Biphenyl amide p38 kinase inhibitors 2: Optimisation and SAR

Article abstract of DOI:10.1016/j.bmcl.2007.10.043

The biphenyl amides are a novel series of p38 MAP kinase inhibitors. Structure-activity relationships of the series against p38α are discussed with reference to the X-ray crystal structure of an example. The series was optimised rapidly to a compound showing oral activity in an in vivo disease model.

Full text of DOI:10.1016/j.bmcl.2007.10.043

Available online at www.sciencedirect.com
Bioorganic & Medicinal Chemistry Letters 18 (2008) 324–328
Biphenyl amide p38 kinase inhibitors 2: Optimisation and SAR
Richard M. Angell, Tony D. Angell, Paul Bamborough,* David Brown,
Murray Brown, Jacky B. Buckton, Stuart G. Cockerill, Chris D. Edwards,
Katherine L. Jones, Tim Longstaff, Penny A. Smee, Kathryn J. Smith,
Don O. Somers, Ann L. Walker and Malcolm Willsonà
GlaxoSmithKline R&D, Medicines Research Centre, Gunnels Wood Road, Stevenage, Hertfordshire SG1 2NY, UK
Received 8 August 2007; revised 10 October 2007; accepted 13 October 2007
Available online 17 October 2007
Abstract—The biphenyl amides are a novel series of p38 MAP kinase inhibitors. Structure–activity relationships of the series against
p38a are discussed with reference to the X-ray crystal structure of an example. The series was optimised rapidly to a compound
showing oral activity in an in vivo disease model.
ꢀ 2007 Elsevier Ltd. All rights reserved.
The biphenyl amides (BPAs) are a novel series of p38a/b
MAP kinase inhibitors discovered through structure-
based focused screening. As described elsewhere, crystal-
lography was used to elucidate the binding mode of 1
and 3.1 This suggested that the part of the molecule that
could be most readily varied was the cyanophenyl amide
group, which lies in the outer lipophilic region of the
ATP-binding site and points out towards solvent
(Fig. 1). Array technology allowed a wide range of sub-
stituents to be introduced at this position using amide
coupling.
Me
NC
O
Figure 1. X-ray complex of p38a with 3, focusing on the amide
portion.
N
H
O
N
Me
3
N
Compounds in Table 1 were prepared by a variety of
similar routes. In the synthesis of 4 (Scheme 1), 3-bro-
mo-4-methylbenzoic acid A was activated using CDI
and coupled with tert-butylcarbazate to give the BOC-
protected amide B. The BOC-group was removed using
TFA to give the hydrazide C, which was refluxed with
triethyl orthoacetate to form the 1,3,4-oxadiazole D. A
Suzuki reaction with 4-carboxyphenylboronic acid un-
der standard conditions produced 16 in good yield.
The acid was converted to the acid chloride E using oxa-
lyl chloride and then stirred with p-anisidine to give 4.
Keywords: p38 kinase; p38a; p38; CSBP; MAP kinase; Inhibitors;
BPA, biphenyl amide; BPAs, biphenyl amides; Protein kinase X-ray
structure; Binding mode; Structure-based drug design; Kinase
selectivity.
*
Corresponding author. Tel.: +44 1438 763246; fax: +44 1438
763352; e-mail: paul.a.bamborough@gsk.com
Present address: Arrow Therapeutics, Britannia House, 7 Trinity
Street, London SE1 1DA, UK.
à
Selected compounds illustrating SAR trends will now
be discussed. Table 1 shows the p38a activity for a
Present address: Genomic Solutions, 1B Blackstone Road, Hunting-
don, Cambridgeshire PE29 6EF, UK.
0960-894X/$ - see front matter ꢀ 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2007.10.043

R. M. Angell et al. / Bioorg. Med. Chem. Lett. 18 (2008) 324–328
325
Table 1. Activity of selected phenyl and benzyl amide analogues
against p38a (lM)2
Apart from phenyl-substituted amides, the best enzyme
activity comes from compounds bearing benzyl amides
(compounds 8–13). As with phenyl amides, the substitu-
tion pattern does not greatly change the activity. Benzyl
amides are comparable to or better than the correspond-
ing phenyl amides (for example 4 and 11, 7 and 8). Even
bulky 4-substituents on the benzylic ring are well toler-
ated (13). These groups presumably pass beyond Val30
towards solvent.
Me
O
X
O
R
N
Me
N
Compound
X
R
IC50
10
Ki
1
NH
3-(4-Me piperazine),
4-OMe
1.6
Further substitution on the benzylic amide nitrogen with
N-methyl does not lead to a significant loss of potency
(compare 14 and 12). This is consistent with the environ-
ment of the amide NH in the X-ray structure of 3. No
protein atom or visible water lies close enough to pre-
vent substitution. However, N-substitution on phenyl
amides causes a greater loss of activity (compare 15
and 3). This can be attributed to intramolecular steric
interactions between the amide methyl group and both
of the adjacent phenyl rings, destabilising the bound
conformation.
2
3
NH
NH
3-(4-Me piperazine)
3-CN
4-OMe
3.1
1.5
2.3
1.6
0.49
0.24
0.36
0.25
4
NH
NH
5
2-OMe
6
NH
NH
3-CONHMe
4-NHSO2Me
4-NHSO2Me
4-H
0.60 0.10
0.28 0.04
0.28 0.04
0.61 0.10
0.37 0.06
0.52 0.08
0.60 0.10
0.87 0.14
0.89 0.14
7
8
NHCH2
NHCH2
NHCH2
NHCH2
NHCH2
NHCH2
9
10
11
12
13
14
15
3-OMe
4-OMe
4-Cl
3-(Methylmorpholine)
Various flexible groups were introduced at the amide
position, intended to project into solvent and to improve
solubility. In contrast to aromatic substituents, small or
flexible substituents have greatly reduced potency (Table
2, 17–20). The precursor acid 16 shows only very weak
activity, while the cyclopropylmethyl 17 is the most po-
tent alkyl amide. This reinforces the need for lipophilic
amide substituents to interact with the ‘outer lipophilic
N(Me)CH2 4-Cl
NMe 3-CN
12
1.9
variety of phenyl and benzyl amide substituents. The
data shown are from a binding assay measuring the
displacement of
ligand.2
a
fluorescent ATP-competitive
Table 2. p38a activity of selected amide analogues (lM)2
Removing the methoxy from the phenylpiperazine ring
of 1 to give 2 increases potency, perhaps because it al-
lows the phenylpiperazine rings to adopt a more pla-
nar conformation. The piperazine ring is not
required for p38a potency (3,4). The outer aromatic
ring makes lipophilic interactions with Val30, but is
accessible to solvent, so the expectation was that there
would be only small substituent effects on the potency.
Electron-withdrawing and donating groups, as well as
larger meta and para substituents, are tolerated with-
out loss of activity (3–7). Modelling from the X-ray
structure of 3 offers no explanation for the increased
affinity of 7.
Me
O
R
O
N
Me
N
Compound
R
IC50
Ki
16
17
18
19
20
OH
>16
3.0
7.1
>16
10
>2.5
0.48
1.1
NH(CH2)-cyclopropyl
NH(CH2)3OH
NH(CH2)2-morpholine
1-Piperidine
>2.5
1.6
Me
Me
Me
Br
Me
Br
O
a
b
c
H
H
OH
N
N
O
N
H
O
Br
NH2
Br
Me
O
O
O
N
N
A
B
C
D
d
Me
Me
Me
O
O
O
f
e
MeO
N
H
Cl
HO
O
O
O
N
N
N
4
E
16
Me
Me
Me
N
N
N
Scheme 1. Reagents and conditions: (a) CDI, tert-butylcarbazate, THF 48%; (b) TFA, 56%; (c) triethyl orthoacetate, 155 ꢁC, 86%;
(d) 4-carboxyphenylboronic acid, Pd(PPh3)4, 2 M sodium carbonate (aq), DME, 91%; (e) oxalyl chloride, DCM, DMF, 100%; (f) p-anisidine,
triethylamine, THF.

326
R. M. Angell et al. / Bioorg. Med. Chem. Lett. 18 (2008) 324–328
O
OH
pocket’ of the ATP site (Val30 or Ala111). Straight-
chain alkyl alcohols (e.g., 18) are tolerated, but with
lowered activity. The flexibility of this group may incur
a penalty due to loss of entropy on binding. Basic chains
including those containing tertiary amines suffer a great-
er loss of activity (e.g., 19). Cyclisation incorporating
the amide nitrogen is also unfavourable (e.g., 20).
OMe
O
OMe
Me
O
Br
q
r,s
t-v
+
26
NH2
Me
Me
B(OH)2
I
NH2
Scheme 3. Reagents and conditions: (q) Pd(PPh3)4, Cs2CO3, ethylene-
glycol–dimethylether, 90 ꢁC, 50%; (r) NaNO2, KI, 5 M HCl (aq),
0–60 ꢁC, 30%; (s) 2 M NaOH (aq), MeOH, 89%; (t) i—thionyl
chloride, reflux; ii—cyclopropylmethyl amine, Na2CO3, DCM, 41%;
(u) i—bis(pinacolato)diboron, KOAc, PdCl2(dppf), DMF, 80 ꢁC, 74%;
(v) i—2-bromo-1H-imidazole, Pd(PPh3)4, 2 M Na2CO3 (aq), DMF;
ii—PdCl2(dppf), 100 ꢁC, 5%.
Having explored SAR at the amide position, the next
changes concentrated on replacing the oxadiazole ring.
The activities of 21–29 are given in Table 3. The 1,3,4-
oxadiazoles (21–24) were synthesised as in Scheme 1,
using different orthoesters at step (c) to vary the substi-
tution in the 5-position. Synthesis of the 1,2,4-oxadiaz-
ole (25) was achieved by oxadiazole formation from
the acid followed by a Suzuki reaction to form the biaryl
ring system (Scheme 2). 26 was synthesised by forming
the pinacolatoborane on the biaryl core and reacting
with 2-bromo-1H-imidazole under Suzuki conditions
(Scheme 3). Reaction of 17 with benzylamine under
microwave conditions gave the benzyl protected 1,3,4-
triazole which was hydrogenated to 27 (Scheme 4).
Me
O
w
27
N
H
O
N
Me
N
Scheme 4. Reagents: (w) i—Benzylamine, NMP, microwave; ii—10%
Pd/C, formic acid, ethanol, 27%.
Table 3. p38a activity (lM) of selected modifications to the biphenyl-
1,3,4-oxadiazole2
Compound 28 was prepared by a similar route to
Scheme 1 using 3-bromobenzoic acid instead of A. 29
was prepared as shown in Scheme 5.
R2
O
N
H
R3
X
In the X-ray structure of 3 (Fig. 2), the methyl on the
oxadiazole (R1 in Table 3) fits snugly into a small lipo-
philic pocket close to Leu171. Replacing this methyl
with hydrogen (21) results in a loss in activity of at least
fivefold relative to 17. Groups such as ethyl, propyl and
butyl (22–24), which modelling suggests are too large to
fit in this pocket without clashing with Leu171, show re-
duced activity with increasing size.
Y
R1
Z
Compound
X
Y
Z
R1
R2
R3
IC50
Ki
17
21
22
23
24
25
26
27
28
29
O
O
O
O
O
N
N
N
O
O
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
O
C
Me
H
Me
Me
Me
Me
Me
Me
Me
Me
H
H
H
H
H
H
H
H
H
H
Me
3.0
>16
10
0.48
>2.5
1.6
Et
Pr
9.7
>16
14
1.5
The nitrogen atoms of the 1,3,4-oxadiazole are within
hydrogen-bonding distance of the backbone NH atoms
of Asp168 and Phe169 (Fig. 2). The oxadiazole oxygen
is close to the sidechain oxygen of Glu71, an unfavour-
able position for an electronegative atom. It was ex-
pected that exchanging other heterocycles for the
oxadiazole would change these interactions and hence
the potency. The 1,2,4-oxadiazole (25) is less potent than
17 by a factor of nearly 5. Two factors can explain this.
First, aromatic oxygen atoms are poor hydrogen-bond
acceptors. The alternative isomer would replace a strong
H-bond from aromatic nitrogen to Phe169 with a weak
one from aromatic oxygen. In addition, the 1,2,4-oxadi-
azole would place nitrogen, a more electronegative atom
than oxygen, close to the acid sidechain of Glu71.
Bu
Me
H
>2.5
2
11
1.7
N
N
N
Me
Me
Me
1.4
>16
>16
0.22
>2.5
>2.5
H
H
g
OH
N
O
+
(OH)2B
(OH)2B
O
25
Br
Br
h-k
l
Me
OH
O
N
N
Me
Me
O
Other heterocycles were prepared, for example the imid-
azole 26, but all are weaker inhibitors than the 1,3,4-
oxadiazole with the exception of the 1,3,4-triazole (27).
The triazole tautomer with the hydrogen at the N4 posi-
tion (Fig. 3) would be able to donate a hydrogen-bond
to Glu71 while retaining both H-bond acceptor func-
tionalities of the 1,3,4-oxadiazole. However, the twofold
Scheme 2. Reagents and conditions: (g) i—HBTU, HOBT, DIPEA,
DMF; ii—cyclopropyl methylamine, 40%; (h) i—oxalyl chloride,
DCM, DMF; ii—0.5 M NH3 in dioxane, 83%; (i) TFAA, pyridine,
DCM, 65%; (j) hydroxylamine hydrochloride, 80 ꢁC, 73%; (k) NaOMe
in MeOH, acetic anhydride, DMF, 50 ꢁC, 41%; (l) Pd(PPh3)4, Cs2CO3,
DMF, 90 ꢁC, 24%.

R. M. Angell et al. / Bioorg. Med. Chem. Lett. 18 (2008) 324–328
327
CN
CN
CN
O
HO
O
HN
HN
O
HN
O
x
y
a-c
Me
29
Me
Br
Me
Me
Br
OH
O
N
Me
N
O
Scheme 5. Reagents and conditions: (x) i—SOCl2; reflux 2 h; ii—3-aminobenzonitrile, triethylamine, THF, 76%; (y) 3-(dihydroxyboranyl)benzoic
acid, Pd(PPh3)4, 2 M sodium carbonate (aq), DME, 60%. Steps (a–c) are as in Scheme 1.
Compounds were assayed for their ability to inhibit the
production of TNFa in LPS-stimulated peripheral
blood mononuclear cells.4 1 did not reproducibly show
activity in this assay at the concentration range tested.
Compounds which were more potent in the enzyme as-
say showed much improved dose–response curves in
PBMCs. For example, 17 has an IC50 of 2.5 lM (mean
of 10 values, standard deviation 0.88).
In addition, 17 shows an excellent pharmacokinetic pro-
file in the rat, displaying good oral bioavailability, a low
plasma clearance, moderate volume of distribution and
a long plasma half-life (Table 4).5
17 was tested in a rat model of acute joint inflamma-
tion.6 The PG–PS (Streptococcal cell wall) reactivation
arthritis model has been shown to be particularly sensi-
Figure 2. X-ray complex of 3 highlighting the oxadiazole interactions.
tive to inhibitors of TNFa and IL-1, and is therefore a
suitable model for testing compounds which affect the
H
N
O
N
N
Ph
Ph
Ph
Ph
N
N
N
N
N
N
N N
H
H
Table 4. Pharmacokinetic parameters of 17 measured in rat5
N1-H
N2-H
N4-H
IV plasma clearance (ml/min/kg)
IV steady state volume of distribution (l/kg)
IV plasma terminal t1/2 (h)
<10
2
Figure 3. Oxadiazole vs triazole tautomers for a model system.
9–13
50%
gain in potency is modest, perhaps because in gas-phase
calculations this N4-H tautomer is significantly higher in
energy than N2-H.3
Oral bioavailability
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
Vehicle
The toluene methyl adjacent to the biaryl bond fills a
small pocket between Ala51 and Thr106.1 28 (R2 = H)
was made to investigate the importance of this and
was very weakly active. This shows the need either to fill
this small pocket, or for the biphenyl to adopt a perpen-
dicular arrangement, for activity. The low activity of 29,
where the methyl was moved onto the adjacent phenyl
ring, suggests that the former is the most important
factor.
Pred 4mg/kg
3.3mg/kg
10mg/kg
30mg/kg
Compounds from the BPA series were highly selective.
17 was tested in 38 other in vitro protein kinase assays,
including CDK2, GSK-3b, JNK isoforms 1–3, Lck,
ROCK1 and VEGFR2, while 10 was tested in 24 assays.
Both compounds bind to p38b with similar affinity to
p38a, with IC50 of 0.7 lM (Ki 0.16 lM) for 10 and
IC50 of 1.8 lM (Ki 0.44 lM) for 17. No other kinase
was inhibited with IC50 below 16 lM, so 10 shows at
least 40-fold selectivity in IC50 against all kinases tested.
0
1
2
3
Day
ED50 = 14.61 (9.52 - 28.06)
Figure 4. Dose-dependent inhibition of ankle swelling in the rat PGPS
model.6

328
R. M. Angell et al. / Bioorg. Med. Chem. Lett. 18 (2008) 324–328
modification of the Cheng–Prusoff equation was used
(Cheng, H. C. Pharmacol. Res. 2005, 50, 21–40).
Ki = IC50*(1+((n[E])/Kd)) where n is the fraction of enzyme
competent to bind fluoroligand. Ki values for 381 com-
pounds in the fluorescence polarisation assay correlate with
those from an activity assay (details to be published) with
r2 = 0.9.
synthesis and activity of pro-inflammatory cytokines.
The compound shows dose-dependent anti-inflamma-
tory activity with an ED50 of 15 mg/kg. Figure 4 shows
results for dosing 17 at 3.3, 10 and 30 mg/kg. Also
shown are results for the vehicle and prednisolone at
4 mg/kg. At 30 mg/kg, activity is comparable to that of
4 mg/kg prednisolone.
3. Gas-phase optimisation of the three phenyltriazole tautom-
ers in the model compound 3-methyl-5-phenyl-1H-1,2,4-
triazole was performed using B3LYP DFT with a 6-31g**+
basis set in JAGUAR v4.0; Schro}dinger Inc.: Portland, OR,
2000. Triazole tautomer N4-H would be capable of
donating the H-bond to Glu71, but its enthalpy of
formation is 7.0 kcal/mol less stable than tautomer N2-H.
Tautomer N1-H is only 0.4 kcal/mol less stable than N2-H.
4. Human peripheral blood mononuclear cells (PBMCs) were
prepared from heparinised human blood from normal
volunteers by centrifugation on hystopaque 1077 at 1000g
for 30 min. The cells were collected from the interface,
washed by centrifugation (1300g, 10 min) and re-suspended
in assay buffer (RPMI1640 containing 10% foetal calf
serum, 1% L-glutamine and 1% penicillin/streptomycin) at
1 · 106 cells/ml. Fifty microliter cells were added to micro-
titre wells containing 50 ll of an appropriately diluted
compound solution (prepared from 10 mM stock in DMSO
by serial dilution in assay buffer giving maximally 0.1%
DMSO final). After 10–15 min incubation, lipopolysaccha-
ride (s. typhosa, sigma) was added giving 1 ng/ml final
concentration. The assay plates were incubated at 37 ꢁC,
5% CO2 for 18–20 h and cell free supernatants collected
following centrifugation at 800g. The supernatant was then
assayed for TNFa using a commercially available ELISA
(Pharmingen).
The in vivo activity is very encouraging for this new tem-
plate, especially considering its modest enzyme potency.
Furthermore, given the selective kinase inhibition profile
of this compound, it is likely that the in vivo activity is
driven by inhibition of p38.
In summary, compound 17 shows significant progress
over compound 1. It is a potent, selective inhibitor of
p38a with cellular activity, oral bioavailability and
activity in an in vivo model of joint inflammation. Fu-
ture publications will describe the continuing develop-
ment of the biphenyl amide series.
Acknowledgment
The authors thank Rick Williamson for suggestions and
for contributing experimental details.
References and notes
1. Angell, R. M.; Bamborough, P.; Cleasby, A.; Cockerill, S.
G.; Jones, K. L.; Mooney, C. J.; Somers, D. O.; Walker, A.
L. Bioorg. Med. Chem. Lett. 2008, 18, 318.
5. Pharmacokinetic parameters in male Lewis rats were
determined following intravenous (iv) and oral (po) admin-
istration at 1 and 3 mg/kg, respectively. Compound was
administered as a solution in 20% DMSO: 80% PEG200
(iv) or 5% DMSO: 40% Vit E: 40% PEG400: 15% Mannitol
(po). Blood was collected over a 24-h time period. Plasma
was prepared following centrifugation and compound
extracted from 50 lL plasma using protein precipitation
with acetonitrile. Samples were then evaporated under
nitrogen and re-suspended in 100 ll of 10:90 acetoni-
trile:water. Analysis was performed using LC–MSMS on
the API365 with a 5 min fast gradient comprising 0.1%
formic acid in water and 0.1% formic acid in acetonitrile
(mobile phases), 20ll injection volume, flow rate 4 ml/min
and ODS3 Prodigy column (5 cm · 2.1 mm, 5 lm). Phar-
macokinetic data were generated using PKTools.
6. PG–PS (peptidoglycan–polysaccharide from Streptococcal
cell walls) was injected intra-articularly and followed two
weeks later by a systemic reactivation using the same
bacterial cell wall component. The reactivated joint
swelling was measured for 3 days after systemic challenge
while dosing orally, twice daily, with the target
compound.
2. (a) Fluorescence polarisation assays used GST-tagged p38a
(activated using MKK6 and re-purified) or GST-tagged
truncated JNK3 (residues 39–402) and an ATP-competitive
rhodamine-green labelled fluoroligand (2-(6-amino-3-imi-
no-3H-xanthen-9-yl)-5-{[({4-[4-(4-Cl-3-hydroxyphenyl)-
5-(4-pyridinyl)-1H-imidazol-2yl]phenyl}methyl)amino]car-
bonyl}benzoic acid). These components were dissolved in a
buffer of final composition 62.5 mM Hepes, pH 7.5,
1.25 mM CHAPS, 1 mM DTT, 12.5 mM MgCl2, with final
concentrations of 12 nM of p38a or 50 nM JNK3 and
5 nM fluoroligand. Thirty microliters of this mixture were
added to wells containing 1 ll of test compound (0.28 nM–
16.6 lM) and incubated for 30–60 min at room tempera-
ture. Fluorescence anisotropy was read in a Molecular
Devices Acquest (excitation 485 nm/emission 535 nm); (b)
The Cheng–Prusoff equation (Cheng, Y.-C.; Prusoff, W. H.
Biochem. Pharmacol. 1973, 22, 3099–3108), Ki =
IC50*(1+([S]/Km)), was used to calculate Ki from deter-
mined IC50 for activity assays. To calculate Ki from
determined IC50 for fluorescence polarisation assays a