Thrombin and lysophosphatidic acid receptors utilize distinct rhoGEFs in prostate cancer cells.

Thrombin and lysophosphatidic acid (LPA) receptors play important roles in vascular biology, development, and cancer. These receptors activate rho via G(12/13) family heterotrimeric G proteins, which are known to directly activate three distinct rho guanine nucleotide exchange factors (rhoGEFs) that contain a regulator of G protein signaling (RGS) domain (RGS-rhoGEFs). However, it is not known which, if any, of these RGS-rhoGEFs (LARG (leukemia-associated rhoGEF), p115rhoGEF, or PDZrhoGEF) plays a role in G protein-coupled receptor-stimulated rho signaling. Using oligonucleotide small interfering RNAs that suppress specific RGS-rhoGEF expression, we show that thrombin receptor stimulation of rho is primarily mediated by LARG in HEK293T and PC-3 prostate cancer cell lines. In contrast, the LPA-stimulated rho response in PC-3 cells is dependent on PDZrhoGEF expression. Suppression of p115rhoGEF had no effect. Thus different rhoGEFs (LARG and PDZrhoGEF) mediate downstream rho signaling by the thrombin and LPA receptors.

The rho family of small GTP-binding proteins plays important roles in many normal biological processes and in cancer (1). This family includes three main groups (rho, rac, and cdc42). rho itself is activated by numerous external stimuli including growth factor receptors, immune receptors, cell adhesion, and G protein-coupled receptors (GPCRs) 1 (2,3). The mechanism of signaling by heterotrimeric G protein-coupled receptors that activate rho has been obscure until recently (3). The discovery of a family of unique rho guanine nucleotide exchange factors (rhoGEFs), p115rhoGEF (4), PDZrhoGEF (5), and LARG (leukemia-associated rhoGEF (6)) suggested a simple mechanism. They contain a regulator of G protein signaling (RGS) domain that binds activated G␣ 12 (7) and G␣ 13 (8) causing rhoGEF activation. Thus, the RGS-rhoGEFs are throught to serve as effectors of activated G␣ 12/13 and as molecular bridges between the heterotrimeric G protein ␣ subunits and rho. A role for these proteins in cellular rho signaling by GPCRs, such as those for thrombin and LPA, has been suggested by studies with dominant negative constructs (9, 10) and inhibition of signaling by expression of the RGS domains, which act as G␣ 12/13 inhibitors (11). There has not, however, been direct proof that these proteins mediate GPCR signals and no information is available about which rhoGEF(s) are downstream of which receptors.
rho activation leads to actin rearrangements and stress fiber formation, gene transcriptional activation, neural process retraction, cell rounding, and smooth muscle contraction (reviewed in Ref. 1). Experimentally, rho activation can be detected directly by measurements of GTP-bound active rho precipitated from cell lysates with effector fusion proteins such as GST-rhotekin (12) or indirectly by any of those functional readouts. The 1321N1 astrocytoma cells is a well studied model of thrombin-induced rho activation (13). Thrombin induces both cell rounding and enhanced cell proliferation in these astrocytoma cells by mechanisms that are independent of known second messengers but are blocked by rho inhibitors.
In this study, we use HEK293T cells and a human prostate cancer cell line, PC-3, to test the role of the three RGS-rho-GEFs, LARG, p115rhoGEF, and PDZrhoGEF, in receptor signaling. HEK293 and PC-3 cells express all three of these proteins with RNA for PDZrhoGEF and LARG being more abundant than that for p115. PC-3 cells overexpress the thrombin receptor (PAR1) and have an increased propensity to metastasize to bone compared with lines that have lower PAR1 expression (14). To demonstrate a role for rhoGEFs in GPCR signaling and to define the specificity of their actions, we prepared 21-nucleotide short interfering RNAs (siRNAs) targeting each of these RGS-rhoGEFs. We show that LARG mediates thrombin responses, while the LPA response is mediated by PDZrhoGEF.

EXPERIMENTAL PROCEDURES
Materials-The three myc-tagged human RGS-rhoGEF plasmids in pcDNAmyc and the SRE.L Luciferase rho reporter construct were described previously (7). The control Renilla luciferase construct, pRL-TK, was purchased from Promega. The C3 exotoxin expression construct in pcDNA3.1 was kindly provided by Dr. John Williams (University of Michigan).
siRNA design and synthesis has been described in detail (15). Briefly, 21-nucleotide synthetic RNA with the following sequences and their complements (plus 3Ј TT overhangs on each) were synthesized by the Qiagen siRNA synthesis service (Xenogene Inc.) and high performance liquid chromatography-purified. The sequence targeted was in the RGS domain for all three RGS-rhoGEFs: p115rhoGEF (CATACCATCTC-TACCGACG), PDZrhoGEF (ACTGAAGTCTCGGCCAGCT), and LARG (GAAACTCGTCGCATCTTCC). Duplexes (with 3Ј TT overhangs) were prepared by annealing sense and antisense strands to form doublestranded RNA. The oligonucleotide pairs were resuspended at 0.3 mg/ml (total) in buffer containing 100 mM potassium acetate, 30 mM * This work was supported by National Institutes of Health Grants GM39561 (to R. R. N.), GM61454 (to T. K.), and GM31954 (to P. C. S.). 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.
Transfection of siRNAs into Cells-For Westen blot analysis, HEK293 cells were transiently transfected in a 6-well plate with 2 g/well of either active rhoGEF siRNAs (LARG, PDZ, p115) or mutant inverted siRNAs and 8 l of LipofectAMINE 2000. After 72 h, protein lysates were prepared as described (17). For luciferase assays in 96-well plates, one rhoGEF plasmid DNA (p115 (2 ng), PDZ (5 ng), or LARG (5 ng)) was introduced into HEK293 cells with 30 ng of rhoGEF siRNAs or inverted mutant siRNAs with 0.2 l of LipofectAMINE 2000 together with the dual luciferase reporters (SRE.L and pRL-TK at 3 ng and 30 ng/well, respectively).
Western Blot Analysis-Western blot analysis is done with RGS-rhoGEF-specific antibodies and ECL detection as described previously (17). The same polyvinylidene difluoride membrane was stripped and re-blotted with anti-G protein ␤ subunit antibody as loading control. Blots were quantitated as described previously (17).
Luciferase Assays-Twenty-four hours post-transfection with firefly and Renilla vectors, HEK293 cells were serum-starved (0.5% serum), and then at 48 h, luciferase activity was determined using the dual luciferase assay kit (Promega, Madison, WI) according to the manufacturer's instructions. The ratio of firefly to Renilla luciferase counts was calculated. Data are expressed as the percent of the control value (samples without siRNA).
GST-rhotekin Pull-down Assay-siRNA (si) or mutant controls (inv) were introduced (10 g/10-mm dish) into HEK293 cells with 30 l of LipofectAMINE 2000 in duplicate 100-mm dishes. Thirty-six hours after transfection, cells were switched to medium with 0.5% serum, then 18 h later stimulated with 300 nM thrombin for 5 min. Cells from each dish were lysed in 0.5 ml of lysis buffer containing 50 mM Tris, pH 7.5, 10 mM MgCl 2 , 0.5 M NaCl, 2% IGEPAL, and 5% sucrose. The active rho was precipitated with GST-rhotekin beads (Cytoskeleton Inc., Denver, CO). Western blots with anti-rhoA antibody were done to assess the amount of active rhoA. Aliquots of total lysate were also analyzed for the amount of rho present.
Cell Rounding Assay-The day before transfection, PC-3 cells were grown on laminin-coated coverslips in 12-well plates until ϳ80% confluent. Cells were then transfected with 0.2 g of EGFP cDNA. To disrupt rho signaling GFP was co-transfected with either a C3 exotoxin expression vector (0.5 g) or with 2.0 g of the indicated siRNA or inverted control along with 8 l/well of LipofectAMINE 2000. At 45 h post-transfection, cultures were changed to serum-free medium and 6 h later treated with either buffer, thrombin (100 nM), or LPA (50 M) for 30 min. Cells were then fixed with 4% paraformaldehyde for 10 min and GFP images obtained using an Olympus fluorescence microscope with a 20ϫ objective. Fluorescent cells were counted (100 -300 per coverslip) and the percentage of rounded cells determined. Cell rounding at 30 min was dose-dependent (EC 50 80 nM for thrombin and 50 M for LPA) and reversible upon removal of thrombin for 45 min (data not shown).
Data Analysis and Statsistics-Quantitative data are means Ϯ S.E. of three or more independent experiments and each luciferase experiment was conducted in triplicate. Two-way analysis of variance with post-tests (Graph Pad Prism 4.0, San Diego, CA) was used to evaluate statistical significance.

RESULTS AND DISCUSSION
All three RGS-rhoGEFs are expressed in HEK293 cells as detected by both Western blot (Fig. 1A) and reverse transcritase-PCR analysis (15). HEK293 cells also exhibit a substantial thrombin-stimulated rho activation ((18) and Fig. 2). To assess the role of RGS-rhoGEFs in receptor signaling, we designed a series of synthetic oligo-siRNAs targeted against the RGS domain of the three human rhoGEFs. Two oligonucleotides each against LARG and PDZrhoGEF and 5 against p115rhoGEF were analyzed (15), and the most active and specific three siRNAs were used in this study. They suppress the expression of their cognate rhoGEF but do not affect either G␤ used as a loading control (Fig. 1A) or the other rhoGEFs (15). Their specificity is also demonstrated in a functional effect on rhomediated gene expression. We used a modified SRE.L luciferase reporter construct (7), which responds to serum response factor-megakaryocytic acute leukemia transcription complexes but not serum response factor-ternary complex factor complexes (19) and provides a rho-dependent transcription response. Luciferase expression is strongly enhanced by transfected wild-type G␣ 13 and the constitutively active G␣ 13 QL mutant (19-and 54-fold over reporter alone, respectively; data not shown), and by transfection of all three of the RGS-rho-GEFs (Fig. 1B). In each case, firefly luciferase expression is reduced to base line by co-transfecting C3 exotoxin (data not shown) consistent with the literature showing that G 13 -and rhoGEF-stimulated gene expression is rho-dependent (9, 18).

FIG. 1. Effects of RGS-rhoGEF siRNAs on rhoGEF protein and rhoGEF-stimulated SRE activity in HEK293 cells.
A, suppression of endogenous rhoGEF proteins by siRNAs. HEK293T cells were transiently transfected with active rhoGEF siRNAs against LARG, PDZrho-GEF, or p115rhoGEF (si) or the related mutant inverted siRNAs (inv) and protein lysates prepared at 72 h. Western blot analysis with RGS-rhoGEF-specific antibodies or a G␤ subunit loading control is shown. B, specificity of siRNA effects on rhoGEF-stimulated SRE.L Luciferase activity. Each rhoGEF plasmid DNA (p115, PDZ, or LARG) was introduced into HEK293T cells using LipofectAMINE 2000 in a 96-well plate along with the dual luciferase reporters (SRE.L and pRL-TK) and a panel of different rhoGEF siRNAs (ϩ) or inverted mutant siRNAs (Ϫ). Firefly and Renilla luciferase activities were measured 48 h post-transfection as described under "Experimental Procedures" and the ratio of firefly to Renilla luciferase counts calculated. The reporters alone gave ratios of 25 Ϯ 3 and each rhoGEF stimulated firefly expression more than 10-fold. Firefly/Renilla ratios for the siRNAs (ϩ and Ϫ) are expressed as a percent of that for buffer controls with only the rhoGEF and reporter cDNAs transfected. Data shown are means Ϯ S.E. of three independent experiments each conducted in triplicate.
The rhoGEF-stimulated luciferase signal is also completely eliminated by 0.5 M latrunculin B (not shown). Co-transfection of the siRNAs targeting LARG, p115rhoGEF, or PDZrhoGEF strongly and specifically reduced the luciferase response to the targeted RGS-rhoGEF but had minimal effects on luciferase expression stimulated by the other two RGS-rhoGEFs (Fig. 1B). As an additional control for specificity, an inverted control siRNA (Ϫ) for each of the active siRNAs (ϩ) was included and had minimal effects. The incomplete inhibition of the luciferase response by the siRNAs may be due to: 1) a strongly amplified signaling cascade which is integrated over 48 h, 2) some maintained expression of RGS-rhoGEF in the presence of siRNA, or 3) incomplete overlap of transfection of the siRNA and the RGS-rhoGEF plasmid. The latter mechanisms seems unlikely given the nearly 90% transfection efficiency of these cells with a GFP reporter plasmid plus the virtually complete suppression of endogenous protein for LARG and p115rhoGEF. The data, however, indicate a substantial and specific suppression of RGS-rhoGEFs by these transfected siRNAs.
To assess the role of the RGS-rhoGEFs in receptor-mediated signaling, we used the thrombin-stimulated activation of rho as detected by precipitation of GTP-bound active rhoA with the effector domain fusion, GST-rhotekin. 2 Thrombin stimulates rho activation in HEK293 cells (Fig. 2) probably via an endogenous PAR1 receptor (20). Transfection of HEK293 cells with the active LARG siRNA (si) shows an essentially complete    Table I. block of thrombin-stimulated rho activation (Fig. 2). Table I quantitates the effects of the three rhoGEF siRNAs and their inverted controls. While this assay exhibits significant variability, only the LARG siRNA reduced the thrombin-stimulated rho activation significantly. To assess signaling by other receptors, we attempted similar measurements with LPA. While there was some stimulation of rhoA activation by LPA in HEK293 cells, it was substantially smaller than that induced by thrombin and not sufficient to permit analysis with the siRNAs. Thus we turned to a different cell line and a different rho response.
rho stimulates several types of cytoskeletal rearrangements. Stress fiber formation in NIH or Swiss 3T3 cells is a classic rho response (21). In other cell types such as the 1321N1 astrocytoma line, cell rounding is observed as a thrombin-stimulated G 12/13 -mediated rho response (10). We found that a similar cell rounding response occurs in the human prostate cancer cell line, PC-3. After thrombin stimulation, there was some appearance of stress fibers following phalloidin staining (data not shown), but the more prominent effect was cell rounding (Fig.  3A). Since the transfection efficiency of the PC-3 cells (30 -50%) was lower than that in HEK293, we used a co-transfected GFP reporter plasmid to permit us to selectively assess the transfected cell population. The cell rounding response to thrombin was reduced by 80 -90% by co-transfecting a C3 exotoxin expression plasmid with the GFP reporter (Fig. 3A). The basal cell shape, however, was not affected by C3 toxin. In addition to thrombin responses, LPA also stimulated PC-3 cell rounding in a rho-dependent manner (Fig. 3A).
The thrombin and LPA responses in PC-3 cells were then tested with the RGS-rhoGEF siRNAs which on Western blots show suppression of the appropriate rhoGEF (data not shown). As expected from the HEK293 data, the LARG siRNA nearly completely abolished the thrombin-stimulated PC-3 cell rounding response (Fig. 3B). This is mediated by the PAR1 type of thrombin receptor, since cell rounding stimulated by the PAR1 agonist peptide SFLLRN was also inhibited by LARG siRNA (Fig. 3C). The negative control inverted LARG siRNA (Ϫ) did not affect thrombin-stimulated cell rounding. Also, the p115rhoGEF and PDZrhoGEF siRNAs did not inhibit the thrombin response indicating that LARG represents the primary downstream pathway from PAR1 to rho activation in PC-3 cells. Surprisingly, the LARG siRNA did not inhibit cell rounding induced by LPA but the PDZrhoGEF siRNA did (Fig.  3B). This shows that the LPA receptor uses a different RGS-rhoGEF to induce rho responses. It is known that PC-3 cells express LPA 1 and LPA 2 but not LPA 3 receptors (22), but we cannot presently say which LPA receptor is mediating this effect. The LPA data also provide additional controls; the LARG siRNA effect on thrombin responses is specific for thrombin and not other rho-activating stimuli, and the lack of effect of PDZrhoGEF siRNA on thrombin responses is not due to incomplete knockdown of the PDZrhoGEF protein as the siRNA was able to disrupt the LPA response.
The mechanism of specificity of PAR1 for LARG and LPA receptor for PDZrhoGEF is unclear. Recently, it has been suggested that thrombin and LPA receptors activate different members of the G 12 family, G␣ 12 and G␣ 13 , respectively (23), and G␣ 12 is known to activate LARG (7). In addition to G protein selectivity, it is possible that there could be direct interactions between the RGS-rhoGEFs and the receptors. Both LARG and PDZrhoGEF have PDZ domains, which bind to plexin B1 and mediate semaphorin signaling (24 -26). The carboxyl termini of LPA1 and LPA2 receptors (EDG2 and EDG4) have classical PDZ interaction motifs (S/T-X-L/V), so direct rhoGEF-receptor interactions may contribute to the specificity.
In summary, we provide the first direct demonstration that RGS-rhoGEFs mediate GPCR signaling to rho as well as showing receptor specificity in the use of RGS-rhoGEFs with PAR1 using LARG in both HEK293 and PC-3 cells and the LPA receptor using PDZrhoGEF in PC-3 prostate cancer cells. Since thrombin stimulates proliferation (27) and vascular endothelial growth factor secretion (28) of prostate cancer cells and prostate cancer metastatic to bone has increased levels of PAR1 (29), the ability to block receptor-stimulated rho signaling could represent an important approach to regulating cancer cell growth and metastasis.