Discovery of a novel series of potent S1P1 agonists

Article abstract of DOI:10.1016/j.bmcl.2010.01.102

The discovery of a novel series of S1P1 agonists is described. Starting from a micromolar HTS positive, iterative optimization gave rise to several single-digit nanomolar S1P1 agonists. The compounds were able to induce internalization of the S1P1 receptor, and a selected compound was shown to be able to induce lymphopenia in mice after oral dosing.

Full text of DOI:10.1016/j.bmcl.2010.01.102

Bioorganic & Medicinal Chemistry Letters 20 (2010) 1516–1519
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of a novel series of potent S1P1 agonists
a
a
a
a
Stefano Crosignani a, , Agnes Bombrun , David Covini , Maurizio Maio , Delphine Marin ,
Anna Quattropani a, Dominique Swinnen a, Don Simpson b, Wolfgang Sauer a, Bernard Françon a,
Thierry Martin a, Yves Cambet a, Anthony Nichols a, Isabelle Martinou a, Fabienne Burgat-Charvillon a,
Delphine Rivron a, Cristina Donini a, Olivier Schott a, Valerie Eligert a, Laurence Novo-Perez a,
Pierre-Alain Vitte a, Jean-François Arrighi a
*
a Merck Serono SA, 9 Chem. des Mines, 1202 Geneva, Switzerland
b BioFocus DPI, Saffron Walden, BC10 1XL, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
The discovery of a novel series of S1P1 agonists is described. Starting from a micromolar HTS positive,
iterative optimization gave rise to several single-digit nanomolar S1P1 agonists. The compounds were
able to induce internalization of the S1P1 receptor, and a selected compound was shown to be able to
induce lymphopenia in mice after oral dosing.
Received 21 December 2009
Revised 15 January 2010
Accepted 16 January 2010
Available online 25 January 2010
Ó 2010 Elsevier Ltd. All rights reserved.
Keywords:
S1P1
Sphingosine-1-phosphate
Structure–activity relationships
Hit-to-lead
Lymphopenia
Sphingosine-1-phosphate is an important chemical mediator,
which exerts its effects via five G-protein coupled receptors (S1P1-
5). In recent years, S1P1 in particular has attracted wide interest as
a therapeutic target, as molecules capable of agonising this receptor
demonstrate a potent lymphopenic effect in vivo.1 In addition, the
investigational drug FTY720 (Fingolimod, 1a, now in Phase 3 clinical
trial forrelapsing-remitting multiplesclerosis) is believed toprovide
therapeuticbenefitviaS1P1:its phosphorylatedform1b(essentially
generated in vivo by sphingosine kinase type 2 activity)2 is in fact a
potent S1P1 agonist and animals treated with FTY720 experience a
strong peripheral blood lymphopenia.3
Phosphorylated FTY720 is poorly selective against the various
S1P receptors, being very active on S1P1, S1P3, S1P4 and S1P5.3
However, it has been demonstrated that only S1P1 activity is nec-
essary to induce lymphopenia, while it has been shown that brady-
cardia induced by S1P receptor non-selective agonists in wild type
mice is suppressed in S1P3 À/À mice.4 To attenuate the potential
risk of bradycardia more selective compounds are needed. In this
Letter, we describe the identification and characterization of new
small molecules which are selective and potent S1P1 agonists
capable of inducing a lymphopenic effect.
A high-throughput screening run on 100,000 compounds (using
a cAMP readout) yielded, among others, compound 2 as positive
(Fig. 1). This compound was considered to be a suitable starting
point for a medicinal chemistry program.
The general synthetic methods employed are summarized in
Scheme 1. The carboxylic acids 5 could be prepared by reaction
of an amine with methyl 2,4-dichloropyrimidine-6-carboxylate,
followed by reductive dechlorination and hydrolysis of the methyl
ester group. These intermediates could be coupled with various
amines to explore the SAR of the R3 substituent, using standard
amide coupling reagents. In particular, we routinely used the poly-
mer-supported Mukaiyama reagent for its ease of use.5 The 2-chlo-
ropyrimidine intermediates 4 could also be used to prepare some
2-substituted compounds via either Pd-catalyzed chemistry or
SNAr reactions.
N
N
CH3(CH2)7
H
N
OR
NH2
HN
O
OH
OH
1a FTY720 R=H
1b (S)-FTY720-P R=PO3H2
2
* Corresponding author. Tel.: +41 224149618; fax: +41 227312179.
E-mail address: stefano.crosignani@merckserono.net (S. Crosignani).
Figure 1. FTY720 (1a), (S)-FTY720-phosphate (1b) and HTS positive 2.
0960-894X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2010.01.102

S. Crosignani et al. / Bioorg. Med. Chem. Lett. 20 (2010) 1516–1519
1517
Cl
Cl
N
N
N
N
H
N
b, c
a
e
d
g
R1
R1
N
N
N
N
R1
Cl
OH
R5
3
OMe
OH
R1
OMe
Cl
N
N
R
R2
2, 10-60
O
R4
R2
O
Cl
N
O
R2
O
O
3
4
5
R3:
R5
N
N
N
N
H
N
H
N
f
N
N
N
N
3
N
R
R2
2, 10-60
O
HO
Cl
O
O
6
7
8
Scheme 1. Reagents and conditions: (a) R1R2NH, TEA, THF; (b) (NH4)+HCOOÀ, Pd/C; (c) NaOH, THF/H2O; (d) R3NH2, PS-Mukaiyama reagent, TEA; (e) PCl5, POCl3; (f) R3NH2,
TEA, DCM; (g) R1R2NH, DIEA, EtOH, 160 °C (microwave irradiation).
To explore the SAR of the R1 and R2 substituents, a different
strategy was employed: 6-hydroxypyrimidine-4-carboxylic acid
was first converted to 6-chloropyrimidine-4-carbonyl chloride 7,
which was reacted with the desired aniline compounds. The inter-
mediates 8 could then be reacted with a variety of amines under
microwave irradiation to give the desired compounds.
number of cyclized substituents, such as variously substituted
piperidines, morpholines or piperazines led to inactive compounds
(results not shown).
Keeping the cyclohexyl group fixed, we then explored the influ-
ence of a second substituent R2 on the nitrogen. Already, a simple
methyl group led to a twofold increase in potency. Bigger increases
in potency could be obtained with longer substituents (ethyl, pro-
pyl, cyclopropylmethylene) giving rise to very potent compounds
(e.g., 18, EC50 25 nM). When in conjunction with second substitu-
ent of sufficient length (e.g., a propyl group), the cycloalkyl group
could be substituted with a non-cyclic group (e.g., a second propyl
group, 19) with an approximately 10-fold drop in potency.
A few modifications of the central core of the molecule were at-
tempted, all leading to dramatic reduction in potency. Even small
substituents on the C2 position of the pyrimidine ring (Me, OMe,
Cl) lead to significant loss of potency (data not shown). The same
was observed after methylation of the amide nitrogen.
At first, the SAR of the R1 substituent was explored (Table 1). Of
the compounds present in the original screening library, only those
possessing a cycloalkyl substituent were found active. The crucial
importance of this substituent was confirmed during our investiga-
tion: smaller substituents lead to a drastic loss of potency (e.g., 9
and 10). Replacing the cycloalkyl substituent with variously substi-
tuted aromatic groups led to poorly active compounds, with the
only exception being the o-tolyl derivative 13, which kept a similar
level of activity. Replacing the cycloalkyl amine moiety with a
At the same time, a preliminary SAR was conducted on R3 (Table
1). We quickly discovered that a small lipophilic substituent on the
position meta to the phenol (R4) was advantageous, leading to a 2–
4-fold increase in potency (20–23). On the other hand, substitution
on the position ortho to the phenol had a markedly inferior effect
(data not shown).
However, our main concern was the presence of the p-anilino-
phenol moiety, both for the possibility of formation of toxic metab-
olites as well as for the possibility of Phase 2 metabolism on the
phenolic group. In fact, the pharmacokinetic (PK) profile of com-
pound 20 was found to be a parameter to be optimized: despite a
Table 1
Compds
R1
R2
R4
R5
S1P1 GTPcS
EC50 (nM)
a
2
9
c-Hexyl
c-Propyl
i-Butyl
c-Pentyl
Ph
o-Tolyl
c-Heptyl
c-Hexyl
c-Hexyl
c-Hexyl
c-Hexyl
Pr
c-Hexyl
c-Hexyl
c-Hexyl
c-Hexyl
c-Hexyl
c-Hexyl
c-Hexyl
c-Hexyl
c-Hexyl
c-Hexyl
c-Hexyl
c-Hexyl
c-Hexyl
c-Hexyl
c-Hexyl
Pr
H
H
H
H
Me
H
H
Me
Me
Et
c-PrCH2
Pr
H
H
H
H
H
H
H
H
H
H
H
H
H
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
1780
>20,000
>20,000
16,700
>20,000
2100
973
700
655
405
25
375
1160
440
390
382
>20,000
>20,000
>20,000
>20,000
960
>20,000
1990
440
Me
Me
H
H
H
Me
H
Me
Me
Me
H
Me
F
Cl
CF3
H
H
F
H
H
H
H
Me
Me
Me
Me
Me
H
H
Me
H
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
medium to low clearance in microsomes (2 and 18 lL/min/mg
respectively in human and rat liver microsomes), the compound
exhibited very high clearance (4.85 L/h/kg) in mice, and significant
amountsof theglucuronide adductcouldbe foundin thePKsamples,
confirming that the phenolic group had to be replaced.
A number of substituents (fluorine, methoxy, carboxy, hydroxy-
methyl or hydroxyethyl, see Table 1, compds 24–27) were tested to
replace the phenol as R5, without much success. Also, moving the
phenol substituent to the meta position of the aniline led to less
potent compounds (data not shown). The data clearly indicated
the preference for an H-bond donor (HBD) as R5.
CH2OH
CH2CH2OH
F
OMe
NHSO2Me
NHCOMe
SO2NH2
SO2NH2
SO2NH2
SO2NH2
SO2NH2
SO2NH2
SO2Me
CONH2
SO2NHMe
CH2NH2
CH2NMe2
The first alternative R5 substituents bearing an HBD that were
investigated were N-arylsulfonamide 28 and N-arylacetamide 29.
Interestingly, the activity of these two compounds seemed to be
dependent on the acidity of the compound, with a pKa similar to
that of the phenol being favoured. So the sulfonamide 28 (EC50
960 nM, calcd pKa 9.9) was found to be significantly more potent
Me
c-PrCH2
Pr
65
1.6
2.9
36
193
73
6
39
Pr
c-Hexyl
c-Hexyl
c-Hexyl
c-Hexyl
c-Hexyl
c-PrCH2
c-PrCH2
c-PrCH2
c-PrCH2
c-PrCH2
than the corresponding acetamide 29 (EC50 >20 lM, calcd pKa
13.6). However, the presence of a 1,4-diamino phenyl unit was
not considered as being of particular advantage in terms of poten-
tially toxic metabolites. In addition, the reversed sulfonamide 30
was also found to possess a similar potency, indicating that there
H
104
a
Average of two experiments. SD were generally within 50% of average.

1518
S. Crosignani et al. / Bioorg. Med. Chem. Lett. 20 (2010) 1516–1519
Table 2
was some scope for changing the position of the HBD. Quite satis-
fyingly, the pharmacokinetic profile of 30 proved to be markedly
superior to the phenol 20, with a low clearance of 0.3 L/h/kg.
This new R5 group was then combined with some of the best
substituents for the lipophilic tail. The SAR was found to be very
similar to that for the phenol subseries, and the compounds bear-
ing a cyclohexyl substituent and a propyl or cyclopropylmethylene
group (33, 34) proved to be very potent, with EC50 of 1.6 and
2.9 nM, respectively. As in the phenol subseries, the derivative
bearing two smaller chains (35) were less active by approximately
one order of magnitude. The fact that this reversed sulfonamide
group was still mainly behaving as an HBD was proved by the fact
that methyl sulfone 36 was less active by ca. 2 orders of magnitude
than the corresponding sulfonamide 33. The primary amide 37,
possessing less acidic protons, also showed an inferior activity.
We also tested some compounds bearing a chargeable nitrogen
to act as HBD (compds 39–40). These compounds retained potency
in the same range as for the corresponding phenol compounds,
although not as good as for the sulfonamide derivatives.
We then explored the possibility of benzofused heterocycles
(Fig. 2) as bioisosteres for the phenolic moiety (Table 2). The corre-
lation between the acidity of the proton and the affinity of the
compounds for the S1P1 receptor was confirmed: indazoles 41
and 42 (calcd pKa 13.6) turned out to be more active than indole
43 (calcd pKa 16.8). Benzotriazole 44 (calcd pKa 8.6) was also less
active, although this could be due to the existence of a second tau-
tomer. Contrary to what was observed before for the indazole-con-
taining compounds, no significant difference of activity was found
between the cyclohexyl-containing compound 45 and its propyl
analogues 46 and 47, with di-propyl analogue 47 being actually
the most potent.
Compds
R1
R2
R3
S1P1 GTPcS
EC50 (nM)
a
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
c-Hexyl
c-Hexyl
c-Hexyl
c-Hexyl
c-Hexyl
Pr
H
Me
H
Me
c-PrCH2
c-PrCH2
Pr
Ab
A
C
D
A
A
A
B
B
B
E
E
2710
418
12,100
7300
9
13
7
211
6
12
12
208
3
33
20
238
35
94
19
3
Pr
c-Hexyl
c-Hexyl
Pr
c-Hexyl
Pr
c-Hexyl
Pr
c-Hexyl
Pr
c-Hexyl
c-Hexyl
c-Hexyl
c-Hexyl
Me
c-PrCH2
c-PrCH2
c-PrCH2
c-PrCH2
c-PrCH2
c-PrCH2
c-PrCH2
c-PrCH2
c-PrCH2
c-PrCH2
c-PrCH2
c-PrCH2
G
G
L
L
M
N
Q
R
a
Average of two experiments. SD were generally within 50% of average.
See Figure 2.
b
Table 3
Selectivity of selected compounds
Compds
S1P1 GTP
EC50 (nM)
c
S
S1P3 GTP
EC50 (nM)
c
S
S1P5 GTPcS
a
a
a
EC50 (nM)
2
18
19
20
31
33
46
49
53
59
60
1780
25
>30,000
>30,000
2320
>30,000
>30,000
>30,000
416
3270
62
1094
236
NDb
191
NDb
258
NDb
96
142
NDb
114
6.3
375
1160
440
1.6
13
6
3
19
3
Pursuing our exploration of R3 groups (Fig. 2), we shifted our
attention to the possibility of integrating a carboxylic acid moiety
in our compounds, to improve their solubility and decrease their
Log D7.4. We chose to integrate this new feature in both our sulfon-
amide and amine-based R3 groups, using spacers of differing length
(51–60) and in both cases we could identify very potent com-
pounds, with a preference being for the compounds with a 2-car-
bon spacer (53, 54, 59). Of particular advantage was, in the case
of the amino-acid spacer, the use of the azetidine linker (60, EC50
3 nM). In this case, the potency of the azetidine compound is
remarkably better than the corresponding dimethylamino com-
pound (39), clearly indicating a significant positive contribution
of the carboxylic acid group to the binding to the S1P1 receptor.
To verify that the effect could be exclusively attributed to the car-
boxylic acid, we also prepared the corresponding azetidine deriva-
tive without the carboxylic acid group, which turned out to be
significantly less active. Inclusion of other solubility-enhancing
substituents on our sulfonamide or amino compounds, such as
hydroxylated alkyl chains, had a detrimental effect on the potency,
as well as a very limited impact on the solubility of the compounds
(data not shown).
0.6
a
Average of two experiments. SD were generally within 50% of average.
Not determined.
b
(e.g., 53) being the least selective. Sulfonamide 33 showed the best
selectivity, with more than four orders of magnitude. Compounds
bearing the cyclohexyl substituent tended to give significantly bet-
ter selectivity versus S1P3 (cf., e.g., 19 vs 18 and 49 vs 46). In gen-
eral, the compounds were not selective versus S1P5, with the
exception of compds 33 and 53 showing a moderate selectivity.
Having discovered a series of very potent analogues, these were
profiled in a battery of in vitro assays to determine their suitability
for in vivo testing. The results of some representative compounds
are summarized in Table 4. As expected, all compounds were lipo-
philic, with Log P values above four for almost all compounds
tested. All compounds lacking an ionizable group showed poor sol-
ubility in aqueous solutions. Predictably, compounds bearing the
Selectivity data for some derivatives can be found in Table 3.
Selectivities versus S1P3 were generally moderate to good, with
the compounds containing both a sulfonamide and acid groups
O
R
N
O
n
O
n
N
H
N
O
OH
N
N
N
OH
S
OH
N
N
H
N
H
R
O
H
M R= H, n= 1
R= Me, n= 1
Q R= H, n= 2
E n= 1
G n= 2
L n= 3
R
A R= H
B R= Me
C
D
N
Figure 2. Alternative R3 groups (see Table 2).

S. Crosignani et al. / Bioorg. Med. Chem. Lett. 20 (2010) 1516–1519
1519
Table 4
nia in vivo. When dosed at 30 mg/kg po, 46 caused a 72% decrease
of blood lymphocytes in mice at 4 h after dosing (without any ef-
fect on erythrocytes and platelets).
Summary of in vitro profiling data
Compds Log P Solubilitya Microsome Clintb
Caco-2 permeability
(
l
g/mL)
(human/rat) (
l
L/min/mg) (Papp,c 10À6 cm/s)
In summary, starting from a weakly potent S1P1 agonist, we
have discovered several single-digit nanomolar agonists. We have
successfully replaced the unwanted phenol group of the original
hit with different groups, which are more acceptable from a toxic-
ity and pharmacokinetic point of view. We have verified that the
agonistic behaviour of our compounds towards the S1P1 receptor
translated as well in the desired internalization of the receptor.
For a selected compound, we confirmed that this functional antag-
onism could induce a significant lymphopenia in mice.
18
19
20
31
33
46
49
53
59
60
7.4
8
22/52
4/7
2/18
43/15
130/14
17/57
<10/26
25 / 25
<10/<10
<10/<10
NDd
21.4
21.8
12.5
NDd
8.5
0.1
2.0
0.8
2.4
NDd 16
4.1
3.6
5.7
4.6
6.1
>32
5
1
11
5
NDd >51
NDd >45
>2.7e 22
a
b
c
Nephelometric kinetic solubility in PBS (1% DMSO).
Intrinsic clearance (see Supplementary data).
Apparent permeability (see Supplementary data).
Not determined.
Supplementary data
d
e
Typical conditions for the synthesis of compounds, as well as
experimental details on the in vitro assays. Full experimental con-
ditions and characterization of all compounds can be found in pat-
ent WO2009019167. Supplementary data associated with this
article can be found, in the online version, at doi:10.1016/
j.bmcl.2010.01.102.
Log D7.4
.
carboxylic acid groups were considerably more soluble; however,
they suffered from a reduced permeability in the Caco-2 assay
compared to the neutral compounds. In most cases, the com-
pounds were sufficiently stable in the presence of human or rat
microsomes, the only exception being compound 33 which showed
instability only in the human system.
In parallel, we verified if the agonistic effect of our compounds
on the S1P1 receptor lead to its internalization, which is believed to
be the cause of the lymphopenic effect in vivo. This was tested
using human osteosarcoma U2OS cells expressing the S1P1 recep-
tor fused to Green Fluorescent Protein (GFP).6 Internalization of the
receptor could be verified by imaging, and the relative EC50s could
be calculated for some compounds, with results in the same range
References and notes
1. (a) Brinkman, V.; Cyster, J. G.; Hla, T. Am. J. Transplant. 2004, 4, 1019; (b) Buzard,
D. J.; Thatte, J.; Lerner, M.; Edwards, J.; Jones, R. M. Exp. Opin. Ther. Pat. 2008, 18,
1141.
2. Zemann, B.; Kinzel, B.; Müller, M.; Reuschel, R.; Mechtcheriakova, D.; Urtz, N.;
Bornancin, F.; Baumruker, T.; Billich, A. Blood 2006, 107, 1454.
3. Mandala, S.; Hajdu, R.; Bergstron, J.; Quackenbush, E.; Xie, J.; Milligan, J.;
Thornton, R.; Shei, G. J.; Card, D.; Keohane, C.; Rosenbach, M.; Hale, J.; Lynch, C.
L.; Rupprecht, K.; Parson, W.; Rosen, H. Science 2002, 296, 346.
4. Forrest, M.; Sun, S.-Y.; Hajdu, R.; Bergstron, J.; Card, D.; Doherty, G.; Hale, J.;
Keohane, C.; Meyers, C.; Milligan, J.; Mills, S.; Nomura, N.; Rosen, H.; Rosenbach,
M.; Shei, G.-J.; Singer, I. I.; Tian, M.; West, S.; White, V.; Xie, J.; Proia, R. L.;
Mandala, S. J. Pharmacol. Exp. Ther. 2004, 309, 758.
5. Crosignani, S.; Gonzalez, J.; Swinnen, D. Org. Lett. 2004, 6, 4579.
6. Jo, E.; Sanna, M. G.; Gonzalez-Cabrera, P. J.; Thangada, S.; Tigyi, S.; Osborne, D. A.;
Hla, T.; Parrill, A. L.; Rosen, H. Chem. Biol. 2005, 12, 703. The U2OS cell line was
obtained from BioImage.
as those obtained in the GTP
46 and 60 had EC50s of 12 and 3 nM, respectively, in the internal-
ization assay (13 and 3 nM, respectively, in the GTP S assay).
cS binding assay. For example, compds
c
Finally, compound 46 was used to test if the effect on S1P1
receptor internalization could translate into significant lymphope-