Structure-based design and subsequent optimization of 2-tolyl-(1,2,3-triazol-1-yl-4-carboxamide) inhibitors of p38 MAP kinase

Article abstract of DOI:10.1016/j.bmcl.2008.04.043

A computer-aided drug design strategy leads to the identification of a new class of p38 inhibitors based on the 2-tolyl-(1,2,3-triazol-1-yl-4-carboxamide) scaffold. The tolyl triazole amides provided a potent platform amenable to optimization. Further exp

Full text of DOI:10.1016/j.bmcl.2008.04.043

Available online at www.sciencedirect.com
Bioorganic & Medicinal Chemistry Letters 18 (2008) 3251–3255
Structure-based design and subsequent optimization
of 2-tolyl-(1,2,3-triazol-1-yl-4-carboxamide) inhibitors
of p38 MAP kinase
D. A. Cogan,a,* R. Aungst,d E. C. Breinlinger,d, T. Fadra,a D. R. Goldberg,a M. H. Hao,a
R. Kroe,c N. Moss,a C. Pargellis,b K. C. Qiana and A. D. Swinamera
aDepartment of Medicinal Chemistry, Boehringer Ingelheim Pharmaceuticals, Inc., Research and Development Center,
900 Ridgebury Road, PO Box 368, Ridgefield, CT 06877, USA
bDepartment of Immunology and Inflammation, Boehringer Ingelheim Pharmaceuticals, Inc., Research and Development Center,
900 Ridgebury Road, Ridgefield, CT 06877, USA
cDepartment of Biologics and Biomolecular Sciences, Boehringer Ingelheim Pharmaceuticals, Inc.,
Research and Development Center, 900 Ridgebury Road, Ridgefield, CT 06877, USA
dMedicinal Chemistry, AMRI, 30 Corporate Circle, Albany, NY 12203, USA
Received 22 February 2008; revised 15 April 2008; accepted 21 April 2008
Available online 25 April 2008
Abstract—A computer-aided drug design strategy leads to the identification of a new class of p38 inhibitors based on the 2-tolyl-
(1,2,3-triazol-1-yl-4-carboxamide) scaffold. The tolyl triazole amides provided a potent platform amenable to optimization. Further
exploration leads to compounds with greater than 100-fold improvement in binding affinity to p38. Derivatives prepared to alter the
physicochemical properties produced inhibitors with IC50’s in human whole blood as low as 83 nM.
ꢀ 2008 Elsevier Ltd. All rights reserved.
The mitogen-activated protein (MAP) kinase p38 is a
key regulator in the signaling pathways controlling the
production of pro-inflammatory cytokines such as
TNF-a and IL-1b.1 Therefore, the identification of p38
inhibitors has been the focus of intense research.2 This
effort has been bolstered by the recent clinical and com-
mercial success of anti-cytokine biologics such as En-
brelꢁ, Remicadeꢁ, Humiraꢁ, and Kineretꢁ for the
treatment of various autoimmune diseases.3 It is antici-
pated that small molecule inhibitors of p38 will provide
at least a similar therapeutic benefit as anti-cytokine
biologics, but with the convenience of oral dosage.
with Met109 in the ATP binding site and hydrophobic
contacts in the adjacent kinase specificity pocket, this
class of compounds takes advantage of additional bind-
ing sites made available by the movement of Phe169.
Thus, the tert-butyl pyrazole replaces Phe169 in the
‘Phe pocket’ while the urea takes part in a hydrogen
bonding network with the backbone NH of Asp168
Met109
Met109
O
Asp168
Asp168
N
N
We have previously described naphthyl urea-based
inhibitors, such as the clinical compound BIRB 796, that
bind to what we call the ‘Phe-out’ conformation of p38.4
In addition to the conventional H-bonding interaction
O
O
O
O
H
N
O
N
S
S
N
H
N
N
O
N
H
N
H
H
O
Glu71
Glu71
Keywords: p38 kinase; p38a; p38; p38 MAPK; MAP kinase; Inhibitors;
Triazoles; Structure-based drug design.
BIRB 796
1
*
Corresponding author. Tel.: +1 203 791 6187; fax: +1 203 791
6072; e-mail: derek.cogan@boehringer-ingelheim.com
Present address: Abbott Bioresearch Center, 381 Plantation St.,
Worcester, MA 01605, USA.
Figure 1. Representative inhibitors highlighting key interactions with
p38. The dotted arc represents the Phe-pocket, and the solid arc
represents the kinase specificity pocket.
0960-894X/$ - see front matter ꢀ 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2008.04.043

3252
D. A. Cogan et al. / Bioorg. Med. Chem. Lett. 18 (2008) 3251–3255
and the carboxylate of Glu71 (Fig. 1).5 By taking
advantage of these interactions, the naphthyl urea class
of inhibitors can achieve KD’s below 1 nM.
The success of BIRB 796 established inhibition of the
Phe-out conformation as a viable path to the clinic.
We were, therefore, interested in further capitalizing
on the potency achievable with Phe-out binders, but also
wished to explore structurally distinct scaffolds. To this
end, we employed a computer aided design strategy that
successfully conceived a new class of p38 inhibitors,
exemplified by 1 (Fig. 1).6 Herein, we describe how we
have designed and optimized another novel and potent
series of Phe-out p38 inhibitors based on the 2-tolyl-
(1,2,3-triazol-1-yl-4-carboxamide) scaffold.
Figure 3. Predicted binding conformation of the designed inhibitor
class highlighting key interactions with p38.
Our strategy used LigBuilder6b software to link two seed
structures set in a model of the p38 active site (Fig. 2).
We chose a tert-butyl phenyl fragment known to bind
in the Phe pocket as one seed.7 To this fragment we ap-
pended an N-formyl group to access the hydrogen
bonds with Glu71 and Asp168 commonly associated
with inhibitors that bind in the Phe-out conformation.
Whereas a pyridyl fragment was chosen as the second
seed in the design of 1, we chose a carbonyl fragment
to access the hydrogen bond with Met109. The software
then used a stochastic approach to attach fragments that
would link the formyl group to the carbonyl, filling the
kinase specificity pocket along the way.
We chose 1,2,3-triazole-4-carboxamides as our specific
inhibitor class due to their synthetic accessibility. This al-
lowed us to quickly test the potency of our design, and
additionally provided a synthetic handle for further
SAR exploration. The synthesis of these inhibitors is out-
lined in Scheme 1. Diazotization of 2-amino-3-methyl
benzoic acid (2) and subsequent displacement with so-
dium azide provided 3. Amide formation with N-(3-ami-
no-5-tert-butyl-2-methoxy-phenyl)-methanesulfonamide6
provided 3-azido-amide 4. Huisgen triazole formation
with ethyl propiolate provided the triazole ester 5. Ester
5 could also be prepared by submitting the azide 3 to
the Huisgen conditions to provide ester 6, followed by
amide formation with N-(3-amino-5-tert-butyl-2-meth-
oxy-phenyl)-methane-sulfonamide. Saponification of 5
provided triazole carboxylic acid 7. Standard amide
coupling procedures provided amides 8–28.
A 4-tolyl group situated in the kinase specificity pocket
emerged from the LigandBuilder dataset as a recurring
and exploitable motif. However, the dataset provided a
variety of impractical flexible groups to connect the tolyl
group to the carbonyl. We modeled options to constrain
these flexible linkages, and predicted that N-linked azoles
would be suitable. The five-membered ring would provide
the necessary rigidity and display the carbonyl at the
proper distance to interact with Met109. Modeling also
indicated that connecting to the tolyl group via an N-link-
age, as opposed to a C-linkage, would favor the desired
out-of-plane conformation. This conformation would
also be reinforced by the methyl of the tolyl core (Fig.
3). We determined previously that the presence of the
methanesulfonamide of the left-hand phenyl imparted
both an improvement in potency and solubility,7 and thus
added this group to our test series.
Binding affinity to p38 was determined by a thermal
denaturation assay.8 Compounds were additionally
O
ii
X
O
O O
S
HO
N3
N
N
H
H
O
2 X = NH2
3 X = N3
i
4
iii
iii
O
O
OEt
R
O
Met109
O
O O
S
Met109
N
N
N
N
ii
N
HO
N
N
N
Asp168
O
Asp168
O
N
H
H
O
O
6
5 R = OEt
7 R = OH
iv
N
v, vi,
or vii
N
H
H
O
O
8 - 28 R = NR1R2
Glu71
Glu71
Scheme 1. Reagents and conditions: (i) NaNO2, 2 M HCl, NaN3
(88%). (ii) a—(COCl)2, cat. DMF; b—N-(3-amino-5-tert-butyl-2-
methoxy-phenyl)-methanesulfonamide, Et3N or 2,6-lutidine (95–
98%). (iii) Ethyl propiolate, DMA, 90 ꢂC (5: 62%; 6: 48%). (iv) NaOH,
MeOH (95%). (v) EDC, HOBt, HNR1R2 (44–62%) (vi) HATU,
HOAT, iPr2NEt, HNR1R2 (50–51%) (vii) a—(COCl)2, cat. DMF. b—
HNR1R2, Et3N or 2,6-lutidine (77%).
First design (ref 6)
Present design
Figure 2. Schematics of the de novo design processes. The dotted arcs
represent the Phe-pocket, and the solid arcs represent the kinase
specificity pocket.

D. A. Cogan et al. / Bioorg. Med. Chem. Lett. 18 (2008) 3251–3255
3253
screened for inhibition of TNF-a production from
LPS-stimulated THP-1 cells and human whole
blood.4 We were delighted to discover that the proto-
typical compound, ester 5 (Table 1), demonstrated
that the design can produce potent inhibitors of
p38 (a Tm of 56.2 ꢂC corresponds to a KD of approx-
imately 8 nM).8 The amides 8–14 demonstrated that a
variety of substitutions are tolerated, and that specific
substitution can lead to an increase in molecular po-
tency.8 Compounds 11 and 12 are noteworthy as
examples with greater potency in whole blood than
BIRB 796.
We obtained a crystal structure of compound 11 bound
to p38.9 As expected, compound 11 bound to the Phe-
out conformation with the Phe-out seed accessing the
requisite hydrophobic and hydrogen-bonding interac-
tions. In addition, the tolyl group fits tightly into the ki-
nase specificity pocket and the right-side amide carbonyl
engages Met109 in a hydrogen bond as we had pre-
Table 1. Molecular and cellular potencies of tolyl triazole p38 inhibitors
O
R
O
O O
S
N
N
N
N
N
H
H
O
a
b
THP-1 IC50 (nM)
c
Human whole blood IC50 (nM)
Compounds
R
Tm (ꢂC)
BIRB 796
5
63.5
56.2
18 2.1
>5000
780
9000 1700
OEt
8
57.3
110 (n = 1)
110 (n = 1)
84 (n = 1)
40 35
8400 1500
2800 830
>15,000
N
H
9
56.6
N
H
10
11
12
13
14
15
16
17
18
19
20
56.8
N
H
N
H
59.7
930 220
260 120
260 84
60.5
14 4.2
N
H
N
H
56.2
14 1.5
N
57.8
29 (n = 1)
27 6.7
>15,000
60.1
2600 1300
470 170
2600 1,000
1,000 630
5600 2700
2800 2500
N
H
64.6
13 1.2
N
H
60.0
35 (n = 1)
14 1.4
N
H
61.6 0.1
56.9
N
H
66 (n = 1)
16 0.71
N
H
N
H
60.8 0.0
a Thermal denaturation. See Ref. 7. Data are from a single experiment unless a standard deviation is provided.
b Inhibition of TNFa production in LPS-challenged THP-1 cells.
c Inhibition of TNFa production in LPS-challenged human whole blood; average of at least three values from at least three individual donors.

3254
D. A. Cogan et al. / Bioorg. Med. Chem. Lett. 18 (2008) 3251–3255
dicted. Moreover, the overall conformation of the crys-
tal structure is similar to the modeled conformation.
Interestingly, the benzyl group also formed an in-
duced-fit interaction with a hydrophobic pocket formed
from the side-chains of Tyr35, Leu167, and Met109
(Fig. 4), providing a rationale for the increase in potency
observed for some of the larger lipophilic amides. Simi-
lar interactions have been reported for ATP-site only
inhibitors of p38.10 The NH of the benzyl amide does
not play a direct role in binding to p38, consistent with
the modest loss on binding potency observed for 14
when compared to 11.
The crystal structure of 11 suggested that substitution at
the a-position of the benzyl group should be tolerated
and might provide a beneficial conformational con-
straint. Compounds 15–20 highlight the effect of a-
methyl substitution (Table 1). The R-enantiomer consis-
tently demonstrated an increase in Tm, although this
benefit did not consistently translate to improved po-
tency in cellular assays (e.g. 15 vs. 16).
By exploring the amide region of the inhibitors, we had
discovered a number of potent compounds with reason-
able potencies in cell assays. However, these compounds
suffered from poor aqueous solubility (typically <1 lg/
mL). Although the crystal structure of 11 indicated that
the amide substituents occupy a largely lipophilic re-
gion, due to the proximity of the solvent interface we felt
that polar substitution may also be tolerated in this re-
gion. Therefore, an additional set of amides with polar
functionality were prepared and were also found to be
potent p38 inhibitors (Table 2). One interesting example
Figure 4. Crystal structure of 11 bound to p38.
Table 2. Molecular and cellular potencies for tolyl triazole p38 inhibitors
O
R
O
O O
S
N
N
N
N
N
H
H
O
a
b
THP-1 IC50 (nM)
c
Human whole blood IC50 (nM)
Compounds
R
Tm (ꢂC)
N
N
N
H
21
57.9
38 11
270 110
720 220
500 220
210 90
N
H
22
58.8
65.1
59.9
56.1
57.4
57.5
74 14
N
23
35 (n = 1)
18 (n = 1)
140 (n = 1)
28 (n = 1)
200 (n = 1)
22 2.1
NH
OH
24
N
H
N
25
790 680
300 180
4500 2100
83 52
N
H
2611
2711
28
N
N
H
N
H
N
N
57.3 1.9
N
H
a,b,c,See footnotes for Table 1.

D. A. Cogan et al. / Bioorg. Med. Chem. Lett. 18 (2008) 3251–3255
3255
Table 3. Relationship between physicochemical properties and potency in human whole blood for select compounds
Compound
Tm (ꢂC)
Human whole
blood IC50 (nM)
Solubility in pH 7.4
buffer (lg/mL)
Caco-2 Papp. · 10À6 (cm/s)
Plasma protein
binding (%)
BIRB796
63.5
60.5
59.9
57.3
780
260
210
83
1.0
bqla
2.0
10
b
99
92
66
66
12
24
28
7.4
13
13.9
a Below quantifiable limit: 0.1 lg/ml.
b Insufficient solubility for assay.
S.; Boehm, J.; Lee, J. C. Nature Rev. Drug Discov. 2003, 2,
717; (d) Ono, K.; Han, J. Cell. Signal. 2000, 12, 1.
2. (a) Dominguez, C.; Powers, D. A.; Tamayo, N. Curr.
Opin. Drug Disc. Dev. 2005, 8, 421; (b) Nikas, S. N.;
Drosos, A. A. Curr. Opin. Invest. Drugs 2004, 5, 1205; (c)
Cirillo, P. F.; Pargellis, C.; Regan, J. Curr. Topics Med.
Chem. 2002, 2, 1021.
is compound 28, whose IC50 in whole blood is 83 nM de-
spite having a Tm 3 ꢂC lower than 12.
Efficacy in the human whole blood assay can be attrib-
uted to a variety of factors, including plasma protein
binding (affinity and kinetics), and membrane perme-
ability, that should be dependant upon a compound’s
physicochemical properties. Table 3 indicates how solu-
bility, permeability, and plasma protein binding relate to
whole blood potency for select compounds. While it ap-
pears that permeability is not a determining factor for
these compounds, the extent of protein binding does
have an impact. As a result, compound 28 is almost
10-fold more potent in whole blood than BIRB 796 de-
spite having a Tm more than 6 ꢂC lower.
3. (a) Klinkhoff, A. Drugs 2004, 64, 1267; (b) Barry, J.;
Kirby, B. Expert Opin. Biol. Ther. 2004, 4, 975.
4. (a) Regan, J.; Breitfelder, S.; Cirillo, P.; Gilmore, T.;
Graham, A. G.; Hickey, E. R.; Klaus, B.; Madwed, J.;
Moriak, M.; Moss, N.; Pargellis, C. A.; Pav, S.; Proto, A.;
Swinamer, A.; Tong, L.; Torcellini, C. J. Med. Chem.
2002, 45, 2994; (b) Regan, J.; Capolino, A.; Cirillo, P. F.;
Gilmore, T.; Graham, A. G.; Hickey, E.; Kroe, R. R.;
Madwed, J.; Moriak, M.; Nelson, R.; Pargellis, C. A.;
Swinamer, A.; Torcellini, C.; Tsang, M.; Moss, N. J. Med.
Chem. 2003, 46, 4676.
It is worth noting that the compounds from this series
typically demonstrate a high level of kinase selectivity,
even among kinases that the naphthyl ureas typically in-
hibit. For example, whereas BIRB 796 has an IC50 for
Jnk2a of 6 nM, the IC50 for compound 28 is 1.1 lM.
In addition, 28 does not demonstrate activity against
the following kinases at 3 lM: MKK1, ERK2, JNK1,
p38c, p38d, MAPKAP-K1a, MAPKAP-K2, MSK1,
PRAK, PKA, PKCa, PDK1, PKBdPH, SGK, S6K1,
GSK3b, ROCK-II, AMPK, CHK1, CK2, PHOS. KI-
NASE, CDK2/cyclin A, CK1, DYRK1A, PP2A, and
NEK6, and the following tested at 10 lM: Btk, Eck,
EGFR, FGFR3, Hek, HGFR, IGF1R, IR, Itk, JAK3,
Lyn, Syk, TXK, VEGFR1.
5. Pargellis, C. A.; Tong, L.; Churchill, L.; Cirillo, P.;
Gilmore, T.; Graham, A. G.; Grob, P. M.; Hickey, E.
R.; Moss, N.; Pav, S.; Regan, J. Nat. Struct. Biol. 2002,
9, 268.
6. (a) Goldberg, D.; Hao, M.-H.; Qian, K. C.; Swinamer, A.
D.; Gao, D. A.; Xiong, Z.; Sarko, C.; Berry, A.; Lord, J.;
Magolda, R. L.; Fadra, T.; Kroe, R. R.; Kukulka, A.;
Madwed, J. B.; Martin, L.; Pargellis, C.; Skow, D.; Song,
J. J.; Tan, Z.; Torcellini, C. A.; Zimmitti, C. S.; Yee, N.
K.; Moss, N. J. Med. Chem. 2007, 50, 4016; (b) Wang, R.;
Gao, Y.; Lai, L. J. Mol. Model 2000, 6, 498.
7. SAR at this region will be disclosed in future publications.
8. A 2.8 ꢂC change in thermal denaturation temperature
corresponds to approximately a 10-fold change in binding
affinity. However, it must be noted that this relationship
was developed for pyrazole naphthyl urea-based inhibitors
(e.g., BIRB 796), and if the thermodynamics of binding for
this class of inhibitors is significantly different, this
relationship may be an approximation. Kroe, R. R.;
Regan, J.; Proto, A.; Peet, G. W.; Roy, T.; Dickert, L.;
Fuschetto, N.; Pargellis, C. A.; Ingraham, R. H. J. Med.
Chem. 2003, 46, 4669.
Our de novo design strategy has provided access to an
additional structurally distinct class of p38 inhibitors.
This work highlights an example of using computer-
aided drug design to generate ideas, and molecular mod-
eling to refine these ideas into viable lead structures. The
design provided potent p38 inhibitors that were amena-
ble to analogue synthesis, allowing the further improve-
ment in physicochemical properties and potency in
human whole blood.
9. The coordinates have been deposited in the RSCB protein
databank (RCSB ID code rcsb047198 and PDB ID code
3CTQ).
10. Fitzgerald, C. E.; Patel, S. B.; Becker, J. W.; Cameron, P.
M.; Zaller, D.; Pikounis, V. B.; O’Keefe, S. J.; Scapin, G.
Nature: Struct. Biol. 2003, 10, 764.
References and notes
11. Compounds 26 and 27 were prepared from their N-Boc
piperidine analogues via Boc-removal (HCl/dioxane; 66%
yield) and reductive amination (formaldehyde, 1% HOAc
in methanol, NaCNBH3, 25% yield).
1. (a) Wagner, G.; Laufer, S. Med. Res. Rev. 2006, 26, 1; (b)
Saklatvala, J. Curr. Opin. Pharm. 2004, 4, 372; (c) Kumar,