Sphingosine 1-Phosphate-induced Cell Proliferation, Survival, and Related Signaling Events Mediated by G Protein-coupled Receptors Edg3 and Edg5*

Sphingosine 1-phosphate (S1P) regulates cell proliferation, apoptosis, motility, and neurite retraction. Contradictory reports propose that S1P acts as either an intracellular second messenger or an extracellular ligand for cell-surface receptors. Hence, the precise signaling mechanisms mediating the diverse cellular effects of S1P remain to be determined. Here, we investigate whether S1P stimulation of cell proliferation, survival, and related signaling events can be mediated by the recently cloned Edg family members of G protein-coupled receptors. We observed that S1P treatment significantly increased proliferation of HTC4 hepatoma cells stably transfected with human S1P receptor Edg3 or Edg5, which was attributable to stimulation of cell growth and inhibition of apoptosis caused by serum starvation. Edg3 and Edg5 transduced S1P-evoked signaling events relevant to cell proliferation and survival, including activation of the ERK/MAP kinases, and immediate-early induction of c-Jun and c-Fos. Trancriptional activation of reporter genes for the c-fospromoter and the serum response element by Edg3 and Edg5 transfected in Jurkat cells was inhibited by pertussis toxin and C3 exoenzyme, implicating Gi/o- and Rho-dependent pathways. Our data also indicated that Edg3 and Edg5 mediated the serum response element activation through transcriptional factors Elk-1 and serum response factor. Thus, specific G protein-coupled receptors Edg3 and Edg5 account for, at least in part, S1P-induced cell proliferation, survival, and related signaling events.

Despite extensive observations in various cell types, the precise signaling mechanisms by which S1P exerts its cellular effects remain undetermined. The major uncertainty is whether S1P acts intracellularly as a second messenger, or extracellularly as a receptor ligand, or both. The second messenger hypothesis is based mainly on the following observations: first, S1P stimulated Ca 2ϩ release directly from isolated endoplasmic reticulum preparations through an inositol trisphosphate-independent mechanism (17,18); second, inhibition of S1P production by sphingosine kinase inhibitors, e.g. N,N-dimethylsphingosine or dihydrosphingosine, blocked cell proliferation induced by platelet-derived growth factor or serum, and Ca 2ϩ mobilization induced by IgE or muscarinic receptor stimulation (21,22). These observations, together with the finding that the intracellular level of S1P was elevated by growth factor stimulation, prompted the assumption that S1P is an intracellular second messenger (23,24). However, the putative intracellular target(s) for S1P has not been identified on a molecular level.
On the other hand, increasing evidence suggests that some of the S1P-elicited effects can be mediated by cell surface G protein-coupled receptors (GPCRs). For example, exogenously added S1P increased [Ca 2ϩ ] i through cell surface GPCR-mediated phospholipase C activation (9,20,25,26). Activation of MAP kinases by S1P was sensitive to pertussis toxin (PTX) inhibition, implicating G i/o -coupled receptors (9,13,27). S1P delivered extracellularly, but not intracellularly, induced morphological changes in cultured neuronal cells (12). Moreover, S1P immobilized on glass particles was able to activate platelets, strongly suggesting an extracellular mechanism (11).
In support of the extracellular ligand mechanism, three closely related seven-transmembrane domain GPCRs in the Edg (Edg ϭ endothelial differentiation gene) family have recently been identified as putative S1P receptors (28 -30). Two of these receptors, Edg3 and H218/AGR16 (herein named Edg5), conferred S1P-induced serum response element (SRE)driven transcription and Ca 2ϩ efflux when expressed in Jurkat cells and Xenopus oocytes, respectively (28). The other receptor, Edg1, induced cell-cell aggregation through enhanced cadherin expression, inhibited adenylyl cyclase and activated MAP kinases via G i protein (29,30).
Nevertheless, the signaling mechanisms by which S1P evokes its complete array of cellular and signaling responses remain to be fully elucidated. Most recently, it was proposed that S1P acts as a dual messenger: as an extracellular ligand for cell surface receptors and as an intracellular second mes-senger mediating distinct sets of signaling and cellular effects (31).
To address this central issue in lysosphingolipid signaling, we examined whether the cloned S1P receptors were capable of mediating S1P-induced proliferation and survival, two of the proposed intracellular effects. We observed that S1P treatment significantly increased cell proliferation and survival of stably transfected HTC4 rat hepatoma cells expressing Edg3 or Edg5. By using biochemical and reporter gene assays, Edg3-and Edg5-evoked signals critical to proliferation and survival were characterized in stably transfected HTC4 cells and transiently transfected Jurkat cells. We observed that both Edg3 and Edg5 mediated activation of the extracellular signal-regulated kinase (ERK), and rapid but transient induction of activator protein-1 (AP-1) transcription factors c-Jun and c-Fos. The major pathways leading to transcriptional activation of the c-fos promoter and the SRE were further dissected by using specific inhibitors against G proteins.

EXPERIMENTAL PROCEDURES
Materials-DNA polymerases, Taq and pfu, were purchased from Stratagene (La Jolla, CA). Marathon-Ready cDNAs of human fetal brain was from CLONTECH (Palo Alto, CA). LipofectAMINE, Opti-MEM medium, and Geneticin (G418 sulfate) were from Life Technologies, Inc. (Gaithersberg, MD). Jurkat leukemic T cells containing the SV40 virus large T antigen (TAg-Jurkat) were obtained from Dr. J. Crabtree (Stanford University). Rat hepatoma HTC4 cells, cell culture media, and fetal bovine serum were from University of California, San Francisco (UCSF), Cell Culture Facilities. S1P was obtained from Biomol (Plymouth, PA). Fatty acid-free human serum albumin (FAF-BSA) was purchased from Sigma. PTX was purchased from CalBiochem (La Jolla, CA). Reporter gene plasmid pfos-luc was a gift from Dr. Q. Hu (UCSF). The PathDetect Elk-1 trans-reporting system was from Stratagene (La Jolla, CA). Expression plasmid pEFC3 containing cDNA for Clostridium botulinum C3 exoenzyme and its control vector pEFmint were generous gifts from Dr. R. Treisman (ICRF, London, United Kingdom). Luciferase assay reagents and the CellTiter 96 AQueous nonradioactive cell proliferation (MTS) assay kits were from Promega (Madison, WI). Antibodies against phosphorylated forms of ERK1 and ERK2 and antibodies against all forms of ERKs were from New England Biolabs (Beverly, MA). Polyclonal antibodies against c-Jun and c-Fos were obtained from Santa Cruz Biotechnologies (Santa Cruz, CA). Peroxidase-conjugated secondary antibodies against mouse and rabbit IgG, enhanced chemiluminescent (ECL) reagents, and Rapid-Hyb solution were purchased from Amersham Pharmacia Biotech. [␣-32 P]dCTP (3000 Ci/mmol) and [methyl-3 H]thymidine (81 Ci/mmol) were obtained from NEN Life Science Products (Boston, MA). The EnzChek Caspase-3 Assay Kit number 2 was obtained from Molecular Probes (Eugene, OR).
Isolation of Human Edg5 cDNA and Stable Transfection in HTC4 Cells-The human ortholog of rat H218/AGR16 was cloned by a combination of RT-PCR and RACE (rapid amplification of cDNA ends). First, a human cDNA fragment was amplified with degenerate primers 5Јctcctg/cgccatc/tgciatc/tgaga/cg and 5Ј-cagc/gia/ca/ga/ca/gtccagc/gaga/ gagc/ga, corresponding to the amino acid sequence LLAIAIER and LLLLDSTC in the third and sixth transmembrane domains of rat H218/AGR16, respectively (32,33). The cDNA template for the PCR reaction (35 cycles of 95°C for 1 min, 55°C for 1 min, 72°C for 2 min on Stratagene's Robocycler) was the reverse-transcribed products of poly(A) ϩ RNAs isolated from human neuroblastoma cell line SK-N-MC. A 400-base pair product was obtained and sequenced, which has a DNA sequence 80% identical to the corresponding region of rat H218/AGR16. The rest of the cDNA sequence was then obtained by 5Ј-and 3Ј-RACE using RACE-ready cDNAs derived from human fetal brain (Marathonready human fetal brain cDNA, CLONTECH). The gene-specific primers used in 5Ј-and 3Ј-RACE (5Ј-gcaggacagtggagcaggcctcga and 5Ј-ctctctacgccaagcattatgtgct, respectively) were derived from the 400-base pair sequence. The RACE reaction conditions were 35 cycles of 95°C for 1 min, 60°C for 1 min, 72°C for 2 min on a Robocycler. RACE products were cloned into pCR2.0 (Invitrogene) and sequenced. To isolate the full-length cDNA, two primers corresponding to the immediate upstream and downstream of the protein coding sequence (5Ј-tcggatccccaccatgggcagcttgtactcg and 5Ј-atctagaccctcagaccaccgtgttgccctc), were used to amplify the Marathon-ready human fetal brain cDNAs (95°C for 1 min, 55°C for 1 min, 72°C for 2 min with pfu polymerase). The resulting 1.1-kilobase product was cut with EcoRI and XbaI and cloned into the pCDEF3 mammalian expression vector.
Multiple lines of stable transfectants of human Edg3 or Edg5 in rat HTC4 hepatoma cells with comparable levels of edg3 or edg5 mRNA were obtained by using the method previously described (34). HTC4 transfectants were maintained in minimum essential medium (MEM) supplemented with 10% fetal bovine serum.
Cell Proliferation Assay-Stable HTC4 cell transfectants plated in 96-well plates were serum-starved for 4 h, and treated with or without S1P in serum-free MEM containing 0.1 mg/ml FAF-BSA for 24 h. Cells were then incubated with either CellTiter 96 AQueous (MTS) solution or 0.2 Ci/ml [ 3 H]thymidine in serum-free MEM. After 4 h of incubation with CellTiter 96 AQueous solution, colored MTS products in the supernatant were transferred into 96-well microtiter plates and absorbance at 490 nm were determined on a Molecular Devices Microplate Reader (Sunnyvale, CA). After 8 h of incubation with [ 3 H]thymidine, cells were washed once with ice-cold phosphate-buffered saline and incubated in 5% trichloroacetic acid for 30 min on ice. After washing twice with phosphate-buffered saline, cell-incorporated [ 3 H]thymidine was solubilized in 0.5 N NaOH, 0.5% SDS. The lysates were transferred into vials, combined with scintillation mixture, and counted on a Beckman scintillation counter (Palo Alto, CA). To obtain cell counts, 5 ϫ 10 4 transfectants plated on 6-well plates were washed and treated with or without 1 M S1P in MEM, 0.1 mg/ml FAF-BSA. After 24 and 48 h, cell numbers in 4 random microscopic fields were counted from each well. The average counts from 6 wells of S1P-treated cells were compared with counts from 6 wells of vehicle-treated controls.
Apoptosis Assays-Stable HTC4 transfectants plated in 6-well plates were serum-starved for 0.5 h, and incubated in serum-free MEM, 0.1 mg/ml FAF-BSA with or without S1P for 12 h. Cells were lysed with 0.5% Triton X-100, 10 mM Tris-HCl, pH 7.5, and 20 mM EDTA for 30 min at 4°C. After centrifugation at 15,000 ϫ g for 20 min to precipitate cellular debris and intact chromosomal DNA, supernatants were extracted with phenol-chloroform. After adding 1/10 volume of 3 M sodium acetate, pH 5.2, nucleic acid was precipitated with isopropyl alcohol. The pellets were resuspended with 10 mM Tris-HCl, pH 7.5, containing RNase A, and loaded on 1.5% agarose gels. The gels were then stained with 1ϫ solution of SYBR Green 1 (Molecular Probes, Eugene, OR), and photographed under UV light. This method preferentially isolates low molecular weight DNA, which enhances detection of a DNA fragmentation "ladder" occurring in apoptosis (35). The activities of caspase-3 (CPP32/apopain)-like proteases were determined by the EnzChek Caspase-3 Assay Kit number 2 (Molecular Probes) following the manufacturer's instruction. Briefly, 5 ϫ 10 5 transfectants plated on 6-well plates were washed and incubated with or without 0.1 or 1 M S1P in MEM, 0.1 mg/ml FAF-BSA. Three hours later, the cells were washed with phosphate-buffered saline, lysed, and caspase-3 activities in the extracts were measured. Fluorescent product of the substrate Z-DEVDrhodamine 110 generated by caspase-3-like proteases in the cell extracts was detected by a Perkin-Elmer LS50B fluorometer with excitation/emission at 496/520 nm. Background fluorescence was determined by including a specific caspase-3 inhibitor (Ac-DEVD-CHO) to the reaction mixtures, and subtracted from the total.
Western Blot and Northern Blot Analysis-Stable HTC4 cell transfectants plated in 6-well plates were serum-starved overnight, and treated with or without S1P in MEM, 0.1 mg/ml FAF-BSA for the indicated time. Cells were lysed on ice with RIPA buffer containing 1% Nonidet P-40, 0.5% deoxycholate, 0.1% SDS, 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 2 mM EGTA, 25 mM sodium fluoride, 1 mM sodium orthovanadate, and 1 ϫ protease inhibitors (Sigma P8340). Lysates of the same amounts of protein were separated on 12% SDS-polyacrylamide gels and transferred onto nitrocellulose membranes. Antibodies against either phosphorylated ERK1/2 or c-Jun were used in Western blot analysis according to the suppliers' instruction. Membranes were developed with ECL detection system after incubation with peroxidaseconjugated secondary antibodies. To normalize the amounts of ERK proteins in each lane, the membranes were stripped and re-blotted with antibodies against all forms of ERK. In some experiments of ERK activation, cells were pretreated with 50 ng/ml PTX for 6 h before S1P stimulation.
For Northern blot analysis, stable HTC4 cell transfectants plated in 6-well plates were serum-starved overnight, and treated with or without S1P in MEM, 0.1 mg/ml FAF-BSA for the indicated time. Total RNAs were isolated with the TRIzol reagent and separated on 1% agarose gel. The transferred Nylon membranes were hybridized with 32 P-labeled rat c-jun cDNA probe amplified from HTC4 cell cDNAs by RT-PCR with primers 5Ј-gggctgctcaagctggcgtcgc and 5Ј-ggcgatccgctccagcttcctt, corresponding to nucleotides 551-572 and 1173-1195 of rat c-jun cDNA (GenBank accession number X17163). Hybridization was performed in Rapid-Hyb solution at 68°C for 1.5 h before washing with 0.1ϫ standard saline citrate, 0.1% SDS at 65°C. The autoradiographs were developed after overnight exposure with one intensifying screen.
Reporter Gene Assay in Transiently Transfected TAg Jurkat Cells-The cis reporter gene plasmids used were pcfos-luc, pSRE-luc (28), and pSRF-luc. The pcfos-luc reporter, in which luciferase expression is controlled by c-fos promoter, was constructed by subcloning the 700-base pair c-fos promoter region from pfos-luc (36), into pGL3 vector (Promega). The pSRF-luc plasmid with 5 copies of SRE.L sequence (5Јatgtactgtatgtccatattaggacatct) (37) upstream of luciferase was obtained from Stratagene. A trans-activating reporter which measures activation of transcription factor Elk-1 is also commercially available (PathDetect Elk-1 trans-reporting system from Stratagene). This system utilizes the pFA-Elk1 plasmid encoding a Gal4/Elk-1 fusion protein containing the Gal4 DNA binding motif and the Elk-1 trans-activation domain. A reporter plasmid pFR-luc containing five copies of the Gal4binding element upstream of luciferase reports the transcriptional activation of Elk-1. Transient transfection and reporter gene assays in TAg-Jurkat cells were done as described previously (28). In some experiments, cells were co-transfected with 1/20 amounts of pEFC3 or its vector control pEFmint to assess the involvement of Rho. The involvement of G i/o was assessed by pretreating some aliquots of transfected cells with 50 ng/ml PTX for 6 h before S1P stimulation.

RESULTS
Cloning of Human Edg5-The sequence of the isolated 1.1kilobase human cDNA (GenBank accession number AF034780) encodes a 353-amino acid protein which is 92% identical to rat H218/AGR16. Its amino acid sequence is 43-44% identical to the human S1P receptors Edg1 and Edg3, and 33-35% identical to the human LPA receptors Edg2 and Edg4, respectively. This protein is presumably the human ortholog of rat H218/ AGR16, and was herein named Edg5 following one working nomenclature system (38).
To study the cellular effects and biochemical signaling of Edg3 and Edg5, we established stably transfected cell lines in HTC4 rat hepatoma cells which do not endogenously respond to S1P in Ca 2ϩ mobilization or any assays in this study. We examined whether extracellular addition of S1P could act on Edg3 or Edg5 expressed in these cells to stimulate proliferation, survival, and relevant signaling pathways. S1P Treatment of Edg3 and Edg5 Transfectants Induced Cell Proliferation-To assess cell proliferation, we first used the CellTiter 96 AQueous proliferation assay that is based on metabolic conversion of a tetrazolium compound MTS to a colored product by living cells. The absorbance intensity of the MTS product is directly proportional to the number of viable cells in culture. As shown in Fig. 1A, S1P treatment in serum-free medium for 24 h increased absorbance in Edg3 and Edg5 transfectants. The effect was significant at 1 nM, reached maximum around 100 nM, and had EC 50 values around 10 nM. In contrast, control vector transfectants did not respond to S1P up to 1000 nM. As control, all transfectants generated similar levels of MTS products when treated with 10% serum, presumably in response to other growth factors in serum (Fig. 1B). The actual total cell numbers were also counted after treatment with 1 M S1P for 24 h and 48 h in serum-free MEM containing 0.1 mg/ml FAF-BSA. At 48 h, cell number of the vector transfectants did not change significantly in the serum-free medium. However, the cell numbers of the Edg3 and Edg5 transfectants increased significantly after S1P treatment compared with the vehicletreated controls (Fig. 1C). These results suggested that S1P stimulated cell proliferation of the HTC4 transfectants expressing Edg3 or Edg5.
To assess the contribution of mitogenesis in the cell number increase by Edg3 and Edg5, we measured the effects of S1P on DNA synthesis. In HTC4 cells stably transfected with Edg3 and Edg5, S1P treatment in serum-free medium increased [ 3 H]thymidine incorporation in a concentration-dependent manner ( Fig. 2A). This effect was significant with 1 and 10 nM S1P in Edg3 and Edg5, respectively, and reached maximum with 1000 nM S1P. In contrast, vector-transfected HTC4 cells did not respond to S1P up to 1000 nM in DNA synthesis. As control, serum induced similar magnitude of responses in vector, Edg3 and Edg5 transfectants (Fig. 2B). S1P Treatment of Edg3 and Edg5 Transfectants Enhanced Cell Survival by Inhibiting Apoptosis-The increase in viable cell numbers with Edg3 and Edg5 transfectants by S1P could also be contributed by greater cell survival. We therefore investigated whether S1P treatment of Edg3 and Edg5 transfectants had reduced cell death during S1P treatment in serumfree conditions. Pilot experiments showed that HTC4 cells cultured in serum-free medium for longer than 48 h underwent significant cell death as measured by the trypan blue exclusion assay, presumably due to withdrawal of survival factors in serum (data not shown). This cell death initiated at around 12 h of serum depletion, as displayed by the appearance of a typical apoptotic DNA ladder (Fig. 3). S1P treatment significantly reduced DNA fragmentation in Edg3 and Edg5 transfectants (Fig. 3A). This effect was more evident with 1 M than with 0.1 M S1P. At 1 M, S1P almost completely abolished (in Edg5) and greatly reduced (in Edg3) DNA fragmentation. In contrast, S1P did not eliminate DNA fragmentation in vector transfectants, although 10% serum did. Structurally similar phospholipids LPA or psychosine (PS) did not have protective effects.
To assess the anti-apoptosis effect of S1P more quantitatively, we measured the enzymatic activities of caspase-3 (CPP32/apopain)-like proteases which are responsible for the manifestation of apoptosis in almost all cell types. After 3 h of serum starvation, caspase-3 activities increased in all transfectants as compared with control cells in 10% serum-containing medium. At 0.1 and 1 M, S1P significantly suppressed the caspase-3 activation in Edg3 and Edg5, but not in vector transfectants (Fig. 3B). These results suggested that S1P treatment significantly inhibited apoptosis caused by growth factor deprivation in Edg3 and Edg5 transfectants as a mechanism for cell survival.
Edg3 and Edg5 Activated ERK-1/2 MAP Kinases in Transfected HTC4 Cells-One of the signaling responses to S1P that has been implicated in cell proliferation and survival is its activation of MAP kinases (9, 13, 27). To see whether Edg3 or Edg5 was able to mediate ERK/MAP kinases activation by S1P, we measured the levels of the active forms of ERKs by Western blot analysis using antibodies against phosphorylated ERK-1/2. Consistent with previous observations in cells expressing endogenous S1P receptors, we observed that 100 nM S1P increased phosphorylation of ERK-1 and ERK-2 in transfectants expressing recombinant Edg3 or Edg5 (Fig. 4A). In contrast, the control vector-transfected HTC4 cells did not respond to 100 nM S1P in ERK phosphorylation, although they still responded to 10% serum (Fig. 4A, bottom panel). The activation of ERK-1 and ERK-2 in Edg3 and Edg5 transfectants were largely blocked by pretreatment with 50 ng/ml PTX for 6 h.
ERK activation by Edg3 and Edg5 was dependent on both S1P concentration and time of incubation. In Edg3, ERK activation reached maximum at 10 nM S1P, whereas it required 100 nM S1P in Edg5 (Fig. 4B). The activation of ERK by S1P was rapid, which was detectable as early as 2 min, reached peak at 10 min, and lasted for at least 30 min (Fig. 4C). In contrast, 1 to 1000 nM S1P did not increase ERK phosphorylation over the same time period in vector-transfected HTC4 cells (data not shown). LPA at 1000 nM did not generate significant activation in ERK phosphorylation in Edg3 or Edg5 transfectants. These results showed that Edg3 and Edg5 specifically mediated S1P-induced ERK activation through G i/o -dependent pathway in HTC4 cell transfectants.

Edg3 and Edg5 Induced AP-1 Transcription Factors c-Jun and c-Fos-One of the significant consequences of ERK activation is the induction of the AP-1 transcription factors c-Jun and c-Fos. It was previously shown that S1P stimulated activity of AP-1 by increasing expression of c-Fos and c-Jun in Swiss 3T3
fibroblasts (39). We therefore examined whether Edg3 or Edg5 was able to mediate S1P-induced increases in mRNA and protein levels of c-Jun and c-Fos in stable HTC4 transfectants. As shown in Fig. 5A, 100 nM S1P increased c-jun mRNA levels rapidly and transiently in Edg3 and Edg5 transfectants, which was significant at 0.5 h, reached a peak at 1 h, and diminished after 2 h. In contrast, vector-transfected HTC4 cells did not generate a significant increase in c-jun mRNA in response to 100 nM S1P, even though they responded to 100 nM 12-Otetradecanoylphorbol-13-acetate and 10% serum (Fig. 5A, bottom panel). Neither LPA nor PS, at 1000 nM, induced significant increases (Fig. 5A). S1P-induced c-Jun activation was also observed on protein level by Western blot analysis (Fig. 5B). After 4 h treatment, S1P concentration dependently increased c-Jun protein levels in Edg3 and Edg5 transfectants. In contrast, S1P at 1000 nM did not generate a significant increase in c-Jun protein in vector-transfected HTC4 cells, albeit 100 nM 12-O-tetradecanoylphorbol-13-acetate and 10% serum did (Fig.  5B, bottom panel). Similar observations were made with c-Fos on both mRNA and protein levels (data not shown). Thus, Edg3 and Edg5 both were capable of inducing c-Jun and c-Fos rapidly and transiently, a characteristic activation of immediateearly genes (IEGs).

The Induction of IEGs by Edg3 and Edg5 Involved G i/o and Rho-To further characterize the induction of IEGs by Edg3
and Edg5, we investigated the receptor-mediated signaling pathways using reporter gene assays. In this regard, the TAg-Jurkat cells provided a useful and convenient system for their low background responses to S1P and the ease of transient transfection for reporter gene assay. This not only was necessary to obtain sufficient reporter gene measurement, but also provided valuable information on the cell-type specificity of Edg receptor signaling. We first observed that Edg3 and Edg5, but not vector transfectants, were able to induce the c-fos promoter reporter gene in response to S1P in a concentrationdependent manner (data not shown). At 100 nM, S1P increased the c-fos promoter reporter gene activity by 5.3 Ϯ 0.3-and 3.9 Ϯ 0.4-fold in Edg3 and Edg5, respectively (Fig. 6A). To assess the role of G proteins in c-fos promoter activation by Edg3 and Edg5, we used PTX and C3 exoenzyme that specifically inhibit G i/o and Rho, respectively. In both Edg3 and Edg5 transfectants, c-fos promoter activation by S1P was partially inhibited by the sole treatment with PTX (50 ng/ml, 6 h) or C3 exoenzyme (co-transfected) (Fig. 6A). Together, PTX and C3 exoenzyme almost completely blocked S1P-induced c-fos promoter activation in Edg3 and Edg5. These results suggested that both G i/o and Rho are partially involved in Edg3-and Edg5-mediated activation of the c-fos promoter. FIG. 4. Edg3-and Edg5-mediated activation of ERK/MAP kinases by S1P and its inhibition by PTX. A, stable transfectants of Edg3, Edg5, and vector in HTC4 hepatoma cells were stimulated with 100 nM S1P for 10 min. Some cells were pretreated with 50 ng/ml PTX for 6 h before S1P treatment. Cell lysates were separated on 12% SDS-polyacrylamide gel electrophoresis and immunoblotted with antiphospho-ERK1/2 antibodies. The positions of phosphorylated ERK1 and ERK2 are indicated. The stripped blots were reprobed with anti-ERK1/2 antibodies to normalize the amounts of ERK1/2 in each lane (labeled ERK1/2). Vector transfectants were also stimulated with 10% serum (labeled ser) as control. B, concentration-dependent responses to 0, 1, 10, 100, and 1000 nM S1P. C, time course of ERK activation induced by 100 nM S1P. The treatment with 1000 nM LPA in Edg3 was 30 min. Similar results were obtained with two other stable transfectant lines in at least three different experiments.
The induction of c-fos and many other IEGs is mediated essentially by the SRE located in their promoters (40). We previously showed that S1P activated the c-fos SRE reporter gene in Jurkat cells transfected with human Edg3 and rat Edg5 (H218/AGR16) (28). To dissect the signaling pathway leading to SRE activation by Edg3 and Edg5, we examined the inhibitory effects of PTX and C3 exoenzyme. PTX and C3 exoen-zyme, when used alone, partially blocked SRE activation in Edg3 and Edg5, and completely blocked it when combined together (Fig. 6B). These results suggested that, similar to the c-fos promoter, activation of the SRE by Edg3 and Edg5 also involved both G i/o -and Rho-dependent pathways.
Edg3 and Edg5 Activated Both Elk-1 and SRF-The SRE is activated co-operatively by two transcription factors: a ternary complex factor such as Elk-1, and the serum response factor (SRF) (37). The Elk-1 is activated by Ras-Raf-ERK pathway, whereas the SRF is activated by downstream effectors of Rho (37). To examine the activation of Elk-1 and SRF by Edg3 and Edg5, we measured Elk-1 trans-activation and SRF reporter gene in TAg-Jurkat cell transfectants after S1P treatment. In cells transfected with Edg3 or Edg5, S1P at 1 to 1000 nM significantly increased activation of the trans-reporter for Elk-1, which was not observed in vector-transfected cells (Fig.  7A). As control, T cell receptor activation resulted in similar levels of Elk-1 activation in vector as in Edg3 and Edg5 transfectants. Involvement of G proteins in Edg3-and Edg5-mediated Elk-1 activation was then examined by PTX treatment. Interestingly, whereas PTX treatment completely blocked Elk-1 activation by Edg5, it did not lower Elk-1 activation by Edg3 (Fig. 7B). Edg3 and Edg5 also mediated activation of the SRF reporter gene in transfected TAg-Jurkat cells in response to S1P in a concentration-dependent manner (Fig. 8A). In con-  6. S1P-induced activation of c-fos promoter and SRE reporter genes, and its inhibition by PTX and C3 exoenzyme in transiently transfected TAg-Jurkat cells. TAg-Jurkat cells were co-transfected with pEdg3/EF3 or pEdg5/EF3 expression plasmids with 1/10 amounts of reporter gene constructs. In some transfections, 1/20 amounts of EFC3 plasmid were included for C3 exoenzyme treatment. After overnight serum starvation, aliquots of transfectants were pretreated with or without 50 ng/ml PTX for 6 h. A, activation of pcfos-luc reporter gene by 100 nM S1P, and its inhibition by PTX and C3. B, activation of pSRE-luc reporter gene by 100 nM S1P, and its inhibition by PTX and C3. The fold stimulation of SRE activation in the control were 3.5 Ϯ 0.02 and 2.9 Ϯ 0.1 for Edg3 and Edg5, respectively. Data shown are mean Ϯ S.E. of triplicate determinations from one representative experiment of three.
trast, vector transfectants did not generate SRF activation with up to 1000 nM S1P, even though they responded to anti-TCR to a similar level as Edg3 and Edg5 transfectants. The SRF activation by Edg3 and Edg5 was completely blocked by co-transfection of C3 exoenzyme, suggesting that Rho is required for Edg3 and Edg5-mediated SRF activation (Fig. 8B).

DISCUSSION
Two distinct mechanisms of S1P action have been proposed, namely an intracellular second messenger mechanism and an extracellular receptor ligand mechanism. A recent report by Van Brocklyn and colleagues (31) subdivided S1P effects into two categories: those correlated with the expression of Edg1 and those did not (31). The Edg1-dependent effects included inhibition of adenylyl cyclase and activation of the ERK-2 MAP kinase, whereas the Edg1-independent effects were Ca 2ϩ mobilization, activation of phospholipase D, tyrosine phosphorylation of p125 FAK , mitogenesis, and suppression of apoptosis. These authors concluded that S1P acts both extracellularly through Edg1, and intracellularly as a second messenger independent of S1P receptors (31). On the other hand, several recent reports showed results that provided evidence for alternative conclusions. These studies showed that Edg1 was capable of eliciting Ca 2ϩ mobilization through coupling to G i/o in certain cell types, and confirmed that Edg3 and Edg5 were also functional S1P receptors (41)(42)(43)(44)(45)(46). These reports expanded the repertoire of receptor-dependent S1P signaling to phospholipase C-mediated Ca 2ϩ mobilization, ERK1/2 activation, and Rho-dependent stress fiber formation.
The reason for these contradictory observations is not entirely clear, but may be explained by the following. First, different cellular or signaling effects may use either intracellular or receptor-mediated mechanisms depending on the specific cell type or heterologous overexpression system. Second, distinct signaling mechanisms may be elicited by different concentrations of S1P, as the extracellular effects of the cloned Edg receptors have EC 50 values around 10 nM, whereas the proposed intracellular effects required more than 1 M. Alternatively, the Edg1-independent effects observed by Van Brocklyn et al. (31) in HL-60 and PC12 cells could be mediated by a small number, but yet functional Edg3 or Edg5 that was undetectable by the receptor binding assay. In fact, later studies (43,47) revealed the presence of mRNAs for Edg3 and Edg5 in both HL-60 and PC12 cells.
Our current study aimed to resolve this central controversy in lysosphingolipid signaling. We observed that Edg3 and Edg5, when expressed in HTC4 cells, were able to mediate S1P-induced cell proliferation and survival. Biochemical signals evoked by Edg3 and Edg5 include Ca 2ϩ , ERK/MAP kinases, Rho, and transcription factors c-Jun, c-Fos, Elk-1, and SRF. These signaling constituents play essential roles in cell growth and survival, and closely resemble the profiles of other GPCRs with mitogenic potentials (48 -50).
We also attempted to dissect the complex G protein-activated pathways leading to the SRE activation which is expected to be cell type specific. Our results clearly implicated the Ras-Raf-FIG. 7. Edg3-and Edg5-mediated activation of Elk-1 transreporter by S1P, and its inhibition by PTX in transiently transfected TAg-Jurkat cells. TAg-Jurkat cells were co-transfected with empty vector, pEdg3/EF3, or pEdg5/EF3 expression plasmids with 1/10 amount of pFR-luc, 1/50 amount of pFA-Elk1 reporter constructs. After overnight serum starvation, aliquots of transfectants were stimulated with S1P. A, concentration-dependent responses of Elk-1 trans-reporter to 0, 1, 10, 100, and 1000 nM S1P. Stimulation with 1/1000 dilution of the C305 ascites anti-T cell receptor (anti-TCR) was used as a control. B, before 100 nM S1P stimulation, some aliquots were pretreated with 50 ng/ml PTX for 6 h. Data shown are mean Ϯ S.E. of triplicate determinations from one representative experiment of three.
FIG. 8. Edg3-and Edg5-mediated activation of SRF by S1P, and its inhibition by co-transfected C3 exoenzyme in TAg-Jurkat cells. TAg-Jurkat cells were co-transfected with empty vector, pEdg3/ EF3, or pEdg5/EF3 expression plasmids with 1/10 amount of pSRF-luc. After overnight serum starvation, aliquots of transfectants were stimulated with S1P. A, concentration-dependent responses of SRF reporter to 0, 1, 10, 100, and 1000 nM S1P. Stimulation with 1/1000 dilution of the C305 anti-T cell receptor (anti-TCR) ascites was used as a control. B, before 100 nM S1P stimulation, some aliquots were co-transfected with 1/20 amounts of EFC3 plasmid encoding the C3 exoenzyme at the beginning of the experiments. Data shown are mean Ϯ S.E. of triplicate determinations from one representative experiment of three.
ERK-Elk-1 and Rho-SRF pathways in this process, which is consistent with the two-pathway model proposed by Treisman and colleagues (37) for LPA. For LPA receptors and other GPCRs, activation of Ras-Raf-ERK can be achieved by either the ␤␥ subunit of G i/o or the ␣ subunit of G q/11 depending on the cell type (49). Our data in HTC4 cells implicated G i/o in the Edg3-and Edg5-mediated ERK activation (Fig. 4A), whereas in Jurkat cells, Edg3-mediated Elk-1 trans-activation may require PTX-insensitive G proteins, most likely G q/11 (Fig. 7B). These results are consistent with our recent finding that Ca 2ϩ mobilization by Edg3 and Edg5 involved G i/o in HTC4 cells, but that by Edg3 in Jurkat cells was PTX-insensitive (44). In other GPCRs, activation of Rho can result from either G␣ 12/13 or G␣ q/11 (51). Our most recent results showed that Edg3 and Edg5 stimulated GTPase activity of G␣ 13 , suggesting that both receptors are capable of activating Rho (52).
Although our data demonstrated that the observed S1P effects were solely dependent on the expression of Edg3 or Edg5, we further investigated whether S1P could be generated intracellularly through stimulation of its own receptors, which then acted inside the cells. Unfortunately, technical impediment hampered the application of necessary reagents and methods in the current system. For instance, sphingosine kinase inhibitors of intracellular S1P generation, such as N,N-dimethylsphingosine, had nonspecific effects on HTC4 cells. At concentrations required to inhibit sphingosine kinases (5-10 M), N,N-dimethylsphingosine directly altered Ca 2ϩ homeostasis (44), a notion supported by a recent finding that the Ca 2ϩ release-activated Ca 2ϩ channel (store-operated Ca 2ϩ current) is a direct target of N,N-dimethylsphingosine and dihydrosphingosine (53). Moreover, it is difficult to avoid leakage while microinjecting S1P inside the transfected cells, which would yield misleading results.
Nonetheless, several theoretical considerations endorse an extracellular ligand role of S1P in, at least some, (patho)physiological settings. First, it is conceivable that S1P generated inside the cells can be excreted, then acts on the cell surface receptors as an autocrine or paracrine factor (3,54). Such mode of action has already been demonstrated for LPA (55), which has GPCRs and many cellular effects similar to S1P (38,56). Second, the concentration of S1P in serum (around 484 nM), likely to be the highest in biological fluids (11), is above the EC 50 levels of the observed receptor-mediated effects, but still below the concentrations required for the proposed intracellular effects. Finally, S1P can be released from platelets in response to other agonists such as thrombin, which makes S1P a likely candidate as an extracellular (patho)physiological mediator (57).
In summary, we demonstrated, for the first time, that Edg3 and Edg5 were capable of mediating S1P-induced cell proliferation and suppression of apoptosis. These cellular effects of S1P were presumably transduced by multiple Edg3-and Edg5elicited signaling constituents, such as Ca 2ϩ , ERK/MAP kinases, Rho, Elk-1, SRF, AP-1, and IEGs. These cellular and signaling responses were solely dependent on the expression of Edg3 or Edg5, strongly supporting the GPCR-mediated mechanism in S1P signaling. Although conclusive attribution of each Edg receptor to a specific S1P effect in vivo awaits targeted gene deletion and specific receptor antagoinsts, our results provide compelling evidence that at least part of the S1P-induced (patho)physiological effects are transduced by GPCRs.