Development of N-2,4-pyrimidine-N-phenyl-N′-alkyl ureas as orally active inhibitors of tumor necrosis factor alpha (TNF-α) synthesis. Part 2

Article abstract of DOI:10.1016/j.bmcl.2006.03.096

A new class of tumor necrosis factor alpha (TNF-α) synthesis inhibitors based on a N-2,4-pyrimidine-N-phenyl-N′-alkyl urea scaffold is described. Many of these compounds showed low-nanomolar activity against lipopolysaccharide stimulated TNF-α production.

Full text of DOI:10.1016/j.bmcl.2006.03.096

Bioorganic & Medicinal Chemistry Letters 16 (2006) 3514–3518
Development of N-2,4-pyrimidine-N-phenyl-N0-alkyl ureas as orally
active inhibitors of tumor necrosis factor
alpha (TNF-a) synthesis. Part 2
Jennifer A. Maier, Todd A. Brugel,* Michael P. Clark, Mark Sabat, Adam Golebiowski,
Roger G. Bookland, Matthew J. Laufersweiler, Steven K. Laughlin,
John C. VanRens, Biswanath De, Lily C. Hsieh, Kimberly K. Brown,
Karen Juergens, Richard L. Walter and Michael J. Janusz
Procter & Gamble Pharmaceuticals, Health Care Research Center, 8700 Mason-Montgomery Rd., Mason, OH 45040, USA
Received 10 March 2006; revised 27 March 2006; accepted 28 March 2006
Available online 24 April 2006
Abstract—A new class of tumor necrosis factor alpha (TNF-a) synthesis inhibitors based on a N-2,4-pyrimidine-N-phenyl-N0-alkyl
urea scaffold is described. Many of these compounds showed low-nanomolar activity against lipopolysaccharide stimulated TNF-a
production. Two analogs were tested in an in vivo rat iodoacetate model of osteoarthritis and shown to be orally efficacious. X-ray
co-crystallization studies with mutated p38a showed that these trisubstituted ureas interact with the ATP-binding pocket in a pseu-
do-bicyclic conformation brought about by the presence of an intramolecular hydrogen bonding interaction.
ꢀ 2006 Elsevier Ltd. All rights reserved.
The over-expression of cytokines, such as tumor necro-
sis factor alpha (TNF-a) and interleukin-1b (IL-1b), has
been implicated in a number of serious inflammatory
disorders.1 Consequently, agents that inhibit the prolif-
eration of these pro-inflammatory cytokines could re-
duce inflammation and prevent further tissue
destruction in diseases such as rheumatoid arthritis
(RA),2, osteoarthritis (OA) and Crohn’s disease.3 When
activated by external stimuli, the serine-threonine kinase
p38 mitogen-activated protein (MAP) kinase perpetu-
ates a phosphorylation cascade among downstream
kinases which activates transcription factors such as
NF-jb leading to ultimate synthesis of TNF-a and
IL-1b.4,5 Many companies have developed small mole-
cule TNF-a production inhibitors that function by com-
petitively binding to the ATP-binding pocket of p38.6
A recent review stated that since 1996, over 234 patents
have been published claiming p38 inhibitors, while
seventeen specific inhibitors have been selected for
development.7 A summary of the clinical development
of five recent candidates (AMG 548, BIRB 796, SCIO
469, SCIO 323, and VX 702) was recently published.8
To date, no company has reported results for a p38 spe-
cific kinase inhibitor past phase II development.
Previously we reported the synthesis and biological
activity of a series of bicyclic pyrazolone-based inhibi-
tors of TNF-a production exemplified by 1 (TNF-a
IC50 = 22 nM, Fig. 1) which was orally active in an
F
F
O
N
O
HN
Cl
N
N
N
N
N
N
NH
OH
NH
OH
1
2
TNF-α IC50 = 22 nM
p38α IC50 = 54 nM
TNF-α = 122 nM
Keywords: Cytokine synthesis inhibitors; p38a kinase; Urea.
Corresponding author. Tel.: +1 513 622 4498; fax: +1 513 622
3681; e-mail: brugel.ta@pg.com
*
Figure 1. Urea-based pro-inflammatory cytokine inhibitor.
0960-894X/$ - see front matter ꢀ 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2006.03.096

J. A. Maier et al. / Bioorg. Med. Chem. Lett. 16 (2006) 3514–3518
3515
Table 1. TNF-a data for urea analogs containing an N0-substituted
benzyl group
in vivo model for osteoarthritis.9 Pyrazolone 1 was also
a potent inhibitor of p38a (IC50 = 54 nM), while show-
ing a good overall profile of kinase selectivity. We fol-
lowed this work with an investigation of an N-2,4-
pyrimidine-N,N0-diphenyl urea scaffold highlighted by
the potent TNF-a synthesis inhibitor 2.10 However,
due to issues with chemical stability, we were unable
to progress compounds of this class any further into
in vivo studies. We now wish to report the development
of new trisubstituted urea analogs which maintain cyto-
kine inhibition, while also showing in vivo efficacy.
R
R2
N
N
HN
N
O
R1
Compound R
R1
R2
TNF-a
IC50 (lM)
a
HO HN
HO HN
HO HN
HO HN
Synthesis of many of the described pyrimidine-ureas was
accomplished in three steps from known starting materi-
als (see Scheme 1) as reported in the previous communi-
cation.10 Regioselective addition of 4-fluoroanilines to
2,4-dichloropyrimidine 3 gave the corresponding 6-anili-
no-4-chloropyrimidine 4. A second nucleophilic addi-
tion of (3S)-3-amino-2-methyl-butan-2-ol to 4 in N-
methyl pyrolidinone (NMP, 135 ꢁC) gave the desired
disubstituted pyrimidine 5. Reaction of 5 with 2-chlo-
robenzyl isocyanate in dichloroethane (DCE) gave the
corresponding trisubstituted ureas 6a.11
6a
6b
6c
6d
6e
6f
4-F
2-Cl
0.014
0.009
0.037
0.015
0.236
0.148
0.056
0.611
2.07
4-MeO
4-MeO
4-MeO
4-F
2-Cl
2-Me
2-MeO
2-Cl
MeO HN
MeO HN
MeO HN
MeO HN
MeO HN
The urea analogs synthesized in this study were all tested
for inhibition of TNF-a in an LPS stimulated human
monocytic (THP-1) whole cell-based assay.12
4-MeO
4-EtO
4-MeO
4-MeO
4-F
2-Cl
Table 1 describes the general SAR among the N0-benzyl-
substituted urea analogs. N0-2-Chlorobenzyl analog 6a
was tested in the cellular assay and showed excellent
potency (IC50 = 14 nM). This represents a near 10-fold
improvement in potency over the related N0-2-chloro-
phenyl analog 2 (IC50 = 122 nM) described previously.10
Additionally, unlike the corresponding N0-2-chlorophe-
nyl analog 2, the benzyl analog proved to be chemically
stable in solution with no decomposition observed after
24 h standing in aqueous buffers between pH 3 and 12.
Presumably, this is due to the presence of an electron-
donating substituent at the N0-position which deacti-
vates the urea toward hydrolysis relative to the N0-aryl
6g
6h
6i
2-Cl
2,4-di-Cl
3,4-di-Cl
2-Cl
H
N
6j
0.016
0.016
O
H
N
6k
4-MeO
2-Cl
N
N
HO
Cl
N
NH
F
N
H
N
NH
F
a
b
N
b
6l
Me2N
4-F
2-Cl
2-Cl
35%
Cl
N
Cl
3%b
3
6m
MeNH
4-MeO
4
5
F
H
N
6n
4-F
2-Cl
25%b
F
F
a Standard deviation for the assay was 30% of mean or less.
b Percentage inhibition at 10 lM.
c
N
N
N
O
NH
HN
analogs. Other N0-2-substituted benzyl analogs contain-
ing the hydroxyl-dimethylpropylamino 2-pyrimidyl
groups and an N-4-methoxyphenyl group (6b–6d) were
synthesized and proved to be nearly equipotent with 6a.
HO
Cl
6a
Scheme 1. Reagents and conditions: (a) EtOH, Na2CO3, 4-fluoroan-
iline; (b) NMP, i-Pr2NEt, (3S)-3-amino-2-methyl-butan-2-olÆHCl,
135 ꢁC; (c) DCE, 2-chlorobenzyl isocyanate, rt to 80 ꢁC.
As with the N0-phenyl analogs, an approximate 10-fold
loss in potency was observed with compounds in which

3516
J. A. Maier et al. / Bioorg. Med. Chem. Lett. 16 (2006) 3514–3518
Table 2. TNF-a data for urea analogs containing N0-alkyl substituents
the hydroxy-dimethylpropylamino group was replaced
with the methoxypropylamino group (6e and 6f). The
one exception in this series was the analog containing
an N-4-ethoxyphenyl group (6g) which proved to be
distinctly active with an IC50 of 56 nM. Two disubsti-
tuted benzyl derivatives were made with the 2,4-dichlo-
robenzyl analog 6h showing greater potency than the
3,4-dichlorobenzyl counterpart 6i, although both were
significantly less potent than the most active analogs
in the series. However, the 2-pyrimidyl tetrahydro-pyr-
anylamino analog 6j and isopropylamino analog 6k
proved to be the most active, non-chiral urea analogs
synthesized to this point. Conversely, other 2-pyrimidyl
derivatives such as the dimethylamino analog 6l,
the methylamino analog 6m, and the 2,6-difluoroani-
line derivative 6n showed negligible activity. This reiter-
ated the need for alkyl branching alpha to the amino
linker.
R
R2
N
N
HN
N
O
R1
R1
Compound
R
R2
TNF-a
IC50 (lM)
a
HO HN
7a
7b
7c
7d
7e
7f
MeO
MeO
MeO
MeO
EtO
F
Et
0.028
0.324
MeO HN
MeO HN
MeO HN
MeO HN
n-Bu
c-hex 0.467
Based on the success of adding an alkyl linker between
N0-urea nitrogen and the phenyl side chain, we decided
to also investigate purely N0-alkyl derivatives. Table 2
lists the results for a variety of simple N0-alkyl ureas.
The alkyl urea 7a proved to be as potent as the corre-
sponding benzyl analog 6b. Use of a longer chain or cyc-
lic alkyl (7b and 7c, respectively) resulted in compounds
with reduced potency. Likewise, switching from hydroxy
dimethylamino to methoxy propylamino also led to a
significant loss in potency as evidenced by the compari-
son between 7d and 7a. However, changing to a 4-eth-
oxyphenyl group (7e) resulted in recovery of activity.
The non-chiral amino analogs 7f and 7g containing
the ethyl urea were also found to maintain decent poten-
cy relative to the most active chiral amino derivatives.
Finally, the di-isopropyl amino analog 7h showed com-
parable activity to 7g, suggesting that some alkyl
branching off the N0-position was well tolerated.
Et
0.819
0.276
0.086
0.080
0.074
Et
H
N
Et
O
H
N
7g
MeO
Et
H
N
7h
F
i-Pr
a Standard deviation for the assay was 30% of mean or less.
For more structurally diverse ureas, a modified synthetic
protocol was developed (Scheme 2). For example, the
appropriate disubstituted pyrimidine 8 could be reacted
with p-nitrophenylchloroformate to give the activated
carbamate 9. Treatment of 9 with 4-aminomethylpyri-
dine gave urea 10a.
NO2
N
N
O
N
N
NH
F
H
N0-Pyrimidyl ureas 10a and 10b (Table 3) were highly
potent, with the 4-substituted analog being more ac-
tive than the corresponding 2-substituted derivative.
The ethyl-Boc-piperazine and methyl-Boc-piperidine
analogs 10c and 10d showed reduced activity. Howev-
er, removal of the Boc protecting group from these
compounds gave analogs 10e and 10f which showed
improved activity versus their protected precursors.
The methyl piperidine 10f (IC50 = 16 nM) was nearly
equipotent with the related unsaturated analog 10a
(IC50 = 4 nM).
N
N
N
F
O
a
H
8
9
NH
N
N
HN
b
N
N
O
Having achieved our goal for this series of attaining
excellent activity in the in vitro cell-based assay for inhi-
bition of TNF-a production, we turned our attention to
evaluating several analogs for their pharmacokinetic
properties as well as in vivo efficacy. The results of these
studies on three analogs are summarized in Table 4. The
F
10a
Scheme 2. Reagents and condition: (a) p-nitrophenylchloroformate,
pyridine, CH2Cl2, rt; (b) 4-aminomethylpyridine, CH2Cl2.

J. A. Maier et al. / Bioorg. Med. Chem. Lett. 16 (2006) 3514–3518
3517
Table 3. TNF-a data for urea analogs containing a 2-pyrimidyl
isopropylamino substituent
in an in vivo rat iodoacetate model (25 mg/kg BID dos-
ing) for osteoarthritis.19,21
An X-ray crystal structure was obtained and solved for
the co-crystal formed between the N0-2-chlorobenzyl
inhibitor 7b and mutated p38a to obtain additional
information concerning the possible mechanism for the
observed TNF-a inhibition for these compounds.15 This
urea analog made many of the necessary interactions for
classical p38 inhibitors (Fig. 2). Primary among these
was the hydrogen bonding motif between the 2-amino-
pyrimidine functionality of the inhibitor and the Met-
109 residue of the hinge region of the ATP-binding
pocket. Additionally, the 4-methoxyphenyl group posi-
tions itself into the well-defined hydrophobic pocket de-
fined by the ‘gatekeeper’ Thr-106 residue. As with the
previously described phenyl urea 2, the conformation
of the urea allows for the N0-urea NH to form an intra-
molecular hydrogen bond with N-3 of the pyrimidine
NH
R1
N
N
HN
N
O
F
a
Compound
R1
TNF-a IC50 (lM)
10a
0.004
0.030
0.277
1.16
N
N
10b
10c
10d
10e
10f
˚
ring (2.69 A N–N distance). This interaction created a
N
N
pseudo-bicyclic arrangement for this compound, which
adds rigidity to the class of inhibitors. Unlike N0-phenyl
urea 2, the N0-benzyl urea carbonyl group did form a di-
rect hydrogen bond to Lys-53. This mimics the interac-
tion observed for the previously described pyrazolone 1
(see molecular overlay in Fig. 2). The 2-chlorobenzyl
group was positioned toward the exposed solvent pocket
of the enzyme.
NBoc
NBoc
NH
•2 HCl
0.119
0.016
•HCl
NH
a Standard deviation for the assay was 30% of mean or less.
N0-chlorobenzyl analog 6b showed low in vitro metabo-
lism in plated rat hepatocytes, however the low solubil-
ity and poor bioavailability prohibited us from
advancing this compound further into in vivo efficacy
studies. N0-Ethyl analogs 7a showed low metabolism,
good solubility, and acceptable bioavailability. When
the chiral amino alcohol substituent in 7a was replaced
with the achiral amino tetrahydropyran group to give
7f, the resulting analog showed an overall reduction in
the described pharmacokinetic parameters. Both of
these analogs showed statistically significant reduction
of cartilage degradation severity compared to control
O
Lys053
Thr106
HN
Table 4. Pharmacokinetic and in vivo data for selected urea analogs
O
Compound Metabolisma Solubility T1/2
F
Cartilage
damage
reductionb (%)
O
His107
(%)
(mg/mL) (h) (%)
NH
HO
6b
7a
7f
32
32
16
<0.01
0.24
0.025
1.9
1.9
NDc
Leu108
NH
O
ND 15–30d 24
NH
H
Å
N
5
Å
0
8
.
8
O
.
0.46 13.2
16
N
2
2
O
N
a Metabolism was measured as percent loss of compound after 4 h
exposure to plated rat (Sprague–Dawley) hepatocytes.
b Reduction in severity of cartilage damage was measured versus
vehicle treated animals in a rat (male Sprague–Dawley) iodoacetate
(IA) model for osteoarthritis at a dose of 25 mg/kg BID (P < 0.05).
c Not determined.
N
N
H
N
H
Å
Met109
Cl
O
9
8
.
2
OH
Figure 2. Co-crystal X-ray structure of 6b (cyan) bound to mutated
p38 with overlay of pyrazolone 1 (magenta).
d Estimated bioavailability based on single po dose.

3518
J. A. Maier et al. / Bioorg. Med. Chem. Lett. 16 (2006) 3514–3518
Mekel, M. J.; Janusz, M. J.; Bioorg. Med. Chem. Lett.
2006, in press, doi:10.1016/j.bmcl.2006.03.095.
We have developed a new class of trisubstituted ureas as
inhibitors of TNF-a production. First generation N0-
phenyl ureas, although showing good potency for inhi-
bition of TNF-a, proved to be chemically unstable.
Through development of N0-benzyl and alkyl urea ana-
logs, high levels of cytokine inhibition were maintained,
while achieving good chemical and metabolic stability.
Several of these analogs showed good solubility and bio-
availability resulting in two analogs, 7a and 7f, showing
oral efficacy in a rat iodoacetate model for osteoarthri-
tis. X-ray crystallography studies with mutated p38
showed a mode of binding for this class of urea inhibi-
tors that mimics that of traditional vicinal bis-aryl
MAP kinase inhibitors. Selected analogs from this class
of TNF-a production inhibitors are being further devel-
oped and optimized for use as a potential treatment for
various inflammatory disorders.
11. All final compounds were characterized by 1H NMR,
HRMS, and HPLC analysis. Purity was assessed as >95%.
12. Duplicate cultures of human monocytic cells (THP-1)13
(2.0 · 105/well) were incubated for 15 min in the presence
or absence of various concentrations of inhibitor before
the stimulation of cytokine release by the addition of
lipopolysaccharide (final LPS concentration, 1 lg/mL).
The amount of TNF-a released was measured 4 h later
using an ELISA (R&D Systems, Minneapolis, MN). The
viability of the cells after the 4 h incubation was measured
using MTS assay14 (Promega Co., Madison, WI).
13. Mohler, K. M.; Sleath, P. R.; Fitzner, J. N.; Cerretti, D.
P.; Alderson, M.; Kerwar, S. S.; Torrance, D. S.; Otten-
Evans, C.; Greenstreet, T.; Weerawarna, K.; Kronhelm, S.
R.; Petersen, M.; Gerhart, M.; Kozlosky, C. J.; March, C.
J.; Black, R. A. Nature 1994, 370, 218.
14. Barltrop, J. A.; Owen, T. C.; Cory, A. H.; Cory, J. G.
Bioorg. Med. Chem. Lett. 1991, 1, 611.
15. The X-ray coordinates have been deposited with the
Protein Data Bank, www.rcsb.org, as entry 2GHM.
Mutated p38 cocrystallization/X-ray crystallography was
conducted as follows: the mutated p38a herein described
is a double mutant (S180A, Y182F) of murine p38a.16
The mutant enzyme cannot be phosphorylated and,
therefore, it is not competent for activation. Protein
expression and purification were carried out as previously
described for the murine enzyme.17 For crystallization,
mutated p38a was incubated overnight (12–16 h) with
1 mM compound. Cocrystals were grown by hanging
drop vapor diffusion using PEG as a precipitating agent
and overall protocols similar to those previously
described for the human enzyme.18 Crystals typically
Acknowledgments
We are grateful to A. L. Roe, C. A. Cruze, W. E.
Schwecke, and C. R. Dietsch for pharmacokinetic stud-
ies, M. Buchalova for chemical stability and solubility
studies, and M. Mekel for X-ray co-crystallization stud-
ies. We thank A. G. Evdokimov and M. E. Pokross for
cloning, expression, and purification of mp38 protein.
We would like to acknowledge that X-ray data were col-
lected at Southeast Regional Collaborative Access Team
(SER-CAT) 22-ID (or 22-BM) beamline at the Ad-
vanced Photon Source, Argonne National Laboratory.
Supporting institutions may be found at www.ser-
cat.org/members.html. Use of the Advanced Photon
Source was supported by the US Department of Energy,
Office of Science, Office of Basic Energy Sciences, under
Contract No. W-31-109-Eng-38.
˚
diffracted to 1.9 A resolution and were of the previously
˚
reported space group: P212121; a = 65.2 A, b = 74.6 A,
˚
0
˚
c = 78.1 A (4 ).
16. Han, J.; Lee, J. D.; Bibbs, L.; Ulevitch, R. J. Science 1994,
265, 808.
17. Wang, Z.; Harkins, P. C.; Ulevitch, R. J.; Han, J.; Cobb,
M. H.; Goldsmith, E. J. Proc. Natl. Acad. Sci. U.S.A.
1997, 94, 2327.
18. Pav, S.; Whit, D. M.; Rogers, S.; Crane, K. M.; Cywin, C.
L.; Davidson, W.; Hopkins, J.; Brown, M. L.; Pargellis, C.
A.; Tong, L. Protein Sci. 1997, 6, 242.
References and notes
1. For a recent review, see: Adams, J. L.; Badger, A. M.;
Kumar, S.; Lee, J. C. In Progress in Medicinal Chemistry;
King, F. D., Oxford, A. W., Eds.; Elsevier: Amsterdam,
2001; Vol. 39, pp 1–60.
2. Paleolog, E. Expert Opin. Investig. Drugs 2003, 12, 1087.
3. Baugh, J. A.; Bucala, R. Curr. Opin. Drug Discov. Devel.
2001, 4, 635.
4. Herlaar, E.; Brown, Z. Mol. Med. Today 1999, 5, 439.
5. Ono, K.; Han, J. Cell. Signalling 2000, 12, 1.
6. Bolos, J. Mini-Rev. Med. Chem. 2005, 5, 857.
7. Goldstein, D. J.; Gabriel, T. Curr. Top. Med. Chem. 2005,
5, 1017.
8. Lee, M. R.; Dominquez, C. Curr. Med. Chem. 2005, 12,
2979.
9. Clark, M. P.; Laughlin, S. K.; Laufersweiler, M. J.;
Bookland, R. G.; Brugel, T. A.; Golebiowski, A.; Sabat,
M.; Townes, J. A.; VanRens, J. C.; Djung, J. F.; Natchus,
M. G.; De, B.; Hsieh, L. C.; Xu, S. C.; Walter, R. L.; Mekel,
M. J.; Heitmeyer, S. A.; Brown, K. K.; Juergens, K.; Taiwo,
Y. O.; Janusz, M. J. J. Med. Chem. 2004, 47, 2724.
10. Brugel, T. A.; Maier, J. A.; Clark, M. P.; Sabat, M.;
Golebiowski, A.; Bookland, R. G.; Laufersweiler, M. J.,
Laughline, S. K., VanRens, J. C.; De, B.; Hsieh, L. C.;
19. Sprague–Dawley male rats (200–225 g) from Harlan
(Oregon, WI) under anesthesia were injected in the
patellar ligament region of the left leg (flexed 90ꢁ at the
knee) with 20 lL of a 10 mg/mL concentration of mono-
sodium iodoacetate (IA) (Aldrich Chemical, Milwaukee,
WI). Animals (groups of 15) were dosed for 7 days BID
(ꢀevery 12 h) with the potential inhibitor (25 mg/kg) or
Vehicle (2.5 mL/kg). Animals were sacrificed on day 22
and the left joint was disarticulated and fixed in 10%
formalin for 24–48 h prior to capturing the image. An
image of the tibial cartilage surface was captured using an
Optimas (Optimas, Media Cybernetics LP, Silver Springs,
MA) image analysis system. Three independent observers
assessed the damage in a blinded manner using a scale of
0–4 of increasing severity (0 = normal; 4 = maximum
severity).20
´
20. Guingamp, C.; Gegout-Pottie, P.; Philippe, L.; Terlain, B.;
Netter, P.; Gillet, P. Arthritis Rheum. 1997, 40, 1670.
21. Janusz, M. J.; Hookfin, E. B.; Heitmeyer, S. A.; Woessner, J.
F.; Freemont, A. J.; Hoyland, J. A.; Brown, K. K.; Hsieh, L.
C.; Almstead, N. G.; De, B.; Natchus, M. G.; Pikul, S.;
Taiwo, Y. O. Osteoarthritis Cartilage 2001, 9, 751.