Syntheses of sphingosine-1-phosphate analogues and their interaction with EDG/S1P receptors

Article abstract of DOI:10.1016/j.bmcl.2004.03.001

Sphingosine-1-phosphate (S1P) is an important regulator of a wide variety of biological processes acting as an endogenous ligand to EDG/S1P receptors. In an effort to establish structure-activity relationship between EDG/S1P and ligands, we report herein

Full text of DOI:10.1016/j.bmcl.2004.03.001

Bioorganic & Medicinal Chemistry Letters 14 (2004) 2499–2503
Syntheses of sphingosine-1-phosphate analogues and their
interaction with EDG/S1P receptors
Hyun-Suk Lim,a Jeong-Ju Park,a Kwangseok Ko,b Mee-Hyun Leeb and
Sung-Kee Chunga,*
aDivision of Molecular and Life Sciences, Pohang University of Science and Technology, San31, Hyoja-Dong, Nam-gu,
Pohang 790-784, South Korea
bC&C Research Laboratories, Hwasung 445-976, South Korea
Received 2 February 2004; revised 28 February 2004; accepted 1 March 2004
Dedicated to Prof. Yong Hae Kim on the occasion of his 65th birthday
Abstract—Sphingosine-1-phosphate (S1P) is an important regulator of a wide variety of biological processes acting as an endo-
genous ligand to EDG/S1P receptors. In an effort to establish structure–activity relationship between EDG/S1P and ligands, we
report herein homology modeling study of EDG-1/S1P1, syntheses of S1P analogues, and cell based binding affinity study for EDG/
S1P receptors.
Ó 2004 Elsevier Ltd. All rights reserved.
Sphingosine-1-phosphate (S1P), one of the sphingolipid
metabolites, is known to act as both an extracellular
mediator and an intracellular second messenger. S1P-
activated extracellular effects are mediated via plasma
reported to date. Thus, it is deemed highly desirable to
obtain some key structure–activity relationship (SAR)
for S1P with respect to each individual EDG/S1P
receptor.
membrane
G
protein-coupled receptor EDG/S1P
family, which include EDG-1/S1P1, EDG-3/S1P3, EDG-
5/S1P2, EDG-6/S1P4, and EDG-8/S1P5 subtypes,
whereas specific intracellular targets remain to be iden-
tified.1 S1P-activated EDG/S1P receptors are linked to
diverse biological responses, such as mitogenesis, dif-
ferentiation, migration, and apoptosis, and thus are
believed to be involved in a variety of pathological
conditions including angiogenesis, inflammation, and
cardiovascular diseases, etc.2 Therefore, S1P analogues
with different specificities and affinities for the different
EDG/S1P receptors would be extremely valuable in
studying which receptor subtypes mediate which bio-
logical responses to S1P. More specifically, discovery of
S1P agonists or antagonists might also provide the basis
for development of novel therapeutic agents. However,
the medicinal chemistry of S1P is not yet well developed,
and there are no selective agonists or antagonists
Recently, we reported syntheses of S1P stereoisomers
and derivatives, and their interaction with EDG/S1P
receptors.3 It was shown that
D-erythro forms of S1P
and dihydro-S1P have higher affinities than the other
stereoisomers, indicating that 3D spatial orientations of
key functionalities, that is, the C2-amino and C3-
hydroxyl groups of S1P, are very important for the
specific binding to EDG/S1P receptors. However, the
possible interaction mode between the C3-hydroxyl
group of S1P and EDG/S1P receptors remains to be
defined. In order to shed a further light to these issues,
and to prepare a basis for rational design of potential
agonists and antagonists for EDG/S1P receptors, we
have carried out a study on the computer modeling of
EDG-1/S1P1 docked with S1P, syntheses of a number of
S1P analogues, and their interaction with EDG/S1P
receptors.
Thus, we have built a homology model of EDG-1/S1P1
receptor, and used it in docking studies with S1P to
understand their specific interactions, specifically the
role of the C3-hydroxyl group of S1P.4 The computer
model has shown three different ion-pairing interactions
Keywords: Sphingosine-1-phosphate analogues; EDG/S1P receptors;
Relative binding affinities; Structure–activity relationships.
* Corresponding author. Tel.: +82-54-279-2103; fax: +82-54-279-3399;
e-mail: skchung@postech.ac.kr
0960-894X/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.03.001

2500
H.-S. Lim et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2499–2503
between EDG-1/S1P1 receptor and the S1P head group;
two ion pairs between Arg120 and Arg292 and the anionic
phosphate of S1P; a single ion pair between Glu121 and
the C2-amino group of S1P, observations that are in
accord with the previous computational modeling study
by Parrill et al.5 In addition to these proposed interac-
tions, the fourth interaction has also been noted: the
interactions between the C3-hydroxyl group of S1P and
the backbone carbonyl group of Phe296 and the hydroxyl
diate in the S1P synthesis,3 were protected with TBS and
the product was alkylated with alkyl iodide in the
presence of NaH/DMF to give 11a–c. Selective depro-
tection of C1-O-silyl group was achieved by treatment of
11a–c with HF/pyridine solution at 0 °C, under which
the C3-O-silyl group remained intact.7 Phosphorylation
of compounds 12a–c, followed by deprotection yielded
the desired N-alkyl-S1P 14a–c (Scheme 2).
98
ꢀ
group of Tyr , both within 3.2 A distance (Fig. 1).
The observation that 4,5-dihydro-S1P is less potent than
S1P itself in its binding affinity, suggests that the double
bond somehow plays some important role in the bind-
ing.3 Thus, we substituted phenyl moieties for the C4–5
double bond of S1P. Synthesis of the phenyl-incorpo-
rated S1P analogues was accomplished via compounds
17a–d following the literature method described by
Chun et al.8 Compounds 18a–d were prepared using the
procedure employed for the preparation of 8a from 2a
(Scheme 3).3;9
Since our previous studies3 and the homology modeling
clearly suggest that the C3-hydroxyl group should play
an important role for the specific binding, we have
prepared some C3-deoxy-S1P analogues. Garner alde-
hyde 1a6 was elaborated via Wittig condensation,
hydrogenation, and phosphorylation to provide com-
pounds 8a and 9a. By employing the same procedures
on the enantiomeric aldehyde 1b, the corresponding
enantiomer 8b was also obtained (Scheme 1).
The computational model also suggests that the ali-
phatic tail part of S1P should positively interact with the
hydrophobic area formed by the seven transmembrane
spanning helices in the EDG-1/S1P1. In order to exam-
ine the hydrophobic interaction involving the tail part,
we synthesized a series of S1P derivatives in which the
n-C13H27 aliphatic long chain was replaced by the
shorter chains such as n-C11H23, n-C7H15, cyclohexyl,
and phenyl-containing side chains including 3-decyl-
oxyphenyl, 4-buthoxyphenyl, and 4-benzyloxyphenyl.
The synthesis started with the b-keto-phosphonate 19,
The theoretical 3-D modeling of EDG-1/S1P1 docked
with S1P also suggests that the positively charged
ammonium group of S1P is essential for the specific
binding. Therefore, it was envisioned that N-alkylation
of S1P might possibly differentiate its binding affinity
and specificity to EDG/S1P subtypes. However, N,N-
disubstitution is clearly not desirable because one
hydrogen on the C2-amino group appears necessary to
interact with the carboxylate of Glu121. Thus, the two
hydroxyl groups of N-Boc-sphingosine 10, an interme-
Figure 1. Theoretical binding conformation of
D
-erythro S1P in the EDG-1/S1P1 active site. Hydrogen bonds are indicated as dashed lines.
(CH2)nR1
CHO
(CH2)nR1
HO
i
ii
O
O
NBoc
NBoc
NHBoc
2a: n = 11, R1 = CH3
3a: n = 0, R1 = Ph
4a: n = 11, R1 = CH3
5a: n = 0, R1 = Ph
1a
O
P
O
(CH2)nR1
(CH2)nR1
iii
iv
H3CO
O
HO
P
O-
O
+
OCH3 NHBoc
NH3
6a: n = 11, R1 = CH3
7a: n = 0, R1 = Ph
8a: n = 11, R1 = CH3
9a: n = 0, R1 = Ph
Scheme 1. Reagents and conditions: (i) (a) Ph3PþC15H31Brꢀ/LHMDS for 2a and 2b, Ph3PþBnBrꢀ/LHMDS for 3a, THF, ꢀ78 °C–rt, 91%, (b) H2,
10% Pd/C, EtOAc, rt, 96%; (ii) LiCl, 90% aq AcOH, rt, 87%; (iii) P(OCH3)3, CBr4, pyridine, 0 °C, 84%; (iv) TMSBr, CH2Cl2, 0 °C, 58%.

H.-S. Lim et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2499–2503
2501
OH
OTBS
OTBS
i
ii
HO
C
13H27
TBSO
C13H27
HO
C13H27
NHBoc
NBoc
NBoc
R2
R2
10
11a: R2 = Me
11b: R2 = Et
11c: R2 = n-Pr
12a: R2 = Me
12b: R2 = Et
12c: R2 = n-Pr
OTBS
OH
O
O
iv
iii
H3CO
P
O
C13H27
HO
P
O
C13H27
OCH3 NBoc
R2
OH
NHR2
14a: R2 = Me
14b: R2 = Et
14c: R2 = n-Pr
13a: R2 = Me
13b: R2 = Et
13c: R2 = n-Pr
Scheme 2. Reagents and conditions: (i) (a) TBSCl, imidazole, DMF, rt, quant, (b) alkyl iodide, NaH, DMF, rt, 89–91%; (ii) HF, pyridine, THF,
0 °C, 63–88%; (iii) P(OCH3)3, CBr4, pyridine, 0 °C, quant; (iv) (a) TBAF, THF, rt, 60–65%, (b) TMSBr, CH2Cl2, rt, 43–54%.
OH
O
CHO
iii
i
ii
R3
R3
O
O
O
iv
O
NBoc
1a
OH
NBoc
NBoc
16a-d
15a-d
OH
O
a : R3 = m-C12H25
b : R3 = p-C12H25
c : R3 = m-OC11H23
d : R3 = p-OC11H23
HO
P O
R3
R3
NBoc
17a-d
OH
NH2
18a-d
Scheme 3. Reagents and conditions: (i) n-BuLi, aryl bromide, THF, ꢀ42 °C, 60–80%; (ii) PCC, CH2Cl2, rt, 80–87%; (iii) DIBAL-H, THF, 0 °C, 93–
98%; (iv) (a) LiCl, 90% aq AcOH, rt, 87–92%, (b) P(OCH3)3, CBr4, pyridine, 0 °C, 79–90%, (c) TMSBr, CH2Cl2, 0 °C, 40–58%.
which was an intermediate in the earlier sphingosine
synthesis in our laboratory.10 The Horner–Wadsworth–
Emmons (HWE) condensation of the phosphonate 19
with various aldehydes provided the corresponding
enones 20a–f in good yields. Removal of the two protect-
ing groups (N-trityl and O-TBS) of 20a–f with 2 N HCl
in MeOH–THF under reflux gave 3-ketosphingosine
derivatives 21a–f. Enones 21a–f were selectively reduced
with Zn(BH4)2 to afford the anti form of sphingosine
analogues 22a–f.10 N-Boc protection of 22a–f, followed
by phosphorylation and deprotections provided the
desired tail-modified S1P analogues 25a–f (Scheme 4).
The binding affinities of the synthetic S1P analogues
were evaluated in vitro for the EDG/S1P receptors by
measuring their ability to displace radioligand, [3H]-S1P
from EDG-1/S1P1, EDG-3/S1P3, and EDG-5/S1P2,
which were expressed in CHO cells.11;12 The results are
shown in Figure 2. Among the C3-deoxy-S1P analogues
having an aliphatic or aromatic side chain, compound
8a having (2R)-configuration and 16 carbon chain
length displayed selective binding affinities for EDG-3/
S1P3 (82%) and EDG-5/S1P2 (84%), ca. three times
higher affinity than the 18 carbon analogue.3 The
enantiomeric compound 8b did not show much of the
O
O
O
O
i
ii
P(OMe)2
TBSO
R4
HO
R4
TBSO
NH2.HCl
NHTr
NHTr
20a-f
21a-f
19
OH
OH
OH
O
iii
v
iv
H3CO
P
O
R4
HO
R4
HO
R4
OCH3 NHBoc
NH2
22a-f
NHBoc
24a-f
23a-f
a : R4 = n-C11H23
b : R4 = n-C7H15
OH
O
P
vi
HO
O
R4
d : R4
=
=
c : R4
e : R4
=
=
OH
NH2
25a-f
OBn
OC10H21
f : R4
OC4H9
Scheme 4. Reagents and conditions: (i) For the reaction with aliphatic aldehydes, R3CHO, DBU, LiCl, THF, rt, 90–95%, for the reaction with
aromatic aldehydes, R3CHO, NaH, THF, rt, quant; (ii) 2 N HCl, THF–MeOH, 40 °C, 80–85%; (iii) Zn(BH4)2, THF, ꢀ78 °C, 75%; (iv) Boc2O, 1 N
NaOH, dioxane–H2O, rt, 85–93%; (v) P(OCH3)3, CBr4, pyridine, 0 °C, 73–85%; (vi) TMSBr, CH2Cl2, rt, 43–54%.

2502
H.-S. Lim et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2499–2503
120
100
80
60
40
20
0
S1P 8a
8b
9a 14a 14b 14c 18a 18b 18c 18d 25a 25b 25c 25d 25e 25f
Compounds
EDG-1
EDG-3
EDG-5
Figure 2. Relative binding affinity of S1P derivatives on specific binding of
D
-erythro S1P to EDG-1, -3, and -5 receptors. CHO cells transfected with
Edg-1, -3, and -5, respectively, were incubated in the presence of 2 nM [3H]-S1P with or without 1 lM of the indicated compounds. Results are
means ꢁ standard deviation of duplicate determinations.
binding affinity, indicating that 3D configuration of C2-
amino group is important even in the C3-deoxy-S1P
analogues.
group, n-C7H15 (25b) substantially reduced binding
affinities, especially for EDG-1/S1P1 subtype (31% for
EDG-1/S1P1, 52% for EDG-3/S1P3, and 63% for EDG-
5/S1P2, respectively). The replacement of the n-C13H27 in
S1P with cyclohexyl (25c), 4-benzyloxyphenyl (25d),
3-decyloxyphenyl (25e), or 4-buthoxyphenyl (25f)
resulted in marked decreases of the binding affinity. These
results suggest that the hydrophobic pocket located in the
transmembrane helices region of EDG/S1P receptors is
somewhat narrow and long, and the hydrophobic inter-
action plays an important role for efficient binding. Thus,
seemingly minor structural changes in the tail part of S1P
can have substantial consequences in the interaction with
the hydrophobic pocket of the receptors.
Very interestingly, N-Me–S1P 14a and N-Et–S1P 14b
showed highly selective affinity for EDG-1/S1P1 (55.1%
and 51.4%, respectively), and they were completely
nonbinding to EDG-3/S1P3 and EDG-5/S1P2, whereas
N-Pr–S1P 14c showed negligible affinity to all subtypes.
On the basis of the 3-D modeling of EDG-1/S1P1
docked with S1P, it was initially thought that there was
not enough room for the N-alkyl group to occupy.
However, the flexible movement of the active site may
allow the N-methyl or N-ethyl group, although the space
surrounding the C2-amino group of S1P appears rela-
tively tight; N-propyl–S1P may be just too bulky to be
accommodated. None of the N-alkyl substituted S1P
analogues was found to bind to EDG-3/S1P3 or EDG-5/
S1P2. The high selectivity of N-Me–S1P 14a and N-
Et–S1P 14b might be significant in the sense that it can
provide a direction to find selective agonists or antago-
nists to EDG-1/S1P1 subtype for the first time.
In summary, we have carried out a homology modeling
study of EDG-1/S1P1, prepared a series of S1P ana-
logues, and evaluated their binding affinity for EDG/
S1P receptors. Based on these results, it might be con-
cluded that (1) the C3-hydroxyl group plays important
role in binding to EDG/S1Ps, specifically to EDG-1/
S1P1, (2) the tail portion (olefin and n-C13H27) of S1P
can be replaced by m-alkylphenyl group without much
impact on the binding affinity, (3) N-Me–S1P and N-Et–
S1P show selective binding affinity for EDG-1/S1P1,
thus providing a possibility to find selective agonists or
antagonists to EDG-1/S1P1 for the first time, and (4) a
fine-tuning of the long alkyl chain may be possible in
terms of the selectivity and the absolute affinity. These
findings are expected to substantially contribute to the
development of potent and selective agonists and
antagonists for EDG/S1P receptors. Further works
along these lines are currently in progress.
In the case of substituted phenyl-incorporated com-
pounds, compound 18a containing m-C12H25 substituted
phenyl moiety showed almost the same affinities to S1P,
albeit nonselectively for three EDG/S1P subtypes,
whereas p-substituted one 18b had much weaker affini-
ties. This observation demonstrates that C4–5 double
bond could be replaced with a phenyl group substituted
with m-alkyl group without decrease of affinities for
EDG/S1Ps.13 Compounds 18c–d with a substituted
alkoxy group showed relatively weak affinities.
We have also tested the tail-modified S1P analogues.
First, the aliphatic long chain of S1P, n-C13H27 was
replaced with a shorter chain such as n-C11H23 or n-C7H15.
Compound 25a having n-C11H23 was found to have
equipotent affinities for S1P (115% for EDG-1/S1P1,
83% for EDG-3/S1P3, and 103% for EDG-5/S1P2,
respectively). However, introducing a shorter alkyl
Acknowledgements
We thank Professor P. G. Suh of POSTECH for pro-
viding CHO cells. Financial support from the Ministry
of Education/BSRI Fund is gratefully acknowledged.

H.-S. Lim et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2499–2503
2503
10. (a) Chung, S. K.; Lee, J. M. Tetrahedron: Asymmetry
1999, 10, 1441; (b) Lee, J. M.; Lim, H. S.; Chung, S. K.
Tetrahedron: Asymmetry 2002, 13, 343.
11. Murata, N.; Sato, K.; Kon, J.; Tomura, H.; Okajima, F.
Anal. Biochem. 2000, 282, 115.
References and notes
1. (a) Spiegel, S.; Milstien, S. FEBS Lett. 2000, 476, 55; (b)
Payne, S. G.; Milstien, S.; Spiegel, S. FEBS Lett. 2002, 531, 54.
2. (a) Lee, M. J.; Thangada, S.; Claffey, K. P.; Ancellin, N.;
Liu, C. H.; Kluk, M.; Volpi, M.; Sha’afi, R. I.; Hla, T. Cell
1999, 99, 301; (b) Pyne, S.; Pyne, N. Pharmacol. Ther.
2000, 88, 115; (c) Levade, T.; Auge, N.; Veldman, R. J.;
Cuvillier, O.; Negre-Salvayre, A.; Salvayre, R. Circ. Res.
2001, 89, 957; (d) Hla, T. Pharmacol. Res. 2003, 47, 401.
3. Lim, H. S.; Oh, Y. S.; Suh, P. G.; Chung, S. K. Bioorg.
Med. Chem. Lett. 2003, 13, 237.
4. The homology model of EDG-1/S1P1 was developed using
Modeller4 (Laboratory of Molecular Biophysics in Rocke-
feller University, New York, USA) and Insight II program
(Accelrys, San Diego, CA, USA) on a Silicon Graphics
Octane workstation (1.2GB RAM, IRIX 6.5). The EDG-1/
S1P1 sequence was obtained from GenBanke
(AAF43420), and the bovine rhodopsin model composed
of seven transmembrane spanning helices (PDB code 1boj)
was used as a template structure for homology modeling.14
5. Parrill, A. L.; Wang, D.; Bautista, D. L.; Brocklyn, J. R.
V.; Lorincz, Z.; Fischer, D. J.; Baker, D. L.; Liliom, K.;
Spiegel, S.; Tigyi, G. J. Biol. Chem. 2000, 275, 39379.
6. Liang, X.; Andersch, J.; Bols, M. J. Chem. Soc., Perkin
Trans. 1 2001, 2136.
12. Binding assay protocol: [3H]-S1P was purchased from
American Radiolabeled Chemicals, Inc. CHO cells (CHO/
Edg-1, CHO/Edg-3, CHO/Edg-5, and CHO/Mock) in
confluent six multiplates were washed twice with the ice-
cold binding buffer consisting of 20 mM Tris–HCl
(pH 7.5), 100 mM NaCl, 15 mM NaF, and 0.4% (w/v)
BSA, and then incubated with the same buffer containing
2 nM [3H]-S1P (about 40,000 dpm per well) and 1 lM of
S1P or S1P derivatives in a final volume of 1.0 mL. The
plates were kept in ice for 90 min, and the cells
were washed twice with the same ice-cold binding buffer
to remove unbounded ligand. The cells were solubilized
with the solubilizing solution composed of 0.1% SDS,
0.4% NaOH, and 2% Na2CO3, and the radioactivity
was measured by a liquid scintillation counter after the
addition of scintillation cocktail solution. The relative
binding affinities of each compound to each EDG/S1P
receptor were presented as percentage to S1P.
13. (a) Koide, Y.; Hasegawa, T.; Takahashi, A.; Endo, A.;
Mochizuki, M.; Nakagawa, M.; Nishida, A. J. Med.
Chem. 2002, 45, 4629; (b) Clemens, J. J.; Davis, M. D.;
Lynch, K. R.; Macdonald, T. L. Bioorg. Med. Chem. Lett.
2003, 13, 3401.
14. The rhodopsin model (PDB code 1boj) used as the
template in this study is not based on a crystal structure
but rather a theoretical model. Because of the potential
problems associated with this discrepancy, a homology
modeling study based on a rhodopsin crystal structure is in
progress.
7. Nicolaou, K. C.; Hepworth, D.; King, N. P.; Raymond,
M.; Finlay, V.; Scarpelli, R.; Manuela, M.; Pereira, A.;
Bollbuck, B.; Bigot, A.; Werschkun, B.; Winssinger, A.
Chem. Eur. J. 2000, 6, 2783.
8. Chun, J.; He, L.; Byun, H. S.; Bittman, R. J. Org. Chem.
2000, 65, 7634.
9. Szulc, Z. M.; Hannun, Y. A.; Bielawska, A. Tetrahedron
Lett. 2000, 41, 7821.