Identification of a buried pocket for potent and selective inhibition of Chk1: Prediction and verification

Article abstract of DOI:10.1016/j.bmc.2005.10.022

Inhibition of the Chk1 kinase by small molecules binding to its active site is a strategy of great therapeutic interest for oncology. We report how computational modelling predicted the binding mode of ligands of special interest to the Chk1 ATP site, for representatives of an indazole series and debromohymenialdisine. These binding modes were subsequently confirmed by X-ray crystallography. The binding mode of a potent indazole derivative involves non-conventional C-H...O and N-H...π-aromatic interactions with the protein. These interactions are formed in a buried pocket at the periphery of the ATP-binding site, the importance of which has previously been overlooked for ligand design against Chk1. It is demonstrated that filling this pocket can confer ligands with dramatically enhanced affinity for Chk1. Structural arguments in conjunction with assay data explain why targeting this pocket is also advantageous for selective binding to Chk1. Structural overlays of known inhibitors complexed with Chk1 show that only the indazole series utilizes the pocket of interest. Therefore, the analysis presented here should prove helpful in guiding future structure-based ligand design efforts against Chk1.

Full text of DOI:10.1016/j.bmc.2005.10.022

Bioorganic & Medicinal Chemistry 14 (2006) 1792–1804
Identification of a buried pocket for potent and selective
inhibition of Chk1: Prediction and verification
Nicolas Foloppe,* Lisa M. Fisher, Geraint Francis, Rob Howes,
Peter Kierstan and Andrew Potter
Vernalis (R&D) Ltd, Granta Park, Abington, Cambridge CB1 6GB, UK
Received 7 June 2005; revised 10 October 2005; accepted 18 October 2005
Available online 9 November 2005
Abstract—Inhibition of the Chk1 kinase by small molecules binding to its active site is a strategy of great therapeutic interest for
oncology. We report how computational modelling predicted the binding mode of ligands of special interest to the Chk1 ATP site,
for representatives of an indazole series and debromohymenialdisine. These binding modes were subsequently confirmed by X-ray
crystallography. The binding mode of a potent indazole derivative involves non-conventional C–Hꢀ ꢀ ꢀO and N–Hꢀ ꢀ ꢀp-aromatic
interactions with the protein. These interactions are formed in a buried pocket at the periphery of the ATP-binding site, the impor-
tance of which has previously been overlooked for ligand design against Chk1. It is demonstrated that filling this pocket can confer
ligands with dramatically enhanced affinity for Chk1. Structural arguments in conjunction with assay data explain why targeting this
pocket is also advantageous for selective binding to Chk1. Structural overlays of known inhibitors complexed with Chk1 show that
only the indazole series utilizes the pocket of interest. Therefore, the analysis presented here should prove helpful in guiding future
structure-based ligand design efforts against Chk1.
ꢀ 2005 Elsevier Ltd. All rights reserved.
1. Introduction
series elaborated around an indazole core, which in-
cludes at least 13 compounds.11,24 Structural information
for DBH bound to Chk1 is also lacking. Importantly, the
indazole series includes compound 1 (Table 1), one of the
most potent Chk1 inhibitors reported to date.
The human Chk1 kinase, which plays an essential role in
the regulation of the cell cycle G2/M checkpoint,1–10 is a
promising target for the development of small molecule
inhibitors against cancer.11–20 Selective Chk1 inhibitors
should also prove useful to study the regulation of the
G2/M checkpoint from a fundamental biology point of
view.
We present how molecular modelling successfully pre-
dicted the binding mode of key representatives of the
indazole series (compounds 1 and 2) and DBH to the
Chk1 ATP-binding site. Analysis of these docking
modes in conjunction with the associated binding affin-
ities led us to formulate a specific working hypothesis
regarding ligand design against Chk1, towards im-
proved binding affinity and selectivity. Specifically, the
comparative modelling of compounds 1 and 2 pinpoint-
ed a buried pocket at the periphery of the Chk1 ATP
site, suggesting convincingly that it is binding to this
pocket which explains the dramatic affinity difference
between these two compounds. A structural analysis of
several kinase active sites, combined with assay data,
also indicated that targeting this pocket ought to be
advantageous for selective binding to Chk1. Therefore,
the modelling work led us to devise a structure-based
strategy of general interest for elaboration of Chk1
inhibitors. This strategy was vindicated when the mod-
elled binding modes were subsequently confirmed with
Only a few chemical classes acting as Chk1 inhibitors
have been reported.9,11,15–17,19,21–26 Many of these
inhibitors are natural products such as debromohymen-
ialdisine (DBH),9 or staurosporine and its deriva-
tives.15–17,21–23 Crystal structures of staurosporine and
two analogs bound to the Chk1 kinase domain have been
described.18 This structural information is, however,
missing for the previously presented drug-like chemical
Abbreviations: DBH, debromohymenialdisine; PDB, Protein data
bank; RMSD, root mean square deviation; SAR, structure–activity
relationship.
Keywords: Drug design; Hydrogen bonds; Molecular modelling;
Molecular
*
recognition;
Oncology;
X-ray
crystallography.
Corresponding author. Tel.: +44 1223 895 338; fax: +44 1223 895
556; e-mail: n.foloppe@vernalis.com
0968-0896/$ - see front matter ꢀ 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2005.10.022

N. Foloppe et al. / Bioorg. Med. Chem. 14 (2006) 1792–1804
Table 1. Structures of compounds 1–3 and their binding affinitiesa to kinases of interest
1793
Compound
Structure
Ki (lM)
IC50 (lM)
Chk1
Chk1
CDK1
CDK2
GSK3b
PDK1
Akt1
PKA
O
O
N
N
N
N
1
2
0.026
(0.005)
0.042
(0.013)
0.040
(0.018)
0.104
(0.039)
0.314
(0.066)
0.443
(0.142)
23.600
(1.840)
171.719
(38.850)
N
N
N
8.580
(3.060)
6.296
(4.160)
0.845
(0.200)
14.309
(5.529)
0.744
(0.166)
2.706
(2.100)
>50
>200
N
N
N
N
O
N
3 (DBH)
0.659
(0.021)
0.353
(0.285)
0.266
(0.096)
0.856
(0.788)
0.168
(0.013)
8.830
(0.495)
0.190
(0.760)
>50
N
O
a Every reported affinity is the average of at least two measurements, with standard deviations given in parentheses.
X-ray crystallography. As predicted, the X-ray binding
mode includes a number of non-conventional protein–
ligand interactions such as C–Hꢀ ꢀ ꢀO and N–Hꢀ ꢀ ꢀp-aro-
matic interactions. The present work also provides new
and careful determinations of the affinities of com-
pounds 1–3 for Chk1.
ed here for the first time, is important to interpret the SAR
of substituents grafted on this core. Indeed, the influence
of the substituents on affinity can only be deduced if the
affinity of the core itself is known. This shows that the
additional methoxyphenol moiety in compound 1 leads
to a 330-fold increase in affinity for Chk1. It is therefore
important to characterize the structural positioning of
this methoxyphenol in its Chk1-binding site, to gain a bet-
ter understanding of the factors which lead to such a dra-
matic increase in affinity.
Structural overlays of the known chemical inhibitors
complexed to Chk1 show that only the indazole series
takes advantage of the pocket of interest. This suggests
that there is scope for further elaboration of other inhib-
itors, and will provide helpful guidance for future struc-
ture-based design efforts towards more potent and
selective Chk1 inhibitors.
DBH was also found to inhibit Chk1 in an ATP-com-
petitive manner, confirming a previous report which
characterized the binding of DBH to Chk1 in terms of
an IC50.9 We obtained a Ki of ꢁ0.6 lM for the affinity
of DBH to Chk1.
2. Results and discussion
2.2. Prediction and confirmation of the binding mode of
the indazole scaffold
Given the excellent agreement between predicted mod-
elled binding modes and their subsequently determined
X-ray counterparts (see Section 2.2), only the X-ray
binding modes are shown (Figs. 1, 2 and 4).
Compounds 1 and 2 were initially docked computational-
ly into the ATP-binding site of the crystal structure of apo
human Chk1 (PDB28 entry 1IA829). This suggested con-
vincing binding modes for compounds 1 and 2, in view
of their sensible interactions with the protein, which could
explain the affinity difference between 1 and 2 (Section
2.4). The root mean square deviation (RMSD) between
the predicted and X-ray coordinates of compounds 1
Compounds 1 and 2 were prepared by previously pub-
lished literature routes24,27 (see Methods and Schemes
1–3).
2.1. Binding affinities of selected indazole compounds and
DBH
˚
and 2 was ꢁ0.4 A (see Methods and Supplementary
Information). Although the X-ray structure of the
Chk1-compound 1 complex is of medium resolution
(Table 2), the crystallographic electron density maps
clearly support the conformation of the methoxyphenol
of compound 1 as shown in Figure 2, as well as the orien-
tation of the protein backbone for residues Ser147 and
Asp148. These peptide units adopt the same conforma-
tion in Chk1 X-ray structures of better resolutions.18,25,29
Compounds 1 and 2 inhibit human Chk1 in an ATP-com-
petitive manner, with Ki values of 0.026 and 8.58 lM,
respectively (Table 1). Each Ki was derived from several
independent measurements confirming that compound 1
is a potent Chk1 inhibitor, although its affinity appears
to be six times less than previously reported.11,24 The affin-
ity for Chk1 of the core of the indazole series (2), present-

1794
N. Foloppe et al. / Bioorg. Med. Chem. 14 (2006) 1792–1804
Tyr86
Glu85
Val68
Cys87
Asn59
Glu55
Figure 1. Crystallographic binding mode of compound 2 in the ATP-binding site of Chk1. For clarity, only selected residues and hydrogen atoms are
shown. Hydrogen bonds between the compound and the kinase backbone are depicted with green dotted lines. The three red spheres represent water
molecules. The shown protein residues are labeled.
A
Glu85
Tyr86
Val68
Cys87
Asn59
Glu55
B
Glu85
Leu84
Tyr86
Asn59
Glu55
Cys87
Asp148
Ser147
Figure 2. (A and B) Different views of the crystallographic binding mode of compound 1 in the ATP-binding site of Chk1. For clarity, only selected
residues and hydrogen atoms are shown. Conventional hydrogen bonds between the compound and the protein are depicted with green dotted lines,
while C–Hꢀ ꢀ ꢀO and N–Hꢀ ꢀ ꢀp-aromatic electron interactions are traced in magenta. For the N–Hꢀ ꢀ ꢀp-aromatic interaction, the phenyl ring is
represented by its centroid. The shown protein residues are labelled.
The indazole core of 1 and 2 largely overlaps with the vol-
ume typically occupied by the adenine base of ATP when
complexed to kinases, with the indazole hydrogen-bond-
ed to the highly conserved kinase backbone motif at the
hinge between the two lobes of the kinase domain
(Fig. 1). The hydrogen-bond interactions between inda-

N. Foloppe et al. / Bioorg. Med. Chem. 14 (2006) 1792–1804
1795
OMOM
OMe
Mts
H
Mts
N
H
N
NH2
N
I
N
I
a
b
N
c
N
N
N
4
6
5
OMOM
OMOM
OMOM
OMe
H
H
H
N
d
N
f
e
N
N
OMe
N
N
OMe
OHC
8
9
I
7
OH
OMOM
OMe
H
N
H
N
h
N
OMe
g
N
H
H
N
N
N
1
N
10
Scheme 1. Reagents and conditions: (a) Concd HCl (aq), NaNO2, KI, H2O, 0–50 ꢁC, 3 h, (39%); (b) NaOtBu, MtsCl, THF, 0–23 ꢁC, 2.5 h, (92%); (c)
(12), PdCl2(PPh3)2, Et3N, K2CO3, dioxane/H2O, 90 ꢁC, 2 h; (d) NaOH, MeOH, 45 ꢁC, 1 h, (27%); (e) I2, NaOH, dioxane, rt, (42%); (f) PhLi, s-BuLi,
DMF, THF, ꢂ78 ꢁC to rt, 18 h; (g) 1,2-diaminobenzene, CuAc2, AcOH, MeOH/H2O, rt, 19 h (21%); (h) TFA/THF, H2O (28%).
OH
OMOM
OMOM
OMe
OMe
OMe
a
b
Br
B
Br
HO
OH
11
12
Scheme 2. Reagents and conditions: (a) methoxymethyl chloride, KOH(s), H2O, Bu4NBr, DCM, 0 ꢁC to rt, 64 h, (93%); (b) n-BuLi, B(OMe)3, THF,
ꢂ78 ꢁC to rt, 20 h, (68%).
O
N
OtBu
N
N
b
a
N
N
N
O
HO2C
O
13
14
15
N
O
N
OtBu
N
N
N
c
d
N
OHC
16
2
Scheme 3. Reagents and conditions: (a) LiAlH4, 0 ꢁC, THF, 50%; (b) (BOC)2O, THF, 60%; (c) IBX, THF, n, 70%; (d) o-phenylenediamine, sulfur,
DMF, 95 ꢁC, 55%.

1796
N. Foloppe et al. / Bioorg. Med. Chem. 14 (2006) 1792–1804
Table 2. Crystallographic data collection and refinement statistics for Chk1 in complex with compounds 1–3
Structure
Chk1 + compound
1
2
3 (DBH)
Data collection statistics
˚
Resolution (A)
Measured reflections
2.60
117777
9736
2.35
2.05
57350
14217
95.1/82.9
2/1
117609
19586
98.1/99.8
4/4
Unique reflections
Completeness: overall/in hrb* (%)
Average multiplicity/in hrb
Mean I/rI: Overall/in hrb
Rmerge: Overall/in hrb (%)
96.7/87.1
4/2
12.7/2.6
8.8/38.8
9.6/3.7
8.9/35.2
14.7/2.4
7.8/45.1
Refinement statistics
Rfree (%)
Rcryst (%)
29.0
22.2
27.5
19.3
26.2
18.2
Rms deviations:
˚
Bonds (A)
0.025
2.289
3.137
2C3L
0.018
1.645
2.424
2C3K
0.022
2.100
3.095
2C3J
Angles (ꢁ)
2
˚
B Factor (A )
PDB Code
Rfree is the R factor calculated using 5% of the reflection data chosen randomly and omitted from the refinement process, whereas Rcryst is calculated
with the remaining data used in the refinement. Rms bond lengths and angles are the deviations from ideal values; the rms deviation in B factors is
calculated between covalently bonded atoms.
* hrb, highest resolution bin.
Val68
Asn59
Glu85
Tyr86
Cys87
Glu55
Figure 3. Superposition of the crystal structures of Chk1 in complex with compound 1 (red), DBH (blue), staurosporine (green; PDB entry 1NVR18
and compound 17 (yellow; PDB entry 2BRB25). Compound 17 is shown in Scheme 4. Only selected protein residues are shown and labelled.
)
zole and protein define the tautomer of the bound form of
the indazole. In the X-ray structure of compound 2, three
water molecules were located in the buried pocket lined
by Glu55, Asn59 and Val68. These water molecules are
displaced by the methoxyphenol of compound 1 when
bound to Chk1 (Fig. 2). This pocket filled by the meth-
oxyphenol of 1 is not occupied by other Chk1 ligands
for which X-ray binding modes are available (Fig. 3).
Therefore, the elucidation of the binding mode of com-
pound 1 provides new structural insights relevant to li-
gand design against Chk1. Transferring elements of the
binding mode of compound 1 to other chemical series
of Chk1 inhibitors, by filling the buried pocket, could
confer increased affinity for Chk1 to these series.
A recently reported inhibitor (18, Scheme 4), with an
affinity of 0.450 lM for Chk1,19 contains a methoxyphe-
nyl analogous to the methoxyphenol of 1. This prompt-
ed us to test whether 18 could be docked in the Chk1
ATP site, with its methoxyphenyl group in a position
similar to that of its counterpart in 1. Two sensible
docking modes of 18 to Chk1 were obtained (see Sup-
porting Information), with the central pyrimidine
hydrogen-bonding the conserved kinase backbone motif
as expected, and reasonable interactions between the
phenyl rings and the protein. The geometric constraints
associated with these docking modes, however, do not
allow the methoxyphenyl group to access the pocket dis-
cussed in this work. Therefore, we predict that the chem-

N. Foloppe et al. / Bioorg. Med. Chem. 14 (2006) 1792–1804
1797
Br
N
OH
HN
N
NH
HN
N
N
O
CO2H
OMe
17
18
OH
O
H
H
N
N
NH
N
OMe
N
H
N
N
20
19
Scheme 4.
˚
ical series represented by 18 is another chemotype which
does not utilize this pocket.
tion was less accurate (RMSD ꢁ1.2 A) than with the
indazole compounds, partly because of flexibility in
the seven-membered ring and the water-mediated hydro-
gen bond between DBH and the protein backbone in the
X-ray structure. Given the pseudo-symmetry of its
hydrogen-bond donating and accepting groups, the
fused ring system of DBH could also dock to the kinase
backbone in another orientation, by flipping around the
2.3. Prediction and confirmation of the binding mode of
DBH
The overall binding mode of DBH was also predicted by
docking to the ATP binding site of the crystal structure
of apo human Chk1 and confirmed by X-ray crystallog-
raphy (Fig. 4). Although essentially correct, this predic-
N
tween the imidazolone ring of DBH and Lys38, Asn135
Cys87–Hꢀ ꢀ ꢀO hydrogen bond. The hydrogen bonds be-
Glu85
Lys38
Tyr86
Asp148
Cys87
Asn135
Figure 4. X-ray binding mode of debromohymenialdisine (DBH) to Chk1. For clarity, only selected residues and hydrogen atoms are shown.
Hydrogen bonds between the DBH and the protein or water molecules are depicted with green dotted lines. The red spheres represent water
molecules. Selected protein residues are labelled. The black arrow points to the carbon which could be derivatised towards the buried pocket
discussed in this work.

1798
N. Foloppe et al. / Bioorg. Med. Chem. 14 (2006) 1792–1804
and Asp148 select between the two possible orientations
of the fused ring system (Fig. 4).
C–H/p interactions, which are known to be more
favourable than the associated van der Waals dispersion
contribution.45–47
The relatively high number of hydrogen bonds between
DBH and Chk1 is associated with a rather modest binding
affinity (Table 1). This is compatible with some evidence
indicating that protein–ligand hydrogen-bonding may fa-
vour protein–ligand association only weakly.25,30–32
DBH does not fill the pocket occupied by the methoxy-
phenol of compound 1. This suggests that opportunities
may exist to enhance DBH affinity against Chk1 by reach-
ing into the pocket of interest. From a purely structural
point of view it may be possible to derivatise from the sev-
en-membered ring which offers at least one C–H vector
pointing towards the pocket (Figs. 3 and 4). In that con-
text the exact conformation of the fused ring of DBH,
when bound to Chk1, is important. The crystallographic
electron density does support the puckering of the seven-
membered ring shown in Figure 4.
The excellent structural complementarity between the
methoxyphenol moiety and its binding pocket, where all
elements of the protein form potentially favourable inter-
actions with the ligand, is consistent with the associated
large difference in Chk1 affinity between 1 and 2. Other
factors contributing to this affinity gain probably include
desolvation of the ligand phenyl and methyl groups, and
possibly an entropy gain upon displacement of the water
molecules48 which otherwise fill this pocket (Fig. 1). Using
relatively apolar ligand functionalities to interact with
buried protein polar groups is likely to be favourable from
a ligand desolvation point of view, and may be a strategy
of general interest to target this pocket.
A notable consequence of the interactions described
above is that an aromatic, rather than an aliphatic, ring
may be required to fill the pocket lined by the side chain
of Leu84 and the backbone of Ser147. Indeed, only the
p electrons of an aromatic moiety would accommodate
a hydrophobic contact on one side while being faced by
a hydrogen bond donor on the other side. Also, basic ste-
reochemistry suggests that an aliphatic ring may be too
thick to fit sterically in the pocket. Pyrazole rings and ana-
logues with both an accepting and a donating nitrogen
may be well suited to hydrogen-bond the side chain of
Asn59.
2.4. Non-conventional C–Hꢀ ꢀ ꢀO and N–Hꢀ ꢀ ꢀp
interactions
The details of the interactions of the methoxyphenol
with the Chk1-binding pocket (Fig. 2) are of great inter-
est because they are associated with a dramatic increase
in affinity for Chk1, as well as an influence on the bind-
ing selectivity across different kinases (Table 1 and Sec-
tion 2.6). The methoxyphenol fits very snugly in its
binding pocket, with its polar groups forming extensive
hydrogen bond interactions with Glu55, Asn59, and
Asp148. All the hydrogen-bonding capabilities of the li-
gand hydroxyl are fulfilled by donating to the carboxyl-
ate of Glu55 and accepting from the side-chain amide of
Asn59 and the backbone amide of Asp148. Asn59 side
chain donates a proton to both the hydroxyl and meth-
oxy oxygens in a bifurcated hydrogen bond. The methyl
group of compound 1 has non-polar contacts with the
side chains of Val68 and Leu84. This methyl group,
polarised by the oxygen atom to which it is attached,
may form favourable weak polar C–Hꢀ ꢀ ꢀO interactions,
with some hydrogen-bonding character, with the car-
bonyl oxygens of Val68 and the side chain of Asn59.
Although this type of interaction has been well docu-
mented in the arena of small organic molecules for some
time,33–37 the recognition of its interest in protein–ligand
recognition for drug design is more recent 38–40 and con-
troversial. It is striking that the methyl group of 1 con-
tacts two carbonyl oxygens, which are otherwise
hydrated (Fig. 1).
2.5. Consistency between SAR and binding mode for the
indazole series
Elements of SAR around the indazole series represented
by compound 1 were summarized previously (Table 5 in
Li and Zhu11). The majority of the analogues of 1 had
affinities for Chk1 in the low nanomolar range. In all
these potent analogues, the phenol ring is derivatized
with relatively small hydrophobic substituents which
fit sterically (not shown) in the pocket discussed in this
work, consistent with the associated high affinities and
the binding mode in Figure 2. The small affinity differ-
ences between these compounds are, however, too subtle
to be explained only based on simple stereochemical
criteria.
Compound 19 (Scheme 4) is an intriguing exception, be-
cause it has a much lower reported affinity for Chk1
(1.1 lM), despite enough room in the protein pocket
to accommodate its pyridone ring. The most stable form
of 19 bound to Chk1 is unclear, due to the tautomeric
equilibrium between pyridone and pyridol. Although
2-pyridone is intrinsically marginally more stable than
2-hydroxypyridine in gas phase,49 the pyridone-pyridol
equilibrium is strongly affected by the environment
and substituents.50 Modelling of the bound pyridone
tautomer shows that its amide nitrogen would be buried
without hydrogen-bonding partner, suggesting that des-
olvation of the pyridone would strongly weaken its
binding to Chk1. Compound 19 would fit better in the
pocket as a pyridol because the pyridol nitrogen could
hydrogen-bond to Lys38 (not shown). However, this is
Arguably even more unusual is the phenyl ring of the
methoxyphenol being sandwiched between the apolar
side chain of Leu84 and the polar amide N–H of
Ser147, with this N–H pointing toward the phenyl p-ar-
omatic electrons. The distance between the nitrogen and
˚
the ring centroid is 3.5 A, with an N–Hꢀ ꢀ ꢀcentroid angle
of 141.5ꢁ. This provides an unambiguous example where
a ligand phenyl ring acts as a hydrogen bond acceptor.
Although this type of interaction has already been
recognised,41–44 very few examples have been reported
mediating protein–ligand complexes. The interaction be-
tween Leu84 and the methoxyphenol might also involve

N. Foloppe et al. / Bioorg. Med. Chem. 14 (2006) 1792–1804
1799
still compatible with the weak affinity of 19 if it exists
primarily as a pyridone in aqueous solution.
transition.58–60 Therefore, selectivity data versus CDK1
are of particular interest. Unfortunately, a crystal struc-
ture of CDK1 is, to our knowledge, not available. There-
fore, an homology model of human CDK1 was built
based on a crystal structure of human cyclin dependent
kinase 2 (CDK2) bound to cyclin A. This utilizes the
66.2% sequence identity between CDK1 and CDK2.
There is no notable sequence difference between CDK1
and CDK2 for the residues lining the pocket of interest,
and the homology model of CDK1 is very similar to its
CDK2 template in this region. The selectivity data are
discussed assuming that compounds 1–3 adopt similar
binding modes with different kinases.
The other notable loss of affinity was observed for the
analogue of 1 where the benzimidazole was replaced
by a phenyl11 (compound 20, Scheme 4), which can be
explained by the binding mode of 1, for instance because
of the lack of hydrogen bond between a phenyl and the
kinase backbone.
To recapitulate, the compounds reported in Table 5
from Li and Zhu11 can be grouped in two sets: (i) those
with high affinity (Ki < 0.1 lM) which are all derivatised
with small hydrophobic substituents which fit in the
pocket of interest and (ii) compounds 19 and 20 (Scheme
4) which have much lower affinities, consistent with the
binding mode of 1 and the above arguments. Therefore,
the major trends in the SAR around compound 1 are
consistent with the binding mode of this compound. De-
tailed calculations, such as free energy perturbation,51,52
may be able to provide additional insights into the sub-
tle affinity differences between the most potent com-
pounds in this series, but this is beyond the scope of
this work. Also, one would need to consider possible
uncertainties in the experimental affinities, given that
the present work finds that the affinity of 1 is six times
less than previously reported.11,24
The lack of affinity of 1 and 2 for Akt (protein kinase B)
and PKA can be understood in view of the steric clash
between the benzimidazole of 1 and 2 and Phe439
(Akt) or Phe327 (PKA) (Fig. 5). With other kinases
compound 2 tends to exhibit affinities closer to that
for Chk1, although showing elements of selectivity. In
particular, compound 2 is 7-fold more potent for
CDK1 than Chk1. This difference is lost with the more
potent compound 1 (Table 1), which implies that larger
potency gains are made for Chk1 than for CDK1 by
binding of the methoxyphenol to its pocket. The notion
that the interactions of the methoxyphenol with its
pocket are less favourable in CDK1 than in Chk1 is con-
sistent with the replacement of Leu84/Asn59 in Chk1 by
Phe80/Leu55 in CDK1 (Fig. 5). The replacement of
Asn59 by an apolar side chain leads to a loss of two con-
ventional hydrogen bonds between the protein and com-
pound 1. In addition, the replacement of Leu84 (Chk1)
by the bulkier Phe80 (CDK1) may lead to steric hin-
drance between compound 1 and CDK1, as suggested
by a comparison of the similar Chk1/CDK2 pair. This
is not sufficient, however, to obtain the desired selectiv-
ity between Chk1 and CDK1 with 1, given that it exhib-
its the same IC50 for both proteins. This result puts the
indazole series of Chk1 inhibitors in a new light, sug-
gesting deficiencies in its therapeutic selectivity profile.
In addition, this chemical series can only be a blunt
instrument if used as a pharmacological tool to study
the mechanisms and regulation of the G2/M checkpoint.
The present work also highlights that obtaining an
X-ray structure for CDK1 would help in the design of
therapeutically relevant Chk1 inhibitors.
2.6. Influence of the pocket on the selectivity of binding
Different kinases are involved in a variety of cell signal-
ling pathways53 but X-ray structures combined with
sequence alignments have shown that many features of
the ATP-binding pocket tend to be conserved across
kinases.54–56 Therefore, selectivity of inhibition for a
particular kinase is a well-known issue, and we investi-
gated this question for Chk1 by comparing the IC50Õs
of compounds 1–3 for several kinases (Table 1). One
of the criteria to select the kinases for which selectivity
data were obtained was that an X-ray structure should
be available for their ATP-binding site, with the excep-
tion of cyclin dependent kinase 1 (CDK1, also called
Cdc2). This should allow insights regarding the influence
on selectivity of the Chk1 pocket filled by the methoxy-
phenol of 1, by comparison to its counterpart in other
kinase structures. The specific role of this pocket for
selectivity may be dissected, to some degree, by compar-
ing the IC50Õs of compounds 1 and 2. This can only be a
qualitative analysis because there is more to selectivity
than simple steric interactions between a ligand and
the residues that it contacts directly. For instance overall
protein dynamics, with elements of induced fit, may con-
tribute to selectivity. It remains that, given the dramatic
potency enhancement achieved by filling the pocket of
interest in Chk1, it is of direct practical interest to esti-
mate the potential of this pocket for selectivity gains
when targeting Chk1.
That the Chk1 pocket lined by Asn59 and Leu84 plays a
role regarding selectivity is confirmed by the IC50Õs of 1
and 2 with CDK2, GSK3b and PDK1. These proteins
have an apolar side chain in place of Asn59 (Chk1), con-
sistent with much lower affinities of 1 for these proteins.
Taken together, these results strongly suggest that the
pocket discussed in this work is of great practical impor-
tance to design ligands with a high degree of selectivity
for Chk1. This may help to convert DBH and other
compounds into a Chk1 selective inhibitor.
In particular, the rationale for Chk1 inhibition in the con-
text of oncology is to abrogate the ability of tumour cells
to arrest in G2 in response to DNA damage.11,12,57 This is
incompatible with a concomitant inhibition of Chk1 and
CDK1, because active CDK1 is necessary for the G2 to M
3. Conclusions
This work reports the crystal structures of DBH and key
representatives of an indazole series of Chk1 inhibitors

1800
N. Foloppe et al. / Bioorg. Med. Chem. 14 (2006) 1792–1804
Chk1: Leu84
Akt: Met229
Chk1: Asn59
CDK1: Phe80
Akt: Leu204
CDK2: Phe80
GSK3β: Leu132
PDK1: Leu159
PKA: Met120
CDK1: Leu55
CDK2: Leu55
GSK3β: Met101
PDK1: Met134
PKA: Leu95
Akt: Phe439
PKA: Phe327
Figure 5. Superposition of Chk1 structure (red) with those of kinases Akt (cyan), CDK1 (magenta), CDK2 (green), GSK3b (light grey), PDK1 (dark
blue) and PKA (orange). X-ray structures of Chk1, Akt, CDK2, GSK3b, PDK1 and PKA are from PDB entries 1IA8,29 1O6L,82 1OI9,67 1Q5K,83
1H1W84 and 1ATP,85 respectively. For CDK1, a homology model based on CDK2 is used. The conserved hinge backbone, to which compound 1 is
hydrogen bonded (green dotted lines) in Chk1, is closely aligned across all proteins. Only selected protein side chains in the vicinity of compound 1
are shown and labelled. These side chains are those structurally aligned with Asn59 and Leu84 of Chk1, which are in direct contact with the
methoxyphenol of compound 1. Phe439 (Akt) and Phe327 (PKA) are also shown because they have no structural equivalent in the other kinases
considered here.
complexed with this human kinase ATP-binding site.
The binding modes of these ligands, and their implica-
tions for ligand design, had been predicted accurately
by computer modelling well before the crystal structures
were obtained, providing an example of a helpful bona
fide prediction verified by experiment. This was supple-
mented by careful measurements of the associated bind-
ing affinities, and the actual affinity of 1 for Chk1 is six
times less than previously reported.11,24 Compound 1,
however, remains one of the highest affinity ligands for
Chk1, and comparing the binding modes and affinities
of 1 and 2 led to new and helpful insights.
cancer probably requires ligands which discriminate be-
tween Chk1 and CDK1.
It is striking that the pocket of interest is not filled by the
ligand in any of the other Chk1-ligand X-ray structures
reported so far, and this covers the majority of the cur-
rent chemical series of Chk1 inhibitors. This means that
there may be scope for further elaborations of some of
these series, including DBH which has a relatively low
molecular weight. The X-ray structure of DBH bound
to Chk1 provides a firm structural basis for derivatisa-
tion of DBH towards the pocket highlighted in the pres-
ent work.
The binding mode of 1 revealed a buried pocket which
can be used to gain dramatic affinity increases against
Chk1. Interestingly, it is compatible with a number of
non-conventional interactions between 1 and the protein
pocket. In particular, this provides a rare non-ambigu-
ous example of an N–Hꢀ ꢀ ꢀp-aromatic interaction with
hydrogen bond character between ligand and protein,
which is probably important to retain high binding affin-
ity while filling this pocket. In addition, C–Hꢀ ꢀ ꢀO inter-
actions were observed, which may also present a weak
hydrogen bond character.
4. Computational and experimental methods
4.1. Molecular modelling
Docking was performed with the program rDock which
is an extension of the program RiboDock,61 using an
empirical scoring function calibrated based on
protein–ligand complexes.62,63 The Monte Carlo/simu-
lated annealing protocol initially used in RiboDock
was replaced by a steady state Genetic Algorithm
(GA) to improve the efficiency of the docking search.
Ligand docking poses were represented using a conven-
tional chromosome representation of translation, rota-
tion and rotatable bond dihedral angles. A single GA
population was used, of size proportional to the number
of ligand rotatable bonds, with a mutation:crossover ra-
tio of 3:2. The overall orientation and internal confor-
mation of the compounds were searched with the GA,
while the protein was kept fixed.
The comparison of the residues lining the Chk1 pocket
of interest to their counterpart in a few other kinases
suggests that targeting this pocket should strongly influ-
ence the binding selectivity. This was born out by com-
paring the affinities of 1 and 2 for these kinases. This
analysis suggests that, although the affinity of 1 is simi-
lar for Chk1 and CDK1, its interactions with the pocket
moderately favour binding to Chk1. This is notable be-
cause the biological rational for targeting Chk1 against

N. Foloppe et al. / Bioorg. Med. Chem. 14 (2006) 1792–1804
1801
Docking used the ATP-binding site in the crystal struc-
ture of apo human Chk1 (PDB28 entry 1IA829). Water
molecules were removed from the coordinates and polar
hydrogens added to the protein using the CHARMm
force-field.64 To specify the docking volume, the ATP-
bound crystal structure of protein kinase A (PDB entry
1ATP) was superimposed onto that of Chk1. There was
unambiguous superimposition of the ATP-binding sites
and the docking volume in Chk1 was defined as the
were performed in a similar manner. For the GSK3b as-
say, GSK3b (Upstate) was used at 4 nM with 2.5 lg phos-
phor-glycogen synthase peptide-2 (Upstate) at a final
ATP concentration of 100 lM. The reaction was incubat-
ed for 30 min at rt. For the PDK1 assay, PDK1 (Upstate)
was used at 14 nM with 5 lM PDKtide (Zinsser Analyti-
cal; sequence YRRAAVPPSPSSLSRHSSPHQ(P)SE-
DEEE) at a final ATP concentration of 10 lM. The
reaction was incubated at 30 ꢁC for 30 min. For the
Akt1 assay, 3 nM Akt1 (Upstate) was used with Histone
H1 as substrate with a final ATP concentration of
200 lM. The reaction was incubated at 30 ꢁC for
40 min. For the CDK2/Cyclin A assay, CDK2/Cyclin A
obtained in-house was used at 9.5 nM with CDKtide as
a substrate at a final ATP concentration of 100 lM. The
reaction was incubated at 30 ꢁC for 40 min.
˚
space within 8 A of the ATP molecule. Before docking,
the three-dimensional structure of the compounds was
built with the software MOE65 and energy minimized
with the Merck molecular force-field.66 Several low ener-
gy conformations were generated for the flexible seven-
membered ring of DBH and all of them were docked.
Each compound was subjected to 200 docking runs,
and the output docking modes were analyzed by visual
inspection in conjunction with the docking scores. The
best scoring docking modes were selected as working
hypothesis for compounds 1–3, because they were struc-
turally sensible and compatible with the few elements of
SAR available.11 In addition, there was relatively little
variation in the docking output and no alternative plau-
sible docking mode was generated for compounds 1 and
2. The RMSD on non-hydrogen atoms between mod-
elled and X-ray binding modes for the ligands were cal-
culated after best fit of the protein structures on all
atoms of the backbone of the conserved kinase motif
(Glu85-Tyr86-Cys87), using MOE. Hydrogen atoms
shown in Figures 1,2 and 4 were added to the X-ray
coordinates with the software MOE and the Merck
molecular force-field. Compound 18 was docked to the
same protein structure used to dock compounds 1–3.
Debromohymenialdisine (DBH) from Stylotella
aurantium was purchased from Calbiochem (Catalogue
No. 252010), with a purity of 96.8% by HPLC.
4.3. Chk1 Ki determinations
Chk1 kinase assays were performed as described previ-
ously25 in a final reaction volume of 50 ll. Chk1 was
used at 18 nM and the reaction was incubated at 30 ꢁC
for 30 min. For Ki determinations a matrix of inhibitor
and substrate concentrations were tested. Inhibitor con-
centrations were tested from four times IC50 with 0.6-
fold dilutions over 7 concentrations, as well as a final
reaction with no inhibitor present. Substrate concentra-
tions ranged from 0.5 · Km to 8 · Km.
Due to the inherent errors that occur with traditional
methods of analysis of linearised versions of the Michae-
lis–Menten equation (Lineweaver–Burk, Eadie–Hof-
stee), more direct methods relying on nonlinear
regression analysis applied to the whole data set simulta-
neously are known to be more accurate.74 We have used
such a non-linear regression analysis for Ki determina-
tions, following a 4-step process:
Modeling of the three-dimensional structure of human
CDK1 was carried out by comparative homology mod-
elling, using as template a crystal structure of CDK2 in
complex with cyclin A (PDB entry 1OI9,67 resolution
˚
2.10 A). This template was selected because the assays
involving CDK1 were performed with an homologous
CDK1/cyclin B complex. The sequence of human
CDK1 was retrieved from the SWISS-PROT data
bank68 (Accession No. P06493), and shares 66.2% iden-
tity with that of CDK2 (shortest sequence used for total
length), which provides a sound basis for comparative
modelling.69,70 This allowed an unambiguous sequence
alignment for the region encompassing the binding site
of interest, and 10 structural models were generated with
previously described algorithms,71,72 as implemented in
MOE.65 The default parameters of the program were
kept. The 10 CDK1 models did not vary significantly
in the region of interest and the best scoring model (as
reported by MOE) was used for comparison with the
other kinases (Fig. 5).
(i) An estimate of the Vmax and Km of Chk1 was
obtained in the absence of inhibitor, using the Michae-
lis–Menten steady state equation (model 253; XLFit
4,75 ID Business Solutions, Guildford, UK).
(ii) For each inhibitor, an initial suggested mode of
action was inferred from analysis of double reciprocal
Lineweaver–Burk plots. A competitive mode of action
was suggested in all cases, however, a non-competitive
mechanism was not ruled out in subsequent steps.
(iii) Then the experimental data were fitted alternatively
to a competitive and a non-competitive equation using a
3D non-linear regression (XLFit 4, IDBS). For the non-
competitive model, Eq. 4 from Kakkar et al.86 was used.
For the competitive model we used the Michaelis–Men-
ten substrate plus inhibitor equation (Model 12; Life
Science Workshop Toolbox, MDL Information Systems
Inc., USA)
Pictures of three-dimensional molecular structures were
prepared with PyMOL.73
4.2. Kinase assays for IC50 determinations
y ¼ V max=ð1 þ ðKm1=x1Þ ꢃ ð1 þ x2=KiÞÞ;
where x1 = [ATP]; x2 = [Inhibitor] and y = CPMcorrected
Enzyme inhibition assays to determine IC50Õs with Chk1,
CDK1/Cyclin B and PKA were performed as described
previously,25 and equivalent assays with other kinases
.
Km1 is the Michaelis constant for the enzyme–substrate

1802
N. Foloppe et al. / Bioorg. Med. Chem. 14 (2006) 1792–1804
pair, and Ki is the dissociation constant for the
enzyme–inhibitor pair. By definition, competitive inhibi-
tion can be surmounted by increasing the concentration
of substrate. Consequently, a competitive inhibitor does
not alter Vmax. The data set was processed against both
models using a free-fit initially and the Km and Vmax in
the absence of inhibitor were obtained. In all cases,
the Vmax obtained from the non-linear regression was
close to the Vmax calculated in step i). Three-dimensional
plots were produced with both the competitive and non-
competitive models with x-axis = [ATP], y-axis = [I]
butyllithium followed by trimethyl borate (Scheme 2).
Biaryl 6 was deprotected in the presence of methanolic
sodium hydroxide to give the free indazole 7, which
upon iodination in the presence of sodium hydroxide
and iodine gave the 3-iodoindazole 8. Reaction of 8 with
a 1:1 mixture of phenyllithium and sec-butyllithium fol-
lowed by quenching with DMF at ꢂ78 ꢁC gave the cor-
responding aldehyde 9. The aldehyde was condensed
with diaminobenzene in a mixture of methanol and
water, yielding 10. This was subsequently deprotected
upon treatment with TFA in THF and water, giving
compound 1. Mass spectrometry and NMR character-
isation were identical to literature:24 MS EI m/e 357
(M++H); 1NMR (MeOD) d: 3.95 (3H, s); 6.85–6.87
(1H, m); 7.15–7.20 (1H, m); 7.25–7.35 (4H, m); 7.55–
7.60 (1H, m), 7.65 (1H, s); 8.45–8.50 (1H, m).
and z-axis = CPMcorrected
.
(iv) To confirm which model fitted the data best, several
statistical parameters were checked: the normalized v2,
F-Test and T-Test. In view of these statistical tests, each
compound fitted preferentially to a competitive mode of
action model, consistent with the X-ray structures. Ki
values were then obtained from the relevant model.
Compound 2 was prepared as follows. The commercially
available acid 13 was reduced with lithium aluminium hy-
dride giving the primary alcohol 14 in 50% yield. This was
then N-protected with BOC anhydride and subsequently
oxidized using the IBX system81 in refluxing THF, yield-
ing the 3-formyl indazole 16. This was then reacted with o-
phenylenediamine and molecular sulfur in DMF at 95 ꢁC
overnight giving the fully deprotected indazole-benzimid-
azole 2. Mass spectrometry and NMR characterisation
were identical to literature27: MS EI m/e 235 (M++H);
1NMR (MeOD) d: 7.21–7.25 (3H, m); 7.35–7.47 (1H,
m); 7.45–7.65 (3H, m), 8.35–8.40 (1H, m).
4.4. Crystallisation and three-dimensional structure
determination
Chk1 1-289 was expressed, purified and crystallised as
previously described.25
Protein–ligand complex data sets were collected at cryo-
temperature. Reservoir solution with 20% glycerol add-
ed was used as the cryoprotectant. Data were collected
either on our in-house detector (RU-H3R rotating an-
ode generator with R-Axis IV++ image plate detector)
or on beamline ID29 at the ESRF (Grenoble). The data
sets were subsequently processed using DENZO,76 in
space group P21 with unit cell dimensions isomorphous
to those of the previously solved apo Chk1 structure.29
The ligand bound structures were solved by molecular
replacement using the apo Chk1 model coordinates
(PDB code: 1AI8). Twenty cycles of rigid-body and re-
strained refinement were carried out followed by model
building using the molecular graphics program O.77
Crystallographic water molecules were added by cycling
REFMAC5 with ARP/wARP.78 The progress of the
refinement was assessed using Rfree and the conventional
R factor. Once refinement of the structure was complet-
ed, the structures were validated using PROCHECK79
and various programs from the CCP4i package.80 Full
data collection and refinement statistics are presented
in Table 2.
Acknowledgments
We thank the support staff of beamline ID29 at the
European Synchrotron Radiation Source (Grenoble,
France) for help with data collection. We thank Jenifer
Borgononi, Claire Nunns, Alan Robertson, Allan Surg-
enor, and Lee Walmsey for assistance with the present
work. We also thank I-Jen Chen and Christine Richard-
son for helpful discussions.
Supplementary data
Supplementary data associated with this article can be
found, in the online version, at doi:10.1016/
j.bmc.2005.10.022.
4.5. Synthetic chemistry
References and notes
Full experimental details for the preparation of com-
pound 1 were described previously,24 and only a summa-
ry of the synthetic route is outlined in Schemes 1, 2 and
the following. 6-Aminoindazole was dissolved in water
and treated with concentrated hydrochloric acid and
sodium nitrite at 0 ꢁC to give the corresponding diazoni-
um salt, which was immediately converted to 4 upon
treatment with potassium iodide. N-protection with
mesitylenesulfonyl chloride and tert-sodium butoxide
in THF gave sulfonamide 5. This was then coupled to
boronic acid 12 using standard Suzuki coupling condi-
tions giving the biaryl 6. The boronic acid 12 was pre-
pared by treatment of protected bromide 11 with
1. Sanchez, Y.; Wong, C.; Thoma, R. S.; Richman, R.; Wu,
Z.; Piwnica-Worms, H.; Elledge, S. J. Science 1977, 277,
1497–1501.
2. Melo, J.; Toczyski, D. Curr. Opin. Cell Biol. 2002, 14, 237–
245.
3. Rhind, N.; Russell, P. J. Cell Sci. 2000, 113, 3889–3896.
4. Chen, Z.; Xiao, Z.; Chen, J.; Ng, S.-C.; Sowin, T.; Sham,
H.; Rosenberg, S.; Fesik, S.; Zhang, H. Mol. Cancer
Therap. 2003, 2, 543–548.
5. Liu, Q.; Guntuku, S.; Cui, X. S.; Matsuoka, S.; Cortez,
D.; Tamai, K.; Luo, G.; Carattini-Rivera, S.; De Mayo,
F.; Bradley, A.; Donehower, L. A.; Elledge, S. J. Genes
Dev. 2000, 14, 2448–2459.

N. Foloppe et al. / Bioorg. Med. Chem. 14 (2006) 1792–1804
1803
6. Zachos, G.; Rainey, M. D.; Gillespie, D. A. F. EMBO J.
2003, 22, 713–723.
7. Luo, Y.; Rockow-Magone, S. K.; Kroeger, P. E.; Frost,
L.; Chen, Z.; Han, E. K.-H.; Ng, S.-C.; Simmer, R. L.;
Giranda, V. L. Neoplasia 2001, 3, 411–419.
8. Lopez-Girona, A.; Tanaka, K.; Chen, X.-B.; Baber, B. A.;
McGowan, C. H.; Russell, P. Proc. Natl. Acad. Sci. U.S.A.
2001, 98, 11289–11294.
9. Curman, D.; Cinel, B.; Williams, D. E.; Rundle, N.;
Block, W. D.; Goodarzi, A. A.; Hutchins, J. R.; Clarke, P.
R.; Zhou, B.-B.; Lees-Miller, S. P.; Andersen, R. J.;
Roberge, M. J. Biol. Chem. 2001, 21, 17914–17919.
10. Furnari, B.; Rhind, N.; Russell, P. Science 1997, 277,
1495–1497.
29. Chen, P.; Luo, C.; Deng, Y.; Ryan, K.; Register, J.;
Margosiak, S.; Tempczyk-Russell, A.; Nguyen, B.; Myers,
P.; Lundgren, K.; Kan, C.-C.; OÕConnor, P. M. Cell 2000,
100, 681–692.
30. Fersht, A. Trends Biochem. Sci. 1987, 12, 301–304.
31. Davis, A. M.; Teague, S. J. Angew. Chem. Int. Ed. 1999,
38, 736–749.
32. Hunenberger, P. H.; Helms, V.; Narayana, N.; Taylor, S.
¨
S.; McCammon, J. A. Biochemistry 1999, 38, 2358–2366.
33. Sutor, D. J. Nature 1962, 195, 68–69.
34. Taylor, R.; Kennard, O. J. Am. Chem. Soc. 1982, 104,
5063–5070.
35. Gu, Y.; Kar, T.; Scheiner, S. J. Am. Chem. Soc. 1999, 121,
9411–9422.
11. Li, Q.; Zhu, G.-D. Curr. Topics Med. Chem. 2002, 2, 939–
971.
12. Tenzer, A.; Pruschy, M. Curr. Med. Chem. 2003, 3, 35–
46.
13. Sausville, E. A.; Arbuck, S. G.; Messmann, R.; Headlee,
D.; Lush, R. D.; Bauer, K.; Murgo, A.; Figg, W. D.;
Lahusen, T.; Jaken, S.; Roberge, M.; Fuse, E.; Kuwabara,
T.; Senderowicz, A. M. J. Clin. Oncol. 2001, 19, 2319–
2333.
14. Senderowicz, A. M. The Oncologist 2002, 7, 12–19.
15. Graves, P. R.; Yu, L.; Schwarz, J. K.; Gales, J.; Sausville,
E. A.; OÕConnor, P. M.; Piwnica-Worms, H. J. Biol.
Chem. 2000, 275, 5600–5605.
16. Busby, E. C.; Leistritz, D. F.; Abraham, R. T.; Karnitz, L.
M.; Sarkaria, J. N. Cancer Res. 2000, 2000, 2108–2112.
17. Jackson, J. R.; Gilmartin, A.; Imburgia, C.; Winkler, J.
D.; Marshall, L. A.; Roshak, A. Cancer Res. 2000, 60,
566–572.
18. Zhao, B.; Bower, M. J.; McDevitt, P. J.; Zhao, H.; Davis,
S. T.; Johanson, K. O.; Green, S. M.; Concha, N. O.;
Zhou, B.-B. J. Biol. Chem. 2002, 277, 44609–44615.
19. Lyne, P. D.; Kenny, P. W.; Cosgrove, D. A.; Deng, C.;
Zabludoff, S.; Wendoloski, J. J.; Ashwell, S. J. Med.
Chem. 2004, 47, 1962–1968.
20. Kawabe, T. Mol. Cancer Ther. 2004, 3, 513–519.
21. Eastman, A.; Kohn, E. A.; Brown, M. K.; Rathman, J.;
Livingstone, M.; Blank, D. H.; Gribble, G. W. Mol.
Cancer Therap. 2002, 1, 1067–1078.
22. Roberge, M.; Berlinck, R. G. S.; Xu, L.; Anderson, H. J.;
Lim, L. Y.; Curman, D.; Stringer, C. M.; Friend, S. H.;
Davies, P.; Vincent, I.; Haggarty, S. J.; Kelly, M. T.;
Britton, R.; Piers, E.; Anderson, R. J. Cancer Res. 1998,
58, 5701–5706.
36. Desiraju, G. R.; T., S. The Weak Hydrogen Bond In
Structural Chemistry and Biology; Oxford Science Publi-
cations, 1999; Vol. 9.
37. Steiner, T. New J. Chem. 1998, 1099–1103.
38. Pierce, A. C.; Sandretto, K. L.; Bemis, G. W. Proteins
2002, 49, 567–576.
39. Moras, D.; Klaholz, B. P. Structure 2002, 10, 1197–1204.
40. Sarkhel, S.; Desiraju, G. R. Proteins 2004, 54, 247–259.
41. Levitt, M.; Perutz, M. F. J. Mol. Biol. 1988, 201, 751–754.
42. Meyer, E. A.; Castellano, R. K.; Diederich, F. Angewandte
Chemie 2003, 42, 1210–1250.
43. Steiner, T.; Koellner, G. J. Mol. Biol. 2001, 305, 535–557.
44. Scheiner, S.; Kar, T.; Pattanayak, J. J. Am. Chem. Soc.
2002, 124, 13257–13264.
45. Brandl, M.; Weiss, M. S.; Jabs, A.; Suhnel, J.; Hilgenfeld,
¨
R. J. Mol. Biol. 2001, 307, 357–377.
46. Nakagawa, Y.; Ire, K.; Yanagita, R. C.; Hhigashi, H.;
Tsuda, K. J. Am. Chem. Soc. 2005, 127, 5746–5747.
47. Nishio, M.; Umezawa, Y.; Hirota, M.; Takeuchi, Y.
Tetrahedron 1995, 51, 8665–8701.
48. Dunitz, J. D. Science 1994, 264, 670.
49. Schlegel, H. B.; Gund, P.; Fluder, E. M. J. Am. Chem.
Soc. 1982, 104, 5347–5351.
50. Clegg, W.; Nichol, G. S. Acta Cryst. E 2004, 60, 1433–1436.
51. Kollman, P. A. Chem. Rev. 1993, 93, 2395–2417.
52. Pearlman, D. A.; Charifson, P. S. J. Med. Chem. 2001, 44,
3417–3423.
53. Manning, G.; Whyte, D. B.; Martinez, R.; Hunter, T.;
Sudarsanam, S. Science 2002, 298, 1912–1934.
54. Toledo, L. M.; Lydon, N. B.; Elbaum, D. Curr. Med.
Chem. 1999, 6, 775–805.
55. Williams, D. H.; Mitchell, T. Curr. Opin. Pharmacol. 2002,
2, 567–573.
23. Kohn, E. A.; Yoo, C. J.; Eastman, A. Cancer Res. 2003,
63, 31–35.
56. Noble, M. E.; Endicott, J. A.; Johnson, L. N. Science
2004, 303, 1800–1805.
24. Kania, R. S.; Bender, S. L.; Borchardt, A.; Braganza, J.
F.; Cripps, S. J.; Hua, Y.; Johnson, M. D.; Johnson, T. O.
J.; Luu, H. T.; Palmer, C. L.; Reich, S. H.; Tempczyk-
Russell, A. M.; Teng, M.; Thomas, C.; Varney, M. D.;
Wallace, M. B. Patent WO 0102369 2001.
57. Shao, R.-G.; Cao, C.-X.; Shimizu, T.; OÕConnor, P. M.;
Kohn, K. W.; Pommier, Y. Cancer Res. 1997, 57,
4029–4035.
58. Jin, P.; Gu, Y.; Morgan, D. O. J. Cell Biol. 1996, 134,
963–970.
25. Foloppe, N.; Fisher, L. M.; Howes, R.; Kierstan, P.;
Potter, A.; Robertson, A. G. S.; Surgenor, A. E. J. Med.
Chem. 2005, 48, 4332–4345.
26. Wang, G.; Li, G.; Mantei, R. A.; Chen, Z.; Kovar, P.; Gu,
W.; Xiao, Z.; Zhang, H.; Sham, H. L.; Sowin, T.;
Rosenberg, S.; Lin, N.-H. J. Med. Chem. 2005, 48,
3118–3121.
27. Edwards, M. L.; Cox, P. J.; Amendola, S.; Deprets, S. D.;
Gillepsy, T. A.; Edlin, C. D.; Morley, A. D.; Gardner, C.
J.; Pedgrift, B.; Bouchard, H.; Babin, D.; Gauzy, L.; Le
Brun, A.; Majid, T. N.; Reader, J. C.; Payne, L. J.; Khan,
N. M.; Cherry, M. Patent WO 03/035065 2003.
28. Berman, H. M.; Westbrook, J.; Feng, Z.; Gilliland, G.;
Bhat, T. N.; Weissig, H.; Shindyalov, I. N.; Bourne, P. E.
Nucleic Acids Res. 2000, 28, 235–242.
59. Poon, R. Y. C.; Jiang, W.; Toyoshima, H.; Hunter, T. J.
Biol. Chem. 1996, 271, 13283–13291.
60. Rhind, N.; Furnari, B.; Russell, P. Genes Dev. 1997, 11,
504–511.
61. Morley, S. D.; Afshar, M. J. Comput. Aided Mol. Des.
2004, 18, 189–208.
62. Barril, X.; Hubbard, R. E.; Morley, D. Mini-Rev. Med.
Chem. 2004, 4, 779–791.
63. Morley, D. in preparation.
64. MacKerell, A. D., Jr.; Bashford, D.; Bellott, M.; Dun-
brack, R. L., Jr.; Evanseck, J. D.; Field, M. J.; Fisher, S.;
Gao, J.; Guo, H.; Ha, S.; Joseph-McCarthy, D.; Kuchnir,
L.; Kuczera, K.; Lau, F. T. K.; Mattos, C.; Michnick, S.;
Ngo, T.; Nguyen, D. T.; Prodhom, B.; Reiher, W. E., III;
Roux, B.; Schlenkrich, M.; Smith, J. C.; Stote, R.; Straub,

1804
N. Foloppe et al. / Bioorg. Med. Chem. 14 (2006) 1792–1804
´
J.; Watanabe, M.; Wiorkiewicz-Kuczera, J.; Yin, D.;
Karplus, M. J. Phys. Chem. B 1998, 102, 3586–3616.
77. Jones, T. A.; Kjelgaard, M. Uppsala University, Uppsala,
Sweden 1993.
65. http://www.chemcomp.com.
78. Perrakis, A.; Morris, R. J. H.; Lamzin, V. S. Nat. Struct.
Biol. 1999, 6, 458–463.
79. Laskowwski, R. A.; MacArthur, M. W.; Moss, D. S.;
Thornton, J. M. J. Appl. Crystallogr. 1993, 26, 283–291.
80. Collaborative Computational Project, N. Acta Cryst. D
1994, 50, 760–763.
81. Frigerio, M.; Santagostino, M.; Sputore, S. J. Org. Chem.
1999, 64, 4537–4538.
82. Yang, J.; Cron, P.; Good, V. M.; Thompson, V.;
Hemmings, V.; Barford, D. Nat. Struct. Biol. 2002, 9,
940–944.
66. Halgren, T. J. Comp. Chem. 1999, 20, 720–729.
67. Hardcastle, I. R.; Arris, C. E.; Bentley, J.; Boyle, F. T.;
Chen, Y.; Curtin, N. J.; Endicott, J. A.; Gibson, A. E.;
Golding, B. T.; Griffin, R. J.; Jewsbury, P.; Meneyrol, J.;
Mesguiche, V.; Newell, D. R.; Noble, M. E. M.; Pratt, D. J.;
Wang, L.-Z.; Whitfield, H. J. J. Med. Chem. 2004, 47, 3710.
68. OÕDonovan, C.; Martin, M. J.; Gattiker, A.; Gasteiger, E.;
Bairoch, A.; Apweiler, R. Brief Bioinform. 2002, 3, 275–
284.
69. Chothia, C.; Lesk, A. M. EMBO J. 1986, 5, 823–826.
´
´
70. Martı-Renom, M. A.; Stuart, A. C.; Fiser, A.; Sanchez,
R.; Melo, F.; Sali, A. Annu. Rev. Biophys. Biomol. Struct.
83. Bhat, R.; Xue, Y.; Berg, S.; Hellberg, S.; Ormo, M.;
Nilsson, Y.; Radesater, A. C.; Jerning, E.; Markgren, P.
O.; Borgegard, T.; Nylof, M.; Gimenez-Cassina, A.;
Hernandez, F.; Lucas, J. J.; Diaz-Nido, J.; Avila, J. J.
Biol. Chem. 2003, 278, 45937–45945.
84. Biondi, R. M.; Komander, D.; Thomas, C. C.; Lizcarno,
J. M.; Deak, M.; Alessi, D. R.; van Aalten, D. M. F.
EMBO J. 2003, 21, 4219–4228.
85. Zheng, J. H.; Trafny, E. A.; Knighton, D. R.; Xuong, N.
H.; Taylor, S. S.; Teneyck, L. F.; Sowadsky, J. M. Acta
Cryst. D Biol. Cryst. 1993, 49, 362.
86. Kakkar, T.; Pak, Y.; Mayersohn, M. J. Pharmacol. Exp.
Ther. 2000, 293, 861–869.
ˇ
2000, 29, 291–325.
71. Levitt, M. J. Mol. Biol. 1992, 226, 507–533.
72. Fechteler, T.; Dengler, U.; Schomberg, D. J. Mol. Biol.
1995, 253, 114–131.
73. Delano, W. L. The PyMOL UserÕs Manual; DeLano
Scientific: San Carlos, CA, USA, 2002.
74. Kakkar, T.; Boxenbaum, H.; Mayersohn, M. Drug Metab.
Dispos. 1999, 27, 756–762.
75. http://www.idbs.com.
76. Otwinowski, Z.; Minor, W. Methods Enzymol. 1997, 276,
307–326.