Nrg-1 belongs to the endothelial differentiation gene family of G protein-coupled sphingosine-1-phosphate receptors.

The previously cloned rat nerve growth factor-regulated G protein-coupled receptor NRG-1 (Glickman, M., Malek, R. L., Kwitek-Black, A. E., Jacob, H. J., and Lee N. H. (1999) Mol. Cell. Neurosci. 14, 141-52), also known as EDG-8, binds sphingosine-1-phosphate (S1P) with high affinity and specificity. In this paper we examined the signal transduction pathways regulated by the binding of S1P to EDG-8. In Chinese hamster ovary cells heterologously expressing EDG-8, S1P inhibited forskolin-induced cAMP accumulation and activated c-Jun NH2-terminal kinase. Surprisingly, S1P inhibited serum-induced activation of extracellular regulated protein kinase 1 and 2 (ERK1/2). Treatment with pertussis toxin, which ADP-ribosylates and inactivates G(i), blocked S1P-mediated inhibition of cAMP accumulation, but had no effect on c-Jun NH2-terminal kinase activation or inhibition of ERK1/2. The inhibitory effect of S1P on ERK1/2 activity was abolished by treatment with orthovanadate, suggesting the involvement of a tyrosine phosphatase. A subunit selective [35S] guanosine 5'-3-O-(thio)triphosphate binding assay demonstrates that EDG-8 activated G(i/o) and G12 but not Gs and G(q/11) in response to S1P. In agreement, EDG-8 did not stimulate phosphoinositide turnover or cAMP accumulation. The ability of S1P to induce mitogenesis in cells expressing the EDG-1 subfamily of G protein-coupled receptors is well characterized. In contrast, S1P inhibited proliferation in Chinese hamster ovary cells expressing EDG-8 but not empty vector. The antiproliferative effect, like S1P-mediated ERK1/2 inhibition, was orthovanadate-sensitive and pertussis toxin-insensitive. Our results indicate that EDG-8, a member of the EDG-1 subfamily, couples to unique signaling pathways.

that EDG-8 is mainly coupled to G i/o and G 12 and inhibits, rather than stimulates, ERK activation in CHO cells. Moreover, S1P markedly inhibits proliferation in CHO cells overexpressing EDG-8 in a G i/o -independent manner.
Transfection of Cells with HA-tagged EDG-8 -A polymerase chain reaction strategy was used to insert a 9-amino acid HA epitope sequence (YPYDVPDYASL) into the COOH terminus of EDG-8. The sense and antisense polymerase chain reaction primers were 5Ј-aaaa-aagcttCAAGGTTCGCATAT AGACCAG-3Ј and 5ЈaaaactcgagCTAAGC-ATAATCTGGAACATCATATGGATAGTCTGTAGCATCAGGCACC-AG-3Ј, respectively. The sense primer has 21 nucleotides of 5Јuntranslated region from the rat EDG-8 cDNA (6). The antisense primer has 27 nucleotides coding for the HA epitope and 7 codons from the COOH terminus of EDG-8 (bold type). Polymerase chain reaction products were cloned into the HindIII and XhoI cloning sites of the mammalian expression vector pcDNA3 (Invitrogen, San Diego, CA), generating HAedg8pcDNA3. The fidelity of the plasmid construct was verified by sequencing. CHO cells were either transiently (2 days) or stably transfected with HAedg8pcDNA3 using LipofectAMINE 2000 (Life Technologies, Inc.). Expression of EDG-8 in transfected cells was verified by immunoblotting using the HA polyclonal antibody.
[ 32 P]S1P Binding-[ 32 P]S1P was prepared as described previously (36). CHO cells seeded at a density of 75,000 cells/cm 2 were transiently transfected with HAedg8pcDNA (1 g/well) and incubated with the indicated concentrations of [ 32 P]S1P. Unlabeled lipid competitors were added as 4 mg/ml fatty acid-free bovine serum albumin complexes, and bound [ 32 P]S1P was quantitated by scintillation counting as described previously (36).
Measurement of cAMP Production-CHO cells were grown in 12-well plates, transiently transfected with HAedg8pcDNA3 or the human ␤ 2adrenergic receptor cDNA in pSVL (37), washed with phosphate-buffered saline, and pretreated with 0.5 mM isobutylmethylxanthine for 20 min at 37°C. To measure inhibition of cAMP formation, EDG-8-expressing cells were incubated with 10 M forskolin in the presence of the indicated concentrations of S1P for 20 min, and cAMP levels were measured as described (38). Measurement of stimulation of cAMP accumulation was performed as above in the absence of forskolin. cAMP accumulation in ␤ 2 -adrenergic receptor-expressing cells was induced with isoproterenol (2 M; 20 min) and served as a positive control.
Measurement of PI Turnover-CHO cells were grown in 12-well plates, transiently transfected with HAedg8pcDNA3 or Rm1pcDNA3 (rat M 1 muscarinic receptor cDNA in pcDNA3; Ref. 38), and washed with phosphate buffered saline and pretreated with 10 mM LiCl for 30 min at 37°C. EDG-8-expressing cells were incubated with the indicated concentrations of S1P for 20 min, and PI turnover was measured as described (38). PI turnover in M 1 muscarinic receptor-expressing cells was stimulated with carbachol (1 mM; 20 min) and served as a positive control.
SDS-PAGE and Immunoblotting-Transiently transfected CHO cells were grown in 35-mm tissue culture dishes, treated with the indicated lipids, and harvested/lysed in 95°C SDS-PAGE sample buffer as described previously (34). Proteins were separated on 10% polyacrylamide gels (Bio-Rad) and transferred to polyvinylidene difluoride membranes (Amersham Pharmacia Biotech). Blots were probed with phosphorylation state-specific MAP kinase primary antibodies. Goat antirabbit or goat anti-mouse horseradish peroxidase-conjugated secondary antibodies allowed detection of proteins by the ECL Plus Detection System (Amersham Pharmacia Biotech). Fluorograms were quantitated by densitometry. Blots were stripped, reprobed with primary antibodies that recognize MAP kinases independently of phosphorylation state, and quantitated. Data are expressed as the means Ϯ S.E. of n independent experiments.
[ 35 S]GTP␥S Binding Assay-CHO cells transfected with HAedg8pcDNA or vector alone were harvested and homogenized in ice-cold 20 mM HEPES, pH 8.0, 1 mM EDTA, 0.1 mM phenylmethylsulfonyl fluoride, 2 g/ml aprotinin, 10 g/ml leupeptin by 15 passages through a 25-gauge needle. Homogenates were centrifuged at 1,000 ϫ g for 5 min, and the resulting supernatant was centrifuged at 25,000 ϫ g for 30 min to obtain the membrane fraction. Membranes (20 g of protein/assay point) were resuspended in 50 l of 50 mM Tris-HCl, pH 7.4, 100 mM NaCl, 2 mM EDTA, 1 M GDP and 3 mM MgCl 2 and incubated with 1 M S1P or vehicle for 10 min at 30°C. Subsequently, 30 nM [ 35 S]GTP␥S (1300 Ci/mmol; PerkinElmer Life Sciences) was added, and membranes were incubated for an additional 10 min at 30°C. Incubations were terminated by adding 500 l of ice-cold 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 20 mM MgCl 2 , 0.5% Nonidet P-40 (Calbiochem), 10 g/ml aprotinin, 100 M GDP, and 100 M GTP. After 30 min, solubilized extracts were incubated for 1 h at 4°C with 10 l of a G␣ subunit-selective antibody. (Santa Cruz). Nonspecific binding was determined by immunoprecipitation with nonimmune serum. Bound [ 35 S]GTP␥S was quantitated by scintillation spectrometry. Nonspecific binding typically represented 10 -20% of total binding and was subtracted from the latter.
Cell Proliferation Assays-Cell proliferation was measured as previously described (39). CHO cells (5,000 cells) were transfected with HAedg8pcDNA or vector alone and plated in 12-well plates containing Ham's F-12 medium supplemented with 10% charcoal-treated dialyzed fetal bovine serum. After 18 h (t ϭ 0), cells were washed twice with Ham's F-12 and grown in Ham's F-12 without serum to measure proliferative effects following S1P treatment or Ham's F-12 with 10% charcoal-treated dialyzed fetal bovine serum to measure the inhibitory effects of S1P on proliferation. S1P and inhibitors were added at the indicated concentrations. After 48 h, cells were washed with phosphatebuffered saline, fixed with 70% ethanol for 10 min, and stained with crystal violet. Incorporated dye was dissolved in 0.1 M sodium citrate in 50% ethanol, pH 4.2, and the absorbance was measured at 540 nm. In parallel experiments, cells were trypsinized and counted in a hemocytometer. Cell numbers in four random microscopic fields were counted from each well. Measurements from the two methods gave identical results. Each determination represents the mean Ϯ S.E. of three to four individual wells.

RESULTS
[ 32 P]SIP Binds to EDG-8 -Many cell lines commonly employed for heterologous expression of receptor genes, including COS and HEK293 cells, respond to S1P (40). In contrast, previous reports have demonstrated that native CHO cells, while expressing detectable levels of EDG-5 mRNA (41), do not bind or readily respond to S1P (20,23,25). In agreement with these studies, we found that native CHO cells transfected with vector exhibited negligible binding of [ 32 P]S1P, as shown by the absence of significant correlation between B and B/F in the Scatchard plot (Fig. 1, left panel). Therefore, CHO cells were subsequently employed for the analysis of EDG-8 function (Tables  I and II). Following transient transfection with an expression vector for HA-tagged EDG-8, CHO cells specifically bound [ 32 P]S1P with high affinity (K d ϭ 6 Ϯ 4 nM; n ϭ 3 independent experiments; Fig. 1, left panel). This K d value is in excellent agreement with a recent study performed in HEK293T cells (14), demonstrating that EDG-8 is indeed a high affinity receptor for S1P in diverse cell types. The level of expression of EDG-8 in CHO cells based on [ 32 P]S1P binding was comparable with that seen in these cells transfected with EDG-1, -3, and -5 (25). In agreement with a recent report in HEK293T cells (14), only unlabeled S1P and dihydro-S1P effectively competed with [ 32 P]S1P for binding to EDG-8 ( Fig. 1, right panel), whereas LPA, sphingosine, and SPC had no significant effects.
Effect of EDG-8 on cAMP Formation-Previously it has been shown that S1P treatment of EDG-transfected CHO cells increases cAMP content in a PTX-insensitive manner with the following rank order of efficacy: EDG-5 Ͼ EDG-3 Ͼ Ͼ EDG-1 (25). Treatment of CHO cells transiently expressing EDG-8 with S1P (1 to 1000 nM), however, had no effect on intracellular cAMP content (Table I), whereas isoproterenol (2 M) caused a robust 6-fold increase in cAMP levels in CHO cells expressing the ␤ 2 -adrenergic receptor. To assess the ability of EDG-8 to mediate inhibition of cAMP accumulation, cAMP levels were increased by treatment with forskolin. Addition of S1P resulted in a dose-dependent, PTX-sensitive, decrease in forskolin-induced cAMP accumulation ( Fig. 2A), whereas in empty vectortransfected cells, S1P, even at a concentration as high as 1 M, had no significant effect on forskolin-induced cAMP accumulation (Fig. 2B). Similar to S1P, dihydro-S1P (1 M) effectively inhibited cAMP formation, whereas low concentrations of LPA, sphingosine, and SPC (10 to 100 nM) had no significant effect, and only small (22%) inhibition was found at a high concentration of LPA (1 M) (Fig. 2B).
Effect of EDG-8 on PI Turnover-In CHO cells transfected with EDG-1, -3, or-5, S1P increases intracellular Ca 2ϩ concentrations caused by the activation of phospholipase C (25). In contrast, S1P, at concentrations ranging from 1 nM to 1 M, did not significantly stimulate PI turnover in EDG-8-expressing cells (Table I), whereas carbachol (1 mM) produced a marked 5 Ϯ 0.9-fold increase in PI turnover in cells transiently expressing the M 1 -muscarinic receptor known to stimulate phospholipase C via G q/11 (n ϭ 3 independent experiments).
In serum-starved EDG-8-expressing CHO cells, S1P did not activate ERK1/2 as measured by either Western blot analysis with phosphorylation state-specific antibodies or immunocomplex kinase assay of whole cell lysates (Table I). Even after prolonged incubations of up to 60 min, S1P had no significant effect on ERK activation. Surprisingly, induction of ERK1/2 phosphorylation by serum was repressed in a time-dependent fashion by S1P (1 M) in CHO cells expressing EDG-8 (Fig. 3A). This inhibitory response was insensitive to PTX (Fig. 3A). Furthermore, serum-induced ERK1/2 phosphorylation was not inhibited by S1P in CHO cells transfected with empty vector (Fig. 3A). Similar to S1P, dihydro-S1P inhibited ERK1/2 phosphorylation by ϳ40% (Fig. 3B). In agreement with their inability to bind to EDG-8, LPA, SPC, and sphingosine had no effects on ERK phosphorylation induced by serum (Fig. 3B). Similar results were obtained by immunocomplex kinase assays (Fig.  4). Once more, S1P but not LPA repressed serum activation of ERK1/2 by ϳ50 -60%.
In contrast, in CHO cells transiently expressing EDG-1, -3, or -5, S1P did not inhibit serum-induced phosphorylation of ERK1/2 as determined by Western blot analysis, even in cells pretreated with PTX (Table I). Identical results were noted by immunocomplex kinase assays (Table I).
Although the binding of S1P to EDG-1, -3, and -5 did not inhibit ERK activation, this is not unprecedented. It has previously been demonstrated that the GPCR, angiotensin II type 2 (AT 2 ) receptor, inhibits serum-activated ERK1/2 in several cell types (48,49). This effect was abolished by inhibition of protein phosphatases. Similarly, the ability of S1P to inhibit serum-activated ERK 1/2 in EDG-8-expressing CHO cells was blocked by orthovanadate and okadaic acid, albeit to a lesser extent (Fig. 4). Neither okadaic acid (5 M) nor orthovanadate (0.1 mM) alone had a significant effect on serum-stimulated ERK1/2 activation (Fig. 4).
EDG-8 Activates SAPK/JNK but Not p38 MAP Kinase-In serum-starved CHO cells transiently expressing EDG-8, S1P induced a time- (Fig. 5A) and dose-dependent (EC 50 ϳ100 nM; data not shown) activation of JNK as measured by an increase in JNK phosphorylation. Increased JNK phosphorylation was apparent as early as 1 min after treatment with S1P (3 Ϯ 0.2-fold over basal; n ϭ 3 independent experiments) and was sustained for at least 60 min. Remarkably, only the p54 isoform of JNK appears to be phosphorylated in a PTX-independent manner, suggesting that EDG-8 may regulate specific JNK isoforms. S1P-stimulated JNK phosphorylation was not observed in parental CHO cells transfected with empty vector (Fig. 5A). This phosphorylation was associated with a 2.8-fold increase in JNK activity (Fig. 6). Similar to S1P, dihydro-S1P also stimulated p54 JNK phosphorylation 2.4-fold, whereas other lysophospholipids did not have any significant effects (Fig. 5B). In agreement with the Western analyses (Fig. 5B), in vitro kinase assays demonstrated that LPA does not stimulate JNK activity in EDG-8-expressing CHO cells (Fig. 6). In contrast to its effect on JNK activity in EDG-8 overexpressing CHO cells, S1P has no significant effect on p38 phosphorylation (data not shown; 1 M S1P; 1-60 min). Immunocomplex kinase assays also confirmed the inability of S1P/EDG-8 to induce p38 activity, whereas the positive control arsenite (50 M; 60 min) potently activated p38 in CHO cells (data not shown).
EDG-8 Coupling to Individual G Proteins-Coupling of EDG-8 to endogenous G protein family members was assessed by a [ 35 S]GTP␥S binding assay (30). COOH-terminal peptidedirected antibodies were used to immunoprecipitate either specific G␣ subunits (e.g. G␣ s , G␣ 12 or G␣ 13 ) or G␣ family members (e.g. G␣ i/o or G␣ q/11 ). S1P-stimulated [ 35 S]GTP␥S binding in the immunoprecipitates was used as a measure of G protein coupling and activation.
As shown in Fig. 7, negligible [ 35 S]GTP␥S binding was evident for G␣ s , G␣ i/o , G␣ q/11 , and G␣ 12 in empty vector-transfected CHO cells treated with S1P. In EDG-8-expressing cells, however, S1P treatment produced a 2-2.5-fold increase in [ 35 S]GTP␥S binding to the G␣ i/o family of G proteins and G␣ 12 (Fig. 7). Whereas S1P stimulation of EDG-8 activated G␣ i/o and G␣ 12 , it had no effect on G␣ s or G␣ q/11 (Fig. 7). In agreement with previous reports (20,23), CHO cells express G␣ s , G␣ i/o , G␣ q/11 , and G␣ 12 proteins (Fig. 7). Thus, the lack of detectable coupling to G␣ s or G␣ q/11 is not due to the absence of these G proteins in CHO cells. Moreover, when membranes from CHO cells expressing the ␤ 2 -adrenergic receptor were incubated with 2 M isoproterenol, a 2.  Present study a 1, positive effect; 2, negative effect; rank order of intrinsic activity is 111 Ͼ 11 Ͼ 1; 1/2, positive and negative effects reported. -, no effect; ND, not determined.
b Activation of adenylyl cyclase was measured by determining cAMP levels in the presence of IBMX after treatment with 1 M S1P for 20 min. c PI turnover was measured as described under "Experimental Procedures." d ERK activation was monitored by phosphorylation-specific antibodies and by immunocomplex kinase assay of 18 -24 h serum-starved, transfected cells treated with 1 M S1P for 1, 5, 10, 30, and 60 min. ERK inhibition was monitored by phosphorylation-specific antibodies and by immunocomplex kinase assay of transfected cells treated with 10% serum for 2 h followed by 1 M S1P for 30 and 60 min. e JNK activation was determined by phosphorylation-specific antibodies and kinase assay following S1P treatment. f p38 MAP kinase activity was measured by phosphorylation state-specific antibodies and immunocomplex kinase assay. g Proliferation was determined as a fold change in cell number following S1P treatment (48 h).  13 , for which no activation could be discerned through EDG-8. EDG-5, a known activator of G␣ 13 (30), was used as a positive control. In EDG-5 transfected CHO cells, S1P did not promote [ 35 S]GTP␥S binding to G␣ 13 . ation via EDG-3, EDG-5, and possibly EDG-1 (27,29,42). Thus, it was of great interest to examine whether binding of S1P to EDG-8 also regulates mitogenic pathways. To assess induction of proliferation, stably transfected cells plated in 12-well plates containing serum-free medium were counted after treatment with 1 M S1P for 48 h. Cell numbers from both control vector and EDG-8 transfectants did not increase in response to S1P (data not shown), suggesting that S1P does not behave as a mitogen at EDG-8.
Interestingly, the proliferative response of EDG-8-transfected CHO cells was 62% lower than vector-transfected cells following mitogenic stimulation with 10% serum for 48 h (Fig.  8). This observation was corroborated by independent experiments, demonstrating that EDG-8-transfected CHO cells exhibit about a 3-fold reduction in [ 3 H]thymidine incorporation compared with untransfected or vector-transfected cells (data not shown). Taken together, our findings suggest that EDG-8 contains intrinsic activity and inhibits serum-stimulated cell proliferation. Furthermore, addition of 1 M S1P significantly inhibited serum-stimulated proliferation of EDG-8-transfected cells (Fig. 8). The anti-proliferative effect induced by S1P binding to EDG-8 was dose-dependent with an EC 50 value around 20 nM (data not shown). In contrast, 1 M S1P treatment did not inhibit proliferation of vector-transfected cells stimulated with 10% serum (Fig. 8). Because biochemical evidence indicates that EDG-8 couples to G i/o , we investigated the possibility that G i/o proteins may be involved in the antiproliferative response induced by S1P. To this end, cells were pretreated with PTX. PTX was completely ineffective in abolishing S1P-induced anti-proliferation (Fig. 8). In contrast, treatment of EDG-8-transfected cells with orthovanadate (1 M) significantly impaired S1P-induced antiproliferative effects, suggesting that this response is dependent on a protein-tyrosine phosphatase (Fig. 8). Neither PTX nor orthovanadate had deleterious effects on CHO overexpressing EDG-8 cells as demonstrated by normal attachment and survival of the cells after 48 h of treatment (data not shown). DISCUSSION This study demonstrates that EDG-8, originally cloned as a GPCR termed NRG-1 (6), binds and is activated by S1P when heterologously expressed in CHO cells. CHO cells were selected to characterize EDG-8 because this cell line does not readily respond to S1P in the absence of exogenous EDG gene transfer (20,23,25). Furthermore, many studies relating to EDG-1, -3, and -5 signaling have been performed in CHO cells, thereby providing a common background for direct comparison to EDG-8 signaling (Tables I and II). The affinity of S1P for EDG-8 (K d ϭ 6 nM) is similar to previously reported affinity values obtained for EDG-1, -3, and -5 (K d ϭ 8 -26 nM) (15,25). Based upon the lysophospholipid ligands that were tested for their ability to displace [ 32 P]S1P binding, the pharmacological profile of EDG-8 closely resembles other S1P-binding EDG-1 receptors. Moreover, similar to our previous results with EDG-1, -3, and -5, SPC (15, 36) did not compete with [ 32 P]S1P

FIG. 3. EDG-8 mediates repression of ERK1/2 phosphorylation.
A, CHO cells transiently transfected with vector (V) or EDG-8 were pretreated with vehicle or PTX (100 ng/ml, 18 h). After serum stimulation (10%; 2 h), cells were incubated with 1 M S1P for the indicated times. A representative Western blot is shown. Relative phospho-ERK1 and phospho-ERK2 levels were normalized to total ERK2 and plotted as the means Ϯ S.E. of three independent experiments. Expression of EDG-8 in transfected cells was verified using an anti-HA antibody. B, CHO cells transiently transfected with EDG-8 were treated with 1 M of the indicated lipids (60 min). ERK1/2 phosphorylation was detected in cell lysates using a phosphorylation-specific ERK antibody. EDG-8 expression was monitored using an anti-HA antibody. A representative Western blot is shown from experiments that were repeated at least twice. dh-S1P, dihydro-S1P; Sph, sphingosine. After serum stimulation (10%; 2 h) cells were incubated with the indicated lipids (1 M, 60 min). ERK1/2 MAP kinase was immunoprecipitated with an anti-ERK1/2 MAP kinase antibody, and an immunocomplex kinase assay was performed as described under "Experimental Procedures" using an Elk-1 fusion protein as a substrate. EDG-8 expression was verified using an anti-HA antibody. B, ERK activity was expressed as a percentage of vehicle-treated cells (mean Ϯ S.E. of n ϭ 3-4). An asterisk indicates statistically significant difference when compared with untreated EDG-8-transfected cells as determined by Student's t test (p Ͻ 0.05). A # indicates statistically significant difference when compared with OA as determined by Student's t test (p Ͻ 0.05). OA, okadaic acid; OV, orthovanadate.
for binding to EDG-8. This is not surprising because recently a unique receptor for SPC, known as ORG1, has been identified (50). In contrast, others have previously demonstrated that SPC activates endogenous G i in HEK293 cells heterologously expressing EDG-1 (30) and induces Ca 2ϩ mobilization in CHO cells transfected with EDG-5 (23). Taken together, the binding data presented here establish that EDG-8 is a specific S1P and  8. EDG-8 inhibition of proliferation is orthovanadate-sensitive and PTX-insensitive. EDG-8-or vector (V)-transfected CHO cells were plated at low density and washed after 18 h. Cells were then cultured with 10% charcoal-treated dialyzed fetal bovine serum in the presence of 1 M S1P or vehicle (t ϭ 0). The inhibitors PTX (100 ng/ml) and orthovanadate (2 M) were added 3 h prior to S1P treatment. After 48 h of S1P or vehicle addition, cell proliferation was quantitated as described under "Experimental Procedures." Data represent the means Ϯ S.E. of three to six independent experiments and are expressed as fold increase in cell number following S1P or vehicle treatment. An asterisk indicates statistically significant difference when compared with transfected cells not treated with S1P as determined by Student's t test (p Ͻ 0.05). OV, orthovanadate. dihydro-S1P receptor.
EDG-8 resembles EDG-1 by inhibiting adenylyl cyclase via a PTX-sensitive mechanism in CHO cells, implicating G i /G o proteins in the signaling process. [ 35 S]GTP␥S binding assays corroborate the coupling of EDG-8 with G i . Interestingly, LPA at a relatively high concentration (1 M) was able to weakly inhibit cAMP formation in CHO cells expressing EDG-8. The ability of LPA to weakly inhibit cAMP formation but not displace [ 32 P]S1P binding in EDG-8-transfected cells (present study) is reminiscent of findings that LPA does not displace [ 32 P]S1P binding in EDG-1-transfected cells but promotes ERK1/2 activation (51). It has been proposed that EDG-1 contains distinct sites for S1P and LPA (51). Our data with EDG-8 supports a similar two-site model. However, this hypothesis awaits definitive characterization of the binding sites by sitedirected mutagenesis. Notwithstanding, our displacement binding studies suggest that EDG-8 is not an LPA receptor. Identical results were obtained in HEK293T cells overexpressing EDG-8 (14). Indeed, it has been convincingly demonstrated that LPA is the high affinity ligand for EDG-2, -4, and -7 (1,5,8).
Although CHO cells transfected with EDG-1, -3, or -5 can activate ERK in response to S1P in a PTX-sensitive and Rasdependent manner (17,20,23,24), we found instead that S1P inhibited ERK activity in EDG-8-transfected CHO cells. Although the majority of GPCRs studied to date (including the EDG receptors) appear to couple positively with ERK1/2 (52), examples of GPCR-mediated repression of ERK1/2 activity have been reported (48,49). In cultured neurons, AT 1 receptors activate ERK1/2, whereas AT 2 receptors inhibit serum-activated ERK1/2 (48). Considering that the MAP kinases are important in growth, differentiation, and apoptosis (52), it has been proposed that the antagonistic modulation of ERK activity by different receptors within the same gene family acts as a molecular counterbalance system (48). Hence, the ability of EDG-1, -3, and -5 to stimulate ERK1/2 activity and EDG-8 to repress it parallels the AT 1 /AT 2 receptor system. The finding that expression of EDG-8 but not EDG-1 or EDG-3 in PC12 cells is chronically down-regulated by growth factors (6) provides a potential mechanism for fine tuning this counterbalance system. Such a scenario has been demonstrated in the AT receptor family, where AT 2 receptor expression is up-regulated following cellular injury to antagonize AT 1 signaling (53).
The exact physiological role of ERK inhibition by GPCRs remains to be elucidated. EDG-1, -3, and -5 have been demonstrated to mediate PTX-sensitive cell proliferation (28,42) in an apparent ERK-dependent manner (27). Moreover, in endothelial cells, growth factor-and S1P-induced DNA synthesis and ERK activation are inhibited by PD98059, an inhibitor of ERK signaling (27,54). Interestingly, EDG-8 mediates antiproliferative effects in a PTX-insensitive and orthovanadate-sensitive manner. These observations suggest that the binding of S1P to EDG-8 might negatively regulate cellular proliferation by reducing ERK1/2 activity via a protein-tyrosine phosphatase. Analogously, AT 1 receptors promote cell proliferation and ERK activation, whereas AT 2 receptors exert antiproliferative effects and suppress ERK activation (48,49,55).
Several studies have linked protein serine/threonine phosphatases and/or protein-tyrosine phosphatases to AT 2 receptormediated repression of serum-activated ERK1/2 (48,49,53,56). Depending on the cell type, the protein phosphatases SHP-1, PP2A, and MKP-1 have each been implicated in inhibition of ERK1/2 activity. The decrease in ERK1/2 activity by AT 2 receptors in cultured neurons occurs through a PTX-and okadaic acid-sensitive pathway, implicating the involvement of PP2A (48). However, in N1E-115 neuroblastoma cells, AT 2 receptors act through a PTX-insensitive and orthovanadatesensitive pathway, suggesting the involvement of SHP-1 (49). Inhibition of ERK1/2 activity by S1P in CHO cells overexpressing EDG-8 was PTX-insensitive, orthovanadate-sensitive, and only weakly inhibited by okadaic acid. Future studies are warranted to define the exact nature of the phosphatase involved in EDG-8-mediated inhibition of the ERKs.
To determine whether potential inhibitory effects on ERK1/2 activity by activation of EDG-1, -3, or -5 might be masked by the interactions of these receptors with G i (resulting in ERK1/2 activation), we treated CHO cells with PTX. Regardless of which EDG receptor was expressed, PTX treatment did not facilitate S1P-mediated repression of serum-activated ERK1/2. These findings were not totally unexpected because EDG-1 appears to couple only to G i (30), and the inhibitory ERK response elicited by EDG-8 is PTX-insensitive. In addition, EDG-3 and EDG-5 are capable of coupling to PI turnover and Ca 2ϩ mobilization (25) via G q/11 (30), both signaling events that can lead to activation of the ERKs (57-59). EDG-8 does not couple to G q/11 , and thus, to PI turnover or Ca 2ϩ mobilization, which likely explains why this receptor does not mediate ERK activation if it is assumed that coupling of EDG-8 to other G protein(s) may mask the G i /ERK activation pathway. G␣ 12 is a likely candidate G protein for the inhibitory effects of EDG-8 on ERK activity. Mutationally activated G␣ 12 and G␣ 13 , but not G␣ q , G␣ s , or G␣ i2 , have been shown to inhibit EGF stimulation of ERK activity in COS-7 cells in a Ras-and Raf-independent manner (60). Furthermore, mutationally activated G␣ 12 and G␣ 13 stimulate JNK via Cdc42 and MEKK (60). Whereas expression of the activated G␣ 12 mutant in HEK293 cells does not appreciably stimulate ERK2 activity, JNK activity is increased by this mutant and involves the small GTPases Rac and Cdc42 (61). These findings are consistent with this study, demonstrating that both ERK inhibition and JNK activation by EDG-8 are PTX-insensitive.
It is of interest that EDG-8 appears to selectively activate the p54 JNK isoform. Other GPCRs, such as the ␣ 1A -adrenergic receptor, differentially activate specific JNK isoforms (62). Another member of the EDG family, EDG-5, has been shown to activate JNKs in a PTX-insensitive manner (23). In that study only the p46 JNK isoform activity was measured.
In contrast to the PTX-insensitive S1P-induced activation of p38 in CHO-EDG-5 cells (23), an increase in p38 activity was not observed in S1P-treated CHO overexpressing EDG-8 transfectants. The lack of p38 activation in edg8-transfected CHO cells is another distinguishing feature of the signaling pathways mediated by EDG-5 and EDG-8 receptors.
While this manuscript was in review, Im et al. (14) described the recloning of NRG-1 (EDG-8) from rat brain. In that report, EDG-8 was shown to bind S1P, inhibit forskolin-stimulated cAMP accumulation in rat hepatoma Rh7777 cells, and couple to G i in Xenopus laevis oocytes, in agreement with our data. In conclusion, we have demonstrated fundamental differences in G protein coupling between the four known high affinity S1P receptors, EDG-1, -3, -5, and-8. Furthermore, we have identified differences in effector pathways that are potentially related to differences in G protein coupling.