Protein-ligand crystal structures can guide the design of selective inhibitors of the FGFR tyrosine kinase

Article abstract of DOI:10.1021/jm3004043

The design of compounds that selectively inhibit a single kinase is a significant challenge, particularly for compounds that bind to the ATP site. We describe here how protein-ligand crystal structure information was able both to rationalize observed selectivity and to guide the design of more selective compounds. Inhibition data from enzyme and cellular screens and the crystal structures of a range of ligands tested during the process of identifying selective inhibitors of FGFR provide a step-by-step illustration of the process. Steric effects were exploited by increasing the size of ligands in specific regions in such a way as to be tolerated in the primary target and not in other related kinases. Kinases are an excellent target class to exploit such approaches because of the conserved fold and small side chain mobility of the active form.

Full text of DOI:10.1021/jm3004043

Article
pubs.acs.org/jmc
Protein−Ligand Crystal Structures Can Guide the Design of Selective
Inhibitors of the FGFR Tyrosine Kinase
Richard A. Norman, Anne-Kathrin Schott, David M. Andrews, Jason Breed, Kevin M. Foote,
Andrew P. Garner, Derek Ogg, Jonathon P. Orme, Jennifer H. Pink, Karen Roberts, David A. Rudge,
Andrew P. Thomas, and Andrew G. Leach*
AstraZeneca Pharmaceuticals, Mereside, Alderley Park, Macclesfield, SK10 4TG, U.K.
S
* Supporting Information
ABSTRACT: The design of compounds that selectively inhibit a single kinase
is a significant challenge, particularly for compounds that bind to the ATP site.
We describe here how protein−ligand crystal structure information was able
both to rationalize observed selectivity and to guide the design of more
selective compounds. Inhibition data from enzyme and cellular screens and the
crystal structures of a range of ligands tested during the process of identifying
selective inhibitors of FGFR provide a step-by-step illustration of the process.
Steric effects were exploited by increasing the size of ligands in specific regions
in such a way as to be tolerated in the primary target and not in other related
kinases. Kinases are an excellent target class to exploit such approaches because
of the conserved fold and small side chain mobility of the active form.
selectivity between FGFR1, IGF1R, and KDR to guide the
design of more selective FGFR1 inhibitors.
INTRODUCTION
■
It has become apparent over recent years that selective
inhibition of a single kinase is a significant challenge.1−4
Several surveys of kinase inhibition profiles and correspondence
(or lack of it) with overall sequence, or just binding site
sequence, demonstrate the challenges involved in achieving
selectivity.2,5−11 The so-called gatekeeper residue is often
highlighted as a key determinant of selectivity.2−4,7,10−14 Such
surveys do not provide a route map for how to get from a given
compound, or set of compounds, with a given profile to a more
selective profile, should that be desired. A compound’s
selectivity profile depends upon the detail of the interactions
of that compound with each binding site.2,11 Crystal structures
of the protein−ligand complex can provide insight and
understanding.5,7,15 Such structures are particularly powerful
for addressing the issue of selectivity among kinases because of
the remarkable degree of conservation of the protein fold
among the various kinases.7 This similarity means that the
structure of a ligand in complex with its desired target kinase in
combination with the crystal structure of a second kinase or a
sequence alignment with a second kinase can provide insights
into how the ligand might be achieving selectivity and further
might suggest structural changes to the ligand that would
improve the selectivity. Some inhibitors that achieve selectivity
rely upon covalent bonding to their target which can make it
more difficult to obtain safe and orally bioavailable com-
pounds.10 Others rely upon variations in the degree of protein
flexibility, which remains difficult to predict a priori.5,10 In the
work reported here, we describe how protein−ligand crystal
structure information was used to rationalize observed
During the course of routine screening of compounds from
the AstraZeneca compound collection against a panel of kinase
assays (employing the activated enzymes), a small number of
pyrazolylaminopyrimidines that inhibited FGFR1 more po-
tently than any of the other kinases in the panel was identified.
A lead identification campaign was initiated with the aim of
identifying a compound that was sufficiently selective to
provide clear evidence for FGFR driven effects in in vivo
studies. The relative merits of “clean” kinase inhibitors that
selectively inhibit one kinase and broad spectrum inhibitors that
inhibit many are a subject of vigorous discussion focused
particularly around the generally perceived trade-off between
efficacy and tolerability.16 In this case, these starting points had
marginal selectivity compared to KDR and were closely related
to compounds that were potent inhibitors of IGF1R, the target
against which these compounds were originally designed.17
Plots of potency for inhibiting these two kinases versus that for
FGFR1 for the whole series of pyrazolylaminopyrimidines
available at the initiation of this program are shown in Figure 1.
FGFR1 has been linked to a number of tumor types including
bladder cancers and represents an attractive target for medicinal
chemistry efforts.18 With this in mind, a crystal structure for a
complex of FGFR1 with one of the initial compounds that had
prominent FGFR1 inhibition was sought.
Received: November 15, 2011
Published: May 21, 2012
© 2012 American Chemical Society
5003
dx.doi.org/10.1021/jm3004043 | J. Med. Chem. 2012, 55, 5003−5012

Journal of Medicinal Chemistry
Article
Figure 1. Data from an isolated enzyme assay for KDR (left) and cellular assay for IGF1R (right) are plotted against the corresponding values in an
isolated enzyme assay for FGFR1. This is the data set that was available at the initiation of a program to identify and design selective inhibitors of
FGFRs. The pIC50 (−log10 IC50) values for compounds tested against both targets are plotted against one another with pyrazolylaminopyrimidines
highlighted as blue + symbols. The black line is the line of equality, with the upper green line indicating the point at which compounds are 10-fold
selective for KDR or IGF1R and the lower line indicating 10-fold selectivity for FGFR1.
Figure 2. (a) Crystal structure of the complex of 1 with FGFR1. There are two protein−ligand complexes in the asymmetric unit, and the
conformations of 1 in both are shown. (b) Crystal structure of PD173074 (2) in complex with FGFR1 (PDB accession code 2FGI).19 (c) Overlaid
structures showing key residues surrounding the binding site in FGFR1 (in black, PDB accession code 1AGW),20 InsR (in orange, PDB accession
code 1IRK),21 and KDR (in purple, PDB accession code 1Y6B).22
observed for those compounds. A molecule of ethylene glycol,
derived from the cryoprotection buffer, is observed in the
structure of 1 that is absent in the public domain structures
(Figure 2a). Compound 1 was known to be an equipotent
inhibitor of FGFR1 and KDR (compare the fifth and sixth
columns in Table 1). The sequence alignment of FGFR1 with
KDR provided a “map” of the binding site shown in Figure 2c
in which the identity of residues close to the ligand in these two
kinases (and also InsR, see below) is indicated. One particular
RESULTS AND DISCUSSION
■
The crystal structure of the complex of pyrazolylaminopyr-
imidine 1 with the FGFR1 kinase domain is shown in Figure
2a. Details of the construct and methods for obtaining this
structure are provided in the Experimental Section. The protein
structure is very similar to that observed for two Sugen
compounds20 and for a compound originally disclosed by the
Parke-Davis company (PD173074 (2),19,23 also shown in
Figure 2b). Compound 1 binds in the ATP site, as also
5004
dx.doi.org/10.1021/jm3004043 | J. Med. Chem. 2012, 55, 5003−5012

Journal of Medicinal Chemistry
Article
Table 1. Variation of Enzyme and Cellular Potency with Substituents R1, R2, and R3 As Indicated for
a
Pyrazolylaminopyrimidines
a
Values given are average pIC50 values (−log10(IC50), see Experimental Section). Letters in brackets beneath each number show results from a
Student’s t-test; compounds that do not share a letter within a column are distinct at 95% confidence.
5005
dx.doi.org/10.1021/jm3004043 | J. Med. Chem. 2012, 55, 5003−5012

Journal of Medicinal Chemistry
Article
sequence difference suggested an opportunity to achieve
selectivity; Ala640 in FGFR1 is Cys1043 in KDR. The fact
that cysteine has a larger side chain than alanine and that this
residue lies in a constricted indentation at the base of the ATP
pocket, which we have styled the “pit”, suggested that
compounds that rigidly protruded in an appropriate direction
to fill this space might be able to achieve selectivity. The
flexibility of compound 1 is illustrated by the two
conformations observed in the two FGFR1−1 complexes in
the crystal asymmetric unit (Figure 2a). The NHCH2CH2
linker is sufficiently flexible that it can presumably adapt to the
different steric environment in the KDR ATP pocket, and
hence, this particular ligand is not well suited to exploit the
opportunity offered by the “pit” region of the binding site that
the pyridyl ring probes. A number of other pyrazolylaminopyr-
imidines also incorporating 5-methyl substituted pyrazoles had
also been screened, and some are highlighted in the first four
rows of Table 1. In this table, the initially available enzyme
inhibition data for FGFR1 and KDR are supplemented with
enzyme data for IGF1R and cellular data for these three kinases
as well as for the insulin receptor tyrosine kinase (InsR).
Compound 3 provides a reference for compounds that do not
protrude into the “pit” at all, the allyl group being too short to
reach this site, and clearly shows in both enzyme and cellular
assays that, like 1, it is an equipotent inhibitor of FGFR1 and
KDR. The selectivity of two compounds (4 and 5) with more
rigid groups at the same position as the CH2CH2-pyridyl group
in 1 was highly suggestive that the difference between FGFR1
and KDR identified in the crystal structure could indeed be
used to gain selectivity. Compounds 4 and 5 have one fewer
methylene groups between the NH and the aromatic ring
compared to 1 and so have one fewer internal degrees of
freedom. Compound 4, which has a methyl group on the
oxazole, manages to increase potency against FGFR1 in a way
that is not matched by an increase against KDR. When the
methyl group is replaced by the substantially larger tert-butyl
group in 5, enzyme potency against FGFR1 is maintained but
there is a decrease in potency against KDR. This is replicated at
the cellular level. To confirm that compounds possessing an
oxazole side chain were engendering selectivity in the way
hypothesized above, the crystal structure of a related
compound, 6 (Table 1), was sought. This compound was
selected by virtue of being potent, soluble, and available in
sufficient quantity.
Figure 3. Compound 6 in complex with FGFR1. The conformations
of 6 from both molecules in the crystallographic dimer are shown.
which is the second most common gatekeeper residue among
tyrosine kinases and the most common across the entire
kinome.12 These include the IGF1R and InsR tyrosine kinases.
These two are singled out in particular because in addition to
having a Met gatekeeper, they have glycine in place of Ala640
such that the features used to achieve selectivity against KDR
would be ineffective against IGF1R and InsR (see Figure 2c).
Rather than having less space in the “pit”, they have slightly
more. The gatekeeper residue is a well established selectivity
handle.2−4,7,10−14 The crystal structure of 1 in complex with
FGFR1 revealed that the 5-substituent on the pyrazole is
directed toward the gatekeeper, Valine 561 in FGFR1. The
crystal structure of 2 in complex with FGFR1 shows that the
3,5-dimethoxyphenyl group is positioned such that it is in
proximity to the valine side chain. It was our hypothesis that
this might not be tolerated by kinases with a methionine
gatekeeper. In line with this hypothesis, 2 is reported to have
activity of >50 μM against both InsR and MEK, two kinases
with a methionine gatekeeper residue.19,24 Comparison of the
two structures shown in Figure 2a and Figure 2b suggests that if
a suitable linker could be found to link the pyrazole that binds
at the hinge and an aryl group, a similar effect might be
exploited in the pyrazolylaminopyrimidines. It was felt that a
two-atom spacer might achieve this, and consequently
compound 7 was selected for screening. By good fortune this
compound already existed in our compound collection,
obviating the need for synthesis. The full data set for
compound 6 (used to generate a FGFR1 complex crystal
structure) is provided alongside those of 4, 7, and 8 which share
the same “pit” group in Table 1.
Going from the CH2CH2Me group in 7 to the CH2CH2Ph
group in 8 results in a >0.5 log unit increase in isolated enzyme
potency against both FGFR1 and KDR (which has the isosteric
Thr gatekeeper). At the same time cellular potency is
maintained. By contrast there is a 1.2 log unit decrease in
enzyme potency against IGF1R and a >0.5 log unit decrease in
cellular potency against the two Met gatekeeper kinases IGF1R
and InsR. A feature has been introduced that is tolerated in
FGFR1 but not in IGF1R or InsR. If the phenyl group on the
CH2CH2 spacer had been able to wrap around the gatekeeper,
then it could protrude into the hydrophobic pocket occupied
snugly by the phenyl ring in 2. The set of hypotheses, relating
binding interactions with desired selectivity, that compound 8
The structure of 6 in complex with FGFR is shown in Figure
3. In contrast to compound 1 only one conformation of the
pyrimidine 2-substituent is observed in the two molecules of
the crystallographic dimer. This supports the hypothesis that
this is a more conformationally rigid group. The structure also
shows that the methyl on the isoxazole is directed toward the
side chain of Ala640, which can be seen in the background,
suggesting a beneficial hydrophobic interaction that would also
favor this conformation. This interaction is precisely what was
hoped for. In KDR the larger cysteine side chain at this position
would either require the ligand to adopt a different
conformation or require a perturbation of the binding at the
hinge (or both), which would carry an energetic penalty
disfavoring binding. When the methyl group on the isoxazole is
enlarged further as exemplified in compound 5, the effect
becomes more pronounced with a decrease in KDR inhibition
potency.
It was hoped that selectivity could be demonstrated against
the set of kinases that possess a methionine gatekeeper residue,
5006
dx.doi.org/10.1021/jm3004043 | J. Med. Chem. 2012, 55, 5003−5012

Journal of Medicinal Chemistry
Article
was selected to test were further probed by obtaining the
structure of the complex of 8 with FGFR1 (shown in Figure 4).
groups on the aromatic ring rather than just one, one cannot
avoid pushing against Val559.
Compounds 8 and 9 have been tested for inhibition of a
larger set of kinases in a single point assay fashion that yields
percentage inhibition at a particular concentration. Compound
8 was tested at 10 μM and compound 9 at 1 μM. The
percentage inhibition is not a reliable way to quantify selectivity
but is a practical way of obtaining a broader view across the
kinome. The kinases have been classified according to whether
they have larger residues than Ala640, Val 561, and Val559 at
the corresponding positions. The mean and range of percentage
inhibitions observed for each class are summarized in the two
Venn diagrams shown in Figure 5. The three residues targeted
Figure 4. Compound 8 in complex with FGFR1. The conformations
of 8 from both molecules in the crystallographic dimer are shown.
The ethylene glycol present in the earlier structures was no
longer observed, having presumably been displaced by the
phenyl ring. The phenyl group does indeed occupy the
hydrophobic pocket, fitting tightly against the gatekeeper,12
while the methyl isoxazole protrudes into the pit. This indicates
that 8 utilizes the various specificity-enhancing features
described above.
Figure 5. Mean percentage inhibition and the range of percentage
inhibition caused by (a) 10 μM 8 and (b) 1 μM 9 against kinases
classified according to whether they have a residue larger than Ala640
at the corresponding position (red circle), a residue larger than Val559
at the corresponding position (yellow circle), and a residue larger than
Val561 at the corresponding position (blue circle).
With encouraging data as presented for compounds 1 and
3−8 in Table 1 and in advance of obtaining the crystal structure
shown in Figure 4, it was speculated that replacing the phenyl
group in compound 8 with a 3,5-dimethoxyphenyl group, as in
2, might contribute further to selectivity, as that group was
found to fit very snugly into a hydrophobic pocket in the crystal
structure shown in Figure 2b. However, the lipophilicity of
compound 8 was higher than desired (clogP = 4.3) and was
implicated as the cause of a number of problems with the
compound,25 including poor solubility. The bromine sub-
stituent on the pyrimidine ring was not considered an essential
element for FGFR inhibition (see data in Table 1 for
compound 10 which is the chloro equivalent of 4 and for
compounds 11, 12, and 13 which are the des-bromo versions of
4, 6, and 8, respectively), while clearly making a significant
contribution to increased lipophilicity (π value for Br is 0.86),26
and so compound 9, in which this atom was removed and the
phenyl ring substituted for 3,5-dimethoxyphenyl, was targeted
for synthesis and tested. The outcome of this testing is
summarized in Table 1. Compound 9 maintains the same level
of inhibition of FGFR in enzyme and cellular assays as 8 while
having the same or improved selectivity compared to KDR,
IGF1R, and InsR. The FGFR1 potency is particularly striking
because the lipophilicity is reduced (clogP for 9 is 3.4) as a
result of the removal of the bromine atom and addition of the
methoxy groups. The crystal structure of 2 in complex with
FGFR1 suggests that one of the two methoxy groups in that
compound presses closely against the side chain of Val559 in
the hydrophobic pocket. This should entail decreased inhibition
of any kinase having a larger group than this at the
corresponding position in the sequence. This provides a third
point at which steric encumbrance might be used to disfavor
binding to kinases other than FGFR. By having two methoxy
in the approach described here could, in principle, permit
selectivity against all except 10 kinases, of which six are tyrosine
kinases (see Supporting Information for details). Gatekeeper
Val561 appears to be the most influential residue, but Ala640 in
the pit is also seen to make a useful contribution to diminishing
inhibition while Val559 is least influential.
The rationalization of the drivers for selectivity suggested by
the crystal structures and probed with enzymatic data permits a
more general understanding of the selectivity in the series of
pyrazolylaminopyrimidines. With this in mind the compounds
in the initial data set shown in Figure 1 have been reanalyzed. A
selection of interesting compounds from that set and some
tested subsequently that were available in sufficient quantity
were selected for screening in the various enzyme assays. Their
inhibition values are summarized in Table 2. The series of
compounds with Me at R2 (14−22) have varying R1
substituents, some of which ought to bind in the “pit”. It is
notable that only the cyclopropyl substituted isoxazole 21 binds
more tightly to FGFR1 than compound 4. In line with the
hypotheses outlined, 14−22 retain low values of inhibition of
KDR and the large amide substituent in 20 is too big to be
tolerated in either FGFR or KDR. While it is unsurprising that
21 inhibits FGFR more potently than 4 without an increase in
potency against KDR, it is surprising that IGF1R which
notionally has a larger “pit” region is not inhibited more
strongly. This is consistent with a key role for hydrophobic
interactions in the “pit” of FGFR involving the methyl group of
the side chain of Ala640. Comparing 14 and 24 shows that
increasing the pyrazole 5-substituent from Me to Et increases
potency for all three kinases to about the same degree.
Compounds 7 and 26 confirm that compared to the 3-pyridyl
group the methylisoxazole favors FGFR and IGF1R as
5007
dx.doi.org/10.1021/jm3004043 | J. Med. Chem. 2012, 55, 5003−5012

Journal of Medicinal Chemistry
Article
b
Table 2. Variation of Enzyme Potency with Substituents R1, R2, and R3 As Indicated for Pyrazolylaminopyrimidines
a
b
Data obtained in duplicate only. Values given are average pIC50 (−log10(IC50), see Experimental Section) from at least three measurements other
than where indicated. Letters in brackets beneath each number show results from a Student’s t-test; compounds that do not share a letter within a
column are distinct at 95% confidence. The letters designate the same classes as shown in Table 1.
compared to KDR, while 27 suggests that changing n-Pr to i-Pr
enhances potency against IGF1R while reducing it for FGFR
and KDR. The pairs 21+28 and 22+29 present a contrast;
changing methyl to methoxypropyl at R2 in 22+29 reduced
5008
dx.doi.org/10.1021/jm3004043 | J. Med. Chem. 2012, 55, 5003−5012

Journal of Medicinal Chemistry
Article
C18 column, or UPLC was performed on an Waters Acquity binary
solvent manager with Acquity PDA and an SQD mass spectrometer
using a 50 mm × 2.1 mm, 1.7 μm BEH column from Waters, and
purities were measured by UV absorption at 254 nm and are ≥95%
unless otherwise stated. NMR spectra were recorded on a Bruker
Av400 or Bruker DRX400 spectrometer at 400 MHz in DMSO-d6 at
303 K unless otherwise indicated. H NMR spectra are reported as
chemical shifts in parts per million (ppm) relative to an internal
solvent reference.
IGF1R potency as is observed on going from 4 to 12, but in
21+28 no change is observed. Compound 30 contrasts
markedly to 14 and 24 because its FGFR potency is much
higher while KDR remains unchanged and IGF1R declines,
recapitulating the particular sensitivity of Met gatekeeper
kinases to this structural change. Retention of potency and
selectivity for the isoxazole isomer 31 compared to 9 confirms
that activity at FGFR1 is not dependent upon the positioning
of these heteroatoms.
1
N4-(3-(3,5-Dimethoxyphenethyl)-1H-pyrazol-5-yl)-N2-((3-
methylisoxazol-5-yl)methyl)pyrimidine-2,4-diamine, 9. Hydro-
gen chloride, 4 M in dioxane (2−3 drops), was added to 3-(3,5-
dimethoxyphenethyl)-1H-pyrazol-5-amine (286 mg, 1.16 mmol) and
4-chloro-N-((3-methylisoxazol-5-yl)methyl)pyrimidin-2-amine (200
mg, 0.89 mmol) in EtOH (10 mL). The solution was heated at reflux
for 20 h and then allowed to cool to room temperature. The solid
product was collected by filtration, washed with EtOH, and dried
under vacuum to afford the crude product as the HCl salt. The solid
was then triturated with MeOH/H2O (∼1:1 mixture) and basified
with 7 M NH3 in MeOH, causing the material to go into solution
briefly and then precipitate out again. The solid was collected by
filtration, washed with a little MeOH followed by water, and then
dried under vacuum to afford the title compound as a pale cream solid
In conclusion, crystal structures of the complex between a
tyrosine kinase and a new lead series have been used to rapidly
identify and design compounds that achieve levels of selectivity
that are sufficient to allow in vivo testing of specific FGFR-
related biological hypotheses. As highlighted by Bamborough et
al., it is often the case that gross sequence level similarities, even
when localized to the binding pocket, can mask single point
differences that are sufficient to permit selective modulation of
potency favoring one kinase over another.2 How this might be
achieved will depend upon the binding mode of the series being
developed and therefore requires detailed knowledge of how
the protein sequence and different parts of the ligand interact.
Crystal structures of appropriate complexes provide just this
kind of information. In this case, the steric bulk of three amino
acid side chains complemented by a specific hydrophobic
interaction has been exploited to gain selectivity. General rules
might be difficult or even impossible to derive, requiring
detailed consideration of each series and its interactions with
each set of kinases. Having crystal structures and high quality
enzyme and cellular data for the various stages of the process of
tuning selectivity provides a good foundation for believing that
the changes measured are caused by the structural changes
described. For this particular approach to be successful, the
binding mode observed in FGFR1 needs to be energetically
favored over the alternative binding modes that a series such as
the one described here might adopt. It further requires that the
key amino acid side chains are not in positions where they can
move freely and at small energetic cost. Kinases in their active
form present a well conserved fold and most of the side chains
in the ATP binding site are well described by the observed
electron density, suggesting that they are a good system in
which to apply this approach to achieving selectivity. The
achievement reported in this communication is illustrated most
clearly by considering where compounds 8 and 9 would be
found if added to the plots in Figure 1. They represent a good
starting point for optimization toward a clinical candidate that
is able to test the hypothesis that inhibition of FGFRs is of
therapeutic value.
1
(288 mg, 74%). H NMR (400 MHz, DMSO/CD3COOD, 30 °C)
2.16 (3H, s), 2.83 (4H, s), 3.70 (6H, s), 4.55 (2H, s), 6.11 (1H, s),
6.18 (1H, bs), 6.27 (1H, bs), 6.31 (1H, t), 6.40 (2H, d), 7.83 (1H, d);
13C NMR (500 MHz, DMSO/CD3COOD, 100 °C) 10.6, 27.1, 34.9,
36.9, 55.1, 94.5, 96.7, 98.4, 102.0, 106.7, 143.5, 144.1, 146.5, 155.5,
159.1, 160.2, 160.7, 161.4, 171.2. HRMS (ESI): calcd for C22H25N7O3,
436.209 16; found, 436.208 86. Anal. Calcd for C22H25N7O3: C, 60.68;
H, 5.79; N, 22.51. Found: C, 60.32, H, 5.80; N, 22.34.
N4-(3-(3,5-Dimethoxyphenethyl)-1H-pyrazol-5-yl)-N2-((3-methyl-
isoxazol-5-yl)methyl)pyrimidine-2,4-diamine, compound 8 (100 mg),
was purified by preparative HPLC (Waters XBridge Prep C18 OBD
column, 5 μm silica, 19 mm diameter, 100 mm length), using
decreasingly polar mixtures of water (containing 1% NH3) and MeCN
as eluents. Fractions containing the desired compound were
evaporated to afford the title compound (53 mg) as a white solid.
Mp 170.3−170.5 °C; 1H NMR (400 MHz, DMSO, 21 °C) 2.16 (3H,
s), 2.82 (4H, s), 3.71 (6H, s), 4.53 (2H, s), 6.11 (1H, s), 6.21 (2H, bs),
6.31 (1H, t), 6.41 (2H, d), 7.83 (1H, d). Anal. Calcd for C22H25N7O3:
C, 60.68; H, 5.79; N, 22.51. Found: C, 60.60, H, 5.74; N, 22.70.
3-(3,5-Dimethoxyphenylethyl-1H-pyrazol-5-amine, used as starting ma-
terial, was prepared as follows: (a) MeCN (2.29 mL, 43.61 mmol) was
added to a slurry of sodium hydride (1.75 g dispersion in mineral oil,
43.61 mmol) in anhydrous toluene (70 mL) and the mixture stirred at
room temperature for 30 min. Ethyl 3-(3,5-dimethoxyphenyl)-
propanoate (8.66 g, 36.34 mmol) in toluene (60 mL) was added,
and the reaction mixture was heated at reflux for 18 h. After the
mixture was cooled and the reaction quenched with a small amount of
water, the solvent was evaporated under reduced pressure. The residue
was dissolved in 2 M HCl (50 mL). The acidic solution was then
extracted twice with EtOAc. The organic extracts were combined,
washed with water followed by brine, and then dried over magnesium
sulfate. After filtering, the solvent was evaporated under reduced
pressure to yield the crude product as a yellow oil. The oil was purified
by flash chromatography on silica, eluting with DCM. Fractions
containing pure product were combined and evaporated to leave a
cream solid (3.76 g, 44%). To the solid (3.72 g, 15.96 mmol) in EtOH
(55 mL) was added hydrazine hydrate (852 μL, 17.56 mmol). The
mixture was heated at reflux for 24 h before allowing it to cool. After
the mixture was evaporated under reduced pressure, the residue was
dissolved in DCM, washed with water followed by brine, dried with
magnesium sulfate, filtered, and then evaporated under reduced
pressure to afford 3-(3,5-dimethoxyphenylethyl-1H-pyrazol-5-amine as
EXPERIMENTAL SECTION
■
Chemistry. All reactions were performed under inert conditions
(nitrogen) unless otherwise stated. All solvents and reagents were
purchased from commercial sources and used without further
purification. Upon workup, organic solvents were typically dried
prior to concentration with anhydrous MgSO4 or Na2SO4. Flash silica
chromatography was typically performed on an Isco Companion, using
Silicycle silica gel, 230−400 mesh, 40−63 μm cartridges, Grace Resolv
silica cartridges, or Isolute Flash Si or Si II cartridges. Reverse phase
chromatography was performed using a Waters XBridge Prep C18
OBD column, 5 μm silica, 19 mm diameter, 100 mm length), using
decreasingly polar mixtures of either water (containing 1% NH3) and
MeCN or water (containing 0.1% formic acid) and MeCN as eluents.
Analytical LC−MS was performed on a Waters 2790 LC instrument
with a 996 PDA and 2000 amu ZQ single quadrupole mass
spectrometer using a Phenomenex Gemini 50 mm × 2.1 mm, 5 μm
1
a pale yellow solid (3.76 g. 42% over two steps). H NMR (300.132
MHz, DMSO) δ 2.64−2.82 (4H, m), 3.71 (6H, s), 4.07−4.72 (2H,
m), 5.20 (1H, s), 6.31 (1H, t), 6.38 (2H, d). MS: m/z 248 (MH+).
4-Chloro-N-((3-methylisoxazol-5-yl)methyl)pyrimidin-2-amine was pre-
pared as follows: To a solution containing 2-[(3-methylisoxazol-5-
5009
dx.doi.org/10.1021/jm3004043 | J. Med. Chem. 2012, 55, 5003−5012

Journal of Medicinal Chemistry
Article
yl)methylamino]pyrimidin-4-ol (8.8 g, 43.52 mmol) and diisopropy-
lethylamine (9.6 mL) in toluene (40 mL) was added phosphorus
oxychloride (4.8 mL, 51.50 mmol) dropwise. The gummy suspension
was heated at 80 °C for 2 h. The mixture was allowed to cool to room
temperature and then poured portionwise into saturated sodium
bicarbonate solution. The product was extracted with EtOAc (×2),
washed with brine, dried over magnesium sulfate, filtered, and then
evaporated to leave a cream solid. The solid was washed with EtOAc
and DCM (plus a few drops of MeOH), and the suspension was
heated to reflux. After filtration, the product was obtained as a cream
solid (1.6 g). The filtrate was loaded onto a silica column and then
eluted with EtOAc. Fractions containing product were combined and
then evaporated. The residue was triturated with Et2O to afford 4-
chloro-N-((3-methylisoxazol-5-yl)methyl)pyrimidin-2-amine as a pale
yellow solid (3.28 g). Total yield = 4.88 g (50%). 1H NMR (400 MHz
DMSO) 2.19 (s, 3H), 4.56 (d, 2H), 6.15 (s, 1H), 6.77 (d, 1H), 8.22 (t,
1H), 8.29 (d, 1H). MS: m/z 225 (MH+).
model-building, using the 2FO − FC and FO − FC electron density
maps as displayed in COOT,33 followed by refinement in REFMAC30
were carried out to produce the final structures (PDB codes 4F63,
4F64, and 4F65). All structural figures were made using PyMOL.34
Biological Evaluation. In all enzyme and cell assays, compound
was tested at a range of concentrations. The mean data values for each
concentration, along with untreated control wells and 100%
inhibition/competition control wells, were used to derive a plot of
inhibition/competition against concentration. Origin software was
used to interpolate IC50 values by nonlinear regression.
Kinase Inhibition Assays Using Caliper Technology. The
inhibitory activity of compound against the kinases FGFR1, KDR, and
IGF1R was determined with Caliper off-chip incubation mobility shift
assays, using a microfluidic chip to measure the conversion of a
fluorescent labeled peptide to a phosphorylated product. Custom
peptide substrates with a fluorescent tag, sequence specific for each
enzyme, were obtained from Cambridge Research Biochemicals
(Cleveland, U.K.). Enzymes, peptide substrates, and ATP were
incubated in the presence of compound diluted in DMSO (1% v/v
DMSO assay final), in white Greiner 384-well low volume plates, in a
total reaction volume of 12 μL. Enzymes and peptides were added
separately to the compound plates and were incubated at room
temperature, and the kinase reaction was then quenched with the
addition of a stop buffer (comprising 100 mM HEPES (pH 7.5), 5%
DMSO, 88 mM EDTA, 0.22% Caliper coating reagent no. 3, 0.033%
Brij-35 solution). Stopped assay plates were then read using the
Caliper LabChip LC3000.
GST-FGFR1 human recombinant kinase domain (residues 456−
765) with an N-terminal GST-tag (expressed from a baculovirus in
Sf21 insect cells) was obtained from Millipore. The kinase reaction
[170 μM ATP, 5 nM enzyme, 1.5 μM substrate] in phosphorylation
buffer (100 mM HEPES (pH 7.5), 5 mM MgCl2, 1 mM DTT, 1 mM
EDTA, 0.22% Caliper coating reagent no. 3, 0.033% Brij-35 solution)
was quenched after a 40 min incubation.
GST-VEGFR2 (KDR) human recombinant kinase domain
(residues 805−1356) with an N-terminal GST-tag (expressed from a
baculovirus in Sf9 insect cells) was obtained from BPS Bioscience Inc.
The kinase reaction (7 μM ATP, 8 nM enzyme, 1.5 μM substrate) in
phosphorylation buffer (50 mM MOPS (pH 6.5), 20 mM MnCl2, 4
mM DTT, 0.05% CHAPSO) was quenched after a 90 min incubation.
IGF-IR human recombinant kinase domain (residues 959 to end)
containing an N-terminal His6-tag (expressed by baculovirus in Sf21
insect cells) was obtained from Upstate. The kinase reaction [7 μM
(below Km) ATP, 15 nM enzyme, 1.5 μM substrate] in
phosphorylation buffer [50 mM MOPS (pH 6.5), 5 mM MnCl2, 1
mM DTT, 0.004% Triton X-100] was quenched after an 80 min
incubation.
Measurement of IGF-1R Phosphorylation. IGF-1R null murine
fibroblasts R+ cells expressing human IGF-1R were from Xiao Tu
(Kimmel Cancer Center, Thomas Jefferson University, Philadelphia,
PA, U.S.). R+ cells were derived from transgenic mouse IGF1R
knockouts that were then transfected with human insulin-like growth
factor 1 receptor (IGF1R). R+ cells were routinely cultured in DMEM
(Gibco) growth medium supplemented with 2 mM L-glutamine
(Invitrogen), 10% (v/v) heat inactivated fetal calf serum (PAA), and
hygromycin B (50 mg/mL) in a 5% CO2 incubator at 37 °C. Cells
were plated at a density of 5 × 103 cells per well in 96-well black
Packard View plates (PerkinElmer) in DMEM supplemented with 1%
heat inactivated FCS and 1% L-glutamine and incubated at 37 °C.
Following overnight culture cells were acoustically dosed using an
Echo 555 (Labcyte), with compounds serially diluted in 100% DMSO.
Following a 30 min compound treatment cells were stimulated for 20
min with a final concentration of 30 nM IGF-l (Gropep’s IMOOl)
diluted in DMEM without serum. Cells were fixed with 4%
formaldehyde for 20 min and treated with 0.05% Triton for 10 min.
Cells were then blocked with 2% bovine serum albumin (BSA) and 2%
goat serum (DAKO Ltd.) in PBS for 1 h and incubated with rabbit
dual phospho specific anti-phospho IGF-1R/IR antibody (Biosource)
(1:350) for 1 h, followed by incubation with anti-rabbit Alexa Fluor
488 secondary antibody (1:1000) for 1 h. Measurement was done
2-[(3-Methylisoxazol-5-yl)methylamino]pyrimidin-4-ol was prepared as
follows: (3-Methylisoxazol-5-yl)methanamine (9.3 g, 83 mmol) and 2-
methylsulfonylpyrimidin-4-ol (9.8 g, 69 mmol) were heated together
at 160 °C for 4 h. The mixture was allowed to cool to room
temperature and then dissolved in DCM and purified by flash
chromatography on silica, eluting with a gradient of 5−15% MeOH in
DCM. Fractions containing product were combined and evaporated to
afford 2-[(3-methylisoxazol-5-yl)methylamino]pyrimidin-4-ol as a
1
brown gum (8.88 g, 62%). H NMR (400 MHz, DMSO) δ 2.19 (s,
3H), 4.57 (s, 2H), 5.6 (d, 1H), 6.19 (s, 1H), 7.03 (bs, 1H), 7.61 (d,
1H), 11 (bs, 1H). MS: m/z 207 (MH+).
Protein Expression and Purification. Human FGFR1 consisting
of residues 458−765 was engineered to incorporate a TEV-cleavable
N-terminal 6×His tag and mutations C488A and C584S.27 Non-
phosphorylated FGFR1 protein was obtained from 24 h growth of
IPTG-induced E. coli BL21 (DE3) Star cells coexpressing the PTP1B
phosphatase gene (induction at 20 °C using 100 μM IPTG).
Following cell lysis in buffer consisting of 20 mM Tris-HCl, pH 7.8,
10 mM imidazole, 300 mM NaCl, 2 mM TCEP and clarification of the
lysate by centrifugation, the FGFR1 protein was bound to a Ni-NTA
column (QIAGEN), washed and step-eluted in 20 mM Tris-HCl, pH
7.8, 250 mM imidazole, 300 mM NaCl, 2 mM TCEP. The protein was
subjected to TEV cleavage after exchange into 20 mM Tris-HCl, pH
7.8, 300 mM NaCl, 2 mM TCEP, 5 mM EDTA and further purified by
ion-exchange chromatography after exchange into 20 mM Tris-HCl,
pH 7.5, eluting bound protein with a 0−1 M NaCl gradient. The final
purification step consisted of size-exclusion chromatography on a
Superdex 200 column pre-equilibrated and run in 20 mM Tris-HCl,
pH 8.0, 20 mM NaCl, 2 mM TCEP buffer followed by concentration
to 10 mg/mL. The FGFR1 protein purity was assessed by SDS−PAGE
and the predicted molecular mass confirmed by liquid chromatog-
raphy−mass spectrometry. Purified protein was snap frozen in liquid
nitrogen and stored at −80 °C for later use in crystallization.
Crystallization and Data Collection. Crystals of FGFR1 were
grown at 4 °C using the hanging drop vapor diffusion method from a
reservoir solution of 16−20% PEG 8000, 100 mM PCTP, pH 6.25−
7.25, 100−300 mM (NH4)2SO4, 25% ethylene glycol. Crystals appear
after 4 days and need a further 7 days to reach the final dimensions of
approximately 100 μm × 20 μm × 20 μm. Crystals were then
subjected to cross-linking with glutaraldehyde28 and transferred into a
soaking solution containing 98% (v/v) reservoir solution and 2% (v/v)
DMSO, to which compounds 1, 6, and 8 were added independently of
each other to a final concentration of 2 mM. Soaks were carried out at
4 °C for 6 h, after which the crystals were flash frozen in liquid
nitrogen. Diffraction data were collected in-house on a Rigaku FRE
rotating anode generator equipped with a Saturn 944 CCD detector or
at the ESRF on beamline ID23-1, processed using MOSFLM29,30 and
scaled and merged using SCALA as implemented in the CCP4
software package.31 The FGFR1-compound crystals belong to space
group C2 (Table S1 in Supporting Information) and contain two
complexes per asymmetric unit. The structures were solved by
molecular replacement using AMORE using the structure of FGFR1
(PDB code 1FGK) as the search model.32 A number of iterations of
5010
dx.doi.org/10.1021/jm3004043 | J. Med. Chem. 2012, 55, 5003−5012

Journal of Medicinal Chemistry
Article
using an Acumen Explorer HTS reader (TTP Labtech Ltd.) at
excitation wavelength of 488 nm and emission wavelength of 530 nm.
Measurement of InsR Phosphorylation. CHOT cells over-
expressing human insulin receptor were obtained from Dr. J. Tavare
(Department of Biochemistry, University of Bristol, Bristol, U.K.).
CHOT cells were routinely cultured in Ham F12 (Invitrogen) growth
medium supplemented with 2 mM L-glutamine (Invitrogen), 10% (v/
v) heat inactivated fetal calf serum (PAA Laboratories), and
hygromycin B (50 μg/mL) in a 5% CO2 incubator at 37 °C. Cells
were plated at a density of 5 × 103 cells per well in 96-well black
Packard View plates (PerkinElmer) in DMEM supplemented with 1%
heat inactivated FCS and 1% L-glutamine and incubated at 37 °C.
Following overnight culture, cells were acoustically dosed using an
Echo 555 (Labcyte), with compounds serially diluted in 100% DMSO.
Following a 60 min compound treatment cells were stimulated for 20
min with a final concentration of 30 nM insulin ligand (Sigma), diluted
in Ham F12 without serum. Cells were fixed with 4% formaldehyde for
20 min and treated with 0.05% Triton for 10 min. Cells were then
blocked with 2% bovine serum albumin (BSA) and 2% goat serum
(DAKO Ltd.) in PBS for 30 min and incubated with rabbit anti-
phospho InsR antibody (AZ proprietary) overnight, followed by
incubation with anti-rabbit Alexa Fluor 488 secondary antibody for 1
h. Measurement was done using an Acumen Explorer HTS reader
(TTP Labtech Ltd.) at excitation wavelength of 488 nm and emission
wavelength of 530 nm.
9, which was studied in duplicate. Individual values of IC50 are
provided in the Supporting Information.
Percentage inhibition data were obtained at the concentrations
mentioned in the text from the National Centre for Protein Kinase
Profiling at the University of Dundee, U.K.
ASSOCIATED CONTENT
* Supporting Information
■
S
Crystallographic parameters, individual IC50 and pIC50 values,
and sequence overlays and percentage inhibition for other
kinases. This material is available free of charge via the Internet
at http://pubs.acs.org.
Accession Codes
The crystal structures of compounds 1, 6, and 8 in complex
with FGFR have been deposited with the RCSB Protein Data
Bank under accession codes 4F63, 4F64, and 4F65,
respectively.
AUTHOR INFORMATION
Corresponding Author
*Phone: +44 1625 231853. E-mail: andrew.leach@astrazeneca.
com.
■
Notes
Measurement of FGFR1 Phosphorylation. Cos-1 cells were
routinely cultured in DMEM growth medium supplemented with 2
mM L-glutamine and 3% fetal calf serum (FCS). For transfection,
Lipofectamine 2000 (Invitrogen) was mixed with OptiMEM
(Invitrogen) and incubated at room temperature for 5 min. DNA
encoding 3′FLAG FGFR1 or empty vector (pcDNA3.2) was diluted
with OptiMEM, and equal volumes of DNA and Lipofectamine 2000
were combined (DNA/lipid = 1:1.2 ratio) and incubated at room
temperature for 20 min. Harvested cells were diluted with 1% FCS/
DMEM. The complexed transfection solution was added to the cell
solution, and the cells were seeded at 1.2 × 104 cells per well in 96-well
plates (Costar) and incubated at 37 °C in a 5% CO2 incubator
overnight. The following day cells were acoustically dosed using an
Echo 555 (Labcyte), with compounds serially diluted in 100% DMSO.
Following a 1 h incubation, medium was removed and cells were fixed
with 100% methanol and incubated for 20 min and then treated with
0.1% Triton for 20 min. Cells were then incubated with monoclonal
anti-phospho-FGFR antibody (Cell Signaling Technology) (1:1000)
for 1 h followed by incubation with anti-mouse Alexa Fluor 594
secondary antibody (1:500) and Hoechst (1:1000) for 1 h.
Measurement was done using an Arrayscan (Cellomics).
Measurement of KDR Phosphorylation. HUVECs were
obtained from Promocell and cultured in MCDB131 (Gibco) growth
medium supplemented with 2 mM glutamine, 10% FCS, and 1%
penicillin−streptomycin (Gibco) in a 5% CO2 incubator at 37 °C.
Cells were plated at a density of 3.5 × 104 cells per well in 24-well
plates in MCDB131medium supplemented with 2 mM glutamine, 1%
FCS, and 1% penicillin−streptomycin. After overnight incubation, the
medium was replaced with 500 μL of serum free medium. Compounds
were added 150 min later (diluted in DMSO to a final concentration
of 10 mM) and incubated for 90 min before stimulation with VEGF
(25 ng/well) for 5 min. Cells were lysed with RIPA buffer (60 mM
Tris, pH 7.4, 150 mM NaCl, 1 mM EDTA, 10× RIPA detergent (10%
NP40 + 2.5% deoxycholate), and phosphatase inhibitor cocktails 1
(Sigma P2850) and 2 (Sigma P5726) and protease inhibitor (Sigma
P8340)). Lysates were then developed according to the human
phospho-VEGF R2 (KDR) ELISA protocol (R&D Systems) until the
plate development stage where SuperSignal (Pierce) was used.
Luminescence measurement was carried out using a Tecan plate
reader.
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The authors thank Jason Kettle, Julie Tucker, and Ray Gairns
for assistance in the preparation of this manuscript and many
co-workers for contributions to the FGF project.
ABBREVIATIONS USED
■
FGFR, fibroblast growth factor receptor (usually referring here
to its tyrosine kinase domain/activity) unless indicated as
FGFR1. FGFR is used to refer collectively to the four isoforms
FGFR1−4; KDR, kinase insert domain receptor (VEGFR2);
IGF1R, insulin-like growth factor I receptor; InsR, insulin
receptor tyrosine kinase; ATP, adenosine triphosphate; IPTG,
isopropyl β-D-thiogalactoside; TCEP, tris(carboxyethyl)-
phosphine; EDTA, ethylenediaminetetraacetic acid; SDS−
PAGE, sodium dodecyl sulfate polyacrylamide gel electro-
phoresis; PCTP, propionic acid, cacodylate, bis-Tris propane;
PEG, polyethylene glycol; HEPES, 4-(2-hydroxyethyl)-1-
piperazineethanesulfonic acid; MOPS, 3-(N-morpholino)-
propanesulfonic acid; DTT, dithiothreitol; DMEM, Dulbecco’s
modified Eagle's medium; PBS, phosphate buffered saline
REFERENCES
■
(1) Aronov, A. M.; Murcko, M. A. Toward a Pharmacophore for
Kinase Frequent Hitters. J. Med. Chem. 2004, 47, 5616−5619.
(2) Bamborough, P.; Drewry, D.; Harper, G.; Smith, G. K.;
Schneider, K. Assessment of Chemical Coverage of Kinome Space
and Its Implications for Kinase Drug Discovery. J. Med. Chem. 2008,
51, 7898−7914.
(3) Fabian, M. A.; Biggs, W. H.,, 3rd; Treiber, D. K.; Atteridge, C. E.;
Azimioara, M. D.; Benedetti, M. G.; Carter, T. A.; Ciceri, P.; Edeen, P.
T.; Floyd, M.; Ford, J. M.; Galvin, M.; Gerlach, J. L.; Grotzfeld, R. M.;
Herrgard, S.; Insko, D. E.; Insko, M. A.; Lai, A. G.; Lelias, J.; Mehta, S.
A.; Milanov, Z. V.; Velasco, A. M.; Wodicka, L. M.; Patel, H. K.;
Zarrinkar, P. P.; Lockhart, D. J. A small molecule-kinase interaction
map for clinical kinase inhibitors. Nat. Biotechnol. 2005, 23, 329−336.
(4) Posy, S. L.; Hermsmeier, M. A.; Vaccaro, W.; Ott, K.; Todderud,
G.; Lippy, J. S.; Trainor, G. L.; Loughney, D. A.; Johnson, S. R. Trends
in Kinase Selectivity: Insights for Target Class-Focused Library
Screening. J. Med. Chem. 2011, 54, 54−66.
All IC50 values were transformed to pIC50 values, and an average of
at least three values for enzyme data and at least two values for cellular
data is reported in the tables other than for cellular data for KDR
where only one measurement was made on all compounds other than
5011
dx.doi.org/10.1021/jm3004043 | J. Med. Chem. 2012, 55, 5003−5012

Journal of Medicinal Chemistry
Article
(5) Xing, L.; Shieh, H. S.; Selness, S. R.; Devraj, R. V.; Walker, J. K.;
Devadas, B.; Hope, H. R.; Compton, R. P.; Schindler, J. F.; Hirsch, J.
L.; Benson, A. G.; Kurumbail, R. G.; Stegeman, R. A.; Williams, J. M.;
Broadus, R. M.; Walden, Z.; Monahan, J. B. Structural Bioinformatics-
Based Prediction of Exceptional Selectivity of p38 MAP Kinase
Inhibitor PH-797804. Biochemistry 2009, 48, 6402−6411.
(6) Vieth, M.; Higgs, R. E.; Robertson, D. H.; Shapiro, M.; Gragg, E.
A.; Hemmerle, H. KinomicsStructural Biology and Chemogenomics
of Kinase Inhibitors and Targets. Biochim. Biophys. Acta 2004, 1697,
243−257.
(7) Liao, J. J. Molecular Recognition of Protein Kinase Binding
Pockets for Design of Potent and Selective Kinase Inhibitors. J. Med.
Chem. 2007, 50, 409−424.
(8) Verkhivker, G. M. Exploring Sequence-Structure Relationships in
the Tyrosine Kinome Space: Functional Classification of the Binding
Specificity Mechanisms for Cancer Therapeutics. Bioinformatics 2007,
23, 1919−1926.
(9) Verkhivker, G. M. Computational Proteomics of Biomolecular
Interactions in the Sequence and Structure Space of the Tyrosine
Kinome: Deciphering the Molecular Basis of the Kinase Inhibitors
Selectivity. Proteins: Struct., Funct., Bioinf. 2007, 66, 912−929.
(10) McInnes, C.; Fischer, P. M. Strategies for the Design of Potent
and Selective Kinase Inhibitors. Curr. Pharm. Des. 2005, 11, 1845−
1863.
(23) Dimitroff, C. J.; Klohs, W.; Sharma, A.; Pera, P.; Driscoll, D.;
Veith, J.; Steinkampf, R.; Schroeder, M.; Klutchko, S.; Sumlin, A.;
Henderson, B.; Dougherty, T. J.; Bernacki, R. J. Anti-Angiogenic
Activity of Selected Receptor Tyrosine Kinase Inhibitors, PD166285
and PD173074: Implications for Combination Treatment with
Photodynamic Therapy. Invest. New Drugs 1999, 17, 121−135.
(24) Thompson, A. M.; Delaney, A. M.; Hamby, J. M.; Schroeder, M.
C.; Spoon, T. A.; Crean, S. M.; Showalter, H. D. H.; Denny, W. A.
Synthesis and Structure−Activity Relationships of Soluble 7-
Substituted 3-(3,5-Dimethoxyphenyl)-1,6-naphthyridin-2-amines and
Related Ureas as Dual Inhibitors of the Fibroblast Growth Factor
Receptor-1 and Vascular Endothelial Growth Factor Receptor-2
Tyrosine Kinases. J. Med. Chem. 2005, 48, 4628−4653.
(25) clogP, version 4.71; Daylight Chemical Information Systems:
Santa Fe, NM.
(26) Hansch, C.; Leo, A.; Unger, S. H.; Kim, K. H.; Nikaitani, D.;
Lien, E. J. Aromatic Substituent Constants for Structure−Activity
Correlations. J. Med. Chem. 1973, 16, 1207−1216.
(27) Mohammadi, M.; Schlessinger, J.; Hubbard, S. R. Structure of
the FGF Receptor Tyrosine Kinase Domain Reveals a Novel
Autoinhibitory Mechanism. Cell 1996, 86, 577−587.
(28) Lusty, C. J. A Gentle Vapor-Diffusion Technique for
Crosslinking of Protein Crystals for Cryocrystallography. J. Appl.
Crystallogr. 1999, 32, 106−112.
(29) Battye, T. G. G.; Kontogiannis, L.; Johnson, O.; Powell, H. R.;
Leslie, A. G. W. iMOSFLM: A New Graphical Interface for
Diffraction-Image Processing with MOSFLM. Acta Crystallogr., Sect.
D: Biol. Crystallogr. 2011, 67, 271−281.
(11) Caffrey, D. R.; Lunney, E. A.; Moshinsky, D. J. Prediction of
Specificity-Determining Residues for Small-Molecule Kinase Inhib-
itors. BMC Bioinf. 2008, 9, 491.
(12) Zuccotto, F.; Ardini, E.; Casale, E.; Angiolini, M. Through the
“Gatekeeper Door”: Exploiting the Active Kinase Conformation. J.
Med. Chem. 2010, 53, 2681−2694.
(13) Weiss, M. M.; Harmange, J.; Polverino, A. J.; Bauer, D.; Berry,
L.; Berry, V.; Borg, G.; Bready, J.; Chen, D.; Choquette, D.; Coxon, A.;
DeMelfi, T.; Doerr, N.; Estrada, J.; Flynn, J.; Graceffa, R. F.; Harriman,
S. P.; Kaufman, S.; La, D. S.; Long, A.; Neervannan, S.; Patel, V. F.;
Potashman, M.; Regal, K.; Roveto, P. M.; Schrag, M. L.; Starnes, C.;
Tasker, A.; Teffera, Y.; Whittington, D. A.; Zanon, R. Evaluation of a
Series of Naphthamides as Potent, Orally Active Vascular Endothelial
Growth Factor Receptor-2 Tyrosine Kinase Inhibitors. J. Med. Chem.
2008, 51, 1668−1680.
(14) Vulpetti, A.; Crivori, P.; Cameron, A.; Bertrand, J.; Brasca, M.
G.; D’Alessio, R.; Pevarello, P. Structure-Based Approaches to Improve
Selectivity: CDK2-GSK3β Binding Site Analysis. J. Chem. Inf. Model.
2005, 45, 1282−1290.
(30) Murshudov, G. N.; Vagin, A. A.; Dodson, E. J. Refinement of
Macromolecular Structures by the Maximum-Likelihood Method. Acta
Crystallogr., Sect. D: Biol. Crystallogr. 1997, 53, 240−255.
(31) Bailey, S. The CCP4 Suite: Programs for Protein Crystallog-
raphy. Acta Crystallogr., Sect. D: Biol. Crystallogr. 1994, 50, 760−763.
(32) Navaza, J. AMoRe: An Automated Package for Molecular
Replacement. Acta Crystallogr., Sect. A: Found. Crystallogr. 1994, 50,
157−163.
(33) Emsley, P.; Cowtan, K. Coot: Model-Building Tools for
Molecular Graphics. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2004,
60, 2126−2132.
(34) The PyMOL Molecular Graphics System, version 1.2r3pre;
Schrodinger, LLC: New York.
̈
(15) Noble, M. E. M.; Endicott, J. A.; Johnson, L. N. Protein Kinase
Inhibitors: Insights into Drug Design from Structure. Science
(Washington, DC, U. S.) 2004, 303, 1800−1805.
(16) Morphy, R. Selectively Nonselective Kinase Inhibition: Striking
the Right Balance. J. Med. Chem. 2010, 53, 1413−1437.
(17) Barlaam, B.; Pape, A.; Thomas, A. PCT Patent 2003048133,
2003.
(18) Turner, N.; Grose, R. Fibroblast growth factor signalling: from
development to cancer. Nat. Rev. Cancer 2010, 10, 116−129.
(19) Mohammadi, M.; Froum, S.; Hamby, J. M.; Schroeder, M. C.;
Panek, R. L.; Lu, G. H.; Eliseenkova, A. V.; Green, D.; Schlessinger, J.;
Hubbard, S. R. Crystal Structure of an Angiogenesis Inhibitor Bound
to the FGF Receptor Tyrosine Kinase Domain. EMBO J. 1998, 17,
5896−5904.
(20) Mohammadi, M.; McMahon, G.; Sun, L.; Tang, C.; Hirth, P.;
Yeh, B. K.; Hubbard, S. R.; Schlessinger, J. Structure of the Tyrosine
Kinase Domain of Fibroblast Growth Factor Receptor in Complex
with Inhibitors. Science (Washington, D.C.) 1997, 276, 955−960.
(21) Hubbard, S. R.; Wei, L.; Ellis, L.; Hendrickson, W. A. Crystal
structure of the tyrosine kinase domain of the human insulin receptor.
Nature (London) 1994, 372, 746−754.
(22) Harris, P. A.; Cheung, M.; Hunter, R. N., III; Brown, M. L.;
Veal, J. M.; Nolte, R. T.; Wang, L.; Liu, W.; Crosby, R. M.; Johnson, J.
H.; Epperly, A. H.; Kumar, R.; Luttrell, D. K.; Stafford, J. A. Discovery
and Evaluation of 2-Anilino-5-aryloxazoles as a Novel Class of
VEGFR2 Kinase Inhibitors. J. Med. Chem. 2005, 48, 1610−1619.
5012
dx.doi.org/10.1021/jm3004043 | J. Med. Chem. 2012, 55, 5003−5012