Optimization of potent dfg-in inhibitors of platelet derived growth factor receptorβ (PDGF-Rβ) guided by water thermodynamics

Article abstract of DOI:10.1021/jm500373x

In this study we report on the hit optimization of substituted 3,5-diaryl-pyrazin-2(1H)-ones toward potent and effective platelet-derived growth factor receptor (PDGF-R) β-inhibitors. Originally, the 3,5-diaryl-pyrazin-2-one core was derived from the marine sponge alkaloid family of hamacanthins. In our first series compound 2 was discovered as a promising hit showing strong activity against PDGF-Rβ in the kinase assay (IC₅₀ = 0.5 μM). Furthermore, 2 was shown to be selective for PDGF-Rβ in a panel of 24 therapeutically relevant protein kinases. Molecular modeling studies on a PDGF-Rβ homology model using prediction of water thermodynamics suggested an optimization strategy for the 3,5-diaryl-pyrazin-2-ones as DFG-in binders by using a phenolic OH function to replace a structural water molecule in the ATP binding site. Indeed, we identified compound 38 as a highly potent inhibitor with an IC₅₀ value of 0.02 μM in a PDGF-Rβ enzymatic assay also showing activity against PDGF-R dependent cancer cells.

Full text of DOI:10.1021/jm500373x

Article
pubs.acs.org/jmc
Optimization of Potent DFG-in Inhibitors of Platelet Derived Growth
Factor Receptorβ (PDGF-Rβ) Guided by Water Thermodynamics
Rebecca Horbert,†,∥ Boris Pinchuk,†,∥ Eugen Johannes,† Joachim Schlosser,† Dorian Schmidt,†
Daniel Cappel,‡ Frank Totzke,§ Christoph Schachtele,§ and Christian Peifer*,†
̈
†Institute of Pharmacy, Christian-Albrechts-University of Kiel, Gutenbergstraße 76, D-24116 Kiel, Germany
‡Schrodinger GmbH, Dynamostraße 13, D-68165 Mannheim, Germany
̈
§ProQinase GmbH, Breisacherstraße 117, D-79106 Freiburg, Germany
S
* Supporting Information
ABSTRACT: In this study we report on the hit optimization of substituted
3,5-diaryl-pyrazin-2(1H)-ones toward potent and effective platelet-derived
growth factor receptor (PDGF-R) β-inhibitors. Originally, the 3,5-diaryl-
pyrazin-2-one core was derived from the marine sponge alkaloid family of
hamacanthins. In our first series compound 2 was discovered as a promising
hit showing strong activity against PDGF-Rβ in the kinase assay (IC50 = 0.5
μM). Furthermore, 2 was shown to be selective for PDGF-Rβ in a panel of 24
therapeutically relevant protein kinases. Molecular modeling studies on a
PDGF-Rβ homology model using prediction of water thermodynamics
suggested an optimization strategy for the 3,5-diaryl-pyrazin-2-ones as DFG-in
binders by using a phenolic OH function to replace a structural water
molecule in the ATP binding site. Indeed, we identified compound 38 as a
highly potent inhibitor with an IC50 value of 0.02 μM in a PDGF-Rβ
enzymatic assay also showing activity against PDGF-R dependent cancer cells.
pound 1, Chart 2), a potent inhibitor of VEGF-R2/3 (IC50
0.03 μM) with good efficacy in cellular assays.7
=
INTRODUCTION
■
In our ongoing study to develop ATP-competitive receptor
tyrosine kinase (RTK) inhibitors with anticancer activity, we
focused on the hamacanthin B family of deep-sea sponge
derived bis-indole alkaloids possessing a 3,5-bisindole-3,4-
dihydropyrazin(1H)-2-one scaffold.1 In line with this notion,
cis-3,4-dihydropyrazin(1H)-2-one hamacanthin B was reported
to be a potent bacterial methicillin-resistant Staphylococcus
protein kinase inhibitor (MRSA-PK inhibitor) with an IC50
value of 0.016 μM and possessing significant selectivity over
human protein kinase isoforms.2 Moreover, substituted (1H)-
pyrazin-2-ones have been investigated as human protein kinase
(PK) inhibitors showing this moiety to be suitable as a core
scaffold for PK inhibitor design.3 Among human PKs in
oncology, overactivated RTK including VEGFR, PDGF-R, and
c-kit are considered to be major targets for the development of
clinically effective inhibitors.4 Thus, many anticancer com-
pounds that are advanced into the clinic show (group-)
selectivity toward VEGF-R, FGF-R, EGF-R, PDGF-R, c-kit, and
Flt-3 (Chart 1).5 Furthermore, inhibitors addressing a single PK
with high selectivity are of significant interest. A recent study
reported about the development of imidazo[1,2-α]pyridines
with potent PDGF-R activity and oral bioavailability.6
RESULTS AND DISCUSSION
■
Chemistry. The test compounds reported in this study were
synthesized by our flexible synthetic platform to produce
targeted 3,5-diaryl-pyrazin-2(1H)-ones (VI, Scheme 1).8 As key
intermediates for the final ring closure, the open-chained
diketoamides (V) were accessible via two different routes. On
the one hand, arylglyoxylic acid (I) was activated by
carbonyldiimidazole (CDI) and coupled with the α-ketoamine
moiety (II) to produce the open-chained diketoamide (V). On
the other hand, CDI mediated coupling of arylglyoxylic acid (I)
and amine (III) gave amide (IV), which yielded ketoamide (V)
upon DDQ-oxidation. Microwave-mediated ring closure in the
final step by using the open-chained diketoamide (V) and
ammonium acetate as nitrogen source yielded the desired 3,5-
diaryl-pyrazin-2(1H)-ones (VI).
Molecular Modeling and Binding Mode. In a prelimi-
nary in vitro screening involving 24 therapeutically relevant PK,
compound 2 was actually determined to potently inhibit
PDGF-Rβ with an IC50 of 0.5 μM.1 Moreover, compound 2
was shown to be selective over the other PK enzymes tested in
Special Issue: New Frontiers in Kinases
Received: March 10, 2014
Therefore, we designed the aryl-substitution patterns of the
6-membered 3,5-diaryl-pyrazin-2(1H)-one 21 based on the
corresponding 5-membered 3,4-diaryl-2H-pyrrole-2-one (com-
© XXXX American Chemical Society
A
dx.doi.org/10.1021/jm500373x | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
Chart 1. Selection of Small Molecule Kinase Inhibitors (smKI, Approved or in Clinical Phases) Targeting PDGF-R (among
Other PKs)
this panel (VEGF-R2 IC50 = 4 μM, VEGF-R3 IC50 = 5 μM,
FLT3 IC50 = 36 μM; IC50-values >100 μM: AKT1, ARK5,
Aurora-A/B, B-RAF-VE, CDK2/CycA, CDK4/CycD1, COT,
EGFR, EPHB4, ERBB2, FAK, IGF1R, SRC, INSR, MET,
PLK1, SAK, TIE2, and CK2a1). Thus, we aimed to further
enhance the potency of this hit compound toward PDGF-Rβ
by using a homology model/docking approach since no X-ray
structure of the PDGF-R PK domain is available in the public
domain. Besides, the vast majority of ATP-competitive
inhibitors can be categorized in type I or type II binders.9
That is, binding two different conformations of the ATP site,
namely, the “DFG-in” (type I) and “DFG-out” (type II)
conformation. In order to address the question to which
PDGF-Rβ-conformation compound 2 is likely to bind, we
compared the molecular structure of 2 to a VEGF-R2-DFG-in-
ligand (type I inhibitor pyridinyl-triazine, pdb 2p2h10) and to a
prototypical DFG-out binder (type II inhibitor imatinib, pdb
2hyy11), respectively (Figure 1).
The binding pose overlay of the inhibitor structures in Figure
1 (shown as bound to the ATP binding pocket) suggests that
compound 2 represents most likely a DFG-in binder. First,
compared to the VEGF-R2-type-I pyridinyl-triazine inhibitor,
compound 2 is matching well in terms of molecular shape.
Furthermore, docked compound 2 is also occupying a
B
dx.doi.org/10.1021/jm500373x | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
plausible DFG-in binding modes were successfully calculated by
docking experiments with 2 modeled in several DFG-in ATP-
binding pockets (not shown) possessing significant sequence
similarity to PDGF-Rβ (VEGF-R2, pdb 2p2h,10 3cjg,14 and
3c7q;15 fms, pdb 3lcd16). Therefore, on the basis of the highly
related RTK VEGF-R2 we generated two PDGF-Rβ homology
models, namely, using VEGF-R2 template structures in the
DFG-in (pdb 2p2h10) and DFG-out conformation (pdb 1t4612),
respectively. However, because only the DFG-in homology
model proved to predict reasonable binding poses for 2 in the
ATP binding pocket, this structure was used for further
modeling studies in this project (Figure 2, pdb 2p2h;
Schrodinger Prime,17 version 3.2, Schrodinger, LLC, New
Chart 2. Chemical Structures of the Potent 3,4-Diaryl-2H-
pyrrole-2-one VEGF-R2/3 Inhibitor 1 and 3,5-Diaryl-
pyrazin-2(1H)-one 2 of This Study Showing Comparable
Decoration Patterns of the Aryl Moieties
̈
̈
York, NY, USA, 2013).
In order to validate the modeled hinge-binding mode of 2 in
the ATP site of our PDGF-Rβ DFG-in homology model
(Figure 2), we first synthesized and evaluated compounds 3
and 4 for their potency against PDGF-Rβ (Chart 3).
Accordingly to our docking results the introduction of the 6-
methyl group at the 3,5-diaryl-pyrazin-2(1H)-one core
(compound 3) is blocking the inhibitor’s dual H-bond
interactions to the hinge region (Glu101/Cys103) and thus
killing the biological activity of 3.
Scheme 1. General Preparation Scheme of 3,5-Diaryl-
pyrazin-2(1H)-ones (VI)
a
Likewise, compared to compound 2, the lactam-regioiso-
meric structure 4 should be able to address an H-bond only to
Glu101 but not be able to accept the H-bond from Cys103 thus
showing reduced biological activity. In fact, the biological
activities of 3 and 4 are in line with the proposed binding mode
demonstrating a valid modeled pose of 2 in the active site of
PDGF-Rβ. Motivated by the in vitro confirmation of our DFG-
in homology model based binding mode, we retained the 3,5-
diaryl-pyrazin-2(1H)-one core and next focused on the 3-aryl
moiety.
a
The ketoamides (V) were prepared in two different routes, either via
CDI mediated coupling of arylglyoxylic acid (I) and α-ketoamines (II)
or via CDI mediated coupling of arylglyoxylic acid (I) and
indolethylamines (III) to produce 2-oxo-2-phenyl-acetamides (IV),
which yielded the corresponding (V) upon DDQ-oxidation. Final
compounds (VI) were prepared in a microwave reaction using
ammonium acetate as nitrogen source (method d). For clarity, the test
compounds were numbered consecutively (2−39). Analytical data for
all relevant compounds is available in Supporting Information.
Optimization of the 3-Aryl Substitution Pattern.
According to the calculated binding mode the 3-aryl part of 2
is situated in the HRII which opens to the solvent (Figure 2).
To further investigate the structure−activity relationships
(SARs) and to optimize the 3-aryl system we used a
straightforward docking/scoring approach. Although no direct
ligand−protein interactions are involved in this area, our
molecular docking campaign showed slight differences in the
predicted binding affinities for variations of the methoxy-aryl
substitution. As a result, compound 2 was top-ranked followed
by 5, 15, and 7 (series 1, Table 1). For compounds 8 and 9
bearing an ortho-methoxy substitution (causing a 1H-pyrazin-2-
one/3-aryl dihedral angle of almost 90°), no binding mode was
predicted in the narrow ATP pocket. Further 3-aryl variations
of 2 showed no significant docking scores. Hence, derived from
the modeling approach we synthesized a selection of
compounds (5−15 showing systematic variations of the 1H-
pyrazin-2-one-3-aryl moiety, series 1, Table 1), which were
subsequently tested against PDGF-Rβ. Actually the docking/
scoring results were reflected by the in vitro testing data. In
accordance with the data of this series the 3-(3,4,5-
trimethoxyphenyl) moiety in 2 was determined to be optimal:
decreasing the number of methoxy groups (5 and 6) as well as
positional changes of the aryl-methoxy groups (7, 8, and 9)
resulted in diminished ligand affinity. Also, compared to hit
compound 2, the biological activity in this series could not be
increased by introducing other moieties into the 3-aryl system
such as morpholine (12), 4-fluor (13), 4-ethyl, or indole (15).
Optimization of the 5-Aryl Substitution Pattern.
Therefore, we retained the 3-(3,4,5-trimethoxyphenyl)-1H-
Figure 1. Left: Overlay of modeled docking pose of 2 (gray color) as
oriented in the ATP binding pocket of VEGF-R2 and crystallo-
graphically determined binding mode of DFG-in/type I VEGF-R2
inhibitor 4-(2-anilinopyridin-3-yl)-N-(3,4,5-trimethoxyphenyl)-1,3,5-
triazin-2-amine (green color, pdb 2p2h). Right: Superposition of
docking pose of 2 (gray color) and DFG-out/type II inhibitor imatinib
(light blue color, pdb 2hyy).
comparable space in the VEGF-R2 binding pocket. Second, the
deep pocket binding part of the well-established DFG-out/type
II-binder imatinib, namely, the 4-[(4-methylpiperazin-1-yl)-
methyl]benzamide moiety, is not present in 2 (Figure 1, right).
Third, docking of compound 2 into PK structures related to
PDGF-Rβ possessing DFG-out conformations (Abl-kinase, pdb
2hyy;11 VEGF-R2, pdb 3vhk9 and 1t46;12 c-src pdb, 3f3w13)
revealed predicted binding poses in the ATP site not consistent
with the originally present type II inhibitors. In contrast,
C
dx.doi.org/10.1021/jm500373x | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
Figure 2. Left: DFG-in homology model of the PDGF-Rβ kinase domain based on the template structure of crystallographically determined VEGF-
R2 (pdb 2p2h) and modeled binding mode of hit compound 2 (brown color) in the ATP binding pocket. Right: Ligand interaction diagram of
modeled binding mode of 2 in the ATP binding pocket of the PDGF-Rβ PK domain. Key amino acid residues and ligand-active site interactions are
shown. The central pyrazin-2(1H)-one core is involved in a dual H-bond interaction to the hinge region residues Glu101 and Cys103. The pyrazin-
2(1H)-one 3-(3,4,5-trimethoxy) part is situated in the hydrophobic region II (HRII), whereas the pyrazin-2(1H)-one-5-(indole-3yl) part is
occupying the hydrophobic pocket I (HPI).
Chart 3. Chemical Structures and Biological Activity against PDGF-Rβ (in Vitro Kinase Assay) of Key 3,5-Diaryl-pyrazin-2(1H)-
ones 2, 3, and 4 to Confirm the Calculated Hinge-Binding Motif of 2 in the Homology-Modeled ATP Binding Pocket of the
PDGF-Rβ PK
pyrazin-2-one system, and for further optimization of 2 we next
concentrated on variations of the 5-indole moiety, which is
situated in the buried HPI (Figure 2). We first generated a
focused virtual set of substituted 5-indole variations of
compound 2 and performed docking experiments using the
Table 1. Series 1: SAR Data of Variations of Hit Compound
2 (PDGF-Rβ IC50 = 0.5 μM) Regarding the Pyrazin-2(1H)-
one-3-aryl Substitution Pattern; Biological Activity Was
Determined for Each Compound against PDGF-Rβ in an in
Vitro Kinase Assay
PDGF-Rβ homology model structure (Schrodinger GlideSP,18
̈
version 5.9, Schrodinger, LLC, New York, NY, USA, 2013).
̈
From the top-ranked docking hits (Figure 3) we prepared 14
synthetically accessible compounds and tested those for their
biological activity against PDGF-Rβ (series 2, Table 2).
In line with the calculated modeling data compounds, 16, 24,
and 27 were indeed determined to show good activity against
PDGF-Rβ (Table 2). The naphtyl moiety of 24 (IC50 PDGF-
Rβ = 1 μM) is filling the HPI, whereas the 5′-hydroxyindole in
16 (IC50 PDGF-Rβ = 0.3 μM) is addressing an additional H-
bond to the gatekeeper residue Thr100 (Figure 3). Only the
biological activity against PDGF-Rβ of compound 16 (showing
the 5′-hydroxyindole decoration) was determined to be in the
sub-μM range (as was the parent compound 2). However, the
flat in vitro SARs for further compounds of this 1H-pyrazin-2-
one-5-aryl series 2 (17−29, see Table 2) poorly matches the
docking/scoring of the compounds. In contrast to the good
correlation of predicted ligand−protein interactions of our
PDGF-Rβ homology model with the SAR (series 1), so far
these results may indicate a structural misarrangement
regarding the HPI in our model. However, the inconsistency
D
dx.doi.org/10.1021/jm500373x | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
Figure 3. Ligand interaction diagram of top-ranked compounds 24 (left) and 16 (right) in the ATP binding site of the PDGF-Rβ homology model.
Key amino acid residues and ligand−active site interactions are shown. GlideSP ranking of the compounds (top-down): 24, 16, 27, 29, 18, 28, 22,
21, 20, 19, 2, 17, 16, 25, and 23.
trimethoxyphenyl)-1H-pyrazin-2-one scaffold (compounds
30−38, series 3, Table 3). Herein, the concept of the 5-phenyl
decoration was based on the idea to investigate SARs in terms
of the displacement of hydration sites. In turn, all compounds
of this project were docked into the ATP pocket of our PDGF-
Rβ homology model. As an outcome, compound 38 was
calculated to be the top hit of this docking campaign. Strikingly,
the predicted high affinity of 38 was confirmed by the in vitro
assay (38 IC50 PDGF-Rβ = 0.02 μM, Table 3) demonstrating
this compound to be the most potent inhibitor of all three
series. The docking pose actually shows that the majority of key
hydration sites calculated by WaterMap were displaced by the
inhibitor (Figure 4). In detail, the 3-(3,4,5-trimethoxyphenyl)
moiety displaces all relevant hydration sites in the HRII. In the
hinge region two of the most unstable hydration sites (colored
in red) are displaced by the lactam moiety of the 1H-pyrazin-2-
one core. This approach also offers a strong structural
explanation for the comparable high affinity of this lactam-
regioisomer (also see 2 versus 4, Chart 3). Most notably, within
the HPI the 4′-phenolic OH of 38 is not only displacing an
unstable hydration site (colored in light red) but this OH group
is also replacing a stable water molecule (colored in green),
positioning the OH function in an H-bond network between
Lys53 and Asp180 (Figure 4). This water replacement should
be unfavorable in terms of entropy but can be compensated by
enthalpy summing up to a total decrease of Gibbs free energy
(ΔG). In addition, the vicinal 3′-methoxy group of 38 displaces
two unstable hydration sites in the rear HPI. However, two
hydration sites imbedded by Val33, Val84, Ile98, and Met74
were not yet affected by the present inhibitors. Hence, the
displacement of these water molecules will be the subject of our
ongoing inhibitor optimization strategy.
Considering the displacement of hydration sites (Figures 4
and 5; for complete data, see Supporting Information), a
differentiated analysis regarding the contribution of various
functional groups at the phenyl moiety toward ligand affinity
can be discussed21 based on the SAR data of series 3 (Table 3).
The phenolic-4′-OH is essential for potent ligand affinity
toward PDGF-Rβ as compounds 35 and 38 show IC50values in
the sub-μM range. In contrast, compounds having 4′-methoxy
(30 and 37) or 4′-chloro (32) substitutions were determined to
be significantly less active. Interestingly, the phenyl-3′-
substituted compounds 31 (3-methoxy, IC50 PDGF-Rβ = 1.4
μM) and 33 (3-chloro, IC50 PDGF-Rβ = 0.8 μM) are moderate
binders. This may be due to displacement of the two hydration
sites #8 and #83 with unfavorable ΔG (see Figure 5 and
Supporting Information) at the rear pocket while leaving the
Table 2. Series 2: SAR Data of Variations of Hit Compound
2 (PDGF-Rβ IC50 = 0.5 μM) Regarding the Pyrazin-2(1H)-
one-5-aryl Substitution Pattern
of the modeling/in vitro SAR correlation for this pocket could
refer to particular parameters triggering ligand affinity, which
are not sufficiently implemented in the applied docking/scoring
process.
WaterMap Calculations. Namely, this phenomen may be
explained by the thermodynamic contribution of displacing
water molecules, predominantly in hydrophobic binding
pockets. It is considered to be a key factor for ligand affinity
(“hydrophobic effect”).19 In order to investigate if entropic
critically water molecules in the ATP binding pocket of PDGF-
Rβ could be predictable parameters for the optimization of 3,5-
diaryl-pyrazin-2(1H)-ones in our project, we calculated
hydration sites in the PDGF-Rβ homology model apo-structure
by using the WaterMap technology17,18 (Figure 4). The
WaterMap method combines calculations for molecular
dynamics, solvent clustering, and statistical thermodynamics
to assess the enthalpy, entropy, and free energy of water
“hydration sites”. Moreover, WaterMap has been successfully
applied to study parameters driving selectivity for ligands in the
ATP pocket of protein kinases.20 Having the WaterMap results
for the PDGF-Rβ homology model apo structure in hand, we
designed a set of 5-phenyl variations of the 3-(3,4,5-
E
dx.doi.org/10.1021/jm500373x | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
Figure 4. Top: Docking pose of 38 (PDGF-Rβ IC50 = 0.02 μM) in the ATP binding pocket of the PDGF-Rβ PK homology model domain showing
the molecular surface (docking by Glide) and overlay of hydration sites (colored spheres; for color code, see below) calculated by WaterMap.
Bottom: Ligand−protein interaction diagram of 38 in the ATP binding pocket of the PDGF-Rβ homology model. Key hydration sites from
WaterMap calculations on the PDGF-Rβ homology model apo-structure are superimposed to illustrate the displacement of hydration sites by the
inhibitor. Hydration sites shown as red spheres represent “unstable” water molecules. Their displacement results in an increase of environmental
entropy (“hydrophobic effect”). Green spheres symbolize “stable” water molecules, which should not be displaced by an inhibitor in terms of
unfavorable enthalpic effects.
structural water #13 between Lys53 and Asp180 in place (with
favorable ΔG). The phenyl-2′-substituted compounds 34 (2-
chloro, IC50 PDGF-Rβ = 5 μM) and 36 (phenolic-2-OH, IC50
PDGF-Rβ = 2 μM) also showing moderate affinity indicate
sterical space at this part of the binding site (hydration site
#34). Thus, the 2′-phenyl position of this series may be suitable
for further ligand optimization regarding interactions to the
HPI.
On the basis of the PDGF-Rβ IC50 values of compounds 31,
35, 38, and 39, we quantified the relative contribution of the 3′-
methoxy and the 4′-hydroxyphenyl substituents toward the
experimental ΔG of these inhibitors. The values of ΔG are
based on the calculation21 of ΔG = −RT·ln IC50, Figure 5).
Within this set of compounds the 4′-hydroxy moiety
contributes approximately −11 kJ/mol to the actual free
binding enthalpy, while the 3′-methoxy moiety (∼−4.3 kJ/mol)
adds 3-fold less value to ΔG. In this regard, the WaterMap
calculations are in good accordance: the 4′-hydroxy moiety is
replacing the hydration site #13 for which an entropic term of
15.5 kJ/mol has been calculated. The 3′-methoxy moiety is
displacing hydration site #83 with a calculated −TΔS of 4.8 kJ/
mol (indicating the methoxy-oxygen to partly compensates the
hydration site oxygen). In addition to the SAR discussed above,
these numbers further highlight the importance of the phenyl
4′-hydroxy moiety in terms of replacing a structural water
molecule.
The most potent PDGF-Rβ-inhibitors of the 1H-pyrazin-2-
ones were tested against the closely related PK VEGF-R2 and
c-kit (Table 4). Herein, compound 38 showed only moderate
10-fold selectivity for PDGF-Rβ over VEGF-R2 and c-kit
suggesting comparable DFG-in binding modes in these kinases,
which is in line with the findings of Furet et al.22 In their study
it is shown that PK having a cysteine residue in the DFG-minus-
1-position favor DFG-in binding of ligands that can form a
sulfur−aromatic interaction to this cysteine. In fact, in the
binding mode of the ligands in our PDGF-Rβ model this is the
case for Cys179 (D180-F181-G182, also see LID Figure 2,
right, and Figure 4, bottom) explaining why in addition to
F
dx.doi.org/10.1021/jm500373x | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
Table 3. Series 3: SAR Data of Variations of Hit Compound
Table 4. Biological Activity of Selected Compounds against
PDGF-Rβ and the Closely Related PK VEGF-R2 and c-kit
(Activated Kinase Assays, for Details See Experimental
Section)
2 (PDGF-Rβ IC50 = 0.5 μM) Regarding the Pyrazin-2(1H)-
a
one-5-phenyl Substitution Pattern
PDGF-Rβ
IC50 (μM)
VEGF-R2
c-kit
#
IC50 (μM)
IC50 (μM)
2
0.5
0.3
1.4
0.8
0.1
0.02
2.3
1.0
23
0.9
2.7
7.4
2.4
0.4
0.2
16
31
33
35
38
56
2.2
0.2
Furthermore, compound 38 was profiled for selectivity at a
concentration of 1 μM in a panel of 300 wild-type kinases.
Herein, 38 most potently inhibited PDGF-Rβ (residual activity
= 1%) but also blocked further PKs, albeit to a lesser extent
(Figure 6, also see Supporting Information). However,
compound 38 could be shown to be a promising hit with
reasonable selectivity for further optimization toward an
effective PDGF-Rβ inhibitor for clinical use.
Cellular Assays. Furthermore, we evaluated a selection of
potent inhibitors of this project for their cytotoxic profiles using
five different cell lines including HL-60, a human myeloblastic
leukemia cell line (Figure 7). It has been shown that the
proliferation and differentiation particularly of the HL-60 cells
depend upon PDGF-R signaling.23 Cells were treated with the
a
The design of the phenyl decoration was based on joined WaterMap/
Glide docking calculations. The biological activity against PDGF-Rβ of
compound 38 (PDGF-Rβ IC50 = 0.02 μM) showing the phenyl-3′-
methoxy-4′-hydroxy decoration could be significantly improved
compared to the parent hit compound 2.
PDGF-Rβ these compounds also potently inhibit VEGF-R2
and c-kit (and Flt3, see Table 4 and Figure 6).
Figure 5. Top: Overlay of 38 (only polar hydrogens shown) and selected WaterMap hydration sites (within 5 Å of the ligand and ΔG > 4.2 kJ/mol
or < −4.2 kJ/mol) in the PDGF-Rβ homology model apo-structure. The hydration sites are numbered consecutively and are colored based on their
respective ΔG values in kJ/mol from red/unstable to green/stable. For the key hydration sites #13 and #83, calculated ΔG, enthalpies, and entropies
(in kJ/mol) with respect to bulk solvent are shown (complete data for all hydration sites can be found in the Supporting Information). Below:
Experimentally determined relative contribution of 3′-methoxy and 4′-hydroxy phenyl-substituents to ΔG of the inhibitors 31 (IC50 = 1.4 μM), 35
(IC50 = 0.1 μM), 38 (IC50 = 0.02 μM), and 39 (IC50 = 8 μM) against PDGF-Rβ.
G
dx.doi.org/10.1021/jm500373x | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
Figure 6. Graphical representation of the selectivity profile of compound 38 at a concentration of 1 μM against 300 wild-type protein kinases (mean
of duplicate measurements). For clarity of representation, not all PK are shown at the y-axis (for full details see Supporting Information). PKs
showing residual activity less than 50%: PDGF-Rβ (1%), RET (5%), HIPK4 (5%), CLK4 (5%), RIPK2 (9%), DYRK2 (9%), ACV-R2B (11%),
LRRK2 (12%), CLK2 (12%), ACV-R1 (13%), DAPK1 (15%), MELK (15%), RPS6KA6 (16%), ACV-RL1 (16%), DYRK1B (16%), FLT3 (16%),
DYRK3 (20%), VEGF-R2 (21%), DAPK3 (21%), PIM3 (22%), CLK1 (24%), DAPK2 (25%), RPS6KA3 (27%), CSF1-R (28%), TGFB-R2 (28%),
RPS6KA2 (33%), HIPK2 (34%), TRK-C (36%), MST2 (36%), FGR (36%), MST1 (38%), FGF-R2 (39%), ARK5 (41%), VEGF-R3 (41%), KIT
(42%), RPS6KA1 (42%), HIPK1 (42%), BLK (43%), DYRK1A (43%), IRAK4 (43%, SNF1LK2 (43%), TRK-A (44%), MERTK (44%), ITK (46%),
CDK5/p35NCK (49%), SRC (49%), DYRK4 (50%), and TRK-B (50%). For assay details and abbreviations also visit www.proqinase.com.
test compounds and their viability was determined after 48 h
incubation. All test compounds were shown to be cytotoxic
against HL-60 cells. Interestingly, the other cells tested were
significantly less affected by treatment of 2, which is in line with
the notion that HL-60 cells depend on PDGF-R signaling.
Compound 2 was determined to have an IC50 value of 0.026
μM against HL-60 cells1 and exhibited a significantly stronger
antiproliferative effect than 38 that had an IC50 value of 3.2 μM
in this assay. However, even if 38 showed to be most potent
against HL-60 cells within the other cell lines tested in this
project, the reduced cellular efficacy of 38 compared to 2 is in
sharp contrast to the data from isolated PDGF-Rβ assay (Table
4). This may be due to inferior ADME properties such as
limited cellular bioavailability, unspecific protein binding, or a
metabolic inactivation particularly of the key phenolic−OH
moiety of 38. Thus, in order to estimate the cellular uptake of
38, an assay was performed using Caco-2 cells25 (Figure 8).
The data shows that compound 38 is slowly penetrating into
the cells reaching a maximum intracellular concentration of
approximately 40% after 60 min of incubation time. The
cellular uptake is paralleled by a decrease of 38 in the assay
medium. However, although this data provides evidence for a
moderate cellular bioavailability of 38, the total recovery of the
inhibitor (in both portions, medium and intracellular) was
determined to be approximately 70%. As discussed above this
may be due to unfavorable ADME properties of 38 such as
nonspecific protein binding and metabolization.
Effect of Compound 38 on the Signal Transduction in
U87 Cells. In order to analyze the effect of 38 on the PDGF
signal transduction, a Western blot experiment was performed
using U87 cells. Herein, the phosphorylation of Akt and ERK
upon PDGF BB stimulation in the absence and presence of 38
was determined in a concentration-dependent manner (Figure
9). In this experiment 38 was able to decrease significantly the
phosphorylation of AKT at a concentration of 3 μM.
Furthermore, the ERK phosphorylation was less affected as
expected for an inhibitor blocking PDGF-R signaling.26
H
dx.doi.org/10.1021/jm500373x | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
Figure 7. Determination of cytotoxic profiles of compounds 2, 16, 31, 35, and 38 in cell viability assays using HL-60, TK 10, 786-0, M14, and MCF-
7 cell lines.
displacing water molecules from key hydration sites could be
calculated by the WaterMap technology. This approach led to
the design of compounds featuring both displacement and
replacement of key hydration sites. Herein, the gain in ligand
affinity can be explained by the displacement of key hydration
sites of the 3,5-diaryl-pyrazin-2(1H)-one core scaffold. More-
over, the replacement of a structural water molecule in the
active site by the 4′-phenolic OH group of 38 in combination
with the displacement of two further unstable hydration sites by
the vicinal 3′-methoxy moiety of 38 essentially contributes to
the ligand’s affinity. This resulted in a 25-fold increase of the
IC50 value for 38 (PDGF-Rβ: IC50 = 0.02 μM) compared to the
CONCLUSIONS
■
In this article we demonstrate the successful optimization of
3,5-diaryl-pyrazin-2(1H)-ones as potent PDGF-Rβ inhibitors.
SARs for three pyrazin-2(1H)-ones series showing variable 3,5-
aryl substitution patterns have been investigated. The 3,4,5-
trimethoxy system at the 3-aryl position proved to be optimal
and was therefore retained for further optimization. This
moiety is situated in the HRII where the molecular modeling
correlated accurately with the biological data. However,
standard docking/scoring only insufficiently reflected the
impact of the “hydrophobic effect” in the HPI where the 5-
aryl system is located. The thermodynamic contribution of
I
dx.doi.org/10.1021/jm500373x | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
autosampler, diode array detection, and an Agilent Zorbax Eclipse
XDB-C8 column (150 × 4.6 mm, 5 μm particle size). Elution was
achieved with a solvent gradient system of water and acetonitrile, with
0.1% of acetic acid and a flow rate of 1 mL/min. The eluent flow was
splitted to the mass spectrometer.
Mass spectra with nominal resolution were recorded with an
Esquire ∼LC mass spectrometer (Bruker Daltonik, Bremen,
Germany), with electrospray ionization operating in the positive ion
mode, with the following parameters: drying gas nitrogen 8 l/min,
nebulizer 35 psi, dry gas heating 350 °C, HV capillary 4000 V, and HV
end plate offset −500 V. GC/MS was performed on a HP6890 Series
System. EI-mass spectra were recorded on a Varian MAT 311A (70
eV). HRMS spectra were recorded on a MAT-95 (Finnigan).
Where appropriate, column chromatography was performed for
crude precursors with Merck silica gel 60 (0.063−0.200 mm) or Acros
organics silica gel (0.060−0.200 mm; pore diameter ca. 60 nm).
Column chromatography for test compounds was performed using a
LaFlash system (VWR) with Merck silica gel 60 (0.015−0.040 mm) or
RP8 columns. The progress of the reactions was monitored by thin-
layer chromatography (TLC) performed with Merck silica gel 60 F-
245 plates. Where necessary, reactions were carried out in a nitrogen
atmosphere using 4 Å molecular sieves. All reagents and solvents were
obtained from commercial sources and used as received (THF was
used after distillation over K/benzophenone). Reagents were
purchased from Sigma-Aldrich Chemie, Steinheim, Germany;
Figure 8. Measurement of the cellular uptake of 38 using a Caco-2
model. Cells were incubated with 38 at a concentration of 50 μM for
90 min, and the amount of 38 in the medium as well as the
intracellular portion of 38 was determined by HPLC analysis (a
control experiment using propranolol was performed; for further
details see Supporting Information).
Lancaster Synthesis, Muhlheim, Germany; or Acros, Nidderau,
̈
Germany.
HPLC analysis was performed on a Hewlett-Packard HP 1090
Series II using a Thermo Betasil C8 (150 × 4.6 mm 5 μM) column
(mobile phase flow 1.5 mL/min, gradient KH2PO4 buffer pH 2.3/
methanol, UV-detection 230/254 nm). All key compounds submitted
to biological assays were proven by this method to show ≥98% purity.
Synthesis of Compounds. Synthesis of arylglyoxylic acids (I,
Scheme 1) was achieved by SeO2-mediated oxidation of the respective
acetophenone in pyridine.24 The α-ketoamines (II, Scheme 1) were
Figure 9. Effect of 38 on the signaling in U87 cells. The cells were
treated with the indicated concentrations of 38 for 60 min at 37 °C
and stimulated with 10 ng/mL PDGF BB for 10 min. Cells were lysed
and lysates immunoblotted with the indicated antibodies (for details
see Supporting Information). Similar results were obtained in
duplicate experiments.
prepared either by Delep
́
ine reaction from acetophenones27 or by
using aryl acids as starting material.28 The open-chained diketoamides
(V, Scheme 1) were synthesized via two different routes: (a) by CDI-
mediated coupling of arylglyoxylic acids I and aryl α-ketoamines II or
(b) by CDI-mediated coupling7 of arylglyoxylic acids I and indole
derivatives III to yield indole-amides IV. Subsequent DDQ oxidation1
of the indole side-chain in IV yielded indole-diketoamides V. All test
compounds were synthesized in the final step accordingly to the
general microwave mediated method using ammonium acetate as
nitrogen source:1 a microwave vial (5 mL) was equipped with
ammonium acetate (10 equiv) and a solution of open-chained
diketoamide V (1 equiv) in 3 mL acetic acid (total volume). The vial
was sealed and stirred at 160 °C for 4 min in a microwave synthesizer
(CEM Discover). The reaction vessel was cooled to rt when H2O was
added to precipitate the raw pyrazinone, which was filtered off and
purified by preparative HPLC (RP-phase) to afford the test compound
in ≥98% purity.
hit compound 2 (PDGF-Rβ: IC50 = 0.5 μM). A selectivity
profile in a panel of 300 PKs was determined for 38 showing
the compound to most potently block PDGF-Rβ activity.
However, the promising activity of 38 in the kinase assay did
not fully translate into potent efficacy in cellular assays as
analyzed by Western blot. The moderate cellular efficacy of 38
may be due to the compound’s limited ADME properties such
as inferior cellular bioavailability based on poor permeability or
the susceptibility to metabolic degradation particularly of the
phenolic hydroxy moiety. Further hit-to-lead development of
this compound series toward potent PDGF-Rβ inhibitors is
therefore essential.
The experimental procedures for the preparation of compounds 5,
6, 8, 12, 13, 15, and 31 are reported in our recent study.7 The
experimental procedure for the preparation of compounds 2 and 16
are reported in the literature.1 The analytical data for the other
compounds of this article (3b/3c/3, 4a/4, 7b/7c/7, 9b/9c/9, 10b/
10c/10, 11b/11c/11, 14b/14c/14, 17b/17c/17, 18a/18, 19a/19,
20b/20c/20, 21a/21, 22a/22, 23a/23, 24a/24, 25a/25, 26a/26, 27a/
27, 28a/28, 29a/29, 30a/30, 32a/32, 33a/33, 34a/34, 35a/35, 36a/
36, 37a/37, 38a/38, and 39a/39) can be found in the Supporting
Information. X-ray data for compounds 24 (CCDC number: 986158)
and 29a (CCDC number: 986159) is also given in the Supporting
Information.
EXPERIMENTAL SECTION
■
1H (300 MHz) and 13C (75 MHz) NMR were recorded on a Bruker
Avance III 300 spectrometer (Rheinstetten, Germany) at 300 K with a
multinuclear probe head using the manufacturer’s pulse programs. The
data are reported as follows: chemical shifts in ppm from Me4Si
(TMS) as external standard, multiplicity, and coupling constant (Hz).
NMR spectra were obtained, and 1H (300 MHz) and 13C spectra (75
MHz) were referenced either to TMS or to internal DMSO-d5 (1H
NMR δ 2.50) and internal DMSO-d6 (13C NMR δ 39.5) or internal
CHCl3 (1H NMR δ 7.26) and internal CDCl3 (13C NMR δ 77.0). All
coupling constants (J values) are quoted in Hz. The following NMR
abbreviations are used: b (broad), s (singlet), d (doublet), t (triplet),
and m (unresolved multiplet). The labeling scheme of structures to
correlate NMR signals can be found in Supporting Information.
Mass spectra of the compounds were recorded after chromato-
graphic separation. Mixtures were separated with an Agilent 1100
HPLC system (Waldbronn, Germany) consisting of a thermostated
Molecular Modeling. All modeling was performed on a DELL 8
core system. For preparation, visualization and building the 3D
structures Maestro (version 9.3) from Schrodinger (Schrodinger, LLC,
̈
̈
New York, NY, 2013) was used. The illustrations of modeling were
generated by Maestro. For compound docking and screening the
Schrodinger “Glide SP” workflow was used.18 The goal of the Glide
̈
J
dx.doi.org/10.1021/jm500373x | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
methodology is to semiquantitatively rank the ability of candidate
ligands to bind to a specified conformation of the protein receptor.29
Prior to determining binding poses of ligands, energetically minimized
compound conformations were generated, docked into the active site,
and subsequently ranked based on their calculated binding affinity.
The homology model for PDGF-Rβ was performed using the Prime
workflow,16,30 primary sequence data of PDGF-Rβ (http://www.
uniprot.org/uniprot/P09619), and the VEGF-R2 template structures
in the DFG-in- (pdb 2p2h) and DFG-out-conformation (pdb 1t46),
respectively.
Glutamax with 10% FCS, 100 μg/mL streptomycin, and 100 U/mL
penicillin G and incubated in a 5% CO2 humidified atmosphere at 37
°C. For proliferation experiments, cells were seeded in 20 μL pro well
into 384-well Greiner 384 CellStar plates (Greiner Bio-One I. AG,
Kremsmunster, AT). In addition to the test plates, one plate was
̈
prepared for the reference measurement at day zero. All plates were
incubated for 24 h at 37 °C in a humidified atmosphere with 5% CO2.
Compounds that were dissolved in 100% DMSO (v/v) were added to
test plates using the Echo 550 Liquid Handler (Labcyte Inc.,
Sunnyvale, UK). The final DMSO concentration in the assay was
0.5% (v/v). The viability of the cells in the day zero control plates
were determined on the same day without adding any compounds.
The CellTiter-Glo Viability Assay was used to determine the viability
of cells using the standard protocol for this assay (Promega Corp.,
Madison, US). The luminescence signal was measured at the EnSpire
Multimode Plate Reader (PerkinElmer, Waltham, US). Test plates
were incubated for further 48 h, and the cell viability was defined as
just described. Measured raw data were converted into percent of cell
growth by using the high control (0.5% DMSO (v/v) without
compound) and the day zero control. For dose−response studies, 11
different concentrations of compounds were tested in quadruplicates.
The IC50 values were calculated using the 4-parameter logistic fit
option of GraphPad Prism 5.
WaterMap computes water properties (location, occupancy,
enthalpy, entropy, and free energy) through a combination of
molecular dynamics, solvent clustering, and statistical thermodynamic
analysis. First, a 2 ns explicit-solvent molecular dynamics simulation of
the protein with the ligand removed is run in order to sample the
configurations of water molecules in the binding site. The coordinates
of the protein are restrained with a 5.0 kcal/mol/A2 harmonic
potential applied to the initial positions of the heavy atoms, which
ensures convergence of the water sampling around the protein
conformation of interest. Waters from approximately 2000 equally
spaced frames from the molecular dynamics simulation are then
spatially clustered to form localized hydration sites, and the
thermodynamic properties of those sites are computed. The enthalpy
is computed as the average nonbonded molecular mechanics
interaction energies of the waters in the hydration site with the rest
of the system. The entropy is computed by numerically integrating a
local expansion of spatial and orientational correlation functions, as
described in the inhomogeneous solvation theory work by Lazaridis.31
The relevant solvation thermodynamic quantities for the ligand are
computed based on the amount of overlap with the hydration sites.
Biological Evaluation. All inhibitor solutions were prepared
freshly in DMSO prior to each experiment and used immediately.
Determination of IC50 Values of Compounds. Recombinant
Protein Kinases. The inhibitory profile of compounds was determined
using the following 24 protein kinases (GenBankAcc. No. available on
http://www.proqinase.com/pages/science): AKT1, ARK5, Aurora-A,
Aurora-B, B-Raf-VE, CDK2/CycA, CDK4/CycD1, CK2-A1, EGF-R,
EPHB4, ERBB2, FAK, IGF1-R, SRC, VEGF-R2, VEGF-R3, FLT3,
INS-R, MET, PDGF-Rβ, PLK1, SAK, TIE2, and COT. All protein
kinases were expressed using human cDNAs in Sf9 insect cells as
recombinat GST-fusion proteins or His-tagged proteins by means of
the baculovirus expression system. Kinases were purified by affinity
chromatography using either GSH-agarose (Sigma) or Ni-NTH-
agarose (Qiagen). The purity and identity of each kinase was
determined by SDS-PAGE/silver staining and Western blot analysis
using specific antibodies.
ASSOCIATED CONTENT
* Supporting Information
■
S
Details of WaterMap calculation, spectroscopic details, IR data,
purity and X-ray analysis for compounds, selectivity profile of
38, measurement of intracellular uptake, and Western blot
analysis. This material is available free of charge via the Internet
at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
*(C.P.) Tel: 0049 431 880 1137. Fax: 0049 431 880 1352. E-
mail: cpeifer@pharmazie.uni-kiel.de.
■
Author Contributions
∥R.H. and B.P. contributed equally to this work. The
manuscript was written through contributions of all authors.
All authors have given approval to the final version of the
manuscript.
Notes
The authors declare no competing financial interest.
Protein Kinase Assay. A proprietary protein kinase assay
(33PanQinase Activity Assay) was used for measuring the kinase
activity of the 24 protein kinases. All protein kinase assays were
performed in 96-well FlashPlatesTM (PerkinElmer/NEN, Boston,
MA, USA) in 50 μL reaction volumes. Assays for all enzymes were
performed in a solution containing 60 mM HEPES-NaOH, pH 7.5, 3
mM MgCl2, 3 mM MnCl2, 3 μM Na-orthovanadate, 1.2 mM DTT, 50
μg/mL PEG20000, 1 μM [γ-33P]-ATP (approximately 5 × 105 cpm
per well), and recombinant protein kinase (50−400 ng). Depending
upon the kinase being assayed, appropriate substrates were used and
were as follows (substrates shown in parentheses): AKT1 (GSK3/
14−-27), ARK5 (autophosphorylation), Aurora-A, Aurora-B (Tetra-
(LRRWSLG)), B-Raf-V600E (MEK1 KM), CDK2/CycA (histone
H1), CDK4/CycD1 (Rb-CTF), CK2-A1 (casein), EGF-R, EPHB4,
ERBB2, FAK, IGF1-R, SRC, VEGF-R2, VEGF-R3 (poly(Glu,Tyr)
4:1), FLT3, INS-R, MET, PDGF-Rβ (poly(Ala,Glu,Lys,Tyr) 6:2:5:1),
PLK1 (casein), SAK (autophosphorylation), TIE2 (poly(Glu,Tyr)
4:1), and COT (autophosphorylation). The IC50 values were
measured by testing 10 concentrations of compounds in singlicate.
The final DMSO concentration in the assay was 1% (v/v). The data
were fitted using the 4-parameter logistic fit option of GraphPad Prism
5.
ACKNOWLEDGMENTS
■
We thank Martin Schutt and Dr. Ulrich Girreser for excellent
̈
technical and analytical assistance at the Institute of Pharmacy
in Kiel, Germany. We gratefully acknowledge the help of Dr.
Dieter Schollmeyer, University of Mainz, Institute for Organic
Chemistry, Germany, for X-ray analysis of compounds.
Financial support by DFG grant (PE 1605/2-1) is greatly
acknowledged.
ABBREVIATIONS USED
■
ADME, adsorption, distribution, metabolism, excretion; AKT1,
v-akt murine thymoma viral oncogene homologue 1 (PKB);
ARK5, AMPK-related protein kinase 5 (NUAK1); aurora,
aurora kinase; B-RAF, v-raf murine sarcoma viral oncogene
homologue B1; CDI, carbonyldiimidazole; CDK, cyclin-
dependent kinase; COT, mitogen-activated protein kinase
kinase kinase 8 (MAP3K8); DDQ, 2,3-dichloro-5,6-dicyano-
1,4-benzoquinone; DFG-in/out, sequence motif of aspartic
acid-phenylalanine-glycine of a loop in the PK domain; DMSO,
dimethyl sulfoxide; EGF-R, epidermal growth factor receptor;
Cell Culture and Proliferative Assays Using HL-60, TK 10,
786-0, M14, and MCF-7 cells. The cells were grown in RPMI 1640
K
dx.doi.org/10.1021/jm500373x | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
D.; Manley, P. W. Structural biology contributions to the discovery of
drugs to treat chronic myelogenous leukaemia. Acta Crystallogr., Sect.
D: Biol. Crystallogr. 2007, 63, 80−93.
(12) Mol, C. D.; Dougan, D. R.; Schneider, T. R.; Skene, R. J.; Kraus,
M. L.; Scheibe, D. N.; Snell, G. P.; Zou, H.; Sang, B. C.; Wilson, K. P.
Structural basis for the autoinhibition and STI-571 inhibition of c-Kit
tyrosine kinase. J. Biol. Chem. 2004, 279, 31655−63.
(13) Getlik, M.; Grutter, C.; Simard, J. R.; Kluter, S.; Rabiller, M.;
Rode, H. B.; Robubi, A.; Rauh, D. Hybrid compound design to
overcome the gatekeeper T338M mutation in cSrc. J. Med. Chem.
2009, 52, 3915−3926.
(14) Harris, P. A.; Boloor, A.; Cheung, M.; Kumar, R.; Crosby, R. M.;
Davis-Ward, R. G.; Epperly, A. H.; Hinkle, K. W.; Hunter, R. N., III;
Johnson, J. H.; Knick, V. B.; Laudeman, C. P.; Luttrell, D. K.; Mook, R.
A.; Nolte, R. T.; Rudolph, S. K.; Szewczyk, J. R.; Truesdale, A. T.; Veal,
J. M.; Wang, L.; Stafford, J. A. Discovery of 5-[[4-[(2,3-dimethyl-2H-
indazol-6-yl)methylamino]-2-pyrimidinyl]amino]-2-methyl-b enzene-
sulfonamide (Pazopanib), a novel and potent vascular endothelial
growth factor receptor inhibitor. J. Med. Chem. 2008, 51, 4632−40.
(15) Hilberg, F.; Roth, G. J.; Krssak, M.; Kautschitsch, S.;
Sommergruber, W.; Tontsch-Grunt, U.; Garin-Chesa, P.; Bader, G.;
Zoephel, A.; Quant, J.; Heckel, A.; Rettig, W. J. BIBF 1120: triple
angiokinase inhibitor with sustained receptor blockade and good
antitumor efficacy. Cancer Res. 2008, 68, 4774−82.
(16) Meyers, M. J.; Pelc, M.; Kamtekar, S.; Day, J.; Poda, G. I.; Hall,
M. K.; Michener, M. L.; Reitz, B. A.; Mathis, K. J.; Pierce, B. S.; Parikh,
M. D.; Mischke, D. A.; Long, S. A.; Parlow, J. J.; Anderson, D. R.;
Thorarensen, A. Structure-based drug design enables conversion of a
DFG-in binding CSF-1R kinase inhibitor to a DFG-out binding mode.
Bioorg. Med. Chem. Lett. 2010, 20, 1543−7.
EPHB, EPH receptor B; FAK, focal adhesion kinase; FLT3,
fms-related tyrosine kinase 3; HPI, hydrophobic pocket I;
HRII, hydrophobic region II; IGF1-R, insulin-like growth factor
1 receptor; INS-R, insulin receptor; MET, met proto-oncogene
(hepatocyte growth factor receptor); PDGF-R, platelet-derived
growth factor receptor; PK, protein kinase; PLK1, polo-like
kinase 1; RTK, receptor tyrosine kinase; SAK, serine/
threonine-protein kinase (PLK4); smKI, small molecule kinase
inhibitor; SRC, v-src sarcoma (Schmidt−Ruppin A-2) viral
oncogene homologue; TIE2, tunica interna endothelial cell
kinase; VEGF-R, vascular endothelial growth factor receptor
REFERENCES
■
(1) Pinchuk, B.; Johannes, E.; Gul, S.; Schlosser, J.; Schaechtele, C.;
Totzke, F.; Peifer, C. Marine derived hamacanthins as lead for the
development of novel PDGF-Rbeta protein kinase inhibitors. Marine
Drugs 2013, 11, 3209−23.
(2) Zoraghi, R.; Worrall, L.; See, R. H.; Strangman, W.; Popplewell,
W. L.; Gong, H. S.; Samaai, T.; Swayze, R. D.; Kaur, S.; Vuckovic, M.;
Finlay, B. B.; Brunham, R. C.; McMaster, W. R.; Davies-Coleman, M.
T.; Strynadka, N. C.; Andersen, R. J.; Reiner, N. E. Methicillin-resistant
Staphylococcus aureus (MRSA) pyruvate kinase as a target for bis-
indole alkaloids with antibacterial activities. J. Biol. Chem. 2011, 286,
44716−44725.
(3) Caldwell, J. J.; Veillard, N.; Collins, I. Design and synthesis of
2(1H)-pyrazinones as inhibitors of protein kinases. Tetrahedron 2012,
68, 9713−9728.
(4) Ibrahim, N.; Yu, Y.; Walsh, W. R.; Yang, J. L. Molecular targeted
therapies for cancer: sorafenib mono-therapy and its combination with
other therapies. Oncology Rep. 2012, 27, 1303−11.
(17) (a) Jacobson, M. P.; Kaminski, G. A.; Friesner, R. A.; Rapp, C. S.
Force field validation using protein side chain prediction. J. Phys.
Chem. B 2002, 106, 11673−11680. (b) Jacobson, M. P.; Pincus, D. L.;
Rapp, C. S.; Day, T. J. F.; Honig, B.; Shaw, D. E.; Friesner, R. A. A
hierarchical approach to all-atom protein loop prediction. Proteins:
Struct., Funct., Bioinf. 2004, 55, 351−367.
(5) Hicken, E. J.; Marmsater, F. P.; Munson, M. C.; Schlachter, S. T.;
Robinson, J. E.; Allen, S.; Burgess, L. E.; DeLisle, R. K.; Rizzi, J. P.;
Topalov, G. T.; Zhao, Q.; Hicks, J. M.; Kallan, N. C.; Tarlton, E.;
Allen, A.; Callejo, M.; Cox, A.; Rana, S.; Klopfenstein, N.; Woessner,
R.; Lyssikatos, J. P. Discovery of a novel class of imidazo[1,2-
a]pyridines with potent PDGFR activity and oral bioavailability. Med.
Chem. Lett. 2014, 5, 78−83.
(18) Friesner, R. A.; Banks, J. L.; Murphy, R. B.; Halgren, T. A.;
Klicic, J. J.; Mainz, D. T.; Repasky, M. P.; Knoll, E. H.; Shelley, M.;
Perry, J. K.; Shaw, D. E.; Francis, P.; Shenkin, P. S. Glide: A new
approach for rapid, accurate docking and scoring. 1. Method and
assessment of docking accuracy. J. Med. Chem. 2004, 47, 1739−1749.
(19) (a) Abel, R.; Young, T.; Farid, R.; Berne, B. J.; Friesner, R. A.
Role of the active-site solvent in the thermodynamics of factor Xa
ligand binding. J. Am. Chem. Soc. 2008, 130, 2817−2831. (b) Young,
T.; Abel, R.; Kim, B.; Berne, B. J.; Friesner, R. A. Motifs for molecular
recognition exploiting hydrophobic enclosure in protein-ligand
binding. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 808−13. (c) Haider,
K.; Huggins, D. J. Combining solvent thermodynamic profiles with
functionality maps of the Hsp90 binding site to predict the
displacement of water molecules. J. Chem. Inf. Modeling 2013, 53,
2571−86. (d) Snyder, P. W.; et al. Mechanism of the hydrophobic
effect in the biomolecular recognition of arylsulfonamides by carbonic
anhydrase. Proc. Natl. Acad. Sci. U.S.A. 2011, 108, 17889−17894.
(e) Breiten, B.; et al. Water networks contribute to enthalpy/entropy
compensation in protein-ligand binding. J. Am. Chem. Soc. 2013, 135,
15579−15584.
(6) (a) Fabbro, D.; Cowan-Jacob, S. W.; Mobitz, H.; Martiny-Baron,
G. Targeting cancer with small-molecular-weight kinase inhibitors.
Methods Mol. Biol. 2012, 795, 1−34. (b) Janne, P. A.; Gray, N.; Jeff
̈
Settleman, J. Chemical structures and known kinase targets for
clinically approved kinase inhibitors. Nat. Rev. Drug Discovery 2009, 8,
709−723. (c) http://www.nlm.nih.gov/cgi.
(7) Peifer, C.; Selig, R.; Kinkel, K.; Ott, D.; Totzke, F.; Schachtele,
C.; Heidenreich, R.; Rocken, M.; Schollmeyer, D.; Laufer, S. Design,
synthesis, and biological evaluation of novel 3-aryl-4-(1H-indole-3yl)-
1,5-dihydro-2H-pyrrole-2-ones as vascular endothelial growth factor
receptor (VEGF-R) inhibitors. J. Med. Chem. 2008, 51, 3814−24.
(8) Johannes, E.; Horbert, R.; Schlosser, J.; Schmidt, D.; Peifer, C.
Effective synthesis of 3,5-diaryl-(1H)-pyrazin-2-ones via microwave
mediated ring closure. Tetrahedron Lett. 2013, 54, 4067−4072.
(9) (a) Dar, A. C.; Shokat, K. M. The evolution of protein kinase
inhibitors from antagonists to agonists of cellular signaling. Annu. Rev.
Biochem. 2011, 80, 769−795. (b) Iwata, H.; Oki, H.; Okada, K.;
Takagi, T.; Tawada, M.; Miyazaki, Y.; Imamura, S.; Hori, A.; Lawson, J.
D.; Hixon, M. S.; Kimura, H.; Miki, H. A back-to-front fragment-based
drug design search strategy targeting the DFG-out pocket of protein
tyrosine kinases. ACS Med. Chem. Lett. 2012, 3, 342−346.
(20) (a) Robinson, D. D.; Sherman, W.; Farid, R. Understanding
kinase selectivity through energetic analysis of binding site waters.
ChemMedChem 2010, 5, 618−27. (b) Myrianthopoulos, V.; Kritsanida,
M.; Gaboriaud-Kolar, N.; Magiatis, P.; Ferandin, Y.; Durieu, E.;
Lozach, O.; Cappel, D.; Soundararajan, M.; Filippakopoulos, P.;
Sherman, W.; Knapp, S.; Meijer, L.; Mikros, E.; Skaltsounis, A. L.
Novel inverse binding mode of indirubin derivatives yields improved
selectivity for DYRK kinases. ACS Med. Chem. Lett. 2013, 4, 22−26.
(21) Abad-Zapatero, C. Ligand efficiency indices for effective drug
discovery. Expert Opin. Drug Discovery 2007, 2, 469−88.
(10) Hodous, B. L.; Geuns-Meyer, S. D.; Hughes, P. E.; Albrecht, B.
K.; Bellon, S.; Bready, J.; Caenepeel, S.; Cee, V. J.; Chaffee, S. C.;
Coxon, A.; Emery, M.; Fretland, J.; Gallant, P.; Gu, Y.; Hoffman, D.;
Johnson, R. E.; Kendall, R.; Kim, J. L.; Long, A. M.; Morrison, M.;
Olivieri, P. R.; Patel, V. F.; Polverino, A.; Rose, P.; Tempest, P.; Wang,
L.; Whittington, D. A.; Zhao, H. L. Evolution of a highly selective and
potent 2-(pyridin-2-yl)-1,3,5-triazine tie-2 kinase inhibitor. J. Med.
Chem. 2007, 50, 611−626.
(22) Furet, P.; Bold, G.; Meyer, T.; Roesel, J.; Guagnano, V.
Aromatic interactions with phenylalanine 691 and cysteine 828: a
concept for FMS-like tyrosine kinase-3 inhibition. Application to the
(11) Cowan-Jacob, S. W.; Fendrich, G.; Floersheimer, A.; Furet, P.;
Liebetanz, J.; Rummel, G.; Rheinberger, P.; Centeleghe, M.; Fabbro,
L
dx.doi.org/10.1021/jm500373x | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
discovery of a new class of potential antileukemia agents. J. Med. Chem.
2006, 49, 4451−4454.
(23) Reiterer, G.; Yen, A. Platelet-derived growth factor receptor
regulates myeloid and monocytic differentiation of HL-60 cells. Cancer
Res. 2007, 67, 7765−72.
(24) Riley, H. L. Oxidation Activity of Selenium Dioxide. Nature
1947, 159, 571−572.
(25) (a) Schwering, U. Dissertation, University of Kiel; http://
macau.uni-kiel.de/receive/dissertation_diss_00005345. (b) Fogh, J.;
Fogh, J. M.; Orfeo, T. One hundred and twenty-seven cultured human
tumor cell lines producing tumors in nude mice. J. Natl. Cancer Inst.
1977, 59, 221−226.
(26) Servidei, T.; Riccardi, A.; Sanguinetti, M.; Dominici, C.;
Riccardi, R. Increased sensitivity to the platelet-derived growth factor
(PDGF) receptor inhibitor STI571 in chemoresistant glioma cells is
associated with enhanced PDGF-BB-mediated signaling and STI571-
induced Akt inactivation. J. Cell. Physiol. 2006, 208, 220−228.
(27) Delepine, M. Sur l′hexamethylene-amines; seld dammonium;
́ ́
action des acides; production d’amines primaires. C. R. Hebd. Seances
Acad. Sci. 1895, 120, 501−502.
(28) Suzuki, M.; Iwasaki, T.; Miyoshi, M.; Okumura, K.; Matsumot,
K. Synthesis of amino-acids and related compounds. New convenient
syntheses of alpha-C-acylamino acids and alpha-amino ketones. J. Org.
Chem. 1973, 38, 3571−3575.
(29) Halgren, T. A.; Murphy, R. B.; Friesner, R. A.; Beard, H. S.;
Frye, L. L.; Pollard, W. T.; Banks, J. L. Glide: A new approach for
rapid, accurate docking and scoring. 2. Enrichment factors in database
screening. J. Med. Chem. 2004, 47, 1750−1759.
(30) Jacobson, M. P.; Friesner, R. A.; Xiang, Z. X.; Honig, B. On the
role of the crystal environment in determining protein side-chain
conformations. J. Mol. Biol. 2002, 320, 597−608.
(31) (a) Lazaridis, T. Inhomogeneous fluid approach to solvation
thermodynamics. 1. Theory. J. Phys. Chem. B 1998, 102, 3531−3541.
(b) Lazaridis, T. Inhomogeneous fluid approach to solvation
thermodynamics. 2. Applications to simple fluids. J. Phys. Chem. B
1998, 102, 3542−3550.
M
dx.doi.org/10.1021/jm500373x | J. Med. Chem. XXXX, XXX, XXX−XXX