Discovery of potent vascular endothelial growth factor receptor-2 inhibitors

Article abstract of DOI:10.1002/cmdc.200900373

Substantial evidence over the last decades has implicated uncontrolled angiogenesis with various pathological states, including cancer. Vascular endothelial growth factor (VEGF) plays a critical role in its regulation. Because the tyrosine kinase VEGF rec

Full text of DOI:10.1002/cmdc.200900373

MED
DOI: 10.1002/cmdc.200900373
Discovery of Potent Vascular Endothelial Growth Factor
Receptor-2 Inhibitors
Athanasios Papakyriakou,[a] Maria E. Katsarou,[a] Maria Belimezi,[b] Michael Karpusas,[c] and
Dionisios Vourloumis*[a]
Substantial evidence over the last decades has implicated un-
controlled angiogenesis with various pathological states, in-
cluding cancer. Vascular endothelial growth factor (VEGF) plays
a critical role in its regulation. Because the tyrosine kinase
VEGF receptor-2 (VEGFR-2) is the major mediator of the mito-
genic, angiogenic, and permeability-enhancing effects of VEGF,
it has become one of the most profound anti-angiogenesis tar-
gets. Inspired by the anthranilamide class of VEGFR-2 inhibi-
tors, we performed a computational analysis of some potent
representative members, using docking and molecular dynam-
ics calculations. Based on the observations drawn from intro-
ducing the effect of the receptor’s flexibility in implicit aque-
ous environment, we designed, synthesized, and characterized
several new analogues of related scaffolds with modifications
in their steric and electronic characteristics. In vitro evaluation
of these compounds revealed several novel VEGFR-2 inhibitors
that are less cytotoxic and more potent than the parent com-
pounds.
Introduction
Vascular endothelial growth factor (VEGF) plays a critical role in
the regulation of angiogenesis, the process that leads to the
formation of new blood vessels.[1] Substantial evidence over
the last decades implicates uncontrolled angiogenesis with
various pathological states such as psoriasis, diabetic retino-
pathy, rheumatoid arthritis, chronic inflammation, and cancer.[2]
VEGF exerts its cellular effects mainly by interacting with two
high-affinity transmembrane tyrosine kinase receptors (RTKs):
VEGFR-1 (Flt-1) and VEGFR-2 (KDR human/Flk-1 mouse).[3] In-
hibition of VEGF or its receptor signaling system is therefore
an attractive target for therapeutic intervention,[4,5] as exempli-
fied by the recent USFDA approval of a humanized anti-VEGF
monoclonal antibody (bevacizumab, Avastinꢀ) as a first-line
treatment for metastatic colorectal cancer, in combination with
chemotherapy.[6]
Based on the universally accepted hypothesis that VEGFR-2
is the major mediator of the mitogenic, angiogenic, and per-
meability-enhancing effects of VEGF,[5] the specific receptor has
become one of the most profound anti-angiogenesis targets.[11]
Consequently, a significant number of small-molecule inhibi-
tors has been developed during the last few years,[12–15] com-
prising derivatives of phthalazines, anthranilamides, disubsti-
tuted ureas, indolin-2-ones, quinazolines, quinolines, pyrimi-
dines, pyridines, triazines, thiazoles, oxazoles, imidazoles, inda-
zoles, indenopyrrolocarbazolones, and other miscellaneous
compounds.[15] The availability of high-resolution X-ray crystal
structures of the kinase domain of human VEGFR-2,[16] as well
as co-crystallization data of the receptor with small-molecule
binders,[17–24] has greatly enhanced important structure–activity
relationship studies.
Both VEGFR-1 and VEGFR-2 have seven Ig-like extracellular
domains, a single-transmembrane region, and a consensus ty-
rosine kinase sequence that is interrupted by a kinase insert
domain.[7] Their kinase domains are highly conserved, with
about 70% sequence identity, whereas the C-terminal region is
the most differentiated.[7] On the surface of vascular endothe-
lial cells, VEGFR-2 undergoes dimerization upon strong binding
of VEGF homodimers, albeit with lower affinity than VEGFR-1.[5]
The activity of both receptors is mediated by ligand-depen-
dent tyrosine autophosphorylation in the intracellular kinase
domain which catalyzes the phosphorylation of cytosolic sub-
strate proteins and downstream cellular events.[8,9] Among sev-
eral tyrosine residues that have been shown to be phosphory-
lated, two major VEGF-dependent autophosphorylation sites
have been identified in VEGFR-2; however, only the autophos-
phorylation of Tyr1175 has been identified as crucial for VEGF-
dependent endothelial cell proliferation.[10]
A representative example of VEGFR-2 inhibitors is the an-
thranilamides, designed based on the hypothesis that an intra-
molecular hydrogen bond between the aniline and the benza-
nilide carbonyl group would favor a conformation with a high
degree of similarity to that of the previously discovered phtha-
lazine class (Scheme 1).[25] Lead optimization studies within the
[a] Dr. A. Papakyriakou, Dr. M. E. Katsarou, Dr. D. Vourloumis
Laboratory of Chemical Biology of Natural Products and
Designed Molecules, Institute of Physical Chemistry, NCSR “Demokritos”
15310 Ag. Paraskevi Attikis, Athens (Greece)
Fax: (+30)210-6511766
E-mail: vourloumis@chem.demokritos.gr
[b] Dr. M. Belimezi
Regulon A.E., Afxentiou 7, 17455 Alimos Attikis, Athens (Greece)
[c] Dr. M. Karpusas
Laboratory of Physics, Department of Science
Agricultural University of Athens, Iera Odos 75, 11855 Athens (Greece)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.200900373.
118
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2010, 5, 118 – 129

Potent VEGFR-2 Inhibitors
modeling of several designed anthranilamide scaffolds with al-
ternative substitution patterns based on AAL993 was pursued.
Based on our observations the central phenyl ring could be re-
placed by thiophene, which could potentially alter the elec-
tronic characteristics of the newly designed ligands, optimizing
their binding affinity. These analogues were synthesized and
screened in vitro for their inhibitory potential toward VEGFR-2.
Furthermore, the analogous anthranilamides were generated
for direct biological comparison. Our overall results were ana-
lyzed by computational methods.
Scheme 1. Evolution of the phthalazine class of VEGFR-2 inhibitors to the
anthranilamides, and the structure of AAL993 as a representative example of
the latter.
Results
anthranilamide series resulted in the identification of AAL993
(Scheme 1) as a highly potent and selective inhibitor of
VEGFRs, with good pharmacological properties and excellent
oral bioavailability, in both rodents and dogs.[26] The X-ray crys-
tal structure of AAL993 in complex with the catalytic domain
of diphosphorylated VEGFR-2 revealed that this molecule binds
to and stabilizes the inactive conformation of the protein.[27]
Despite the high structural conservation of RTKs bound to the
ATP-active configuration, the inactive states of these enzymes
often adopt distinct conformations, presenting large variations
in the size and shape of the ATP-binding pocket.[28] As a conse-
quence, it was suggested that targeting such inactive confor-
mations would provide an attractive strategy for achieving
target selectivity.[28]
As reported for the AAL993–VEGFR-2 complex,[27] the inhibitor
binds to the protein through three hydrogen bonds between:
a) the pyridine N atom and the backbone NH of Cys919 in the
hinge region of the protein, b) the anthranilamide C=O group
and the backbone NH of Asp1046 of the conserved DFG motif,
and c) the anthranilamide NH and the side chain carboxylate
of Glu885 of the aC helix. The trifluoromethylphenyl moiety of
AAL993 is accommodated within a lipophilic pocket defined
by residues Ile888, Leu889, Ile892, Val898, Val899, Leu1019,
and Ile1044, while the phenyl ring of the anthranilamide
moiety is sandwiched between the hydrophobic side chain
components of Val914 and Lys868. For the initial stages of our
modeling effort, we prepared a model of the AAL993–VEGFR-2
complex based on the available X-ray structures of other com-
plexes (see Computational methods in the Experimental Section
and Figure 1A). The docking results were analyzed based on
the reported characteristics of the compound’s interaction
with the biological target. Thus, the choice of the docked con-
formation was based solely on structural and not energetic cri-
In our studies we employed docking calculations in conjunc-
tion with molecular dynamics (MD), aiming to provide recep-
tor–drug models that implicate both the flexibility of the
kinase domain and the effect of the polar solvent in the ob-
served binding affinity of the inhibitors.[29] Initially, molecular
Figure 1. A) Model of VEGFR-2 kinase domain based on the X-ray crystal structure (PDB entry 1YWN).[18] The location of AAL993 in the ATP binding site is illus-
trated by the light green surface. Also indicated are the activation loop, the catalytic loop, the aC helix, the Gly-rich loop, and the two Tyr residues of the acti-
vation loop that were found to be autophosphorylated. B) Close-up view of AAL993 (stick representation; C atoms in orange) docked into the ATP binding
site of VEGFR-2. Residues of the kinase domain (C atoms in light green) that mediate key H-bonding interactions are labeled. The trifluoromethylphenyl
moiety of AAL993 is buried inside a hydrophobic pocket, which is depicted as a surface. Nitrogen atoms are shown in blue, oxygen in red, sulfur in yellow,
and fluorine in green.
ChemMedChem 2010, 5, 118 – 129
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
119

D. Vourloumis et al.
MED
teria, that is, higher-binding-energy models were discarded
when they did not exhibit the three key interactions men-
tioned above (Figure 1B).
the selected aldehydes comprise heteroatoms capable of inter-
acting with Cys919 of the hinge region through hydrogen
bonding. The differential topology of the heteroatoms would
serve in the investigation of the target’s flexibility as well as
the solvation effects. Aldehydes b and f were specifically em-
ployed to explore the potential of extensive hydrophobic con-
tacts compensating for the loss of the aforementioned key in-
teraction.
Our MD calculations employed the inactive “DFG-out” con-
formation of VEGFR-2 as previously reported.[27] A 10-ns MD
simulation, carried out with AAL993 docked at the ATP binding
site, predicted that some of the previously mentioned key in-
teractions are not well conserved in the presence of solvent
(figure S1, Supporting Information). In particular, the distance
between the anthranilamide NH and the carboxylate of
Glu885 fluctuates largely as a function of the simulation time.
Similarly, the pyridine N atom and the side chain amide of
Cys919 are within H-bonding distance for less than 50% of the
simulation time. Because Cys919 is one of the two residues,
the other being Glu917, in the hinge region that interacts with
the adenine of ATP, it represents a primary target for ligand
design. On the other hand, the trifluoromethylphenyl group of
the inhibitor is well accommodated within the lipophilic
pocket that consists of residues Ile888, Leu889, Ile892, Val898,
Val899, Leu1019, and Ile1044, thus maintaining their hydro-
phobic contacts throughout the major part of the MD simula-
tion (figure S2, Supporting Information). Based on these obser-
vations we hypothesized that substituting either the pyridine
group or the central phenyl group in the anthranilamide series
might yield improved inhibitors of VEGFR-2.
The thiophene-containing analogues were synthesized ac-
cording to the procedure presented in Scheme 3. Specifically,
protection of methyl-3-aminothiophene-2-carboxylate 5 with
Synthesis
The targeted anthranilamides were readily prepared in two
steps,[26] as presented in Scheme 2. Thus, ring opening of isato-
ic anhydride 2 with 3-trifluoromethylaniline 1 afforded amide
3, which upon reductive alkylation with aldehydes RCHO (a–j)
produced the corresponding anthranilamides 4a–4j. AAL993
was also synthesized by the same reaction sequence (4a in
Scheme 2) for direct biological comparisons. The majority of
Scheme 3. Reagents and conditions: a) 5 (1.0 equiv), Boc2O (3.1 equiv), Et3N
(20% in CH2Cl2), DMAP (1.6 equiv), 238C, 4 h, 80%; b) NaOH (0.6m in H2O/
MeOH 1:20), 238C, 6 h, 98%; c) 6 (1.0 equiv), 1 (3.0 equiv), DIC (4.0 equiv),
DIPEA (4.1 equiv), DMAP (3.0 equiv), THF (0.60m), 708C, 2 h, 52%; d) TFA
(26% in CH2Cl2), 238C, 30 min, 99%; e) AcOH (4.1 equiv), RCHO (1.4 equiv),
MeOH (0.07m), 238C, 12 h, then NaCNBH3 (4.7 equiv), 238C, 24 h, 57–95%.
Boc=t-butyloxycarbonyl; DIC=N,N’-diisopropylcarbodiimide; DIPEA=N,N-
diisopropylethylamine; DMAP=N,N-dimethylpyridin-4-amine; THF=tetrahy-
drofuran; TFA=trifluoroacetic acid.
Boc2O afforded the corresponding N-protected ester, which
was then hydrolyzed with sodium hydroxide in methanol to
yield the N-protected aminothiophene-2-carboxylic acid 6.
Coupling of 6 with 3-trifluoromethylaniline 1, mediated by DIC,
followed by the removal of the protecting carbamate with
TFA, furnished aromatic amine 8 in 52% yield for the two
steps. Finally, 8 was employed in the reductive alkylation reac-
tions with aldehydes RCHO (a–j, Scheme 3) to form com-
pounds 9a–9j in 57–95% yield.
Biological evaluation
Scheme 2. Reagents and conditions: a) 1 (1.0 equiv), 2 (1.0 equiv), DMF
(0.62m), 1008C, 24 h, 36%; b) AcOH (3.0 equiv), RCHO (1.3 equiv), MeOH
(0.07m), 238C, 8 h, then NaCNBH3 (4.4 equiv), 238C, 24 h, 28–91%.
DMF=N,N-dimethylformamide.
The ability of the synthetic compounds to inhibit the prolifera-
tion of cell growth was determined by MTT assay, as described
120
www.chemmedchem.org
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2010, 5, 118 – 129

Potent VEGFR-2 Inhibitors
previously.[30] MTT is absorbed by the mitochondria, where it is
transformed into formazan by succinic dehydrogenase. By as-
sessing mitochondrial dehydrogenase activity, the amount of
viable cells in a cell population after treatment with each com-
pound for various time periods can be measured. Therefore,
the sensitivity of the cells to the concentration and incubation
time with the various compounds can be determined. The ab-
sorbance of each cell lysate solution was measured at l=
550 nm. The results from the MTT assays are expressed as the
means of the absorption values A550 ꢀSD using data from two
independent experiments, which had each been repeated
three times. CC50 values, representing the concentration of
each compound required for a 50% decrease in cell viability,
were estimated by sigmoid fit, and are listed in Table 1.
kinase assay kit (Cell Signaling Technology, USA), following the
colorimetric ELISA protocol that was provided by the manufac-
turer. Recombinant human GST–VEGFR-2 kinase (Val789–
Val1356), biotinylated peptide substrate and phosphotyrosine
antibody, for the detection of phosphorylated substrate, were
all supplied within the kit. Percent inhibition values were calcu-
lated using Equation (1), as detailed in the Experimental Sec-
tion below. Each IC50 value was determined by plotting the
percent inhibition as a function of log([compound] in nm)
using a sigmoid fit and are presented in Table 1.
The four most potent compounds, 4a (AAL993), 4e, 9d, and
9j, were evaluated for their effect on angiogenesis by using an
endothelial tube formation assay on an ECMatrixTM (Chemi-
conꢀ). This solid gel consists of basement proteins (e.g., lami-
nin, collagen type IV), growth factors (e.g., TGF-b), and proteo-
lytic enzymes such as plasminogen and matrix metalloproteas-
es that promote tube formation,
IC50 values for the inhibition of VEGFR-2 (KDR) kinase by indi-
vidual compounds were measured with the HTScanꢀ VEGFR-2
a procedure requiring cell adhe-
sion, migration, differentiation,
Table 1. Potency data for the synthetic analogues 4a–4j and 9a–9j.
and growth. Organization pat-
terns in cells were observed
after 4 h of culture. Qualitative
assessment of images acquired
at 6 h showed that 9d and 9j
significantly inhibit the forma-
tion of capillary tubes, relative
Compd
R
CC50 [mm]
IC50 [nm]
Compd
R
CC50 [mm]
IC50 [nm]
>500
to compound 4a and the con-
trol (Figure 2A,C). Quantitative
analysis was also possible with
an end-point assay, in which the
capillary tube branch points
were counted at 6 h. In the ab-
sence of VEGF, 4a inhibited an-
giogenesis up to 60%, and 9j
up to 25% at both 10 and
50 mgmLꢁ1, while 9d inhibited
angiogenesis up to 35 and 58%
at 10 and 50 mgmLꢁ1, respec-
tively, relative to control (Fig-
ure 2B). Incubation of cells in
combination with 50 ngmLꢁ1
VEGF (Figure 2D) almost entire-
ly inhibited tube formation up
to 90% for compound 9d at
both concentrations. Significant
inhibition was observed for 4a
up to 65 and 80%, and for 9j
up to 80 and 65% at 10 and
50 mgmLꢁ1, respectively, relative
to control. Compound 4e ex-
hibited the lowest inhibitory
effect among the four inhibitors
tested (data not shown).
4a
146
40
9a
99
4b
4c
141
38
490
9b
9c
264
43
>500
>500
>500
4d
4e
36
>500
9d
9e
109
152
<1
105
3
>500
4 f
340
>500
9 f
>300
>500
4g
4h
349
34
>500
>500
9g
9h
285
93
>500
>500
4i
4j
229
153
>500
>500
9i
9j
138
104
>500
70
ChemMedChem 2010, 5, 118 – 129
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
121

D. Vourloumis et al.
MED
Figure 2. Tube formation assay after culture on ECMatrixTM for 6 h. HUVECs were stained with calcein and visualized by fluorescence microscopy. A) Control
cells and cells treated with AAL993 (4a), 9d, or 9j at 10 or 50 mgmLꢁ1, as indicated. B) Number of capillary tube branch points of treated HUVECs compared
with untreated cells (relative to control, set at 100%). C) Control cells and cells treated with AAL993 (4a), 9d, or 9j at 10 or 50 mgmLꢁ1, as indicated, in the
presence of VEGF (50 ngmLꢁ1). D) Number of capillary tube branch points of treated HUVECs compared with untreated cells (relative to control, set at 100%)
in the presence of VEGF (50 ngmLꢁ1). Data in graphs are expressed as the mean ꢀSD of three independent experiments; p<0.01 relative to control.
Moreover, the side chain carboxylate groups of Glu885 and
Discussion
Asp1046 were found to be hydrogen bonded via solvent inter-
Our efforts to evaluate our results and extract useful conclu-
sions regarding important conformational and structural data
in the biological affinities of the synthetic analogues toward
the protein target were assisted by computational methods.
Although the absence of an exact crystal structure renders this
method highly speculative, important interactions may be
identified in the process, inspiring further development and di-
recting future synthetic endeavors.
actions (W5 and W6), while the backbone NH group of
Asp1046 interacts directly with the anthranilamide C=O group.
Specifically, substitution of the 4-pyridine functionality in
AAL993 (4a) by 3-pyridine in compound 4e resulted in a
higher-affinity VEGFR-2 inhibitor (Table 1). This effect cannot be
explained by examining the initial docked conformation of 4e,
owing to the clear resemblance with AAL993, which lacks the
intermolecular hydrogen bond between the pyridine N atom
and the NH group of Cys919. However, during the 10-ns MD
calculations, it was recognized that the missing interaction was
compensated by solvent molecules (W1–W3 in Figure 3), which
mediate H-bonding contacts between the pyridine N atom
and both the NH and C=O groups of Cys919. This effect is ob-
served in approximately 80% of the total simulation time. In-
terestingly, the direct hydrogen bonding of the anthranilamide
NH with the carboxylate of Glu885 is also substituted by a sol-
vent-mediated interaction via a long-lived water molecule (3–
4 ns, figure S3, Supporting Information), that is, W4 in Figure 3.
Figure 3. Snapshot at the end of a 10-ns MD simulation illustrating com-
pound 4e within the ATP binding cleft of VEGFR-2. Key residues of the
kinase domain and water molecules (W1–W6) that mediate putative H-bond-
ing interactions are depicted. The hydrophobic residues that interact with
the trifluoromethylphenyl ring of 4e are omitted for clarity.
122
www.chemmedchem.org
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2010, 5, 118 – 129

Potent VEGFR-2 Inhibitors
Notably, the trifluoromethylphenyl ring retains the same hydro-
phobic interactions throughout the MD simulation, similar to
AAL993 (Figure 1B). The above observations point to a binding
model for compound 4e, which mimics the same interactions
as those reported for 4a (AAL993), some of which, however,
could be mediated by solvent molecules.
be oriented toward the hinge region, as observed throughout
our MD simulations (Figure 4B). There it probably participates
in two H-bonding interactions with the amide group of
Cys919 and with the backbone of Glu917. In this conforma-
tion, the amide C=O of 9d may accept two hydrogen bonds
from Asp1046 NH and Cys1045 SH, while the amide NH may
interact via water-mediated contacts with the carboxylate of
Glu885. Further stabilization of Glu885 and Asp1046 side
chains may involve solvent molecules, such as W5 shown in
Figure 4B. A putative network of solvent molecules (W1–W4 in
Figure 4B) is also predicted to mediate H-bonding interactions
between the hydroxy group of 9d and the backbone of
Phe1047, Gly1048 (DFG motif), and Arg1032. It is therefore no-
table that in any of the predicted conformations, compound
9d interacts with important residues of the ATP binding site,
an effect that justifies its sub-nanomolar IC50 value for VEGFR-2
(Table 1).
Docking of compound 9d revealed a binding motif in which
the trifluoromethylphenyl group is accommodated by the hy-
drophobic pocket while the amide NH is hydrogen bonded to
Glu885, and the amide C=O group to the backbone NH of
Asp1046 (Figure 4A). Additional stabilization of Asp1046 may
be provided by the hydroxy group of 9d through a hydrogen
bond with its backbone carbonyl. At the same conformation,
the nitro group forms ionic interactions with Arg1032, which is
part of the catalytic loop, and is highly conserved among pro-
tein tyrosine kinases. In previous reports, Arg1032 was found
to be hydrogen bonded to the side chain carboxylate of
Asp1028 (HRD motif), which is essential for catalysis of the
phosphotransfer reaction.[31] Alternatively, the nitro group may
Compound 9j is predicted to bind within the ATP binding
site in the same orientation as previously described (Figure 5).
The trifluoromethylphenyl group is buried in the hydrophobic
Figure 5. Intermolecular interactions of the simulated 9j–VEGFR-2 complex.
Key residues of the kinase domain are labeled, and residues that interact by
hydrophobic contacts are omitted for clarity.
pocket, the amide NH is hydrogen bonded to the carboxylate
of Glu885, and the amide C=O accepts a hydrogen bond from
the Asp1046 backbone. Additionally, the amino group of 9j
may contribute to an increased stabilization of Asp1046
through hydrogen bonding. Interestingly, the thiophene ring is
positioned ideally to interact with the terminal amine of
Lys868, which provides electrostatic stabilization to Glu885. In
this configuration, the furan oxygen atom of 9j may readily
accept a hydrogen bond from the side chain carboxamide of
Asn1033. An additional interaction with Cys1045 (Figure 5)
might be possible, as the furan oxygen is in the proximity of
its sulfhydryl group (4.0 ꢂ).
Figure 4. The different binding orientations of compound 9d sampled
during the MD simulations: docked configuration at A) 0 ns, and B) 10 ns.
Residues of the VEGFR-2 kinase domain that provide key interactions and
water molecules that mediate putative H-bonding interactions are depicted.
The side chain atoms of Phe1047 and the residues that provide hydropho-
bic contacts are omitted for clarity.
Concerning the synthetic analogues that lack hydrogen
bonding potential (Schemes 2 and 3), as indicated by the per-
fluorobenzyl groups (in 4b and 9b) or the bulkier biphenyls
ChemMedChem 2010, 5, 118 – 129
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
123

D. Vourloumis et al.
MED
(in 4 f and 9 f), it is clear that the hydrophobic interactions ini-
tially targeted cannot compensate for the interaction with
Cys919 of the hinge region. Similarly, the sulfur-bearing aro-
matic thiazoles 4i and 9i, as well as the benzothiophenes 4g
and 9g, probably adopt unfavorable orientations to accept hy-
drogen bonds due to electronic or steric reasons, respectively.
Compounds with phenolic hydroxy groups at the para position
(4c, 4h, and 9h) were also found to be inactive (Table 1), indi-
cating that steric hindrance does not allow the formation of
the aforementioned hydrogen bond; on the other hand, such
hydroxy groups contribute to the observed enhanced cytotox-
icity (Table 1).
Experimental Section
Chemistry
Unless otherwise noted, all solvents and reagents for organic syn-
thesis were obtained from commercial suppliers and were used
without further purification. All reactions were carried out under a
dry Ar atmosphere with anhydrous, freshly distilled solvents under
anhydrous conditions unless otherwise noted. All reactions were
stirred with Teflon-coated magnetic stir bars, and temperatures
were measured externally. Reactions requiring anhydrous condi-
tions were carried out in oven-dried (1208C, 24 h) or flame-dried
(vacuum <0.5 Torr) glassware. Yields refer to chromatographically
and spectroscopically (1H NMR) homogeneous materials. Reactions
were monitored by thin-layer chromatography (TLC) carried out on
0.25 mm silica gel plates (60 F254, E. Merck). Silica gel (60, particle
size: 0.040–0.063 mm, E. Merck) was used for flash column chroma-
tography.
Great variation in the inhibitory potential between the two
series is observed with R=3- or 4-pyridine (Table 1). Specifical-
ly, the anthranilamide compounds 4a and 4e are highly active
nanomolar inhibitors, while the respective thiophene-contain-
ing analogues 9a and 9e were found to be ineffective VEGFR-
2 binders. Similarly, the inactive anthranilamides with R=2-hy-
droxy-5-nitrobenzyl or 2-furan are transformed into the nano-
molar thiophene-containing inhibitors 9d and 9j, respectively
(Table 1). Clearly, there is a significant difference in the bound
configurations between the two subfamilies. These results
direct us to propose that substitution of the central benzene
group by thiophene causes a change in the electronic proper-
ties of the designed scaffolds (figures S4 and S5, Supporting In-
formation). It is therefore plausible that a change in the geom-
etry of the intramolecular hydrogen bond at the constrained
anthranilamides might result in altered conformational free-
dom of the thiophene-containing scaffolds, which might be re-
sponsible for additional interactions with key residues at the
catalytic cleft. Furthermore, the thiophene sulfur atom might
be directly implicated in H-bonding interactions with critical
residues, as in the case of 9j, which is predicted to provide ad-
ditional stabilization for the Lys868···Glu885 pair (Figure 5).
NMR spectra were recorded on a Bruker Avance III-250 or a Bruker
Avance DRX-500 instrument, as noted individually. All spectra were
recorded at 298 K using deuterated solvents as internal standards:
[CDCl3, 7.26 ppm (1H) and 77.16 ppm (13C); [D6]DMSO, 2.50 ppm
(1H) and 29.8 ppm (13C)]. Chemical shift values (d) are given within
an accuracy of 0.01 ppm for 1H NMR and 0.1 ppm for 13C NMR
data, while the coupling constants (J) are within 0.1 Hz. Multiplici-
ties are designated as singlet (s), doublet (d), triplet (t), or multiplet
(m). Broad or obscured peaks are indicated as “b” or “obs”, respec-
tively. 2D NMR 1H–1H COSY and 1H–13C HMQC correlation spectra
were also recorded at 298 K in order to assist in peak assignment.
The 1H and 13C NMR spectra of the most potent compounds 4a,
4e, 9d, and 9j are given in the Supporting Information.
2-Amino-N-[3-(trifluoromethyl)phenyl]benzamide (3).[26] A mix-
ture of 3-trifluoromethylaniline 1 (500 mg, 3.1 mmol) and isatoic
anhydride 2 (500 mg, 1.0 equiv, 3.1 mmol) in DMF (5 mL) was
stirred at 1008C for 24 h. The solvent was evaporated under re-
duced pressure to produce a residue which was dissolved in EtOAc
(25 mL) and washed with saturated aqueous NH4Cl (2ꢃ15 mL) and
saturated NaCl (2ꢃ15 mL). The solution was dried using anhydrous
MgSO4, filtered, and the solvent was evaporated to provide the
crude product, which was eluted from 10% EtOAc in hexanes to
yield 3 as colorless needles (315 mg, 36%); 1H NMR ([D6]DMSO,
500 MHz): d=6.37 (bs, 2H), 6.60 (dd, J=7.6, 7.4 Hz, 1H), 6.76 (d,
J=8.4 Hz, 1H), 7.22 (dd, J=8.0, 7.6 Hz, 1H), 7.42 (d, J=7.7 Hz, 1H),
7.57 (dd, J=8.0, 7.7 Hz, 1H), 7.65 (d, J=8.0 Hz, 1H), 7.97 (d, J=
8.1 Hz, 1H), 8.20 (s, 1H), 10.27 (s, 1H).
Conclusions
Three novel low-nanomolar-range VEGFR-2 inhibitors have
been discovered through a structure-based approach assisted
by computational methods. MD analysis of the VEGFR-2–
AAL993 complex has inspired the design and synthesis of a
thiophene series of VEGFR-2 inhibitors with altered steric, elec-
tronic, and H-bonding requirements relative to the parent an-
thranilamides. The latter became clear by direct comparison of
the biological activities between closely related analogues. In
vitro evaluation of the two series both by KDR-specific kinase
assays and HUVEC tube formation assays revealed their high
anti-angiogenic potency. Moreover, a combination of docking
and MD was used in order to formulate a reasonable explana-
tion for the obtained results. Additional examples to assist us
in the evaluation of this hypothesis and the establishment of a
valid structure–activity relationship analysis, along with related
selectivity studies and in vivo experiments, are currently under
investigation.
Compounds (4a–4j). A solution of 3 (210 mg, 0.75 mmol) and
AcOH (50 mL, 2.9 mmol) in MeOH (10 mL) was divided among 10
flasks. Each solution was treated with 0.10 mmol of the corre-
sponding aldehyde and stirred at ambient temperature for 8 h.
The resulting reaction mixtures were treated with 1.0 mL of a solu-
tion of NaCNBH3 (210 mg, 3.3 mmol) in MeOH (10 mL) and stirred
for a further 24 h at ambient temperature. The solvent was evapo-
rated, and the residues were diluted with CH2Cl2 (15 mL) and
washed with saturated aqueous NaHCO3 (2ꢃ10 mL) followed by
saturated NaCl (2ꢃ10 mL). The organic layer was dried over
MgSO4, filtered, and concentrated under reduced pressure. All
compounds were purified by column chromatography, eluted from
EtOAc/hexanes:
2-[(4-Pyridyl)methyl]amino-N-[3-(trifluoromethyl)phenyl]benza-
mide (4a, AAL993). Colorless solid; Rf =0.20 (EtOAc/hexanes, 1:1);
1
(17 mg, 61%): H NMR ([D6]DMSO, 500 MHz): d=4.50 (d, J=6.1 Hz,
2H), 6.55 (d, J=8.5 Hz, 1H), 6.67 (dd, J=7.4, 7.4 Hz, 1H), 7.26 (dd,
124
www.chemmedchem.org
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2010, 5, 118 – 129

Potent VEGFR-2 Inhibitors
J=7.8, 7.4 Hz, 1H), 7.34 (bd, J=5.6 Hz, 2H), 7.45 (d, J=7.7 Hz, 1H),
7.59 (t, J=8.0 Hz, 1H), 7.73 (dd, J=7.9, 1.1 Hz, 1H), 7.94 (t, J=
6.2 Hz, 1H), 7.97 (bd, J=8.2 Hz, 1H), 8.26 (bs, 1H), 8.49 (bd, J=
5.8 Hz, 2H), 10.44 (s, 1H); 13C NMR (CDCl3, 62.5 MHz): d=46.1,
112.5, 115.1, 116.0, 117.5 (q), 121.2 (q), 122.1, 123.7, 127.7, 129.7,
133.8, 138.6, 148.6, 149.7, 150.1, 168.4; HRMS (ESI) m/z calcd for
C20H16F3N3ONa [M+Na]+: 394.1138, found: 394.1137.
1:2); (17 mg, 53%): 1H NMR (CDCl3, 500 MHz): d=4.64 (bs, 2H),
6.69 (dd, J=7.4, 7.4 Hz, 1H), 6.79 (d, J=8.4 Hz, 1H), 7.31–7.42 (m,
5H), 7.46 (dd, J=7.8, 8.0 Hz, 1H), 7.54 (d, J=7.6 Hz, 1H), 7.74 (d,
J=7.9 Hz, 1H), 7.81 (d, J=7.8 Hz, 1H), 7.86 (s, 1H), 7.87 (d, J=
7.4 Hz, 1H), 7.98 (s, 1H), 8.02 (bm, 1H); 13C NMR (CDCl3, 62.5 MHz):
d=42.0, 112.8, 115.0, 115.7, 117.3 (q), 121.1 (q), 121.6, 123.1, 123.4,
123.7, 124.3, 124.7, 127.5, 129.7, 133.2, 133.9, 137.9, 138.5, 141.1,
150.0, 168.2; HRMS (ESI) m/z calcd for C23H17F3N2OSNa [M+Na]+:
449.0906, found: 449.0904.
2-(Perfluorobenzyl)amino-N-[3-(trifluoromethyl)phenyl]benza-
mide (4b). Colorless solid; Rf =0.75 (EtOAc/hexanes, 1:4); (24 mg,
1
70%): H NMR (CDCl3, 500 MHz): d=4.52 (bs, 2H), 6.73 (dd, J=7.7,
2-(4-Hydroxybenzyl)amino-N-[3-(trifluoromethyl)phenyl]benza-
mide (4h). Yellow solid; Rf =0.55 (EtOAc/hexanes, 1:2); (20 mg,
7.4 Hz, 1H), 6.87 (d, J=8.4 Hz, 1H), 7.40 (m, 2H), 7.48 (dd, J=8.0,
7.9 Hz, 1H), 7.51 (d, J=7.9 Hz, 1H), 7.74 (d, J=8.0 Hz, 1H), 7.84 (s,
1H), 7.87 (bs, 1H); 13C NMR (CDCl3, 62.5 MHz): d=35.0, 111.8, 115.6,
116.4, 117.4 (q), 121.3 (q), 123.8, 127.7, 129.7, 129.7, 134.0, 138.4,
148.9, 168.0; HRMS (ESI) m/z calcd for C21H12F8N2ONa [M+Na]+:
483.0714, found: 483.0713.
1
69%): H NMR (CDCl3, 500 MHz): d=4.33 (bs, 2H), 6.66 (dd, J=7.4,
7.4 Hz, 1H), 6.71 (d, J=8.4 Hz, 1H), 6.76 (bd, J=8.4 Hz, 2H), 7.21
(bd, J=8.4 Hz, 2H), 7.31 (m, 1H), 7.39 (d, J=7.6 Hz, 1H), 7.48 (dd,
J=8.0, 7.9 Hz, 1H), 7.50 (dd, J=7.8, 1.0 Hz, 1H), 7.74 (d, J=7.9,
1H), 7.86 (s, 1H), 7.91 (bs, 1H); 13C NMR (CDCl3, 62.5 MHz): d=46.9,
112.8, 114.8, 115.4, 115.7, 117.4 (q), 121.1 (q), 123.7, 127.5, 128.8,
129.7, 130.9, 133.8, 138.5, 150.0, 154.9, 168.4; HRMS (ESI) m/z calcd
for C21H17F3N2O2Na [M+Na]+: 409.1134, found: 409.1136.
2-(3,5-Dibromo-4-hydroxybenzyl)amino-N-[3-(trifluoromethyl)-
phenyl]benzamide (4c). Yellow solid; Rf =0.45 (EtOAc/hexanes,
1:1); (23 mg, 56%): 1H NMR (CDCl3, 500 MHz): d=4.32 (bs, 2H),
5.85 (bs, 1H), 6.59 (d, J=8.4 Hz, 1H), 6.70 (dd, J=7.9, 7.5 Hz, 1H),
7.31 (dd, J=8.4, 7.8, 1H), 7.41 (d, J=7.7 Hz, 1H), 7.44 (bs, 2H), 7.49
(dd, J=8.0, 7.6 Hz, 1H), 7.53 (d, J=7.6 Hz, 1H), 7.74 (d, J=7.4 Hz,
1H), 7.90 (s, 2H), 7.99 (bs, 1H); 13C NMR (CDCl3, 62.5 MHz): d=45.8,
110.3, 112.7, 115.0, 116.0, 117.4 (q), 121.2 (q), 123.7, 127.5, 129.7,
130.6, 133.3, 133.9, 138.5, 148.6, 149.7, 168.2; HRMS (ESI) m/z calcd
for C21H15Br2F3N2O2Na [M+Na]+: 564.9345, found: 564.9347.
2-[(Thiazole-2-yl)methyl]amino-N-[3-(trifluoromethyl)phenyl]ben-
zamide (4i). Light yellow solid; Rf =0.40 (EtOAc/hexanes, 1:2);
(14 mg, 50%): 1H NMR (CDCl3, 500 MHz): d=4.77 (bs, 2H), 7.71–
7.76 (m, 2H), 7.26 (bm, 1H), 7.32 (dd, J=8.0, 7.6 Hz, 1H), 7.41 (d,
J=7.6 Hz, 1H), 7.49 (dd, J=8.0, 7.8 Hz, 1H), 7.54 (d, J=7.8, 1H),
7.73–7.79 (m, 2H), 7.93 (s, 1H), 8.05 (bs, 1H); 13C NMR (CDCl3,
62.5 MHz): d=45.6, 112.9, 116.7, 116.0, 117.4 (q), 119.4, 121.2 (q),
123.7, 127.6, 129.7, 133.8, 138.6, 142.7, 149.2, 168.1, 171.4; HRMS
(ESI) m/z calcd for C18H14F3N3OSNa [M+Na]+: 400.0702, found:
400.0702.
2-(2-Hydroxy-5-nitrobenzylamino)-N-[3-(trifluoromethyl)phenyl]-
benzamide (4d). Bright yellow solid; Rf =0.40 (EtOAc/hexanes,
1:1); (32 mg, 91%): 1H NMR (CDCl3, 500 MHz): d=4.51 (bs, 2H),
6.85 (m, 3H), 7.36 (dd, J=7.8, 7.8 Hz, 1H), 7.41 (d, J=7.8 Hz, 1H),
7.48 (dd, J=8.0, 7.8 Hz, 1H), 7.58 (d, J=7.8 Hz, 1H), 7.72 (d, J=
8.0 Hz, 1H), 7.90 (s, 1H), 8.03 (dd, J=8.9, 2.5 Hz, 1H), 8.08 (s, 1H),
8.13 (d, J=2.5 Hz, 1H); 13C NMR (CDCl3, 62.5 MHz): d=46.0, 114.5,
116.7, 117.1, 117.7 (q), 118.0, 118.7, 121.6 (q), 123.9, 124.3, 124.8,
125.2, 125.8, 127.6, 129.8, 133.9, 138.1, 141.1, 149.1, 162.0, 168.1;
HRMS (ESI) m/z calcd for C21H16F3N3O4Na [M+Na]+: 454.0985,
found: 454.0987.
2-[(Furan-2-yl)methyl]amino-N-[3-(trifluoromethyl)phenyl]benza-
mide (4j). Yellow solid; Rf =0.55 (EtOAc/hexanes, 1:2); (15 mg,
55%): 1H NMR (CDCl3, 500 MHz): d=4.40 (d, J=5.1 Hz, 2H), 6.25
(m, 1H), 6.31 (m, 1H), 6.70 (m, 1H), 6.82 (d, J=8.4 Hz, 1H), 7.32–
7.38 (m, 2H), 7.47 (t, J=8.0 Hz, 2H), 7.51 (dd, J=7.9, 1.4 Hz, 1H),
7.74 (d, J=7.3 Hz, 2H), 7.88 (bs, 1H), 7.95 (bs, 1H); 13C NMR (CDCl3,
62.5 MHz): d=40.7, 109.7, 110.6, 112.6, 115.2, 120.4, 123.2 (q), 123.6
(q), 127.6, 129.2, 129.6, 133.8, 142.1, 143.3, 149.7, 152.3, 168.1;
HRMS (ESI) m/z calcd for C19H15F3N2O2Na [M+Na]+: 383.0978,
found: 383.0977.
2-[(3-Pyridyl)methyl]amino-N-[3-(trifluoromethyl)phenyl]benza-
mide (4e). Colorless solid; Rf =0.25 (EtOAc/hexanes, 1:1); (18 mg,
64%): 1H NMR (CDCl3, 500 MHz): d=4.47 (bs, 2H), 6.64 (d, J=
8.4 Hz, 1H), 6.69 (dd, J=7.6, 7.4 Hz, 1H), 7.26–7.33 (m, 2H), 7.41 (d,
J=7.6 Hz, 1H), 7.49 (dd, J=8.0, 7.9 Hz, 1H), 7.55 (d, J=7.9 Hz, 1H),
7.72 (d, J=7.8 Hz, 1H), 7.75 (d, J=8.0 Hz, 1H), 7.91 (s, 1H), 8.03
(bs, 1H), 8.08 (s, 1H), 8.51 (d, J=3.8 Hz, 1H), 8.62 (s, 1H); 13C NMR
(CDCl3, 62.5 MHz): d=44.8, 112.5, 115.2, 116.0, 117.5 (q), 121.2 (q),
123.7, 123.9, 127.6, 129.7, 133.9, 134.8, 135.3, 138.6, 146.0, 148.5,
148.7, 149.7, 168.3; HRMS (ESI) m/z calcd for C20H16F3N3ONa
[M+Na]+: 394.1138, found: 394.1136.
3-(tert-Butoxycarbonylamino)thiophene-2-carboxylic acid (6).[32]
Di-tert-butyl dicarbonate (Boc2O, 3.4 g, 15.6 mmol) was added
dropwise to a stirred solution of methyl 3-aminothiophene-2-car-
boxylate 5 (800 mg, 5.1 mmol), DMAP (1.0 g, 8.2 mmol), Et3N
(2.0 mL, 14.4 mmol), and CH2Cl2 (8 mL) under an Ar atmosphere.
After 4 h at ambient temperature, the orange solution was extract-
ed with ice-cold HCl (1n, 3ꢃ7 mL), ice-cold NaOH (1n, 3ꢃ7 mL),
and brine (3ꢃ7 mL). The organic extract was dried over MgSO4, fil-
tered, concentrated under reduced pressure, and the correspond-
ing protected amine was purified by column chromatography
(10% EtOAc/hexanes). The resulting yellow product was then dis-
solved in MeOH (60 mL) and treated with NaOH (0.6m in H2O/
MeOH 1:20, 3.0 mL). After stirring for 6 h at ambient temperature,
the solution was diluted with H2O, and MeOH was evaporated
under reduced pressure. The aqueous phase was washed with Et2O
(2ꢃ10 mL), then cooled in an ice bath and acidified to pH 2 using
H2SO4 from a 10% v/v solution. The precipitate was immediately
extracted with EtOAc (3ꢃ10 mL), dried over MgSO4, filtered, and
concentrated under reduced pressure to yield 6 (980 mg, 4 mmol,
2-(3,3-Diphenylallylamino)-N-[3-(trifluoromethyl)phenyl]benza-
mide (4 f). Yellow solid; Rf =0.65 (EtOAc/hexanes, 1:2); (10 mg,
28%): 1H NMR (CDCl3, 500 MHz): d=3.91 (d, J=6.7 Hz, 2H), 6.20
(dd, J=6.8, 6.7 Hz, 1H), 6.75 (m, 2H), 7.16–7.25 (m, 7H), 7.29–7.43
(m, 6H), 7.49 (dd, J=7.9, 7.8 Hz, 1H), 7.60 (d, J=7.4 Hz, 1H), 7.80
(d, J=7.9 Hz, 1H), 7.91 (s, 1H), 8.08 (bs, 1H); 13C NMR (CDCl3,
62.5 MHz): d=42.9, 112.3, 115.2, 117.4 (q), 121.1 (q), 123.8, 126.4,
127.6, 127.7, 128.3, 128.5, 128.8, 129.7, 133.9, 138.5, 138.9, 141.9,
144.3, 149.7, 168.2; HRMS (ESI) m/z calcd for C29H23F3N2ONa
[M+Na]+: 495.1655, found: 495.1656.
1
2-(Benzo[b]thiophen-3-ylmethylamino)-N-[3-(trifluoromethyl)-
phenyl]benzamide (4g). White solid; Rf =0.85 (EtOAc/hexanes,
78% over two steps): H NMR (CDCl3, 500 MHz): d=1.54 (bs, 9H),
7.53 (d, J=5.4 Hz, 1H), 7.92 (d, J=5.4 Hz, 1H), 9.19 (s, 1H).
ChemMedChem 2010, 5, 118 – 129
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
125

D. Vourloumis et al.
MED
3-Amino-N-[3-(trifluoromethyl)phenyl]thiophene-2-carboxamide
(8). A mixture of 6 (730 mg, 3.0 mmol) and N,N’-diisopropylcarbo-
diimide (DIC, 1.86 mL, 12.0 mmol) was dried by azeotropic evapo-
ration twice with toluene. Another mixture of 3-(trifluoromethyl)a-
niline 1 (1.12 mL, 9.0 mmol), DMAP (1.10 g, 9.0 mmol), and DIPEA
(2.10 mL, 12.3 mmol) was also dried the same way. The two mix-
tures were combined in dry THF (5 mL) under an Ar atmosphere
and heated at 708C for 2 h. The reaction was then cooled to ambi-
ent temperature, diluted in EtOAc (30 mL), and extracted with satu-
rated aqueous NaHCO3 (10 mL), NH4Cl (2ꢃ10 mL), and NaCl (2ꢃ
10 mL). After drying the organic layer with MgSO4, filtering, and
concentrating, the coupled product was purified by column chro-
matography (10!15% EtOAc/hexanes) to yield compound 7. Con-
sequently, protected amine 7 was dissolved in CH2Cl2 (20 mL), and
TFA (7 mL, 26%) was added in three portions over 30 min and
stirred at ambient temperature. The solution was then concentrat-
ed to dryness under reduced pressure to yield amine 8 (445 mg,
52% over two steps): 1H NMR (CDCl3, 500 MHz): d=6.69 (d, J=
5.4 Hz, 1H), 7.31 (d, J=5.4 Hz, 1H), 7.38 (bs, 1H), 7.42 (d, J=7.9 Hz,
1H), 7.48 (dd, J=8.0, 7.9 Hz, 1H), 7.68 (d, J=8.0 Hz, 1H), 7.78 (s,
1H).
HRMS (ESI) m/z calcd for C19H15F3N2O3SNa [M+Na]+: 431.0648,
found: 431.0648.
3-(2-Hydroxy-5-nitrobenzylamino)-N-[3-(trifluoromethyl)phe-
nyl]thiophene-2-carboxamide (9d). Light yellow solid; Rf =0.35
1
(EtOAc/hexanes, 1:1); (26 mg, 85%): H NMR (CDCl3, 500 MHz): d=
4.52 (bs, 2H), 6.76 (d, J=5.4 Hz, 1H), 6.82 (d, J=8.8 Hz, 1H), 7.30
(s, 1H), 7.32–7.37 (m, 2H), 7.43 (dd, J=7.9, 7.8 Hz, 1H), 7.65 (d, J=
7.9 Hz, 1H), 7.76 (m, 1H), 7.83 (s, 1H), 8.02 (dd, J=8.9, 2.5 Hz, 1H),
8.11 (m, 1H); 13C NMR (CDCl3, 62.5 MHz): d=47.4, 105.2, 116.7,
117.7 (q), 119.0, 121.3 (q), 123.9, 124.5, 124.9, 125.5, 129.4, 129.7,
138.1, 141.3, 155.3, 161.7, 163.3; HRMS (ESI) m/z calcd for
C19H14F3N3O4SNa [M+Na]+: 460.0549, found: 460.0546.
3-(Pyridin-3-ylmethylamino)-N-[3-(trifluoromethyl)phenyl]thio-
phene-2-carboxamide (9e). Yellow solid; Rf =0.20 (EtOAc/hexanes,
1
1:1); (25 mg, 95%): H NMR (CDCl3, 500 MHz): d=4.52 (d, J=6.0 Hz,
2H), 6.63 (d, J=5.4 Hz, 1H), 7.21 (bs, 1H), 7.24–7.29 (m, 2H), 7.36
(d, J=7.6 Hz, 1H), 7.45 (dd, J=8.0, 7.8 Hz, 1H), 7.69 (dd, J=8.9,
8.8 Hz, 2H), 7.87 (s, 1H), 7.90 (m, 1H), 8.53 (m, 1H), 8.60 (s, 1H);
13C NMR (CDCl3, 62.5 MHz): d=46.7, 101.2, 117.1 (q), 117.8, 120.7
(q), 123.4, 123.9, 125.2, 129.1, 129.6, 135.0, 138.7, 148.8, 148.9,
155.8, 163.6; HRMS (ESI) m/z calcd for C18H14F3N2OSNa [M+Na]+:
400.0702, found: 400.0701.
Compounds (9a–9j). A solution of 8 (200 mg, 0.7 mmol) and
AcOH (50 mL, 2.9 mmol) in MeOH (10 mL) was divided among 10
flasks. Each solution was treated with 0.1 mmol of each aldehyde
and stirred at room temperature for 12 h. In each reaction, 1.0 mL
of a solution of NaCNBH3 (210 mg, 3.3 mmol) in MeOH (10 mL) was
added, and stirring was continued for a further 24 h at ambient
temperature. The solvent was evaporated, and the residues were
diluted with CH2Cl2 (15 mL) and extracted with saturated aqueous
NaHCO3 (2ꢃ10 mL) followed by saturated NaCl (2ꢃ10 mL). The or-
ganic layer was dried over MgSO4, filtered, and concentrated under
reduced pressure. All products were purified by column chroma-
tography (EtOAc/hexanes).
3-(3,3-Diphenylallylamino)-N-[3-(trifluoromethyl)phenyl]thio-
phene-2-carboxamide (9 f). Bright yellow solid; Rf =0.75 (EtOAc/
1
hexanes, 1:2); (30 mg, 89%): H NMR (CDCl3, 500 MHz): d=3.89 (t,
J=6.1 Hz, 2H), 6.07 (t, J=6.6 Hz, 1H), 6.46 (d, J=5.5 Hz, 1H), 7.04
(s, 1H), 7.10–7.21 (m, 8H), 7.24–7.38 (m, 5H), 7.49 (m, 1H), 7.62 (d,
J=7.8 Hz, 1H), 7.78 (s, 1H); 13C NMR (CDCl3, 62.5 MHz): d=44.6,
100.4, 117.0 (q), 118.1, 120.5 (q), 123.3, 125.9, 127.6, 127.7, 127.8,
128.3, 128.5, 128.8, 129.6, 129.9, 138.8, 139.2, 141.7, 144.5, 155.9,
163.5; HRMS (ESI) m/z calcd for C27H21F3N2OSNa [M+Na]+:
501.1219, found: 501.1217.
3-(Benzo[b]thiophen-3-ylmethylamino)-N-[3-(trifluoromethyl)-
phenyl]thiophene-2-carboxamide (9g). Yellow solid; Rf =0.60
(EtOAc/hexanes, 1:3); (21 mg, 69%): H NMR (CDCl3, 500 MHz): d=
3-(Pyridin-4-ylmethylamino)-N-[3-(trifluoromethyl)phenyl]thio-
phene-2-carboxamide (9a). Orange solid; Rf =0.20 (EtOAc/hex-
anes, 1:1); (15 mg, 57%): H NMR (CDCl3, 500 MHz): d=4.54 (d, J=
1
1
4.72 (d, J=5.3 Hz, 2H), 6.73 (d, J=5.4 Hz, 1H), 7.19 (s, 1H), 7.27 (d,
J=5.4 Hz, 1H), 7.31–7.45 (m, 5H), 7.70 (d, J=7.0 Hz, 1H), 7.79 (d,
J=7.7 Hz, 1H), 7.83–7.90 (m, 3H); 13C NMR (CDCl3, 62.5 MHz): d=
44.0, 117.1 (q), 118.1, 120.6 (q), 121.5, 123.2, 123.4, 123.5, 124.4,
124.8, 128.9, 129.6, 133.5, 137.7, 138.7, 141.1, 156.1, 163.6; HRMS
(ESI) m/z calcd for C21H15F3N2OS2Na [M+Na]+: 455.0470, found:
455.0472.
6.4 Hz, 2H), 6.54 (d, J=5.4 Hz, 1H), 7.23 (bs, 1H), 7.27–7.30 (m,
3H), 7.38 (d, J=7.9 Hz, 1H), 7.47 (dd, J=8.0, 7.9 Hz, 1H), 7.72 (d,
J=8.0 Hz, 1H), 7.91 (s, 1H), 7.99 (t, J=5.8 Hz, 1H), 8.57 (d, J=
5.2 Hz, 2H); 13C NMR (CDCl3, 62.5 MHz): d=48.0, 101.3, 117.2 (q),
117.7, 120.8 (q), 122.1, 123.4, 129.1, 129.7, 138.6, 149.3, 149.7,
155.7, 163.6; HRMS (ESI) m/z calcd for C18H14F3N3OSNa [M+Na]+:
400.0702, found: 400.0702.
3-(4-Hydroxybenzylamino)-N-[3-(trifluoromethyl)phenyl]thio-
phene-2-carboxamide (9h). Light yellow solid; Rf =0.30 (EtOAc/
hexanes, 1:1); (25 mg, 91%): H NMR (CDCl3, 500 MHz): d=4.40 (d,
J=5.0 Hz, 2H), 6.68 (d, J=5.4 Hz, 1H), 6.76 (d, J=8.3 Hz, 2H),
7.15–7.19 (m, 3H), 7.25 (d, J=5.4 Hz, 1H), 7.34 (d, J=7.7 Hz, 1H),
7.43 (dd, J=7.9, 7.7 Hz, 1H), 7.67–7.74 (m, 2H), 7.83 (s, 1H);
13C NMR (CDCl3, 62.5 MHz): d=48.7, 100.3, 115.7, 117.2 (q), 118.2,
120.6 (q), 123.5 128.7, 129.0, 129.6, 130.9, 138.7, 155.2, 156.3, 163.7;
HRMS (ESI) m/z calcd for C19H15F3N2O2SNa [M+Na]+: 415.0698,
found: 415.0700.
3-(Perfluorobenzylamino)-N-[3-(trifluoromethyl)phenyl]thio-
phene-2-carboxamide (9b). Light yellow solid; Rf =0.70 (EtOAc/
hexanes, 1:4); (29 mg, 89%): H NMR (CDCl3, 500 MHz): d=4.59 (d,
J=6.7 Hz, 2H), 6.86 (d, J=5.4 Hz, 1H), 7.13 (bs, 1H), 7.32–7.37 (m,
2H), 7.44 (dd, J=8.0, 7.9 Hz, 1H), 7.71 (d, J=8.0 Hz, 1H), 7.81 (bs,
1H), 7.84 (m, 1H); 13C NMR (CDCl3, 62.5 MHz): d=36.9, 102.0, 117.1
(q), 117.3, 120.8 (q), 123.5, 129.1, 129.7, 138.5, 154.8, 163.3; HRMS
(ESI) m/z calcd for C19H10F8N2OSNa [M+Na]+: 489.0278, found:
489.0280.
1
1
3-(2,5-Dihydroxybenzylamino)-N-[3-(trifluoromethyl)phenyl]thio-
phene-2-carboxamide (9c). Yellow solid; Rf =0.40 (EtOAc/hexanes,
1:1); (18 mg, 63%): 1H NMR (CDCl3, 500 MHz): d=4.46 (bs, 2H),
6.63–6.72 (m, 3H), 6.77 (d, J=5.4 Hz, 1H), 7.19 (s, 1H), 7.28 (d, J=
5.4 Hz, 1H), 7.37 (d, J=7.4 Hz, 1H), 7.46 (dd, J=8.0, 7.8 Hz, 1H),
7.68 (bs, 1H), 7.71 (d, J=7.5 Hz, 1H), 7.82 (s, 1H); 13C NMR (CDCl3,
62.5 MHz): d=46.7, 102.0, 115.5, 117.2, 117.3 (q), 119.1, 120.9 (q),
123.6, 125.2, 128.9, 129.7, 135.4, 138.6, 148.8, 149.5, 156.0, 163.3;
3-(Thiazol-2-ylmethylamino)-N-[3-(trifluoromethyl)phenyl]thio-
phene-2-carboxamide (9i). Light yellow solid; Rf =0.20 (EtOAc/
hexanes, 1:2); (20 mg, 75%): H NMR (CDCl3, 500 MHz): d=4.81 (d,
J=6.3 Hz, 2H), 6.72 (d, J=5.4 Hz, 1H), 7.22 (s, 1H), 7.25–7.29 (m,
2H), 7.36 (d, J=7.6 Hz, 1H), 7.46 (dd, J=7.8, 7.6 Hz, 1H), 7.72 (d,
J=7.8 Hz, 1H), 7.75 (d, J=3.2 Hz, 1H), 7.89 (s, 1H), 8.09 (m, 1H);
13C NMR (CDCl3, 62.5 MHz): d=47.4, 102.3, 117.2 (q), 118.2, 119.6,
120.8 (q), 123.4, 128.8, 129.6, 138.7, 142.9, 155.4, 163.4, 170.9;
1
126
www.chemmedchem.org
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2010, 5, 118 – 129

Potent VEGFR-2 Inhibitors
HRMS (ESI) m/z calcd for C16H12F3N3OS2Na [M+Na]+: 406.0266,
found: 406.0267.
50 kcalmolꢁ1 ꢂꢁ2 force constant to the Ca atoms that were deter-
mined in the crystal structure. Subsequently, a second round of
1000 steps of unrestrained energy minimization with conjugate
gradients was performed, which resulted in a model with heavy-
atom RMSD=0.49 ꢂ with respect to the crystal structure. This
structure was used for docking calculations and subsequent MD
simulations. A model of VEGFR-2 kinase domain with both Tyr1054
and Tyr1059 phosphorylated was prepared in the same way. The
program AutoDock 3[43] was used for docking of the inhibitors to
the model of the VEGFR-2 kinase domain, and AutoDockTools 1.5
was used for examination of the docking results. Protein and li-
gands were treated with the united-atom approximation by merg-
ing all nonpolar hydrogen atoms. Kollman partial charges were as-
signed to protein atoms, AM1-BCC charges to ligands, while charg-
es and van der Waals parameters for the two phosphate groups
were taken from the AMBER database. The interaction grid maps
were centered at the ATP binding site and comprised 81ꢃ81ꢃ81
points of 0.25 ꢂ spacing. The Lamarckian genetic algorithm was
used with the following parameters: a population size of 50 indi-
viduals, a maximum number of 1.5ꢃ106 energy evaluations and a
maximum number of 27000 generations, an elitism value of 1.0, a
mutation rate of 0.02, and a crossover rate of 0.80. For most of the
calculations, 100 docking rounds were performed with step sizes
of 0.25 ꢂ for translations and 58 for orientations and torsions.
Docked conformations were clustered within 1.5 ꢂ RMS positional
deviation tolerance.
3-(Furan-2-ylmethylamino)-N-[3-(trifluoromethyl)phenyl]thio-
phene-2-carboxamide (9j). Yellow solid; Rf =0.35 (EtOAc/hexanes,
1:4); (16 mg, 62%): 1H NMR (CDCl3, 500 MHz): d=4.45 (bs, 2H),
6.22 (m, 1H), 6.31 (m, 1H), 6.80 (d, J=5.4 Hz, 1H), 7.17 (s, 1H), 7.29
(d, J=5.4 Hz, 1H), 7.33–7.37 (m, 2H), 7.44 (dd, J=8.0, 7.9, 1H), 7.70
(d, J=7.9 Hz, 2H), 7.86 (s, 1H); 13C NMR (CDCl3, 62.5 MHz): d=42.5,
101.2, 107.2, 110.5, 117.0 (q), 118.1, 120.6 (q), 123.3, 128.8, 129.6,
138.8, 142.3, 152.5, 155.8, 163.5; HRMS (ESI) m/z calcd for
C17H13F3N2O2S Na [M+Na]+: 389.0542, found: 389.0540.
Computational methods
Because the X-ray crystal structure coordinates of AAL993 in com-
plex with the catalytic domain of VEGFR-2 are not yet available at
the PDB,[27] the highest-resolution (1.71 ꢂ) structure, with the
fewest missing residues, of a 4-aminofuro[2,3-d]pyrimidine com-
pound with VEGFR-2 (PDB entry 1YWN) was used.[18] The previously
determined crystal structure of the VEGFR-2 kinase domain free of
ligand (PDB entry 1VR2)[16] was discarded owing to: a) lower resolu-
tion (2.40 ꢂ), b) more missing residues in the activation loop, and
c) a greater number of disordered side chain atoms. However, su-
perimposition of 1VR2 with 1YWN displays minor differences be-
tween the two structures (backbone RMSD=0.50 ꢂ). The residue
numbering used herein is according to 1VR2, which differs from
1YWN by +2. Water molecules and ligands were deleted from the
PDB file. Given that the constructs used for crystallization lack the
central 50 residues (residues 942–991) of the 68-residue kinase
insert domain, residues Phe935, Val936, and Pro937 were removed
to facilitate linking of Glu934 and Phe999 (Ca–Ca distance of
4.8 ꢂ). The missing residues Phe1047–Asp1052 (where Phe1047
and Gly1048 belong to the DFG motif) were modeled using the
corresponding coordinates from PDB entry 2OH4,[21] with VEGFR-2
adopting the DFG-out conformation as previously reported for
AAL993.[27] For the missing residues Lys871–Ala874, Ca atoms
were added manually, and the rest of the missing atoms were au-
tomatically completed by the XLEaP module of AMBER 9.[33] Histi-
dine protonation states were predicted at pH 7.0 using the pro-
gram H++ with standard parameters for the Poisson–Boltzmann
electrostatics calculation method.[34] After visual inspection for H-
bonding ability of each histidine with neighboring residues, all
nine histidines were set to be protonated at Ne2 only. The other
basic and acidic residues were assigned their standard protonation
states. The parm99 AMBER force field was employed for the pro-
tein atoms,[35] and the TIP3P water model was used for the sol-
vent.[36] The phosphorylation site residues Tyr1054 and Tyr1059
were assigned AMBER force field parameters as reported previous-
ly.[37] VEGFR-2 ligands taken from crystal structures were assigned
AM1-BCC charges[38] and GAFF parameters[39] using the ANTE-
CHAMBER module of AMBER 9. Coordinates for the designed scaf-
folds were produced from SMILES representation using
OMEGA 2.2,[40] and were then assigned AM1-BCC charges using the
MOLCHARGE 1.3 module of QUACPAC.[41] The electrostatic poten-
tial of geometry-optimized ligands was calculated at the Hartree–
Fock level with 6-31G* basis using Gaussian 98.
Molecular dynamics (MD) calculations were carried out using the
SANDER and PMEMD programs of AMBER 9.[33] Periodic boundary
conditions were imposed by the particle mesh Ewald method, with
a limit of 8 ꢂ for the direct space sum. Numerical integration was
performed with a 2-fs time step, and all bonds involving hydrogen
atoms were constrained with SHAKE. Temperature and pressure
controls were imposed using a Berendsen-type algorithm[44] with
coupling constants of 1.0 and 2.0 ps, respectively. Either the free
protein or the protein–ligand complexes were immersed in isomet-
ric truncated octahedron water boxes, and then an appropriate
number of counterions were added to neutralize the system
charge. The following procedure was carried out to optimize the
water positions and to equilibrate the temperature and pressure of
the systems: a) energy minimization for 1000 steps using the
steepest-descent method with harmonic restraints of 50 kcal
molꢁ1 ꢂꢁ2 force constant on all solute atoms; b) restrained constant
volume dynamics (NVT ensemble) at 300 K for 30 ps; c) a second
energy minimization for 1000 steps with 10 kcalmolꢁ1 ꢂꢁ2 restraints
on all protein Ca atoms; d) the temperature was then gradually in-
creased to 300 K within six rounds of 5-ps constant volume dynam-
ics (NVT), while solute atoms were restrained with 10 kcal
molꢁ1 ꢂꢁ2; e) restraints were then gradually released within 20 ps in
the NVT ensemble at 300 K; f) the density of the systems was in-
creased from ~0.88 to ~1.04 gcmꢁ3 during 150 ps of constant
pressure dynamics (NPT ensemble). Subsequently, production runs
were carried out under physiological conditions (300 K, 1 atm) in
the NPT ensemble for a total time of 10 ns. The translational
center-of-mass motion was removed every 1000 steps, and trajec-
tories were updated every 500 steps. Analysis and visual inspection
of the MD trajectories were performed using the PTRAJ module of
AMBER 9 and VMD 1.8.6.[45] Figures were also prepared with VMD,
while plots were generated using the program GRACE. All calcula-
tions were carried out on an Intel Quad Core workstation, running
an X86_64 Linux 2.6-smp kernel, while AMBER 9 was compiled
using the Intel Fortran 9.1 compilers.
To relax the protein from crystal packing contacts and to optimize
the position of the modeled atoms, energy minimization was per-
formed in implicit solvent using a generalized Born solvation
model (GBHCT [42]
with a cutoff of 16 ꢂ for the nonbonded interac-
)
tions. An initial 2000-step round of minimization was carried out
using the steepest-descent method with positional restraints of
ChemMedChem 2010, 5, 118 – 129
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
127

D. Vourloumis et al.
MED
transferred to a Delfiaꢀ 96-well streptavidin-coated yellow plate
(PerkinElmer Life Sciences). After 1 h at room temperature, the so-
lution was removed, and the plate was washed with PBS/T (1ꢃPBS
with 0.05% Tween-20; 3ꢃ200 mL per well). The wells were then in-
cubated with primary antibody (100 mL per well), phosphotyrosine
mAb (P-Tyr-100), 1:1000 in PBS/T with 1% BSA, for 1 h at room
temperature. The solutions were then removed, and the wells
were washed with PBS/T (3ꢃ200 mL per well). The wells were then
incubated at room temperature for 30 min with anti-mouse IgG
HRP-labeled secondary antibody (100 mL per well), 1:500 in PBS/T
with 1% BSA (Cell Signaling Technology, USA). Unbound antibody
was removed, and the plate was washed with PBS/T (5ꢃ200 mL
per well). TMB substrate (100 mL; Cell Signaling Technology, USA)
was added to each well, and the plate was incubated at room tem-
perature for 5 min. Stop solution (100 mL; Cell Signaling Technolo-
gy, USA) was added to each well, and after incubating at room
temperature for 5 min, the absorbance at 450 nm was recorded
with an ELISA plate reader. Percent inhibition values were calculat-
ed according to Equation (1):
Biology and screening assays
Cell culture. HeLa (human cervical epithelioid carcinoma) cells
were cultured in Dulbecco’s modified Eagle’s medium (DMEM) sup-
plemented with 2 mm l-glutamine (PAA) and 10% fetal calf serum
(PAA). HUVECp (human umbilical vein endothelial cells pooled
from multiple isolates; Cascade Biologics) were cultured in Medium
200 supplemented with LSGS (Cascade Biologics). Cells used in ex-
periments were between passages 3 and 5. All cells were main-
tained at 378C under 5% CO2 as exponentially growing monolay-
ers.
Cell proliferation assays. The ability of the compounds to inhibit
proliferation of cell growth was determined using the MTT assay, as
described previously.[30] Briefly, 4ꢃ105 cells were trypsinized and
seeded into 96-well microplates (Orange Scientific, Belgium) in
complete medium up to a volume of 100 mL. Cells were incubated
for 24 h at 378C in an atmosphere containing 5% CO2. Medium
was removed, and fresh medium was added to each well, contain-
ing the appropriate concentration of each compound in a total
volume of 100 mL, and cells were incubated for an additional 24 h.
Afterward the supernatant was removed, and cells were washed
twice with 200 mL 1ꢃPBS. 1-(4,5-Dimethylthiazol-2-yl)-3,5-diphenyl-
formazan (MTT; Sigma Chemical Co., St. Louis, MO, USA) diluted in
complete medium to a final concentration of 1 mgmLꢁ1, was
added to each well in a volume of 100 mL. The plate was incubated
for 1.5 h at 378C, so that MTT (yellow) can be transformed into for-
mazan crystals (purple) by the viable cells. The supernatant was re-
moved, and the formazan crystals were solubilized for 10 min
upon addition of 100 mL DMSO. The absorbance of each cell lysate
solution was measured at 550 nm. Results from the MTT assays are
expressed as the mean value of the absorption at 550 nm ꢀ stan-
dard deviation from two independent experiments, which were re-
peated three times each. IC50 values represent the concentration of
each compound required for 50% decrease in cell viability, and
were estimated using sigmoid fitting.
% Inhibition ¼ f1ꢁ½ðAmeasuredꢁAminÞ=ðAmaxꢁAminÞꢂg ꢃ 100
ð1Þ
for which Ameasured is the mean value from two independent experi-
ments. Each IC50 value was determined by plotting the percent in-
hibition as a function of log([compound] in nm) using a sigmoidal
fit, and are presented in Table 1.
In vitro HUVEC tube formation assays. The in vitro angiogenesis
assay kit, using ECMatrixTM gel (Chemiconꢀ), was used according to
the manufacturer’s protocol. Briefly, matrix solution was solidified
on 96-well plates. HUVEC (up to passage 4) cell suspension of 1ꢃ
105 cellsmLꢁ1 was prepared in medium containing potential angio-
genesis inhibitors at concentrations of 10 and 50 mm, in the pres-
ence or absence of VEGF at 50 ngmLꢁ1. All compounds were dilut-
ed in 4% DMSO, (final concentration of DMSO in culture was 1%).
Cells were cultured in duplicates at a concentration of 1ꢃ104 cells
per well in ECMatrixTM-coated plates. Tube formation and cellular
networks were monitored under a light microscope. Cells were
stained with calcein, and images were acquired with a digital
camera under a fluorescence microscope.
In vitro kinase assays. IC50 values for the inhibition of VEGFR-2
(KDR) kinase using individual compounds were measured with an
HTScanꢀ VEGFR-2 kinase assay kit (Cell Signaling Technology, USA),
following the colorimetric ELISA protocol provided by the manu-
facturer. Recombinant human GST–VEGFR-2 kinase (Val789–
Val1356), biotinylated peptide substrate, and phosphotyrosine anti-
body for the detection of phosphorylated substrate were all sup-
plied within the kit. Before the reaction, a 2ꢃ ATP/substrate cock-
tail ([ATP]=40 mm, [substrate]=3 mm) and a 4ꢃ reaction cocktail
([enzyme]=8 ngmLꢁ1 in DTT/kinase buffer) were prepared. All com-
pounds were dissolved in DMSO and diluted with deionized H2O
to a final DMSO concentration of 4% (v/v). Control experiments
were carried out by adding DMSO alone; that is, no test com-
pounds were added into reactions containing either 2ꢃ ATP/sub-
strate and 4ꢃ reaction cocktail (Amax control), or 2ꢃ ATP/substrate
only (Amin control); 4ꢃ reaction cocktail (12.5 mL) was incubated at
room temperature for 5 min with 12.5 mL of the pre-diluted com-
pound of interest (concentrations per well ranging from 1 to
500 nm) as well as with 12.5 mL of Amax and Amin control samples;
2ꢃ ATP/substrate cocktail (25 mL) was added to pre-incubated
cocktail/compound (25 mL per well) in a 96-well reaction plate. The
final assay conditions per well for a 50-mL reaction were the follow-
ing: 60 mm HEPES (pH 7.5), 5 mm MgCl2, 5 mm MnCl2, 3 mm
Na3VO4, 1.25 mm DTT, 20 mm ATP, 1.5 mm peptide, and 100 ng
VEGFR-2 kinase. The reaction plate was incubated at room temper-
ature for 30 min. Each reaction was terminated by the addition of
’stop’ buffer (50 mL per well; 50 mm EDTA, pH 8). Subsequently,
25 mL of each reaction and 75 mL deionized H2O per well were
Acknowledgements
This work was supported by the research project ENTER2004
(04EP63), which is co-financed by the E.U.–European Social Fund
(75%) and the Greek Ministry of Development–GSRT (25%).
Keywords: angiogenesis · drug design · inhibitors · medicinal
chemistry · molecular dynamics · VEGFR-2
[1] D. J. Hicklin, L. M. Ellis, J. Clin. Oncol. 2004, 23, 1011–1027.
[2] N. Ferrara, R. S. Kerbel, Nature 2005, 438, 967–974.
[3] G. Neufeld, T. Cohen, S. Gengrinovitch, Z. Poltorak, FASEB J. 1999, 13, 9–
22.
[4] N. Ferrara, Nat. Rev. Cancer 2002, 2, 795–803.
[5] N. Ferrara, Endocr. Rev. 2004, 25, 581–611.
[6] N. Ferrara, K. J. Hillan, H. P. Gerber, W. Novotny, Nat. Rev. Drug Discovery
2004, 3, 391–400.
[7] M. Shibuya, Int. J. Biochem. Cell Biol. 2001, 33, 409–420.
[8] N. Ferrara, H. P. Gerber, J. LeCouter, Nat. Med. 2003, 9, 669–676.
[9] T. Tammela, B. Enholm, K. Alitalo, K. Paavonen, Cardiovasc. Res. 2005, 65,
550–563.
[10] T. Takahashi, S. Yamaguchi, K. Chida, M. Shibuya, EMBO J. 2001, 20,
2768–2778.
128
www.chemmedchem.org
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2010, 5, 118 – 129

Potent VEGFR-2 Inhibitors
[11] T. L. Underiner, B. Ruggeri, D. E. Gingrich, Curr. Med. Chem. 2004, 11,
731–745.
[12] G. Kesisis, H. Broxterman, G. Giaccone, Curr. Pharm. Des. 2007, 13,
2795–2809.
[13] H. Zhong, J. P. Bowen, Curr. Top. Med. Chem. 2007, 7, 1379–1393.
[14] A. Kiselyov, K. V. Balakin, S. E. Tkachenko, Expert Opin. Invest. Drugs
2007, 16, 83–107.
mann, G. Martiny-Baron, J. Mestan, J. Rçsel, M. Sills, D. Stover, F. Acemo-
glu, E. Boss, R. Emmenegger, L. Lꢅsser, E. Masso, R. Roth, C. Schlachter,
W. Vetterli, D. Wyss, J. M. Wood, J. Med. Chem. 2000, 43, 2310–2323.
[26] P. W. Manley, P. Furet, G. Bold, J. Brꢄggen, J. Mestan, T. Meyer, C. R.
Schnell, J. Wood, M. Haberey, A. Huth, M. Krꢄger, A. Menrad, E. Ottow,
D. Seidelmann, G. Siemeister, K. H. Thierauch, J. Med. Chem. 2002, 45,
5687–5693.
[15] S. Schenone, F. Bondavalli, M. Botta, Curr. Med. Chem. 2007, 14, 2495–
2516.
[16] M. A. McTigue, J. A. Wickersham, C. Pinko, R. E. Showalter, C. Parast V, A.
Tempczyk-Russell, M. R. Gehring, B. Mroczkowski, K. Chen-Chen, J. E. Vil-
lafranca, K. Appelt, Structure 1999, 7, 319–330.
[17] P. A. Harris, M. Cheung, R. N. Hunter Iii, M. L. Brown, J. M. Veal, R. T.
Nolte, L. Wang, W. Liu, R. M. Crosby, J. H. Johnson, A. H. Epperly, R.
Kumar, D. K. Luttrell, J. A. Stafford, J. Med. Chem. 2005, 48, 1610–1619.
[18] Y. Miyazaki, S. Matsunaga, J. Tang, Y. Maeda, M. Nakano, R. J. Philippe,
M. Shibahara, W. Liu, H. Sato, L. Wang, R. T. Nolte, Bioorg. Med. Chem.
Lett. 2005, 15, 2203–2207.
[27] P. W. Manley, G. Bold, J. Brꢄggen, G. Fendrich, P. Furet, J. Mestan, C.
Schnell, B. Stolz, T. Meyer, B. Meyhack, W. Stark, A. Strauss, J. Wood, Bio-
chim. Biophys. Acta Proteins Proteomics 2004, 1697, 17–27.
[28] M. Huse, J. Kuriyan, Cell 2002, 109, 275–282.
[29] P. Cozzini, G. E. Kellogg, F. Spyrakis, D. J. Abraham, G. Costantino, A.
Emerson, F. Fanelli, H. Gohlke, L. A. Kuhn, G. M. Morris, M. Orozco, T. A.
Pertinhez, M. Rizzi, C. A. Sotriffer, J. Med. Chem. 2008, 51, 6237–6255.
[30] T. Mosmann, J. Immunol. Methods 1983, 65, 55–63.
[31] L. N. Johnson, M. E. M. Noble, D. J. Owen, Cell 1996, 85, 149–158.
[32] S. Foister, M. A. Marques, R. M. Doss, P. B. Dervan, Bioorg. Med. Chem.
2003, 11, 4333–4340.
[19] B. L. Hodous, S. D. Geuns-Meyer, P. E. Hughes, B. K. Albrecht, S. Bellon, J.
Bready, S. Caenepeel, V. J. Cee, S. C. Chaffee, A. Coxon, M. Emery, J. Fret-
land, P. Gallant, Y. Gu, D. Hoffman, R. E. Johnson, R. Kendall, J. L. Kim,
A. M. Long, M. Morrison, P. R. Olivieri, V. F. Patel, A. Polverino, P. Rose, P.
Tempest, L. Wang, D. A. Whittington, H. Zhao, J. Med. Chem. 2007, 50,
611–626.
[20] M. H. Potashman, J. Bready, A. Coxon, T. M. Demelfi, Jr., L. DiPietro, N.
Doerr, D. Elbaum, J. Estrada, P. Gallant, J. Germain, Y. Gu, J. C. Harmange,
S. A. Kaufman, R. Kendall, J. L. Kim, G. N. Kumar, A. M. Long, S. Neer-
vannan, V. F. Patel, A. Polverino, P. Rose, S. Van Der Plas, D. Whittington,
R. Zanon, H. Zhao, J. Med. Chem. 2007, 50, 4351–4373.
[21] M. Hasegawa, N. Nishigaki, Y. Washio, K. Kano, P. A. Harris, H. Sato, I.
Mori, R. I. West, M. Shibahara, H. Toyoda, L. Wang, R. T. Nolte, J. M. Veal,
M. Cheung, J. Med. Chem. 2007, 50, 4453–4470.
[22] J. C. Harmange, M. M. Weiss, J. Germain, A. J. Polverino, G. Borg, J.
Bready, D. Chen, D. Choquette, A. Coxon, T. DeMelfi, L. DiPietro, N.
Doerr, J. Estrada, J. Flynn, R. F. Graceffa, S. P. Harriman, S. Kaufman, D. S.
La, A. Long, M. W. Martin, S. Neervannan, V. F. Patel, M. Potashman, K.
Regal, P. M. Roveto, M. L. Schrag, C. Starnes, A. Tasker, Y. Teffera, L.
Wang, R. D. White, D. A. Whittington, R. Zanon, J. Med. Chem. 2008, 51,
1649–1667.
[33] D. A. Case, T. E. Cheatham Iii, T. Darden, H. Gohlke, R. Luo, K. M.
Merz, Jr., A. Onufriev, C. Simmerling, B. Wang, R. J. Woods, J. Comput.
Chem. 2005, 26, 1668–1688.
[34] J. C. Gordon, J. B. Myers, T. Folta, V. Shoja, L. S. Heath, A. Onufriev, Nucle-
ic Acids Res. 2005, 33, W368–W371.
[35] W. D. Cornell, P. Cieplak, C. I. Bayly, I. R. Gould, K. M. Merz, Jr., D. M. Fer-
guson, D. C. Spellmeyer, T. Fox, J. W. Caldwell, P. A. Kollman, J. Am.
Chem. Soc. 1995, 117, 5179–5197.
[36] W. L. Jorgensen, J. Chandrasekhar, J. D. Madura, R. W. Impey, M. L. Klein,
J. Chem. Phys. 1983, 79, 926–935.
[37] N. Homeyer, A. H. C. Horn, H. Lanig, H. Sticht, J. Mol. Model. 2006, 12,
281–289.
[38] A. Jakalian, D. B. Jack, C. I. Bayly, J. Comput. Chem. 2002, 23, 1623–1641.
[39] J. Wang, R. M. Wolf, J. W. Caldwell, P. A. Kollman, D. A. Case, J. Comput.
Chem. 2004, 25, 1157–1174.
[40] R. Kristam, V. J. Gillet, R. A. Lewis, D. Thorner, J. Chem. Inf. Model. 2005,
45, 461–476.
[41] OpenEye Scientific Software, http://eyesopen.com (accessed October
28, 2009).
[42] V. Tsui, D. A. Case, Biopolymers 2000, 56, 275–291.
[43] G. M. Morris, D. S. Goodsell, R. S. Halliday, R. Huey, W. E. Hart, R. K.
Belew, A. J. Olson, J. Comput. Chem. 1998, 19, 1639–1662.
[44] H. J. C. Berendsen, J. P. M. Postma, W. F. Van Gunsteren, A. Dinola, J. R.
Haak, J. Chem. Phys. 1984, 81, 3684–3690.
[23] D. S. La, J. Belzile, J. V. Bready, A. Coxon, T. DeMelfi, N. Doerr, J. Estrada,
J. C. Flynn, S. R. Flynn, R. F. Graceffa, S. P. Harriman, J. F. Larrow, A. M.
Long, M. W. Martin, M. J. Morrison, V. F. Patel, P. M. Roveto, L. Wang,
M. M. Weiss, D. A. Whittington, Y. Teffera, Z. Zhao, A. J. Polverino, J. C.
Harmange, J. Med. Chem. 2008, 51, 1695–1705.
[45] W. Humphrey, A. Dalke, K. Schulten, J. Mol. Graph. 1996, 14, 33–38.
[24] E. Hu, A. Tasker, R. D. White, R. K. Kunz, J. Human, N. Chen, R. Bꢄrli, R.
Hungate, P. Novak, A. Itano, X. Zhang, V. Yu, Y. Nguyen, Y. Tudor, M.
Plant, S. Flynn, Y. Xu, K. L. Meagher, D. A. Whittington, G. Y. Ng, J. Med.
Chem. 2008, 51, 3065–3068.
[25] G. Bold, K. H. Altmann, J. Frei, M. Lang, P. W. Manley, P. Traxler, B. Wiet-
feld, J. Brꢄggen, E. Buchdunger, R. Cozens, S. Ferrari, P. Furet, F. Hof-
Received: September 4, 2009
Revised: October 22, 2009
Published online on November 17, 2009
ChemMedChem 2010, 5, 118 – 129
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
129