Identification of benzoxazole analogs as novel, S1P3 sparing S1P1 agonists

Article abstract of DOI:10.1016/j.bmcl.2012.04.095

A novel series of benzoxazole-derived S1P₁ agonists were designed based on scaffold hopping molecular design strategy combined with computational approaches. Extensive SAR studies led to the discovery of compound 17d as a selective S1P₁ agonist (over S1P₃) with high CNS penetration and favorable DMPK properties. 17d also demonstrated in vivo pharmacological efficacy to reduce blood lymphocyte in mice after oral administration.

Full text of DOI:10.1016/j.bmcl.2012.04.095

Bioorganic & Medicinal Chemistry Letters 22 (2012) 3973–3977
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Identification of benzoxazole analogs as novel, S1P3 sparing S1P1 agonists
Guanghui Deng, Qinghua Meng, Qian Liu, Xuesong Xu, Qiongfeng Xu, Feng Ren, Taylor B. Guo,
⇑
Hongtao Lu, Jia-Ning Xiang, John D. Elliott, Xichen Lin
Research and Development, GlaxoSmithKline Pharmaceuticals, 898 Halei Road, Zhangjiang Hi-tech Park, Pudong, Shanghai 201023, China
a r t i c l e i n f o
a b s t r a c t
Article history:
A novel series of benzoxazole-derived S1P1 agonists were designed based on scaffold hopping molecular
design strategy combined with computational approaches. Extensive SAR studies led to the discovery of
compound 17d as a selective S1P1 agonist (over S1P3) with high CNS penetration and favorable DMPK
properties. 17d also demonstrated in vivo pharmacological efficacy to reduce blood lymphocyte in mice
after oral administration.
Received 28 March 2012
Revised 19 April 2012
Accepted 20 April 2012
Available online 28 April 2012
Ó 2012 Elsevier Ltd. All rights reserved.
Keywords:
Sphingosine-1-phosphate (S1P)
Agonist
Lymphopenia
Multiple sclerosis
Modulation of biological actions of sphingosine-1-phosphate
receptor-1 (S1P1), a member of the G protein coupled receptor
(GPCR) superfamily, has emerged as a new paradigm for the dis-
covery of therapeutic agents in autoimmune diseases.1 Recently,
a novel S1P1 agonist, FTY720 (Fig. 1), was approved as the first oral
drug for the treatment of relapsing/remitting multiple sclerosis
(RRMS).2,3 The pharmacological efficacy of FTY720 was believed
to be due to its agonistic activity at S1P1 when phosphorylated in
vivo. S1P1 agonism could redirect the trafficking of peripheral T
cells and B cells from systemic circulation into secondary lymphoid
organs, thus reduce lymphocyte infiltration into CNS resulting in
the observed efficacy in preclinical animal models for MS.4 The pri-
mary adverse effect observed in clinical trials of FTY720 is tran-
sient and asymptomatic bradycardia. Its S1P3 agonism (FTY720-P
S1P3 EC50 ꢀ3 nM and S1P1 EC50 ꢀ0.3 nM)2 was thought to be
responsible for the hemodynamic and pulmonary side effects seen
in clinical trials.5–7 Hence S1P1 agonists with selectivity over S1P3
are preferred for development.
plified by 1, Fig. 2).1,12 Compound 1 showed excellent S1P1 potency
(pEC50 = 11, Tango assay13) and good selectivity over S1P3 (106-
14
fold). The 3-chlorinated indole derivative 2 demonstrated similar
potency and selectivity (S1P1 pEC50 = 10.6; S1P3 pEC50 <5), which
suggests that small substitutions at the 3-position of the indole
ring are tolerated. In order to further simplify the structure, we
hypothesize the cyclization of indole with oxazole would maintain
similar conformation and provide less unnecessary carbons (Fig. 3).
Since alkyl amine in 4 is known to have very different physico-
chemical properties compared with the indole nitrogen in 3, we
decided to synthesize and examine the potency of the none-amine
5a first which is also more synthetically feasible.
Our first task is to develop a convergent synthetic route target-
ing compound 5a. As outlined in Scheme 1, starting from commer-
cial available material 6, acidic hydrolysis followed by selective
iodination and reduction provided aniline 9. Amide bond forma-
tion between aniline 9 and acid chloride 11 (prepared from acid
10) gave amide 12. Intermediate 12 was then subjected to the
cyclization condition to give aza-benzoxazole 13. Treatment of
13 with zinc reagent in the presence of suitable Pd catalyst resulted
in the coupling product 14. The final target molecules 5a, 5c–f
were synthesized through ester hydrolysis. Compound 5b was ob-
tained from 13 via lithiation followed by quenching with dry ice.
With the synthetic approach successfully developed, we inves-
tigated the SAR on the carboxylic side chain. The first compound 5a
demonstrated reasonable potency (pEC50 6.8) in the S1P1 Tango as-
say13 without S1P3 activity (Table 1). Extension or reduction one
carbon of the carboxylic acid side chain (5e, 5f) maintained compa-
rable S1P1 potency and selectivity over S1P3. However, if the side
chain was further reduced (5b, 5c, and 5d) the enzymatic potency
Various efforts have been followed as S1P1 agonists with limited
pharmacophores, including CS-0777,8 ACT-128800,9 AMG 369,10 or
PF-99111 ( Fig. 1). Herein, we reported the discovery of a novel
benzoxazole series as selective S1P1 agonists over S1P3. The series
demonstrated good overall pharmacokinetic properties, high CNS
penetration property, and good in vivo efficacy in the mouse lym-
phopenia study.
In our previous study, we had discovered a series of indole-
1,2,4-oxadiazole compounds as orally active S1P1 agonists (exem-
⇑
Corresponding author. Tel.: +86 21 6159 0718; fax: +86 21 6159 0730.
E-mail address: xichen.2.lin@gsk.com (X. Lin).
0960-894X/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmcl.2012.04.095

3974
G. Deng et al. / Bioorg. Med. Chem. Lett. 22 (2012) 3973–3977
NH2
OH
HO
N
NH2
O
OH
FTY720
CS-0777
HO
O
N
Cl
OH
N
N
S
O
OH
N
S
N
F
O
AMG369
ACT-128800
O
N
N
H
N
OH
O
PF-991
Figure 1. Structure of known S1P1 agonists.
was decreased dramatically (pEC50 <5.5). Given the poor yield
(<10%) leading to compound 5a in the Negishi coupling step, com-
pound 5f was then selected as a starting point for further optimi-
zation due to its synthetic easiness.
Cl
1O
N
Cl
3'
N2
O
OH
N1'
Starting from 5f, we turned our attention to optimize the left
hand side of the molecule in order to improve S1P1 potency (Table
2). The chloro-pyridine analogue (15a) proved to be least potent.
Replacing the chloride with a more lipophilic trifluoromethyl
group resulted in a much improved potency (15b). Its pyridine
analogue demonstrated a slightly improved S1P1 activity (15c).
The most potent compound (15d) was obtained by the replace-
ment of the isopropyl group with a phenyl group, with a subn-
anomolar S1P1 agonistic activity. The potency decreased
drastically when one of the phenyl group was replaced by its bio-
isoster, thiophene (15e). It was noteworthy that all compounds
were selective S1P1 agonists against S1P3.
O
2
Cl
X
O
N
N
O
OH
N
O
3
With the left & right hand molecule optimized, we then focused
on optimizing the core. The aza-benzoxazole core proved to be la-
bile in basic conditions (cleaved oxazole byproduct was observed
in the ester hydrolysis). In order to improve its stability, we then
tried to replace the core with other bicyclic heteroaromatic rings,
while maintaining S1P1 potency. As shown in Table 3, all the bicy-
clic cores (16a–e) resulted in 10- to 1000-fold loss of potency ex-
cept the benzoxazole derivative 15f (pEC50 9.4). Compound 16f
demonstrated high S1P1 potency comparable to that of the aza-
benzoxazole lead compound (15d) and good selectivity over
S1P3. In addition, the benzoxazole core was observed to be more
stable than the aza-benzoxazole structure under basic condition.
With the successful optimization of left hand side (LHS) and core,
Cl
Cl
N
O
N
O
H
N
OH
OH
4
O
N
O
N
O
5a
O
Figure 3. Design of aza-benzoxazole compound 5a.
Cl
Cl
O
O
N
N
OH
O
N
OH
O
N
N
Cl
O
N
O
2 S1P1 pEC50 = 10.6
1
S1P1 pEC50 = 11
Figure 2. The S1P1 agonizing activity of compounds 1 and 2.

G. Deng et al. / Bioorg. Med. Chem. Lett. 22 (2012) 3973–3977
3975
H2N
HO
I
O2N
HO
O2N
I
O2N
Cl
c
b
a
N
N
N
8
HO
N
9
7
6
Cl
Cl
Cl
O
e
O
d
H
N
O
I
Cl
OH
O
HO
12
N
N
O
O
N
10
11
Cl
f
Cl
O
Cl
O
O
N
N
i
h
O
I
O
O
n
n
13
N
O
N
OH
O
n=1-5
14a-e
O
n = 1-5
5a, 5c-f
g
Cl
O
N
O
COOH
N
5b
Scheme 1. Synthesis of compound 5a and its analogues. Reagents and conditions (a) concd HCl, 112 °C, 20 h, 100%; (b) I2, H5IO6, AcOH, H2SO4, H2O, 90 °C , 20 h, 95%; (c) Fe,
HCl, MeOH, reflux, 1 h, 85%; (d) (COCl)2, DMF, DCM, rt, 2 h; (e)TEA, DCM, rt, overnight; (f) Ph3P, C2Cl6, TEA, DCM, rt, 50% of d, e and f; (g) nBuLi, dry ice, À78 °C, 26.5%; (h)
Pd2(dba)3, Cs2CO3, (t-Bu)3P-HBF4, zinc reagent, THF, 50 °C; (i) NaOH, iPrOH, H2O, rt, 4–35% of h and i.
Table 1
S1P1 and S1P3 agonizing activity of compounds 5a–f
Cl
Table 2
O
S1P1 and S1P3 agonizing activity of compounds 15a–e
O
O
N
OH
OH
O
R
N
n
N
O
N
a
b
No.
n
S1P1 pEC50
S1P3 pEC50
N
5a
5b
5c
5d
5e
5f
4
0
1
2
3
5
6.8
<5.5
<5.5
<5.5
6.0
<5
<5
<5
<5
<5
<5
No.
R
S1P1 pEC50
S1P3 pEC50
Cl
O
O
O
15a
6.0
7.7
8.2
9.5
<5
6.8
The Tango assay13 for S1P1.
a
CF3
CF3
S1P3 Ca2+ mobilization in hS1P3 assay (in 96-well microplates).14
b
15b
15c
15d
<5
<5
<5
the subsequent SAR study was thus focused back at the right hand
side (RHS) of 16f while fixing the LHS and the benzoxazole core.
To optimize RHS of 16f, structure-based design was utilized.
Without S1P1 crystal structure at the time of the research, homol-
ogy modeling was utilized.15 The proposed binding mode of 16f is
shown in Figure 4.
N
CF
3
The binding mode suggests that the carboxylic acid group of 16f
occupy the region of the lipid membrane close to the extracellular
loops and form ionic interaction with Arg120/Arg292 while the
remaining ligand structure extend toward the center of the mem-
brane. This binding mode allows the charged carboxylic acid moi-
ety located in the shallow and less hydrophobic region of the
pocket while the more hydrophobic aromatic ring system
CF3
15e
6.4
<5
S

3976
G. Deng et al. / Bioorg. Med. Chem. Lett. 22 (2012) 3973–3977
Table 3
Table 4
S1P1 and S1P3 agonizing activity of core replacement compounds 16a–f
S1P1 and S1P3 activity of compounds 17a–n
CF3
CF3
O
OH
N
O
Core
RHS
a
No.
RHS
S1P1 pEC50
S1P3 pEC50
6.6
No.
Core
S1P1 pEC50
8.0
S1P3 pEC50
<5.0
O
O
H
N
N
17a
10.5
9.5
O
16a
S
N
N
17b
17c
6.0
6.7
N
N
N
O
16b
16c
16d
6.4
6.9
6.9
<5.0
<5.0
<5.0
H
N
OH
10.5
9.2
O
N
N
N
N
O
HN
17d
<5.5
O
N
S
O
O
16e
16f
7.2
9.4
<5.0
<5.0
17e
17f
8.9
9.0
5.9
5.7
N
O
N
O
N
N
O
O
17g
10.1
6.0
O
O
N
O
17h
17i
17j
9.4
9.8
9.1
5.7
6.3
6.3
N
N
O
O
O
N
O
O
17k
9.3
5.7
N
O
O
O
N
17l
8.0
9.4
10
5.7
6.1
6.9
O
O
N
17m
17n
O
O
N
Figure 4. Predicted binding mode for 16f in S1P1 homology model.
a
S1P3 Ca2+ mobilization in hS1P3 assay (in 384-well microplates).
penetrates into the more lipophilic region of the membrane. The
binding mode is also consistent with those reported up to date.16,17
Glu121, a residue with negatively charged side chain, is located clo-
sely to Arg120, and might interacts with Arg120 when Arg120 is free
of contact. This observation suggests that a positively charged moi-
ety, for example basic amine, in the RHS may even enhance ligand
binding affinity in S1P1 if accessible to Glu121. Based on this find-
ing, we tried to introduce a nitrogen atom into RHS chain to vali-
date this hypothesis.
As shown in Table 4, insertion of N into the side chain provided
analogue 17a with a 10-fold increase in S1P1 potency, while at the
same time the undesirable S1P3 activity was also increased by at
least 10-fold. The S1P1 selectivity over S1P3 still met our criterion
of over 1000-fold. The tertiary amine analogue (17b) showed sim-
ilar S1P1 potency to16f. Moving N from right to left side provided
analogue 17c showing similar profiles with its parent compound
17a. The piperidine acid derivative (17d) proved to be potent

G. Deng et al. / Bioorg. Med. Chem. Lett. 22 (2012) 3973–3977
3977
Table 5
The mouse PK, brain–blood ratio and lymphopenia
No.
Mouse PK (2 mg/kg)
Br/Br ratio
Lymphopenia
37% (4 h), >100% (24 h)
17d
T1/2: 5.40 h; F: >100% Cmax 1.07
l
g/mL AUC0–24 h: 10.3
lg h/mL
1.45 (1 h), 3.73 (4 h)
S1P1 agonist without S1P3 agonistic activity. Extension of the acid
side chain by one carbon (17e) resulted in lower S1P1 potency. A
similar S1P1 potency was observed when the piperidine N was
moved inside ring (17f). The potency was much improved by fur-
ther extension of the side chain by one carbon (17g) but the com-
pound was also suffered from undesirable S1P3 activity. The
piperazine analogue (17h) demonstrated comparable S1P1 potency
however still with S1P3 activity. Similarly, the S1P3 activity could
not be mitigated by changing the side chain length (17i and 17j)
or moving the carboxylic acid around the piperidine ring (17k).
The azetidine analogue (17l) was found to be much less potent
compared with the piperidine derivative (17d), probably due to a
shorter side chain. When the side chain length was increased by
one or two carbon, the S1P1 potency was improved drastically
(17m and 17n). Both compounds still showed significant S1P3 ago-
nistic activity. The amino acid compounds in Table 4 suggested
that nitrogen atom in the proper position would facilitate the bind-
ing affinity and improve potency, which confirmed the modeling
findings.
Compound 17d was selected for further progress in in vivo
studies due to its good S1P1 potency (pEC50 = 9.2), minimal unde-
sirable S1P3 activity (pEC50 <5.5), and good overall physiochemical
properties. In mouse CNS penetration study (iv dosing), compound
17d showed good brain to blood ratio (BBR at 1 h and 4 h post dos-
ing are 1.45 and 3.73, respectively; Table 5). The mouse PK study
(p.o. dosing, 2 mg/kg) revealed that this compound had very good
oral bioavailability (ꢀ100%) and a moderate half-life (5.40 h, p.o.)
in correspondence with its low clearance (3.68 mL/min/kg, iv)
and moderate volume of distribution (1.48 L/kg, iv). The compound
was evaluated in the in vivo efficacy study. In the acute mouse
lymphopenia model, 17d turned out to be efficacious at 1 mg/kg
(37% lymphocyte reduction at 4 h). And the lymphopenia effect
was transient. The lymphocytes count measured at 24 h was back
to normal (100% of baseline). Indeed, this transient lymphopenia
profile was desired for MS treatment, and the sustained lymphope-
nia observed from FTY720 (blood lymphocytes were reduced to
20% of baseline and after drug cessation lymphocyte counts only
return to baseline values after 4–8 weeks) was believed to be
responsible for its serious infection side effects in clinical trial.18,19
In summary, we have discovered a novel benzoxazole series of
selective S1P1 agonist based on knowledge-based scaffold hopping
combined with computational approaches. Among these ana-
logues, compound 17d was identified to have good S1P1 agonistic
potency with S1P3 selectivity. Compound 17d demonstrated high
brain penetration and favorable pharmacokinetic properties in
mice. Oral dosing of compound 17d in mice resulted in a statisti-
cally significant reduction in circulating lymphocytes at 4 h post-
dosing and back to normal at 24 h.
References and notes
1. Meng, Q.; Zhao, B.; Xu, Q.; Xu, X.; Deng, G.; Li, C.; Luan, L.; Ren, F.; Wang, H.; Xu,
H.; Xu, Y.; Zhang, H.; Xiang, J.-N.; Elliott, J. D.; Guo, T. B.; Zhao, Y.; Zhang, W.; Lu,
H.; Lin, X. Bioorg. Med. Chem. Lett. 2012, 22, 2794.
2. Brinkmann, V.; Billich, A.; Baumruker, T.; Heining, P.; Schmouder, R.; Francis,
G.; Aradhye, S.; Burtin, P. Nat. Rev. 2010, 9, 883.
3. FDA approves first oral drug to reduce MS relapses; FDA news release Sept. 22,
2010;
(http://www.fda.gov/NewsEvents/Newsroom/PressAnnouncements/
ucm226755.htm).
4. Rosen, H.; Alfonso, C.; Surh, C. D.; McHeyzer-Williams, M. G. PNAS 2003, 100,
10907.
5. Hale, J. J.; Doherty, G.; Toth, L.; Mills, S. G.; Hajdu, R.; Keohane, C. A.; Rosenbach,
M.; Milligan, J.; Shei, G.-J.; Chrebet, G.; Bergstrom, J.; Card, D.; Forrest, M.; Sun,
S.-Y.; West, S.; Xie, H.; Nomura, N.; Rosen, H.; Mandalab, S. Bioorg. Med. Chem.
Lett. 2004, 14, 3501.
6. Buzard, D.; Han, S.; Thoresen, L.; Moody, J.; Lopez, L.; Kawasaki, A.; Schrader, T.;
Sage, C.; Gao, Y.; Edwards, J.; Barden, J.; Thatte, J.; Fu, L.; Solomon, M.; Liu, L.;
Al-Shamma, H.; Gatlin, J.; Le, M.; Xing, C.; Espinola, S.; Jones, R. M. Bioorg. Med.
Chem. Lett. 2011, 21, 6013.
7. Kurata, H.; Kusumi, K.; Otsuki, K.; Suzuki, R.; Kurono, M.; Takada, Y.; Shioya, H.;
Komiya, T.; Mizuno, H.; Ono, T.; Hagiya, H.; Minami, M.; Nakade, S.; Habashit,
H. Bioorg. Med. Chem. Lett. 2011, 21, 3885.
8. Nishi, T.; Miyazaki, S.; Takemoto, T.; Suzuki, K.; Iio, Y.; Nakajima, K.; Ohnuki, T.;
Kawase, Y.; Nara, F.; Inaba, S.; Izumi, T.; Yuita, H.; Oshima, K.; Doi, H.; Inoue, R.;
Tomisato, W.; Kagari, T.; Shimozato, T. ACS Med. Chem. Lett. 2011, 2, 368.
9. Bolli, M. H.; Abele, S.; Binkert, C.; Bravo, R.; Buchmann, S.; Bur, D.; Gatfield, J.;
Hess, P.; Kohl, C.; Mangold, C.; Mathys, B.; Menyhart, K.; Müller, C.; Nayler, O.;
Scherz, M.; Schmidt, G.; Sippel, V.; Steiner, B.; Strasser, D.; Treiber, A.; Weller, T.
J. Med. Chem. 2010, 53, 4198.
10. Cee, V. J.; Frohn, M.; Lanman, B. A.; Golden, J.; muller, K.; Neira, S.; Pickrell, A.;
Arnett, H.; Buys, J.; Gore, A.; Fiorino, M.; Horner, M.; Itano, A.; Lee, M. R.;
McElvain, M.; Middleton, S.; Schrag, M.; Rivenzon-Segal, D.; Vargas, H. M.; Xu,
H.; Xu, Y.; Zang, X.; Siu, J.; Wong, M.; Burli, R. W. ACS Med. Chem. Lett. 2011, 2,
107.
11. Walker, J. K. Discovery of a potent and selective S1P1 receptor agonist for the
treatment of rheumatoid arthritis ACS meeting Mar 21–25, 2010; San
Francisco, CA.
12. Ahmed, M.; Myatt, J.; Norton, D.; Rivers, D. A. From PCT Int. Appl. WO
2008074821A1, 2008.
13. S1P1 Tango assay was conducted according to manufacturer’s protocol (Catalog
K1520, Invitrogen, Carlsbad, CA).
14. S1P3 GeneBLAzer assay was conducted according to manufacturer’s protocol
(Catalog K1713, Invitrogen, Carlsbad, CA).
15. We built a S1P1 homology model using Rhodopsin as a template. Sequence
alignment and initial homology modeling were performed using the Molecular
Operating Environment (MOE, 2009) from the Chemical Computing Group, Inc.
Initial molecular docking attempt for compound 16f with Glide molecular
docking method failed to produce successful placement of the molecule within
the ligand binding site, mainly due to the limited space of the accessible ligand
binding pocket. To further explore 16f’s possible binding mode in S1P1,
induced fit docking, a more advanced in-silico docking protocol that treats
protein structure flexibility, was adopted.
16. Pan, S.; Mi, Y.; Pally, C.; Beerli, C.; Chen, A.; Guerini, D.; Hinterding, K.;
Nuesslein-Hildesheim, B.; Tuntland, T.; Lefebvre, S.; Liu, Y.; Gao, W.; Chu, A.;
Brinkmann, V.; Bruns, C.; Streiff, M.; Cannet, C.; Cooke, N.; Gray, N. Chem. Biol.
2006, 13, 1227.
17. Schürer, S. C.; Brown, S. J.; Cabrera, P. G.; Schaeffer, M.-T.; Chapman, J.; Jo, E.;
Chase, P.; Spicer, T.; Hodder, P.; Rosen, H. ACS Chem. Biol. 2008, 3, 486.
18. Morris, M. A.; Gibb, D. R.; Picard, F.; Brinkmann, V.; Straume, M.; Ley, K. Eur. J.
Immunol. 2005, 35, 3570.
19. Park, S. I.; Felipe, C. R.; Machado, P. G.; Garcia, R.; Skerjanec, A.; Schmouder, R.;
Tedesco-Silva, H., Jr.; Medina-Pestana, J. O. Braz. J. Med. Biol. Res. 2005, 38, 683.
Acknowledgments
We thank Drs. Eric Yang, Hong Lu, Zongping Zhang, Jiansong
Yang, Jinqiang Zhang, Jiansong Yang, Wei Zhang, Frankie S. Mak,
for their helpful comments and discussions.