Activated Gα₁₃ Impairs Cell Invasiveness through p190RhoGAP-Mediated Inhibition of RhoA Activity

The GTPase RhoA is a downstream target of heterotrimeric G₁₃ proteins and plays key roles in cell migration and invasion. Here, we show that expression in human melanoma cells of a constitutively active, GTPase-deficient Gα₁₃ form (Gα₁₃QL) or lysophosphatidylcholine (LPC)-promoted signaling through Gα₁₃-coupled receptors led to a blockade of chemokine-stimulated RhoA activation and cell invasion that was rescued by active RhoA. Melanoma cells expressing Gα₁₃QL or cells stimulated with LPC displayed an increase in p190RhoGAP activation, and defects in RhoA activation and invasion were recovered by knocking down p190RhoGAP expression, thus identifying this GTPase-activating protein (GAP) protein as a downstream Gα₁₃ target that is responsible for these inhibitory responses. In addition, defective stress fiber assembly and reduced migration speed underlay inefficient invasion of Gα₁₃QL melanoma cells. Importantly, Gα₁₃QL expression in melanoma cells led to impairment in lung metastasis associated with prolonged survival in SCID mice. The data indicate that Gα₁₃-dependent downstream effects on RhoA activation and invasion tightly depend on cell type–specific GAP activities and that Gα₁₃-p190RhoGAP signaling might represent a potential target for intervention in melanoma metastasis. [Cancer Res 2008;68(20):8221–30]

Rho GTPases control the dynamics of the actin cytoskeleton during cell migration, contributing to the development of cell protrusions and adhesion at the leading edge and to retraction at the cell rear (1, 2). Their activity is tightly regulated by guanine-nucleotide exchange factors (GEF), which stimulate exchange of bound GDP by GTP, and inhibited by GTPase-activating proteins (GAP), which promote GTP hydrolysis (3, 4). Therefore, cell migration is finely regulated by the balance between GEF and GAP activities on Rho GTPases.

Activation of cell migration machinery is required at different steps of metastasis, such as in invasion across basement membranes and interstitial tissues, intravasation into blood or lymphatic circulation, and extravasation and invasion through subendothelial basement membranes for colonization of distant organs (5, 6). Involvement of Rho GTPases in cancer is well documented (reviewed in ref. 7), providing control of both cell migration and growth. RhoA and RhoC are highly expressed in colon, breast, and lung carcinomas (8, 9), whereas overexpression of RhoC in melanoma leads to enhancement of cell metastasis (10).

Chemokines are chemotactic cytokines which stimulate cell migration and activation and exert their functions upon binding to heterotrimeric guanine nucleotide-binding (G) protein-coupled receptors (GPCR; refs. 11, 12). Activation of Rho GTPases represents a target of chemokine signaling, providing cell adhesion and directionality during migration. Most solid cancer cells express CXCR4, a receptor for the chemokine CXCL12 (also called SDF-1), which is expressed in lungs, bone marrow, and liver (13). Similarly to the key role of chemokines for the homing of immune cells during immune surveillance, chemokines contribute to tumor cell trafficking and colonization of organs during metastasis (14, 15).

CXCR4 is expressed on melanoma, a highly aggressive cancer when metastasis starts (16), providing activation of the cell invasive machinery upon interaction with CXCL12 (13, 17–19). CXCL12 stimulates RhoA activation on melanoma cells, and expression of a mutant dominant-negative (DN) form of Rho abolishes chemokine-promoted invasion (18). In addition, it was shown that the GEF Vav2 is an upstream molecule regulating CXCL12-mediated RhoA activation in these cells (20).

Interaction of chemokine receptor with their agonists elicits different cellular responses, which depend on initial receptor-heterotrimeric G-protein binding (11). Heterotrimeric G proteins consist of a α subunit and a complex formed by β and γ subunits (21, 22). Basally, Gβγ and GDP-bound Gα are associated, and upon interaction with an activated receptor, GTP replaces GDP and GTP-Gα dissociates from the Gβγ-dimer. These two elements interact with effector proteins leading to the activation of distinct signaling pathways. The GTPase activity inherent to Gα limits G-protein activation, as GTP hydrolysis causes reassociation of GDP-Gα and Gβγ. G proteins are classified into four subfamilies, G_s, G_i/o, G_q/11, and G_12/13, according to the Gα protein present in the complex (21, 22).

In addition to its well-known role on cell growth and transformation, G_12/13 proteins also regulate RhoA activation and cell migration (23). Thus, expression of constitutively active (CA) mutant forms of Gα₁₂ and Gα₁₃, which lack GTPase activity due to substitution of a glutamine (Q231 or Q226) to a leucine residue, stimulates Rho-dependent stress fiber and focal adhesion assembly (24, 25). In prostate and breast cancer, a key role for Gα_12/13 activation during tumor cell motility has been proposed (26, 27), as it was shown that G₁₂ signaling promotes Rho-mediated cell invasion in vitro.

In the present work, we studied how Gα₁₃ activation controls chemokine-stimulated human melanoma cell invasion and metastasis. Interestingly, we found that expression of a GTPase-deficient Gα₁₃ form (Gα₁₃QL) impaired melanoma cell invasion and in vivo metastasis that was associated with inhibition of RhoA activation. We provide here mechanistic characterization of the molecular components involved in these unexpected effects. The data suggest that cancer cell–type specific differences might underlie distinct Gα₁₃-dependent activation of signaling pathways leading to different biological responses.

Cells, antibodies, and reagents. The human melanoma cell lines BLM (28) and MeWo were cultured as reported (18). MDA-MB-231 human breast cancer and mouse fibroblastic Swiss 3T3 cells were grown in DMEM supplemented with 10% fetal bovine serum (BioWhittaker). Anti-Gα₁₃, anti-Gα₁₂, anti-RhoA, and anti-phosphotyrosine antibodies were purchased from Santa Cruz Biotechnology, anti-green fluorescent protein (GFP) from Molecular Probes, antibodies to paxillin from BD Biosciences, anti–β-actin from Sigma-Aldrich, and anti-p190RhoGAP from Upstate. Control P3X63 monoclonal antibody (mAb) was from Dr. Francisco Sánchez-Madrid (Hospital de la Princesa). Anti-CXCR4 mAb and CXCL12 were obtained from R&D Systems. Lysophosphatidylcholine (LPC) was from Sigma-Aldrich, U46619 from Alexis Biochemicals, and PP2 and PP3 from Calbiochem-Novabiochem Co.

Vectors, RNA interference, and transfections. Vectors coding for GFP-fused forms of wild-type Gα₁₃ (Gα₁₃wt) or Gα₁₃QL were generated by subcloning Gα₁₃ cDNAs (gifts from Dr. Piero Crespo, Universidad de Cantabria, and Dr. Silvio Gutkind, NIH) into pEGFP-C1 (Clontech). Vectors coding for Gα₁₃wt or Gα₁₂QL were purchased from UMR cDNA. Vectors coding for GFP-fused forms of wild-type RhoA and Rac1, DN N19-RhoA, and activated V14-RhoA and V12-Rac1 were gifts from Dr. Francisco Sánchez-Madrid. For small interfering RNA (siRNA), we used a control siRNA duplex (29) and designed three target-specific siRNA duplexes against Gα₁₃ mRNA. Sense strands were Gα₁₃ (1) 5′-GGCAUCCAUGAAUACGACUdTdT-3′, targets bases 610-630; Gα₁₃ (2) 5′-CUUUAUGGUGACCUCCUAUdTdT-3′, targets bases 1575-1595; and Gα₁₃ (3) 5′-GUUCAUGGUAUCGUGCAUGdTdT-3′, targets bases 2894-2914. In addition, we used two siRNAs against p190RhoGAP, p190A (1) and p190A (2), siGENOME duplex D-004158-03 and D-004158-04, respectively (Dharmacon). Cells were transiently transfected with expression vectors or siRNA (100 nmol/L) using Lipofectamine (Invitrogen Corp.) or X-tremeGENE (Roche Diagnostics), respectively, according to manufacturer's instructions. Transfectants were tested in the different assays 48 h posttransfection.

Invasion assays. Invasions were done as earlier reported (18). Briefly, cells were loaded on the upper compartments of invasion chambers coated with Matrigel (BD Biosciences). The lower compartments were filled with invasion medium with or without CXCL12. Invasive cells were fixed, stained, and counted under a microscope.

Immunoprecipitation, Western blotting, and GTPase assays. For immunoprecipitation, melanoma cells were lysed (30) and extracts were incubated with antibodies followed by specific coupling to protein A-Sepharose beads (Amersham Pharmacia Biotech). Proteins were eluted in Laemmli buffer, resolved by SDS-PAGE, and subjected to Western blotting with primary antibodies, followed by incubation with horseradish peroxidase–conjugated secondary antibodies and detection with SuperSignal chemiluminiscent substrate (Pierce). GTPase activity assays were performed, as previously reported (18). In brief, cells were incubated with or without CXCL12, and upon cell lysis, aliquots from extracts were kept for total lysate controls and the remaining volume incubated with GST-C21 (for RhoA) or GST-PAK-CD (for Rac1) fusion proteins (31) in the presence of glutathione-agarose beads. After elution of bound proteins, they were subjected to Western blotting with anti-RhoA or anti-Rac1 antibodies.

Retroviral gene transfer and animal studies. Gα₁₃wt or Gα₁₃QL cDNAs were cloned into pRETRO-bsd vector (gift from Dr. Reuven Agami, National Cancer Institute, Amsterdam, the Netherlands). Vectors were cotransfected by Lipofectamine with pNGVL-VSV-G and pNGVL-gag-pol vectors (gifts from Dr. Rafael Delgado, Hospital 12 de Octubre) into 293FT packaging cells. After 48 h, conditioned medium containing viral particles was used to infect BLM cells, which were selected with blasticidin (Invitrogen Corp.) for 3 wk. For xenografting studies, 6-wk to 9-wk sex-matched BALB/c SCID mice (Harlan), bred and maintained under specific pathogen-free conditions at the Centro de Investigaciones Biológicas Animal Facility, were injected s.c. in the lateral thoracic wall or i.v. into the tail vein with 1 × 10⁶ cells in 0.2 mL PBS. The Consejo Superior de Investigaciones Científicas Ethics Committee approved the protocols used for experiments with mice. Mice were inspected on a daily basis for local tumor growth and general condition and were killed when signs of respiratory stress were noted or when s.c. tumors reached a volume of 2.5 cm³.

Statistical analyses. Data were analyzed by one-way ANOVA followed by Tukey-Kramer multiple comparisons. In both analyses, the minimum acceptable level of significance was P < 0.05.

Expression of Gα₁₃QL in melanoma leads to inhibition of chemokine-promoted cell invasion and RhoA activation. The chemokine CXCL12 activates RhoA in the highly aggressive BLM human melanoma cell line, and expression of DN Rho results in inhibition of invasion toward the chemokine (18). As Rho constitutes a well-known downstream effector of Gα₁₃ (23), we expressed in BLM cells the Gα₁₃ Q226L (Gα₁₃QL) mutant, which leads to a CA GTPase-deficient form of Gα₁₃ in fibroblasts (24), and tested cell invasion and RhoA activation in response to CXCL12. Gα₁₃QL BLM transfectants displayed a large inhibition of Matrigel invasion toward the chemokine compared with Gα₁₃wt or mock cell invasion (Fig. 1A). Expression of endogenous, as well as GFP-fused forms of Gα₁₃, was monitored by Western blotting (Supplementary Fig. S1A, left), showing similar expression levels of Gα₁₃wt and Gα₁₃QL species, and control experiments revealed that their expression did not affect cell viability (Supplementary Fig. S1B). Notably, whereas RhoA was activated by CXCL12 in mock and Gα₁₃wt cells (7-fold to 8-fold), activation was deficient in Gα₁₃QL melanoma transfectants (Fig. 1B). Instead, all transfectants showed similar levels of chemokine-promoted activation of Rac1 (Supplementary Fig. S1C), another RhoGTPase that is involved in ruffle and lamelipodia formation and which controls cell migration. Moreover, expression in BLM cells of Gα₁₂QL similarly caused defective RhoA activation compared with mock and Gα₁₃wt transfectants (Fig. 1C). We also found that impairment in RhoA activation on Gα₁₃QL transfectants was associated with inhibition in chemokine-promoted spreading on fibronectin (Supplementary Fig. S1D).

Consistent with previous results (24, 27), Gα₁₃QL expression in mouse fibroblastic Swiss 3T3 and human MDA-MB-231 breast carcinoma cells (Supplementary Fig. S1A, right) led to a robust induction in basal RhoA activation compared with Gα₁₃wt transfectants (4-fold to 6-fold; Fig. 1D), and activation was further increased by CXCL12 in MDA-MB-231 cells. In contrast, no RhoA activation was detected on MeWo melanoma cells expressing Gα₁₃QL.

To confirm that inhibition of melanoma cell invasion and RhoA activation could be directly attributed to Gα₁₃QL, we transfected Gα₁₃ siRNA to silence Gα₁₃ expression. Gα₁₃ siRNA (3) inhibited by >80% the expression of endogenous Gα₁₃, whereas Gα₁₃ siRNA (1) and (2) did not affect its expression (Fig. 2A,, top). Importantly, Gα₁₃ (3) siRNA knocked down the expression of Gα₁₃QL in addition of the endogenous counterpart (Fig. 2A,, bottom), and it reverted Gα₁₃QL-dependent inhibition of invasion in cotransfection experiments (Fig. 2B). Moreover, Gα₁₃ silencing rescued CXCL12-triggered RhoA activation that is impaired in Gα₁₃QL transfectants, without affecting Rac activation (Fig. 2C), suggesting that RhoA-dependent invasion is interfered by Gα₁₃QL expression. To prove this hypothesis, we cotransfected Gα₁₃QL, Gα₁₃wt, or empty vector with wild-type, DN, or CA forms of RhoA and checked transfectant invasion to CXCL12. Results showed that Gα₁₃QL-dependent defective invasion was recovered when Rho CA, but not Rac CA, forms were cotransfected (Fig. 2D and Supplementary Fig. S2A and B). In addition, inhibition of melanoma cell invasion by Rho DN expression in Gα₁₃wt and mock transfectants reached levels similar to those attained by Gα₁₃QL counterparts, again pointing to the inhibitory action of Gα₁₃QL on RhoA activation. These data indicate that inefficient Gα₁₃QL transfectant invasion is a direct consequence of inhibition of RhoA activation.

The results from Fig. 2B already suggested that chemokine-promoted melanoma cell invasion might be independent of endogenous Gα₁₃ function. Both Gα_i and Gα₁₃ associated to CXCR4 in BLM cells, and exposure to CXCL12 transiently increased this association (Supplementary Fig. S3A). However, CXCL12-stimulated cell invasion was not affected by silencing the endogenous Gα₁₃ expression (Supplementary Fig. S3B). Activation of heterotrimeric G proteins leads to separation of Gα subunits from Gβγ dimers, which are both capable of downstream signaling (21, 22). Pertussis toxin (PTx) prevents this dissociation, inhibiting Gα_i-triggered and Gβγ-triggered signaling. Abolishment by PTx of BLM cell invasion induced by CXCL12 correlated with a large inhibition of RhoA activation in PTx-treated cells (Supplementary Fig. S3C). In addition, BLM transfection, either with a vector encoding human transducin Gα_t subunit, a protein that associates with Gβγdimers (32), or with a vector coding for the carboxy terminus of GPCR kinase (GRK2-CT; ref. 33), both treatments causing suppression of Gβγ-dependent responses, did not alter cell invasion toward CXCL12 (Supplementary Fig. S3D). These data show that chemokine-stimulated melanoma cell invasion involving RhoA activation predominantly depends on Gα_i-mediated responses and that endogenous Gα₁₃ or Gβγ do not play relevant roles in this process. Furthermore, they indicate that Gα₁₃QL actions functionally oppose to CXCL12-promoted, Gα_i-dependent signaling.

p190RhoGAP mediates Gα₁₃QL-dependent inhibition of RhoA activation and melanoma cell invasion. GAP proteins stimulate the slow intrinsic rate of GTP hydrolysis of Rho GTPases, leading to Rho GTPase inactivation. A candidate molecule mediating Gα₁₃QL-dependent inhibition of RhoA activation is p190RhoGAP, a protein containing a GAP domain on its COOH terminal region (34, 35), and shows preferential activity on RhoA after its phosphorylation on Y1105 (36). We found that p190RhoGAP is expressed in BLM melanoma cells and that CXCL12 stimulates its tyrosine phosphorylation (Fig. 3A,, left). Furthermore, Gα₁₃QL BLM transfectants displayed higher p190RhoGAP phosphorylation and increased binding to RhoA than Gα₁₃wt and mock counterparts, both in the absence or presence of CXCL12 (Fig. 3A,, right). Importantly, p190RhoGAP silencing restored the defective Gα₁₃QL cell invasion toward CXCL12, without affecting invasion of Gα₁₃wt and mock transfectants, and this effect was associated with recovery of RhoA activation by CXCL12 in Gα₁₃QL cells (Fig. 3B,, top and bottom, and 3C, top and bottom). Therefore, these results identify p190RhoGAP activation as a mechanism responsible for impaired RhoA activation and invasion promoted by CXCL12 in Gα₁₃QL transfectants. As Src proteins are involved in p190RhoGAP tyrosine phosphorylation (37, 38), we addressed whether they might mediate Gα₁₃QL-dependent phosphorylation of the GAP protein by using the PP2 Src inhibitor. Treatment of Gα₁₃QL transfectants with PP2 abolished p190RhoGAP tyrosine phosphorylation in comparison with cells treated with the control nonblocking PP3 reagent (Fig. 3D), suggesting that Src proteins are highly candidates to mediate Gα₁₃QL-promoted p190RhoGAP phosphorylation in melanoma cells.

The Vav proteins are GEF members whose phosphorylation is required for their GEF activity toward Rho GTPases (39). As Vav2 is needed for chemokine-stimulated RhoA activation and BLM cell invasion (20), we tested whether its activation was altered in Gα₁₃QL cells. We found that CXCL12-stimulated Vav2 phosphorylation was comparable in the three transfectants and that similar amounts of RhoA were recovered in anti-Vav2 immunoprecipitates from cell lysates after exposure to CXCL12 (Supplementary Fig. S4). These results indicate that chemokine-promoted Vav2 activation and association with RhoA is not altered in Gα₁₃QL melanoma transfectants.

LPC inhibits chemokine-stimulated melanoma cell invasion and RhoA activation through Gα₁₃-p190RhoGAP activation. We next investigated whether extracellular ligands that activate Gα₁₃ upon interaction with their receptors could mimic the biological actions of Gα₁₃QL on melanoma cells. LPC and thromboxane A2 (TXA₂) are bioactive lipids that exert their functions involving G₁₃-mediated, as well as G_q/11-mediated, signaling (40). G2A and GPR4 are postulated to function as receptors mediating LPC actions (41), whereas TPα and TPβ are receptors for TXA₂ (42, 43). PCR analyses revealed that BLM cells express GPR4, G2A, and TPα, but we did not detect TPβ (Supplementary Fig. S5). Invasion of BLM cells to CXCL12 was significantly inhibited by LPC and the TXA₂ analogue U46619 (Fig. 4A,, left). LPC-mediated impairment in invasion was rescued when Gα₁₃ expression was silenced (Fig. 4A,, right). Instead, Gα₁₃ knocking down did not recover invasion of U46619-incubated cells, indicating that LPC actions were dependent on Gα₁₃-mediated signaling, whereas U46619 blocking of cell invasion was Gα₁₃-independent. Notably, CXCL12-promoted RhoA activation was inhibited by LPC, which was rescued by Gα₁₃ silencing (Fig. 4B and C). Control experiments showed that Rac1 activation by CXCL12 was not altered in LPC-treated cells (Fig. 4C), and incubation with LPC did not affect BLM viability (Supplementary Table S1). Moreover, LPC stimulated p190RhoGAP tyrosine phosphorylation and its binding to RhoA and cooperated with CXCL12 in further augmenting this phosphorylation (Fig. 4D). LPC-dependent phosphorylation of p190RhoGAP was mediated by Gα₁₃, as it was absent in Gα₁₃-silenced cells (Fig. 4D). Together, these data indicate that blockade by LPC of chemokine-promoted melanoma cell invasion involves Gα₁₃-dependent activation of p190RhoGAP and subsequent inhibition of RhoA activation.

Expression of Gα₁₃QL in melanoma inhibits lung metastasis. We next used retroviral gene transfer to generate BLM transfectants stably expressing Gα₁₃QL to investigate its influence on in vivo metastasis. Total Gα₁₃ protein was overexpressed in Gα₁₃wt and Gα₁₃QL stable transfectants compared with mock counterparts (Supplementary Fig. S6A), which was associated with impaired Gα₁₃QL transfectant invasion toward CXCL12 and decreased two-dimensional migration (Fig. 5A and Supplementary Fig. S6B). Furthermore, these transfectants displayed defective RhoA activation in response to the chemokine (Fig. 5B), which correlated with inefficient actin stress fiber formation on Matrigel, as well as with inhibition in the generation of focal contacts, as detected with anti-vinculin antibodies (Supplementary Fig. S6C). No significant differences in the cell cycle of 5′-bromodeoxyuridine–labeled transfectants were found, with doubling times of ∼22 hours. In addition, expression levels of CXCR4 and integrin β1 were similar among these transfectants (Supplementary Fig. S6D and E).

S.c. inoculation of melanoma Gα₁₃QL, Gα₁₃wt, or mock stable transfectants into SCID mice gave rise to primary tumors with similar growth rates (Fig. 5C,, left). However, two of seven Gα₁₃wt and mock-injected mice developed several lung metastatic nodes, whereas there was no lung colonization in mice injected with Gα₁₃QL transfectants (Fig. 5C , right). Melanoma cells derived from s.c. tumors retained overexpression of Gα₁₃ and displayed invasiveness across Matrigel similarly to their original transfectants (Supplementary Fig. S7A), indicating that they conserved the invasive properties during tumor growth.

Upon 4 weeks of i.v. inoculation with Gα₁₃wt or mock transfectants, all mice progressively developed breathing difficulties, which was associated with formation of lung metastatic nodes, with no signs of additional metastases in other organs. After 7 to 8 weeks, there were no surviving Gα₁₃wt or mock mice (Fig. 5D,, left). Instead, Gα₁₃QL mice had a longer disease-free period (6–7 weeks), as well as a significantly prolonged survival, with the last mice of this group dying after 13 weeks. Appearance and abundance of metastatic nodes in lungs from these mice were similar to Gα₁₃wt or mock counterparts (Fig. 5D , right). The human origin of the tumors was confirmed by their expression of Gα₁₃ and glyceraldehyde-3-phosphate dehydrogenase mRNA detected by PCR with specific primers (Supplementary Fig. S7B). These data indicate that expression of Gα₁₃QL on melamoma cells impairs their in vivo lung metastasis.

Dynamic reorganization of the actin cytoskeleton by Rho GTPases drives cells in motion (1, 2). Rho GTPase activation is finely regulated by the functional interplay between GEFs and GAPs, and thus migratory responses closely reflect their balance of activities. Heterotrimeric G₁₂ proteins constitute upstream molecules, which control Rho activation after interaction of GPCR with their agonists (23). Here, we show that expression in melanoma cells of Gα₁₃QL, a GTPase-deficient form of Gα₁₃, led to inhibition of chemokine-promoted RhoA activation and cell invasion, which was dependent on p190RhoGAP function (Fig. 6). Defective RhoA activation in melanoma cells was also detected after expression of Gα₁₂QL, indicating that activated Gα₁₃ and Gα₁₂ trigger similar functional responses. Contrary to melanoma, we found that expression of Gα₁₃QL in Swiss 3T3 fibroblasts or in MDA-MB231 breast carcinoma cells recapitulated the reported Rho activation (24, 27), indicating that cell type–specific differences might exist in the molecular signals regulating RhoA activation. Whereas RhoA activation by CXCL12 was deficient in Gα₁₃QL melanoma cells, activation of Rac was unaltered, suggesting that stimulation of these GTPases either occurs independently of each other or that Rac activation is an upstream event of Rho activation.

Defective RhoA activation and invasion was a direct effect of Gα₁₃QL rather than from endogenous Gα₁₃ function, as both responses were blocked when Gα₁₃ siRNA was transfected together with Gα₁₃QL, but not in single Gα₁₃ siRNA transfectants. Although CXCL12 stimulated Gα₁₃ and Gα_i association with CXCR4, chemokine-promoted RhoA activation was largely inhibited by pertussis toxin, indicating a predominant Gα_i involvement. Importantly, expression of a CA Rho form rescued Gα₁₃QL-dependent inhibition on invasion, confirming that impairment in Rho activation underlay the defective Gα₁₃QL melanoma cell invasion.

Potential mechanisms mediating Gα₁₃QL-dependent defective RhoA activation in melanoma include activation of RhoGAP function or inhibition of RhoGEF activity. p190RhoGAP has GAP activity preferentially on Rho, this activity being stimulated by tyrosine phosphorylation (34, 36, 44). Stimulation by CXCL12 or Gα₁₃QL expression enhanced tyrosine phosphorylation of p190RhoGAP in melanoma cells. Furthermore, we found higher amounts of RhoA associated with p190RhoGAP in Gα₁₃QL transfectants than in mock or Gα₁₃wt counterparts, and silencing p190RhoGAP expression in Gα₁₃QL cells led to rescue of RhoA activation and invasion in response to CXCL12, therefore identifying p190RhoGAP as responsible for the inhibitory effects due to Gα₁₃QL (Fig. 6). Thus, higher GAP activity of p190RhoGAP toward RhoA in Gα₁₃QL melanoma transfectants might constitute one of the mechanisms accounting for decreased RhoA activation compared with Swiss 3T3 and MDA-MB-231 Gα₁₃QL counterparts. As mentioned in a recent review (23), although it is assumed that G₁₂-mediated signaling causes Rho activation and in vitro invasion of breast and prostate cancer cells (26, 27), this might not be always the rule. Thus, glioblastoma cells stimulated with sphingosine-1-phosphate display G₁₂-mediated inhibition of cell migration (45), suggesting that tumor type–specific differences in RhoA activation might be responsible for the distinct invasion responses.

Silencing the GEF Vav2 in BLM cells blocks CXCL12-promoted RhoA activation (20). However, Vav2 activation and association with RhoA were similar in Gα₁₃QL, Gα₁₃wt, and mock transfectants, indicating that Gα₁₃QL-mediated blockade in RhoA activation was not due to alterations in Vav2 function. It is surprising that RGS domain-containing RhoGEFs, such as p115RhoGEF, PDZ-RhoGEF, and LARG, known to interact with Gα_12/13 proteins and stimulate Rho activation (23), apparently do not importantly contribute to Rho activation of melanoma cells in response to chemokines. Should any of these RhoGEFs mediate Gα₁₃QL actions in BLM cells, we would theoretically expect increased RhoA activation and invasion as a result of cooperation with Vav proteins, which is not what we observe. Our previous work with Vav-silenced cells indicated that Vav proteins are main GEFs involved in chemokine-promoted RhoA activation and invasion of melanoma cells (20), and the present data suggest that RGS-RhoGEF role in Gα₁₃QL transfectants might be obscured by the potent p190RhoGAP activity. Another potential mechanism which could contribute to Gα₁₃QL-dependent, p190RhoGAP-mediated inhibition of Rho activation could arise from the fact that active Gα₁₃ binds to E-cadherin (46) and that p120ctn links p190RhoGAP to the E-cadherin complex in Rac-activated cells, causing Rho inactivation (47). However, BLM cells or Gα₁₃ transfectants express very low amounts of cell membrane E-cadherin (not shown), suggesting the unlikelihood of defective RhoA activation by E-cadherin–associated p190RhoGAP in Gα₁₃QL cells.

The above data raised the interesting possibility that stimuli that activate Gα₁₃ after interaction with their GPCR could limit chemokine-promoted RhoA activation and invasion of melanoma cells. We found that stimulation by CXCL12 of BLM cell invasion and Rho activation was opposed by LPC and the TXA₂ analogue U46619, which are lipids that bind to Gα₁₃-coupled receptors (40). Candidate LPC receptors GPR4 and G2A (41) were found to be expressed on BLM melanoma cells, and knocking down Gα₁₃ led to rescue of both invasion and Rho activation in response to CXCL12, indicating that LPC-dependent inhibition was mediated by Gα₁₃, potentially involving GPR4 and/or G2A. Although BLM cells also express the TPα receptor for TXA₂, Gα₁₃ silencing did not result in recovering invasion that was inhibited by U46619, suggesting that this TXA₂ analogue might be exerting its inhibition through G_q, an additional G protein mediating TXA₂ actions (48). Importantly, interference by LPC of chemokine-promoted melanoma cell invasion correlated with LPC-dependent, Gα₁₃-mediated stimulation of p190RhoGAP tyrosine phosphorylation and blockade of RhoA activation. These results show that activation in melanoma cells of the Gα₁₃-p190RhoGAP route by extracellular stimuli opposes chemokine-triggered RhoA activation and cell invasion in vitro and indicate that LPC-dependent and Gα₁₃QL-dependent inhibition of RhoA activation and invasion share a common control point that is p190RhoGAP.

Therefore, these data suggest that a tight balance of RhoA activation due to GEF and GAP actions controls melanoma cell invasion (Fig. 6). On one hand, CXCL12 stimulates Gα_i-dependent Vav2-RhoA activation, as well as tyrosine phosphorylation of p190RhoGAP (ref. 20, and this work), overall causing RhoA activation and stimulation of cell invasion, indicating that, under these conditions, GEF-promoted activation prevails over GAP activity toward Rho. When Gα₁₃ is activated, either through Gα₁₃QL expression or by LPC, p190RhoGAP-mediated RhoA inactivation overcomes CXCL12-dependent stimulation, causing defective RhoA activation and impaired invasion. Initial characterization of the molecular components involved in p190RhoGAP tyrosine phosphorylation in melanoma cells led to the identification of Src proteins as likely candidates to mediate this phosphorylation, as suggested from the results obtained with Src inhibitors. p190RhoGAP is a well-known Src substrate (37, 38), and thus, Src-dependent phosphorylation represents a potential mechanism regulating the GAP activity on Rho and, hence, the invasion of melanoma cells.

To investigate if Gα₁₃QL-dependent impairment in in vitro melanoma cell invasion could influence in vivo metastasis, we generated transfectants stably overexpressing Gα₁₃QL or Gα₁₃wt to be used in xenograft studies. Gα₁₃QL stable transfectants retained the inhibition of invasion and RhoA activation in response to CXCL12, which was associated to a decrease in transfectant migration speed.

BLM human melanoma cells were originally selected by their high potential to disseminate into lungs of immunodeficient mice (28). CXCR4 strongly contributes to BLM cell metastasis into lungs, as its silencing leads to inhibition of melanoma lung colonization.¹

Although there were no significant differences in primary tumor growth between s.c. inoculated Gα₁₃ BLM transfectants, confirming the in vitro results, near a third of mice injected with Gα₁₃wt or mock transfectants developed melanoma lung metastatic nodes, whereas no lung metastases were observed in mice inoculated with Gα₁₃QL melanoma cells. When Gα₁₃QL transfectants were i.v. inoculated, we found a significantly prolonged disease-free and survival period, compared with Gα₁₃wt or mock counterparts. These results suggest that activated Gα₁₃ might represent a molecular brake for initiation of invasive steps leading to melanoma metastasis. If melanoma cells expressing activated Gα₁₃ reach blood circulation, deficient homing into lungs occurs, suggesting that extravasation and/or tissue invasion capabilities might be affected. Based on the in vitro results, it can be proposed that p190RhoGAP-mediated inhibition of RhoA activation in Gα₁₃QL melanoma cells could impair stress fiber assembly, hindering cell motility and leading to inefficient invasion and metastasis in vivo. Although our results strongly suggest that Gα₁₃QL expression mainly affects cell motility, we cannot, however, exclude that Gα₁₃ activation might impair, to some extent, melanoma cell survival in the lungs.

Little is known on the expression pattern of G₁₂ and p190RhoGAP proteins in human tumor samples. Histopathologic analyses using human tissue specimen revealed higher Gα₁₃ expression in breast and prostate cancer (26, 27), whereas p190RhoGAP has been proposed to play tumor suppressor roles in glioma (35, 49). Further analyses are required to better define the function of these proteins in tumor cell growth and metastasis, and it will be relevant to functionally correlate their roles in in vivo models of melanoma metastasis.

From the above data, it is tempting to speculate that Gα₁₃-coupled receptors might represent a cell entrance way to inhibit RhoA activation via stimulation of p190RhoGAP. GPCRs represent important regulators of tumor cell growth and metastasis, and as recently proposed, they might provide powerful opportunities for cancer prevention and treatment (50). The present results should contribute to widen our knowledge of key signaling components activated through these receptors whose function is needed for efficient tumor cell invasion.

No potential conflicts of interest were disclosed.

Grant support: Ministerio de Educación y Ciencia grant SAF2005-02119 (J. Teixidó), Fundación de Investigación Médica Mutua Madrileña grants (J. Teixidó), and Fundación de Investigación Científica de la Asociación Española contra el Cáncer grants (R.A. Bartolomé).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Dr. Paloma Sánchez-Mateos for helpful discussions, Ma. Teresa Seisdedos for confocal microscopy, and Manuel Moreno-Calle, María Herrera-Hernández, and Nohemí Arellano-Sánchez for technical assistance in the animal facility.