Inhibitory and Stimulatory Regulation of Rac and Cell Motility by the G_12/13-Rho and G_i Pathways Integrated Downstream of a Single G Protein-Coupled Sphingosine-1-Phosphate Receptor Isoform

Materials.

S1P and 1-monooleoyl lysophosphatidic acid (LPA) were purchased from Biomol (Plymouth Meeting, Pa.) and Avanti (Birmingham, Ala.), respectively. They were dissolved, aliquoted, and stored as described previously (29). Recombinant human insulin-like growth factor I was purchased from R&D Systems. A rabbit polyclonal anti-Gα_q/11 and a mouse monoclonal anti-Rac antibody were purchased from Upstate Biotechnology. Rabbit polyclonal antibodies against Gα_s/olf (C-18), Gα_i3 (C-10), anti-Gα₁₂ (S-20), Gα₁₃ (A-20), and GRK2 (H-222) and a mouse monoclonal anti-RhoA antibody were bought from Santa Cruz Biotechnology. A rabbit polyclonal anti-Gα₁₃ antibody (371778) was bought from Calbiochem. A mouse monoclonal anti-ERK1/2 antibody (clone 03-6600) was obtained from Zymed Laboratories Inc. An anti-FLAG M2 antibody and tetramethyl rhodamine isocyanate (TRITC)-labeled phalloidin were obtained from Sigma. PTX was bought from List Biological Laboratories. AlCl₃ and NaF were obtained from Wako Pure Chemicals (Osaka, Japan) and were added to media at a molar ratio of 1:4 (AlCl₃ to NaF) to generate AlF₄⁻. Y-27632 and HA-1077 were supplied by Mitsubishi Pharma (Wako, Japan) and Asahi Chemical Industry (Fuji, Japan), respectively. Botulinum C3 toxin, the glutathione S-transferase (GST)-human PAK1 (amino acids 75 to 131) fusion protein, GST-mouse rhotekin (amino acids 7 to 89), and a mouse monoclonal anti-myc antibody (9E10) were prepared as described previously (41).

Plasmids, adenoviruses, and transfections.

pME18S-myc-N¹⁹RhoA, pME18S-myc-V¹⁴RhoA, pGEX-2T-rhotekin, pGEX-2T-PAK, and an adenovirus encoding myc-N¹⁹RhoA were described previously (31, 33, 35). cDNAs encoding full-length mouse Gα_s, Gα_i2, Gα₁₂, Gα₁₃, and Gα_q, and the C-terminal peptide of human β-adrenergic receptor kinase (βARK-CT; βARK residues 495 to 689), were obtained by reverse transcription-PCR (RT-PCR) from total mouse brain RNA and human brain RNA (Sawady Technology, Tokyo, Japan), respectively. PCR-based methods were used to generate the cDNAs encoding myc-tagged C-terminal regions of Gα_s (residues 319 to 377), Gα₁₂ (residues 326 to 379), Gα₁₃ (residues 321 to 377), and Gα_q (residues 306 to 359), which were designated Gα_s-CT, Gα₁₂-CT, Gα₁₃-CT, and Gα_q-CT; S1P₂ with a FLAG-epitope at its N terminus (FLAG-S1P₂); and fusion receptors S1P₂-Gα₁₂, S1P₂-Gα₁₃, and S1P₂-Gα_q, which have full-length Gα₁₂, Gα₁₃, and Gα_q, respectively, fused to the C terminus of S1P₂. The cDNAs of full-length Gα proteins, their C termini, and the FLAG-S1P₂ and S1P₂-Gα fusion receptors were ligated onto the mammalian expression vector pCAGGS (a gift from M. Miyazaki, Osaka University Medical School) to generate pCAGGS-Gα_i2, pCAGGS-Gα₁₂, pCAGGS-Gα₁₃, pCAGGS-Gα_q, pCAGGS-Gα_s-CT, pCAGGS-Gα₁₂-CT, pCAGGS-Gα₁₃-CT, pCAGGS-Gα_q-CT, pCAGGS-FLAG-S1P₂, pCAGGS-S1P₂-Gα₁₂, pCAGGS-S1P₂-Gα₁₃, and pCAGGS-S1P₂-Gα_q, respectively. βARK-CT cDNA was ligated onto the mammalian expression vector pME18S (a gift from K. Maruyama, Tokyo Medical and Dental University) to generate pME18S-βARK-CT. Replication-deficient adenoviruses encoding Gα₁₂-CT, Gα₁₃-CT, and Gα_q-CT were generated and amplified as described previously (8). pCAGGS-LacZ and an adenovirus encoding LacZ were kindly donated by I. Saito (Institute of Medical Sciences, University of Tokyo).

The cells were infected with adenoviruses at a multiplicity of infection of 200 by incubating cells with an adenovirus-containing medium for 1 h, which conferred successful gene transduction in nearly 100% of cells. After recovery in growth medium for 24 h, the cells were serum deprived for 24 h before experiments.

Transient transfection with expression plasmid vectors was carried out by using LipofectAMINE (Invitrogen) 48 h before each experiment. To study the migration of transiently transfected cells, the cells were cotransfected with either one of the Gα-CT expression plasmids, the βARK-CT expression plasmid, or the empty vector and pCAGGS-LacZ as a transfection marker (31) for 3 h. In some experiments in which the actin cytoskeleton was evaluated (Fig. 2C), the green fluorescent protein (GFP) expression vector pEGFP-C1 (Clontech) was employed as a transfection marker. After recovery in growth medium for 21 h, the cells were serum deprived for 24 h.

To establish CHO cells that stably express S1P₂-Gα fusion receptors, cells were cotransfected with pCAGGS-S1P₂-Gα and the neomycin resistance gene expression vector pKM3 (27) and were selected in the presence of 0.7 mg of G418 (Nacalai, Kyoto, Japan)/ml. To establish CHO-S1P₂ cells that stably express full-length Gα protein and Gα_q-CT, CHO-S1P₂ cells were cotransfected with either pCAGGS-Gα, pCAGGS-Gα_q-CT, or the Zeocin resistance gene expression vector pCMV/Zeo (Invitrogen) and were selected in the presence of 50 μg of Zeocin (Invitrogen)/ml and 0.7 mg of G418/ml. Cloned cells were isolated and tested for expression of transduced genes.

In the experiments using CHO-S1P₂ cells that express N¹⁹Rho or V¹⁴Rho, CHO-S1P₂ cells were cotransfected with either pME18S-myc-N¹⁹RhoA or pME18S-myc-V¹⁴RhoA and pCMV/Zeo and were selected in the presence of 50 μg of Zeocin/ml. The Zeocin-resistant cell populations were employed in these experiments.

Cells.

CHO-K1 (CHO) cells, Swiss 3T3 cells, and COS7 cells were grown in Ham's F-12 (CHO) or Dulbecco's modified Eagle medium (3T3 and COS7) supplemented with 10% fetal bovine serum (Equitech-Bio, Ingram, Tex.), 100 U of penicillin/ml, and 100 μg of streptomycin/ml (Wako Pure Chemicals). CHO cells that stably overexpress either S1P₁, S1P₂, or S1P₃, i.e., CHO-S1P₁, CHO-S1P₂, and CHO-S1P₃ cells, respectively, have been described previously (13, 29, 30) and were maintained in the presence of 0.7 mg of G418/ml. Cells were treated with C3 toxin (10 μg/ml) in F-12 medium containing 10% fetal bovine serum for 48 h and then in serum-free F-12 medium for a further 24 h. PTX (200 ng/ml) treatment was carried out by incubating cells in serum-free F-12 medium containing PTX for 24 h before experiments.

Transwell migration assay.

Chemotactic migration of cells was measured in a modified Boyden chamber (Neuroprobe) using polycarbonate filters with 8-μm pores as described in detail previously (31, 33). In migration assays using transiently transfected cell populations, the migratory cells attached to the lower side of the membrane were subjected to stainining with 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside as a substrate. The number of migratory cells staining positive with 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside was determined by using a microplate reader as described above.

Determination of the activities of Rho and Rac.

Pulldown assay methods to determine GTP-bound active forms of Rho and Rac have been described in detail previously (31, 33, 35). Briefly, cell extracts were incubated with the GST-rhotekin Rho-binding domain (for determination of Rho activity) or the GST-PAK CRIB domain (for determination of Rac activity) immobilized to glutathione-Sepharose 4B beads (Pharmacia Amersham Biotech) at 4°C for 45 min, followed by three washes. Bound Rho and Rac proteins were quantitatively detected by Western blotting using specific, monoclonal antibodies against RhoA and Rac.

Western blotting, [Ca²⁺]_i measurement, and fluorescence microscopy.

Western blotting was performed as described previously (29). The band shift of activated p42 and p44 extracellular signal-regulated kinase (ERK) was detected by Western blot analysis of total cell lysates with a mouse monoclonal anti-ERK antibody (8). Intracellular free Ca²⁺ concentration ([Ca²⁺]_i) was measured as described previously (29) in Fura-2-loaded cells with a CAF-110 spectrofluorimeter (Japan Spectroscopy, Inc., Tokyo, Japan) with excitation at 340 and 380 nm and emission at 500 nm.

To evaluate actin cytoskeletons, cells were transfected as indicated 48 h before experiments and were serum starved for 24 h. After treatment with receptor agonists and/or Rho kinase inhibitors for indicated times, the cells were fixed in 3.7% formaldehyde in phosphate-buffered saline and processed as described previously (41). F-actin was visualized with TRITC-labeled phalloidin under an inverted fluorescence microscope IX70 (Olympus, Tokyo, Japan).

Statistics.

Values are presented as means ± standard errors of three or more determinations and are representative of at least two independent experiments with similar results.

AlF₄⁻ mimics S1P₂ actions in inhibiting Rac and migration in IGF I-stimulated cells.

Most of the actions of GPCRs are mediated through heterotrimeric G protein coupling. However, recent studies provide evidence that mechanisms independent of heterotrimeric G protein coupling mediate certain actions of GPCRs, which include regulation of a Na⁺-H⁺ exchanger and a Ca²⁺ channel, as well as activation of STAT (5). To understand in more depth the molecular mechanisms underlying GPCR-mediated inhibition of cell migration, we first investigated whether S1P₂-mediated Rac inhibition occurs through heterotrimeric G protein coupling, and, if so, which heterotrimeric G protein mediates Rac inhibition. In CHO cells expressing the S1P₂ receptor (CHO-S1P₂), but not in naive CHO cells or CHO cells expressing S1P₁ or S1P₃, S1P dose-dependently and potently inhibited IGF I-directed chemotaxis, which is a Rac-dependent process (31). Direct stimulation of the heterotrimeric G proteins with AlF₄⁻ (17) dose-dependently inhibited chemotaxis of CHO-S1P₂ cells toward IGF I, like S1P stimulation of the S1P₂ receptor (Fig. 1A). AlF₄⁻ also mimicked the action of S1P₂ in inhibiting IGF I-induced increases in cellular amounts of a GTP-bound, active form of Rac (GTP-Rac) (Fig. 1B) and in stimulating RhoA (Fig. 1C), ERK (Fig. 1D), and Ca²⁺ mobilization (data not shown). These observations strongly suggest that the heterotrimeric G protein(s) mediated inhibition of Rac and cell migration as well as the other actions of S1P₂.

Endogenous G₁₂ and G₁₃ couple S1P₂ to inhibition of cell migration.

To examine which member of the heterotrimeric G proteins is responsible for mediating suppression of cell migration, we determined the effects of specific inhibition of receptor-G protein coupling by either transient expression of C-terminal peptides (1, 11) of heterotrimeric G protein α subunits (Gα-CTs) or pretreatment with PTX. Shown in Fig. 2A are Western blot analyses of the expression of Gα_s-CT, Gα_q-CT, Gα₁₂-CT, and Gα₁₃-CT. CHO-S1P₂ cells were cotransfected with one of these peptides and β-galactosidase (LacZ) and subjected to a migration assay (31). As in naive CHO-S1P₂ cells (31), S1P inhibited IGF I-directed chemotaxis in vector-transfected CHO-S1P₂ cells with a bell-shaped dose-response curve and maximal inhibition at 10⁻⁷ M (Fig. 2B).Neither expression of any of these C-terminal peptides nor pretreatment with PTX affected chemotaxis toward IGF I in the absence of S1P (Fig. 2B). Expression of Gα₁₂-CT or Gα₁₃-CT, but not Gα_s-CT or Gα_q-CT, specifically abolished the S1P inhibition of IGF I-directed chemotaxis. We confirmed that expression of Gα_q-CT effectively inhibited the S1P-induced increase in [Ca²⁺]_i, a G_q-mediated response, compared to that with the vector control in S1P-expressing cells (Fig. 2C). Inhibition of G_i/o by PTX pretreatment, and expression of βARK-CT, which acts as a scavenger for βγ subunits (20), to a lesser extent, rather potentiated S1P inhibition of IGF I-directed chemotaxis at lower S1P concentrations. PTX pretreatment or the expression of βARK-CT substantially attenuated S1P-induced ERK activation (data not shown), confirming the effectiveness of PTX and βARK at inhibiting G_i. These observations are consistent with the notion that endogenously expressed G₁₂ and G₁₃ are responsible for mediating a signal that leads to inhibition of migration upon S1P stimulation of the S1P₂ receptor.

It has been reported for Swiss 3T3 fibroblasts (12) that endothelin-1 (ET1) and LPA induce stress fiber formation through G₁₂ and G₁₃, respectively. By employing Swiss 3T3 cells and these GPCR agonists, we determined the specificity of the inhibitory actions of Gα₁₂-CT and Gα₁₃-CT. We observed that expression of Gα₁₃-CT abolished LPA-induced stress fiber formation (Fig. 2Da, b, e, and f) but not ET1-induced stress fiber formation (Fig. 2Di, j, m, and n), while expression of Gα₁₂-CT abolished both ET1- and LPA-induced stress fiber formation (Fig. 2Dc, d, k, and l). The results indicate that the Gα₁₃-CT peptide acts as a selective inhibitor for G₁₃ whereas the Gα₁₂-CT peptide acts as an inhibitor for both G₁₂ and G₁₃. Together with the observation using βARK-CT (Fig. 2B), we conclude that the α subunits, but not the βγ dimer, of G₁₃ or both G₁₂ and G₁₃ mediate inhibition of migration.

Endogenous G₁₂ and G₁₃ couple S1P₂ to inhibition of Rac and stimulation of Rho.

We next expressed myc-tagged Gα₁₂-CT, Gα₁₃-CT, and Gα_q-CT, and LacZ as a control, by adenovirus-mediated gene transduction in CHO-S1P₂ cells, and we determined the effects of their expression on the activities of RhoA and Rac (Fig. 3) and also on the actin cytoskeleton (Fig. 4). Expression of the inhibitor proteins was confirmed by Western blot analysis (Fig. 3A). Expression of any of these C-terminal peptides did not affect IGF I-induced Rac activation (Fig. 3B) or membrane ruffling (Fig. 4) in the absence of S1P, nor did it affect S1P stimulation of ERK (Fig. 3D), indicating that their expression did not compromise cellular activity in a nonspecific manner. Interestingly, however, expression of either Gα₁₂-CT or Gα₁₃-CT, but not Gα_q-CT or LacZ, abolished S1P inhibition of Rac activation in response to IGF I (Fig. 3B). Expression of either Gα₁₂-CT or Gα₁₃-CT, but not Gα_q-CT, also greatly inhibited S1P-induced RhoA stimulation compared to LacZ transfection (Fig. 3C). Consistent with the effects on Rho and Rac, expression of Gα₁₂-CT or Gα₁₃-CT, but not of Gα_q-CT or LacZ, abolished S1P suppression of peripheral actin filament assembly in response to IGF I, as well as S1P stimulation of stress fiber formation (Fig. 4). These results are consistent with the observations on cell migration and indicate that G₁₂ and G₁₃ are responsible for mediating suppressive effects of S1P on cellular Rac activity, membrane ruffling, and cell migration in CHO-S1P₂ cells.

S1P₂-Gα₁₂ and S1P₂-Gα₁₃ fusion receptors, but not S1P₂-Gα_q, mediate inhibition of Rac and migration.

Consistent with the observations showing functional coupling of S1P₂ to G₁₂ and G₁₃ (Fig. 2 to 4), we observed coimmunoprecipitation of S1P₂ and either G₁₂ or G₁₃ from the cells coexpressing these molecules (Fig. 5A). We further studied and compared the effects of S1P on IGF I-directed migration in CHO cells that stably expressed fusion receptors designated S1P₂-Gα₁₂, S1P₂-Gα₁₃, and S1P₂-Gα_q, which have either of the full-length Gα subunit sequences fused to the C terminus of S1P₂ (37). In CHO cells expressing either the S1P₂-Gα₁₂ or the S1P₂-Gα₁₃ fusion receptor, S1P potently inhibited chemotaxis toward IGF I (Fig. 5B). Adenovirus-mediated expression of Gα₁₂-CT or Gα₁₃-CT completely abolished S1P inhibition of IGF I-directed chemotaxis in cells expressing the respective fusion receptor S1P₂-Gα₁₂ or S1P₂-Gα₁₃. Cells expressing S1P₂-Gα_q showed a prominent increase in the [Ca²⁺]_i in response to S1P (Fig. 5C); however, they responded to S1P with only a marginal inhibition of cell migration (Fig. 5B). In vector-transfected cells and cells expressing either of the three fusion receptors, IGF I similarly stimulated Rac activity (Fig. 5D). In cells expressing either S1P₂-Gα₁₂ or S1P₂-Gα₁₃, but not vector-transfected cells or cells expressing S1P₂-Gα_q, S1P induced a nearly complete inhibition of Rac activation in response to IGF I, as in CHO-S1P₂ cells (Fig. 5D). As expected, cells expressing S1P₂-Gα₁₂ or S1P₂-Gα₁₃, but not vector-transfected cells or cells expressing S1P₂-Gα_q, showed strong activation of RhoA in response to S1P (Fig. 5D). These observations provided further evidence that G_12/13 coupled with S1P₂ for inhibition of Rac and migration.

Rho mediates inhibition of cellular Rac activity and migration through a mechanism not involving Rho kinase.

We tested the possibility that Rho, which is an effector of G_12/13, was involved in the signaling pathway leading to suppression of Rac activity. Pretreatment of S1P₂-expressing cells with botulinum C3 toxin, which inactivates Rho, did not affect migration in response to IGF I; however, it completely abolished the S1P₂-mediated inhibition of IGF I-directed migration (Fig. 6A). Similarly, adenovirus-mediated expression of N¹⁹RhoA did not affect chemotaxis toward IGF I but totally abolished S1P inhibition of IGF I-directed migration (Fig. 6A). Consistent with these observations, either C3 treatment or N¹⁹RhoA expression abolished S1P inhibition of IGF I-induced Rac stimulation, while these treatments did not affect Rac activation in response to IGF I alone (Fig. 6B). On the other hand, we observed that cells that stably expressed a myc-tagged constitutively active RhoA mutant, myc-V¹⁴RhoA, showed reductions in both chemotaxis and Rac activation in response to IGF I, compared to vector-control cells (Fig. 6C and 6D).

We next studied whether Rho kinase, which is one of the direct effectors of Rho (23, 46), was involved in S1P₂-mediated suppression of Rac activity and migration. It was previously reported that a Rho kinase inhibitor abolished V¹⁴Rho-induced inhibition of Rac activity (50). HA-1077 and Y-27638, which are two structurally unrelated Rho kinase inhibitors (26, 49), effectively inhibited S1P-induced stress fiber formation in CHO-S1P₂ cells (Fig. 7); however, they were totally ineffective in preventing the S1P₂-mediated inhibition of IGF I-directed cell migration (Fig. 6E), Rac activation (Fig. 6F), or membrane ruffling (Fig. 7). In contrast, C3 toxin abolished the S1P₂-mediated inhibition of IGF I-induced membrane ruffling (Fig. 7f) as well as induction of stress fiber formation (Fig. 7j). S1P₂ mediates actin cytoskeletal changes in two ways through Rho: it inhibits membrane ruffling in IGF I-stimulated cells, and it stimulates stress fiber formation. These observations clearly indicate that inhibition of membrane ruffle formation requires a Rho-dependent signal other than the Rho kinase pathway, whereas induction of stress fiber formation requires Rho kinase.

Overexpression of Gα_i counteracts AlF₄⁻- and S1P-induced inhibition of Rac and migration.

S1P₂ couples not only to G_12/13 but also to G_i (13, 30, 49). PTX pretreatment of CHO-S1P₂cells potentiated S1P inhibition of chemotaxis (Fig. 1B), suggesting that G_i conveyed a signal which counteracted inhibition of migration. To study in more depth the role of Gα_i in the regulation of Rac and cell migration by the S1P receptor, we evaluated the effects of overexpression of Gα_i2, which is an endogenously expressed Gα_i isoform in CHO cells, on AlF₄⁻- and S1P-induced inhibition of Rac and migration in CHO-S1P₂ cells (Fig. 8). In cells overexpressing Gα_i2, S1P by itself stimulated chemotaxis and Rac moderately, unlike the situation in the vector control cells (data not shown). Overexpression of Gα_i2 nearly completely abolished AlF₄⁻- and S1P-induced inhibition of chemotaxis (Fig. 8B and C). The results contrast sharply with those obtained with overexpression of either Gα₁₂or Gα₁₃, which potentiated AlF₄⁻- and S1P-induced inhibition of IGF I-directed chemotaxis (Fig. 8B and C). Overexpression of Gα_q did not affect inhibition by AlF₄⁻ or S1P. In agreement with these observations on migration, overexpression of Gα_i2 nearly abolished S1P inhibition of IGF I-induced stimulation of Rac, whereas overexpression of Gα₁₂ and Gα₁₃ potentiated S1P inhibition of IGF I stimulation of Rac compared to that in vector control cells (Fig. 8D). Expression of Gα_q did not affect S1P inhibition of Rac activity. Overexpression of these Gα subunits did not affect chemotaxis or Rac activation in response to IGF I alone (Fig. 8C and D). Thus, G_i generates a stimulatory signal for Rac and consequently migration to antagonize G_12/13-mediated inhibition of Rac and migration in S1P receptor signaling.

S1P₃ mediates S1P inhibition, rather than stimulation, of Rac and cell migration upon G_i inactivation.

As we reported previously (31), S1P by itself directed chemotaxis in both S1P₁- and S1P₃-expressing CHO cells (CHO-S1P₁ and CHO-S1P₃ cells, respectively) (Fig. 9A and B). This contrasts with CHO-S1P₂ cells, in which S1P by itself does not stimulate chemotaxis (31). PTX pretreatment of both CHO-S1P₁ and CHO-S1P₃ cells totally inhibited migration toward S1P, indicating that S1P-directed chemotaxis is G_i/o dependent (Fig. 9A and B). When CHO-S1P₁ cells were stimulated with a combination of S1P and IGF I, they showed a chemotactic response slightly greater than that to stimulation with IGF I alone (Fig. 9C). PTX pretreatment did not affect CHO-S1P₁ cell migration stimulated by IGF I alone, and it slightly reduced cell migration stimulated by the combination of S1P and IGF I. In CHO-S1P₃ cells, the combination of S1P and IGF I induced a stimulation of chemotaxis slightly larger than that with IGF I alone, as in CHO-S1P₁ cells (Fig. 9D). PTX did not affect the response to IGF I alone in this cell type, either. In contrast to CHO-S1P₁ cells, however, inactivation of G_i with PTX in CHO-S1P₃ cells dramatically changed the responses to S1P: the cells showed inhibition, rather than stimulation, of IGF I-directed chemotaxis in response to S1P, with a nearly complete inhibition at 10⁻⁶ M (Fig. 9D). Moreover, adenovirus-mediated expression of either Gα₁₂-CT or Gα₁₃-CT abolished S1P inhibition of IGF I-directed chemotaxis in PTX-treated CHO-S1P₃ cells (Fig. 9D).

In CHO-S1P₁ cells S1P by itself stimulated Rac, which was totally inhibited by PTX (Fig. 10A). IGF I also stimulated Rac, but in a PTX-insensitive manner. The combination of S1P and IGF I induced a slightly but consistently greater response in Rac activation compared to that with stimulation by either alone. In CHO-S1P₃ cells S1P by itself stimulated Rac in a PTX-sensitive manner (Fig. 10C), which was similar to the findings for CHO-S1P₁ cells. In contrast to the situation in PTX-pretreated CHO-S1P₁ cells, however, in PTX-pretreated CHO-S1P₃ cells S1P markedly inhibited IGF I-stimulation of Rac (Fig. 10C), which was consistent with the S1P inhibition of IGF I-directed chemotaxis. Expression of either Gα₁₂-CT or Gα₁₃-CT abolished the S1P inhibition of Rac in PTX-pretreated, IGF I-stimulated CHO-S1P₃ cells (Fig. 10C), just as with cell migration (Fig. 9D). In CHO-S1P₃ cells, S1P induced stimulation of RhoA via G_12/13 irrespective of PTX pretreatment (Fig. 10D). S1P did not change the level of GTP-RhoA in CHO-S1P₁ cells (Fig. 10B). Thus, inactivation of G_i in CHO-S1P₃ cells converts S1P-induced stimulation of Rac and migration to inhibition of Rac and migration. These observations indicate that the S1P receptor isoforms S1P₁, S1P₂, and S1P₃ exert distinct regulatory actions on Rac and cell migration through differential coupling to G_12/13 and G_i, which convey signals to inhibit and stimulate Rac and migration, respectively.