Biochemical and biophysical characterization of unique switch pocket inhibitors of p38α

Article abstract of DOI:10.1016/j.bmcl.2010.04.097

Herein we describe the identification and characterization of a class of molecules that are believed to extend into a region of p38 known as the 'switch pocket'. Although these molecules lack a canonical hinge binding motif, they show K_i values as low as 100 nM against p38. We show that molecules that interact with this region of the protein demonstrate different binding kinetics than a canonical ATP mimetic, as well as a wide range of kinome profiles. Thus, the switch pocket presents new opportunities for kinome selectivity which could result in unique biochemical responses and offer new opportunities in the field of kinase drug discovery.

Full text of DOI:10.1016/j.bmcl.2010.04.097

Bioorganic & Medicinal Chemistry Letters 20 (2010) 5787–5792
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Biochemical and biophysical characterization of unique switch pocket
inhibitors of p38
a
*
Steven L. Swann , Philip J. Merta, Lemma Kifle, Duncan Groebe, Kathy Sarris, Philip J. Hajduk, Chaohong Sun
Global Pharmaceutical Research and Development, Abbott Laboratories, Dept., Bldg. AP10-L04L, Abbott Park, IL 60064-6217, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
Herein we describe the identification and characterization of a class of molecules that are believed to
extend into a region of p38 known as the ‘switch pocket’. Although these molecules lack a canonical hinge
binding motif, they show Ki values as low as 100 nM against p38. We show that molecules that interact
with this region of the protein demonstrate different binding kinetics than a canonical ATP mimetic, as
well as a wide range of kinome profiles. Thus, the switch pocket presents new opportunities for kinome
selectivity which could result in unique biochemical responses and offer new opportunities in the field of
kinase drug discovery.
Received 18 February 2010
Revised 20 April 2010
Accepted 21 April 2010
Available online 28 April 2010
Keywords:
Kinase
Switch pocket
p38
Ó 2010 Elsevier Ltd. All rights reserved.
DFG-out
The spatial and temporal control of protein kinases can have
significant effects on cell division, growth, development and viabil-
ity.1 In terms of drug discovery, kinases offer the potential for mod-
ulating a wide range of phenotypic effects in practically every
therapeutic area. The past 15 years of kinase drug discovery has
clearly shown that the kinase ATP binding pocket is typically a very
druggable entity, and direct competition with ATP binding can
translate into subnanomolar inhibition of kinase activity. Unfortu-
nately, the homology of the ATP binding site across the kinome
makes it very difficult to design selective inhibitors or predict
activity across the other members of the kinase family. Also, trans-
lating kinase selectivity into a desirable pharmacological and tox-
icological profile can be very difficult. Thus, although kinases
serve as attractive targets for drug discovery, the pleiotropic role
of these enzymes continues to translate into significant off-target
and toxic effects.
In the past five years, allosteric inhibition of kinases has
emerged as an intriguing new approach to drug discovery because
targeting domains outside of the ATP binding region, which can
have high levels of homology across the kinome, may afford in-
creased selectivity and reduced toxicity. The discovery of mole-
cules that do not bind directly to the canonical ATP site was first
N
O
Cl
O
O
N
F
O
N
N
N
O
N
N
NH
N
N
S
N
O
N
N
N
S
O
O
1
2
3
Figure 1. Structures of known p38a inhibitors: (1) SB 203580; (2) Doramapimod (BIRB-796); (3) Deciphera switch pocket inhibitor (Deciphera Pharmaceuticals LLC).
* Corresponding author. Tel.: +1 847 937 9626.
E-mail address: Steven.Swann@abbott.com (S.L. Swann).
0960-894X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2010.04.097

5788
S. L. Swann et al. / Bioorg. Med. Chem. Lett. 20 (2010) 5787–5792
Figure 4. Model of various amide-based switch pocket inhibitors (lines) overlayed
with BIRB-796 (yellow sticks) suggesting potential hydrogen bonding interactions
with Asp160 and Arg70 of the switch pocket.
iting only four other kinases with an IC50 less than 1 lM in an
Figure 2. Crystal structure of p38a (1KV2) highlighting the (A) ATP site (red mesh),
DFG-out site (dark blue spheres) and switch pocket (light blue spheres).
in-house kinome panel of 50 kinases. Although no further indica-
tion of the progression of this compound has been disclosed, allo-
steric inhibition of p38 as well as other kinases, was now seen as a
viable approach to identifying novel drug candidates. Unfortu-
nately, despite only a few additional reports of non-ATP competi-
tive kinase inhibitors,3 there is a need for additional biochemical
characterization of these types of molecules.
successfully disclosed by Boehringer Ingelheim in 2002 where
their Phase IIb/III candidate Doramapimod, a p38a inhibitor, was
in development for psoriasis, rheumatoid arthritis and Crohn’s dis-
ease.2 This molecule exhibited very good kinome selectivity inhib-
Figure 3. (A) X-ray structure of BIRB-796 in p38
a
(1KV2); (B) Deciphera switch pocket inhibitor docked in p38
a
suggesting potential hydrogen bonding interactions with
Arg70 of the switch pocket.
O
C
N
O
H2N
NH
N
N
NH2
N
AcOH / EtOH
HO
HO
4
5
O
O
O
H2HR or HNRR2
HATU
O
N
N
H
N
H
N
N
N
H
N
N
H
R/H
N
R2
HO
7-31
6
O
O
Scheme 1. Synthesis of designed library of p38a
switch pocket compounds.11

S. L. Swann et al. / Bioorg. Med. Chem. Lett. 20 (2010) 5787–5792
5789
In light of this, there is a clear need to further identify and char-
acterize molecules that bind to distinct allosteric sites outside of
the kinase ATP binding site. Here we describe the biophysical
and biochemical characterization of a class of small molecules that
have unique biochemical consequences as well as present new
opportunities for selectivity across the kinome.
P38 kinase, or MAPK14, plays a critical role in the production of
proinflammatory cytokines such as TNFa and IL-1b and is activated
inhibit p38
a
by binding to a novel site outside of the canonical ATP
by a variety of cellular stresses including osmotic shock, lipop-
binding pocket known as the switch pocket. Although this unique
class of compounds lacks a canonical hinge binding motif, some
olysacchardies (LPS) and growth factors.1,4 Although there are four
highly similar isoforms of p38 (p38a, p38b, p38c, and p38d, the
demonstrate low nanomolar inhibition of p38
different binding kinetics, as compared to canonical ATP competi-
tive inhibitors. In addition, interacting with the switch pocket may
a
and exhibit very
alpha isoform is believed to be the key isoform involved in the
inflammatory response and thus may offer a novel target for treat-
ment of a variety of diseases including Rheumatoid arthritis and
Table 1
Binding affinities of switch pocket inhibitors of p38
a
in a TR-FRET assay
O
N
N
H
N
H
N
R/H
N
R2
O
Compounds
X
Y
p38
a
Binding Kia
(
lM)
Compounds
X
Y
p38a Binding Ki (lM)
N
SB 203580 (1)
—
—
0.057
18
H
0.327
N
O
BIRB-796 (2)
Deciphera (3)
—
—
—
—
0.022
0.099
19
20
H
H
0.336
0.433
N
H
N
Ph
7
8
H
N
0.086
0.089
21
22
H
0.471
0.715
OH
NH2
–(CH2)3NH2
–CH3
N
O
HO
9
H
H
0.112
0.140
23
24
–CH2CH3
0.735
0.786
S
O
O
N
N
10
H
H2N
NH2
N
O
11
12
–CH2CH3
H
0.157
0.176
25
26
H
0.885
1.39
N
O
N
–CH2CH3
13
14
H
H
–CH2CN
0.252
0.283
27
28
H
H
1.40
1.45
NH
NH2
–(CH2)3OH
O
O
15
16
17
H
0.291
0.301
0.312
29
30
31
H
H
H
>3.9
N
N
H
>3.9
>3.9
NH2
O
–CH2(CH3)2
a
Ki values were calculated using the Cheng–Prusoff equation (Ki = (IC50)/(1 + [probe]/Kd).

5790
S. L. Swann et al. / Bioorg. Med. Chem. Lett. 20 (2010) 5787–5792
Crohn’s disease. To date, the scientific literature is replete with
publications highlighting the discovery and optimization of com-
pounds that inhibit p38a, largely through binding to the canonical
ATP site of the kinase. However, this large body of research has yet
to yield a successful p38 inhibitor as a marketed drug,5–7 as a result
of either significant toxic side effects or, more recently, a lack of
clinical efficacy.
itive ATP site binder, BIRB-796(2), and the Deciphera compound(3)
as control compounds (Table 1). Both SB 203580 and BIRB-796
showed inhibitory activity that reflected their reported literature
values of 57 nM and 22 nM, respectively, whereas the representa-
tive sulfonylurea analog from Deciphera exhibited a Ki value of
99 nM in our assay. By comparison, the amide based inhibitors
exhibited Ki values ranging from 86 nM to greater than 3 lM,
One of the first reported p38 inhibitors was Smith Kline Bee-
cham’s SB203580 (Fig. 1-1), whose characterization was first re-
ported in 1993 as a molecule that inhibited the phosphorylation
of HSP-27 by a protein kinase, termed at the time, ‘reactive kinase
(RK)’.1 In the next 10 years, over 10,000 published articles and pat-
ents highlighted the discovery and characterization of other p38
inhibitors that exclusively bind the ATP binding site. In 2002, the
report of Boehringer Ingelheim‘s allosteric inhibitor BIRB-796
(Fig. 1-2) was the first reported kinase inhibitor to extend outside
of the ATP binding site, 10 years removed from the characteriza-
tion of SB 203580 (Fig. 1-1).
depending on the amide substituent. Interestingly, according to
this data, the switch pocket seems to be relatively tolerant of var-
ious alkyl or aryl substituted amides. However, there is a clear
preference for more polar side chains which likely extend into a
solvent exposed region of the active site and even potentially form
specific interactions with the more hydrophilic portion of the pro-
tein. In contrast, larger and more extended hydrophobic com-
pounds tend bind much more weakly to p38, typically showing
Ki values above 1
In light of the fact that molecules that bind the DFG-in and DFG-
out conformations of p38 quite often exhibit very different kinet-
lM.
a
Interestingly, two years later, the first in a larger class of patent
applications from Deciphera Pharmaceuticals LLC (Lawrence, Kan-
sas)8 disclosed a novel class of molecules (Example Fig. 1-3) that
are believed to extend into a novel allosteric binding site of inac-
tive p38 known as the switch pocket. Although these molecules
bear a resemblance to the pyrazolyl urea inhibitors originally dis-
closed by Boehringer Ingelheim, they contain functionality that
purportedly extends into a previously unexplored region of the
kinase, as well as lack of a canonical hinge binding interaction.
The switch pocket is believed to extend into a region which is
adjacent to the ATP binding site but partially overlaps with the
well known ‘BIRB-site’ in the DFG-out conformation of the protein
( Fig. 2). Interacting with the switch pocket may have unique
biochemical consequences as well as present new opportunities
for selectivity across the kinome.
ics of binding, we wanted to further evaluate a subset of our
compounds using surface plasmon resonance.10 To this end, the
binding kinetics of three compounds with relatively diverse side
chains, were evaluated using SPR, along with SB 203580(1), a
competitive ATP site binder, and the Deciphera compound(3). This
data is summarized in Table 2.
SB203580, which binds to the canonical ATP binding site dem-
onstrates a 50–200-fold faster on-rate as compared to all of the
other compounds in this study. The observed slow on-rate for
BIRB-796 is a consequence of the local protein rearrangement
Table 2
Kinetic binding data generated using surface plasmon resonance for two previously
reported p38
Deciphera Switch pocket inhibitor(3) and 3 amide-based switch pocket inhibitors (7,
12, 13), against immobilized p38
a inhibitors, SB203580(1) and BIRB-796(2), as well as a representative
a
Even though public crystal structure coordinates have yet to be
made available for inhibitors of this type, the structural similarity
of these compounds to the original allosteric BIRB inhibitors indi-
cate a very similar binding mode. In addition, the recent issued
patents8b around the switch pocket, among other things, indicate
that Arginine 70 of helix 3 in the kinase domain plays a critical role
Surface plasmon resonance
TR-FRET
p38 Binding
Compounds
Kon (1/ms)
Koff (1/s)
KD (nM)
a
Ki (nM)
SB 203580 (1)
BIRB-796 (2)7
Deciphera (3)
Compound 7
Compound 12
Compound 13
2138 ꢀ 103
85 ꢀ 103
.0088
4.1
.098
564
14
10
55
57
9
1360
86
176
250
.0083 ꢀ 10ꢁ3
3.7 ꢀ 103
56.0 ꢀ 103
53.9 ꢀ 103
34.4 ꢀ 103
.002
.0007
.0005
.0019
in the affinity of these molecules for p38a. This further suggests
that these molecules bind to the inactive form of the kinase, and
extend into a region similar to BIRB-796. Thus, using the available
information in the patent literature,8 along with the landmark
crystal structure of BIRB-796(1KV2), we developed a working mod-
el (Fig. 3B) of the switch pocket. This model is depicted in Figure 3B
where the pocket is shaped largely by the DFG-loop (tan), the Gly-
cine-Rich- or P- Loop (purple) and helix 3 of the protein. In this
particular molecule, the cyclic sulfonylurea is in proximity to
ASP168 of the DFG-loop, as well as a proximal ARG70 on helix 3,
both of which may serve as hydrogen bonding partners for the sul-
fonylurea moiety.
From this, we designed a small library of compounds that would
further probe the switch pocket region of p38 (Fig. 4). Our analogs
were designed with an amide linker at the 30 position of the phenyl
pyrazole which could potentially maintain H-bonds to the Arg70 as
well as Asp168 of the DFG-loop, and would allow for rapid analog-
ing to explore SAR in this region of the active site. The general
synthesis of the proposed analogs is depicted in Scheme 1. Com-
mercially available 3-hydrazinylbenzoic acid (4) was treated with
pivaloyl cyanide in the presence of acetic acid in ethanol yielding
5 in good yields. The amino pyrazole 5 was then converted to the
urea using 1-napthylisocyanate preserving the acid as a handle
18
16
14
12
10
8
SB 203580
Compound 7
RU
6
4
2
0
-100 -50
0
50
100 150 200 250 300 350 400 450
-2
Time (s)
for further elaboration to
carboxamides.
a small library of substituted
Figure 5. SPR traces of SB203580 (red) and compound
immobilized p38 illustrating different binding kinetics for canonical hinge binders
as compared to a representative switch pocket inhibitor.
7 (blue) binding to
The focused library of compounds was assayed for their ability
a
bind p38
a
in a TR-FRET assay9 along with SB203580(1), a compet-

S. L. Swann et al. / Bioorg. Med. Chem. Lett. 20 (2010) 5787–5792
5791
Table 3
Profile of a subset of 17 switch pocket compounds against a panel of 50 kinases in a TR-FRET assay format
p38α
Compound 13
Compound 17
Compound 18
Compound 19
Compound 20
Compound 15
Deciphera (3)
Compound 9
Compound 7
Compound 11
Compound 8
Compound 14
Compound 16
Compound 10
SB203580 (1)
BIRB-796 (2)
Compound 12
<10nm
>10uM
Ki
Heat map colors are on a logarithmic scale ranging from red (Ki 610 nM) to green (Ki >10 lM).
required to adopt the DFG-out form of the kinase. Thus, the fact
that the Deciphera inhibitors (3) as well as compounds 7, 12 and
13 demonstrate slower on-rates is additional evidence that these
a few kinases in our assay. By contrast, the kinome profile of the
amide-based switch pocket inhibitors can vary dramatically based
on the amide side chain. For example, the most non-selective
inhibitors, namely compounds 8, 10, 14 and 16, each of which in-
compounds also bind the DFG-out form of p38a.
The SPR traces of SB203580 (red) and compound 7 (blue) are
shown in Figure 5 to illustrate how compounds with comparable
binding affinities, can have very different binding kinetics. One com-
pelling aspect of this data is the 40ꢀ faster on-rate of SB203580 for
the kinase, where the protein is completely saturated within 50 s
of compound injection. In contrast compound 7 takes twice as long
to completely saturate the protein, however, the >10ꢀ slower
off-rate is quite evident such that the inhibitor is never completely
displaced from the protein surface, even after 4 min.
Finally, a subset of our switch pocket inhibitors were then fur-
ther evaluated against a panel of 50 kinases, along with SB203580,
the Deciphera inhibitor (3) and BIRB-796, to further assess the
implications of binding to the switch pocket across a representa-
tive subset of the kinome. This data is shown in Table 3.
hibit at least 10 kinases with a Ki value less than 1
exclusively contain amides with basic amines. However, other
compounds that are much more selective for p38 (compounds
7, 9, 12, 15, 17, 18, 19) contain a wide variety of side chains, some
of which also include basic amines. Although there seems to be no
clear structure–activity relationship to explain the variation in
kinome profiles of these compounds, undoubtedly there are oppor-
tunities to achieve different selectivity profiles by exploiting this
portion of the kinase active site. This is an area of kinase research
yet to be explored in any depth.
In summary, we have identified and characterized novel mole-
cules that are believed to extend into a region of p38 known as the
switch pocket. Although these molecules lack a canonical hinge
binding motif, they still show Ki values as low as from 100 nM. It
is clear that molecules that interact with this region of the protein
demonstrate different binding kinetics than a canonical ATP
mimetic, as well as a wide range of kinome profiles. Thus the
switch pocket presents new opportunities for kinome selectivity
lM, seem to
a
As expected, the kinome profile of SB203580 and BIRB-796 are
different, a likely result of the preferential difference in protein
conformation. The Deciphera inhibitor seems to be a generally
weaker inhibitor, showing only modest inhibitory activity against

5792
S. L. Swann et al. / Bioorg. Med. Chem. Lett. 20 (2010) 5787–5792
8. (a) Flynn, D. L.; Petillo, P. A. WO2004/061084; (b) Flynn, D. L.; Petillo, P. A.
WO2007/008917; (c) Flynn, D. L., Petillo, P. A. WO 2008/0248548.
which could result in unique biochemical responses and offer new
opportunities as in the field of kinase drug discovery.
9. TR-FRET competition binding was performed using 2.5 nM p38 Alpha kinase
(full-length, GST-tagged protein; Invitrogen cat. no. PV3304), 100 nM tracer
199 (Alexa-647-labeled probe molecule that binds the ATP binding site at a Kd
of approximately 64 nM; Invitrogen cat. no. PV5830), 2 nM Europium-labeled-
anti-GST antibody (Invitrogen cat. no. PV5594), in a reaction buffer consisting
of 50 mM HEPES pH 7.5/10 mM MgCl 2/1 mM EGTA/0.01% Brij-35/0.01% BSA/
1 mM dTT. Compounds are tested at different ligand concentrations ranging
Acknowledgement
The authors would like to thank Dr. Stevan Djuric for helpful
insights and discussions during the course of this work.
from 0.0001 to 10
lM in 10ꢀ dilutions. Reaction mixtures are incubated for
2.5 h and fluorescence is then measured on a PerkinElmer Envision plate
reader. All calculations are performed on the ratio of the two emission values
(665 nm from the bound tracer 199‘, 616 nm from the Eu-anti-GST antibody).
Successful competition by the test compound versus the fluorescent probe for
the ATP binding site of the kinase is quantified by calculating the EC50 value
based on a 4 parameter logistic curve fit. A Ki value is then calculated using the
Cheng–Prusoff equation (Ki = (IC50)/(1 + [tracer 199]/Kd). Finally‘, results are
visualized in a heatmap which is generated the pKi values (pKi = 6 ꢁ log [Ki]).
10. Regan, J.; Pargellis, C. A.; Cirillo, P. F.; Gilmore, T.; Hickey, E. R.; Peet, G. W.;
Proto, A.; Swinamer, A.; Moss, N. Bioorg. Med. Chem. Lett. 2003, 13, 3101.
11. Reagents and conditions: (a) pivaloyl cyanide, A., EtOH, 81%; (b) 1-
isocyanatonaphthalene, pyridine, THF, 82%; (c) HATU, DIEA, DMA, amine.
References and notes
1. Ono, K.; Han, J. Cell. Signalling 2000, 12, 1.
2. Pargellis, C.; Tong, L.; Churchill, L.; Cirillo, P. F.; Gilmore, T.; Graham, A. G.; Grob,
P. M.; Hickey, E. R.; Moss, N.; Pav, S.; Regan, J. Nat. Struct. Biol. 2002, 9, 268.
3. Lindsley, C. W.; Barnett, S.; Yaroschak, M.; Bilodeau, M. T.; Layton, M. E. Curr.
Top. Med. Chem. 2007, 7, 1349.
4. Kumar, S.; Boehm, J.; Lee, J. C. Nat. Rev. Drug Disc. 2003, 2, 717.
5. Dominguez, C.; Powers, D. A.; Tamayo, N. Curr. Opin. Drug Disc. Dev. 2005, 8,
421.
6. Genovese, M. C. Arthritis Rheum. 2009, 60, 317.
7. Cohen, S. B.; Cheng, T. T.; Chindalore, V.; Damjanov, N.; Burgos-Vargas, R.;
Delora, P.; Zimany, K.; Travers, H.; Caulfield, J. P. Arthritis Rheum. 2009, 60, 335.