Synthesis and SAR of p38α MAP kinase inhibitors based on heterobicyclic scaffolds

Article abstract of DOI:10.1016/j.bmcl.2007.07.029

The synthesis and structure-activity relationships (SAR) of p38α MAP kinase inhibitors based on heterobicyclic scaffolds are described. This effort led to the identification of compound (21) as a potent inhibitor of p38α MAP kinase with good cellular potency toward the inhibition of TNF-α production. X-ray co-crystallography of an oxalamide analog (24) bound to unphosphorylated p38α is also disclosed.

Full text of DOI:10.1016/j.bmcl.2007.07.029

Bioorganic & Medicinal Chemistry Letters 17 (2007) 5019–5024
Synthesis and SAR of p38a MAP kinase inhibitors based
on heterobicyclic scaffolds
T. G. Murali Dhar,* Stephen T. Wrobleski, Shuqun Lin, Joseph A. Furch,
David S. Nirschl, Yi Fan, Gordon Todderud, Sidney Pitt, Arthur M. Doweyko,
John S. Sack, Arvind Mathur, Murray McKinnon, Joel C. Barrish, John H. Dodd,
Gary L. Schieven and Katerina Leftheris
Bristol-Myers Squibb Pharmaceutical Research Institute, Princeton, NJ 08543-4000, USA
Received 22 May 2007; revised 9 July 2007; accepted 9 July 2007
Available online 21 July 2007
Abstract—The synthesis and structure–activity relationships (SAR) of p38a MAP kinase inhibitors based on heterobicyclic scaffolds
are described. This effort led to the identification of compound (21) as a potent inhibitor of p38a MAP kinase with good cellular
potency toward the inhibition of TNF-a production. X-ray co-crystallography of an oxalamide analog (24) bound to unphosphor-
ylated p38a is also disclosed.
Ó 2007 Elsevier Ltd. All rights reserved.
A significant body of the literature has underscored the
importance of proinflammatory cytokines TNF-a and
IL-1b as important mediators in the pathophysiology
of chronic inflammatory diseases like rheumatoid arthri-
tis, Crohn’s disease, and psoriasis.1 The regulatory ap-
proval and clinical success of anti-cytokine biological
agents anakinra (Kineret),2 an IL-1 receptor antagonist,
etanercept (Enbrel),3 a soluble TNF receptor fusion pro-
tein, infliximab (Remicade),4 and adalimumab (Humir-
a), both TNF-a monoclonal antibodies, has validated
the concept of inhibiting proinflammatory cytokines in
the treatment of chronic inflammatory diseases. This
has sparked significant effort in both academia and
industry to identify orally bioavailable small molecule
inhibitors of TNF-a and IL-1b roduction.5
appropriate target for anti-inflammatory therapy. Be-
cause of this, there has been an intense effort in the phar-
maceutical industry to identify safe and effective p38a
inhibitors that has resulted in the identification of sev-
eral clinical compounds.7 We recently disclosed a series
of 5-cyanopyrimidines and pyrrolo[2,1-f][1,2,4]triazines
as novel inhibitors of the p38a MAP kinase.8 This letter
describes the synthesis and preliminary structure–activ-
ity relationships (SAR) of p38a inhibitors based on a
heterobicyclic scaffold (A) as depicted in Figure 1.
Figure 1 outlines the rationale for the synthesis of hete-
robicyclic p38a inhibitors. Scios recently reported the
synthesis and structure–activity relationships (SAR) of
indole based heterocyclic inhibitors of p38a.9 The
authors initially identified compound (1) as a submi-
cromolar inhibitor of p38a. They hypothesized that
the introduction of methyl groups on the piperidine ring
of compound (1) would restrict conformational mobility
and potentially lead to analogs with increased potency.
Based on this rationale compound (2) was synthesized
and was found to be ꢀ14-fold more potent than com-
pound (1) as shown in Figure 1.
The mitogen-activated protein kinase (MAPK) p38 is
one of the most extensively studied kinase partly be-
cause of its important role as a key enzyme in the pro-
duction of proinflammatory cytokines such as TNF-a
and IL-1b.6 Of the four different isoforms of the p38
family (a, b, c, and d), p38a is ubiquitously expressed
and is generally considered the most important isoform
in the inflammatory signal transduction pathway and an
Further modifications to the indole scaffold eventually
led to Scios’s first p38 clinical compound, Scios-469.10
Keywords: p38a; MAP kinase inhibitors; TNFa; X-ray co-crystallography.
Corresponding author. Tel.: +1 609 252 4158; fax: +1 609 252
7410; e-mail: murali.dhar@bms.com
In our recent efforts focused on designing novel classes
of p38 inhibitors, we envisioned that incorporation of
*
0960-894X/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2007.07.029

5020
T. G. Murali Dhar et al. / Bioorg. Med. Chem. Lett. 17 (2007) 5019–5024
Insert methyl groups
on piperidine ring to
N
N
O
N
O
O
restrict conformational
O
mobility (Scios work) 9
F
O
OMe
N
N
SAR
N
N
H
N
O
N
Cl
NH
1
2 ( )
IC50 = 70 nM
Scios - 469
IC50 = 950 nM
p38α IC50 = 9 nM
Add ring constraint between
benzylic methylene and
piperizine ring to restrict
conformational
mobility (present work)
R'
R'
R'
R'
N
N
N
N
N
Y
X
O
N
O
N
Z
potential interaction
O
N
O
N
N
( )n
with Lys 53 when
Z = N
X
X
X
X
Examples
X, Y, Z = C, N
n = 1,2
N
N
N
R
R
N
R
p38α IC50 = ?
R
A1
A2
A3
A
Figure 1.
a ring constraint between the benzyl methylene position
and a proximal piperazine or piperidine ring as depicted
in compounds of type A 1–3 (Fig. 1) would also decrease
conformational mobility, and potentially lead to a novel
series of heterobicyclic p38a inhibitors. Molecular mod-
eling studies suggested that this type of ring constraint
may be advantageous for two main reasons: (1) to lock
the terminal aryl group in a conformation favorable for
occupying the deep hydrophobic pocket of p38a as de-
fined by amino acids Thr106, Lys53, Leu75, Leu86,
Leu104, and Val105; (2) to allow the possibility of pick-
ing up an additional favorable H-bond to Lys53 via a
nitrogen atom in compounds of type A 1–3 (when
Z = N). Based on this rationale we chose to synthesize
the pyrazolo[4,3-c]pyridine (A1), imidazo[1,5-a]pyrazine
(A2), and pyrido[3,4-d]pyrimidine (A3) for further eval-
uation. The synthesis of scaffolds A1, A2, and A3 is out-
lined in Schemes 1–3. Since the synthesis of the
imidazo[1,5-a]pyrazine core (A2) incorporates a hydro-
genation step, only fluorine substituents were employed
on the phenyl moiety that was designed to project into
the deep hydrophobic pocket.
R
O
O
R'
N
a
b
N
BocN
BocN
N
BocN
R
R
d
c
R'
N
N
R'
N
N
O
N
HN
R''
Scheme 1. Reagents and conditions: (a) tert-butoxy bis(dimethylamino)methane, 110 °C, 16 h (46%); (b) subst. or unsubst. phenylhydrazine, AcOH,
aq MeOH, rt, 1 h (ꢀ50–70%); (c) anhyd HCl, dioxane, 40 °C, 2 h (quantitative); (d) R0COCl, CH2Cl2, (iPr)2EtNH2, rt, 0.5 h.

T. G. Murali Dhar et al. / Bioorg. Med. Chem. Lett. 17 (2007) 5019–5024
5021
F
N
N
N
N
N
N
a, b
c
d, e
N
N
N
NH2
F
N
N
Cl
2HCl
N
N
F
F
g
f
F
F
N
N
N
N
HN
O
N
R
Scheme 2. Reagents and conditions: (a) NCS, CCl4, (C6H5CO)2O2, reflux 20 h; (b) hexamethylenetetramine (HMTA), toluene, 100 °C, 20 h (62% for
steps a, b); (c) EtOH, concd HCl, reflux 6 h, rt, 85 h; (d) 2,4-difluorobenzoyl chloride, (i-Pr)2EtNH2, dioxane, DMF, rt, 18 h (65%); (e) POCl3,
110 °C, 4 h (40%); (f) H2, 10% Pd–C, MeOH, 40 psi, 20 h (quantitative); (g) RCOCl, CH2Cl2, (i-Pr)2EtNH2or RCO2H, DMF, BOP, Et3N.
R
O
CO2Et
N
a
NH
b, c
BnN
BnN
BnN
O
N
N
•HCl
R
R
e
d
N
N
O
N
HN
N
N
R'
Scheme 3. Reagents and conditions: (a) formamidine acetate, NaOMe, MeOH, rt, 20 h (70%); (b) POCl3, (i-Pr)2EtNH2, toluene, 105 °C, 2 h (99%);
(c) subst. or unsubst. boronic acid or ester, 2 M aq K2CO3, Pd(PPh3)4, EtOH/toluene (ꢀ1:9), 110 °C, 10 h; (d) 1-chloroethylchloroformate, (i-
Pr)2EtNH2, CH2Cl2, 0 °C to rt, 20 h.
The indole carboxylic acids required to synthesize the
corresponding carboxamide analogs were either com-
mercially available or were prepared by methods known
in the literature.11
of the indole ring for optimal potency, the initial trends
with the pyrazolo[4,3-c]pyridine (Table 1) suggested that
incorporation of a substituent at the 6-position of the in-
dole ring did not significantly alter the potency in this
series (e.g., compare 11 with 12, Table 1). Second, C-3,
C-4, and C-6 substituted indole carboxamides (Table
1, analogs 7, 9, and 10, respectively) were nearly equipo-
tent in the p38a enzyme assay (40–55 nM). This result is
in contrast to the piperazine amides related to 2 where
the C-3, C-4, and C-6 substituted indole carboxamides
displayed decreased potencies (1.5–4.7 lM)9 in the en-
zyme assay. Only the C-2 substituted indole isomer 6
in the pyrazolo[4,3-c]pyridine series appeared signifi-
cantly less active (ꢀ4- to 5-fold) than the corresponding
C-3, C-4, and C-6 substituted indole carboxamide ana-
logs. These results suggest that incorporation of the ring
constraint as outlined in Figure 1 leads to analogs that
are more tolerant to amide substituent modifications rel-
ative to the piperazine amide series. This observation
may be due to a combination of conformational restric-
The preliminary SAR for the pyrazolo[4,3-c]pyridine
(A1) and imidazo[1,5-a]pyrazine (A2) are outlined in
Tables 1 and 2, respectively.
Although the initial compounds tested in this series were
moderately potent inhibitors of p38a,12 we were pleased
to see that many of the indole based analogs (7–13 Table 1)
achieved potencies similar to compound 2, thus support-
ing our aforementioned hypothesis of conformational
restriction of the piperidine ring via ring fusion
(Fig. 1). In addition, SAR trends that appear to diverge
from previously reported9 SARin the compound 2 series
were also observed. First, whereas analogs of 2 appear
to require a combination of the dimethyl groups on
the piperazine ring and a substituent on the 6-position

5022
T. G. Murali Dhar et al. / Bioorg. Med. Chem. Lett. 17 (2007) 5019–5024
Table 1. Pyrazolo[4,3-c]pyridine (A1) SAR
Table 2. Imidazo[1,5-a]pyrazine (A2) SAR
R'
F
F
R
N
N
N
N
O
N
O
N
R
R''
Compound
R
P38a C50 (nM)
Compound
R, R0
Cl, F
R00
P38a C50 (nM)
F
F
F
F
3
97
15
593
F
F
4
5
Cl, H
Cl, H
68
F
F
16
17
18
19
167
38
Cl
Cl
112
N
OMe
6
Cl, H
NH
NH
N
154
50
N
Cl
164
N
H
7
8
Cl, H
Cl, H
55
40
N
OMe
F
F
F
F
9
10
11
12
13
14
Cl, H
Cl, H
Cl, H
Cl, H
Cl, H
Cl, H
37
40
N
N
N
N
H
N
O
N
a, b
O
N
OMe
OMe
O
O
N
H
17
N
20
N
N
30
N
N
Scheme 4. Reagents and conditions: (a) (COCl)2, CH2Cl2, rt, 30 min;
(b) dimethylamine (2.0 M in THF), 0 °C (52% for steps a, b).
26
Cl
similar sequence was employed for the preparation of
the pyrazolo[4,3-c]pyridine 26 and27 and the pyri-
do[3,4-d]pyrimidine analog 28.
39
N
The SAR for the indole based pyrazolo[4,3-c]pyridines
(A1), imidazo[1,5-a]pyrazines (A2), and pyrido[3,4-
d]pyrimidines (A3) having an oxalamide moiety at the
C-3 position is outlined in Table 3 and Figure 2.
137
N
H
N
OMe
tion that forces the chlorophenyl moiety to occupy the
hydrophobic pocket in the desired orientation as well
as the potential additional H-bond interaction between
the pyrazole N-2 nitrogen and the Lys53 residue (vide
infra).
As is evident from Table 3 and Figure 2, incorporation
of the oxalamide side chain at the C-3 position of the in-
dole moiety improves the p38a enzyme potency in both
the pyrazolo[4,3-c]pyridine (A1) and imidazo[1,5-a]pyra-
zine (A2) series 3- to 10-fold (e.g., compare 14 in Table 1
with 26a in Figure 2, and 18 in Table 2 with 22 in Table
3). Consistent with the SAR outlined earlier for the pyr-
azolo[4,3-c]pyridine, the deschloro analog 27 is equipo-
tent to compound 26b (Fig. 2).
In order to further improve the potency within the pyr-
azolo[4,3-c]pyridine (A1) and imidazo[1,5-a]pyrazine
(A2) series, we decided to investigate the SAR of indole
based analogs which incorporate an oxalamide side
chain at the C-3 position due to the recent disclosure
of related oxalamides as potent inhibitors of p38a.10
Installation of the oxalamide side chain was undertaken
as described in the conversion of 17–20 (Scheme 4). A
To understand the binding mode of the pyrazolo[4,3-
c]pyridine (A1) and imidazo[1,5-a]pyrazine (A2) class
of p38a inhibitors we performed X-ray co-crystallo-

T. G. Murali Dhar et al. / Bioorg. Med. Chem. Lett. 17 (2007) 5019–5024
Table 3. Imidazo[1,5-a]pyrazine (A2) SAR with oxalamide side chain
5023
F
F
N
N
O
N
X
R
O
O
R''
N
R'
Compound
X
R
R0
R00
P38a C50 (nM)
20
CH
OMe
Me
H
13
N
21
CH
CH
N
Cl
11
14
29
13
16
N
N
N
Figure 3. Binding interactions between 24 and unphosphorylated p38
˚
22
Cl
Me
H
a based on X-ray crystallographic analysis (1.7 A resolution). Hydro-
gen bond distances are given in angstroms with key protein residues
labeled.
23
OMe
OMe
OMe
gatekeeper residue Thr106 (not shown) and the carbon
chain of Lys53. The inner walls of the cavity are made
up of Leu75, Leu86, Leu104, and Val 105. The methoxy
moiety sits in a shallow cavity formed by Ile84 and
Ala157. The terminal oxalamide appears to extend into
solvent and does not engage in any specific interaction
with the protein. However, a close inspection of the
binding site in contact with the oxalamide group reveals
that, although there are no specific H-bonds between
protein and the oxalamide moiety, the group fits into a
shallow hydrophobic groove on the surface made up
of P-loop Val30 and hinge region Leu108/Gly110. Effec-
tively, this interaction results in changing the exposed
surface of the protein–ligand complex (in the contact re-
gion) from hydrophobic to hydrophilic. The end result is
a modest hydrophobic effect wherein binding affinity
may be enhanced by the removal of ordered solvent
water molecules at the lip of the binding site.
N
N
24( )
25( )
N
H
HO
HO
N
Me
graphic studies of compound 24 (Table 3) with purified,
unphosphorylated p38a. The key binding interactions
between compound 24 and the p38a enzyme are illus-
trated in Figure 3.13
The X-ray structure of 24 complexed with p38a reveals a
combination of H-bonding and hydrophobic interac-
tions. The hinge region residues, Met109 and Gly110,
form H-bonds between their backbone NHs and the
central amide carbonyl of the inhibitor. The exposed
imidazo nitrogen of the inhibitor engages in a H-bond
to a water molecule which is itself H-bonded to Lys53
and Asp168. The pendant 2,4-difluorophenyl group is
located in a deep hydrophobic cavity flanked by the
Although incorporation of the oxalamide side chain
contributed to a 3- to 10-fold boost in potency in the
pyrazolo[4,3-c]pyridine (A1) and imidazo[1,5-a]pyrazine
F
F
Cl
N
Cl
N
N
N
N
O
N
O
N
X
O
N
N
Cl
R
O
O
O
O
O
O
N
N
N
N
N
NH
R'
27, p38αIC50 = 37 nM
28, p38α IC50 = 195 nM
X = N, R = OMe, R' = H
26a, p38α IC50 = 29 nM
X = CH, R = Cl, R' = Me
26b, p38α IC50 = 33 nM
Figure 2. Pyrazolo[4,3-c]pyridine (A1) and pyrido[3,4-d]pyrimidine (A3) SAR with oxalamide side chain.

5024
T. G. Murali Dhar et al. / Bioorg. Med. Chem. Lett. 17 (2007) 5019–5024
Gabriel, T. Curr. Top. Med. Chem. 2005, 5, 1017; (c)
Diller, D. J.; Lin, T. H.; Metzger, A. Curr. Top. Med.
Chem. 2005, 5, 953; (d) Dominguez, C.; Powers, D. A.;
Tamayo, N. Curr. Opin. Drug Discov. Dev. 2005, 8, 421;
(e) Wagner, G.; Laufer, S. Med. Res. Rev. 2006, 26, 1.
6. (a) Adams, J. L.; Badger, A. M.; Kumar, S.; Lee, J. C.
Prog. Med. Chem. 2001, 38, 1; (b) Lee, J. C.; Laydon, J. T.;
McDonnell, P. C.; Gallagher, T. F.; Kumar, S.; Green, D.;
McNulty, D.; Blumenthal, M. J.; Heyes, J. R. Nature
1994, 372, 739.
(A2) series, the corresponding pyrido[3,4-d]pyrimidine
analog 28 was significantly less potent (Fig. 2). In light
of the aforementioned X-ray co-crystal structure for
the imidazo[1,5-a]pyrazine analog 24, this result was
surprising since all key interactions of compound 28
with the enzyme appear to be in place for optimal po-
tency. A plausible explanation could be that the H-
bonding trajectory of either the imidazo or pyrazolo
nitrogens toward the available water molecule held by
Asp168/Lys53 is not reproduced when the nitrogen is
in a pyrimidine ring. In the former case, the plane of
interaction is in line with the water (OH) and is about
1.7 A in length (based on the X-ray structure). When
the pyrimidine analog was modeled, the resulting trajec-
tory is significantly off-plane and the distance increases
˚
to 2.1 A. Thus, the pyrimidine core offers a less efficient
H-bonding opportunity to this water molecule.
7. (a) Goldstein, D. M.; Gabriel, T. Curr. Top. Med. Chem.
2005, 5, 1017; (b) Dominguez, C.; Powers, D. A.; Tamayo,
N. Curr. Opin. Drug Discov. Dev. 2005, 8, 421.
˚
8. (a) Liu, C.; Wrobleski, S. T.; Lin, J.; Ahmed, G.; Metzger,
A.; Wityak, J.; Gillooly, K. M.; Shuster, D. J.; McIntyre,
K. W.; Pitt, S.; Shen, D. R.; Zhang, R. F.; Zhang, H.;
Doweyko, A. M.; Diller, D.; Henderson, I.; Barrish, J. C.;
Dodd, J. H.; Schieven, G. L.; Leftheris, K. J. Med. Chem.
2005, 48, 6261; (b) Hynes, J., Jr.; Dyckman, A. D.; Lin, S.;
Wrobleski, S. T.; Wu, H.; Gillooly, K. M.; Lonial, H.;
Loo, D.; McIntyre, K. W.; Pitt, S.; Shen, D. R.; Shuster,
D. J.; Zhang, X.; Behnia, K.; Marathe, P. H.; Doweyko,
A.; Barrish, J.; Dodd, J.; Schieven, G.; Leftheris, K. J.
Med. Chem, submitted for publication.
9. Mavunkel, B. J.; Chakravarty, S.; Perumattam, J. J.;
Luedtke, G. R.; Liang, X.; Lim, D.; Xu, Y.-J.; Laney, M.;
Liu, D. Y.; Schreiner, G. F.; Lewicki, J. A.; Dugar, S.
Biorg. Med. Chem. Lett. 2003, 13, 3087.
10. Dugar, S. 7th World Congress on Inflammation, Mel-
bourne, Australia., August 20–24, 2005.
11. Dugar, S. WO2004/032874.
In conclusion, we have identified novel series of pyrazol-
o[4,3-c]pyridine (A1), imidazo[1,5-a]pyrazine (A2) and
pyrido[3,4-d]pyrimidine (A3)-based p38a inhibitors.
Incorporation of an oxalamide moiety at the C-3 posi-
tion of the indole significantly improved the enzyme po-
tency in these series. One of the compounds was a potent
inhibitor of TNF-a release (compound 21, Table 3,
IC50 = 460 nM) in human peripheral blood mononu-
clear cells (PBMCs).14
12. p38 Enzyme assay: the assays were performed in U-
bottom 384-well plates. The final assay volume was 30 ll
prepared from 15 ll additions of enzyme and substrates
(fluoresceinated peptide FL-IPTSPITTTYFFFKKK-OH
and ATP) and test compounds in assay buffer (100 mM
Hepes, pH 7.2, 10 mM MgCl2, 0.015% Brij35, and 4 mM
DTT). The reaction was initiated by the combination of
bacterially expressed, activated p38 with substrates and
test compounds. The reaction mixture was incubated at
room temperature for 60 min and terminated by adding
30 ll of 35 mM EDTA to each sample. The reaction
mixture was analyzed on the Caliper LabChip 3000 by
electrophoretic separation of the fluorescent substrate and
phosphorylated product. Inhibition data were calculated
by comparison to no enzyme control reactions for 100%
inhibition and vehicle-only reactions for 0% inhibition.
The final concentrations of reagents in the assays are ATP,
20 lM; FL-IPTSPITTTYFFFKKK-OH, 1.5 lM; p38,
6 nM; and DMSO, 0.6%.
Acknowledgments
The authors thank their colleagues Susan E. Kiefer, John
A. Newitt, and Kevin Kish for their efforts in expressing
and purifying p38a protein and producing p38a co-crys-
tals with compound 24 that were used in this study.
References and notes
1. Wagner, G.; Laufer, S. Med. Res. Rev. 2006, 26, 1.
2. (a) Kavanaugh, A. Adv. Ther. 2006, 23, 208; (b) Fleisch-
mann, R. Expert Rev. Clin. Immunol. 2006, 2, 331.
3. Grunke, M.; Kalden, J. R. Expert Rev. Clin. Immunol.
2005, 1, 313.
4. (a) Saleem, B.; Mackie, S.; Emery, P. Expert Rev. Clin.
Immunol. 2006, 2, 193; (b) Cottone, M.; Mocciaro, F.;
Modesto, I. Expert Opin. Biol. Ther. 2006, 6, 401.
5. For reviews of this area see: (a) Hynes, J., Jr.; Leftheris, K.
Curr. Top. Med. Chem. 2005, 5, 967; (b) Goldstein, D. M.;
13. PDB deposition number is 2QD9.
14. Dyckman, A.; Hynes, J.; Leftheris, K.; Liu, C.; Wrobleski,
S. T. WO 090912, 2003, Chem. Abstr. 2003, 139, 292275.