Potent S1P receptor agonists replicate the pharmacologic actions of the novel immune modulator FTY720

Article abstract of DOI:10.1016/j.bmcl.2004.02.106

Alteration in lymphocyte trafficking and prevention of graft rejection in rodents observed on exposure to FTY720 (1) or its corresponding phosphate ester 2 can be induced by the systemic administration of potent sphingosine-1- phosphate receptor agonists

Full text of DOI:10.1016/j.bmcl.2004.02.106

Bioorganic & Medicinal Chemistry Letters 14 (2004) 3351–3355
Potent S1P receptor agonists replicate the pharmacologic
actions of the novel immune modulator FTY720
Jeffrey J. Hale,a,* William Neway,a Sander G. Mills,a Richard Hajdu,b
Carol Ann Keohane,b Mark Rosenbach,b James Milligan,b Gan-Ju Shei,b
Gary Chrebet,b James Bergstrom,b Deborah Card,b Gloria C. Koo,c Sam L. Koprak,c
Jesse J. Jackson,b Hugh Rosenb, and Suzanne Mandalab
aDepartment of Medicinal Chemistry, Merck Research Laboratories, Rahway, NJ 07065, USA
bDepartment of Immunology and Rheumatology Research, Merck Research Laboratories, Rahway, NJ 07065, USA
cDepartment of Atherosclerosis and Endocrinology Research, Merck Research Laboratories, Rahway, NJ 07065, USA
Received 12 December 2003; accepted 17 February 2004
Abstract—Alteration in lymphocyte trafficking and prevention of graft rejection in rodents observed on exposure to FTY720 (1) or
its corresponding phosphate ester 2 can be induced by the systemic administration of potent sphingosine-1-phosphate receptor
agonists exemplified by 19. The similar S1P receptor profiles of 2 and 19 coupled with their comparable potency in vivo supports a
connection between S1P receptor agonism and immunosuppressive efficacy.
Ó 2004 Elsevier Ltd. All rights reserved.
FTY720 (1) is a novel, orally active immunosuppressive
agent that is currently in Phase II clinical trials for the
prevention of allograft rejection.1 While the pharmaco-
dynamic effects that follow the systemic administration
of 1 have been described,2;3 entry into the clinic appar-
ently preceded an understanding of the mechanism(s) by
which this compound worked. Compound 1 has since
been reported to be rapidly metabolized in the blood of
a variety of species (including man) to the corresponding
phosphate ester 2, which is a potent agonist of four of
the five known sphingosine-1-phosphate (S1P) recep-
tors.4;5 Agonism of S1P receptors appears to alter the
trafficking of circulating lymphocytes and this has been
postulated to contribute to the immunosuppressive
efficacy of 1. With a molecular target connected to the
actions of 1, we sought to obtain S1P receptor agonists
without the complication of the observed in vivo equi-
librium between 1 and 2 so that we could directly study
the pharmacologic effects of the agonism of S1P recep-
tors. Our efforts to obtain such compounds and the
details of their characterization are the subject of this
report.
Our initial medicinal chemistry efforts were focused on
obtaining nonhydrolyzable phosphonate analogs of 2
that had comparable S1P receptor affinities and phar-
macokinetic profiles making them suitable for use as
molecular tools. An obvious target was phosphonate 3;4
this compound was found to lower circulating lympho-
cytes in rodents similarly to 1. Relatively higher doses
were required to elicit the lymphopenic response and
* Corresponding author. Tel.: +1-732-594-2916; fax: +1-732-594-5966;
e-mail: jeffrey_hale@merck.com
Present address: The Scripps Research Institute, ICND-156, 10550
N. Torrey Pines Road, La Jolla, CA 92037, USA.
0960-894X/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.02.106

3352
J. J. Hale et al. / Bioorg. Med. Chem. Lett. 14 (2004) 3351–3355
this could be attributed in part to its significantly lower
affinity for S1P receptors as compared to 2 (Table 1).
With this data in hand, we initiated efforts to further
refine the structure of 3 with the goal of obtaining
analogs with enhanced potency in vitro and in vivo.
The syntheses of some of the various test compounds are
shown in Scheme 1. Hydrolysis/decarboxylation of di-
ethyl 2-(acetylamino)-2-(2-(4-octylphenyl)ethyl) malon-
ate6 afforded amino acid 4 which was a key intermediate
required to prepare many of the desired targets. N-
Protection of 4 as the benzyl carbamate followed by
carboxyl reduction and dibenzylphosphorylation gave 5.
Global deprotection with sodium/ammonia afforded
phosphate ester 6. Amino acid 4 was readily elaborated
to N-Boc aldehyde 7 which could be subjected to vari-
ous homologation sequences to give analogs such as
c-amino acid 8, a,b-unsaturated phosphonate 10, or
phosphonate 11. Addition of dibenzyl methylphospho-
nate to 7 followed by deprotection afforded b-hydroxy
phosphonate analog 9. Conversion of amino acid 4 to
N-Boc carboxylate 12 followed by elaboration to b-
amino acid 13 featured an Arndt–Eisert rearrangement.
Table 1. Inhibition (IC50, nM) of [33P]-S1P binding to S1P receptorsa
R
R (Compd)
S1P1
S1P3
S1P4
S1P5
S1P
2
0.16
0.28
8.7
0.04
6.3
23
15
0.23
0.77
230
3
510
210
PO3H2
O
0.76
0.71
68
1.7
1.8
NH2
6
OH
PO3H2
PO3H2
14
140
240
NH2
9
8.4
280
720
62
240
100
76
NH2
11
PO3H2
PO3H2
14
140
NH2
10
0.16
1700
1900
710
0.10
4.8
23
9.8
NH2 OH
14
SO3H
7600
>10,000
6600
18
>10,000
>10,000
>10,000
16
4400
3300
1000
4.5
NH2
15
CO2H
NH2
13
CO2H
NH2
8
PO3H2
NH2 OH
Scheme 1. Reagents and conditions: (a) conc HCl, 100 °C (100%); (b)
benzyl chloroformate, dioxane/aq NaOH, rt (70%); (c) isobutyl chlo-
roformate, TEA, THF, 0 °C, then NaBH4, THF/H2O; (d) dibenzyl
(N,N-diisopropyl)phosphoramidite, 1H-tetrazole, CH2Cl2, 0 °C, then
MCPBA (58%, two steps); (e) Na/NH3, THF, )33 °C (73%); (f) di-t-
butyldicarbonate, dioxane/aq NaOH (100%); (g) (COCl)2, DMSO,
DIEA, )78 to 0 °C (80%, two steps); (h) trimethyl phosphonoacetate,
NaHMDS, THF, 0 °C to rt (72%); (i) H2, 10% Pd/C, MeOH; (j) TFA,
CH2Cl2; (k) NaOH, aq MeOH (8: 80%, three steps, 13: 79%, two
steps); (l) dibenzyl methylphosphonate, LDA, THF, )78 °C; (m) iodo-
trimethylsilane, CH2Cl2 (9: 10%, 10: 73%, two steps, 11: 56%, two
steps, 14: 81%); (n) tetraethyl methylenediphosphonate, NaHMDS,
THF; (o) isobutyl chloroformate, NMM, CH2Cl2/ether, then CH2N2
(86%); (p) silver benzoate, TEA, MeOH (87%); (q) DIBALH, CH2Cl2,
)78 °C to rt (91%); (r) diethyl phosphite, NaHMDS, THF, 0 °C (63%,
ds ¼ 3:1); (s) I2, PPh3, imidazole, CH2Cl2 (76%); (t) Na2SO3, aq EtOH,
70 °C (42%). All compounds are racemic or equal mixtures of diaste-
reomers.
19
PO3H2
85
280
33
NH2 OH
20
PO3H2
17
320
800
73
NH2 OH
21
PO3H2
15
320
500
27
NH2 OH
22
a Displacement of [33P]-labeled sphingosine-1-phosphate (S1P) by test
compounds from human S1P receptors expressed on CHO cell
membranes. All new compounds had S1P2 IC50 > 10 lM. Data are
reported as mean for n ¼ 3 determinations. SD were generally 20%
of the average. See Ref. 3 for assay protocol.

J. J. Hale et al. / Bioorg. Med. Chem. Lett. 14 (2004) 3351–3355
3353
Compound 12 could also be carried through straight-
forward sequences to give either a-hydroxy phospho-
nate analog 14 or sulfonic acid 15. The individual
diastereomers of 14 were synthesized as shown in
Scheme 2. The stereocenter of the carbon bearing the
amino group was set during a Friedel–Crafts reaction
between octylbenzene and enantiopure ())-2-tri-
fluoroacetamidosuccinic anhydride7 to give 16. Ketone
reduction followed by switching of N-protecting groups
afforded the chiral N-Boc amino acid 17. A sequence of
reactions analogous to those used in the preparation of
14 afforded 18 as a mixture of diastereomers which were
separated using silica gel chromatography. Global de-
protection of the individual diastereomers with iodotri-
methylsilane afforded 19 and 20. Diastereomers 21 and
22 were obtained by employing the (R)-anhydride in the
Friedel–Crafts reaction. The configuration of the carbon
bearing the hydroxy group in these compounds was
determined by carrying out NOE experiments on the
cyclic carbamates 23 and 24 which were prepared as
shown.8
approximately 10-fold higher affinity for both S1P3 and
S1P4 as compared to 2 while the profile of des-hydroxy-
methyl phosphonate 11 was almost identical to that of 3.
The structural requirements of a bioisosteric replace-
ment for the phosphate ester of either 2 or 6 appear to
be fairly restricted as only a-hydroxy phosphonate
analog 14 was found to have affinities for the various
S1P receptors approaching those of 2 and 6. The data
for the individual a-hydroxy phosphonate diastereomers
19–22 indicate that the 2-(R), 4-(S)-stereochemistry of
19 is preferred for maximal S1P receptor affinity and the
configuration of these centers has little effect on selecting
either for or against the different S1P receptor sub-types.
Additionally, aside from a 6-fold lower affinity for S1P5,
the S1P receptor profile of diastereomer 19 was found to
be quite comparable to that of 2.
Recent data demonstrate that hematopoietic cells
genetically deficient in S1P1 receptors have defects in
thymic emigration and recirculation similar to lympho-
cytes from normal mice treated with 1 which indicates
that an agonist-driven functional antagonism of S1P1 is
an important component in the efficacy of 1.10;11 We
have found that cell surface expression of FLAG-tagged
S1P1 on HEK293 cells is down-regulated for sustained
periods in response to treatment with 2 and 19. S1P1 is
also desensitized by 1 as reported,12 but with delayed
kinetics that are consistent with the requirement for
phosphorylation to the active form.
Ligand competition studies between [33P]-S1P and 2,
and 3 and all of the new compounds were carried out for
each of the five human S1P receptors stably expressed
in Chinese Hamster Ovary (CHO) cell membranes.4 S1P
receptor agonism by the test compounds was also
determined by measurement of ligand-induced [35S]-50-
O-3-thiotriphosphate (GTPcS) binding; all of the com-
pounds tested were found to be agonists of S1P recep-
tors (data not shown).9 Several things were readily
observed of the S1P receptor data (Table 1). Deletion of
the hydroxymethyl group of either 2 or 3 was found to
have a minimal effect on S1P receptor affinity. Phos-
phate 6 had only 3-fold lower affinity for S1P1 and
The immunosuppressive efficacy of 1 has been proposed
to arise from its ability to promote the sequestration of
CD4þ and CD8þ T cells and B cells in secondary lym-
phoid organs which prevents their infiltration into
transplanted or antigen-bearing nonlymphoid tissues.3
Measurement of blood lymphocyte counts in rodents
after the administration of test compounds provides a
convenient surrogate marker for efficacy that is ame-
nable to the screening of multiple analogs in vivo.13
Intravenous administration of either 1 or 2 to mice or
rats has been previously found to lead to a rapid low-
ering of circulating lymphocytes with the nadir being
reached in 2–3 h.4 Each of the a-hydroxy phosphonate
analogs 19–22 was found to maximally lower circulating
lymphocytes in a manner similar to 2 at a three hour
time point after iv administration to mice. Pharmaco-
dynamic dose-titration data for 19–22 (Table 2) indicate
that in vivo potency tracks well with S1P receptor
affinity. The comparable rat pharmacokinetic profiles of
2, 19, and 20 are also consistent with this observation.
Circulating lymphocytes were found to be minimally
lowered 3 h after administration of phosphate ester 6 to
mice at a dose of 10 mpk iv (Table 2); this result appears
to be in agreement with the lack of immunosuppressive
efficacy in rodents that has been reported for the cor-
responding amino alcohol.2 It should be noted that the
attenuated pharmacodynamic activity of 6 as compared
to analogs with comparable S1P receptor affinities (e.g.,
2 or 19) can not be explained by species differences as all
of the compounds described here have very similar
affinities for the mouse, rat, and human S1P receptors.
Pharmacokinetic factors, such as enhanced metabolism
Scheme 2. Reagents and conditions: (a) 2-(R)-Trifluoro-
acetamidosuccinic anhydride, AlCl3, nitromethane, CH2Cl2 (70%); (b)
H2, 10% Pd/C, HOAc; (c) NaOH, aq dioxane, then di-t-butyl dicar-
bonate; (d) see Scheme 1; (e) separate diastereomers; (f) TMS-I,
CH2Cl2 (77%); (g) HCl, EtOH; (h) COCl2, DIEA, MeCN, rt (23: 74%,
two steps, 24: 53%, two steps).

3354
J. J. Hale et al. / Bioorg. Med. Chem. Lett. 14 (2004) 3351–3355
Table 2. Mouse peripheral lymphocyte loweringa (PLL) and rat pharmacokineticb data for selected S1P receptor agonists
R
R (Compd)
Murine PLLED50 (mpk iv)
0.06
Rat PK(1.0 mpk iv)
2
Clp ¼ 17.8 mL/min/kg
Vdss ¼ 5.9 L/kg, t1=2 ¼ 7:6 h
PO3H2
O
35% @ 10 mpkc
Clp ¼ 29.8 mL/min/kg
NH2
Vdss ¼ 35.1 L/kg, t1=2 ¼ 4:2 h
6
PO3H2
0.15
0.6
Clp ¼ 5.2 mL/min/kg
NH2 OH
Vdss ¼ 1.4 L/kg, t1=2 ¼ 5:7 h
19
PO3H2
Clp ¼ 4.3 mL/min/kg
NH2 OH
Vdss ¼ 0.7 L/kg, t1=2 ¼ 3:6 h
20
PO3H2
3.0
2.5
ndd
ndd
NH2 OH
21
PO3H2
NH2 OH
22
a Individual data points for dose titrations were the average percentage decrease of peripheral blood lymphocyte counts in n ¼ 3 animals versus
control (n ¼ 3) 3 h after iv administration of the test compound. SD were generally 20% of the average. See Ref. 13.
b See Ref. 14.
c Percentage decrease of peripheral blood lymphocyte counts in test animals (n ¼ 3) versus control.
d Not determined.
of 6 in the mouse or the large steady state volume of
distribution observed for this compound in the rat, or as
yet undetermined subtleties between S1P receptor
selectivity and alterations in lymphocyte trafficking are
potential sources that could require further investiga-
tion.
2 also behaves similarly in two rodent models for the
prevention of allograft rejection. These data support the
connection between S1P receptor agonism and the
immunosuppressive efficacy of FTY720. Investigations
into the relationships between S1P receptor subtype
selectivity and the immunomodulatory effects of S1P
receptor agonists are underway.
Diastereomer 19 and FTY720 (1) were further charac-
terized in rodent models of immunosuppressive efficacy.
In an in vivo mixed lymphocyte reaction,15 1 and 19
reduced the alloantigen stimulated increase in the
draining popliteal lymph node following footpad injec-
tion of Fisher splenocytes into Lewis rats. Mean lymph
node weight ratios of intra-subject allogeneic to synge-
neic splenocyte challenges for 1 and 19 were 1.0
(0.3 mpk/day sc) and 1.6 (0.5 mpk/day sc), respectively,
compared to 3.4 for vehicle controls. In a mouse thyroid
allograft model,16 1 and 19 dose dependently prolonged
graft function. Similar maximal efficacy was achieved
with both compounds at 3 mpk/day, and 1 appeared to
be 3- to 10-fold more potent than 19.
Acknowledgements
The authors would like to thank Dr. George Doss for
conducting and interpreting the data from the NMR
experiments used to assign the structures of 23 and 24.
References and notes
1. Novarik, J. M.; Burtin, P. Expert Opin. Emerging Drugs
2003, 8, 47–62.
2. Kiuchi, M.; Adachi, K.; Kohara, T.; Minoguchi, M.;
Hanano, T.; Aoki, Y.; Mishina, T.; Arita, M.; Nakao, N.;
Ohtsuki, M.; Hoshino, Y.; Teshima, K.; Chiba, K.;
Sasaki, S.; Fujita, T. J. Med. Chem. 2000, 43, 2946–2961.
3. Chiba, K.; Yanagawa, Y.; Masubuchi, Y.; Kataoka, H.;
Kawaguchi, T.; Ohtsuki, M.; Hoshino, Y. J. Immunol.
1998, 160, 5037–5044.
4. Mandala, S.; Hajdu, R.; Bergstrom, 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.; Parsons, W.; Rosen, H. Science 2002, 296,
346–349.
In conclusion, nonhydrolyzable phosphonate analogs of
FTY720 phosphate (2) have been identified as potent
agonists of S1P receptors. While one report describing
S1P receptor agonists has appeared,17 no in vivo char-
acterization of those compounds was disclosed. Sys-
temic administration of the compounds described herein
results in a replication of the primary pharmacodynamic
effect of 1 in rodents, which is a dose-responsive low-
ering of circulating lymphocytes. An analog (19) with an
S1P receptor profile comparable to bioactive metabolite

J. J. Hale et al. / Bioorg. Med. Chem. Lett. 14 (2004) 3351–3355
3355
5. Brinkmann, V.; Davis, M. D.; Heise, C. E.; Albert, R.;
Cottens, S.; Hof, R.; Bruns, C.; Prieschl, E.; Baumruker,
T.; Hiestand, P.; Foster, C. A.; Zollinger, M.; Lynch, K.
R. J. Biol. Chem. 2002, 277, 21453–21457.
6. Durand, P.; Peralba, P.; Sierra, F.; Renaut, P. Synthesis
2000, 505–506.
11. Allende, M.; Dreier, J. L.; Mandala, S.; Proia, R. L. J.
Biol. Chem., 2004, M314291200.
€
12. Graler, M. H.; Goetzl, E. J. FASEB J., 2004, express
article doi:10.1096/fj.03-0910fje, published online January
8, 2004.
13. Mice were dosed intravenously into the tail vein with
0.1 mL of test compound dissolved in vehicle (2% (w/v)
hydroxypropyl-b-cyclodextrin (Cerestar, USA), 0.12 M
NaCl). For screening purposes, lymphocyte lowering was
assessed at 3 h post dose. After rendering each mouse
unconscious by CO2 to effect, the chest was opened,
0.5 mL of blood was withdrawn via direct cardiac punc-
ture, and the blood was immediately stabilized with
EDTA and hematology evaluated using a clinical hema-
tology autoanalyzer calibrated for performing murine
differential counts (H2000, CARESIDE, Culver City,
CA). Reduction in lymphocytes by test compound treat-
ment was established by comparison of hematological
parameters of three mice versus three vehicle treated mice.
14. Plasma compound concentration measurements used to
calculate pharmacokinetic parameters were obtained after
iv administration of test compounds via a cannula that
had been previously implanted in the femoral vein of male
Sprague-Dawley rats (n ¼ 2). Compounds 2 and 6 were
formulated at 1.0 mg/mL in 2% hydroxymethyl-b-cyclo-
dextrin/10 nM Na2CO3. Compounds 19 and 20 were
formulated at 1.0 mg/mL in ethanol/PEG-400/water
(20:30:50, v/v/v).
7. Sahoo, S. P.; Caldwell, C. G.; Chapman, K. T.; Durette, P.
L.; Esser, C. K.; Kopka, I. E.; Polo, S. A.; Sperow, K. M.;
Niedzwiecki, L. M.; Izquierdo-Martin, M.; Chang, B. C.;
Harrison, R. K.; Stein, R. L.; MacCoss, M.; Hagmann, W.
K. Bioorg. Med. Chem. Lett. 1995, 5, 2441–2446.
8. NOE data were collected at 600 MHz in CDCl3 with a
mixing time of 0.3 s. The observed NOEs indicated below
were used to assign the structures of 23 and 24 and are
consistent with those reported for related compounds. See
Shi, Z.; Harrison, B. A.; Verdine, G. L. Org. Lett. 2003, 5,
633–636.
NOE
NOE
H
H
H
H
O
H
H
O
Ar
PO(OCH2CH3)2
Ar
PO(OCH2CH3)2
HN
HN
O
O
23
24
15. Dorsch, S. E.; Roser, B. Aust. J. Exp. Biol. Med. Sci. 1974,
52, 253–264.
16. Talento, A. M.; Nguyen, T.; Blake, A.; Sirotina, C.;
Fioravanti, D.; Burkholder, R.; Gibson, N. H.; Sigal, I.;
Springer, M. S.; Koo, G. C. Transplantation 1993, 55, 418–
422.
17. Clemens, J. J.; Davis, M. D.; Lynch, K. R.; Macdonald,
T. L. Bioorg. Med. Chem. Lett. 2003, 13, 3401–
3404.
9. All test compounds were found to be agonists of human
S1P1, S1P3, S1P4 and S1P5 receptors as evidenced by their
ability to induce levels of GTPcS binding comparable to
S1P. The magnitudes of the calculated EC50 values from
these assays were generally 5-fold of the IC50 values.
10. Matloubian, M.; Lo, C. G.; Cinamon, G.; Lesneski, M. J.;
Xu, Y.; Brinkmann, V.; Allende, M. L.; Proia, R. L.;
Cyster, J. G. Nature 2004, 427, 355–360.