Sphingosine 1-Phosphate Analogs as Receptor Antagonists*

Sphingosine 1-phosphate (S1P) is a lysophospholipid mediator that evokes a variety of cell and tissue responses via a set of cell surface receptors. The recent development of S1P receptor agonists, led by the immunomodulatory pro-drug FTY720, has revealed that S1P signaling is an important regulator of lymphocyte trafficking. With the twin goals of understanding structure-activity relationships of S1P ligands and developing tool compounds to explore S1P biology, we synthesized and tested numerous S1P analogs. We report herein that a subset of our aryl amide-containing compounds are antagonists at the S1P1 and S1P3 receptors. The lead compound in series, VPC23019, was found in broken cell and whole cell assays to behave as a competitive antagonist at the S1P1 and S1P3 receptors. The structure-activity relationship of this series is steep; for example, a slight modification of the lead compound resulted in VPC25239, which was one log order more potent at the S1P3 receptor. These new chemical entities will enable further understanding of S1P signaling and provide leads for further S1P receptor antagonist development.

Sphingosine 1-phosphate (S1P) 1 is a lysophospholipid mediator that evokes a variety of cellular responses by stimulation of five members of the endothelial cell differentiation gene receptor family. The endothelial cell differentiation gene receptors are G-protein coupled receptors that, upon stimulation, propagate second messenger signals via activation of heterotrimeric G-protein ␣ subunits and ␤-␥ dimers. Ultimately, this S1P-driven signaling results in cell survival, increased cell migration, and, often, mitogenesis. The recent development of agonists targeting S1P receptors has provided insight regarding the role of this signaling system in physiologic homeostasis. For example, the immunomodulator FTY720 (2-amino-2-[2-(4octylphenyl)ethyl]propane-1,3-diol), which is a pan S1P receptor agonist following phosphorylation, revealed that S1P tone influences lymphocyte trafficking (1)(2)(3)(4). The utility of an S1P receptor agonist was unexpected; indeed, prior speculation focused on the potential (as yet unrealized) for S1P antagonists as anti-angiogenic agents.
Recent findings also suggest a physiological influence for S1P in the vasculature.
Although not yet explored in detail, it has been hypothesized that S1P may exert anti-inflammatory actions on endothelial cells through its release from high density lipoprotein (5). A recent study found that S1P inhibited tumor necrosis factor ␣-mediated monocyte-endothelial cell adhesion. 2 Furthermore, an S1P 1 receptor antagonist described herein blocked the anti-inflammatory action of S1P, thereby providing evidence that this effect maps to the S1P 1 receptor. If verified, these results would expand the role of the S1P 1 receptor to include influencing monocyte extravasation and would further highlight how the development of S1P receptor-specific compounds is expanding our understanding of the biology of this important signaling system. To further characterize the biology associated with individual S1P receptors, we have undertaken a program to develop S1P analogs with the twin goals of expanding the structureactivity relationships associated with S1P receptor interactions and identifying receptor-specific compounds. Our studies have led to the identification of a series of S1P analogs that behave as antagonists at two of the five S1P receptors.

EXPERIMENTAL PROCEDURES
Materials-Chemicals for syntheses were purchased from Aldrich, Sigma, Advanced ChemTech Chemical Company (Louisville, KY), and/or NovaBiochem and were used without further purification.
[␥-32 P]ATP and ␥-35 S-GTP were purchased from Amersham Biosciences. CyQuant cell proliferation assay kit and Fluo-4AM calcium indicator were purchased from Molecular Probes (Eugene, OR). Chinese hamster ovary and T24 cells were purchased from the American Type Culture Collection (Manassas, VA). HEK293T cells were a gift from Dr. Judy White (Department of Cell Biology, University of Virginia). Tissue culture media and normal fetal bovine serum (FBS) were obtained from Invitrogen. Charcoal/dextran-stripped FBS was obtained from Gemini Bio-Products (Woodland, CA). G-protein ␣, ␤, and ␥ DNAs were a gift from Dr. Doug Bayliss (Dept. of Pharmacology, University of Virginia). Sphingosine 1-phosphate was purchased from Avanti Polar Lipids (Alabaster, AL).
Syntheses of VPC23019, VPC23031, VPC25239, and VPC23089 -The synthetic route to the meta-substituted compounds VPC23031, VPC25239, and VPC23019 was initiated with a Sonogashira coupling (6) of 3-iodo-1-nitrobenzene with the appropriate terminal alkyne. The resulting adducts were then subjected to simultaneous hydrogenation of the nitro group and the triple bond to generate the meta-substituted anilines. The anilines were next coupled to a protected serine, and the ensuing amides underwent hydrogenolysis to afford the free alcohols. The alcohols were subsequently phosphorylated, oxidized with hydrogen peroxide, and then subjected to acid-catalyzed global deprotection to provide the final products, VPC23031, VPC25239, and VPC23019. Synthesis of the ortho-substituted compound, VPC23089, commenced with the union of 2-iodoaniline and 1-octyne via a Sonogashira coupling. The ensuing aniline was then coupled to a protected serine utilizing the benzotriazol-1-yl-oxytripyrrolidino-phosphonium hexafluorophosphate reagent. The resulting amide was then subjected to a hydrogenation/ hydrogenolysis step to remove the benzyl ether-protecting group and simultaneously reduce the aryl triple bond. The liberated alcohol was next phosphorylated, oxidized with hydrogen peroxide, and then sub-jected to acid-catalyzed global deprotection to provide the final product, VPC23089. NMR and mass spectrometry were used to confirm all structures. VPC23019 will be available from Avanti Polar Lipids.
Transient Expression in HEK293T Cells-The appropriate receptor plasmid DNA (encoding human S1P 1 , human S1P 2 , human S1P 3 , human S1P 4 , human S1P 5 , human LPA 1 , human LPA 2 , or human LPA 3 receptors) was mixed with equal amounts of expression plasmids encoding human G i2 ␣ (for S1P 3 , a mutated (C352F) rat G i2 ␣ was used), cow ␤ 1 , and cow ␥ 2 proteins, and these DNAs were used to transfect monolayers of HEK293T cells (where "T" indicates expression of the SV-40 virus large T antigen) using the calcium phosphate precipitate method (7). After about 60 h, cells were harvested, and membranes were prepared, aliquoted, and stored at Ϫ70°C until use (8). Transfection of receptor and G-protein was confirmed with the ␥-35 S-GTP binding assay (described below), as analysis of HEK293T cells transfected with G-proteins alone did not respond to agonist stimulation (see Fig.  2A, inset).
Stable Expression in T24 Cells-T24 cell monolayers were cotransfected with the human S1P 1 , S1P 2 , and S1P 3 receptor-encoding DNAs and the pIRESpuro2 plasmid DNA (Clontech, San Jose, CA) using either Lipofectamine 2000 (Invitrogen) or FuGENE 6 (Roche Applied Science). Clonal populations expressing the puromycin acetyltransferase gene were selected by addition of puromycin (Sigma-Aldrich) to the culture medium. T24 cells were grown in monolayers at 37°C in a 5% CO 2 /95% air atmosphere in growth medium consisting of 95% Dulbecco's modified Eagle's medium/F-12 medium and 10% charcoal/dextran-stripped FBS.
Cell Migration Assay-Cell migration assays were performed using modified Boyden chambers (tissue culture-treated with a 24-mm diameter, a 0.010-mm thickness, and 0.008-mm pores, Transwell®, Costar Corp., Cambridge, MA) containing polycarbonate membranes that were coated on the underside with 0.1% gelatin. The underside of the polycarbonate membranes was rinsed once with migration medium (Dulbecco's modified Eagle's medium/F-12 without Phenol Red and 0.1% fatty acid-free BSA) and then immersed in the lower chamber containing 2 ml of migration medium. T24 cells transfected stably with human S1P 1 receptor DNA were grown in Dulbecco's modified Eagle's medium/ F-12 medium containing charcoal/dextran-stripped FBS and 0.010 mg/ml puromycin to 100% confluence in 150 ϫ 25-mm tissue culture plates and serum starved for at least 12 h. Serum-starved cells were removed from culture dishes with 10ϫ trypsin-EDTA (Invitrogen), washed once with migration medium, and resuspended in migration medium (10 6 cells/ml). One milliliter of the cell suspension was added to the upper migration chamber while the S1P agonist VPC22277 (10 nM) was added to the lower chamber. Cells were allowed to migrate to the underside of the membrane for 4 h at 37°C in the presence or absence of antagonist (VPC23019 (0 -1000 nM), VPC23019, VPC23031, VPC23089, and VPC25239 (50 nM each)), which were added to the lower chamber. The migrated cells attached to the bottom surface of the membrane were removed with 10ϫ trypsin-EDTA, their mass was determined by combining 0.1 ml of cell suspension with an equal volume of CyQuant dye solution (3.0 ml of 2ϫ lysis buffer and 0.015 ml of CyQuant dye), and the resulting fluorescence was quantified using the FlexStation TM fluorimeter (Molecular Devices, Menlo Park, CA). Each determination represents the average of two individual migration chambers. For determination of the reversibility of the antagonism associated with VPC23019, cells were incubated with 0.01 mM VPC23019 at 37°C for 30 min. The monolayer was washed three times with phosphate-buffered saline and processed immediately for the cell migration assay, as described above.
Measurement of Intracellular Calcium Mobilization-A FlexStation TM fluorimeter was used to measure intracellular calcium in native T24 cells and T24 cells stably transfected with either human S1P 2 or human S1P 3 receptor DNA. Cells were seeded (ϳ50,000 cells/well) in 96-well, clear bottom black microplates (Corning Costar Corp.) and left overnight at 37°C. The cells were dye-loaded with 0.004 mM Fluo-4AM ester in a loading buffer (Hanks' balanced salt solution, pH 6.4, containing 20 mM HEPES, 0.1% fatty acid-free BSA, and 2.5 mM probenecid) for 30 min at 37°C. After washing cell monolayers three times with phosphate-buffered saline, loading buffer was added, and the cells were exposed to sets of compounds for 3 min at 25°C in the FlexStation TM . In all cases, each concentration of every compound was tested at least in triplicate. For determination of the reversibility of the antagonism associated with VPC23019 (0.010 mM), the compound was added in combination with loading dye to the cells and incubated at 37°C for 30 min. The cells were washed with phosphate-buffered saline and exposed to compounds immediately, as described above.
Determination of the Binding Constant for VPC23019 at the S1P 1 and S1P 3 Receptors-The binding constant (K b ) for VPC23019 at the S1P 1 and S1P 3 receptors was determined by Schild analyses from curves that were fitted using the nonlinear regression method discussed by Lew and Angus (9). Briefly, nonlinear analysis of the best fit line generated by plotting the negative log of the EC 50 values obtained from agonist dose-response curves, in the absence and presence of varying concentrations of antagonist, was plotted against the concentration of antagonist to give the K b value. An F-test analysis was also performed to establish whether the antagonist did or did not meet the criteria of a simple competitive interaction. S1P Radiolabeling-[ 32 P]S1P was prepared by incubating sphingosine and [␥-32 P]ATP with cell lysate from HEK293T cells transfected transiently with human sphingosine kinase type 2 DNA. The 0.2-ml reaction contained 0.025 mM sphingosine, 1 mCi of [␥-32 P]ATP (7000 Ci/mmol), and kinase buffer (10 mM Mg(C 2 H 3 O 2 ) 2 in 50 mM Tris, pH 7.5, 10 mM NaF, and 2 mM semicarbizide). The reaction was initiated by the addition of 0.02 mg of cell lysate protein and incubated at 37°C for at least 30 min. The [ 32 P]S1P was extracted by the addition of 1 N HCl and 2.0 M each KCl, methanol, and chloroform to the reaction mixture. The mixture was then vortexed and centrifuged at 1000 ϫ g for 5-10 min. The organic layer was isolated, and the extraction procedure was repeated two additional times with the remaining aqueous fraction. The combined organic fractions were dried under a stream of nitrogen gas and resuspended in aqueous 0.1% fatty acid-free BSA. The specific activity of the product, [ 32 P]S1P, is estimated to be that of the radiolabeled substrate, [␥-32 P]ATP, i.e. 7000 Ci/mmol.
[ 32 P]S1P Binding Assay-Membranes containing 0.005 mg of protein from HEK293T cells transfected transiently with both receptor and G-protein DNAs were incubated in 0.5 ml of binding buffer (50 mM HEPES, 100 mM NaCl, 10 mM MgCl 2 , pH 7.5), 50 pM [ 32 P]S1P, and the indicated lipid(s) for 1 h at room temperature. Bound ligand was separated from free ligand by rapid filtration and analyzed in a liquid scintillation counter. Nonspecific binding was determined as residual binding of radioligand in the presence of excess S1P to membranes, both heat-denatured and non-heat-denatured, from HEK293T cells transfected transiently with receptor and G-protein DNAs; it was typically 60% of total binding. The binding constant (K i ) associated with the ligand-receptor interaction was determined from the IC 50 using the Chang-Prusoff equation (K i ϭ IC 50 /(1 ϩ [L]/K d ). In applying this equation, the concentration of radioligand (L) is 0.05 nM and the K d value used was that reported for the S1P-S1P 1 receptor interaction, i.e. 8.1 nM (10).
Statistical Analysis-The EC 50 and IC 50 values for all dose response curves were determined by nonlinear regression analysis of all data using the Graphpad Prism program. The error associated with the data collected is reported as the standard error of the mean (S.E.).

VPC23019 Is
Devoid of Agonism at the S1P 1 and S1P 3 Receptors-In the course of our examinations of S1P analog structure-activity relationship, we discovered that the aryl-amide compound VPC23019 (Fig. 1) lacked agonist activity at the S1P 1 ( Fig. 2A) and S1P 3 (Fig. 2B) receptors in a broken cell ␥-35 S-GTP binding assay. Indeed, a profile suggesting inverse agonism was observed at VPC23019 concentrations greater than 100 nM in this assay; however, the alternate explanation, such as a neutral antagonist blocking an endogenous agonist in our preparation, cannot be discounted from present data. The lack of agonist activity was confirmed in whole cell assays using T24 cells (derived from bladder carcinoma) in which either the S1P 1 (T24-S1P 1 ) or S1P 3 (T24-S1P 3 ) receptor was expressed stably. Reverse transciptase-PCR analysis detected only S1P 2 receptor mRNA endogenously in T24 cells; however, although the efficacy of the response was the same, S1P was at least 100-fold more potent when recombinant S1P 1 , S1P 2 (data not shown), or S1P 3 (Fig. 2D) receptors were expressed stably. Thus, the T24 cell system provides an opportunity to interro-gate individual S1P receptors introduced by transfection. Numerous studies have demonstrated that S1P can promote cell migration; however, it was shown recently that S1P can also inhibit migration, possibly through stimulation of the S1P 2 receptor (11). To circumvent this inhibitory effect, we used VPC22277 (Fig. 1), an S1P analog that is an agonist at the S1P 1 and S1P 3 receptors, but not the S1P 2 receptor (Table I). (All of the compounds in the aryl amide series (12) are devoid of detectable activity at the S1P 2 receptor.) We found that migration of T24-S1P 1 cells could be induced by VPC22277, whereas no migration was evoked in response to VPC23019 (Fig. 2C). Similarly, in whole cell calcium mobilization studies using T24-S1P 3 cells, dose-dependent calcium mobilization was observed with S1P (Fig. 2D). Thus, VPC23019 is devoid of agonist activity at the S1P 1 and S1P 3 receptors using both broken cell and whole cell assays.
VPC23019 Blocks Agonist Activity at the S1P 1 and S1P 3 Receptors-The finding that VPC23019 exhibited possible inverse agonist activity at the S1P 1 or S1P 3 receptors prompted us to investigate whether this compound blocked agonist activity. Using the ␥-35 S-GTP binding assay, we found that S1P concentration effect curves generated in the presence of VPC23019 at either the S1P 1 (Fig. 3A) or S1P 3 (Fig. 3B) recep-tors produced a concentration-dependent, parallel rightward shift in the curves. This shift in agonist-mediated responses was also observed in two whole cell assays, cell migration (Fig.  3C) and calcium mobilization (data not shown). VPC23019 neither exhibited agonist activity at the LPA 1-3 endothelial cell differentiation gene family receptors at concentrations up to 0.03 mM nor blocked the action at these sites (data not shown).
VPC23019 S1P Receptor Affinity-Schild analyses of the antagonist activity associated with VPC23019 in the ␥-35 S-GTP binding assay gave pK b values at the S1P 1 (Fig. 3A) and S1P 3 (Fig. 3B) receptors of 7.49 Ϯ 0.15 and 5.98 Ϯ 0.08, respectively. Additionally, the nonlinear regression method of Lew and Angus (9), which predicts whether a compound behaves as a competitive antagonist, suggested that VPC23019 behaves as a competitive antagonist at both receptors. Schild analysis of calcium mobilization in T24-S1P 3 cells gave a K b value for VPC23019 that was ϳ10-fold less than that observed in the ␥-35 S-GTP binding assay. However, the nonlinear regression analysis indicated that VPC23019 did not behave as a competitive antagonist in the calcium mobilization assay with T24-S1P 3 cells. Importantly, both whole cell S1P receptor assays (Ca 2ϩ mobilization (Fig. 3D, S1P 3 ) and cell migration (S1P 1 , not shown)) recovered fully after washing out the antagonist. FIG. 2. VPC23019 lacks agonist activity at the S1P 1 and S1P 3 receptors. HEK293T cells were transfected transiently with equal amounts of human S1P 1 or S1P 3 receptor and G i2 ␣, G␤ 1 , and G␥ 2 plasmid DNAs. Membranes were collected after 60 h. Receptor activation was determined using a broken cell binding assay measuring the binding of ␥-35 S-GTP to the membrane as a function of lipid concentration. Concentration-dependent stimulation of S1P 1 (A) and S1P 3 (B) receptors was observed with S1P (filled circles) but not VPC23019 (open circles). When receptor plasmid DNA was excluded, no significant binding of ␥-35 S-GTP was observed with 0.010 mM S1P (A, inset), which demonstrates that the activity is a function of receptor expression. Binding of ␥-35 S-GTP was observed with 0.010 mM S1P in HEK393T cells transfected transiently with only receptor and G i2 ␣ plasmid DNA; however, the response was at least 3-fold less than that of cells where both receptor and all three G-protein plasmid DNAs were added (A, inset). Data points are in triplicate and are representative of two independent experiments. The percent activation is based on normalization of disintegrations/min values obtained from the minimum and maximum S1P concentration. Typical values for 0 and 100% binding were ϳ300 and 3000 dpm/well, respectively, for both the human S1P 1 and S1P 3 receptors. C, migration of T24 cells transfected stably with human S1P 1 receptor was observed with the S1P 1 agonist VPC22277 (10 nM) but not VPC23019 (1-1000 nM). Data points are in duplicate and are representative of two independent experiments. The percent migrating cells is based on normalization of relative fluorescence unit (RFU) values obtained from the RFU values when migration was observed with 0.1% BSA carrier (minimum) and VPC22277 (maximum). Typical values for 0 and 100% migration were ϳ30,000 and 100,000 RFU/well, respectively. D, concentration-dependent calcium mobilization of untransfected T24 Thus, the antagonist activity exhibited by VPC23019 is reversible as well as fully surmountable, which are essential criteria for a competitive antagonist.
To measure the affinity of VPC23019 for the S1P 1 and S1P 3 receptors directly, we examined the ligand-receptor interaction associated with the S1P 1 and S1P 3 receptors via a receptor binding assay using [ 32 P]S1P in competition with S1P and VPC23019. Analysis of S1P in the radioligand binding assay (Fig. 4A) yielded pK i values of 8.96 Ϯ 0.14 and 8.12 Ϯ 0.06 at the S1P 1 and S1P 3 receptors, respectively. These values are in agreement with the published pK d values for radiolabeled S1P binding to these receptors (10,13,14). The radioligand binding assay also revealed an excellent correlation between the pK i and the pK b for VPC23019 generated from the Schild analysis at both the S1P 1 and S1P 3 receptors, i.e. pK ii values of 7.86 Ϯ 0.16 and 5.93 Ϯ 0.19, respectively ( Fig. 4B and Table II). Finally, VPC23019 was also found to be devoid of agonist activity at the S1P 2 receptor, and radioligand binding studies with the S1P 2 receptor revealed that VPC23019 did not influence the binding of [ 32 P]S1P to the S1P 2 receptor at concentrations up to 0.010 mM (data not shown).
Changes in Analog Structure Alter Activity at the S1P 1 and S1P 3 Receptors-Previous studies with para-substituted aryl amide compounds (VPC23019 is meta-substituted, Fig. 1) revealed that they are agonists at all of the S1P receptors, except at the S1P 2 receptor, where they are inactive (12). Furthermore, different potencies and efficacies were observed with changes in either the length of the longest alkyl chain or the spatial configuration about the amino carbon (12). It is reasonable, therefore, to expect that such changes would also influence antagonist activity. Thus, we synthesized several structural analogs of VPC23019 and tested the effect on potency at either the S1P 1 or S1P 3 receptor.
If the primary amine of VPC23019 is placed in the opposite spatial configuration (VPC25027), the compound is at least 1 log order less potent than VPC23019 as an antagonist at the S1P 1 and S1P 3 receptors (data not shown). Extension of the alkyl chain located at the third phenyl carbon (meta to the amide) beyond eight carbon atoms resulted in a switch from antagonist to agonist activity at both the S1P 1 and S1P 3 receptors (Table I). Conversely, shortening of the alkyl chain by one (VPC25239) or two (VPC23031) carbon atoms or moving it to the second phenyl carbon (ortho-VPC23089) resulted in an antagonist profile similar to that observed previously with VPC23019 at both the S1P 1 and S1P 3 receptors (data not shown). Furthermore, these analogs blocked agonist-mediated migration (Fig. 5A) and calcium mobilization (Fig. 5B) in T24 cells stably expressing either the S1P 1 or S1P 3 receptor, respectively. None of these synthetic maneuvers resulted in an antagonist with improved potency except for VPC25239 (Table  II), which was equal in potency to the lead compound at the S1P 1 receptor but about 1 log order more potent at the S1P 3 receptor (Table II). Thus, VPC25239 is an equipotent S1P 1 / S1P 3 receptor antagonist, whereas the other antagonists in the series are more potent at the S1P 1 receptor. The affinity of VPC23031 observed at the S1P 3 receptor (Table II) does not appear to correlate with the ability of the compound to effectively shift the agonist-mediated activation of the receptor (Fig.  5B). The reason for this disconnect is unclear to us presently.
As observed previously with VPC23019, all of the compounds in this series behaved as agonists at the S1P 4 and S1P 5 receptors, and no agonist activity was observed with the S1P 2 receptor (Table I). Finally, modification of the phosphate head group of VPC23019 (e.g. phosphonate) resulted in compounds with an agonist and antagonist activity profile similar to that observed with VPC23019. 3

DISCUSSION
Understanding of the physiological role of the S1P receptors has been greatly enhanced by using sphingosine-and S1Prelated analogs. An example of this is FTY720, a sphingosine-like lipid that has been shown to prolong allograft survival (15)(16)(17)(18)(19)(20)(21)(22) and is efficacious in autoimmune disease models such as experimental autoimmune encephalomyelitis (1,23) and the non-obese diabetic mouse (24,25). Recent reports demonstrate that FTY720 is a pro-drug; the immunomodulatory effect associated with its administration requires phosphorylation (1,2,26). The resultant S1P analog generated (FTY720-phosphate) was found to be a potent agonist at all of the S1P receptors except the S1P 2 receptor (1, 2). Furthermore, studies performed with the S1P 1 receptorselective agonist SEW2871 (5-(4-phenyl-5-trifluoromethylthiophen-2-yl)-3-(3-trifluoromethyl-phenyl)-(1,2,4)oxadiazole) suggest that lymphopenia, which is an index of FTY720 action, is a result of activation of the S1P 1 receptor (4). Furthermore, lymphocyte egress from thymus and peripheral lymphatic tissues is dependent on the S1P 1 receptor (3). Conversely, the toxicity associated with S1P agonist administration in rodents (27,  cells (inset) and T24 cells transfected stably with human S1P 3 receptor was observed with S1P (filled circles) but not VPC23019 (open circles). The small increase in activity at the highest concentration of VPC23019 (10,000 nM) was observed with and without expression of the S1P 3 receptor (inset) and thus does not represent agonist activity at this receptor. Data points are in triplicate and are representative of two independent experiments. The percent activation is based on normalization of RFU values obtained from the minimum and maximum S1P concentration. Typical values for 0 and 100% calcium mobilization were ϳ400 and 4000 RFU/well, respectively. 28) may be because of stimulation of the cardiac S1P 3 receptors (29). These studies highlight the potential associated with the further characterization of related molecular entities. We have characterized the activity of a series of S1P analogs that are antagonists at two of the five S1P receptors. Our results show that two of these analogs, VPC23019 and VPC25239, are reasonably potent at the S1P 1 receptor (i.e. K i Ͻ 50 nM), and VPC25239 is also potent (K i Ͻ 100 nM) at the S1P 3 receptor. By comparison, the K d value for the S1P-S1P 1 receptor interaction is reported to be 8.1 nM (10). Our previous results with para-substituted aryl amide analogs demonstrate that these are uniform S1P receptor agonists (12). The antagonist activity observed with the compounds presented here resulted from minor changes in the aryl amide structures. Given our earlier observations, it was counterintuitive that antagonists would be realized by 1) positioning of the primary amine in a configuration opposite that of naturally occurring sphingosine and S1P, 2) placing the phenyl substituents in a 1,3 (meta) configuration, and 3) limiting the alkyl chain to no more than eight carbon atoms. In view of our present findings, it is possible that one of the enantiomers of the meta-substituted analog of FTY720 described by Kiuchi et al. (15) is an antagonist (following phosphorylation) for one or more S1P receptors; however, the whole animal assays used to evaluate their compounds would not have allowed for this determination.
The availability of an S1P receptor antagonist should provide a useful tool for studies of S1P biology. Two predictions made using S1P agonists and genetic models are prominent. First, the history of S1P and the S1P 1 receptor in endothelial biology, which is reinforced by the defect in vascular maturation observed in S1P 1 receptor gene ablated mice (3), has led to the suggestion that S1P 1 receptor antagonists might prove FIG. 3. Antagonism at the S1P 1 and S1P 3 receptors by VPC23019. A and B, HEK293T cells were transfected transiently with equal amounts of human S1P 1 or S1P 3 receptor and G i2 ␣, G␤ 1 , and G␥ 2 plasmid DNAs. Membranes were collected after 60 h. Receptor activation was determined using a broken cell binding assay measuring the binding of ␥-35 S-GTP to the membrane as a function of agonist (S1P) stimulation. Blockade of S1P stimulation at S1P 1 and S1P 3 in the ␥-35 S-GTP broken cell binding assay was performed in the presence of 10,000 nM (open circles), 1000 nM (open squares), 100 nM (filled squares), or 0 nM (filled circles) of VPC23019. The binding constant (pK b ) is reported as pK b Ϯ S.E. Data points are in hextuplicate and are representative of two independent experiments for each receptor. The percent activation is based on normalization of disintegrations/min values obtained from the minimum and maximum S1P concentration. Typical values for 0 and 100% binding were ϳ300 and 3000 dpm/well, respectively, for both the human S1P 1 and S1P 3 receptors. C, blockade of the migration of T24 cells transfected stably with human S1P 1 receptor obtained with the S1P 1 agonist VPC22277 (10 nM) was observed with 10, 100, and 1000 nM concentrations of VPC23019. Data points are in duplicate and are representative of two independent experiments. The percent migrating cells is based on normalization of RFU values obtained from the RFU values when the migration was observed with BSA (minimum) and VPC22277 (maximum). Typical values for 0 and 100% migration were ϳ30,000 and 100,000 RFU/well, respectively. D, Ca 2ϩ mobilization observed with T24 cells transfected stably with human S1P 3 receptor (solid line) was not altered by pretreatment with VPC23019 (10,000 nM, dashed line) followed by washout. Data points are in triplicate and are representative of two independent experiments. The percent activation is based on normalization of RFU values obtained from the minimum and maximum S1P concentration. Typical values for 0 and 100% calcium mobilization were ϳ400 and 4000 RFU/well, respectively. useful as anti-angiogenic agents. Second, the suggestion that phospho-FTY720 exerts its effect on lymphocyte trafficking by desensitizing T-lymphocyte S1P 1 receptors (and thereby functionally antagonizing S1P signaling in T-cells (3,30)), suggests that a S1P 1 receptor antagonist should also evoke the lymphopenia that is characteristic of S1P 1 receptor agonists, such as phospho-FTY720 or SEW2871 (4). The presence of the phosphate monoester in our current set of antagonist compounds, however, might result in their rapid hydrolysis by cell surface lipid phosphatases and thus make their use in vivo problematic. Thus, the testing of these ideas may require the synthesis of similar compounds containing hydrolysis-resistant phosphate analogs.
In summary, the results presented here clearly illustrate the discrete chemical space associated with the interaction between these ligands and the S1P receptors. As demonstrated amply by FTY720, such chemical tools can provide unique insights into the physiological roles of S1P and expand our  4. Displacement of radiolabeled S1P by VPC23019 at S1P 1 and S1P 3 . HEK293T cells were transfected transiently with equal amounts of human S1P 1 or S1P 3 receptor and G i2 ␣, G␤ 1 , and G␥ 2 plasmid DNAs. Membranes were collected after 60 h. Displacement of radiolabeled S1P was determined using a membrane binding assay measuring the binding of [ 32 P]S1P to the receptor. Dosedependent displacement of [ 32 P]S1P was observed with S1P (A) and VPC23019 (B) for both S1P 1 (filled circles) and S1P 3 (open circles) receptors. The pK i values are the Ϫlog of the inhibitory binding constant (K i ) and are reported as pK i Ϯ S.E. Data points are in triplicate and are representative of two independent experiments for each receptor. The percent binding is based on normalization of disintegrations/min values obtained from the minimum and maximum. Typical values for 0 and 100% binding were ϳ10,000 and 30,000 dpm/tube, respectively, for both the human S1P 1 and S1P 3 receptors. Nonspecific binding was determined as residual binding of radioligand in the presence of excess S1P to membranes, both heat-denatured and non-heat-denatured, from HEK293T cells transfected transiently with receptor and G-protein DNAs; it was typically 60% of total binding. understanding of S1P receptor signaling, a significant undertaking given their potential roles as therapeutic targets. FIG. 5. Agonist response at the S1P 1 and S1P 3 receptors is altered by VPC23019-related analogs. A, blockade of the migration of T24 cells transfected stably with human S1P 1 receptor obtained with the S1P 1 agonist VPC22277 (10 nM) was observed with VPC23019, VPC23031, VPC23089, and VPC25239 (50 nM). Data points are in duplicate and are representative of two independent experiments. The percent migrating cells is based on normalization of RFU values obtained from the RFU values obtained when the migration was observed with BSA (minimum) and VPC22277 (maximum). Typical values for 0 and 100% migration were ϳ30,000 and 100,000 RFU/ well, respectively. B, blockade of Ca 2ϩ mobilization via stimulation of T24 cells transfected stably with human S1P 3 receptor was performed in the absence (filled circles) or presence of 10,000 nM VPC23031 (filled squares), VPC23089 (open circles), and VPC25239 (open squares). Data points are in triplicate and are representative of two independent experiments. The percent activation is based on normalization of RFU values obtained from the minimum and maximum S1P concentration. Typical values for 0 and 100% calcium mobilization were ϳ400 and 4000 RFU/well, respectively.