Discovery of a novel chemotype of tyrosine kinase inhibitors by fragment-based docking and molecular dynamics

Article abstract of DOI:10.1021/ml3001984

We have discovered a novel chemical class of inhibitors of the EphB4 tyrosine kinase by fragment-based high-throughput docking followed by explicit solvent molecular dynamics simulations for assessment of the binding mode. The synthesis of a single derivative (compound 7) of the hit identified in silico has resulted in an improvement of the inhibitory potency in an enzymatic assay from 8.4 μM to 160 nM and a ligand efficiency of 0.39 kcal/mol per non-hydrogen atom. Such remarkable improvement in affinity is due to an additional hydroxyl group involved in two favorable (buried) hydrogen bonds as predicted by molecular dynamics and validated by the crystal structure of the complex with EphA3 solved at 1.7 a resolution.

Full text of DOI:10.1021/ml3001984

Letter
pubs.acs.org/acsmedchemlett
Discovery of a Novel Chemotype of Tyrosine Kinase Inhibitors by
Fragment-Based Docking and Molecular Dynamics
Hongtao Zhao,† Jing Dong,† Karine Lafleur,‡ Cristina Nevado,*,‡ and Amedeo Caflisch*,†
‡
†Department of Biochemistry and Department of Organic Chemistry, University of Zurich, Winterthurerstrasse 190, CH-8057
Zurich, Switzerland
S
* Supporting Information
ABSTRACT: We have discovered a novel chemical class of
inhibitors of the EphB4 tyrosine kinase by fragment-based
high-throughput docking followed by explicit solvent molec-
ular dynamics simulations for assessment of the binding mode.
The synthesis of a single derivative (compound 7) of the hit
identified in silico has resulted in an improvement of the
inhibitory potency in an enzymatic assay from 8.4 μM to 160
nM and a ligand efficiency of 0.39 kcal/mol per non-hydrogen
atom. Such remarkable improvement in affinity is due to an
additional hydroxyl group involved in two favorable (buried)
hydrogen bonds as predicted by molecular dynamics and
validated by the crystal structure of the complex with EphA3 solved at 1.7 Å resolution.
KEYWORDS: In silico screening, EphB4 kinase, angiogenesis, cancer, explicit solvent MD
he Eph-ephrin system, including the EphA2 and EphB4
Treceptors, plays a critical role in tumor and vascular
functions during carcinogenesis.1,2 Recently, it has been shown
that delivery of chemotherapeutic drugs by an EphA2 targeting
peptide into EphA2-expressing cancer cells led to dramatically
improved efficacy in inhibiting tumor growth.3 So far, a few Eph
inhibitors have been identified, including the marketed drug
Dasatinib (Figure S1 in the Supporting Information).4−12
Although their role is still controversial for certain types of
cancer, e.g., non small cell lung cancer,13 the identification of
selective inhibitors of Eph tyrosine kinases will help to elucidate
their involvement in deregulated signaling.
Previously, we have developed an efficient in silico procedure
called ALTA, which stays for anchor-based library tailoring
approach, to interrogate a library of compounds for high-
throughput docking.14 First, small and mainly rigid virtual
fragments are docked in the binding site. The fragments with
most favorable calculated binding free energy (anchors) are
used to identify the compounds with 2D structure containing
one of these anchors, which are then submitted to flexible-
ligand docking. In this letter, we report a new approach for in
silico screening based on the synergistic combination of the
ALTA procedure for docking followed by explicit solvent
molecular dynamics simulations for further validation of the
binding poses.
The flowchart of the ALTA procedure is shown in Figure 1.
First, the nearly 9 million compounds in the ZINC-all now
library15 (version of August 2011) were decomposed into
563,774 fragments by in house developed software (Figure S2
in the Supporting Information). Just like its in vitro counterpart
of fragment-based drug discovery,16,17 the success of the ALTA
Figure 1. Flowchart of the ALTA virtual screening approach for the
tyrosine kinase EphB4.
in silico screening approach depends on the choice of
fragments. The use of virtual fragments by computational
decomposition of a real compound library offers opportunities
to explore a much greater fragmental space, with no limitations
in availability. To obtain fragments with high chemical richness
Received: July 16, 2012
Accepted: August 23, 2012
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ml3001984 | ACS Med. Chem. Lett. XXXX, XXX, XXX−XXX

ACS Medicinal Chemistry Letters
Letter
Table 1. Novel EphB4 Kinase Inhibitors Discovered by the ALTA Virtual Screening Approach with Hit Optimization Colored
in Green
a
b
Enzymatic assay based on FRET carried out in house using Invitrogen EphB4 assay kit at an ATP concentration of 30 μM. Enzymatic assay with
c
radioactive ATP at 1 μM concentration carried out at Reaction Biology Corp. The IC50 values were determined in a 10-dose response. Ligand
efficiency in kcal/mol per heavy atom. Percentage inhibition at compound concentration of 30 μM and ATP concentration of 30 μM. N.A.: not
d
e
available.
that can serve as a starting point either directly for hit
optimization or for identification of their “parent” compounds,
we developed a new decomposition protocol whose main
difference from our previous approach18 is the preservation of
longer substituents (e.g., N-methylurea) on ring systems
(details of algorithm explained in Figure S2 in the Supporting
Information). Most of the fragments obtained by the new
decomposition algorithm have a molecular weight ranging
between 150 and 300 g/mol, possess fewer than five rotatable
bonds, and do not have any formal charge (Figure S3 in the
Supporting Information). Second, this set of fragments was
reduced to a kinase-focused library of 63,252 fragments by
retaining only those with molecular weight smaller than 300 g/
mol, a maximum of three rotatable bonds, more than one ring,
and the capability to form two hydrogen bonds with the
backbone polar groups of the so-called hinge region. For the
latter criterion acidic CH groups (e.g., in aromatic rings) were
also considered as donors. The requirement of having more
than one ring helps to direct the search toward chemical space
less affected by the crowded intellectual property coverage,
given the diversity in fused rings. Moreover, one-ring anchor
fragments are too small for providing enough binding energy.
Use of a target-focused fragment library is computationally
more efficient than docking the entire library of fragments.
The kinase-focused fragment library was then docked into
the ATP binding site of EphB4 (PDB code 2VWX) by
AutoDock4,19 followed by ranking according to a previously
reported scoring function.8 The 998 fragments with an
estimated binding energy lower than −5 kcal/mol were used
to identify their “parent” compounds in the ZINC-all now
library which yielded a total of 19,427 compounds. Flexible-
ligand docking of these compounds followed by scoring8 and
elimination of those mentioned in patents related to kinases
resulted in four scaffolds, which were further investigated by
explicit solvent molecular dynamics simulations using the all-
atom CHARMM PARAM22 force field20 and the TIP3P model
of water.21 Two molecules sharing the pyrimidoisoquinolinone
scaffold showed stable intermolecular hydrogen bonds with the
hinge region (bottom panels of Figure S4 in the Supporting
Information). Thus, commercially available compounds 1−6,
bearing a pyrimidoisoquinolinone scaffold, were purchased and
tested (Table 1). Of these six compounds, 1, 2, and 5 showed
inhibition of the EphB4 tyrosine kinase in the low micromolar
range in two different enzymatic assays based on FRET (carried
out in house) and radioactive ATP (performed at Reaction
Biology Corp.), respectively.
The previously reported optimization of another chemical
class of EphB4 inhibitors discovered by the ALTA approach7,14
suggested to us that the addition of a hydroxyl group in
position 5 of the phenyl ring in compound 1 would increase the
binding affinity. This hydroxyl group is expected to be involved
in two additional hydrogen bonds with the side chain of Glu664
and the backbone NH of the Asp in the DFG motif, as
observed in so-called type I1/2 inhibitors22 of tyrosine kinase
with small gatekeeper residue (e.g., Thr) and DFG-in
conformation. As predicted, compound 7 shows four stable
hydrogen bonds with the ATP binding site (two with the hinge
region and two involving the additional hydroxyl group) in two
300-ns molecular dynamics simulations with explicit solvent
(Figure 2). Note that these molecular dynamics runs were
started from the binding mode obtained by docking and were
carried out before chemical synthesis of compound 7.
The synthesis of compound 7 started with the saponification
of ethyl 2-oxocyclohexanecarboxylate 8 in the presence of
sodium hydroxide. Reaction of the β-keto acid 9 with acetone
in the presence of acetic anhydride and concentrated sulfuric
B
dx.doi.org/10.1021/ml3001984 | ACS Med. Chem. Lett. XXXX, XXX, XXX−XXX

ACS Medicinal Chemistry Letters
Letter
Figure 2. 2D illustration of binding mode of inhibitor 7 in the ATP binding site of EphB4 (top) and time series of intermolecular hydrogen bonds,
i.e., distances between donor and acceptor atoms, in two molecular dynamics simulations started with different random values of the initial atomic
velocities (bottom). The choice of colors is consistent in top and bottom panels while the green horizontal lines in the latter are drawn to illustrate
that the hydrogen bonds are always formed except for transient ruptures.
a
Scheme 1. Synthesis of Tautomers 11 and 11′
a
Reagents and conditions: (a) NaOH, water, 25 °C, 12 h, 61%; (b) acetone, H2SO4, acetic anhydride, 0 °C, 4 h, 93%; (c) 5-methoxy-2-
methylaniline, o-xylene, MW, 150 °C, 5 min, 46%.
a
Scheme 2. Synthesis of Target Compound 7
a
Reagents and conditions: (a) malononitrile, piperidine, EtOH, 100 °C, 2 h, 82%; (b) formamide, 210 °C, 7 h, 63%; (c) BBr3, CH2Cl2, 25 °C, 16 h,
28%.
acid provided the activated acetal 10,23 which reacted with 5-
methoxy-2-methylaniline under microwave irradiation to afford
a tautomeric mixture of amides 11 and 11′ (Scheme 1).24
Reaction of the mixture with malononitrile provided
intermediate 12, which upon condensation with formamide
afforded compound 13.25 Final demethylation for the methoxy
group in the aromatic ring was obtained in the presence of BBr3
(Scheme 2).
It was gratifying to observe that in the enzymatic assay the
potency of compound 7 is about 50 times higher than that of
the parent compound 1 (Table 1). This improvement
corresponds to a difference in binding free energy of about
2.5 kcal/mol, which is consistent with the two additional
hydrogen bonds of the hydroxyl group of 7 as predicted by the
docking and molecular dynamics simulations. The final
validation was obtained by the determination of the crystal
C
dx.doi.org/10.1021/ml3001984 | ACS Med. Chem. Lett. XXXX, XXX, XXX−XXX

ACS Medicinal Chemistry Letters
Letter
structure of the complex between the catalytic domain of
EphA3 and inhibitor 7 (resolution of 1.7 Å, see Supporting
Information for details of X-ray crystallography). The binding
mode in the crystal structure confirmed the predicted binding
pose of inhibitor 7 into EphB4 (DFG-in conformation) and the
four intermolecular hydrogen bonds observed in the molecular
dynamics runs (Figure 3). The EphA3 tyrosine kinase was used
with compound 7 has definitively validated the binding mode
predicted in silico.
ASSOCIATED CONTENT
* Supporting Information
■
S
Chemical synthesis, crystallization, similarity comparison with
known Eph inhibitors, description of the algorithm used to
decompose a compound library, distribution of physicochem-
ical properties of the library of anchor fragments, and MD
screening of candidate compounds. This material is available
free of charge via the Internet at http://pubs.acs.org.
Accession Codes
The atomic coordinates and structure factors of EphA3 in
complex with compound 7 have been deposited with the
Protein Data Bank as entry 4G2F.
AUTHOR INFORMATION
Corresponding Author
*C.N.: phone, (41) 446353945; fax, (41) 446353948; e-mail,
nevado@oci.uzh.ch. A.C.: phone, (41) 446355521; fax, (41)
446356862; e-mail, caflisch@bioc.uzh.ch.
■
Figure 3. (Left) Crystal structure of inhibitor 7 bound to the kinase
domain of EphA3. Dashed lines represent intermolecular hydrogen
bonds. (Right) Superposition of pose of 7 in EphB4 obtained by
docking and molecular dynamics (carbon atoms in magenta) onto the
pose of 7 in EphA3 obtained by X-ray crystallography (carbon atoms
in cyan). The overlap was obtained by structural alignment of the
crystal structure of the EphA3/inhibitor 7 complex to the in silico
predicted complex of 7 and EphB4 using only the Cα atoms of the two
kinase domains.
Funding
This work was supported by grants of the Swiss National
Science Foundation and the Sino-Swiss Science and Technol-
ogy Cooperation to A.C.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank Dr. Danzhi Huang for interesting discussions and Mr.
Tim Knehans for the help on calculation of similarity index. We
thank Xiao-Dan Li and Chitra Rajendran (Paul Scherrer
Institute) for interesting discussions and help with X-ray data
collection in a preliminary phase of this work. The simulations
for determination of the crystal structure because of its
straightforward expression in Escherichia coli10 while expression
of EphB4 has been reported only in insect cells.5 Note that 32
of the 36 residues in the ATP binding site of EphA3 are
identical to those of EphB4.26 Moreover, the side chains
involved in binding compound 7 are identical in EphA3 and
EphB4.
were carried out on the Schrodinger cluster at the University of
Zurich.
̈
The affinity of compound 7 was further tested on a panel of
five tyrosine kinases, known as validated drug targets in cancer
therapy. The IC50 value in enzymatic assay in the presence of
the recombinant catalytic domain and 1 μM radiolabeled ATP
is 0.338 μM for Src, 0.864 μM for Abl1, 1.38 μM for Lck, 1.62
μM for EGFR, while no inhibition was observed for IGF1R.
Thus compound 7 has higher affinity for EphB4 than for these
five tyrosine kinases.
Compound 7 was also tested in a cell-based assay. The
cellular EphB4 phosphorylation assay (performed at ProQinase
GmbH) makes use of the murine embryonal fibroblast cell line
(MEF), which expresses a high level of exogenously introduced
full-length human EphB4. Stimulation of these cells with 2 μg/
mL soluble human Ephrin-B2/Fc chimera for 2 h results in
receptor tyrosine autophosphorylation. Compound 7 shows a
cellular IC50 of 0.93 μM, indicating its potential for future
optimization.
In conclusion, we have discovered a new chemical class of
kinase inhibitors by fragment-based docking into the ATP-
binding site of the EphB4 kinase and molecular dynamics
simulations for further assessment of the binding mode. The
chemical synthesis of a single derivative (compound 7) by
addition of just one hydroxyl group, designed on the basis of
the simulation results, has resulted in a novel chemotype of
nanomolar inhibitors of the EphB4 tyrosine kinase. Finally, the
determination of the crystal structure of the complex of EphA3
ABBREVIATIONS
■
EphB4, erythropoietin producing human hepatocellular carci-
noma receptor B4; ALTA, anchor-based library tailoring
approach; MD, molecular dynamics
REFERENCES
■
(1) Noberini, R.; Lamberto, I.; Pasquale, E. B. Targeting Eph
receptors with peptides and small molecules: progress and challenges.
Semin. Cell Dev. Biol. 2012, 23, 51−7.
(2) Tognolini, M.; Incerti, M.; Hassan-Mohamed, I.; Giorgio, C.;
Russo, S.; Bruni, R.; Lelli, B.; Bracci, L.; Noberini, R.; Pasquale, E. B.;
Barocelli, E.; Vicini, P.; Mor, M.; Lodola, A. Structure-activity
relationships and mechanism of action of Eph-ephrin antagonists:
interaction of cholanic acid with the EphA2 receptor. ChemMedChem
2012, 7, 1071−83.
(3) Wang, S.; Placzek, W. J.; Stebbins, J. L.; Mitra, S.; Noberini, R.;
Koolpe, M.; Zhang, Z.; Dahl, R.; Pasquale, E. B.; Pellecchia, M. Novel
targeted system to deliver chemotherapeutic drugs to EphA2-
expressing cancer cells. J. Med. Chem. 2012, 55, 2427−36.
(4) Miyazaki, Y.; Nakano, M.; Sato, H.; Truesdale, A. T.; Stuart, J. D.;
Nartey, E. N.; Hightower, K. E.; Kane-Carson, L. Design and effective
synthesis of novel templates, 3,7-diphenyl-4-amino-thieno and furo-
[3,2-c]pyridines as protein kinase inhibitors and in vitro evaluation
targeting angiogenetic kinases. Bioorg. Med. Chem. Lett. 2007, 17, 250−
4.
(5) Bardelle, C.; Cross, D.; Davenport, S.; Kettle, J. G.; Ko, E. J.;
Leach, A. G.; Mortlock, A.; Read, J.; Roberts, N. J.; Robins, P.;
D
dx.doi.org/10.1021/ml3001984 | ACS Med. Chem. Lett. XXXX, XXX, XXX−XXX

ACS Medicinal Chemistry Letters
Letter
Williams, E. J. Inhibitors of the tyrosine kinase EphB4. Part 1:
Structure-based design and optimization of a series of 2,4-bis-
anilinopyrimidines. Bioorg. Med. Chem. Lett. 2008, 18, 2776−80.
(6) Sun, L.; Liang, C.; Shirazian, S.; Zhou, Y.; Miller, T.; Cui, J.;
Fukuda, J. Y.; Chu, J. Y.; Nematalla, A.; Wang, X.; Chen, H.; Sistla, A.;
Luu, T. C.; Tang, F.; Wei, J.; Tang, C. Discovery of 5-[5-fluoro-2-oxo-
1,2- dihydroindol-(3Z)-ylidenemethyl]-2,4- dimethyl-1H-pyrrole-3-
carboxylic acid (2-diethylaminoethyl)amide, a novel tyrosine kinase
inhibitor targeting vascular endothelial and platelet-derived growth
factor receptor tyrosine kinase. J. Med. Chem. 2003, 46, 1116−9.
(7) Lafleur, K.; Huang, D.; Zhou, T.; Caflisch, A.; Nevado, C.
Structure-based optimization of potent and selective inhibitors of the
tyrosine kinase erythropoietin producing human hepatocellular
carcinoma receptor B4 (EphB4). J. Med. Chem. 2009, 52, 6433−46.
(8) Zhao, H.; Huang, D. Hydrogen bonding penalty upon ligand
binding. PLoS One 2011, 6, e19923.
(23) Sato, M.; Ogasawara, H.; Oi, K.; Kato, T. Synthesis of 1,3-
Dioxin-4-One Derivatives. Chem. Pharm. Bull. 1983, 31, 1896−1901.
(24) Miriyala, B.; Williamson, J. S. An efficient and rapid synthesis of
beta-carboxamide derivatives using 2,2-dimethyl-2H,4H-1,3-dioxin-4-
ones by microwave irradiation. Tetrahedron Lett. 2003, 44, 7957−7959.
(25) Bogdanowiczszwed, K.; Policht, A. The Reaction of
Malononitrile with Enamines of Some Cyclic Beta-Keto Acid Anilides
and Amidines. J. Prakt. Chem. 1984, 326, 721−728.
(26) Huang, D.; Zhou, T.; Lafleur, K.; Nevado, C.; Caflisch, A.
Kinase selectivity potential for inhibitors targeting the ATP binding
site: a network analysis. Bioinformatics 2010, 26, 198−204.
(9) van Linden, O. P.; Farenc, C.; Zoutman, W. H.; Hameetman, L.;
Wijtmans, M.; Leurs, R.; Tensen, C. P.; Siegal, G.; de Esch, I. J.
Fragment based lead discovery of small molecule inhibitors for the
EPHA4 receptor tyrosine kinase. Eur. J. Med. Chem. 2012, 47, 493−
500.
(10) Choi, Y.; Syeda, F.; Walker, J. R.; Finerty, P. J., Jr.; Cuerrier, D.;
Wojciechowski, A.; Liu, Q.; Dhe-Paganon, S.; Gray, N. S. Discovery
and structural analysis of Eph receptor tyrosine kinase inhibitors.
Bioorg. Med. Chem. Lett. 2009, 19, 4467−70.
(11) Martiny-Baron, G.; Holzer, P.; Billy, E.; Schnell, C.; Brueggen,
J.; Ferretti, M.; Schmiedeberg, N.; Wood, J. M.; Furet, P.; Imbach, P.
The small molecule specific EphB4 kinase inhibitor NVP-BHG712
inhibits VEGF driven angiogenesis. Angiogenesis 2010, 13, 259−67.
(12) Zhao, H.; Huang, D.; Caflisch, A. Discovery of tyrosine kinase
inhibitors by docking into an inactive kinase conformation generated
by molecular dynamics. ChemMedChem 2012, DOI: 10.1002/
cmdc.201200331.
(13) Li, G.; Ji, X. D.; Gao, H.; Zhao, J. S.; Xu, J. F.; Sun, Z. J.; Deng,
Y. Z.; Shi, S.; Feng, Y. X.; Zhu, Y. Q.; Wang, T.; Li, J. J.; Xie, D. EphB3
suppresses non-small-cell lung cancer metastasis via a PP2A/RACK1/
Akt signalling complex. Nat. Commun. 2012, 3, 667.
(14) Kolb, P.; Kipouros, C. B.; Huang, D.; Caflisch, A. Structure-
based tailoring of compound libraries for high-throughput screening:
discovery of novel EphB4 kinase inhibitors. Proteins 2008, 73, 11−8.
(15) Irwin, J. J.; Shoichet, B. K. ZINCa free database of
commercially available compounds for virtual screening. J. Chem. Inf.
Model. 2005, 45, 177−82.
(16) Congreve, M.; Chessari, G.; Tisi, D.; Woodhead, A. J. Recent
developments in fragment-based drug discovery. J. Med. Chem. 2008,
51, 3661−80.
(17) Roughley, S. D.; Hubbard, R. E. How well can fragments explore
accessed chemical space? A case study from heat shock protein 90. J.
Med. Chem. 2011, 54, 3989−4005.
(18) Kolb, P.; Caflisch, A. Automatic and efficient decomposition of
two-dimensional structures of small molecules for fragment-based
high-throughput docking. J. Med. Chem. 2006, 49, 7384−92.
(19) Goodsell, D. S.; Olson, A. J. Automated docking of substrates to
proteins by simulated annealing. Proteins 1990, 8, 195−202.
(20) Brooks, B. R.; Brooks, C. L., 3rd; Mackerell, A. D., Jr.; Nilsson,
L.; Petrella, R. J.; Roux, B.; Won, Y.; Archontis, G.; Bartels, C.;
Boresch, S.; Caflisch, A.; Caves, L.; Cui, Q.; Dinner, A. R.; Feig, M.;
Fischer, S.; Gao, J.; Hodoscek, M.; Im, W.; Kuczera, K.; Lazaridis, T.;
Ma, J.; Ovchinnikov, V.; Paci, E.; Pastor, R. W.; Post, C. B.; Pu, J. Z.;
Schaefer, M.; Tidor, B.; Venable, R. M.; Woodcock, H. L.; Wu, X.;
Yang, W.; York, D. M.; Karplus, M. CHARMM: the biomolecular
simulation program. J. Comput. Chem. 2009, 30, 1545−614.
(21) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R.
W.; Klein, M. L. Comparison of Simple Potential Functions for
Simulating Liquid Water. J. Chem. Phys. 1983, 79, 926−935.
(22) Zuccotto, F.; Ardini, E.; Casale, E.; Angiolini, M. Through the
“gatekeeper door”: exploiting the active kinase conformation. J. Med.
Chem. 2010, 53, 2681−94.
E
dx.doi.org/10.1021/ml3001984 | ACS Med. Chem. Lett. XXXX, XXX, XXX−XXX