Discovery of picomolar ABL kinase inhibitors equipotent for wild type and T315I mutant via structure-based de novo design

Article abstract of DOI:10.1021/ja311756u

Although the constitutively activated break-point cluster region-Abelson (ABL) tyrosine kinase is known to cause chronic myelogenous leukemia (CML), the prevalence of drug-resistant ABL mutants has made it difficult to develop effective anti-CML drugs. With the aim to identify new lead compounds for anti-CML drugs, we carried out a structure-based de novo design using the scoring function improved by implementing an accurate solvation free energy term. This approach led to the identification of ABL inhibitors equipotent for the wild type and the most drug-resistant T315I mutant of ABL at the picomolar level. Decomposition analysis of the binding free energy showed that a decrease in the desolvation cost for binding in the ATP-binding site could be as important as the strengthening of enzyme-inhibitor interaction to enhance the potency of an ABL inhibitor with structural modifications. A similar energetic feature was also observed in free energy perturbation (FEP) calculations. Consistent with the previous experimental and computational studies, the hydrogen bond interactions with the backbone groups of Met318 proved to be the most significant binding forces to stabilize the inhibitors in the ATP-binding sites of the wild type and T315I mutant. The results of molecular dynamics simulations indicated that the dynamic stabilities of the hydrogen bonds between the inhibitors and Met318 should also be considered in designing the potent common inhibitors of the wild-type and T315I mutant of ABL.

Full text of DOI:10.1021/ja311756u

Article
pubs.acs.org/JACS
Discovery of Picomolar ABL Kinase Inhibitors Equipotent for Wild
Type and T315I Mutant via Structure-Based de Novo Design
Hwangseo Park,*,† Seunghee Hong,‡ Jinhee Kim,‡ and Sungwoo Hong*,‡
†Department of Bioscience and Biotechnology, Sejong University, Seoul 143-747, Korea
‡Department of Chemistry, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, 305-701, Korea
S
* Supporting Information
ABSTRACT: Although the constitutively activated break-point cluster
region−Abelson (ABL) tyrosine kinase is known to cause chronic
myelogenous leukemia (CML), the prevalence of drug-resistant ABL
mutants has made it difficult to develop effective anti-CML drugs. With
the aim to identify new lead compounds for anti-CML drugs, we
carried out a structure-based de novo design using the scoring function
improved by implementing an accurate solvation free energy term.
This approach led to the identification of ABL inhibitors equipotent for
the wild type and the most drug-resistant T315I mutant of ABL at the
picomolar level. Decomposition analysis of the binding free energy
showed that a decrease in the desolvation cost for binding in the ATP-
binding site could be as important as the strengthening of enzyme−
inhibitor interaction to enhance the potency of an ABL inhibitor with
structural modifications. A similar energetic feature was also observed in free energy perturbation (FEP) calculations. Consistent
with the previous experimental and computational studies, the hydrogen bond interactions with the backbone groups of Met318
proved to be the most significant binding forces to stabilize the inhibitors in the ATP-binding sites of the wild type and T315I
mutant. The results of molecular dynamics simulations indicated that the dynamic stabilities of the hydrogen bonds between the
inhibitors and Met318 should also be considered in designing the potent common inhibitors of the wild-type and T315I mutant
of ABL.
Included in these second-generation ABL inhibitors are
dasatinib,7 nilotinib,8 bafetib,9 bosutinib,10 PPY-A,11 tozaser-
tib,12 danusertib,13 AT-9283,14 and KW-2449.15 Among the
various drug-resistant variants of ABL, T315I mutant has been
the most difficult one to deactivate using ATP-competitive ABL
inhibitors. This mutation occurs at the gatekeeper site and
eliminates a critical hydrogen bond required for tight binding of
the inhibitors by changing the geometry of the ATP-binding
pocket.16 Nonetheless, the abundance of structural information
about the interactions between the wild-type and mutant ABL
kinases and their small-molecule inhibitors has made it possible
to discover new anti-CML drugs. Indeed, such structural
information has enabled the design of common inhibitors for
the wild type and T315I mutant of ABL17 and has elucidated
the mechanism by which the known inhibitors function.18
Recently, we reported that the antimicrotuble agent,
nocodazole in Figure 1, was a common inhibitor of the wild
type and T315I mutant of ABL with the associated Kd values of
∼100 nM.19 Besides the high inhibitory activities, it has
desirable physicochemical properties as a drug candidate and
low molecular weight of 301.3 amu. Therefore, it is anticipated
to serve as a good inhibitor scaffold from which more potent
INTRODUCTION
■
The reciprocal translocation between the Abelson gene on
chromosome 9 and break-point cluster region gene on
chromosome 22 leads to the construction of Philadelphia
chromosome. This recombined gene produces the fusion
protein (ABL) that is a constitutively active tyrosine kinase and
is responsible for the pathogenesis of chronic myelogenous
leukemia (CML).1 Because ABL is involved in most cases of
CML, the inhibition of ABL activity offers a novel strategy for
the development of therapeutics to treat CML.2 Considerable
efforts have therefore been devoted to the discovery of small-
molecule ABL inhibitors as recently reviewed in a compre-
hensive fashion.3 These scientific endeavors have identified the
potent first-generation ABL inhibitors including imatinib, which
was approved for the clinical treatment of CML.4
However, CML cells can develop resistance to the first-
generation inhibitor drugs by introducing point mutations in
the kinase domain of ABL, which has an effect of impeding the
binding of inhibitors in the ATP-binding site. At least 100
different point mutations have been identified in CML patients
who display resistance to imatinib.5 The emergence of drug
resistance has complicated the clinical treatment of CML using
the first-generation anti-CML drugs and motivated the
discovery of new ABL inhibitors with high potency against
both the imatinib-resistant mutants and the wild type of ABL.6
Received: December 3, 2012
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja311756u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
in the de novo design of nocodazole derivatives to identify the
common potent inhibitors of the wild type and T315I mutant.
To obtain the all-atom models for the two receptors, hydrogen
atoms were added to each protein atom in the wild type and T315I
mutant. A special attention was paid to assign the protonation states of
the ionizable Asp, Glu, His, and Lys residues in the original X-ray
crystal structures. The side chains of Asp and Glu residues were
assumed to be neutral if one of their carboxylate oxygens pointed
toward a hydrogen-bond accepting group including the backbone
aminocarbonyl oxygen at a distance within 3.5 Å, a generally accepted
distance limit for a hydrogen bond of moderate strength.23 Similarly,
the lysine side chains were assumed to be protonated unless the NZ
atom was in proximity to a hydrogen-bond donating group. The same
procedure was also applied to determine the protonation states of ND
and NE atoms in His residues.
The structure-based de novo design of new ABL inhibitors was
performed in two steps. First, a variety of nocodazole derivatives were
generated with the LigBuilder program24 using the calculated ABL-
nocodazole and T315I-nocodazole complexes obtained in the previous
study.19 To score the derivatives of a given inhibitor scaffold according
to the binding affinity for the target protein, the program employs the
empirical binding free energy function including van der Waals,
hydrogen bond, electrostatic, and entropic terms.25 As a starting point
to generate the nocodazole derivatives, we analyzed the binding
pockets in the ATP-binding sites of the wild type and T315I mutant
using the POCKET module. The structures of the wild type and
T315I mutant in complex with nocodazole were used as the inputs to
identify the key interaction residues in the ATP-binding sites. The next
step involved the generation of various nocodazole derivatives by
applying a genetic algorithm in which the structures were evolved by
changing the chemical groups at certain substitution positions with the
molecular core being held constant. In this structure-based de novo
design, we explored the substituent space on nocodazole at three
positions to identify the derivatives with enhanced inhibitory activity
against the wild type and T315I mutant of ABL. This could be possible
by comparing the calculated binding free energies of the derivatives
with respect to the wild type and the mutant. In the first step of de
novo design process, the bioavailability rules were also applied to
screen the approximately 100 000 derivatives of nocodazole that were
expected to have desirable physicochemical properties as a drug
candidate.26
Figure 1. Chemical structure of nocodazole.
common inhibitors can be derivatized. On the basis of the
structure-based de novo design and the synthesis of various
nocodazole derivatives, we aim in the present study to identify
new common inhibitors of wild type and T315I mutant of ABL
that can develop into new anti-CML drugs. Computer-aided
drug design has not always been successful because of the
inaccuracy in the scoring function, which leads to a weak
correlation between the computational predictions and
experimental results for binding affinities.20 Our de novo
design approach is distinguished from others by the
implementation of an accurate solvation model in the binding
free energy calculations between ABL and the putative ligands.
This modification of the scoring function seems to improve the
potential for designing the new inhibitors with high activity.21
On the basis of docking analysis, molecular dynamics (MD)
simulations, and free energy perturbation (FEP) calculations,
we also address the energetic and structural features relevant to
the stabilization of the newly identified inhibitors in the ATP-
binding sites of the wild type and T315I mutant of ABL.
MATERIALS AND METHODS
■
De Novo Design. The conformation of Asp381-Phe382-Gly383
(DFG) motif in the activation loop is very important for ABL kinase
activity because this motif is positioned in proximity to the ATP-
binding site. The side chain of Asp381 is directed toward the ATP-
binding site in the active conformation to coordinate an ATP-bound
Mg2+ ion, which is referred to as the “DFG-in” conformation. The side
chains of Asp381 and Phe382 swap positions in the inactive (DFG-
out) conformation, which leaves the former pointing in opposite
direction with respect to the ATP-binding site. The active
conformations of the wild type (Figure 2) and T315I mutant of
Scoring of the Nocodazole Derivatives. Although the effects of
ligand solvation have been shown to be critically important in
protein−ligand association,21 the current scoring function of the
LigBuilder program lacks a solvation term. Therefore, the derivatives
of nocodazole generated with LigBuilder were further screened with a
new binding free energy function constructed by combining an
appropriate solvation free energy term to the original scoring function
of the AutoDock program.27 This modified scoring function can be
expressed in the following form.
⎛
⎜
⎞
Aij
B
ij
r 6
aq
⎟
⎟
ΔGbind = WvdW ∑ ∑ ⎜r 12
−
+ Whbond
ij
⎝
ij
ij ⎠
i=1 j=1
⎛
⎝ ij
⎞
qiqj
Cij
r 12
D
r 10
⎜
⎜
⎟
⎟
∑ ∑ E(t)
−
+ W ∑ ∑
+ W Ntor
elec
tor
ε(r )r
ij ij
ij
⎠
i=1 j=1
i=1 j=1
2/2σ2
+ W ∑ Si(Omax − ∑ Vje−r
)
ij
sol
i
i=1
j≠i
(1)
Figure 2. Ribbon representation of ABL in complex with a potent
inhibitor (29). The positions of gatekeeper site, Thr315, Gly loop, and
DFG motif are indicated.
Here, WvdW, Whbond, Welec, Wtor, and Wsol are the weighting factors of
van der Waals, hydrogen bond, electrostatic interactions, torsional
term, and desolvation energy of inhibitors, respectively. rij represents
the interatomic distance, and Aij, Bij, Cij, and Dij are related to the
depths of the potential energy well and the equilibrium separations
between the protein and ligand atoms. The hydrogen bond term has
an additional weighting factor, E(t), representing the angle-dependent
directionality. Cubic equation approach was applied to obtain the
dielectric constant required in computing the electrostatic interactions
between ABL and nocodazole derivatives.28 Gasteiger−Marsili atomic
ABL were selected as the target proteins in this study. We prepared the
two receptor models in the “DFG-in” conformation from the X-ray
crystal structure of the wild type (residues 229−500, 2.40 Å
resolution) in complex with dasatinib (PDB code 2GQG)22 and that
of T315I mutant (residues 228−515, 1.95 Å resolution) in complex
with PPY-A (PDB code 2QOH).11 These structural models were used
B
dx.doi.org/10.1021/ja311756u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
charges29 were then used for both proteins and ligands to compute the
electrostatic interaction energies. In the entropic term, Ntor is the
number of rotatable bonds in the ligand. In the desolvation term, Si
and Vi are the solvation parameter and the fragmental volume of atom
i,30 respectively, while Omi ax stands for the maximum atomic occupancy.
In the calculation of molecular solvation free energy of nocodazole
derivatives, we used the atomic parameters developed by Choi et al.31
This modification of the solvation free energy term seems to have an
effect of increasing the accuracy in the scoring function because the
underestimation of ligand solvation often leads to the overestimation
of the binding affinity of a ligand with many polar atoms.21 Indeed, the
superiority of this modified scoring function to the previous one was
well-appreciated in recent studies for virtual screening of kinase
inhibitors.32 Using the binding free energy function in eq 1, the
nocodazole derivatives generated with LigBuilder were scored
according to the binding affinities for the wild type and T315I mutant
of ABL. In this second step of de novo design process, 100 top-ranked
derivatives were selected as virtual hits, and the availability of a
synthetic path was determined. About thirty of the designed
nocodazole derivatives that passed this filter were synthesized and
tested for inhibitory activity against the wild type and T315I mutant of
ABL.
to be consistent with the standard AMBER force field. The starting
coordinates of MD simulations were prepared from docking
simulations of the nocodazole derivatives in the ATP-binding sites
of the wild type and T315I mutant. After the addition of twelve
sodium ions for charge neutralization with the AMBER force field
parameters, the all-atom models for the wild type and T315I mutant in
complex with nocodazole derivatives were immersed in rectangular
boxes of dimension 73.8 × 89.6 × 81.7 Å containing 12 904 TIP3P36
water molecules. After 200 cycles of energy minimization to remove
the poor steric contacts, we equilibrated the whole system beginning
with 20 ps equilibration dynamics of the solvent molecules at 300 K.
The next step involved the equilibration of the solute with a fixed
configuration of the solvent molecules consecutively at 10, 50, 100,
150, 200, 250, and 300 K for 10 ps at each temperature. Then, the
equilibration dynamics of the entire system was performed at 300 K
for 100 ps. Following the equilibration procedure, 20.2 ns production
dynamics simulations were carried out with periodic boundary
conditions in the NPT ensemble. The temperature and pressure
were kept at 300 K and 1 atm using Berendsen temperature coupling37
and isotropic molecule-based scaling, respectively. The SHAKE
algorithm,38 with a tolerance of 10−6 Å, was applied to fix all bond
lengths involving the hydrogen atom. We used the time step of 2.0 fs
and the nonbond-interaction cutoff radius of 12 Å; the trajectory was
sampled every 0.4 ps (200-step intervals) for analysis.
Chemical Synthesis and Enzyme Assays. The de novo design
process identified a variety of promising aromatic ring systems that can
be substituted at C6 position of the benzimidazole core. The general
synthetic route for the preparation of benzimidazole derivatives is
shown in Scheme 1. Functionalization of the C6-position was
FEP Calculations. To further understand the energetic features
associated with the inhibition of the wild type and T315I mutant of
ABL by nocodazole derivatives, FEP calculations were also carried out
based on MD simulations. As the starting structures of these MD-FEP
calculations in aqueous solution, we used the final structures of
enzyme−inhibitor complexes obtained in the equilibration dynamics.
MD-FEP calculations were performed using a thermodynamic cycle-
perturbation approach, which is depicted in Figure 3. The change in
Scheme 1. Synthetic Route of Benzimidazole and
a
Benzothiazole Derivatives
Figure 3. Thermodynamic cycle used in calculating the free energy
change in aqueous solution (ΔGu), that in the ATP-binding site of
ABL (ΔGb), and the difference in binding free energies (ΔΔGbind
between the two inhibitors (I1 and I2).
)
a
Reagents and conditions: (a) Pd(dppf)Cl2·CH2Cl2, arylboronic acid,
K2CO3, 1,4-dioxane/H2O = 3:10, 100 °C, 4 h; (b) RCON=C(SMe)-
NHCOR (R = OMe (A), NHEt (B)), AcOH/H2O = 3:1, 100 °C, 5 h;
(c) BrCN, MeCN/H2O, rt, overnight; (d) CDI, DMF, rt, 5 h, then
MeNH2; (e) EtNCO, 1,4-dioxane, 90 °C, 15 h, then H2O, 100 °C, 3 h;
(f) Ac2O, pyridine, rt, 8 h.
binding free energy (ΔΔGbind) between the two inhibitors can be
computed through the nonphysical paths connecting the desired initial
and final states. More specifically, this approach enables the calculation
of ΔΔGbind between the two structurally similar inhibitors by
computationally simulating the transformation of one into the other.
Because the free energy is a state function and, therefore, the sum of all
energies in the thermodynamic cycle in Figure 3 is zero, ΔΔGbind can
be expressed as follows.
accomplished by direct Suzuki couplings of 4-bromo-1,2-diaminoben-
zene (2) with corresponding boronic acids, followed by the
condensation reaction with appropriate reagents A or B to facilitate
the conversions of o-phenylenediamines directly to the desired product
6 in high yields. The benzothiazole derivatives were synthesized from
6-bromo-2-aminobenzothiazole 4 because they were expected to be
potent ABL inhibitors in the de novo design and docking simulations.
After the direct cross-coupling reactions with 4-pyridyl- and phenyl-
boronates under standard coupling conditions, 2-amino group of 5
were further functionalized to either amide or urea groups through the
reactions with ethyl isocyanate or anhydride. The synthesized
derivatives were then tested to determine the IC50 values against
both the wild type and T315I mutant of ABL.33
MD Simulations. MD simulations of the wild type and T315I
mutant of ABL in complex with nocodazole derivatives were carried
out using the AMBER program (version 12) with the force field
parameters reported by Cornell et al.34 To obtain the potential
parameters for nocodazole derivatives unavailable in the force field
database, we followed the procedure suggested by Fox and Kollman35
ΔΔGbind = ΔG2 − ΔG1 = ΔGb − ΔGu
(2)
Here, ΔG1 and ΔG2 refer to the binding free energies of inhibitors I1
and I2, respectively, and ΔGu and ΔGb to the free energies associated
with the structural transformation of I1 into I2 in aqueous solution and
in the ATP-binding site of ABL.
To determine the ΔΔGbind value between I1 and I2, therefore, the
free energy changes in transforming I1 into I2 should be calculated
both in aqueous solution and in the ATP-binding site of ABL. The free
energy change for converting I1 into I2 was computed by perturbing
the Hamiltonian of I1 (initial state) into that of I2 (final state). This
transformation could be accomplished through the parametrization of
the terms comprising the interaction potentials of the system with the
change of a state variable (λ) that mapped onto reactant (λ = 0) and
product (λ = 1) states. The total free energy change for the mutation
from the initial to the final state was then computed by summing the
incremental free energy changes over several windows visited by λ as it
C
dx.doi.org/10.1021/ja311756u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Table 1. Structures and Inhibitory Activities of Representative Nocodazole Derivatives
a
Asterisk indicates the atom attached to the position of substitution.
changed from 0 to 1. Staring from the initial structures, we performed
1680 ps of perturbation for a selected pair of nocodazole derivatives.
Each perturbation consisted of 21 windows with 40 000 steps of
equilibration and 40 000 steps of data collection. We used the time
step of 1 fs and the nonbond-interaction cutoff radius of 12 Å. A
doublewide sampling procedure was performed for all of the structural
transformations, and the reported results were based on the averages
from the backward and forward simulations. As a rough estimation of
the statistical error, we used half the difference between the absolute
values of forward and reverse free energy changes, and the error in
ΔΔGbind was calculated as the square root of the sum of the squares of
the individual errors in ΔGu and ΔGb
methoxy group in nocodazole (1) to ethylamine in 7 leads to
more than 20-fold increase in inhibitory activity both for the
wild type and for T315I mutant. This significant enhancement
of potency is most likely to stem from the strengthening of the
hydrogen bonds with protein groups because the terminal
carbamate group in 1 was found to establish multiple hydrogen
bonds at the entrance of ATP-binding site of ABL.19 In
contrast, the substitution of 3,4-dimethoxybenzene in 8 for the
thiophene-2-one group in 1 leads to the loss of potency.
However, the inhibitory activity can be recovered by the
additional replacement of terminal methoxy group with
ethylamine in 9, which confirms that the formation of strong
hydrogen bonds in the ATP-binding site are critically important
for the effective inhibition of the wild type and T315I mutant
ABL. The high inhibitory activities of 14 and 15 toward both
the wild type and T315I mutant indicate that 2-methoxyphenyl
and 2-ethoxyphenyl groups can serve as the best substituents at
W position of the inhibitor scaffold (Table 1). Most
remarkably, the change of the central aromatic group from
benzimidazole in 9 and 11 to benzothiazole in 27 and 28 leads
to the increase in the inhibitory activity by a factor of 50−100.
The combination of 2-methoxybenzene group, benzothiazole
core, and ethylamine moiety at the three substitution positions
RESULTS AND DISCUSSION
■
Inhibitory Activities of Nocodazole Derivatives.
Among the nocodazole derivatives prepared and tested for
the inhibitory activity, more than ten compounds were found to
have a high potency against both the wild type and T315I
mutant of ABL at the submicromolar level. Table 1 lists the
structures and IC50 values of the representative inhibitors. Most
nocodazole derivatives exhibit a higher potency against the wild
type than against T315I mutant. This result is consistent with
the difficulty associated with achieving a high inhibitory activity
for T315I mutant.39 We note that the change of the terminal
D
dx.doi.org/10.1021/ja311756u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
yields the most potent inhibitor (29) with picomolar activity.
Judging from the exceptionally high inhibitory activities against
both the wild type and T315I mutant, 29 deserves
consideration for further investigation toward the development
of new anti-CML medicines.
Binding Mode Analysis of the Newly Identified
Inhibitors. To obtain some energetic and structural insight
into the inhibitory mechanisms of the newly identified common
inhibitors of the wild type and T315I mutant of ABL, their
binding modes in the ATP-binding sites were investigated using
the modified AutoDock scoring function. The calculated
binding modes of 1, 7, and 9 with respect to the wild type
and T315I mutant are compared in Figure 4. All three
It is interesting to note that the number of hydrogen bonds
increases from two to three as the inhibitor changes from 1 to 7
both for the wild type and for T315I mutant. This trend can be
attributed to the change of the terminal methoxy group in 1 to
the ethylamine group that can establish an additional hydrogen
bond with the backbone aminocarbonyl oxygen of Met318.
This strengthening of hydrogen bond interactions is apparently
responsible for the higher inhibitory activities of 7 than 1
toward the wild type and T315I mutant. In the calculated
structure of ABL-9 complex, however, only two hydrogen
bonds are established with Met318 due to the disappearance of
a hydrogen bond between the aminocarbonyl oxygen of
Met318 and the hydrogen atom (H1) on the terminal
ethylamine moiety of 9. As can be inferred from the
superimposed structures of 1, 7, and 9 in Figure 4a, the
absence of such a hydrogen bond seems to be caused by the
enlargement of the terminal aromatic group from thiophene-2-
one group to 3,4-dimethoxybenzene, which induces the
positional shift of the inhibitor from the ATP-binding site to
bulk solvent. This indicates that the size of terminal aromatic
group should be restricted for the optimal inhibitory activity of
nocodazole derivatives. Beside the hydrogen bond interactions,
the inhibitors can be further stabilized in the ATP-binding sites
via the hydrophobic interactions of their nonpolar groups with
the side chains of Leu248, Tyr253, Val256, Ala269, Val299,
Phe317, Leu370, Ala380, and Phe382. Overall, the structural
features identified in docking simulations indicate that the high
inhibitory activities of certain nocodazole derivatives against the
wild-type and T315I mutant of ABL stem from the multiple
hydrogen bonds and hydrophobic interactions established
simultaneously in the ATP-binding sites.
Despite several similarities in binding modes toward the wild-
type and T315I mutant, some differences are also observed
both in hydrogen bond and in hydrophobic interactions. For
example, two additional residues (Met290 and Ile315) are
found at the interface of van der Waals contacts between the
inhibitors and T315I mutant (Figure 4b), indicating that the
hydrophobic interactions must be established in the stronger
fashion in T315I-inhibitor than in ABL-inhibitor complexes. On
the other hand, the hydrogen bonds between Met318 and the
inhibitors get weaker with the change of the receptor from the
wild type to T315I mutant. The two N−H···N hydrogen bond
distances increase from, respectively, 2.06 Å and 2.07 Å in ABL-
1 and ABL-7 complexes to 2.30 Å and 2.20 Å in T315I-1 and
T315I-7 complexes. The central N−H···O hydrogen bonds also
get weaker as the receptor changes from the wild type to T315I
mutant: the associated H···O distances increase from 2.03 and
1.95 Å in ABL-1 and ABL-7 complexes to 2.17 and 2.28 Å in
T315I-1 and T315I-7 complexes, respectively. The strengthen-
ing of hydrophobic interactions in T315I-inhibitor complexes is
likely to be insufficient to compensate for the weakening of the
hydrogen bonds because 1 and 7 have a lower inhibitory
activity for T315I mutant than for the wild type (Table 1).
Unlike the binding mode in ABL-9 complex, all three hydrogen
bonds are established in T315I-9 complex in a stable form as in
T315I-7 complex. Nonetheless, the binding mode of 9 in the
ATP-binding site of T315I mutant differs significantly from
those of 1 and 7 in that 3,4-dimethoxybenzene group of 9 stays
distant from the ATP-binding site and points toward the DFG
motif located at the top of C-terminal domain. As can be seen
in Figure 4b, 3,4-dimethoxyphenyl moiety of 9 is less stabilized
in the ATP-binding site than thiophene-2-one group of 1 and 7
because of the increase in repulsive interactions with the side
Figure 4. Calculated binding modes of 1, 7, and 9 in the ATP-binding
sites of (a) wild type and (b) T315 mutant of ABL. Carbon atoms of
ABL, 1, 7, and 9 are indicated in cyan, black, pink, and green,
respectively. Each dotted line indicates a hydrogen bond.
inhibitors appear to form close contacts with residues Leu248−
Tyr256, Ile313−Asn322, and Ala269−Lys271, which belong to
the Gly loop, the gatekeeper site, and a β-sheet in the N-
terminal domain, respectively. It is a common structural feature
in the calculated binding modes of three inhibitors for the wild
type and T315I mutant that a nitrogen in the imidazole ring
(N1) and the adjacent amidic hydrogen (H2) form a hydrogen
bond with the backbone amidic hydrogen and the amino-
carbonyl oxygen of Met318, respectively. This is consistent with
the previous X-ray crystallographic studies on ABL-inhibitor
interactions in which the formation of multiple hydrogen bonds
with the backbone groups of Met318 was shown to be
necessary for tight binding of the inhibitors in the ATP-binding
site.11,22
E
dx.doi.org/10.1021/ja311756u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
chain of Ile315, which prevents a bulky group from being
accommodated in the ATP-binding site. Such a difficulty in the
stabilization of terminal 3,4-dimethoxybenzene group in the
ATP-binding site may be an explanation for the significant
decrease in the inhibitory activity for T315I mutant in going
from 1 and 7 to 9. The importance of reducing the repulsive
interactions in designing the T315I mutant inhibitors was also
implicated in recent X-ray crystallographic studies of ABL in
complex with bosutinib.40
and the relatively unfavorable interaction in the ATP-binding
site predicted in docking simulations (Figure 4b).
Figure 6 shows the calculated binding modes 11, 28, and 29
in the ATP-binding sites of the wild type and T315I mutant of
To examine the dynamic stabilities of the proposed binding
modes shown in Figure 4, we carried out 20 ns MD simulations
of the wild type and T315I mutant of ABL in complex with 1,
7, and 9. Compared in Figure 5 are the time evolutions of the
Figure 6. Calculated binding modes of 11, 28, and 29 in the ATP-
binding sites of (a) wild type and (b) T315 mutant of ABL. Carbon
atoms of ABL, 11, 28, and 29 are indicated in cyan, black, pink, and
green, respectively. Each dotted line indicates a hydrogen bond.
ABL. The overall structural features derived from docking
simulations confirm that the high inhibitory activities of
nocodazole derivatives toward the wild type and T315I mutant
stem from the multiple hydrogen bonds and hydrophobic
interactions established simultaneously in the ATP-binding site.
As a consequence of the change of the aromatic group in W
position from 3,4-dimethoxybenzene in 9 to pyridine in 11, the
third hydrogen bond is established in a stable form between the
terminal ethylamine group of the inhibitor and the amino-
carbonyl oxygen of Met318 of the wild type (Figure 6a). This is
consistent with the higher inhibitory activity of 11 than 9 with
respect to the wild type (Table 1). The substitution of pyridine
in 11 for 3,4-dimethoxybenzene in 9 has also an effect of
stabilizing the inhibitor in the ATP binding site of T315I
mutant by reducing the repulsive interactions with the side
chain of Ile315. This feature may account for the 10-fold
increase in the inhibitory activity of 9 over 11 with respect to
T315I mutant. Although the docking simulations of 1, 7, and 9
provided some insight into the relative potencies observed over
the series, the significant differences in IC50 values of 11, 28,
and 29 are difficult to explain because their calculated binding
modes with respect to the wild type and T315I mutant are
quite similar. For example, the urea moiety common in all three
inhibitors establishes three hydrogen bonds with backbone
Figure 5. Time evolutions of the root-mean-square deviation
(RMSDinit) values for heavy atoms of 1, 7, and 9 in the ATP-binding
sites of (a) wild type and (b) T315I mutant of ABL.
root-mean-square deviations (RMSDinit) for the heavy atoms of
1, 7, and 9 in the ATP-binding sites of the wild type and T315I
mutant of ABL. The RMSDinit values of the three inhibitors in
ABL-1, ABL-7, and ABL-9 complexes remain lower than those
in T315I-1, T315I-7, and T315I-9 complexes in the majority of
simulation time, which indicates that the inhibitors bind more
tightly in the ATP-binding site of the wild type than in that of
T315I mutant. This is consistent with a little higher inhibitory
activities of 1, 7, and 9 against the wild type than the mutant
(Table 1). It is also noteworthy that the RMSDinit values fall
into 0.8 Å in most enzyme−inhibitor complexes with the
average values of 0.41−0.61 Å except for the T315I-9 complex,
which indicates an insignificant positional shift of the inhibitors
during the entire course of simulation. On the other hand, 9
exhibits a high motional amplitude in the ATP-binding site of
T315I mutant with the average RMSDinit value of 0.71 Å
(Figure 5b). This dynamic instability may be invoked to explain
the low inhibitory activity of 9 against T315I mutant (Table 1)
F
dx.doi.org/10.1021/ja311756u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
groups of Met318 at the entrance of ATP-binding site while
their nonpolar groups are stabilized at the ATP-binding site by
hydrophobic interactions with the side chains of Leu248,
Tyr253, Val256, Ala269, Met290, Val299, Ile315, Phe317,
Leu370, Ala380, and Phe382. Therefore, further energetic and
structural analyses were performed to elucidate the differences
in inhibitory activities of the derivatives with similar binding
modes.
Decomposition Analysis of Binding Free Energy. We
now turn to the energetic features associated with binding of
the newly identified inhibitors in the ATP-binding sites of the
wild type and T315I mutant of ABL. Table 2 lists the calculated
desolvation cost. For this reason, the ΔGabq values of 9
associated with the binding in the ATP-binding sites of the wild
type and T315I mutant become less favorable than the
corresponding values of 7 by 2.9 and 3.8 kcal/mol, respectively,
which is consistent with more than 10-fold decrease in
inhibitory activity (Table 1). As can be inferred from the
decrease in ΔGgbas and the increase in ΔGsol values in going
from 9 to 11, the change of 3,4-dimethoxybenzene in 9 to
pyridine in 11 enhances the inhibitory activity by strengthening
the enzyme−inhibitor interactions and by reducing the
desolvation cost for complexation.
With respect to the importance of the hydrogen bonds with
the backbone groups of Met318, we note that ΔGgbas values
become more favorable from −20.9 and −20.7 kcal/mol in
ABL-1 and T315I-1 complexes to −23.2 and −23.5 kcal/mol in
ABL-7 and T315I-7 complexes, respectively. Because the
calculated binding modes of the two inhibitors are very similar
except for the number of hydrogen bonds with Met318 (Figure
4), the gain of extra stabilization energy by 2.3 (wild type) and
2.8 kcal/mol (T315I mutant) with the change of the inhibitor
from 1 to 7 can be attributed to the formation of an additional
hydrogen bond in the ABL-7 and T315I-7 complexes. Thus,
the differences in the structural and energetic features
associated with the binding modes of 1 and 7 indicate that
the hydrogen bonds with Met318 should be considered as the
key interactions in designing the potent ABL inhibitors.
Consistent with a significant increase in the inhibitory
activities for the wild type and T315I mutant, the ΔGabq values
of 28 are lower than those of 11 by 2.6 kcal/mol for the wild
type and by 2.9 kcal/mol for T315I mutant. We note in this
regard that the ΔGsol value of 28 is 2.8 kcal/mol higher than
that of 11 whereas the ΔGgbas values of the two inhibitors are
similar for both receptors. This difference indicates that the
substantially enhanced inhibitory activity because of the
substitution of benzothiazole group in 28 for benzimidazole
ring in 11 stems from the reduced desolation cost for
complexation rather than the strengthening of enzyme−
inhibitor interactions. The comparable ΔGgbas values of 11
and 28 are actually not surprising because their calculated
binding modes are similar for both the wild type and T315I
mutant as shown in Figure 6. The large difference in ΔGsol
values of 11 and 28 can also be understood because imidazole
is soluble a little in water while thiazole has only a micromolar
water-solubility.41 Thus, the energetic features found in this
study demonstrate that the potency of an inhibitor can be
optimized by reducing the desolvation cost for enzyme−
inhibitor complexation, as well as by strengthening the
interactions between the inhibitor and the target protein. In
contrast to more than 10-fold increase in the inhibitory activity
because of the replacement of the terminal pyridine ring in 28
with 2-methoxyphenyl group in 29, the ΔGabq values of 28 and
29 are predicted to be similar. This disagreement indicates that
despite the improvement with the solvation term, the scoring
function remains insufficient by itself to fully characterize the
relative potencies of nocodazole derivatives.
Table 2. Calculated Binding Free Energies of ABL
Inhibitors
a
ΔGgbas
wild type
ΔGabq
wild type
compd
T315I
ΔGsol
T315I
1
−20.9
−23.2
−22.6
−23.3
−23.1
−23.8
−20.7
−23.5
−22.0
−23.2
−23.3
−23.5
−12.9
−13.9
−16.2
−15.3
−12.5
−12.8
−8.0
−9.3
−6.4
−8.0
−10.6
−11.0
−7.8
−9.6
−5.8
−7.9
−10.8
−10.7
7
9
11
28
29
a
Calculated binding free energy in the gas phase (ΔGgbas), solvation
free energy (ΔGsol), and binding free energy in solution (ΔGabq) of the
selected inhibitors with respect to the wild type and T315I mutant of
ABL. The ΔGabq values indicate the final binding free energies for the
best binding configurations. All energy values are given in kcal/mol.
binding free energies in solution and solvation free energies of
the six inhibitors mentioned above. Bearing in mind that the
binding free energy of a protein−ligand complex in aqueous
solution (ΔGbaq) may be approximated as the difference
between that in the gas phase (ΔGgbas) and the solvation free
energy of the ligand (ΔGsol), we computed the two energy
components separately to estimate their relative contributions
to ΔGabq. When the two energy components for the six
inhibitors are compared, it follows immediately that ΔGsol
values correspond to 50−70% of ΔGgbas and are larger in
magnitude than ΔGabq. This indicates that the solvation free
energy should be considered as a significant energy component
in the scoring function for the binding affinities of ABL
inhibitors. To enhance the potency of an ABL inhibitor with
structural modifications, therefore, the resulting increase in the
strength of enzyme−inhibitor interactions should be sufficient
to surmount the increased stabilization in solution.
The calculated binding free energies of the six inhibitors
compare reasonably well with their relative inhibitory activities
and structural features found in docking simulations. For
example, ΔGgbas and ΔGsol values of 7 are lower those of 1 by
2.3−2.8 and 1.0 kcal/mol, respectively, which leads to the
lowering of ΔGabq in going from 1 to 7. The larger decrease in
ΔGgbas than ΔGsol indicates that the strengthening of enzyme−
inhibitor interactions because of the substitution of ethylamine
group in 7 for the terminal methoxy group in 1 should be
sufficient to overcome the increase in desolvation cost for the
enzyme−inhibitor complex formation. This can be invoked to
explain the higher inhibitory activity of 7 than 1. The
enlargement of the thiophene-2-one group in 7 to 3,4-
dimethoxybenzene in 9 leads to the increase and the decrease
in ΔGgbas and ΔGsol values, respectively, which indicates the
weakening of enzyme−inhibitor interactions and the increase in
MD Simulations. We attempted to identify the dynamic
properties of enzyme−inhibitor complexes to elucidate the
difference in inhibitory activities of 28 and 29 by conducting
MD simulations of the wild type and T315I mutant of ABL in
complex with 28 and 29. The dynamic stabilities of the wild
type and T315I mutant in solution were evaluated by
calculating the time evolutions of RMSDinit values of backbone
Cα atoms from the starting structures of production dynamics
G
dx.doi.org/10.1021/ja311756u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
for ABL-28, T315I-28, ABL-29, and T315I-29 complexes. As
shown in Figure 7a, the RMSDinit values remain within 2.5 Å
during the entire course of simulation in all four cases,
exhibiting a convergent behavior with respect to the simulation
time. These results indicate that the protein structures can be
maintained stable in all four cases without a significant
conformational change under the simulation conditions
described above. It is also noteworthy that the RMSDinit values
for the heavy atoms of 28 and 29 fall into 1 Å and are
maintained lower than those of Cα atoms of the wild type and
T315I mutant of ABL throughout the entire simulation time
(Figure 7b). This suggests that the movements of 28 and 29
within the ATP-binding sites are insignificant when compared
to the conformational changes of the wild type and T315I
mutant, which is consistent with their high inhibitory activities.
Because the hydrogen bond interactions with backbone
groups of Met318 were found to be the most important
interactions to stabilize the inhibitors in the ATP-binding sites
of the wild type and T315I mutant, the difference in the
inhibitory activities of 28 and 29 may be elucidated by
comparing the dynamic properties of the three hydrogen bonds
shown in Figure 6. Figure 8 displays the time dependences of
the interatomic distances associated with the three hydrogen
bonds. A common feature of ABL-28, T315I-28, ABL-29, and
T315I-29 complexes is that the central hydrogen bond
established between the H2 atom of the inhibitors and the
aminocarbonyl oxygen of Met318 is dynamically most stable
with the time-averaged distances of 1.9−2.0 Å. This hydrogen
bond is found to be preserved for more than 95% of simulation
time in all four cases when the distance limit of N−H···O
hydrogen bond is assumed to be 2.5 Å. The hydrogen bond
between N1 atom of the inhibitors and amidic hydrogen of
Met318 also exhibits a dynamic stability in all four enzyme−
inhibitor complexes. It appears to be established for 94.2%,
93.9%, 92.4%, and 96.5% of residence time in ABL-28, T315I-
28, ABL-29, and T315I-29 complexes, respectively.
Figure 7. Time evolutions of RMSDinit values of (a) protein backbone
Cα atoms and (b) heavy atoms of 28 and 29 from the starting
structures of production dynamics.
Figure 8. Time evolutions of the interatomic distances associated with the hydrogen bond interactions of 28 and 29 with backbone amide group of
Met318 in (a) ABL-28, (b) T315I-28, (c) ABL-29, and (d) T315I-29 complexes. See Figure 6 for the identification of atoms.
H
dx.doi.org/10.1021/ja311756u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Table 3. Calculated Free Energy Changes in Solution (ΔGu) and in the ATP-Binding Sites of Wild Type (WT) and T315I
Mutant ABL (ΔGb), and the Change in Binding Free Energies (ΔΔGbind) Associated with the Structural Transformation
a
between the Two Structurally Similar Inhibitors
perturbation
1 → 7
9 → 11
11 → 28
28 → 29
ΔGu
ΔGb
−7.3 0.0
−8.3 0.0
−8.6 0.1
−1.0 0.0
−1.3 0.1
4.4 0.5
5.1 0.7
5.6 1.1
0.7 0.9
1.2 1.2
1.3 0.0
0.5 0.0
1.4 0.3
−1.6 0.4
−1.3 0.2
−3.0 0.5
−2.7 0.4
WT
T315I
WT
0.5 0.0
ΔΔGbind
−0.8 0.0
−0.8 0.0
T315I
a
All energy values are given in kcal/mol.
Another common feature of the four enzyme−inhibitor
additional hydrogen bond with the backbone amide group of
Met318. These results confirm the significant role of Met318 in
the stabilizing the inhibitor through the formation of multiple
hydrogen bonds. A similar result was also obtained in the
molecular modeling studies reported by Cortopassi and co-
workers.42
complexes lies in the dynamic instability of the hydrogen bond
between H1 atom of the inhibitors and the aminocarbonyl
oxygen of Met318. This hydrogen bond is located at the
entrance of the ATP-binding site and is exposed to bulk
solvent, which allows the hydrogen bond to be ruptured by
solvent molecules. It is found to be maintained only for 63.7
and 64.3% of simulation time in ABL-28 and T315I-28
complexes, respectively. However, the dynamic stability of the
hydrogen bond increases significantly with the change of the
inhibitor from 28 to 29. The H1···OC hydrogen bonds are
preserved for 85.6 and 87.8% of residence time in ABL-29 and
T315I-29 complexes, respectively. Therefore, the higher
inhibitory activity of 29 compared to 28 can be attributed to
the strengthening of the hydrogen bond between H1 atom and
the aminocarbonyl oxygen of Met318.
FEP Calculations. Table 3 lists the calculated ΔΔGbind
values with respect to the wild type and T315I mutant of ABL
between the two structurally similar nocodazole derivatives
shown in Table 1. Consistent with more than 20-fold increase
of the inhibitory activity in going from 1 to 7, the calculated
binding free energy (ΔGbind) values become more favorable in
1 → 7 perturbation by 1.0 and 1.3 kcal/mol with respect to the
wild type and T315I mutant of ABL, respectively. Because ΔGb
is more negative than ΔGu both for the wild type and for T315I
mutant, the increase of inhibitor potency in going from 1 to 7
can be attributed to the greater stabilization in the ATP-binding
sites than in solution with the change of the terminal methoxy
group in 1 to the ethylamine moiety in 7. This result also
indicates that to enhance the inhibitory activity, the structural
modification of an inhibitor should be made in such a way for
the strengthening of enzyme−inhibitor interactions to over-
balance the extra stabilization of the new inhibitor in solution.
In case of 9 → 11 perturbation, ΔΔGbind values become
unpredictable within the framework of FEP theory because of
the large statistical errors in both ΔGu and ΔGb. We note in
this regard that the two terminal methoxy moieties in 9 should
disappear in 11 with the additional change of a carbon on the
terminal phenyl ring to a nitrogen atom. Such a structural
variation seems to be too large to be described properly by FEP
calculations. On the other hand, all statistical errors fall into
0.05 kcal/mol in 11 → 28 FEP calculations. In this case the
magnitudes of ΔGu and ΔGb are also found to be smallest
among the four perturbations under consideration, which is not
surprising because the structural transformation of 11 into 28
involves only the change of the −NH− group on the
benzimidazole ring to a sulfur atom. The more positive value
of ΔGu than ΔGb in 11 → 28 perturbation implies that the
inhibitor can become more potent because it is less destabilized
in the ATP-binding sites than in aqueous solution. This result
exemplifies the importance of reducing the desolation cost for
enzyme−inhibitor complexation to improve the inhibitor
potency, which is consistent with the decomposition analysis
of binding free energies in Table 2.
The higher dynamic stability of the hydrogen bonds between
29 and the wild type and T315I mutant of ABL than those for
28 can be understood by comparing the RMSDinit values of the
inhibitors. As shown in Figure 7b, the RMSDinit values of 29 are
dynamically more stable than those of 28 when bound to the
ATP-binding site of the wild type and the mutant. The time-
averaged RMSDinit values amount to only 0.32 and 0.30 Å for
ABL-29 and T315I-29 complexes, as compared to 0.41 and
0.49 Å for ABL-28 and T315I-28 complexes, respectively.
Therefore, 29 is expected to be motionally more restricted in
the ATP-binding sites than 28. This indicates that 29 can be
bound in the ATP binding sites more tightly than 28, which is
consistent with the establishment of the stronger hydrogen
bonds in ABL-29 and T315I-29 complexes than in ABL-28 and
T315I-28 complexes. Thus, dynamic properties of enzyme−
inhibitor complexes provide useful information for designing
new potent inhibitors of the wild type and T315I mutant of
ABL as well as the binding free energies and the binding modes
estimated from docking simulations.
To further address the importance of the hydrogen bonds in
the potencies of the inhibitors, we compare the binding free
energies of ABL-29 and T315I-29 complexes including all three
hydrogen bonds with those of ABL-28 and T315I-28
complexes in which only two hydrogen bonds are involved.
For this purpose, additional docking simulations of 28 and 29
were carried out in the ATP-binding sites of the wild type and
T315I mutant of ABL. As the starting structures, we used the
MD trajectory snapshots for ABL-29 and T315I-29 complexes
in which 29 formed all three hydrogen bonds with Met318 and
those for ABL-28 and T315I-28 complexes in which one N−
H···O hydrogen bond was broken. The calculated binding free
energy appears to become more favorable from −12.1 to −13.0
kcal/mol in going from ABL-28 to ABL-29 complex. Similarly,
T315I-29 complex is found to be more stable than T315I-28
complex by 1.3 kcal/mol because of the establishment of an
The results of 28 → 29 FEP calculations differ from those of
binding free energy decomposition analysis in that ΔGu and
ΔGb values become positive and negative, respectively, as 28
transforms into 29. This indicates that the resulting improve-
ment of inhibitory activity can be attributed to the combined
effects of the destabilization of the inhibitor in bulk solvent and
the stabilization of the enzyme−inhibitor complex. Thus, the
I
dx.doi.org/10.1021/ja311756u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(4) Capdeville, R.; Buchdunger, E.; Zimmermann, J.; Matter, A. Nat.
Rev. Drug Discov. 2002, 1, 493.
(5) O’Hare, T.; Eide, C. A.; Deininger, M. W. Blood 2007, 110, 2242.
(6) Quintas-Cardama, A.; Kantarjian, H.; Cortes, J. Nat. Rev. Drug
Discovery 2007, 6, 834.
(7) Lombardo, L. J.; Lee, F. Y.; Chen, P.; Norris, D.; Barrish, J. C.;
Behnia, K.; Castaneda, S.; Cornelius, L. A. M.; Das, J.; Doweyko, A.
M.; Fairchild, C.; Hunt, J. T.; Inigo, I.; Johnston, K.; Kamath, A.; Kan,
D.; Klei, H.; Marathe, P.; Pang, S.; Peterson, R.; Pitt, S.; Schieven, G.
L.; Schmidt, R. J.; Tokarski, J.; Wen, M. L.; Wityak, J.; Borzilleri, R. M.
J. Med. Chem. 2004, 47, 6658.
results of FEP calculations for nocodazole derivatives confirm
that the enhancement of inhibitor potency can be achieved by
the reduction of the desolvation cost for enzyme−inhibitor
complexation, as well as by the strengthening of the interactions
between the inhibitor and the amino-acid residues in the ATP-
binding site.
CONCLUSIONS
■
We have identified new common inhibitors of the wild-type
and T315I mutant of ABL by applying a structure-based de
novo design using nocodazole as the inhibitor scaffold. Because
of the improvement of the scoring function by implementing a
proper solvation free energy term, the majority of the designed
derivatives are found to have a higher potency than nocodazole.
In particular, compound 29 reveals picomolar inhibitory activity
against both the wild type and T315I mutant, indicating that it
may serve as a new lead compound for the discovery of anti-
CML drugs. Decomposition analysis of the binding free
energies indicates that in order to enhance the potency of an
ABL inhibitor with structural modifications, the strengthening
of enzyme−inhibitor interactions should be greater than the
increase in desolvation cost for binding in the ATP-binding site.
Consistent with the previous experimental and computational
studies, the hydrogen bond interactions with backbone groups
of Met318 are found to be the most significant binding forces
associated with the stabilization the inhibitors in the ATP-
binding sites of the wild type and T315I mutant. Besides the
strength of the hydrogen bonds with Met318, dynamic
stabilities of the hydrogen bonds appear to be also important
for the effective inhibition of the wild type and T315I mutant of
ABL. The results of FEP calculations for nocodazole derivatives
confirm that the inhibitory activity can be enhanced not only by
the strengthening of the binding in the ATP-binding site but
also by the reduction of the desolvation cost for enzyme−
inhibitor complexation.
(8) Weisberg, E.; Manley, P. W.; Breitenstein, W.; Bruggen, J.;
̈
Cowan-Jacob, S. W.; Ray, A.; Huntly, B.; Fabbro, D.; Fendrich, G.;
Hall-Meyers, E.; Kung, A. L.; Mestan, J.; Daley, G. Q.; Callahan, L.;
Catley, L.; Cavazza, C.; Azam, M.; Neuberg, D.; Wright, R. D.;
Gilliland, D. G.; Griffin, J. D. Cancer Cell 2005, 7, 129.
(9) Kimura, S.; Naito, H.; Segawa, H.; Kuroda, J.; Yuasa, T.; Sato, K.;
Yokota, A.; Kamitsuji, Y.; Kawata, E.; Ashihara, E.; Nakaya, Y.;
Naruoka, H.; Wakayama, T.; Nasu, K.; Asaki, T.; Niwa, T.;
Hirabayashi, K.; Maekawa, T. Blood 2005, 106, 3948.
(10) Boschelli, D. H.; Ye, F.; Wang, Y. D.; Dutia, M.; Johnson, S. L.;
Wu, B.; Miller, K.; Powell, D. W.; Yaczko, D.; Young, M.; Tischler, M.;
Arndt, K.; Discafani, C.; Etienne, C.; Gibbons, J.; Grod, J.; Lucas, J.;
Weber, J. M.; Boschelli, F. J. Med. Chem. 2001, 44, 3965.
(11) Zhou, T.; Parillon, L.; Li, F.; Wang, Y.; Keats, J.; Lamore, S.; Xu,
Q.; Shakespeare, W.; Dalgarno, D.; Zhu, X. Chem. Biol. Drug Des.
2007, 70, 171.
(12) Giles, F. J.; Cortes, J.; Jones, D.; Bergstrom, D.; Kantarjian, H.;
Freedman, S. J. Blood 2007, 109, 500.
(13) Gontarewicz, A.; Balabanov, S.; Keller, G.; Colombo, R.;
Graziano, A.; Pesenti, E.; Benten, D.; Bokemeyer, C.; Fiedler, W.;
Moll, J.; Brummendorf, T. H. Blood 2008, 111, 4355.
̈
(14) Howard, S.; Berdini, V.; Boulstridge, J. A.; Carr, M. G.; Cross, D.
M.; Curry, J.; Devine, L. A.; Early, T. R.; Fazal, L.; Gill, A. L.;
Heathcote, M.; Maman, S.; Matthews, J. E.; McMenamin, R. L.;
Navarro, E. F.; O’Brien, M. A.; O’Reilly, M.; Rees, D. C.; Reule, M.;
Tisi, D.; Williams, G.; Vinkovic,
52, 379.
́
M.; Wyatt, P. G. J. Med. Chem. 2009,
(15) Shiotsu, Y.; Kiyoi, H.; Ishikawa, Y.; Tanizaki, R.; Shimizu, M.;
Umehara, H.; Ishii, K.; Mori, Y.; Ozeki, K.; Minami, Y.; Abe, A.;
Maeda, H.; Akiyama, T.; Kanda, Y.; Sato, Y.; Akinaga, S.; Naoe, T.
Blood 2009, 114, 1607.
ASSOCIATED CONTENT
■
S
* Supporting Information
(16) Quintas-Cardama, A.; Cortes, J. Clin. Cancer Res. 2008, 14, 4392.
(17) (a) Park, H.; Hong, S.; Hong, S. J. Comput.-Aided Mol. Des.
2012, 26, 983. (b) Choi, H. G.; Ren, P.; Adrian, F.; Sun, F.; Lee, H. S.;
Wang, X.; Ding, Q.; Zhang, G.; Xie, Y.; Zhang, J.; Liu, Y.; Tuntland,
T.; Warmuth, M.; Manley, P. W.; Mestan, J.; Gray, N. S.; Sim, T. J.
Med. Chem. 2010, 53, 5439. (c) Huang, W. S.; Metcalf, C. A.;
Sundaramoorthi, R.; Wang, Y.; Zou, D.; Thomas, R. M.; Zhu, X.; Cai,
L.; Wen, D.; Liu, S.; Romero, J.; Qi, J.; Chen, I.; Banda, G.; Lentini, S.
P.; Das, S.; Xu, Q.; Keats, J.; Wang, F.; Wardwell, S.; Ning, Y.;
Snodgrass, J. T.; Broudy, M. I.; Russian, K.; Zhou, T.; Commodore, L.;
Narasimhan, N. I.; Mohemmad, Q. K.; Iuliucci, J.; Rivera, V. M.;
Dalgarno, D. C.; Sawyer, T. K.; Clackson, T.; Shakespeare, W. C. J.
Med. Chem. 2010, 53, 4701. (d) Thomas, M.; Huang, W. S.; Wen, D.;
Zhu, X.; Wang, Y.; Metcalf, C. A.; Liu, S.; Chen, I.; Romero, J.; Zou,
D.; Sundaramoorthi, R.; Li, F.; Qi, J.; Cai, L.; Zhou, T.; Commodore,
L.; Xu, Q.; Keats, J.; Wang, F.; Wardwell, S.; Ning, Y.; Snodgrass, J. T.;
Broudy, M. I.; Russian, K.; Iuliucci, J.; Rivera, V. M.; Sawyer, T. K.;
Dalgarno, D. C.; Clackson, T.; Shakespeare, W. C. Bioorg. Med. Chem.
Lett. 2011, 21, 3743. (e) Deng, X.; Lim, S. M.; Zhang, J.; Gray, N. S.
Bioorg. Med. Chem. Lett. 2010, 20, 4196. (f) Cao, J.; Fine, R.; Gritzen,
C.; Hood, J.; Kang, X.; Klebansky, B.; Lohse, D.; Mak, C. C.;
McPherson, A.; Noronha, G.; Palanki, M. S. S.; Pathak, V. P.; Renick,
J.; Soll, R.; Zeng, B.; Zhu, H. Bioorg. Med. Chem. Lett. 2007, 17, 5812.
(18) (a) Gambacorti-Passerini, C.; Gasser, M.; Ahmed, S.; Assouline,
Experimental procedures, characterization data, and copies of
1H and 13C NMR spectra for all compounds. This material is
available free of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Author
hspark@sejong.ac.kr; hongorg@kaist.ac.kr
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
Dedicated to Professor Deukjoon Kim on the occasion of his
retirement. This research was supported by National Research
Foundation of Korea (NRF) funded by the Ministry of
Education, Science and Technology (NRF-2012-0008440,
2011-0016436, 2011-0020322) and the Institute for Basic
Science (IBS).
REFERENCES
■
(1) Quintas-Cardama, A.; Cortes, J. Blood 2009, 113, 1619.
(2) Eck, M. J.; Manley, P. W. Curr. Opin.Cell Biol. 2009, 21, 288.
(3) Schenone, S.; Bruno, O.; Radi, M.; Botta, M. Med. Res. Rev. 2010,
31, 1.
S.; Scapozza, L. Leukemia 2005, 19, 1267. (b) Lu, S.; Luo, Q.; Hao, X.;
̈
Li, X.; Ji, L.; Zheng, W.; Wang, F. Bioorg. Med. Chem. Lett. 2011, 21,
6964. (c) Wang, D.; Zhang, Z.; Lu, X.; Feng, Y.; Luo, K.; Gan, J.;
J
dx.doi.org/10.1021/ja311756u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Yingxue, L.; Wan, J.; Li, X.; Zhang, F.; Tu, Z.; Cai, Q.; Ren, X.; Ding,
K. Bioorg. Med. Chem. Lett. 2011, 21, 1965.
(19) Park, H.; Hong, S.; Hong, S. ChemMedChem 2012, 7, 53.
(20) Warren, G. L.; Andrews, C. W.; Capelli, A. M.; Clarke, B.;
LaLonde, J.; Lambert, M. H.; Lindvall, M.; Nevins, N.; Semus, S. F.;
Senger, S.; Tedesco, G.; Wall, I. D.; Woolven, J. M.; Peishoff, C. E.;
Head, M. S. J. Med. Chem. 2006, 49, 5912.
(21) Shoichet, B. K.; Leach, A. R.; Kuntz, I. D. Proteins 1999, 34, 4.
(22) Tokarski, J. S.; Newitt, J. A.; Chang, C. Y. J.; Cheng, J. D.;
Wittekind, M.; Kiefer, S. E.; Kish, K.; Lee, F. Y. F.; Borzillerri, R.;
Lombardo, L. J.; Xie, D.; Zhang, Y.; Klei, H. E. Cancer Res. 2006, 66,
5790.
(23) Jeffrey, G. A. An Introduction to Hydrogen Bonding; Oxford
University Press: Oxford,U.K., 1997.
(24) Wang, R.; Gao, Y.; Lai, L. J. Mol. Model. 2000, 6, 498.
(25) Wang, R.; Liu, L.; Lai, L. J. Mol. Model. 1998, 6, 379.
(26) Lipinski, C. A.; Lombardo, F.; Dominy, B. W.; Feeney, P. J. Adv.
Drug Delivery Rev. 1997, 23, 3.
(27) Morris, G. M.; Goodsell, D. S.; Halliday, R. S.; Huey, R.; Hart,
W. E.; Belew, R. K.; Olson, A. J. J. Comput. Chem. 1998, 19, 1639.
(28) Park, H.; Jeon, J. H. Phys. Rev. E 2007, 75, 021916.
(29) Gasteiger, J.; Marsili, M. Tetrahedron 1980, 36, 3219.
(30) Stouten, P. F. W.; Frommel, C.; Nakamura, H.; Sander, C. Mol.
̈
Simul. 1993, 10, 97.
(31) Choi, H.; Kang, H.; Park, H. J. Cheminformatics 2013, 5, 8.
(32) Park, H.; Chi, O.; Kim, J.; Hong, S. J. Chem. Inf. Model. 2011, 51,
2986.
(33) The IC50 determinations were performed at the Reaction
Biology Corp. See the Supporting Information for details.
(34) Case, D. A.; Cheatham, T. E., III; Darden, T.; Gohlke, H.; Luo,
R.; Merz, K. M., Jr.; Onufriev, A.; Simmerling, C.; Wang, B.; Woods, R.
J. J. Comput. Chem. 2005, 26, 1668.
(35) Fox, T.; Kollman, P. A. J. Phys. Chem. B 1998, 102, 8070.
(36) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R.
W.; Klein, M. L. J. Chem. Phys. 1983, 79, 926.
(37) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.;
DiNola, A.; Haak, J. R. J. Chem. Phys. 1984, 81, 3684.
(38) Ryckaert, J.-P.; Ciccotti, G.; Berendsen, H. J. C. J. Comput. Phys.
1977, 23, 327.
(39) O’Hare, T.; Heide, C. A.; Deininger, M. W. Expert Opin. Invest.
Drugs 2008, 17, 865.
(40) Levinson, N. M.; Boxer, S. G. PloS One 2012, 7, e29828.
(41) Li, C.-M.; Chen, J.; Lu, Y.; Narayanan, R.; Parke, D. N.; Li, W.;
Ahn, S.; Miller, D. D.; Dalton, J. T. Drug Metab. Dispos. 2011, 39,
1833.
(42) Cortopassi, W. A.; Feital, R. J. C.; Medeiros, D. J.; Guizado, T.
R. C.; Franca̧ , T. C. C.; Pimentel, A. S. Mol. Simul. 2012, 38, 1132.
K
dx.doi.org/10.1021/ja311756u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX