Gastrin-stimulated Gα13 Activation of Rgnef Protein (ArhGEF28) in DLD-1 Colon Carcinoma Cells*

Background: Rgnef (ArhGEF28) is activated downstream of gastrin and the cholecystokinin receptor to promote colon carcinoma tumor progression. Results: Rgnef activation by Gα13 triggers FAK and paxillin tyrosine phosphorylation in response to gastrin. A C-terminal Rgnef region is necessary for linkage to Gα13. Conclusion: Rgnef is an effector of Gα13 signaling. Significance: Gα13 and Rgnef are implicated in colon carcinoma. The guanine nucleotide exchange factor Rgnef (also known as ArhGEF28 or p190RhoGEF) promotes colon carcinoma cell motility and tumor progression via interaction with focal adhesion kinase (FAK). Mechanisms of Rgnef activation downstream of integrin or G protein-coupled receptors remain undefined. In the absence of a recognized G protein signaling homology domain in Rgnef, no proximal linkage to G proteins was known. Utilizing multiple methods, we have identified Rgnef as a new effector for Gα13 downstream of gastrin and the type 2 cholecystokinin receptor. In DLD-1 colon carcinoma cells depleted of Gα13, gastrin-induced FAK Tyr(P)-397 and paxillin Tyr(P)-31 phosphorylation were reduced. RhoA GTP binding and promoter activity were increased by Rgnef in combination with active Gα13. Rgnef co-immunoprecipitated with activated Gα13Q226L but not Gα12Q229L. The Rgnef C-terminal (CT, 1279–1582) region was sufficient for co-immunoprecipitation, and Rgnef-CT exogenous expression prevented Gα13-stimulated SRE activity. A domain at the C terminus of the protein close to the FAK binding domain is necessary to bind to Gα13. Point mutations of Rgnef-CT residues disrupt association with active Gα13 but not Gαq. These results show that Rgnef functions as an effector of Gα13 signaling and that this linkage may mediate FAK activation in DLD-1 colon carcinoma cells.

Intestinal homeostasis is a dynamic process in which stem cells at the base of crypts divide and migrate to the apical region, where they differentiate. During tumor progression, cells can undergo an epithelial to mesenchymal transition associated with increased cell motility. Changes in cytoskeletal dynamics required for cell migration are coordinated in large part by RhoGTPases (1) whose involvement in cancer is well documented (2). Although activation of RhoGTPases can occur by inhibiting GTPase-activating proteins or guanosine nucleotide dissociation inhibitors, a number of studies suggest that the Dbl family of exchange factors known as GEFs 5 are the primary mediators of RhoGTPase activation (3).
Rgnef, also termed p190RhoGEF or ArhGEF28, is a member of the Dbl family of RhoGEFs (4). Elevated Rgnef expression promotes colorectal carcinoma cell motility, invasion, and tumor progression (5)(6)(7). Rgnef also possesses anti-apoptotic activity (8) and is implicated in the pathogenesis of motor neuron degeneration (9), in the regulation of synapse formation (10), and in signaling associated with dendritic morphogenesis (11,12). Rgnef knock-out yields homozygous-null mice at less than expected Mendelian ratios, and Rgnef-null fibroblasts exhibit decreased RhoA GTPase activity and cell motility downstream of integrins (6). A link with FAK activity has been described in both normal and tumor cell motility. Rgnef is activated by GPCR signaling cascades (7), but the mechanism of activation downstream of GPCRs remains unclear. In particular, Rgnef responds to gastrin-type 2 cholecystokinin receptor (CCK2R) stimulating signaling events (7).
As a member of the Dbl family (13), p190RhoGEF/Rgnef is characterized by a central tandem DH-PH domain and also contains an N-terminal leucine-rich region and a large C-ter-minal region that may form an ␣-helical coiled-coil domain of unknown function. The DH-PH catalytic region is responsible for RhoGEF activity (14,15). Rgnef has the highest amino acid similarity to p114RhoGEF (ArhGEF18), GEF-H1 (ArhGEF2), and Lbc (ArhGEF13). Other related RhoGEFs with similar subunits are p115RhoGEF (ArhGEF1), PDZ-RhoGEF (ArhGEF11), and LARG (ArhGEF12) (16,17) that contain a regulator of G protein signaling homology (RH) domain located within the N-terminal region that confers the capability to bind to G proteins. All of these RhoGEFs preferentially activate RhoA. Lbc has also been characterized as a downstream effector of G␣ 12 (18), and p114RhoGEF can bind to G␤␥ subunits (19). Rgnef lacks a recognized RH domain in its N-terminal region and thus has not been considered a downstream effector of G␣ 12/13 .
The heterotrimeric G proteins, G␣ 12 and G␣ 13 , activate RhoA via RH-containing RhoGEFs resulting in actin stress fiber formation and focal adhesion assembly (20,21). Soon after their discovery, both G␣ 12 and G␣ 13 were demonstrated to induce oncogenic transformation (22)(23)(24). Several studies have demonstrated roles for G␣ 12 or G␣ 13 in cell proliferation and migration (25). Additionally, G␣ 12 may alter cell adhesion in a RhoGEF-independent manner (26 -28). Invasive breast cancer and prostate adenocarcinoma cells express high levels of G␣ 12 (25,29,30). These findings support the hypothesis that GPCRs may signal through G 12 proteins to promote tumor progression (31).
Here, we show that gastrin stimulation of DLD-1 colon carcinoma cells acts through CCK2R to promote G␣ 13 -stimulated FAK Tyr-97 and paxillin Tyr-31 tyrosine phosphorylation, RhoA GTP binding, and the activation of SRE reporter transcription. We find that G␣ 13 activates Rgnef utilizing a previously uncharacterized domain within the Rgnef C-terminal region in close proximity to the Rgnef-FAK-binding site. These results establish Rgnef as a new effector of G␣ 13 in the control of cell migration downstream of gastrin.
Cells-HEK293, HEK293T, and DLD-1 cells were grown in DMEM (Invitrogen) supplemented with 10% fetal bovine serum (FBS, Invitrogen), 2 mM L-glutamine (Invitrogen) and 100 units of penicillin/streptomycin (Lonza) in a humidified atmosphere of 5% CO 2 in air at 37°C. Prior to cell treatments, cells were serum-starved for 16 h. Cells were stimulated with gastrin at 200 nM in serum-free medium for the indicated time periods. For transient transfections, cells were plated at 60 -70% confluence, and after 24 h, cells were transfected utilizing FuGENE 6 (Roche Applied Science) or Metafectene Pro (Biontex) according to the manufacturer's instructions. Protein expression was evaluated 24 -48 h after transfection.
shRNA Knockdown-G␣ 13 short hairpin-interfering RNA (shRNA) targeted against human GNA13 (TRCN0000036885, Mission shRNA, Sigma) was used for stable knockdown in DLD-1 cells. For lentivirus production, HEK293T were cotransfected with pLKO.1 vectors containing the shRNA vector together with pCMV⌬R8.91 (HIV Gag and Pol) and pVSVG at a ratio of 4:3:1, respectively. Medium was collected every 24 h for 2 days and sterile-filtered (0.45-m MCE filter, Millipore). Viral particles were stored at Ϫ80°C. To infect cells by spinoculation, 1 ml of the viral particle supernatant was added to cells, and plates were centrifuged at 1200 ϫ g for 2 h at room temperature. After 24 h, infected cells were selected by adding 2 g/ml of puromycin (Sigma). The efficiency of shRNA knockdown was confirmed by SDS-PAGE and Western blot analysis of protein extracts.
Immunoprecipitation and Western Blots Analysis-Cells were lysed in RIPA buffer (50 mM Tris, pH 7.4, 0.3 M NaCl, 0.1% SDS, 0.5% DOC, 10 mM MgCl 2 , 1 mM Na 3 VO 4 , 10 mM NaF, 1% n-dodecyl ␤-D-maltoside, and protease inhibitors). To undergo AlF 4 Ϫ treatment, 30 M AlCl 3 and 5 mM NaF were added to lysis and wash buffers. Lysates were pre-cleared with 50 l of IgG-free BSA (10 mg/ml) (Sigma) and 50 l of protein A-Sepharose (Roche Applied Science) for 1.5 h at 4°C on a rotating wheel followed by centrifugation. Specific antibodies were added to lysate supernatants and incubated overnight at 4°C with rotation. Protein A-Sepharose (Sigma) or protein G-agarose (Roche Applied Science) beads were added and incubated for 1.5 h at 4°C with rotation. Samples were centrifuged for 5 min at 425 ϫ g, and the beads were washed three times with gentle shaking in RIPA wash buffer (50 mM Tris, pH 7.4, 0.3 M NaCl, 0.1% SDS, 0.5% DOC, 10 mM MgCl 2 , 1 mM Na 3 VO 4 ), followed by aspiration, resuspension in SDS loading buffer, and heating at 100°C. Cell lysates and immunoprecipitates were loaded in SDS-PAGE, immunoblotted to Immobilon-FL membranes (Millipore), and analyzed with the indicated antibodies. Western blots were visualized by infrared detection (Odyssey System) and quantified by Image Studio software (LI-COR) (35).
RhoA Activation Assays-Cells were transfected with Rgnef and/or G␣ 13 plasmids utilizing Metafectene TM Pro (Biontex) and serum-starved overnight 24 h post-transfection. After cell stimulation with gastrin, cells were lysed in RBD2 lysis buffer (50 mM Tris-HCl, pH 7.2, 1% Triton X-100, 0.5% NaDOC, 0.1% SDS, 500 mM NaCl, 10 mM MgCl 2 , and protease inhibitors) and clarified by centrifugation (13,000 ϫ g for 5 min at 4°C). Modified Bradford assays were used to determine protein concentration (Bio-Rad). Aliquots were mixed with SDS sample buffer and stored at Ϫ20°C as total protein lysates. Lysates of equal protein content were incubated with 30 -40 g of GST-Rhotekin RBD immobilized to glutathione-Sepharose 4B beads for 45-60 min at 4°C by rotation. Beads were washed four times with 4°C RBD wash buffer and activated RhoA eluted with SDS sample buffer addition and detected by monoclonal anti-RhoA immunoblotting.
Affinity Precipitation of Activated GEFs-HEK293 cells expressing Rgnef and G␣ proteins were lysed in ice-cold RhoA G17A lysis buffer. Lysates were clarified by centrifugation (16,000 ϫ g for 1 min at 4°C) and pre-cleared by incubation with 50 l of GST bound to glutathione-Sepharose (1 mg/ml) for 10 min at 4°C. Modified Bradford assays were used to determine protein concentration (Bio-Rad). GST-RhoA G17A beads (10 g) were added to each lysate, rotated for 45-60 min at 4°C, and washed three times with GST-RhoA G17A lysis buffer. Bead-associated protein complexes were separated by SDS-PAGE followed by anti-HA tag or Rgnef immunoblotting.
SRE Luciferase Assays-The SRE reporter was designed to monitor the activity of serum-response factor (SRF)-mediated signal transduction. pSRE.L luciferase reporter plasmid encodes for firefly luciferase positioned downstream of a mutant SRE that contains SRF-binding sites but eliminates the ternary complex factor-binding site (37). RhoA activation stimulates SRE transcriptional activation (38). HEK293 cells were transfected with the indicated plasmids with pSRE.L (0.1 g) and the Renilla luciferase thymidine kinase (pRL-TK) (0.01 g). Modeling the Rgnef-G␣ 13 Complex-RH domain sequences for human p115RhoGEF (ArhGEF1), murine Lsc (ArhGEF1), PDZ-RhoGEF (ArhGEF11), and LARG (ArhGEF12) were retrieved from the Uniprot (UniProt-Consortium, 2010). Entries tagged as reviewed were saved (Q9NZN5, Q8R4H2, Q92888, O15085, Q9ES67, Q61210, and Q9Z1I6) and aligned using ClustalW (39). Next, we aligned the C-terminal portion of Rgnef (entries P97433, Q8N1W1, and P0C6P5) to the RH domains of the p115RhoGEF family. Relying on this alignment, we constructed a computational model of the Rgnef RH-like domain bound to G␣ 13 using Modeler (40) based on the homology to the RH-like domain of ArhGEF11, for which the structure in complex with G␣ 13 (Protein Data Bank code 3CX8) is available (41,42). The molecular complex was refined using energy minimization with AMBER 12.0 employing the Amberff99SB-ILDN force field (43).
Statistical Analysis-Data are presented as mean values or as fold induction Ϯ S.E. or Ϯ S.D. as indicated in each figure legend. The sums of all the individual stimulatory effects were compared with the combined effects by using the unpaired t test with two-tailed p values. Differences were considered statistically significant when p Ͻ 0.05 (noted with *), p Ͻ 0.005 (noted with **), and p Ͻ 0.001 (noted with ***).

Results
Gastrin Enhances Paxillin Tyrosine Phosphorylation through CCK2R-DLD-1 colon carcinoma cells will form focal adhesions with enhanced paxillin tyrosine phosphorylation associated with increased cell motility in response to gastrin stimulation (7). To evaluate the involvement of the G protein-coupled CCK2R in gastrin-induced motility-associated signaling, paxillin Tyr-31 phosphorylation was determined by phosphospecific immunoblotting (Fig. 1). Gastrin increased paxillin Tyr(P)-31 phosphorylation within 15-30 min that was sustained to 60 min (Fig. 1A). The CCK2R antagonist YM022 inhibited gastrininduced paxillin Tyr-31 phosphorylation (Fig. 1, B and C), suggesting that this linkage was mediated by CCK2R. Conversely, addition of the CCK1R antagonist SR27897 increased paxillin Tyr(P)-31 phosphorylation by 54% with gastrin co-stimulation. Accordingly, addition of the CCK1R agonist SR146131 did not enhance paxillin Tyr(P)-31 phosphorylation (Fig. 1, D and E), suggesting that gastrin induction of paxillin phosphorylation is dependent on CCK2R (but not CCK1R) in DLD-1 cells.
The signaling pathway downstream gastrin, CCK2R, and RhoA have not been established. Increased transcriptional activity of an SRE-luciferase reporter containing SRF-binding sites is a defined signaling linkage downstream of RhoA GTPase activation (44). In DLD-1 cells, gastrin-stimulated SRE activity was reduced by the YM022 CCK2R antagonist but not by the SR27897 CCK1R antagonist (Fig. 1F). Compared with gastrin, addition of the SR146131 CCK1R agonist only weakly enhanced SRE promoter activity (Fig. 1G). Together, these results support the notion that gastrin-treated DLD-1 cells signal through CCK2R to stimulate paxillin tyrosine phosphorylation and likely RhoA activation leading to an enhanced SRE transcriptional response.
Gastrin-stimulated Rho Activation, FAK, and Paxillin Phosphorylation Involved G␣ 13 -To test whether G␣ 13 could have any role in paxillin-Tyr(P)-31 phosphorylation and in Rho activation of SRE transcription activity following CCK2R stimulation, we generated G␣ 13 shRNA targeted against the coding region of G␣ 13 for stable knockdown in DLD-1 cells. Expression of G␣ 13 -shRNA resulted in Ͼ75% stable reduction in G␣ 13 protein expression compared with Scr shRNA-expressing and Mock-transduced cells ( Fig. 2A). G␣ 13 -shRNA did not affect other G␣ subunit expressions ( Fig. 2A). DLD-1-depleted G␣ 13 cells were analyzed for effects on gastrin-stimulated paxillin Tyr-31 phosphorylation. Compared with Mock-transduced and Scr shRNA-expressing DLD-1 cells, G␣ 13 -shRNA paxillin Tyr-31 phosphorylation was selectively reduced upon G␣ 13 knockdown (Fig. 2, B and C). Together, these results support a gastrin, CCK2R, and G␣ 13 signaling linkage upstream of paxillin tyrosine phosphorylation in DLD-1 cells.
Phosphorylation of FAK at Tyr-397 also occurs in response to gastrin stimulation of DLD-1 cells, and pharmacological FAK inhibition can prevent gastrin-stimulated paxillin tyrosine phosphorylation and cell motility (7). G␣ 13 -shRNA expression reduced gastrin-stimulated FAK Tyr-397 phosphorylation without effects on total FAK expression (Fig. 2D). Gastrin-stimulated SRE-luciferase activity was also reduced in the presence of G␣ 13 shRNA compared with mock DLD-1-infected cells (Fig. 2E). Together, these results support a gastrin, CCK2R, G␣ 13 , FAK, and paxillin signaling linkage that may impact gastrin-stimulated SRE transcriptional activity.
G␣ 13 Activates Rgnef and RhoA in a Linkage to Enhanced SRE Transcription-Rgnef functions to promote paxillin tyrosine phosphorylation downstream of integrins in fibroblasts and gastrin-CCK2R stimulation in DLD-1 colon carcinoma cells (5-7). Because we found that G␣ 13 is important for gastrinstimulated paxillin tyrosine phosphorylation and enhanced SRE activity, we investigated whether G␣ 13 could activate Rgnef via affinity binding to a purified nucleotide-free point mutant (G17A) of RhoA (Fig. 3). In cells expressing CCK2R, gastrin stimulation increased Rgnef binding to RhoA G17A (Fig.  3A). In cells expressing G␣ 13 or G␣ q , Rgnef possessed a higher basal level of activity as determined by RhoA G17A binding. AlF 4 Ϫ addition to activate the G␣ proteins resulted in strong Rgnef binding to RhoA G17A (Fig. 3A). Importantly, expression of constitutively active G␣ 13 Q226L and G␣ q R183C point mutants promoted Rgnef binding to RhoA G17A (Fig. 3B).
To analyze whether G␣ 13 could activate Rho downstream of Rgnef, we measured the activity of RhoA in an affinity binding pulldown assay (Fig. 3C). A combined expression of G␣13 plus Rgnef promoted RhoA activation (Fig. 3C). Importantly, Rgnef expression in combination with G␣12 expression did not result in significant RhoA activation compared with the combination of G␣ 13 and Rgnef (Fig. 3D). We evaluated the effect of G␤1␥1 expression (Fig. 3D) because p114RhoGEF, another member of the Rgnef subfamily, was shown to activate RhoA via G␤␥ (19). However, for Rgnef we could not detect additional Rho activity in the presence of G␤1␥1. The decreased activity observed, which was not always consistent, could be an indirect effect of G␤1␥1 expression. These results support the notion that there may be a selective signaling linkage between G␣ 13 , G␣ q , and Rgnef.
We next performed SRE-luciferase assay in the presence of Rgnef and active forms of G␣ 13 or G␣ q proteins as alternative measure of RhoA signaling (Fig. 3E). Combined Rgnef expres-sion with active G␣ 13 or G␣ q induced a synergistic increase in SRE-luciferase activity. As internal control, we analyzed the expression of active G␣ 13 with its well known downstream Rho-GEF effectors. Expression of LARG or Rgnef resulted in increased SRE-luciferase activity compared with expression of active G␣ 13 alone (Fig. 3F). Taken together, these results suggest that G␣ 13 and G␣ q , but not G␣ 12 or G␤1␥1, stimulate Rgnef activity leading to RhoA activation.
G␣ 13 Immunoprecipitated with the N-and C-terminal Region of Rgnef-To determine whether Rgnef might act as a downstream effector for G␣ 13 , co-immunoprecipitation (IP) analyses were performed with endogenous and exogenous proteins. In DLD-1 cells endogenous G␣ 13 protein was associated with a 190-kDa anti-Rgnef immunoreactive protein (Fig. 4A). A similar band was observed with exogenous HA-Rgnef that immunoprecipitated with endogenous G␣ 13 in DLD-1 cells. Additionally, AlF 4 Ϫ stimulation of cells enhanced the amount of G␣ 13 co-immunoprecipitating with antibodies to Rgnef (Fig.  4B). This result suggests that G␣ 13 needs to be activated downstream of GPCRs to immunoprecipitated with Rgnef. Specificity of G␣ and Rgnef immunoprecipitations was tested by cotransfection and co-IP studies. HA-Rgnef associated with active forms of G␣ 13 and G␣ q but not with G␣ Z (Fig. 4, C and D). No consistent immunoprecipitation was observed with the active forms of G␣ 12 (Fig. 4C), in agreement with the activity data (Fig.  4C). Taken together, the results suggest that active G␣ 13 and G␣ q form a complex with Rgnef and that G␣ 13 and G␣ q are upstream activators of Rgnef. 13 -To determine the Rgnef region that mediates G␣ 13 immunoprecipitation, a series of HA-or mCherry-tagged N-or C-terminal truncated Rgnef constructs was created and verified to express proteins of different but expected sizes (Fig.   FIGURE 3. Gastrin, G␣ 13 , and G␣ q promote Rgnef and RhoA activation. A and B, affinity binding to a nucleotide-free mutant of RhoA (G17A) was used to evaluate Rgnef activation. HEK293 cells were transfected with expression vectors for HA-Rgnef, CCK2R, G␣ 13 wt, or G␣ q wt (A) or HA-Rgnef and G␣ 13 Q226L or G␣ q R183C (B) as indicated. Cells were either left unstimulated or stimulated with 100 nM gastrin or 10 M AlF 4

Rgnef C-terminal Domain Immunoprecipitated with Activated G␣
Ϫ for 40 min. Active Rgnef bound to GST-RhoA G17A beads was visualized by anti-HA immunoblotting (top blots), and protein expression in total cell lysates was evaluated by anti-HA, -G␣ 13 , and -G␣ q immunoblotting. Experiments were repeated three times with similar results. C and D, active RhoA was measured by the GST-Rhotekin RBD pulldown assay. HEK293 cells were transfected with expression vectors for HA-Rgnef, G␣ 13 wt, or G␣ q wt (C) or mCherry-Rgnef, FLAG-G␤1, G␥1, or G␣ 13 wt (D) as indicated. Active RhoA in the pulldown was visualized by anti-RhoA immunoblotting (top blots), and protein expression in total cell lysates was evaluated by anti-RhoA, -mCherry, -FLAG, and -G␣ 13 immunoblotting. Experiments were repeated three times with similar results. E and F, SRE reporter transcriptional activity was measured in HEK293 cells transfected with expression vectors for HA-Rgnef, G␣ 13 Q226L, and G␣ q R183C (E) or expression vectors for G␣ 13 QL, PDZ-RhoGEF, LARG, or Rgnef (F) as indicated. Cells were serum-starved overnight, and luciferase activity was measured. Control empty vector-transfected cell values were set to 1. Data are means Ϯ S.E. fold induction (relative units, R.U.) from three independent experiments (*, p Ͻ 0.05; two-tailed t test). Anti-HA (Rgnef), -G␣ 13 , and -G␣ q immunoblotting was used to verify transfected construct expression. 5A). Anti-HA co-IP analyses show that G␣ 13 immunoprecipitated with the full-length and the C-terminal (1185-1693) region of Rgnef (Fig. 5B). A weaker co-immunoprecipitation was also observed with the N-terminal region (1-1184) of Rgnef. Because the Rgnef C-terminal region interacts with G␣ 13 , we tested whether Rgnef-CT overexpression could block G␣ 13 activity. Co-expression of HA-Rgnef-Ct(1185-1693), HA-Rgnef-Ct*(1279 -1582), or HA-Rgnef-Ct⌬FAK(1302-1582) in HEK293 cells produced a significant reduction of G␣ 13 Q226Linduced SRE-luciferase activity (Fig. 5C). These results support the hypothesis that Rgnef may be proximally activated by G␣ 13 Q226L and that overexpression of the putative Rgnef-Ct interacting region (1279 -1582) blocks G␣ 13 Q226L-induced SRE activity independently of FAK binding to Rgnef (Fig. 5C).
To determine whether the Rgnef region that blocks G␣ 13 Q226L-induced SRE activity can also form a complex with G␣ 13 Q226L, HEK293 co-transfection and co-IP experiments were performed (Fig. 5D). mCherry-Rgnef-Ct* and mCherry-Rgnef⌬FAK associated with antibodies to G␣ 13 . Additionally, mCherry-Rgnef⌬FAK associated in a complex with endogenous G␣ 13 . Taken together, these results show that the Rgnefinteractive (and dominant-negative acting) domain for G␣ 13 -induced signaling occurs independently of FAK binding to Rgnef.
Domain in the Coiled Coil of p190RhoGEF C Terminus-To determine potential residues responsible for Rgnef-Ct association with G␣ 13 , we performed sequence alignments with other RhoGEFs known to interact with G␣ proteins. In particular, we identified a pattern of sequence similarity with the RH domaincontaining subfamily of RhoGEFs: p115, Lsc, PDZ, and LARG (Fig. 6). Alignment is based upon the crystal structure of the RH domain of p115GEF (41,42). Despite low 32% sequence identities shared between the RH domains of p115RhoGEF and PDZ-RhoGEF, these domains share nearly identical three-dimensional structures, a root mean square deviation of 0.608 Å on C␣ values (41,42).
By performing computer modeling using the coordinates of the p115 RhoGEF and PDZ RhoGEF RH domains, we were able to generate a Rgnef-Ct three-dimensional model (Fig. 7A, yellow) that contains similar secondary structure elements with p115 RhoGEF (Fig. 7A, blue) and PDZ RhoGEF (Fig. 7A,  salmon). In particular, the organization of Rgnef residues in the arrangement of helices ␣1, ␣3, ␣8, and ␣9 suggests that a similar binding interface may form with G␣ 13 (Fig. 7B). Computer Ϫ . IP analyses with anti-HA antibodies were used to detect complexes with G␣ 13 with anti-HA and -G␣ 13 immunoblotting. Protein expression levels were evaluated immunoblotting of total lysates. C and D, HEK293 cells were transfected with expression vectors encoding HA-Rgnef, G␣ 12 Q229L, G␣ 13 Q226L, or G␣ q R183C (C) or HA-Rgnef, G␣ 13 Q226L, or G␣z-EE (D) as indicated. C, IP analyses with anti-HA antibodies were used to detect complexes with G␣ 12 , G␣ 13 , and G␣ q by anti-HA and -G␣ 12 , -G␣ 13 , and -G␣ q immunoblotting. D, IP analyses with anti-HA antibodies were used to detect complexes with G␣ 13 and G␣ z by anti-HA, -G␣ 13 , and -EE tag immunoblotting. Protein expression levels were evaluated by immunoblotting of total cell lysates. A-D, experiments were repeated at least three times with similar results. JUNE 12, 2015 • VOLUME 290 • NUMBER 24 modeling reveals potential sites of G␣ 13 interaction mediated by both hydrophobic or ionic amino acid residues within the putative Rgnef-Ct ␣3, ␣8, and ␣9 modeled structure (Fig. 7B).
Because both mutationally activated G␣ 13 and G␣ q could activate Rgnef (Fig. 3), we determined whether Rgnef-5M mutations disrupted binding to both G␣ 13 and G␣ q by HEK co-expression and co-immunoprecipitation analyses (Fig. 7E). Surprisingly, G␣ q immunoprecipitated equally well with Rgnef-WT and Rgnef-5M, whereas G␣ 13 showed selectively decreased association with Rgnef-5M. To determine whether alterations in global protein folding may underlie binding differences between Rgnef-WT and Rgenf-5M, RhoGEF (RhoA G17A) and SRE activity assays were performed (Fig. 7F). Notably, Rgnef-5M exhibited equivalent binding to RhoA G17A and significantly enhanced SRE activity in an equivalent manner compared with Rgnef-WT.
Taken together, these results support the notion that Rgnef residues Glu-1484, His-1396, Glu-1478, Leu-1486, and Arg-1490 (mutated in Rgnef-5M) facilitate a binding and signaling linkage with G␣ 13 . Of course, Rgnef and G␣ 13 association may also be indirect, and these residues within the Rgnef-Ct coiledcoil domain are important in mediating this linkage. Nevertheless, these results strongly support a role for Rgnef as a downstream effector for G␣ 13 , and we have identified a new domain within Rgnef that is important for G␣ 13 signaling.

Discussion
Although the trophic properties of gastrin are well established (45,46), the intracellular and molecular mechanisms by which gastrin modulates cell growth in the gastrointestinal tract have yet to be fully elucidated. It has been reported that stimulation of CCK2R by gastrin activates several signal transduction pathways involved in cell proliferation and migration, such as the mitogen-activated protein kinase (MAPK) that includes ERK, JNK, p38 kinase, and RhoA activation (47). Here, we show that G␣ 13 activates Rgnef, which in turn activates RhoA in human colon cancer DLD-1 cells. Knockdown of G␣ 13 prevented CCK2R-induced gastrin activation of RhoA, FAK-Tyr(P)-397, and paxillin-Tyr(P)-31 tyrosine phosphorylation. Taken together, the results suggest that the signaling pathway between gastrin-induced CCK2R and RhoA downstream activity involves the activation of Rgnef by G␣ 13 .
Lacking a recognized RH-like domain, Rgnef was not considered a putative effector of G␣ 13/12 . Rgnef family members AKAP-lbc and p114RhoGEF have been previously shown to be activated by G proteins (18,19). Our studies place Rgnef as the downstream activator of both G␣ 13 and G␣ q . However, our data are only suggestive of Rgnef as a direct effector, and biochemical experiments with purified proteins have been problematic to date. Our results show that both the N terminus with the DH-PH domain and the C-terminal domains of Rgnef are each sufficient to immunoprecipitate with G␣ 13 . A multiple sequence alignment led us to identify a region at the C terminus of Rgnef with weak homology to the RH domain. RH domains of RH-RhoGEFs substantially differ among the family (48) indi-cating that structural adaptations in G␣ 13 could allow interaction with several RH domains, despite their considerable divergence. In this study, we showed that this domain is necessary for G␣ 13 effects. The fact that the mutation of residues located at this region reduce its binding to G␣ 13 validates the proposed model for the Rgnef/G␣ 13 . A similar region located at the C-terminal region of AKAP-lbc was suggested to be involved in G␣ 12 interaction (18). However, similar coiled-coil structures in other RhoGEFs proteins have been associated with hetero-and homodimerization (49,50). Heterodimerization of Rgnef with known RH-RhoGEFs could also explain the effects obtained in this work. Nevertheless, Rgnef is the first RH-GEF demonstrated to be downstream of CCK2R. We also found that G␣ 13 immunoprecipitated with the N terminus of Rgnef containing the DH and PH domain, which should not dimerize. Previous reports on LARG showed simultaneously interaction with the RH domain (at the N-terminal domain of LARG) and DH-PH domain (51). Similarly, it has been shown that activated G␣ 13 can interact with the RH domain and the DH-PH domains of p115RhoGEF (45,46,52,53). Further investigations will examine the functional role of the interaction through the region containing the DH-PH domains, if the association of Rgnef with G␣ 13 is direct, and determine whether Rgnef mediates GTPaseactivating protein activity toward G␣ 13 .
Our results show that in addition to G␣ 13 , G␣ q can also act as an upstream activator of Rgnef. Interestingly, previous reports show that both G␣ 13 and G␣ q linked CCK stimulation to the activation of Rac in pancreatic acini (54). For many years, the role of G␣ q in RhoA stimulation was under discussion, but investigations supported that G␣ q can induce Rho-dependent responses via cooperation with several RH-RhoGEFs, including Lbc-RhoGEF and LARG (20,37,55). Later on, it was shown that p63RhoGEF interacts and is activated by G␣ q (56,57). Resolution of the crystal structure of the G␣ i/q -p63RhoGEF complex revealed that G␣ q interacts with the DH and PH domains of p63RhoGEF (57). Consistently, the functional interaction between G␣ q and LARG requires only the DH-PH domains of LARG. Thus, the interaction between Rgnef and G␣ q and the fact that G␣ q promotes RhoA activation through Rgnef are not surprising. Moreover, the results with the Rgnef-5M point to a clear difference in the mechanism of activation of Rgnef between G␣ q and G␣ 13 . Further work will be necessary to address which domain in Rgnef is involved in the activation by G␣ q and/or whether p63RhoGEF is also part of this pathway.
Previous studies demonstrated that Rgnef is a key regulator of Rho reactivation, focal adhesion establishment (5), and cell migration downstream of integrins (6). Rgnef can directly bind FAK through a motif in the Rgnef C-terminal region (64), and Rgnef-null cells showed fewer early adhesions upon fibronectin replating (6). Rgnef binding to membrane lipids is needed for activation of FAK at the initial steps of early adhesion, but this effect is independent of its RhoGEF activity (58). Rgnef-GEF activity and Rho activation are required later in cell migration for the process of stability and maturation of adhesion. Interestingly, it has been shown that integrins are noncanonical G␣ 13 -coupled receptors that provide a mechanism for dynamic regulation of RhoA (59), which poses the question whether the link between G␣ 13 and Rgnef could also be active downstream of integrin adhesions. . E, HEK293 cells expressing HA-Rgnef or HA-Rgnef5M with or without G␣ q R183C or G␣ 13 Q226L were co-immunoprecipitated with HA tag antibodies and detected by Li-COR immunoblotting. The arrowhead points to an unspecific band detected with the anti-G q monoclonal antibody. F, HEK293 cells expressing HA-Rgnef or HA-Rgnef 5M were incubated with RhoAG17A-GST beads and visualized by anti-HA tag immunoblotting. G, HEK293 cells were transfected with pSRE.L and pRL-TK, together with either empty vector (Ctl), vector encoding Rgnef or Rgnef5M and SRF activities were measured. Data are means Ϯ S.E. of three independent experiments, each conducted in duplicate (***, p Ͻ 0.001).
In DLD-1 cells, gastrin enhances FAK and paxillin phosphorylation and increases focal adhesion formation linked to the dissolution of cell-cell contacts associated with the loss of E-cadherin surface expression, a process that requires Rgnef (7). Importantly, our results show that activation of G␣ 13 downstream of gastrin leads to FAK and paxillin phosphorylation and is required for Rgnef activation and Rho activity. Taking into account those results and published records, we propose a model (Fig. 8) whereby, upon gastrin stimulation, G␣ 13 directs Rgnef to the plasma membrane where it associates with and activates FAK. Next, G␣ 13 GTP promotes Rgnef-GEF activity, leading to RhoA activation, which in turn can induce downstream paxillin phosphorylation and, as result of those processes, cell motility (which may involve FAK dissociation). Previous results demonstrated that Rgnef⌬FAK did not block the phosphorylation of paxillin downstream of gastrin (7). However, G␣ 13 -induced-SRE activity was inhibited by Rgnef⌬FAK indicating that analyses of SRE luciferase activity may not always reflect changes in paxillin tyrosine phosphorylation despite both being enhanced by gastrin or activated G␣13 construct expression.
RhoA activation is implicated in colorectal tumor progression (2), and the levels of Rgnef mRNA and protein increase with colon carcinoma tumor progression (7). Although CCK2R is frequently overexpressed in cancers and the oncogenic properties of gastrin in the colonic mucosa of mouse are well established (60), its role in humans is still controversial (61). Nevertheless, the presence of CCK2R is recognized to provide a growth advantage to tumors where it is expressed (60). Together our results suggest a mechanism whereby gastrin stimulation of G␣ 13 activates Rgnef implicating a role in human colon carcinoma. Little is known about the specific functions of G␣ 13 in cancer, and those studies have focused on G␣ 12 (29). It is documented that G␣ 13 expression is markedly increased during prostate cancer progression and that micro RNAs regulate its expression post-transcriptionally (62). G␣ 13 also mediates lysophosphatidic acid-stimulated invasive migration of pancreatic cancer cells (63). The future analysis of G␣ 13 levels will help to elucidate whether its expression varies during colon carcinoma progression and further elucidate its function in colon cancer.