AMC. These conditions are more than 1000-fold excess
of inhibitor over enzyme with saturating concentration
of substrate, both requirements for monitoring the
slow-binding behavior of the inhibitor.
˚
2.3 A. Crystallographic and data collection parameters
are summarized in Table 7. Cruzain is inhibited by
AQ581332 when a covalent attachment is formed
between the two molecules through a Michael addition,
involving nucleophilic attack by the Sc of the active
site cysteine residue, Cys25, at the vinyl carbon indicated
in Figure 6. Cbz is the P3 moiety of AQ581332, occupy-
ing the S3 subsite of the enzyme. The phenyl ring of the
Cbz group makes weak hydrophobic contact with
˚
Leu67, 3.7 A away while phenylalanine in the P2 subunit
2.7. Crystallographic data
of the inhibitor takes advantage of cruzainÕs hydropho-
bic S2 pocket (Fig. 7). Constructive hydrophobic
interactions are also made with Leu67, Ala133 and
Leu157. Additionally, Glu205 at the base of the S2
pocket is bent out of the cleft and does not interfere with
the aromatic ring. The ethoxy tail of the inhibitor that
was expected to interact with the S10 region of cruzain18
˚
A 2.3 A crystallographic structure for one member of
this series, AQ581332, a vinyl a-ketoester complexed
to cruzain was determined. The inhibitor co-crystallized
in the P21 space group with one molecule in the asym-
metric unit. Diffraction was observed to a maximum reso-

2148
Y. Choe et al. / Bioorg. Med. Chem. 13 (2005) 2141–2156
Table 6. Inhibition of cruzain by members of P10 optimization library
O
O
H
H
N
N
R
O
N
H
O
O
P10 (R)
Compound
AQ797526
Cruzain IC50 (nM)
0.878
P10 (R)
Compound
AQ903086
Cruzain IC50 (nM)
0.949
OCH3
OCH3
O
O
S
AQ903084
AQ840158
AQ840159
0.644
0.667
0.673
AQ840162
AQ840163
AQ903087
0.955
0.975
1.115
NH 2
OCH3
OCF3
OCH3
O
O
S
AQ840160
AQ840161
0.688
0.720
AQ840164
AQ903085
1.273
1.507
OCH3
OCH3
CF3
AQ903082
AQ903091
AQ903079
0.776
0.794
0.813
A840165
1.734
2.405
3.987
AQ903083
AQ903077
CF3
OCH 3
AQ903090
AQ903081
AQ903080
AQ903078
0.817
0.846
0.860
0.911
AQ903088
AQ840166
AQ903089
4.529
7.474
F
10.509
F
o
o
was not visible in the crystal structure of the complex,
which may suggest that the ethoxy group is exposed
freely to solution rather than bound to the enzyme at
this site.

Y. Choe et al. / Bioorg. Med. Chem. 13 (2005) 2141–2156
2149
Table 7. Crystallographic parameters, data collection, and refinement
statistics
Crystallographic parameters
Unit cell dimensions
˚
45.720
50.675
45.576
116.74
P21
a, b, c (A) and (b)ꢁ
Space group
Molecules per A/U
1
Data collection statistics
˚
Resolution limit (A)
2.3
25,466
Figure 4. Cell culture test result for selected compounds.
Number of unique reflections
(F > 0rjFj)
Rmerge (%)a
9.2
Redundancy of reflections
Overall
Highest resolution shell
5.5
2.3
hI/r(I)i
Overall
Highest resolution shell
8.8
3.6
Completeness
Overall (%)
93.3
88.2
Highest resolution shell (%)
Resolution range; highest shell (A)
˚
2.37–2.30
Refinement statistics
˚
Resolution range (A)
40.8–2.3
13.3
21.8
b
Rcryst (%)
c
Rfree (%)
Figure 5. Progress curve of the hydrolysis of Cbz-Phe-Arg-AMC
(50 lM) by cruzain displays slow-binding inhibition of cruzain by a a-
ketoamide AQ581332. r: No inhibitor, h: 3.125 lM, n: 6.25 lM, s:
12.5 lM, · :25 lM.
Rmsd
˚
Bond lengths (A)
0.02
1.88
Bond angles (ꢁ)
Water molecules
105
2
˚
Average B factor (A)
Several interactions exist between the inhibitor and the
main chain of cruzain. At distances of 3.4, 3.4, and
Complex
Inhibitor
14.12
22.12
˚
3.7 A, the carbonyl oxygen of Gly66 contacts O1, O3,
and N1 of the inhibitor, respectively. Contacts of 3.3
ꢀ
ꢁ
ꢂ
P P
P
a Rmerge
¼
j ðIhi ꢀ hIhiÞ
j
hI, where Ih is the mean structure
h
i
factor intensity of i observations of symmetry-related reflections with
Bragg intensity index h.
b Rcryst
˚
and 3.1 A are also made from O1 and O3 to the peptide
ꢀ
P
ꢁ
P P
P
P
nitrogen of Gly66. N2 of the inhibitor generates a
hydrogen bond to the carbonyl oxygen of Asp158 at a
¼
jj ðF obs j ꢀ j F calcd jjÞ=j F obs j , where Fobs and
h
i
Fcalcd are the observed and the calculated structure factor
˚
distance of 3.1 A while an ordered water molecule situ-
magnitudes.
¼
.
c Rfree
jj F obsðhklÞ j ꢀk j F
j F obsðhklÞ j, where the s
ðhklÞs
˚
ated near the active site cleft forms a 3.3 A hydrogen
ðhklÞ
set of reflections is omitted from the refinement process; 10% of the
ðhklÞs
bond to O2 of the inhibitor. The remaining constellation
of electrostatic interactions is made between AQ581332
and side chains of amino acid residues of cruzain, as
illustrated in Figure 8. Additional interactions include
data were included in the s set.
are proving to be promising lead compounds for drug
development. The approach has been greatly augmented
by the advance of structural biology, high-speed parallel
synthesis, and computational chemistry. In the case of
HIV-1 protease inhibitor development, which is consid-
ered a successful example, structure-based inhibitor de-
sign has focused on either transition state analogs,
substrate analogs, or de novo designed compounds.
Transition-state analog inhibitors are generally peptide
derivatives containing an electrophilic carbonyl group
in place of the scissile amide bond of the substrate.20
The design of transition state analogue inhibitors of
HIV-1 protease has also centered on attempts to replace
the scissile amide bond with nonhydrolyzable peptide
isosteres such as reduced amide, hydroxyethylamines,
and statin derivatives.21
˚
a weak 3.8 A hydrogen bond between O4 and Nd1 of
His159, a second hydrogen bond between O4 and Ne2
˚
of Gln19 at a distance of 2.9 A. Final coordinates have
been deposited with the Protein Data Bank (PDB) and
are accessible under the ID number 1U9Q.
3. Discussion
Herein, we have described the synthesis, identification,
optimization, and crystallographically determined bind-
ing mode of a new series of a-keto-based cruzain inhibi-
tors. As various essential roles are revealed that
proteases play in both normal and pathological states,
specific protease inhibitors are being developed and

2150
Y. Choe et al. / Bioorg. Med. Chem. 13 (2005) 2141–2156
apparent that reactive ketone-containing compounds
P2
could serve as inhibitors for a broad range of enzymes.
Peptidyl a-ketoester inhibitors of elastase and peptidyl
a-ketoamide inhibitors of calpains and other cysteine
proteases such as cathepsin B and papain have been
developed.25 More recently, a-ketoesters and a-dike-
tones have been shown to be very effective inhibitors
of a-chymotrypsin and calpain.24 A substrate-derived
ketone inhibitor was also reported to strongly inhibit
angiotensin converting enzyme (ACE), a metallopro-
tease, while efficient inhibitors of an aspartic protease,
renin, were prepared by replacing the scissile amide
bond of its substrate with a keto-methylene isostere.26,27
Cruzain
O4
O2
H
O6
N2
O1 N1
H
O3
O5
P1
P3
Figure 6. AQ581332. Point of attachment of cruzain is indicated.
Atoms within brackets were not visible in X-ray structure.
The screening of protease targeted libraries coupled with
the exploration of SAR and the structure-guided optimi-
zation of the lead compound series led to a novel series
of potent small-molecular peptidyl inhibitors of cruzain.
Initially a-ketoamide compounds were identified and
subsequently various a-keto analogs such as a-ketoest-
ers, a-ketoacids, and aldehydes with IC50 values in the
low nanomolar range were synthesized. Through the
optimization effort at each of the P1, P2, P3, and P10
subunits, L-valine, L-phenylalanine, Cbz-group, and p-
substituted benzylamines were determined to be the
optimal substituents respectively. A total of 20 com-
pounds displayed picomolar potency in in vitro assays.
Three inhibitors that were tested in cell culture demon-
strated anti-trypanosomal activity in cell culture test.
It is worth noting that a major determinant of potency
is the phenylalanine at P2. This is in agreement with
the well known preference for phenylalanine in clan
CA cysteine proteases. The S2 pocket of cruzain is lined
with hydrophobic residues such as Leu67, Ala133, and
Leu157 that readily accommodate the aromatic side-
chain of the substrates and inhibitors.7,16
Figure 7. Surface representation of the cruzain active site region.
Hydrophobic residues are indicated in green. (Note that the sides of
the S2 pocket are lined with hydrophobic patches.) This figure was
prepared with GRASP.
Kinetics studies and the crystallographic structure of the
a-ketoester, AQ581332 bound to cruzain clearly demon-
strate the inhibition mechanism of this series of inhibi-
Initially, the peptidyl a-ketoester functional scaffold was
reported as a potent transition state analog inhibitor of
serine proteases. The design was based on the crystal
structure of trypsin complexed with an a-ketoacid (4-
amidinophenyl pyruvate, APPA).22 The active site serine
adds to the ketone carbonyl group of APPA to give a
tetrahedral adduct with the oxyanion stabilized by inter-
acting with trypsinÕs oxyanion hole. In addition,
cyclotheonamide A and B, the natural thrombin inhibi-
tors isolated from a marine sponge, both contain the a-
keto amide functional group.20 Since peptide aldehydes,
with peptide sequences based on the substrate specificity
of calpain, were found to be potent reversible inhibitors
of calpain I and II with a Ki as low as 36 nM,23 various
transition state inhibitors for cysteine proteases includ-
ing peptide aldehydes, peptide fluoroalkyl ketones,
dicarbonyl derivatives, peptide a-ketoesters, a-keto-
amides, and a-ketoacids have been developed. In each
case, the inhibitors are known to inhibit cysteine prote-
ases and serine proteases by the reversible formation of
a covalent tetrahedral hemithioketal or hemiketal. They
result from the attack of the sulfhydryl or hydroxy
group of the active site cysteine residue or serine residue
at an electrophile center of the inhibitor.24 It became
tors. The inhibitor forms
a tetrahedral enzyme–
inhibitor complex that results from the nucleophilic
addition of the active site cysteine sulfhydryl to the elec-
trophilic vinyl carbon of the inhibitor. This mimics the
catalytic mechanism of cysteine proteases in which
the cysteine sulfhydryl attacks the amide carbonyl of
the scissile bond. Also, the kinetic studies indicate
AQ581332 is a slow-binding inhibitor. Slow-binding
inhibitors do not instantly show a steady-state initial
velocity. Instead, these inhibitors exhibit a slow onset
of inhibition with time until a steady-state initial velocity
is attained.28 The slow-binding phenomenon could be
due to a conformational change of the enzyme that is re-
quired for the optimum binding of the inhibitor and has
previously been observed with other transition-state
inhibitors of proteases. Examples include the binding
of peptide ketoacids and ketoesters to papain and
cathepsin B.25
Three inhibitors representing the different inhibitor
chemotypes were tested in cell culture using macrophage
infected with T. cruzi. Each of the inhibitors that were
tested exhibited a dose dependent inhibition of trypano-

Y. Choe et al. / Bioorg. Med. Chem. 13 (2005) 2141–2156
2151
Figure 8. Cruzain active site with an inhibitor bound. Cruzain is shown in ribbon representation and the inhibitor is shown in ball-and-stick
representation. Amino acid side-chains involved in binding are highlighted in magenta. Bound water molecule is displayed in red. Hydrogen bonds
are indicated by dashed lines. This figure was prepared with the programs MOLSCRIPT and RASTER 3D.
somal activity. It appears the intracellular cycle of the
parasite was slowed by the inhibitors and, although the
exact mechanism of the action is yet to be clarified, cells
survived longer. In order to effectively inhibit cruzain in
vivo, an inhibitor must penetrate cell membranes.
Among the inhibitors tested, a-ketoamide (AQ665183),
which is more hydrophobic and possibly more mem-
brane permeable displayed stronger trypanostatic activ-
ity than a-ketoester (AQ581332) or a-keto acid
(AQ616476), which has poor membrane permeability
as illustrated in a platelet membrane permeability as-
say.14,28 It is not uncommon that potent inhibitors from
in vitro screening prove to be ineffective against intracell-
ular proteases due to poor membrane permeability. To
enhance membrane permeability, further optimization
of the inhibitors will be necessary. This effort will focus
on the preparation of more neutral, hydrophobic com-
pounds. In addition, it has been observed that the ester
bond of a-ketoester tends to be rapidly metabolized in
vivo.20 Therefore, further optimization to improve the
specificity and potency of inhibitors may focus on mainly
peptidyl a-ketoamides for better membrane permeabil-
ity. However, the hydrophobicity of the inhibitors will
also have to be carefully balanced with aqueous solubil-
ity to achieve a potential therapeutic agent.
SAR has been established using high speed chemistry,
three-dimensional enzyme structure guided modeling,
and docking approaches. A total of 20 compounds dis-
played picomolar potency in in vitro assays and three
inhibitors representing different a-keto-based inhibitor
chemotypes demonstrated anti-trypanosomal activity
in cell culture test, which supports the validity and use-
˚
fulness of the approach. A 2.3 A crystal structure of
cruzain covalently bound to one of the compounds is
also reported. The compounds are potent transition
state analog inhibitors, which reversibly inhibit cruzain
by covalently binding to the sulfhydryl moiety of active
site cysteine. These potent and novel inhibitors of cruz-
ain are expected to provide a promising entry point for
the development of new anti-trypanosomal chemother-
apy and useful probes for the assessment of the enzymeÕs
role in Chagas disease. So far, various chemical back-
bones such as fluoromethyl ketones, vinyl sulfones, thio
semicarbazones have been developed as cruzain inhibi-
tors. The potent a-keto based inhibitors which have
been developed from this study expand the diversity of
chemical scaffolds available for developing cruzain
inhibitors and may be useful to overcome ever increas-
ing drug resistance of T. cruzi.
5. Experimental
5.1. Materials
4. Conclusion
A novel series of peptide based a-ketoamide, a-keto-
ester, a-ketoacid, and aldehyde analogs has been devel-
oped and shown to be potent inhibitors of cruzain, a
major cysteine protease involved in Chagas disease.
To aid the development and the optimization of the
inhibitors at the P1, P2, P3, and P10 positions, a specific
Cruzain was expressed, purified as previously described
and stored in 20 mM bis–tris buffer pH 5.8 at
4 ꢁC.6,16,29,30 All chemicals for buffer preparation were
purchased from Sigma unless described otherwise. The
substrate, Cbz-Phe-Arg-AMC, was purchased from

2152
Y. Choe et al. / Bioorg. Med. Chem. 13 (2005) 2141–2156
molecular probes and 10 mM stock solution in neat
DMSO was stored at ꢀ20 ꢁC.
where Kꢂi is the apparent dissociation constant for the
enzyme–inhibitor complex, t0 is the control (uninhib-
ited) velocity, ti is the inhibited steady-state velocity,
[I0] is the total inhibitor concentration, and [E0] is the
total cruzain concentration. The Kꢂi values for various
substrate concentrations were plotted against the sub-
5.2. In vitro cruzain activity assay
A 96-well formatted fluorescent assay was used to moni-
tor the time course of the activity assay. For Km deter-
mination, initial velocities at a total of 11 substrate
concentrations in the range from 0.15 to 30 lM were
measured. The stock solutions were prepared in 100%
DMSO, and 5 lL of each were added to the reaction
mixture to produce a final mixture with 5% DMSO
total. In the case of initial screening experiments, a
substrate concentration of 3 lM (2 · Km) was used.
Briefly, inhibitors in DMSO 5 lL were added to
175 lL of substrate solution (substrate in 5 lL 100%
DMSO) and assay buffer (100 mM NaOAc, pH 5.5,
10 mM EDTA, and 10 mM DTT) and the well was thor-
oughly mixed. The reaction was initiated by the addition
of cruzain (30 lL, 1.4 nM in assay buffer, preincubated
in a siliconized microcentrifuge tube on ice for at least
10 min). The final assay volume was 210 lL and the final
DMSO concentration was less than 5%. The initial
velocity was monitored for 5 min at 22 ꢁC on a BMG
FluoStar spectrofluorimeter (BMG Labtechnologies)
by following the increase in the fluorescent signal (exci-
tation at 390 nm and emission at 460 nm). Since cruzain
is highly hydrophobic like other members of the papain
family, Costar NBS (nonbinding surface) plates were
used or 0.005% Brij35 was included in the assay buffer
in order to prevent any potential aggregation.
strate concentration and
displayed a y-intercept, which is an estimate of Ki of
AQ581332 for cruzain.
a linear regression line
5.4. Cell culture test
Growth inhibition of T. cruzi amastigotes and toxicity
of the inhibitors were examined by checking the survival
of a mammalian cell culture model infected by T.cruzi
trypomastigotes. J774 mouse macrophage cell line was
cultured in RPMI 1640 medium with 5% heat-inacti-
vated FBS (RPMI medium). For growth inhibition
studies, J774 macrophages were irradiated (3000 rad)
to arrest the cell cycle and cultured in six-well plates
for 24 h at 37 ꢁC. The cultures were infected with 106
T. cruzi trypomastigotes/well of the Y strain for three
hours. Monolayers were washed with RPMI medium
and treated with the inhibitors at 20–30 lM in RPMI
medium (triplicate wells per inhibitor). Inhibitor stocks
were made at 100 mM in DMSO and control without
inhibitors included DMSO (0.01–0.02% v/v). Trypo-
mastigotes output, an indicative of the completion of the
parasite intracellular cycle, was counted in inhibitor-
treated and untreated cultures to measure the growth
inhibition of intracellular amastigote stage of the para-
site by the inhibitors. The culture medium containing
inhibitor or DMSO as the control was replaced daily
and inhibitor efficacy was monitored daily. Survival time
is defined as the time before the cell monolayer is de-
stroyed by the infection. Untreated cells were completely
destroyed after 6 days in this assay indicating the para-
site completed the intracellular cycle and released from
the host cells at the end of the 6 day cycle.33
5.3. Inhibition assays
A solution of substrate (3 lM, 5 lL) was premixed with
inhibitor (1 nM to 10 lM, 5 lL/concentration/well) and
assay buffer (180 lL). The reaction was initiated by the
addition of Cruzain (30 lL, 1.4 nM). Initially, cruzain
IC50 values for all compounds were determined and then
Ki was determined for the most active analogs. The Ki
values were determined by measuring the initial velocity
at varying substrate concentrations in the presence of dif-
ferent inhibitor concentrations and analyzing the result
using the program developed by Cleland.31 A double re-
ciprocal plot was generated in order to demonstrate the
mode of inhibition. The kinetic fluorescence measure-
ment for the slow-binding inhibitor AQ581332 was car-
ried out twice in duplicate using a SpectraMax Gemini
fluorescence spectrometer (Molecular Devices) with an
excitation wavelength of 380 nm, an emission wavelength
of 460 nm, and a 435 nm cutoff filter. Progress curves for
the inhibition of cruzain by AQ581332 followed typical
slow-binding kinetics. Resulting steady-state rates were
fit by nonlinear regression to the following equation for
several substrate concentrations (5, 10, 25, 50, 75, and
100 lM) using the program KALEIDAGRAPH (Synergy
software) according to Eggers et al.32
5.5. Modeling
The modeling initially began with the published struc-
ture of cruzain complexed to Cbz-Tyr-Ala-FMK (PDB
code: 1AIM).16 Other cysteinyl protease-covalent inhib-
itor structures, such as a cathepsin K-hydrazide complex
(PDB code: 1AYW), were used to guide the develop-
ment of a Ôprime–nonprimeÕ spanning model for
a-ketoamide binding to cruzain. Additional parameters
for inter- and intra-atomic torsional, angle, and distance
constraints for the bound a-ketoamide group were taken
from serine protease-a-ketoamide complexes, such as
the trypsin–cyclotheonamide A complex (PDB code:
1TYN). The structural model for the covalent bound
AQ665183 to cruzain (Fig. 3) was constructed using
the backbone of the crystallographic ligand from the
1AIM structure and making modifications as needed
to construct the AQ665183 compound. Torsional, angle,
and distance constraints derived from other complexes
were applied and the AQ665183–cruzain complex con-
structed was subjected to several cycles of molecular
mechanics-based energy minimization using SYBYL (Tri-
pos Inc.) in order to arrive at the final structural model.
ti=t0 ¼ 1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
½E0ꢁ þ ½I0ꢁ þ Kꢂi ꢀ ð½E0ꢁ þ ½I0ꢁ þ Kꢂi Þ ꢀ 4½E0ꢁ½I0ꢁ
ꢀ
2½E0ꢁ

Y. Choe et al. / Bioorg. Med. Chem. 13 (2005) 2141–2156
2153
5.6. Preparation, purification, and crystallization of
cruzain bound to inhibitors
strong difference electron density peaks (P3r) and with-
in acceptable hydrogen-bonding distance and geometry.
Cruzain was expressed and purified as previously de-
scribed.6,16,34 Active cruzain was reversibly inhibited
with excess methyl methanethiolsulfonate (MMTS).
This prevented further proteolytic activity of the enzyme
while purification was completed. After passage over a
5.8. Synthesis and analysis
All commercial reagents were used as received from their
respective suppliers unless otherwise specified. Flash col-
umn chromatography was performed using silica gel 60
(Merck Art 9385). 1H NMR spectra were recorded on a
Varian Gemini 300 MHz NMR system. Deuterated sol-
vents were obtained from Cambridge Isotope Laborato-
ries, Inc. and tetramethylsilane was used as an internal
standard. Chemical shifts are reported in ppm (d) down-
field relative to tetramethylsilane and the coupling con-
stants are given in Hertz. HPLC purity analysis was run
using UV detection at 254 nm on a Shimadzu 10 A analyt-
ical chromatography system fitted with a YMC reverse
phase (C18, 3 lm, 3 · 100 mm) column and/or ELSD
on a Sedex. The retention times reported in this experi-
mental were obtained using the following conditions.
Samples of the reaction product (5 lL, 1 mg/mL) in
DMSO were injected onto the column. The flow rate
was 0.75 mL per minute. The linear gradient elution for
the acetonitrile/water eluent was from 20% to 100% aceto-
nitrile over 2.5 min. The elution solvents contained 0.1%
trifuoroacetic acid. Mass spectra were obtained on a
Micromass Platform 2 using a Gilson 215 autosampler
and an HP 1050 LC. Ionization of the compounds was
achieved either by electrospray (ES+) or chemical ioniza-
tion (APCI, N2) techniques. The following abbreviations
are used: TFA (trifluoroacetic acid), DIEA (N,N-diiso-
propylethylamine), HBTU (O-benzotriazol-1yl-N,N,
N0,N0-tetramethyuronium hexafluorophosphate.
¨
MonoQ column on an AKTA Explorer system (Amer-
sham Pharmacia Biotech Inc.), fractions containing pure
cruzain bound to MMTS were concentrated via vacuum
dialysis against 2 mM bis–tris pH 5.8 to a final concen-
tration of 10–12 mg/mL for crystallization. During con-
centration, DTT was added to a concentration of 5 mM
to remove the MMTS. The inhibitor AQ581332, dis-
solved in DMSO, was added in molar excess quantities
to achieve complexation. Crystals were obtained by
the hanging drop method against a grid of 0.6–1.0 M so-
dium citrate, pH 5.5–7.0. Drops of cruzain bound inhib-
itor that were equilibrated at 18 ꢁC for 24 h were
microseeded with existing crystals of cruzain bound to
another covalently bound inhibitor.16 After 7–12 days,
small crystals of cruzain bound to a vinyl sulfone-based
inhibitor appeared. These crystals were crushed and di-
luted 104 or 105-fold to produce seeding solutions. New
hanging drops of cruzain bound to inhibitor that had
been equilibrated at 18 ꢁC for 24 h were microseeded
with one microliter of diluted seeding stock solution.
Data collection quality crystals (100–200 lm per edge)
grew to maximum size within 3 weeks.
5.7. Data collection, structure solution, and crystallo-
graphic refinement
Data were collected at room temperature on an R-axis
IIc detector fitted with Yale MSC mirrors. A Rigaku
RU-H3R rotating anode operating at 100 mA, 50 kV
was used for this complex. Data were integrated and
scaled using the HKL package of software.35 Table 5
summarizes data processing statistics. The structure
was solved by molecular replacement using AMoRe.36
5.8.1. Preparation of 2-(ethoxycarbonyl)-2-oxoethylidene-
triphenylphosphorane. A solution of ethyl bromopyru-
vate (100 g, 0.5127 mol) in dry carbon tetrachloride
(200 mL) was added dropwise over 3 h to a stirred and
cooled (0 ꢁC) solution of PPh3 (135 g, 0.514 mol) in
dry carbon tetrachloride (1.5 L). The reaction warmed
to rt and then stirred for an additional 36 h. The super-
natant was decanted from the yellow hygroscopic
crystals, which were then washed with anhydrous ether
(3 · 400 mL) by trituration and decantation. The result-
ing dried sticky solid was dissolved in methanol
(600 mL) and the solution cooled to 0 ꢁC. The pH was
adjusted to 10 by gradual addition of iced aqueous so-
dium carbonate (1 N). The solution was diluted to 4 L
with ice water and stirred for 1 h at 0 ꢁC, after which
the precipitate was collected and washed with cold
water. It was then recrystallized from hot ethanol and
water. The resulting crystals separated out as plates
(120 g). A second crop (11 g) was obtained as described.
The total yield was 131 g (75%). Mp 182–183 ꢁC.
˚
The 1.6 A structure of cruzain bound to a covalently
bound peptidyl inhibitor was used as a search model,
excluding the inhibitor molecule and all water mole-
cules.16 One unique solution was easily found for the
complex, with a resulting R-factor prior to refinement
of 0.296% and correlation coefficient of 0.758. The start-
ing model and parameter set for the inhibitor molecule
was generated with Moloc.37 Initial placement of the
inhibitor model into the structure was made via inspec-
tion of Fo–Fc difference electron density maps calculated
from the molecular replacement solution and contoured
at 3r. Subsequent model adjustment and refinement
(positional and individual B-factor) were accomplished
with CHAIN and X-PLOR.38 A simulated annealing
1
˚
omit map calculated with a 5 A radius around the omit-
LC/MS (ES+) 378.5. H NMR (300 MHz, CDCl3): d
ted inhibitor molecule were used to confirm placement
and orientation of the inhibitor molecule. Final rounds
of refinement were completed with the Refmac module
of CCP4i39 and electron density maps were visualized
with Quanta.40 Water molecules were added automati-
cally with the X-solvate module of Quanta as well as
manually with the requirement that they be located in
1.35 (t, J1 = 7.9 Hz, 3H, –CH3), 4.26 (dd, J1 = 7.9 Hz,
2H, –CH2–), 4.84 (d, J1 = 25.7 Hz, 1H, CH), 7.42–7.73
(m, 15H, –C6H5). C23H21O3P (376.38).
5.8.2. Preparation of peptidyl vinyl-a-keto esters. A
solution of the commercially available aldehyde Z-
Phe CHO (0.200 g, 0.706 mmol, 1.0 equiv) and the

2154
Y. Choe et al. / Bioorg. Med. Chem. 13 (2005) 2141–2156
(ethoxyoxalyl)methylene–triphenylphosphorane (0.533 g,
1.412 mmol, 2.0 equiv) in anhydrous dioxane (10 mL)
was heated under reflux for 10 h. The pale yellow solu-
tion was reduced in vacuo, the residue triturated with
diethyl ether (50 mL) and the mixture cooled to 0 ꢁC.
The precipitated triphenylphosphine oxide was filtered
off and washed with cold diethyl ether (2 · 20 mL) and
the combined filtrate and washings were concentrated
in vacuo to a thick oil, which after purification over
silica (EtOAc–hexane, 1:3) yielded the vinyl a-keto ester
as a white solid (0.1885 g, 70%).
J1 = 7.14 Hz, 3H, –CH–CH3), 1.32–1.4 (m, 3H, O–
CH2–CH3), 2.9–3.15 (m, 2H, –CH2–Ph), 4.32–4.38 (m,
2H, O–CH2–CH3), 4.45 (m, 1H, –CH–CH2–Ph), 5.09
(s, 2H, O–CH2–Ph), 5.32 (br s, 1H, N–H), 6.2 (br s,
1H, N–H), 7.20–7.35 (m, 10H, aromatic). MS (EI), m/
z (relative intensity): 428 ([M+], 2).
5.8.4.2. Z-Phe-Ala-a-keto-acid. Purified by reverse
phase chromatography. Yield 69%. 1H NMR (300
MHz, CDCl3): d 1.21 (t, J1 = 7.71 Hz, 3H, CH–CH3),
3.04 (m, 2H, –CH2–Ph), 4.48 (m, 1H, –CH–CH2–Ph),
5.05 (m, 3H, O–CH2–Ph and CH–CH2–Ph), 7.16–7.32
(m, 10H, aromatic). MS (EI), m/z (relative intensity):
399 ([M+], 2).
1
LC/MS (AP+) 382.3. H NMR (300 MHz, CDCl3): d
1.36 (t, J = 7.9 Hz, 3H, –CH3), 2.95 (d, J = 3 Hz, 2H,
–CH2–C6H5), 4.33 (q; J = 7.9 Hz, –CH2), 4.71–4.83 (m,
1H, –CH–), 4.88 (s, 1H, NH), 5.08 (s, 2H, –CH2–
C6H5), 6.72 (d, J = 17.8 Hz, 1H, –CH), 7.08–7.41 (m,
11H, –CH and –C6H5).
5.8.4.3. Z-Phe-Ala-a-keto-methylamide. Purified by
column chromatography (3% MeOH in CH2Cl2). Yield
50%. 1H NMR (300 MHz, CDCl3):
d 1.30 (d,
J1 = 7.23 Hz, 1.5H, –CH–CH3), 1.45 (d, J1 = 7.15 Hz,
1.5H, –CH–CH3), 3.05–3.19 (m, 2H, –CH–CH2–Ph),
4.32–4.51 (m, 3H, O–CH2–CH3 and NH–CH2–Ph),
5.12 (m, 2H, O–CH2–Ph), 5.24 (m, 1H, –CH–), 5.42
(br s, 1H, N–H), 6.3 and 6.5 (br s, 1H, N–H), 7.23–
7.38 (m, 15H, aromatic). MS (EI), m/z (relative inten-
sity): 488 ([M+], 2).
A solution of the aldehyde Z-Phe-Ala-CHO (0.3 g,
0.847 mmol, 1.0 equiv) and (ethoxyoxalyl)methylene–
triphenylphosphorane (0.639 g, 1.693 mmol, 2.0 equiv)
in anhydrous dioxane (10 mL) were treated as described
earlier. Purification over silica (EtOAc–hexane, 1:3)
yielded the corresponding vinyl a-keto ester as a white
solid (0.2491 g, 65%) of the vinyl a-keto ester.
5.8.4.4. Z-Phe-Ala-a-keto-benzylamide. Purified by
column chromatography (3% MeOH in CH2Cl2). Yield
1
LC/MS (ES+) 452.9. H NMR (300 MHz, CDCl3): d
1.21 (d, J = Hz, 3H, –CH3), 1.42 (dd, J1 = J2 = 7.9
Hz, 3H, –CH3), 3.10 (dd, J1 = Hz, J2 = Hz, 2H,
–CH2–C6H5), 4.38 (dd, J1 = 15.9 Hz, J2 = 7.9 Hz,
–CH2), 4.45 (q, J = 7.5 Hz, 1H, –CH), 4.72 (q,
J = 7.5 Hz, 1H, –CH), 5.11 (s, 2H, –CH2–C6H5), 5.60
(d, J = 7.5 Hz, 1H, NH), 6.24 (d, J = 7.5 Hz, 1H, NH),
6.59 (d, J = 17.7 Hz, 1H, –CH), 6.89 (dd, J1 = 17.7 Hz,
J2 = 5.9 Hz, 1H, –CH), 7.16–7.45 (m, 10H, –C6H5).
72%. 1H NMR (300 MHz, CDCl3):
d 1.30 (d,
J1 = 7.23 Hz, 1.5H, –CH–CH3), 1.41 (d, J1 = 7.15 Hz,
1.5H, –CH–CH3), 3.05–3.19 (m, 2H, –CH–CH2–Ph),
4.32–4.51 (m, 3H, O–CH2–CH3 and NH–CH2–Ph),
5.12 (m, 2H, O–CH2–Ph), 5.24 (m, 1H, –CH–), 5.42
(br s, 1H, N–H), 6.3 and 6.5 (br s, 1H, N–H), 7.23–
7.38 (m, 15H, aromatic). MS (EI), m/z (relative inten-
sity): 488 ([M+], 2).
5.8.3. Preparation of ML-library. To a cooled solution
(0 ꢁC) of the Cbz-protected a-ketoamide (0.6 g.
2.56 mmol, 1.0 equiv) in dry dichloromethane (10 mL)
was added BCl3 (1.20 g, 10.24 mmol, 4.0 equiv). The
reaction mixture was stirred at room temperature for
14 h. The solvent was removed under reduced pressure
and the resulting residue was dissolved in DMF
(10.25 mL) to obtain a 0.25 M reaction solution.
5.8.4.5. Z-Phe-Val-a-keto-benzylamide. Purified by
recrystallization from (CH2Cl2–hexane). Yield 60%. H
1
NMR (300 MHz, CDCl3): d 1.30 (d, J1 = 7.23 Hz, 1.5H,
–CH–CH3), 1.41 (d, J1 = 7.15 Hz, 1.5H, –CH–CH3),
3.05–3.19 (m, 2H, –CH–CH2–Ph), 4.32–4.51 (m, 3H,
O–CH2–CH3 and NH–CH2–Ph), 5.12 (m, 2H,
O–CH2–Ph), 5.24 (m, 1H, –CH–), 5.42 (br s, 1H, N–
H), 6.3 and 6.5 (br s, 1H, N–H), 7.23–7.38 (m, 15H, aro-
matic). MS (EI), m/z (relative intensity): 488 ([M+], 2).
A solution of Cbz-protected amino acid in DMF was
stirred with 1.1 equiv of DIEA (0.25 M DMF solution)
and 1.05 equiv HBTU (0.25 M DMF solution) at room
temperature. After 15 min a 1.1 equiv of the Cbz-depro-
tected a-keto amide (0.25 M DMF solution) was added
and stirred for 12 h at room temperature. The resulting
mixtures were diluted with ethyl acetate (10 mL) and ex-
tracted with 1 N NaHCO3 (10 mL) solution and water
(10 mL). After removal of the solvent in vacuo the resi-
dues were analyzed by LC–MS and the crude products
were isolated by preparative HPLC.
5.8.5. General method for the synthesis of peptide
alcohols. A solution of an amino alcohol (1.26 mmol)
in 1,4-dioxane (5 mL) was added dropwise to a solution
of a commercially available activated N-protected amino
acid succinimide ester (1.26 mmol) in 1,4-dioxane
(5 mL). It was stirred at room temperature for 12 h.
Diethylether was added (40 mL) and the organic layer
washed with 1 N HCl (10 mL), saturated NaHCO3
(10 mL), saturated NaCl (10 mL), and dried over
MgSO4. After removal of the solvent, the crude products
were characterized and used without further purification.
5.8.4. Characterization of the Z-Phe-Ala-a-keto
inhibitors
5.8.5.1. 3-[N-(Benzyloxycarbonyl-(S)-phenylalanyl)-
amino]propan-1-ol). Yield 76%. 1H NMR (300 MHz,
CDCl3): d 1.54 (dd, J1 = J2 = 6.03 Hz, 2H, –CH2–),
2.56 (br s, 1H, O–H), 3.05 (m, 2H, –CH2–), 3.30 (dd,
5.8.4.1. Z-Phe-Ala-a-keto-ethylester. Purified by col-
umn chromatography (EtOAc/hexane = 1/1). Yield
75%. 1H NMR (300 MHz, CDCl3):
d 1.25 (t,

Y. Choe et al. / Bioorg. Med. Chem. 13 (2005) 2141–2156
2155
J1 = J2 = 6.03 Hz, 2H, –CH2–), 3.48 (dd, J1 = J2 =
6.03 Hz, 2H, –CH2–OH), 3.48 (dd, J1 = J2 = 5.49 Hz,
2H, –CH2–), 4.47 (dd, J1 = J2 = 6.06 Hz, 1H, –CH–
CH2–Ph), 5.08 (s, O–CH2–Ph), 5.4 (br s, 1H, N–H),
6.17 (br s, 1H, N–H), 7.18–7.35 (m, 10H, aromatic).
MS (EI), m/z (relative intensity): 357 ([M+], 2).
dropwise over 20 min with the peptide alcohol
(1.58 mmol) in dry CH2Cl2 (9 mL) and stirred for one
additional hour at ꢀ63 ꢁC. Triethylamine (0.6672 mL,
4.632 mmol, 4 equiv) was finally added dropwise over
20 min and it was stirred additional 30 min at ꢀ63 ꢁC.
The reaction mixture was let warm up to rt, poured into
a citric acid solution (0.24 N, 10 mL), the organic layer
was separated and the aqueous layer was extracted with
EtOAC (3 · 10 mL). The combined organic extracts
were washed with saturated NaHCO3 (10 mL), satu-
rated NaCl (10 mL) and dried over MgSO4. The solvent
was removed under reduced pressure to yield the crude
aldehyde.
5.8.5.2. 2-[N-(Benzyloxycarbonyl-(S)-phenylalanyl)-
amino]ethan-1-ol).
Purified
by
recrystallization
(EtOAc–pentane). Yield 83%. 1H NMR (300 MHz,
CDCl3): d 2.15 (br s, 1H, O–H), 3.01 (dd, J1 =
J2 = 7.83 Hz, 1H, –CH2–Ph), 3.14 (dd, J1 = J2 =
6.5 Hz, 1H, –CH2–Ph), 3.26–3.35 (m, 2H, –CH2–NH–),
3.48–3.56 (m, 2H, –CH2–OH), 4.35 (dd, J1 =
J2 = 6.6 Hz, 1H, –CH–CH2–Ph), 5.08 (s, O–CH2–Ph),
5.35 (br s, 1H, N–H), 6.00 (br s, 1H, N–H), 7.19–7.37
(m, 10H, aromatic). MS (EI), m/z (relative intensity):
343 ([M+], 2).
5.8.6.1. 3-[N-(Benzyloxycarbonyl-(S)-phenylalanyl)-
amino]propan-1-al). Yield 96%. 1H NMR (300 MHz,
CDCl3): d 2.0 (m, 2H, –CH2–), 2.93 (m, 1H, –CH2–),
3.05 (m, 1H, –CH2–), 3.32–3.48 (br m, 2H, –CH2–Ph),
4.26 (m, 2H, –CH–CH2–Ph), 5.11 (s, O–CH2–Ph), 5.25
(br s, 1H, N–H), 6.02 (br s, 1H, N–H), 7.17–7.41 (m,
10H, aromatic), 9.51 (s, 1H, –CHO). MS (EI), m/z (rel-
ative intensity): 355 ([M+], 2).
5.8.5.3. (2S)-2-[N-(Benzyloxycarbonyl-(S)-phenylalan-
yl)amino]-3-methylpentan-1-ol). Yield 95%. 1H NMR
(300 MHz, CDCl3): d 0.7–0.84 (m, 6H, 2 · CH3) 0.9–
1.03 (m, 1H, CH3–CH–CH2–) 1.25–1.35 (m, 1H, –CH–
CH2–CH3), 1.4–1.52 (m, 1H, –CH–CH2–CH3), 1.8 (br
s, 1H, O–H), 3.00 (dd, J1 = J2 = 7.83 Hz, 1H, –CH2–
Ph), 3.14 (dd, J1 = J2 = 6.5 Hz, 1H, –CH2–Ph), 3.43
(m, 2H, –CH2–OH), 3.65 (m, 1H, –CH–), 4.35 (dd,
J1 = J2 = 6.6 Hz, 1H, –CH–CH2–Ph), 5.11 (s, O–CH2–
Ph), 5.39 (br s, 1H, N–H), 5.62 (br s, 1H, N–H), 7.23–
7.34 (m, 10H, aromatic). MS (EI), m/z (relative inten-
sity): 389 ([M+], 2).
5.8.6.2.
2-[N-(Benzyloxycarbonyl-(S)-phenylalan-
1
yl)amino]ethan-1-al). Yield 98%. H NMR (300 MHz,
CDCl3): d 3.01–3.12 (br m, 2H, –CH2–Ph), 3.26 (m,
2H, –CH2–NH–), 4.45 (m, 1H, –CH–CH2–Ph), 5.08 (s,
O–CH2–Ph), 5.35 (br s, 1H, N–H), 5.82 (br s, 1H, N–
H), 7.17–7.41 (m, 10H, aromatic), 9.53 (s, 1H, –CHO).
MS (EI), m/z (relative intensity): 341 ([M+], 2).
5.8.5.4. (2S)-2-[N-(Benzyloxycarbonyl-(S)-phenylalan-
yl)amino]-3-phenylpropan-1-ol). Yield 53%. 1H NMR
(300 MHz, CDCl3): d 1.78 (br s, 1H, O–H), 2.97 (dd,
J1 = J2 = 8.16 Hz, 1H, –CH2–Ph), 3.02 (m, 1H, –CH2–
Ph), 2.71 (m, 2H, –CH2–Ph), 3.39 (m, 2H, –CH2–OH),
5.8.6.3. (2S)-2-[N-(Benzyloxycarbonyl-(S)-phenylalan-
yl)amino]-3-methylpentan-1-al). Yield 98%. 1H NMR
(300 MHz, CDCl3): d 0.82–0.91 (m, 6H, 2 · CH3)
1.1–1.18 (m, 1H, CH3–CH–CH2–) 1.32–1.45 (m, 1H,
–CH–CH2–CH3), 1.9–2.02 (m, 1H, –CH–CH2–CH3),
3.08 (m, 2H, –CH2–Ph), 4.42–4.58 (2m, 2H, –CH– and
–CH–CH2–Ph), 5.08 (s, O–CH2–Ph), 5.45 (br s, 1H,
N–H), 6.5 (br s, 1H, N–H), 7.16–7.33 (m, 10H, aro-
matic), 9.47 (s, 1H, –CHO). MS (EI), m/z (relative inten-
sity): 397 ([M+], 2).
4.05
(m,
1H,
–CH–CH2–Ph),
4.28
(dd,
J1 = J2 = 6.6 Hz, 1H, –CH–CH2–Ph), 5.08 (s, O–CH2–
Ph), 5.25 (br s, 1H, N–H), 5.74 (br s, 1H, N–H), 7.08–
7.36 (m, 15H, aromatic). MS (EI), m/z (relative inten-
sity): 433 ([M+ ], 2).
5.8.5.5. (2S)-2-[N-(Benzyloxycarbonyl-(S)-phenylalan-
yl)amino]-3-cyclohexylpropan-1-ol). Yield 99%. 1H
NMR (300 MHz, CDCl3): d 0.8–0.95 (m, 2H, –CH2–,
cyclohexyl) 1.1–1.28 (m, 5H, –CH–CH2–CH2–, cyclo-
hexyl), 1.53–1.72 (br m, 4H, –CH2–CH2–, cyclohexyl),
3.0 (m, 1H, –CH2–Ph), 3.18 3.0 (m, 1H, –CH2–Ph),
3.32 (m, 1H, –CH2–OH), 3.45 (m, 1H, –CH2–OH), 1.8
(br s, 1H, O–H), 3.95 (m, 1H, –CH–), 4.35 (dd,
J1 = J2 = 6.6 Hz, 1H, –CH–CH2–Ph), 5.12 (s, O–CH2–
Ph), 5.35 (br m, 1H, N–H), 5.51 (br m, 1H, N–H),
7.26–7.36 (m, 10H, aromatic). MS (EI), m/z (relative
intensity): 439 ([M+], 2).
5.8.6.4. (2S)-2-[N-(Benzyloxycarbonyl-(S)-phenylalanyl)-
amino]-3-phenylpropan-1-al). Yield 93%. 1H NMR
(300 MHz, CDCl3): d 3.03–3.18 (m, 4H, 2 · –CH2–Ph),
4.41 (m, 1H, –CH–CH2–Ph), 4.6 (m, 1H, –CH–CH2–
Ph), 5.07 (s, O–CH2–Ph), 5.23 (br s, 1H, N–H), 6.25
(br s, 1H, N–H), 7.00–7.34 (m, 15H, aromatic), 9.46
(s, 1H, –CHO). MS (EI), m/z (relative intensity): 431
([M+], 2).
5.8.6.5. (2S)-2-[N-(Benzyloxycarbonyl-(S)-phenylalan-
yl)amino]-3-cyclohexylpropan-1-al). Yield 98%. 1H
NMR (300 MHz, CDCl3): d 0.85–0.97 (m, 2H, –CH2–,
cyclohexyl) 1.13–1.26 (m, 4H, –CH2–CH2–, cyclohexyl),
1.59–1.66 (br m, 5H, –CH–CH2–CH2–, cyclohexyl), 3.02
(m, 2H, –CH2–Ph), 4.45 (m, 2H, –CH–CH2–Ph and
–CH–CH2–cyclohexyl), 5.10 (s, O–CH2–Ph), 5.31 (br
m, 1H, N–H), 6.1 (br m, 1H, N–H), 7.18–7.36 (m,
10H, aromatic), 9.39 (s, 1H, –CHO). MS (EI), m/z (rel-
ative intensity): 439 ([M+], 2).
5.8.6. General method for the Swern oxidation of peptide
alcohols under formation of peptide aldehydes. Dimethyl-
sulfide anhydrous (178 mL, 2.49 mmol, 1.5 equiv) in dry
CH2Cl2 (3.6 mL) was added dropwise over 10 min to a
cooled (ꢀ63 ꢁC) solution of oxalyl chloride (1.78 mol,
1.5 equiv) in dry CH2Cl2 (3.6 mL). After additional
20 min at ꢀ63 ꢁC, the reaction mixture was treated

2156
Y. Choe et al. / Bioorg. Med. Chem. 13 (2005) 2141–2156
20. Li, Z.; Patil, G. S.; Golubski, Z. E.; Hori, H.; Tehrani, K.;
Acknowledgements
Foreman, J. E.; Eveleth, D. D.; Bartus, R. T.; Powers, J.
C. J. Med. Chem. 1993, 36, 3472.
21. Rutenber, E. E.; McPhee, F.; Kaplan, A. P.; Gallion, S.
L., Jr.; Hogan, J. C., Jr.; Craik, C. S.; Stroud, R. M.
Bioorg. Med. Chem. 1996, 4, 1545.
The authors would like to thank the Analytical and Pro-
duction Groups at ArQule, Inc. for their support. This
work was supported by a NIH grant P01-AI35707 to
C.S.C. and J.C.E.
22. Walter, J.; Bode, W. Hoppe Seylers Z. Physiol. Chem.
1983, 364, 949.
23. Sasaki, T.; Kishi, M.; Saito, M.; Tanaka, T.; Higuchi, N.;
Kominami, E.; Katunuma, N.; Murachi, T. J. Enzyme
Inhib. 1990, 3, 195.
References and notes
1. World Health Organization: Special Program for Research
and Training in Tropical Diseases (TDR).
2. Du, X.; Guo, C.; Hansell, E.; Doyle, P. S.; Caffrey, C. R.;
Holler, T. P.; McKerrow, J. H.; Cohen, F. E. J. Med.
Chem. 2002, 45, 2695.
3. Engel, J. C.; Doyle, P. S.; Palmer, J.; Hsieh, I.; Bainton, D.
F.; McKerrow, J. H. J. Cell Sci. 1998, 111(Pt5),
597.
24. Angelastro, M. R.; Mehdi, S.; Burkhart, J. P.; Peet, N. P.;
Bey, P. J. Med. Chem. 1990, 33, 11.
25. Hu, L. Y.; Abeles, R. H. Arch. Biochem. Biophys. 1990,
281, 271.
26. Almquist, R. G.; Chao, W. R.; Ellis, M. E.; Johnson, H.
L. J. Med. Chem. 1980, 23, 1392.
27. Szelke, M.; Leckie, B.; Hallett, A.; Jones, D. M.; Sueiras,
J.; Atrash, B.; Lever, A. F. Nature 1982, 299, 555.
28. Harbeson, S. L.; Abelleira, S. M.; Akiyama, A.; Barrett,
R., III; Carroll, R. M.; Straub, J. A.; Tkacz, J. N.; Wu, C.;
Musso, G. F. J. Med. Chem. 1994, 37, 2918.
29. LaLonde, J. M.; Zhao, B.; Smith, W. W.; Janson, C. A.;
DesJarlais, R. L.; Tomaszek, T. A.; Carr, T. J.;
Thompson, S. K.; Oh, H. J.; Yamashita, D. S.; Veber,
D. F.; Abdel-Meguid, S. S. J. Med. Chem. 1998, 41,
4567.
4. Engel, J. C.; Doyle, P. S.; Hsieh, I.; McKerrow, J. H. J.
Exp. Med. 1998, 188, 725.
5. Itow, S.; Camargo, E. P. J. Protozool. 1977, 24, 591.
6. Eakin, A. E.; Mills, A. A.; Harth, G.; McKerrow, J. H.;
Craik, C. S. J. Biol. Chem. 1992, 267, 7411.
7. McGrath, M. E.; Eakin, A. E.; Engel, J. C.; McKerrow, J.
H.; Craik, C. S.; Fletterick, R. J. J. Mol. Biol. 1995, 247,
251.
8. Roush, W. R.; Gonzalez, F. V.; McKerrow, J. H.; Hansell,
E. Bioorg. Med. Chem. Lett. 1998, 8, 2809.
9. Scheidt, K. A.; Roush, W. R.; McKerrow, J. H.; Selzer, P.
M.; Hansell, E.; Rosenthal, P. J. Bioorg. Med. Chem. 1998,
6, 2477.
30. Thompson, S. K.; Smith, W. W.; Zhao, B.; Halbert, S. M.;
Tomaszek, T. A.; Tew, D. G.; Levy, M. A.; Janson, C. A.;
DÕAlessio, K. J.; McQueney, M. S.; Kurdyla, J.; Jones, C.
S.; DesJarlais, R. L.; Abdel-Meguid, S. S.; Veber, D. F. J.
Med. Chem. 1998, 41, 3923.
10. Coffen, D. L.; Baldino, C. M.; Lange, M.; Tilton, R. F.;
Tu, C. Med. Chem. Res. 1998, 8(4/5), 206.
11. Baldino, C. M.; Coffen, D. L.; Chipman, S. D.; Cheng, H.
U.S. Patent, 2000.
31. Cleland, W. W. In Methods in Enzymology; Purich, D. L.,
Ed.; Academic, 1979; Vol. 103.
32. Eggers, C. T.; Wang, S. X.; Fletterick, R. J.; Craik, C. S. J.
Mol. Biol. 2001, 308, 975.
12. Baldino, C. M. J. Comb. Chem. 2000, 2, 89.
13. Baldino, C. M.; Caserta, J.; Goetzinger, W.; Harris, M.;
Hartsough, D.; Yohannes, D.; Yu, L.; Kyranos, J. N.
Curr. Drug Discov. 2004, 15.
14. Li, Z.; Ortega-Vilain, A. C.; Patil, G. S.; Chu, D. L.;
Foreman, J. E.; Eveleth, D. D.; Powers, J. C. J. Med.
Chem. 1996, 39, 4089.
15. Reetz, M. T.; Griebenow, N. Liebigs Ann. 1996, 335.
16. Gillmor, S. A.; Craik, C. S.; Fletterick, R. J. Protein Sci.
1997, 6, 1603.
17. Wassermann, H. H.; Petersen, A. K. Tetrahedron Lett.
1997, 38, 953.
33. Du, X.; Hansell, E.; Engel, J. C.; Caffrey, C. R.; Cohen, F.
E.; McKerrow, J. H. Chem. Biol. 2000, 7, 733.
34. Eakin, A. E.; McGrath, M. E.; McKerrow, J. H.;
Fletterick, R. J.; Craik, C. S. J. Biol. Chem. 1993, 268,
6115.
35. Otwinowski, Z. In Proceedings of the CCP4 Study
Weekend: Data Collection and Processing; Sawyer, L.
Isaacs N., Bailey, S., Eds.; Daresbury Laboratory: War-
rington, UK, 1993.
36. Navaza, J. Acta Crystallogr. 1994, 50, 157.
37. Gerber, P. R.; Muller, K. J. Comput. Aided Mol. Des.
1995, 9, 251.
18. Brinen, L. S.; Hansell, E.; Cheng, J.; Roush, W. R.;
McKerrow, J. H.; Fletterick, R. J. Struct. Fold. Des. 2000,
8, 831.
38. Sack, J. S. J. Mol. Graphics 1988, 6, 244.
39. Potterton, E.; Briggs, P.; Turkenburg, M.; Dodson, E.
Acta Crystallogr. 2003, D59, 1131.
19. Shoichet, B. K.; Kuntz, I. D. Protein Eng. 1993, 6, 723.
40. Accelrys; San Diego, CA, USA, 2000.