Discovery and characterization of novel allosteric FAK inhibitors

Article abstract of DOI:10.1016/j.ejmech.2012.06.035

Focal adhesion kinase (FAK) regulates cell survival and proliferation pathways. Here we report the discovery of a highly selective series of 1,5-dihydropyrazolo[4,3-c][2,1]benzothiazines that demonstrate a novel mode of allosteric inhibition of FAK. These compounds showed slow dissociation from unphosphorylated FAK and were noncompetitive with ATP after long preincubation. Co-crystal structural analysis revealed that the compounds target a novel allosteric site within the C-lobe of the kinase domain, which induces disruption of ATP pocket formation leading to the inhibition of kinase activity. The potency of allosteric inhibition was reduced by phosphorylation of FAK. Coupled SAR analysis revealed that N-substitution of the fused pyrazole is critical to achieve allosteric binding and high selectivity among kinases.

Full text of DOI:10.1016/j.ejmech.2012.06.035

European Journal of Medicinal Chemistry 61 (2013) 49e60
Contents lists available at SciVerse ScienceDirect
European Journal of Medicinal Chemistry
journal homepage: http://www.elsevier.com/locate/ejmech
Original article
Discovery and characterization of novel allosteric FAK inhibitors
Misa Iwatani a, Hidehisa Iwata a, Atsutoshi Okabe a, Robert J. Skene b, Naoki Tomita a, Yoko Hayashi a,
,
,
*
Yoshio Aramaki a, David J. Hosfield b 1, Akira Hori a, Atsuo Baba a, Hiroshi Miki a
a Pharmaceutical Research Division, Takeda Pharmaceutical Company, Ltd., 26-1, Muraoka-Higashi 2-chome, Fujisawa, Kanagawa 251-8555, Japan
b Takeda California, 10410 Science Center Drive, San Diego, CA 92121, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
Focal adhesion kinase (FAK) regulates cell survival and proliferation pathways. Here we report the
discovery of a highly selective series of 1,5-dihydropyrazolo[4,3-c][2,1]benzothiazines that demonstrate
a novel mode of allosteric inhibition of FAK. These compounds showed slow dissociation from
unphosphorylated FAK and were noncompetitive with ATP after long preincubation. Co-crystal structural
analysis revealed that the compounds target a novel allosteric site within the C-lobe of the kinase
domain, which induces disruption of ATP pocket formation leading to the inhibition of kinase activity.
The potency of allosteric inhibition was reduced by phosphorylation of FAK. Coupled SAR analysis
revealed that N-substitution of the fused pyrazole is critical to achieve allosteric binding and high
selectivity among kinases.
Received 28 March 2012
Received in revised form
14 June 2012
Accepted 16 June 2012
Available online 26 June 2012
Keywords:
Focal adhesion kinase
Allosteric inhibitor
Slow dissociation
Cancer
Ó 2012 Elsevier Masson SAS. All rights reserved.
1. Introduction
Kinases are believed to exist in a state of equilibrium between
their active and inactive forms [8]. When the activation loop of
Focal adhesion kinase (FAK) is a non-receptor tyrosine kinase
that is found at integrin clusters, which are so-called focal adhe-
sions. FAK is ubiquitously expressed in the body tissues, and
functions as a switch and scaffold for the transduction of signals
from diverse stimuli, including integrins and growth factor recep-
tors [1]. Studies on defective or abnormal expression of FAK have
shown changes of cellular phenotype and have demonstrated a key
role of FAK in cellular functions. In several types of cancer, such as
cancer of the thyroid, prostate, colon and ovary, FAK expression or
activity is increased and is correlated with aggressive or metastatic
disease [2e6]. In addition, FAK has been shown to mediate both
tumor formation and malignant progression in a mouse model of
cancer [7]. Accordingly, specific inhibition of FAK activity could be
a new strategy for cancer therapy [1].
a kinase adjacent to its ATP-binding cleft is phosphorylated, cata-
lytic activity increases and the kinase is activated [8]. The activation
loop of a kinase generally contains 3 highly conserved residues
(Asp-Phe-Gly) that form the so-called DFG motif, the conformation
of which reflects the activation status. The active DFG-in confor-
mation is very similar among most kinases, and small molecules
targeting this DFG-in conformation are categorized as type I
inhibitors. Type I inhibitors bind to the ATP-binding site and
compete directly with ATP. Most of the kinase inhibitors developed
so far fall into this category, such as the EGFR inhibitor Gefitinib [9].
On the other hand, the inactive DFG-out conformation is unique to
each kinase. Type II inhibitors target the DFG-out conformation,
and have emerged as a new strategy for obtaining high selectivity
for a specific kinase. Type II inhibitors are a class of elongated
compounds that not only bind to the ATP-binding site, but also
extend into an adjacent deep pocket formed by conformational
rearrangement of the activation loop DFG motif. The BCR-ABL
inhibitor STI-571 [10], the p38 kinase inhibitor BIRB796 [11] and
VEGFR/PDGFR inhibitor TAK-593 [12] are categorized as type II
inhibitors. The inhibitors inducing conformational change of the
target kinase demonstrate slow dissociation, such as Lapatinib,
BIRB796 and TAK-593 [11e13]. Finally, several type III inhibitors
have been identified that bind to an allosteric site within the kinase
domain distant from the ATP-binding site and thus do not compete
directly with ATP. The MEK inhibitor PD184352 [14] and the AKT
Abbreviations: FAK, focal adhesion kinase; ATP, adenosine 50-triphosphate;
DMSO, dimethyl sulfoxide; DTT, DL-dithiothreitol; EDTA, ethylenediamine-
N,N,N0,N0-tetraacetic acid; MgCl2, magnesium chloride; MnCl2, manganese chlo-
ride; Tris, tris(hydroxymethyl)aminomethane; Tween 20, polyoxyethylene(20)sor-
bitan monolaurate; IC50, inhibitor concentration producing 50% inhibition; DFG,
Asp-Phe-Gly; HRD, His-Arg-Asp; PYK2, proline-rich tyrosine kinase 2; Src, v-src
avian sarcoma viral oncogene homolog (Schmidt-Ruppin A-2) 1.
* Corresponding author. Tel.: þ81 466 32 2747; fax: þ81 466 29 4469.
E-mail address: miki_hiroshi@takeda.co.jp (H. Miki).
Present address: Molecular Tagging Systems Inc., San Diego, CA 92101, United
1
States.
0223-5234/$ e see front matter Ó 2012 Elsevier Masson SAS. All rights reserved.
http://dx.doi.org/10.1016/j.ejmech.2012.06.035

50
M. Iwatani et al. / European Journal of Medicinal Chemistry 61 (2013) 49e60
inhibitor Akt-I-1 [15,16] are examples of type III inhibitors that are
highly selective among kinases.
compounds 1 and 2 were found to have a unique mode of action of
FAK with slow dissociation from enzyme.
Several FAK inhibitors have already been developed for the
treatment of solid tumors. PF-562771 inhibits FAK and proline-rich
tyrosine kinase 2 (PYK2) [17], while TAE226 inhibits FAK and
insulin-like growth factor 1 receptor (IGF-IR) [18]. Both compounds
are ATP-competitive type I inhibitors and bind to the ATP pocket in
a similar fashion [19,20]. In this study, we performed high-
throughput screening to discover novel FAK inhibitors. As
a result, we identified tricyclic sulfonamides (compounds 1 and 2)
that displayed slow dissociation. Enzymatic analysis and protein
crystal structure determination revealed that these compounds
were novel allosteric FAK inhibitors with high kinase selectivity,
which could be useful lead compounds for developing selective FAK
inhibitors.
3.2. ATP-noncompetitive inhibition requires inhibitor preincubation
To further investigate the inhibition mechanism of the 1,5-
dihydropyrazolo[4,3-c][2,1]benzothiazine derivatives, we per-
formed ATP competition assay at different time of preincubation of
the enzyme with compound 1 or 2. Apparent values of Km (Kmapp
)
under various concentrations of compounds were determined by
MichaeliseMenten analysis. After 60 min preincubation, Kmapp values
were almost the same regardless the concentration of compound 1
(Fig. 2A), which is consistent with ATP-noncompetitive inhibition of
compound 1 demonstrated in Fig. 1B. In constrast, without pre-
incubation time, we found increasing Kampp values depending on the
concentration of compound 1 (Fig. 2B). These results suggested that
compound 1 inhibited FAK competitively with ATP unless compound
and enzyme coexist before enzymatic reaction. Compound 2 had
same ATP inhibition mechanism as compound 1(Figure S1). Taken
together, ATP-competitive inhibition without preincubation time of
enzyme and compounds indicated that compounds 1 and 2 compete
with ATP and a slow dissociation property of these compounds
causes apparent ATP-noncompetitive like inhibition under 60 min
preincubation of FAK with the compounds.
2. Chemistry
Synthesis of compounds 1 and 2 is described in Scheme 1.
Methyl 2-amino-5-bromobenzoate 3 was converted into the cor-
responding N-methyl methansulfonamide, which was treated with
sodium hydride to afford the ketosultam 4 in 59% yield. Treatment
of 4 with N,N-dimethylformamide dimethyl acetal (DMFDMA),
followed by cyclocondensation with hydrazine hydrate or ethyl-
hydrazine provided 1,5-dihydropyrazolo[4,3-c][2,1]benzothiazine 5
or 6 in 88e92% yield. After protection of pyrazole, bromide 5 was
subjected to Pd-catalyzed coupling with 4-ethylphenylboronic acid,
and treated with HCl to afford compound 1 in 58% yield. Compound
2 was prepared from bromide 6 in a similar manner without
protection and deprotection in 72% yield.
3.3. Crystal structure analysis reveals that compounds 1 and 2 bind
the novel allosteric site in FAK
The intriguing kinetic properties exemplified by this series of
compound prompted us to investigate the binding mode of this
chemotype in crystals of the unphosphorylated FAK kinase domain
(Table S1). The crystal structures revealed that while FAK kinase
domain adopts the same overall fold as those determined in the
presence of ATP [21], the inhibitors bind in a novel hydrophobic
pocket w15 Å from the ATP-binding site (Fig. 3A). The pocket is
located within the kinase C-lobe, and extends to the interdomain
cleft where hydrophobic residues from the C-helix cap the binding
site. Consistent with the kinetic data, inhibitor binding involves
significant conformational rearrangements of two key catalytic
regions. Residues 543e552 of the catalytic HRD-loop adopt
a dramatically new conformation relative to the ATP-bound struc-
ture, where instead of folding back on the C-lobe, it now extends
toward the C-helix (Fig. 3B). In this position Ile547, directly adjacent
the HRD motif, has moved w6 Å away, providing access of
compounds 1 and 2 to the allosteric pocket. Notably, this HRD-loop
conformational change is coupled to additional rearrangements of
the DFG motif and activation loop. In the complex, the DFG motif is
shifted from its active “in” position into a DFG-out conformation
not yet observed in other kinases. In the crystal structure, the
N-terminus folds back toward the nucleotide binding site and
adopts an anti-parallel beta-strand conformation that hydrogen-
bonds to backbone atoms of residues within the Gly-rich loop
(Fig. 3C). In this position both Phe565 and Leu567 project into the
nucleotide binding site where they sterically prevent ATP binding.
3. Results
3.1. Tricyclic sulfonamides show slow dissociation from FAK
As part of a high-throughput lead identification effort, we
identified
a novel 1,5-dihydropyrazolo[4,3-c][2,1]benzothiazine
scaffold by screening our internal compound library against the
unphosphorylated FAK kinase domain (Fig. 1A). Follow up studies
identified the ATP-noncompetitive compounds 1 and 2, which
inhibited FAK kinase with IC50 values in the low micromolar range.
To investigate the time-independence of 1 and 2, we compared
inhibition profiles when the enzyme and inhibitor had been first
pre-incubated for either 5 or 60 min. When the ATP concentration
was low (0.5
mM) preincubation for 5 min had little effect on
activity. When the ATP concentration was high (1000
mM), pre-
incubation for 60 min had a slight effect on inhibitory activity; 2.4
and 3.0 fold stronger inhibition in compound 1 and 2, respectively
(Fig. 1B, C, Table 1). It is known from the literature that time-
dependent kinase inhibitors -such as Lapatinib [13], BIRB796 [11]
and TAK-593 [12] e often dissociate very slowly from the enzyme
due to rate-determining conformational changes required for
binding. In our studies both compounds showed slow dissociation
kinetics with half lives of 19.4 min (Fig. 1D). Taken together,
Scheme 1. Synthesis of compounds 1 and 2 (a) MsCl, Et3N, THF; (b) K2CO3, MeOH; (c) MeI, K2CO3, DMF; (d) NaH, DMF; (e) DMFDMA, THF; (f) RNHNH2, EtOH, reflux; (g) i) 5, DHP,
TFA; ii) 4-EtPhB(OH)2, Pd(OAc)2, PPh3, Na2CO3, DME, H2O, microwave; iii) HCl, MeOH; (h) 6, 4-EtPhB(OH)2, Pd(OAc)2, PPh3, Na2CO3, DME, H2O, microwave.

M. Iwatani et al. / European Journal of Medicinal Chemistry 61 (2013) 49e60
51
Fig. 1. Inhibition of compounds 1 and 2 with slow dissociation rates. (A) Chemical structure of compounds 1 and 2. Inhibition curves of compounds 1 (B) and 2 (C) against the
unphosphorylated kinase domain of FAK are shown. Inhibition was assessed under the following conditions: 0.5 M ATP with 5 min of preincubation: closed circles (C); 0.5 M ATP
with 60 min of preincubation: open circles (B); 1000 M ATP with 5 min of preincubation: closed squares (-); and 1000 M ATP with 60 min of preincubation: open squares (,).
m
m
m
m
The 100-fold dilution assay for FAK with compound 1 (C), compound 2 (☒), or no inhibitor as a control ( ) is shown in (D). The phosphorylation status of the kinase domain of FAK
7
was confirmed by immunoblotting (see at Lane 1 in Fig. 4AeD).
The binding interactions between compounds 1 and 2 in the
FAK allosteric site are mostly hydrophobic in nature. Exposure of
the allosteric pocket by the dramatic movement of Ile547 is
accompanied by an equally significant movement of Arg550 in
toward the allosteric binding site, where it packs along the tricyclic
core of compounds 1 and 2. A single polar interaction is observed
between the pyrazole N atom of compound 1 and a tightly bound
water molecule that links the inhibitor to backbone atoms of HRD-
loop residues Phe542 and His544 (Fig. 3D). These water-mediated
interactions are not crucial for inhibitor binding, however, as the
pyrazole ethyl of compound 2 is accommodated in the same
pocket, requiring only a small displacement of the loop immedi-
ately upstream of the HRD-motif (Fig. 3E). Notably, the terminal
ethylphenyl group of compounds 1 and 2 occupy the DFG-in
pocket, thereby dislodging Phe565 and enabling the structural
rearrangement of the activation loop and observed occlusion of the
FAK active site. These two structures thus identify the structure
and conformational changes of
a newly identified allosteric
binding site which is unique to FAK and not yet observed in any
other protein kinase.
3.4. Phosphorylation of FAK reduces allosteric inhibition
In cells, FAK kinase activity is regulated by phosphorylation of
tyrosine. In response to extracellular stimuli, FAK is autophos-
phorylated at Tyr397, providing a high affinity to binding site for Src
homology 2 (SH2)-domain containing proteins [22]. Association of
Src kinase with this site further activates FAK activity by phos-
phorylating residues Tyr576 and Tyr577 located within the acti-
vation loop of FAK [23]. As it is known that inhibitors that target
Table 1
Inhibitory activity of compound 1 and 2 against unphosphorylated kinase domain of FAK under each condition.
IC50
(
m
M)a
pre-incubation 5 min
ATP 0.5
pre-incubation 60 min
ATP 0.5
pre-incubation 5 min
ATP 1000
pre-incubation 60 min
ATP 1000
Compound
m
M
m
M
m
M
mM
1
2
1.1 (0.91e1.3)
1.6 (1.3e2.0)
0.99 (0.81e1.2)
1.3 (1.0e1.8)
2.1 (1.2e3.6)
3.0 (1.7e5.2)
0.88 (0.43e1.8)
1.0 (0.50e2.0)
a
Estimated values and 95% confidence interval in parentheses.

52
M. Iwatani et al. / European Journal of Medicinal Chemistry 61 (2013) 49e60
DFG-in conformation, which is the shared structure among kinases.
As shown in Fig. 5, compound 1 binds to the ATP-binding site and
the N2 nitrogen atom of the pyrazole ring forms a hydrogen bond
with the main chain of Glu500. On the other hand, compound 2 did
not seem to bind to the ATP-binding site because the ethyl
substituent of the pyrazole ring hindered formation of the
hydrogen bond with Glu500. Taken together with ATP-competitive
inhibition of compound
1 against phosphorylated FAK, we
hypothesized that N1-free 1,5-dihydropyrazolo[4,3-c][2,1]benzo-
thiazine 1 binds to the ATP-binding site when FAK is phosphory-
lated, whereas N1-substituted analogs like compound 2 cannot
inhibit phosphorylated FAK because of a lack of affinity toward the
ATP-binding site.
3.6. The N1-ethyl analog 2 has higher kinase selectivity
Several type III kinase inhibitors, such as allosteric inhibitors of
MEK1/2 and Akt1/2, have been reported to display very high kinase
selectivity [14,16,24e26]. Our data suggested that compound 1
shows different modes of inhibition for phosphorylated and
unphosphorylated FAK. To investigate the inhibitory activity of our
compounds against other kinases, kinase panel tests of compounds
1 and 2 were carried out. The panel consisted of 288 kinases (261
wild-type and 27 mutant kinases) supplied as the HotSpotSM Kinase
Profiling service by Reaction Biology Corporation. For these kinase
profiling tests, the kinase domain of FAK enzyme was used, whose
phosphorylation status was confirmed by immunoblotting
(Figure S2).
Fig. 2. ATP competition assay with compound 1. Relative kinase activity is plotted
against the ATP concentration in the presence of various concentrations of compound
1 with 60 min of preincubation (A) or without preincubation (B). Kinase activity
Among the 288 kinases tested, only 11 kinases including FAK
showed more than 50% inhibition by compound 1 (Table S2). These
were FAK (80%), CHK2 (50%), FGR (53%), FYN (61%), GCK/MAP4K2
(60%), HGK/MAP4K4 (75%), MLCK/MYLK (55%), MLK1/MAP3K9
without any inhibitor in the presence of 1000 mM ATP was set as 1.
inactive conformations have reduced activities against post-
translational modifications that enforce the active conformation,
we examined the inhibitory activity of compounds 1 and 2 using
phosphorylated FAK.
(77%), YES (71%), c-Kit (D816V) (64%), and PDGFRa (D842V) (74%).
PYK2, the only member of the FAK family with 60% amino acid
identity in its kinase domain [27,28], was only moderately inhibited
by compound 1 (25% inhibition). These results suggested that
compound 1 shows relatively high selectivity for FAK over other
kinases.
Next, we determined the IC50 values for compounds 1 and 2
against the 11 kinases and PYK2. In this kinase profiling test, the
IC50 value for compound 1 against FAK was almost equal to that for
We used two constructs of full-length FAK, phosphorylated and
unphosphorylated enzymes. By immunoblotting method, we
confirmed that Src-activated FAK enzyme was phosphorylated in
Tyr397, Tyr576/577 and Tyr925 (Lane 3 in Fig. 4AeD), while full
length or kinase domain of FAK enzyme untreated with Src was not
(Lane 1,2 in Fig. 4AeD). Using unphosphorylated full-length FAK,
we observed ATP-noncompetitive and time-dependent inhibition
profiles of compounds 1 and 2 similar to what was observed with
the isolated kinase domain. (Fig. 4F, H). When assayed against
phosphorylated FAK, however, compound 1 was less potent and
compound 2 was almost completely inactive (Fig. 4E, G). The IC50
ratios of the phosphorylated full length enzyme versus the
unphosphorylated full length enzyme were 4.1 (compound 1) and
compound 2 with 4.2
mM and 8.7 mM, respectively (Table 3).
Compound 1 inhibited the 10 kinases at micromolar levels,
consistent with the results of the larger kinase profiling study. In
contrast, compound 2 did not show more than 50% inhibition of
these 10 kinases as well as PYK2 even at 10 mM. The inhibitory
activity of compound 1 against PYK2 was independent of the
phosphorylation status, suggesting that there was little influence
on its binding of the conformational change due to phosphorylation
(Table 3, Figure S3). These results indicated that our allosteric
inhibitors, especially compound 2, had high kinase selectivity.
Taken together with the differing inhibitory activity of compounds
1 and 2 for phosphorylated FAK and the possible binding of
compound 1 to the ATP-binding site shown by docking model
analysis, we speculated that the N1-free analog 1 bound to the
ATP-binding site of all 10 kinases that it inhibited. Therefore, we
postulated that N-substitution of the pyrazole ring was necessary to
increase kinase selectivity.
>9 (compound 2) under 0.5 mM ATP and 60 min preincubation time
(Table 2). Importantly, the reduced affinity observed for compound
1 was accompanied with a change in inhibition mechanism as
compound 1 inhibits phosphorylated FAK competitively with
respect to ATP (Fig. 4E). Taken together, these results suggested that
both compounds 1 and 2 could target the FAK allosteric site in the
unphosphorylated enzyme, but only compound 1 would be able to
inhibit the enzyme when FAK adopt its active conformation.
3.5. Docking study of compound 1 with DFG-in FAK enzyme
4. Discussion
In assays using phosphorylated FAK, compound 2 showed no
inhibitory activity and compound 1 showed ATP-competitive
inhibition. These findings indicated a different mode of binding to
phosphorylated FAK by compound 1. To further investigate binding
to phosphorylated FAK, we docked compound 1 into FAK in the
We discovered novel allosteric FAK inhibitors with a mode of
binding that has not previously been reported for FAK or other
kinases. These tricyclic sulfonamides can be categorized as type III
inhibitors, and their allosteric binding indirectly competes with the

M. Iwatani et al. / European Journal of Medicinal Chemistry 61 (2013) 49e60
53
Fig. 3. Co-crystal structure of the compounds with FAK. (A) Ribbon diagram depiction of the FAK kinase domain in complex with compound 1. The unique HRD-loop conformation
is colored in dark blue (residues 543e552). The Allosteric binding site is labeled in blue and ATP-binding site is labeled in red. (B) Overlay of the ATP-bound FAK kinsase domain with
the FAK-compound 1 structure. The light green ribbon represents the HRD-loop of the ATP-bound DFG-in conformation and the blue ribbon represents the position of the HRD-loop
after binding compound 1 (residues 543e552). (C) Detailed view of the nucleotide binding site of FAK (in the presence of compound 1), showing gate keeper residue Met499 and
hinge residues Glu500, Cys502, and the steric occlusion of ATP-binding by activation loop residues Phe565 and Leu567 (C). Detailed allosteric binding site interactions observed in
the structure of FAK with compound 1(D) and compound 2 (E).
binding of ATP. They showed slow dissociation from FAK, leading to
apparent ATP-noncompetitive inhibition. To detect such slowly
dissociating allosteric inhibitors by screening in the present study,
the following points were critical. The first point is that we used the
unphosphorylated enzyme. Phosphorylation status affects the
enzyme conformation, with the DFG-in form being the phosphor-
ylated kinase and the DFG-out form being the unphosphorylated
kinase [8]. Most type II inhibitors are known to have stronger
affinity for the unphosphorylated kinase [10,29]. Likewise,
compounds 1 and 2 inhibited unphosphorylated FAK more strongly
than phosphorylated FAK, so that they might not have been
detected if we had used phosphorylated FAK for screening.
Although autophosphorylation is known to be important for full
kinase activity, some unphosphorylated tyrosine kinases display
kinase activity [10,12,29]. In fact, kinase activity of full-length
unphosphorylated FAK was detectable by using the high sensi-
tivity AlphaScreenÒ assay system. The kinase domain of FAK has
relatively strong activity even when it is unphosphorylated, prob-
ably due to absence of the FERM domain that is involved in auto-
inhibition of kinase activity by preventing autophosphorylation

54
M. Iwatani et al. / European Journal of Medicinal Chemistry 61 (2013) 49e60
Fig. 4. Phosphorylation status of FAK affects on the inhibitory activity of the compounds. Phosphorylation of the tyrosine sites, Tyr397 (A), Tyr576/577 (B), and Tyr925 (C) was
detected by immunoblotting. (D) Fluorescent staining was used to estimate total FAK. Lane 1: the unphosphorylated kinase domain of FAK, Lane 2: unphosphorylated full-length
FAK, Lane 3: phosphorylated full-length FAK. Inhibition curves for phosphorylated full-length FAK (compound 1; E, compound 2; G) and unphosphorylated full-length FAK
(compound 1; F, compound 2; H) are shown. Inhibition was assessed under the following conditions: 0.5
mM ATP with 5 min of preincubation: closed circles (C); 0.5
mM ATP with
60 min of preincubation: open circles (B); 1000 M ATP with 5 min of preincubation: closed squares (-); and 1000
m
mM ATP with 60 min of preincubation: open squares (,).
[30]. When kinases have low kinase activity in the unphosphory-
lated state, a binding assay system could be useful to find type II or
type III inhibitors [31,32].
ATP-noncompetitive inhibition after a long preincubation time, but
inhibited ATP in a competitive manner without preincubation [12],
because of binding to both the hinge region and the adjacent
backpocket by this type II inhibitor. Thus, managing the pre-
incubation time for enzyme assay is critical to find slowly dissoci-
ating inhibitors and understand their mode of inhibition.
Third, the combination of enzymatic analysis for the mode of
inhibition and structural analysis of co-crystals of the inhibitor
with enzyme was necessary to understand the binding behavior of
these tricyclic sulfonamides. Enzymatic experiments, such as the
ATP competition assay and time-dependent assay, did not confirm
allosteric inhibition. Additional co-crystal structural analysis
revealed that these compounds bind to a novel allosteric site and
Second, we varied the preincubation time prior to initiation of
the enzyme reaction. Compounds 1 and 2 both inhibited FAK in
a preincubation time-dependent manner and their inhibitory
activity was stronger after a long preincubation time. The dilution
assay identified these compounds as slowly dissociating inhibitors.
Thus, screening with a long preincubation time helps to find slowly
dissociating inhibitors. Without preincubation, on the other hand,
compounds 1 and 2 showed ATP-competitive inhibition, leading
presumption that their binding site competed with ATP either
directly or indirectly. The slowly dissociating TAK-593 also showed
Table 2
Inhibitory activity of compound 1 and 2 depending on phosphorylation status of FAK.
IC50
pre-incubation 60 min
(m
M)a
pre-incubation 5 min
pre-incubation 5 min
pre-incubation 60 min
Compound
ATP 0.5
mM
ATP 0.5
m
M
ATP 1000
m
M
ATP 1000 mM
Phosphorylated full length
1
2
1
2
16.3 (13.9e20.6)
>30
4.2 (3.4e5.2)
>10
10.7 (9.0e12.6)
>30
2.6 (2.2e3.0)
3.2 (2.7e3.8)
>30
>30
>10
>10
>30
>30
5.6 (4.1e7.8)
3.2 (2.4e4.2)
Unphosphorylated full length
a
Estimated values and 95% confidence interval in parentheses.

M. Iwatani et al. / European Journal of Medicinal Chemistry 61 (2013) 49e60
55
hinge region when the pyrazole nitrogens are free. This hypothesis
was based on three experimental results: 1) Compound 1 had
remaining inhibitory activity against phosphorylated FAK and
shifted slightly ATP-competitive inhibition. 2) Docking study
revealed that compound 1 have potential to bind to the hinge
region of DFG-in form of FAK. 3) Compound 1 inhibited some
kinases but compound 2 did not. It is known that the binding mode
of a kinase inhibitor differs depending on the target kinases. For
example, the BCR-ABL inhibitor STI-571 binds to ABL and c-Kit in
the DFG-out conformation, but binds to the DFG-in conformation of
SYK [34]. On the contrary, tricyclic sulfonamide 1 was postulated to
have two binding modes for a single target kinase dependent on its
phosphorylation status. Since we could observe only the allosteric
binding mode by co-crystal structural analysis, further investiga-
tion is needed to assess possible binding of N1-free analogs to the
ATP-binding site.
Although all kinases possess a conserved HRD-loop, our allo-
steric inhibitors, especially compound 2, had great selectivity over
288 kinases in the kinase panel assay. Compound 2 did not even
inhibit PYK2, which is surprising because the amino acid residues
that interact with the allosteric inhibitors are relatively well-
conserved in PYK2 (Table S3). The inhibitory activity of
compound 1 was similar for phosphorylated or unphosphorylated
PYK2, suggesting that its binding was not influenced by confor-
mational change related to the phosphorylation status. Small
sequence differences or dynamic conformational differences might
disrupt allosteric binding with PYK2 or other kinases. Since N1-free
pyrazole analogs have the potential to bind to the hinge region of
kinases, N1-substituted tricyclic sulfonamides, such as compound 2
may provide suitable scaffolds for developing highly selective FAK
inhibitors.
Fig. 5. Putative binding mode of compound 1 to FAK in the DFG-in conformation.
Ribbon diagram depiction of the FAK kinase domain in DFG-in conformation in
complex with compound 1. The docked compound 1 is colored yellow carbons. The
dotted line represents a hydrogen bond linking the inhibitor to the main chain of
Glu500.
disrupt ATP pocket formation, leading to the inhibition of kinase
activity. A similar inhibition mechanism has also been reported for
the kinesin spindle protein (KSP) inhibitor GSK-1, which binds to
the allosteric site of KSP but displays ATP-competitive inhibition
[33]. Detail enzymatic studies revealed that our allosteric FAK
inhibitors displayed slow dissociation. Taken together with co-
crystal structural analysis, we could speculate that the inhibitor
binding to allosteric site blocks ATP-binding indirectly, and
their slow dissociation results in apparent ATP-noncompetitive
inhibition.
The model of FAK allosteric inhibition was shown in Fig. 6.
Compounds 1 and 2 bound to the allosteric site on unphosphory-
lated FAK with slow dissociation rate and the potency of allosteric
binding diminished against phosphorylated FAK. The unique
feature of this tricyclic scaffold is that it also seems to bind to the
5. Conclusion
We discovered type III FAK inhibitors targeting a novel binding
site within the C-lobe of the kinase domain that has not been re-
ported for FAK or other kinases. These 1,5-dihydropyrazolo[4,3-c]
[2,1]benzothiazine derivatives showed slow dissociation, which
leads to apparent ATP-noncompetitive inhibition since the inhib-
itor binding to allosteric site blocks ATP-binding indirectly. Inter-
estingly, phosphorylation status affects the affinity or mode of
inhibition of these compounds for FAK. Compound 2, the N1-ethyl
analog, greatly reduced their potency, whereas the the N1-free
analog 1 inhibited phosphorylated FAK in an ATP-competitive
manner. A docking study and kinase profiling suggested that the
N1-free analog 1 have the potential to bind with the hinge regions.
Therefore, N-substitution of the pyrazole ring seems critical to
achieve allosteric binding and high kinase selectivity. In conclusion,
the kinetic and structural data reported here reveal a novel mode of
inhibition and define a new structural model for designing selective
FAK inhibitors as novel cancer therapeutics.
Table 3
Inhibitory activity of compound 1 and 2 against 12 kinases.
IC50(m
M)a
Kinases
Compound 1
Compound 2
FAK/PTK2
CHK2
FGR
4.2 (3.8e4.6)
3.4 (2.7e4.3)
1.9 (1.8e2.0)
2.8 (2.6e3.3)
3.0 (2.7e3.5)
1.6 (1.5e1.8)
4.0 (3.5e4.6)
0.37 (0.34e0.41)
1.8 (1.4e2.2)
2.8 (2.4e3.3)
1.7 (1.4e1.9)
10 (5.4e18.3)
6.8 (5.3e8.6)
8.7 (7.4e10.3)
>10
>10
>10
>10
>10
>10
>10
>10
>10
>10
>30
>30
6. Experimental protocols
FYN
GCK/MAP4K2
HGK/MAP4K4
MLCK/MYLK
MLK1
6.1. Chemistry
6.1.1. General
YES
Proton nuclear magnetic resonance (1H NMR, 300 MHz) spectra
and carbon nuclear magnetic resonance (13C NMR, 75 or 100 MHz)
spectra were determined with a Bruker DPX-300, AV-400, or
AVANCE II spectrometer. Chemical shifts are reported in parts per
c-Kit (D816V)
PDGFRa (D824V)
PYK2b
PYK2c
million (ppm) downfield from tetramethylsilane (d) as the internal
a
Estimated values and 95% confidence interval in parentheses.
Kinase activity of phophorylated PYK2 was measured by HTRF method.
Kinase activity of unphosphorylated PYK2 was measured by AlphaScreen assay
standard in deuterated solvent and coupling constants (J) are re-
ported in Hertz (Hz). The following abbreviations are used for spin
multiplicity: s ¼ singlet, d ¼ doublet, t ¼ triplet, q ¼ quartet,
b
c
method.

56
M. Iwatani et al. / European Journal of Medicinal Chemistry 61 (2013) 49e60
Fig. 6. Model of the allosteric inhibition of FAK.
m ¼ multiplet and br ¼ broad. LCMS was taken on a Shimadzu
Corporation LC-MS system (Column: CAPCELL PAK UG-120 ODS
(Shiseido Co., Ltd., 1.5 mm i.d. x 35 nm) operating in ESI (þ) ioni-
zation mode). Samples were eluted with a gradient of 10e95% or
5e90% acetonitrile containing 0.05% trifluoroacetic acid in water
containing 0.05% trifluoroacetic acid for 2, 3.5 or 4 min, and ultra
violet (UV) absorbance of 220 nm and 254 nm were collected by
a Shimadzu LC-10AD. Melting points, determined on BÜCHI
Melting Point M-545 or SRS OptiMelt, are uncorrected. Elemental
analyses were carried out by Takeda Analytical Research Labora-
tories, Ltd. or Discovery Research Laboratories, Takeda Pharma-
ceutical Company Ltd. Analytical thin layer chromatography (TLC)
was performed on silica gel 60 F254 plates (Merck). Flash column
chromatography separations were performed on Purif-Pack (SI or
NH, Shoko Scientific Co., Ltd), Silica gel 60 (Merck) or Chromatorex
Silica gel (100e200 mesh, Fuji Silysia Chemical Ltd.). All commer-
cially available solvents and reagents were used without further
purification.
the mixture was stirred at room temperature for 2 h. The reaction
mixture was poured into water, and extracted twice with ethyl
acetate. The organic layers were combined, washed with water and
brine, dried over MgSO4 and concentrated in vacuo. The residue was
purified by column chromatography (33% ethyl acetate in hexane)
to give the desired N-methylsulfonamide as a white powder (9.31 g,
77%). 1H NMR (CDCl3)
d 2.97 (3H, s), 3.28 (3H, s), 3.93 (3H, s), 7.33
(1H, d, J ¼ 8.5 Hz), 7.67 (1H, dd, J ¼ 8.5, 2.4 Hz), 8.04 (1H, d,
J ¼ 2.4 Hz).
To an ice-cooled solution of the N-methylsulfonamide (9.00 g,
27.9 mmol) in DMF (90 ml) was added sodium hydride (60% w/w in
mineral oil, 1.29 g, 32.1 mmol) and the mixture was stirred at room
temperature for 4 h. The reaction mixture was poured into ice-
water, and extracted twice with ethyl acetate. The organic layers
were combined, washed with brine, dried over MgSO4 and
concentrated in vacuo. The residue was purified by column chro-
matography (33% ethyl acetate in hexane) to give the title
compound as a white powder (7.75 g, 96%). 1H NMR (CDCl3)
d 3.44
(3H, s), 4.34 (2H, s), 7.05 (1H, d, J ¼ 8.9 Hz), 7.77 (1H, dd, J ¼ 8.9,
6.1.2. 6-Bromo-1-methyl-1H-2,1-benzothiazin-4(3H)-one 2,2-
dioxide (4)
2.4 Hz), 8.25 (1H, d, J ¼ 2.4 Hz).
To an ice-cooled mixture of methyl 2-amino-5-bromobenzoate
(3) (50.0 g, 217 mmol) and triethylamine (90.0 ml, 646 mol) in
THF (650 ml) was added dropwise methanesulfonyl chloride
(34.0 ml, 436 mmol) over 10 min. The mixture was stirred at room
temperature for 2 h. The reaction mixture was concentrated in
vacuo and the residue was suspended in MeOH (800 ml). To the
suspension was added a solution of K2CO3 (200 g, 1450 mmol) in
water (600 ml), and resulting white suspension was stirred at room
temperature for 16 h. Resulting solid was collected by filtration and
washed with water to give a crude product. The filtrate was
concentrated in vacuo, and the residue was suspended in water.
Insoluble solid was collected by filtration and combined with the
crude product. The product was washed with water and dried in
vacuo to give the desired methansulfonamide as a white powder
6.1.3. 8-Bromo-5-methyl-1,5-dihydropyrazolo[4,3-c][2,1]
benzothiazine 4,4-dioxide (5)
To a solution of 6-bromo-1-methyl-1H-2,1-benzothiazin-4(3H)-
one 2,2-dioxide (4) (6.10 g, 21.0 mmol) in THF (20 ml) was added
N,N-dimethylformamide dimethyl acetal (3.35 ml, 25.2 mmol) at
0
ꢀC. The mixture was stirred at room temperature for 3 h and
concentrated in vacuo. The residue was suspended in EtOH
(500 ml), and hydrazine hydrate (1.12 ml, 23.1 mmol) was added to
the mixture. The mixture was refluxed for 30 min, cooled to room
temperature, and concentrated in vacuo. The residual solid was
collected by filtration, washed with ethyl acetate, and dried in vacuo
to give the title compound as a white powder (6.08 g, 92%). 1H NMR
(DMSO-d6)
d
3.32 (3 H, s), 7.37 (1 H, d, J ¼ 8.9), 7.67 (1 H, dd, J ¼ 8.9,
2.5 Hz), 8.16 (1 H, d, J ¼ 2.5 Hz), 8.48 (1 H, s).
(53.3 g, 80%). 1H NMR (DMSO-d6)
d 3.18 (3H, s), 3.88 (3H, s), 7.54
(1H, d, J ¼ 8.9 Hz), 7.83 (1H, dd, J ¼ 8.9, 2.5 Hz), 8.02 (1H, d,
6.1.4. 8-(4-Ethylphenyl)-5-methyl-1,5-dihydropyrazolo[4,3-c][2,1]
benzothiazine 4,4-dioxide (1)
To a solution of 8-bromo-5-methyl-1,5-dihydropyrazolo[4,3-c]
[2,1]benzothiazine 4,4-dioxide (5) (2.86 g, 9.12 mmol) and 3,4-
dihydro-2H-pyran (1.24 ml, 13.6 mmol) in THF (30 ml) was added
trifluoroacetic acid (0.743 ml, 10.0 mmol) at 0 ꢀC. The mixture was
J ¼ 2.5 Hz), 9.99 (1H, s).
To a suspension of the methansulfonamide (11.6 g, 37.6 mmol)
in DMF (116 ml) was added K2CO3 (10.4 g, 75.2 mmol), and the
mixture was stirred at room temperature on a water bath. To the
suspension was added methyl iodide (2.81 ml, 45.2 mmol) and

M. Iwatani et al. / European Journal of Medicinal Chemistry 61 (2013) 49e60
57
heated at 80 ꢀC and stirred for 5 h. The reaction mixture was
neutralized by the addition of triethylamine and diluted with ethyl
acetate. The organic layer was washed with brine, dried over
MgSO4, and concentrated in vacuo. The residual solid was collected
by filtration, washed with ethyl acetateehexane (1:2), and dried in
vacuo to give the desired N-protected pyrazole (protected position
was not determined) as a white powder (3.52 g, 97%). 1H NMR
4-ethylphenylboronic acid (105 mg, 0.700 mmol), Pd(OAc)2
(13.1 mg, 0.0583 mmol), triphenylphosphine (30.7 mg,
0.117 mmol), 2 M Na2CO3 aq. (0.877 ml, 1.75 mmol) and 1,2-
dimethoxyethane (1.00 ml) was charged into a reaction vial and
flashed with nitrogen. The vial was sealed and heated at 150 ꢀC for
20 min under microwave irradiation. After cooling, the reaction
mixture was poured into saturated NaHCO3 aq. and extracted twice
with ethyl acetate. The organic layers were combined, washed with
brine, dried over MgSO4 and concentrated in vacuo. The residue was
purified by column chromatography (0e50% ethyl acetate in
hexane) to give the title compound as colorless crystals (154 mg,
72%). Mp. 130e133 ꢀC. LC-MS: m/z ¼ 368 (MHþ). 1H NMR (CDCl3)
(300 MHz, DMSO-d6)
d 1.53e1.63 (2 H, m), 1.64e1.80 (1 H, m),
1.91e2.08 (2 H, m), 2.12e2.29 (1 H, m), 3.33 (3 H, s), 3.64e3.75 (1 H,
m), 3.93e4.03 (1 H, m), 5.61e5.67 (1 H, m), 7.44 (1 H, d, J ¼ 8.9 Hz),
7.74 (1 H, dd, J ¼ 8.9, 2.4 Hz), 8.09 (1 H, d, J ¼ 2.4 Hz), 8.97 (1 H, s).
A mixture of the N-protected pyrazole (500 mg, 1.26 mmol),
4-ethylphenylboronic acid (226 mg, 1.51 mmol), Pd(OAc)2 (28.2 mg,
0.126 mmol), triphenylphosphine (65.9 mg, 0.251 mmol), 2 M
Na2CO3 aq. (1.88 ml, 3.77 mmol) and 1,2-dimethoxyethane
(2.50 ml) was charged into a reaction vial and flashed with
nitrogen. The vial was sealed and heated at 150 ꢀC for 30 min under
microwave irradiation. After cooling, the reaction mixture was
poured into saturated NaHCO3 aq., and extracted twice with ethyl
acetate. The organic layers were combined, washed with brine,
dried over MgSO4 and concentrated in vacuo. The residue was
purified by column chromatography (2e35% ethyl acetate in
hexane) to give the desired coupled product as a colorless gum
d
1.29 (3H, t, J ¼ 7.5 Hz), 1.63 (3H, t, J ¼ 7.3 Hz), 2.72 (2H, q,
J ¼ 7.5 Hz), 3.44 (3H, s), 4.57 (2H, q, J ¼ 7.3 Hz), 7.33 (2H, d,
J ¼ 8.3 Hz), 7.38 (1H, d, J ¼ 8.7 Hz), 7.45e7.55 (2H, m), 7.71 (1H, dd,
J ¼ 8.7, 2.0 Hz), 7.90e8.00 (2H, m). 13C NMR (75 MHz, CDCl3)
d 15.27,
15.56, 28.54, 31.80, 47.18, 116.21, 117.46, 119.92, 122.99, 126.88 (2C),
128.75 (2C), 129.26, 132.15, 136.96, 137.03, 137.31, 138.38, 144.28.
Anal. Calcd for C20H21N3O2S: C, 65.37; H, 5.76; N, 11.44. Found: C,
65.10, H, 5.92, N, 11.27.
6.2. Reagents
(430 mg, 81%). 1H NMR (CDCl3)
d
1.21e1.29 (3H, m), 1.57e1.86 (3H,
Poly-(GT)-biotin was purchased from Cisbio as a substrate for
FAK and PYK2. To detect the phosphorylated substrate, cryptate-
labeled PT66 (Cisbio, France) and streptavidin-Xlent! (Cisbio)
were used for the phosphorylated full length FAK kinase assay. To
detect the unphosphorylated kinase domain of FAK, unphosphory-
lated full-length FAK and PYK2, we used the AlphaScreenÒ Phos-
photyrosine (PT66) assay Kit (PerkinElmer, MA) or AlphaScreenÒ
Phosphotyrosine (P-Tyr-100) assay Kit (PerkinElmer). In immuno-
blotting analysis, rabbit anti- FAK phosphorylated Tyr397 antibody
(Invitrogen, CA), rabbit anti-FAK phosphorylated Tyr576/577 anti-
body (Cell Signaling, MA), and rabbit anti-FAK phosphorylated
Tyr925 antibody (Cell Signaling) were used to determine the
phosphorylation status of each FAK enzyme. For PYK2, rabbit anti-
PYK2 phosphorylated Tyr402 (Cell Signaling), rabbit anti-PYK2
phosphorylated Tyr579/580 (Invitrogen), and anti-PYK2 antibody
(Cell Signaling) were used.
m), 1.94e2.32 (3H, m), 2.69 (2H, q, J ¼ 7.5 Hz), 3.41 (3H, s),
3.66e3.88 (1H, m), 4.04e4.12 (1H, m), 5.52 (1H, dd, J ¼ 8.7, 3.2 Hz),
7.23e7.45 (3H, m), 7.57 (2H, d, J ¼ 8.3 Hz), 7.65 (1H, dd, J ¼ 8.3,
2.3 Hz), 8.17 (1H, s), 8.33 (1H, d, J ¼ 2.3 Hz).
A suspension of the coupled product (430 mg, 1.02 mmol) in 10%
HCl/MeOH (4.00 ml, 8.67 mmol) was stirred at room temperature
for 5 h. The reaction mixture was concentrated in vacuo, neutralized
with saturated NaHCO3 aq., and extracted twice with ethyl acetate.
The organic layers were combined, washed with brine, dried over
MgSO4 and concentrated in vacuo. The residue was purified by
column chromatography (2e50% ethyl acetate in hexane) and
recrystallized from ethyl acetate/hexane to give the title compound
as colorless crystals (256 mg, 74%). mp 197e199 ꢀC. LC-MS:
m/z ¼ 340 (MHþ). 1H NMR (CDCl3)
d
1.30 (3H, t, J ¼ 7.6 Hz), 2.72
(2H, q, J ¼ 7.7 Hz), 3.48 (3H, s), 7.33 (3H, m), 7.57 (2H, d, J ¼ 8.1 Hz),
7.72 (1H, dd, J ¼ 8.6, 2.2 Hz), 8.14 (1H, s), 8.22 (1H, br. s.), 10.57 (1H,
br. s.). 13C NMR (100 MHz, DMSO-d6 þ DCl)
d
15.44, 27.70, 30.83,
6.3. Preparation of enzymes
114.81, 116.53, 119.31, 121.49, 126.38 (2 C), 128.30, 128.40 (2 C),
129.54, 135.51, 135.96, 137.91, 141.64, 143.34. Note: One drop of
conc. DCl/D2O was added to obtain a sharp spectrum. Peaks at
116.53, 129.54 and 141.64 were not observed in the spectrum
without DCl, probably due to the tautomerization of the pyrazole
ring. Anal. Calcd for C18H17N3O2S: C, 63.70; H, 5.05; N, 12.38. Found:
C, 63.69; H, 4.86; N, 12.48.
Three forms of human FAK enzyme were prepared, including the
unphosphorylated FAK kinase domain, unphosphorylated full-
length FAK, and phosphorylated full-length FAK. The expression
plasmid for the FAK kinase domain (residues 411e686, Genebank
Accession No. NM_005607) was constructed in a pFastBacHTb
vector using a baculovirus expression system. FLAG-tagged full-
length FAK (residues 1e1052) was constructed in a pFTF (ꢁ) vector
using a baculovirus expression system. GST-tagged full-length FAK
activated by Src was supplied by Invitrogen. 6His-tagged full-length
human unphosphorylated PYK2 (residues 1e967, Genebank
Accession No. NM_173176) and GST-tagged full length human
phosphorylated PYK2 (residues 1e967) were purchased from Mil-
lipore and Carna Biosciences, respectively.
6.1.5. 8-Bromo-1-ethyl-5-methyl-1,5-dihydropyrazolo[4,3-c][2,1]
benzothiazine 4,4-dioxide (6)
The title compound was prepared from 6-bromo-1-methyl-1H-
2,1-benzothiazin-4(3H)-one 2,2-dioxide (4) and ethyl hydrazine in
a similar manner to 8-bromo-5-methyl-1,5-dihydropyrazolo[4,3-c]
[2,1]benzothiazine 4,4-dioxide (5).
Colorless crystals, 88% yield. Mp. 158e165 ꢀC. 1H NMR (CDCl3)
d
1.64 (3H, t, J ¼ 7.3 Hz), 3.42 (3H, s), 4.55 (2H, q, J ¼ 7.3 Hz), 7.23 (1H,
6.4. Immunoblotting
d, J ¼ 8.9 Hz), 7.65 (1H, dd, J ¼ 8.9, 2.3 Hz), 7.89 (1H, d, J ¼ 2.3 Hz),
7.96 (1H, s). Anal. Calcd for C12H12BrN3O2S: C, 42.12; H, 3.53; N,
12.28. Found: C, 42.04; H, 3.43; N, 12.22.
The phosphorylation status of the FAK and PYK2 enzymes was
determined by immunoblotting. 10 ng of FAK or PYK2 was sepa-
rated by SDS-PAGE under reducing conditions and transferred to
polyvinylidene difluoride membranes. The membranes were
blocked with BlockAce (DS Pharma Biomedical, Japan) or phos-
phoBlocker (Cell BioLabs, CA) for 1 h at room temperature. For
detection of the primary antibody, membranes were incubated
6.1.6. 1-Ethyl-8-(4-ethylphenyl)-5-methyl-1,5-dihydropyrazolo
[4,3-c][2,1]benzothiazine 4,4-dioxide (2)
A mixture of 8-bromo-1-ethyl-5-methyl-1,5-dihydropyrazolo
[4,3-c][2,1]benzothiazine 4,4-dioxide (6) (200 mg, 0.584 mmol),

58
M. Iwatani et al. / European Journal of Medicinal Chemistry 61 (2013) 49e60
with antibody diluted 1:1000 with immune enhancer solution
(Wako, Japan) overnight at 4 ꢀC. After washing the membranes with
TBST (2 mM Tris (pH 7.4), 50 mM NaCl, 0.01% (v/v) Tween 20),
incubation was done with anti-rabbit HRP-conjugated antibody
diluted 1:1000 with immune enhancer solution (Wako) for 1 h at
room temperature. After washing, membranes were developed
with 20X LumiGLOÒ Reagent and 20X Peroxide kit (Cell Signaling).
To estimate the total amount of FAK enzyme, we performed SDS-
PAGE with FAK (300 ng) under reducing conditions. Gels were
stained with ORIOLE fluorescent gel stain (Bio-Rad, CA) and visu-
alized under UV light.
phase). In this equation, vi was defined as zero, and vs was the slope
of the curve without compound obtained by linear regression [35].
6.7. ATP competition assay
FAK kinase activity was measured at different ATP concentra-
tions with or without preincubation of the enzyme and test
compound using the AlphaScreenÒ assay system, as described
previously. Data were fitted to a competitive kinetic mechanism of
inhibition in GraphPad Prism 5 Software (Ver 5.01).
6.8. Kinase panel assay
6.5. In vitro kinase assays
We used the HotSpotSM Kinase Profiling service (Reaction
Biology Corporation, PA) to assay 288 kinases, including 261 wild-
type kinases and 27 mutant kinases. The kinase buffer consisted of
20 mM HEPES (pH 7.5), 10 mM MgCl2, 1 mM EGTA, 0.02% Brij 35,
0.02 mg/ml BSA, 0.1 mM Na3VO4, 2 mM DTT, and 1% DMSO.
Assays for the unphosphorylated kinase domain of FAK,
unphosphorylated full-length FAK, and PYK2 were performed using
the AlphaScreenÒ assay system. The kinase buffer consisted of
50 mM TriseHCl (pH 7.5), 5 mM MgCl2, 5 mM MnCl2, 2 mM DTT,
and 0.01% Tween20. Optimal concentrations of enzymes or
substrates were 30 ng/mL for the unphosphorylated kinase domain
or unphosphorylated full-length enzyme and 0.1
(GT)-biotin. After preincubation of the enzyme, poly-(GT)-biotin,
and compounds for 5 or 60 min at room temperature, the kinase
Inhibitory activity was examined by detecting [
activity. Kinase reactions were performed in duplicate with 1
ATP and 3 M of the test compound. To evaluate the IC50 values for
compounds 1 and 2 against 11 kinases (CHK2, FAK, FGR, FYN,
MAP4K2, MAP4K4, MYLK, MLK1, YES, c-Kit (D816V), PDGFR
(D824V)), reactions were carried out using ATP at 1 M and five
concentrations (10, 3.3, 1.1, 0.11, and 0.011 M) of each compound.
g-
33P] ATP radio-
M of
m
m
g/mL for poly-
m
a
reaction was initiation by the addition of 0.5 or 1000
mM ATP and
m
then was terminated by adding final concentration of 100 mM
EDTA. Phosphorylated poly-(GT)-biotin was detected by using the
AlphaScreenÒ Phosphotyrosine (PT66) assay Kit for FAK and the
AlphaScreenÒ Phosphotyrosine (P-Tyr-100) assay Kit for PYK2 in
62.5 mM HEPES (pH 7.5) with 250 mM NaCl and 0.1% BSA. The
plates were incubated at room temperature in the dark for more
than 12 h and then read with an EnVision 2102 Multilabel Reader
(PerkinElmer). For phosphorylated full-length FAK, we used the
HTRFÒ detection system. The kinase buffer consisted of 50 mM
TriseHCl (pH 7.5), 5 mM MgCl2, 5 mM MnCl2, 2 mM DTT, 0.01%
Tween20, and 0.01% BSA. Optimal concentrations for the kinase
reaction was 30 ng/mL of phosphorylated full-length FAK and
m
An activity of 0% represented all of the components (including the
enzyme, substrate, and ATP) in the presence of EDTA, while 100%
activity was defined as the total reaction. Curve fitting was per-
formed by using the “Sigmoidal dose-response (variable slope)” in
GraphPad Prism software, with the top and bottom of the curve
constrained at 100 and 0, respectively. If inhibitory activity at 10
was less than 50%, IC50 values could not be estimated.
mM
6.9. Crystallization, data collection and structure solution of FAK
Crystals of the FAK in complex with compound 1 and compound
2 were prepared by incubation of 10 mg/ml protein with 1 mM
compound from 50 mM DMSO stock solutions and left on ice for
1 h. Initial crystal trials were conducted utilizing Takeda California’s
automated nanovolume crystallization technology platform. Large
crystals suitable for data collection were obtained from reservoir
solution containing 100 mM HEPES (pH 7.5), 17% PEG 4000, 8.5%
isopropanol and 15% glycerol. Crystals selected for data collection
were flash frozen in mother liquor with liquid nitrogen in an
Advanced Light Source (ALS) compatible crystal mounting cassette.
Diffraction data for compound 1 was collected from cryo-cooled
crystals at the ALS beamline 5.0.3. Diffraction data for compound
2 was collected from cryo-cooled crystals at Stanford Synchrotron
Radiation Lightsource (SSRL) beamline 7e1. Data were reduced
using the HLK2000 software package. The structures were deter-
mined by molecular replacement using the program MOLREP from
the CCP4 program suite employing the coordinates of FAK (PDB
ID:1MP8) as a search model [21]. Subsequent structure refinement
and model re-building were conducted utilizing REFMAC and
XtalView [36] software packages. All solved structures displayed
between 99.5% and 100% of the non-glycine residues in the most
favored and allowed regions of the Ramachandran plot, and
stereochemical verification was performed by PROCHECK [37]. Data
collection and refinement statistics are included in Table S1, and
Fig. 3 were generated using PyMol.
5 mg/mL of poly-(GT)-biotin. As described above, the kinase reaction
was started by adding ATP to the enzyme, poly-(GT)-biotin, and test
compounds, and was terminated by adding 60 mM EDTA diluted
with the detection buffer (50 mM HEPES (pH 7.0), 0.02% NaN3,
0.1% BSA, and 0.8 M KF). The detection mixture contained crypate-
labeled PT66 and streptavidin-Xlent!. After incubation at room
temperature for 1 h, the plates were read with an EnVision 2102
Multilabel Reader. We set the reaction without enzyme as 100%
inhibitory activity and the total reaction as 0% inhibitory activity.
Sigmoidal curves, IC50 values, and 95% confidence intervals were
estimated using the “Sigmoidal dose-response (variable slope)” in
GraphPad Prism 5 Software (Ver 5.01), with the top and bottom of
the curve constrained at 100 and 0, respectively. If the inhibitory
activity was less than 50% at the highest tested concentration of the
compound, IC50 values were not determined due to inaccuracy.
6.6. 100-fold dilution assay
FAK kinase activity was measured by the AlphaScreenÒ assay
system, as described previously. Unphosphorylated FAK kinase
domain enzyme (300 ng/mL) and a 10-fold higher concentration
than its IC50 value were incubated for 60 min to allow the
enzymeeinhibitor complex to reach equilibrium. The complex was
diluted to 1:100 with 1 mM ATP and 0.1 mg/mL poly-(GT)-biotin to
initiate the kinase reaction. The rate constant for onset of inhibition
(kobs) and the dissociation half-life (t1/2) were determined by the
following equation: P ¼ vs*t þ (vievs) *[1-exp (-kobs*t)]/kobs, t1/
6.10. Docking study
¼ 0.693/kobs. (P; product, t; time, vi; initial velocity, vs; steady state
We used the X-ray co-crystal structure of FAK in the Protein Data
Bank (PDB ID: 2ETM), which shows the DFG-in conformation. The
2
velocity, kobs; the rate constant for conversion from vi phase to vs

M. Iwatani et al. / European Journal of Medicinal Chemistry 61 (2013) 49e60
59
L. Shewchuk, A unique structure for epidermal growth factor receptor bound
to GW572016 (Lapatinib): relationships among protein conformation, inhib-
itor off-rate, and receptor activity in tumor cells, Cancer Res. 64 (2004)
6652e6659.
protein structure for the docking study was prepared with Protein
Preparation Workflow in Schrödinger software, as described below.
(1) All crystallographic waters were removed and chain A was kept.
(2) Hydrogens were added. (3) The orientation of hydroxyl groups,
Asn, and Gln, and the protonation states of His were optimized. (4)
Constrained energy minimization was performed with the
OPLS_2005 force field, setting the max the root-mean square
deviation (RMSD) to 0.30 for the heavy atoms. Compound 1 was
sketched in Maestro. Molecular docking was performed with Glide
SP (version 5.6), using a van der Waals radius scaling factor of 0.8
for the non-polar atoms of the ligand. The docking box was
centered on the ligand of the crystal structure. The best scoring
pose of GlideScore was evaluated.
[14] J.F. Ohren, H. Chen, A. Pavlovsky, C. Whitehead, E. Zhang, P. Kuffa, C. Yan,
P. McConnell, C. Spessard, C. Banotai, W.T. Mueller, A. Delaney, C. Omer,
J. Sebolt-Leopold, D.T. Dudley, I.K. Leung, C. Flamme, J. Warmus, M. Kaufman,
S. Barrett, H. Tecle, C.A. Hasemann, Structures of human MAP kinase kinase 1
(MEK1) and MEK2 describe novel noncompetitive kinase inhibition, Nat.
Struct. Mol. Biol. 11 (2004) 1192e1197.
[15] S.F. Barnett, D. Defeo-Jones, S. Fu, P.J. Hancock, K.M. Haskell, R.E. Jones,
J.A. Kahana, A.M. Kral, K. Leander, L.L. Lee, J. Malinowski, E.M. McAvoy,
D.D. Nahas, R.G. Robinson, H.E. Huber, Identification and characterization of
pleckstrin-homology-domain-dependent and isoenzyme-specific Akt inhibi-
tors, Biochem. J. 385 (2005) 399e408.
[16] C.W. Lindsley, Z. Zhao, W.H. Leister, R.G. Robinson, S.F. Barnett, D. Defeo-Jones,
R.E. Jones, G.D. Hartman, J.R. Huff, H.E. Huber, M.E. Duggan, Allosteric Akt
(PKB) inhibitors: discovery and SAR of isozyme selective inhibitors, Bioorg.
Med. Chem. Lett. 15 (2005) 761e764.
[17] J.K. Slack-Davis, K.H. Martin, R.W. Tilghman, M. Iwanicki, E.J. Ung, C. Autry,
M.J. Luzzio, B. Cooper, J.C. Kath, W.G. Roberts, J.T. Parsons, Cellular charac-
terization of a novel focal adhesion kinase inhibitor, J. Biol. Chem. 282 (2007)
14845e14852.
[18] T.J. Liu, T. LaFortune, T. Honda, O. Ohmori, S. Hatakeyama, T. Meyer, D. Jackson,
J. de Groot, W.K. Yung, Inhibition of both focal adhesion kinase and insulin-
like growth factor-I receptor kinase suppresses glioma proliferation in vitro
and in vivo, Mol. Cancer Ther. 6 (2007) 1357e1367.
[19] D. Lietha, M.J. Eck, Crystal structures of the FAK kinase in complex with
TAE226 and related bis-anilino pyrimidine inhibitors reveal a helical DFG
conformation, PLoS One 3 (2008) e3800.
[20] W.G. Roberts, E. Ung, P. Whalen, B. Cooper, C. Hulford, C. Autry, D. Richter,
E. Emerson, J. Lin, J. Kath, K. Coleman, L. Yao, L. Martinez-Alsina, M. Lorenzen,
M. Berliner, M. Luzzio, N. Patel, E. Schmitt, S. LaGreca, J. Jani, M. Wessel,
E. Marr, M. Griffor, F. Vajdos, Antitumor activity and pharmacology of
a selective focal adhesion kinase inhibitor, PF-562,271, Cancer Res. 68 (2008)
1935e1944.
[21] J. Nowakowski, C.N. Cronin, D.E. McRee, M.W. Knuth, C.G. Nelson,
N.P. Pavletich, J. Rogers, B.C. Sang, D.N. Scheibe, R.V. Swanson, D.A. Thompson,
Structures of the cancer-related Aurora-A, FAK, and EphA2 protein kinases
from nanovolume crystallography, Structure 10 (2002) 1659e1667.
[22] M.D. Schaller, J.D. Hildebrand, J.D. Shannon, J.W. Fox, R.R. Vines, J.T. Parsons,
Autophosphorylation of the focal adhesion kinase, pp125FAK, directs SH2-
dependent binding of pp60src, Mol. Cell Biol. 14 (1994) 1680e1688.
[23] M.B. Calalb, T.R. Polte, S.K. Hanks, Tyrosine phosphorylation of focal adhesion
kinase at sites in the catalytic domain regulates kinase activity: a role for Src
family kinases, Mol. Cell Biol. 15 (1995) 954e963.
[24] E.M. Wallace, J. Lyssikatos, J.F. Blake, J. Seo, H.W. Yang, T.C. Yeh, M. Perrier,
H. Jarski, V. Marsh, G. Poch, M.G. Livingston, J. Otten, G. Hingorani,
R. Woessner, P. Lee, J. Winkler, K. Koch, Potent and selective mitogen-
activated protein kinase kinase (MEK) 1,2 inhibitors. 1. 4-(4-bromo-2-
fluorophenylamino)-1- methylpyridin-2(1H)-ones, J. Med. Chem. 49 (2006)
441e444.
[25] M.A. Bogoyevitch, D.P. Fairlie, A new paradigm for protein kinase inhibition:
blocking phosphorylation without directly targeting ATP binding, Drug Dis-
cov. Today 12 (2007) 622e633.
[26] S.F. Barnett, M.T. Bilodeau, C.W. Lindsley, The Akt/PKB family of protein
kinases: a review of small molecule inhibitors and progress towards target
validation, Curr. Top. Med. Chem. 5 (2005) 109e125.
[27] I.H. Gelman, Pyk 2 FAKs, any two FAKs, Cell Biol. Int. 27 (2003) 507e510.
[28] H. Avraham, S.Y. Park, K. Schinkmann, S. Avraham, RAFTK/Pyk2-mediated
cellular signalling, Cell Signal. 12 (2000) 123e133.
[29] K.S. Gajiwala, J.C. Wu, J. Christensen, G.D. Deshmukh, W. Diehl, J.P. DiNitto,
J.M. English, M.J. Greig, Y.A. He, S.L. Jacques, E.A. Lunney, M. McTigue,
D. Molina, T. Quenzer, P.A. Wells, X. Yu, Y. Zhang, A. Zou, M.R. Emmett,
A.G. Marshall, H.M. Zhang, G.D. Demetri, KIT kinase mutants show unique
mechanisms of drug resistance to imatinib and sunitinib in gastrointestinal
stromal tumor patients, Proc. Natl. Acad. Sci. U. S. A. 106 (2009)
1542e1547.
Acknowledgments
We thank Clifford D. Mol and Gyorgy Snell for analysis of crystal
structure and X-ray data collections. We thank Motoo Iida and
Yasumi Kumagai for analysis of 13C NMR data. We thank Katsunori
Sasa and Akihiko Naotsuka for high-throughput screening and
Taeko Yoshida for assistance with the enzymatic assays. We also
thank Seiji Yamasaki and Kouya Kimoto for assessing physicality of
compounds, and Tomohiro Kawamoto and Mark Hixon for helpful
discussion about data analysis and revision of the manuscript.
Appendix A. Supplementary data
Supplementary data related to this article can be found online at
http://dx.doi.org/10.1016/j.ejmech.2012.06.035.
References
[1] G.W. McLean, N.O. Carragher, E. Avizienyte, J. Evans, V.G. Brunton, M.C. Frame,
The role of focal-adhesion kinase in cancer - a new therapeutic opportunity,
Nat. Rev. Cancer 5 (2005) 505e515.
[2] L.V. Owens, L. Xu, G.A. Dent, X. Yang, G.C. Sturge, R.J. Craven, W.G. Cance, Focal
adhesion kinase as a marker of invasive potential in differentiated human
thyroid cancer, Ann. Surg. Oncol. 3 (1996) 100e105.
[3] C. Recher, L. Ysebaert, O. Beyne-Rauzy, V. Mansat-De Mas, J.B. Ruidavets,
P. Cariven, C. Demur, B. Payrastre, G. Laurent, C. Racaud-Sultan, Expression of
focal adhesion kinase in acute myeloid leukemia is associated with enhanced
blast migration, increased cellularity, and poor prognosis, Cancer Res. 64
(2004) 3191e3197.
[4] W.G. Cance, J.E. Harris, M.V. Iacocca, E. Roche, X. Yang, J. Chang, S. Simkins,
L. Xu, Immunohistochemical analyses of focal adhesion kinase expression in
benign and malignant human breast and colon tissues: correlation with
preinvasive and invasive phenotypes, Clin. Cancer Res. 6 (2000) 2417e2423.
[5] P.L. Judson, X. He, W.G. Cance, L. Van Le, Overexpression of focal adhesion
kinase, a protein tyrosine kinase, in ovarian carcinoma, Cancer 86 (1999)
1551e1556.
[6] L. Tremblay, W. Hauck, A.G. Aprikian, L.R. Begin, A. Chapdelaine, S. Chevalier,
Focal adhesion kinase (pp125FAK) expression, activation and association with
paxillin and p50CSK in human metastatic prostate carcinoma, Int. J. Cancer 68
(1996) 164e171.
[7] G.W. McLean, N.H. Komiyama, B. Serrels, H. Asano, L. Reynolds, F. Conti,
K. Hodivala-Dilke, D. Metzger, P. Chambon, S.G. Grant, M.C. Frame, Specific
deletion of focal adhesion kinase suppresses tumor formation and blocks
malignant progression, Genes Dev. 18 (2004) 2998e3003.
[30] D. Lietha, X. Cai, D.F. Ceccarelli, Y. Li, M.D. Schaller, M.J. Eck, Structural basis for
the autoinhibition of focal adhesion kinase, Cell 129 (2007) 1177e1187.
[31] J.R. Simard, C. Grutter, V. Pawar, B. Aust, A. Wolf, M. Rabiller, S. Wulfert,
A. Robubi, S. Kluter, C. Ottmann, D. Rauh, High-throughput screening to
identify inhibitors which stabilize inactive kinase conformations in p38alpha,
J. Am. Chem. Soc. 131 (2009) 18478e18488.
[8] L.N. Johnson, M.E. Noble, D.J. Owen, Active and inactive protein kinases:
structural basis for regulation, Cell 85 (1996) 149e158.
[9] A.E. Wakeling, S.P. Guy, J.R. Woodburn, S.E. Ashton, B.J. Curry, A.J. Barker,
K.H. Gibson, ZD1839 (Iressa): an orally active inhibitor of epidermal growth
factor signaling with potential for cancer therapy, Cancer Res. 62 (2002)
5749e5754.
[32] S. Kluter, C. Grutter, T. Naqvi, M. Rabiller, J.R. Simard, V. Pawar, M. Getlik,
D. Rauh, Displacement assay for the detection of stabilizers of inactive kinase
conformations, J. Med. Chem. 53 (2010) 357e367.
[10] T. Schindler, W. Bornmann, P. Pellicena, W.T. Miller, B. Clarkson, J. Kuriyan,
Structural mechanism for STI-571 inhibition of abelson tyrosine kinase,
Science 289 (2000) 1938e1942.
[33] L. Luo, C.A. Parrish, N. Nevins, D.E. McNulty, A.M. Chaudhari, J.D. Carson,
V. Sudakin, A.N. Shaw, R. Lehr, H. Zhao, S. Sweitzer, L. Lad, K.W. Wood,
R. Sakowicz, R.S. Annan, P.S. Huang, J.R. Jackson, D. Dhanak, R.A. Copeland,
K.R. Auger, ATP-competitive inhibitors of the mitotic kinesin KSP that function
via an allosteric mechanism, Nat. Chem. Biol. 3 (2007) 722e726.
[34] S. Atwell, J.M. Adams, J. Badger, M.D. Buchanan, I.K. Feil, K.J. Froning, X. Gao,
J. Hendle, K. Keegan, B.C. Leon, H.J. Muller-Dieckmann, V.L. Nienaber,
B.W. Noland, K. Post, K.R. Rajashankar, A. Ramos, M. Russell, S.K. Burley,
[11] C. Pargellis, L. Tong, L. Churchill, P.F. Cirillo, T. Gilmore, A.G. Graham,
P.M. Grob, E.R. Hickey, N. Moss, S. Pav, J. Regan, Inhibition of p38 MAP kinase
by utilizing a novel allosteric binding site, Nat. Struct. Biol. 9 (2002) 268e272.
[12] H. Iwata, S. Imamura, A. Hori, M.S. Hixon, H. Kimura, H. Miki, Biochemical
characterization of TAK-593, a novel VEGFR/PDGFR inhibitor with a two-step
slow binding mechanism, Biochemistry 50 (2011) 738e751.
[13] E.R. Wood, A.T. Truesdale, O.B. McDonald, D. Yuan, A. Hassell, S.H. Dickerson,
B. Ellis, C. Pennisi, E. Horne, K. Lackey, K.J. Alligood, D.W. Rusnak, T.M. Gilmer,

60
M. Iwatani et al. / European Journal of Medicinal Chemistry 61 (2013) 49e60
S.G. Buchanan, A novel mode of Gleevec binding is revealed by the structure
of spleen tyrosine kinase, J. Biol. Chem. 279 (2004) 55827e55832.
[35] R.A. Copeland, Evaluation of enzyme inhibitors in drug discovery. A guide for
medicinal chemists and pharmacologists, Methods Biochem. Anal. 46 (2005)
1e265.
[36] D.E. McRee, XtalView/XfiteA versatile program for manipulating atomic
coordinates and electron density, J. Struct. Biol. 125 (1999) 156e165.
[37] R.A. Laskowski, M.W. MacArthur, D.S. Moss, J.M. Thornton, PROCHECK:
a program to check the stereochemical quality of protein structures, J. Appl.
Crystallogr. 26 (1993) 283e291.