Identification of Critical Residues in Gα13 for Stimulation of p115RhoGEF Activity and the Structure of the Gα13-p115RhoGEF Regulator of G Protein Signaling Homology (RH) Domain Complex*

RH-RhoGEFs are a family of guanine nucleotide exchange factors that contain a regulator of G protein signaling homology (RH) domain. The heterotrimeric G protein Gα13 stimulates the guanine nucleotide exchange factor (GEF) activity of RH-RhoGEFs, leading to activation of RhoA. The mechanism by which Gα13 stimulates the GEF activity of RH-RhoGEFs, such as p115RhoGEF, has not yet been fully elucidated. Here, specific residues in Gα13 that mediate activation of p115RhoGEF are identified. Mutation of these residues significantly impairs binding of Gα13 to p115RhoGEF as well as stimulation of GEF activity. These data suggest that the exchange activity of p115RhoGEF is stimulated allosterically by Gα13 and not through its interaction with a secondary binding site. A crystal structure of Gα13 bound to the RH domain of p115RhoGEF is also presented, which differs from a previously crystallized complex with a Gα13-Gαi1 chimera. Taken together, these data provide new insight into the mechanism by which p115RhoGEF is activated by Gα13.


RH-RhoGEFs are a family of guanine nucleotide exchange factors that contain a regulator of G protein signaling homology (RH) domain. The heterotrimeric G protein G␣
stimulates the guanine nucleotide exchange factor (GEF) activity of RH-RhoGEFs, leading to activation of RhoA. The mechanism by which G␣ 13 stimulates the GEF activity of RH-RhoGEFs, such as p115RhoGEF, has not yet been fully elucidated. Here, specific residues in G␣ 13 that mediate activation of p115RhoGEF are identified. Mutation of these residues significantly impairs binding of G␣ 13 to p115RhoGEF as well as stimulation of GEF activity. These data suggest that the exchange activity of p115RhoGEF is stimulated allosterically by G␣ 13 and not through its interaction with a secondary binding site. A crystal structure of G␣ 13 bound to the RH domain of p115RhoGEF is also presented, which differs from a previously crystallized complex with a G␣ 13 -G␣ i1 chimera. Taken together, these data provide new insight into the mechanism by which p115RhoGEF is activated by G␣ 13 .
Heterotrimeric guanine nucleotide binding proteins (G proteins), composed of ␣, ␤, and ␥ subunits, act as molecular switches that cycle between an inactive, GDP-bound state and an active, GTP-bound state upon stimulation of G protein-cou-pled receptors (1). Once activated, the GTP-bound G␣ subunit dissociates from the G␤␥ dimer, both of which regulate the activity of multiple intracellular effectors to elicit cellular responses. The duration of signaling mediated by G␣ is dictated by the lifetime of bound GTP. All G␣ subunits have intrinsic GTPase activity that hydrolyzes GTP to GDP, and this rate of hydrolysis is enhanced by GAPs, 2 such as RGS proteins (2), which bind the switch regions of activated G␣ subunits, stabilize the transition state for GTP hydrolysis, and thus accelerate the hydrolysis of GTP to GDP (3,4). Once in the GDP-bound state, G␣ subunits reassociate with G␤␥ and are capable of initiating another round of signaling.
G␣ 13 is one of two members of the G 12 family of G proteins (5) and has been implicated in regulating multiple cellular processes that depend on activation of the monomeric GTPase RhoA, including gene transcription, embryogenesis, and rearrangement of the actin cytoskeleton (6 -8). A direct link between G␣ 13 and RhoA was established with the discovery of p115RhoGEF, the founding member of a family of RhoA-specific GEFs containing an RH domain in their N termini (RH-RhoGEFs) (9). Together with p115RhoGEF, PDZ-RhoGEF and LARG constitute the RH-RhoGEF family in mammals (10,11). The RH domain of p115RhoGEF has low sequence similarity to canonical RGS proteins but nevertheless functions as a GAP for G␣ 13 (12). Although the structure of the RH domain of p115RhoGEF is similar to that of other RGS proteins (13), additional elements flanking the RGS box are required for GAP activity (14) and stability of the isolated domain (15). In addition to its RH domain, p115RhoGEF also contains the tandem DH/PH domains characteristic of Dbl family RhoGEFs. The nucleotide exchange activity of p115RhoGEF can be directly stimulated by G␣ 13 in vitro (9), and G␣ 13 acts synergistically with p115RhoGEF to activate Rho-dependent signaling in cells (16). Thus, p115RhoGEF functions as both an effector and a GAP for G␣ 13 . Although the relationship between p115RhoGEF and G␣ 13 has been well established, the precise mechanism by which G␣ 13 regulates the activity of p115RhoGEF remains to be fully elucidated.
The molecular mechanisms regulating p115RhoGEF activity are likely to be complex and may involve multiple intermolecular interfaces with G␣ 13 . It has been clearly demonstrated that G␣ 13 binds directly to the RH domain of p115RhoGEF (12,15). Furthermore, this domain is required for both basal and G␣ 13 stimulated nucleotide exchange activity, suggesting that it plays a critical role in the activation mechanism (15). A structure of the RH domain of p115RhoGEF bound to an AlF 4 Ϫ -activated G␣ 13 -G␣ i1 chimera has been solved by x-ray crystallography and revealed that the ␣ subunit engages this domain through two distinct interfaces (17). In addition to interacting with switch regions I and II of G␣ 13/i1 through an N-terminal extension of the RGS box, the RH domain also docks into the hydrophobic groove between the ␣ 2 and ␣ 3 helices of G␣ 13/i1 , a highly conserved effector interface among other G␣-effector pairs (18 -21). However, many of the residues in the ␣ 3 helix of this chimera are derived from G␣ i1 , and its ability to activate p115RhoGEF has not been clearly demonstrated. Thus, the role of residues in the ␣ 3 helix in regulating p115RhoGEF activity is unknown. G␣ 13 has also been reported to bind to the isolated DH/PH domains of p115RhoGEF in vitro (15); however, it has no capacity to directly stimulate the GEF activity of this fragment. Therefore, the role of additional binding sites outside of the RH domain is also unclear.
Here, specific residues in G␣ 13 that mediate activation of p115RhoGEF are identified. We demonstrate that mutation of these residues, which bind to the RH domain of p115RhoGEF, significantly impairs binding of G␣ 13 to p115RhoGEF as well as stimulation of GEF activity in cells and in vitro. Our results suggest that the initial step in stimulation of exchange activity by G␣ 13 involves allosteric regulation through interaction with the RH domain of p115RhoGEF. We also present a crystal structure of G␣ 13 bound to the RH domain of p115RhoGEF, which differs in several respects from the previously crystallized complex using the G␣ 13/i1 chimera (17) and reinforces the results of our biochemical analysis. This structure provides a more accurate picture of the G␣ 13 -p115RhoGEF RH interface, whereas the biochemical data provide new insight into the mechanism by which p115RhoGEF is activated by G␣ 13 .
Full-length G␣ 13 -Full-length G␣ 13 (G␣ 13 FL) was generated by first introducing a silent mutation into pCMV5-G␣ 13 by QuikChange site-directed mutagenesis that eliminates an internal HindIII site using the primers 5Ј-GCCCGAGA-GAAGCTCCATATTCCCTGGGG and 5Ј-CCCCAGGG-AATATGGAGCTTCTCTCGGGC. Mutated bases are underlined. KpnI and HindIII sites were then introduced at the 5Ј-and 3Ј-ends, respectively, of G␣ 13 by PCR using pCMV5-G␣ 13 lacking the internal HindIII site as the template and the primers 5Ј-GGTACCGCGGACTTCCTGCCGTC and 5Ј-GACCA-AGCTTTCACTGCAGCATGAGCTG. The PCR products were digested with KpnI and HindIII and ligated into the pFast-Bac HT(Ϫ) vector containing the N-terminal ␣ 1 helix of G␣ i1 such that the construct contains, from the N terminus, a His 6 tag, residues 1-28 of G␣ i1 , a TEV protease site, and residues 2-377 of G␣ 13 .
p115RhoGEF⌬C-Tyr 763 of p115RhoGEF was mutated to a stop codon using QuikChange site-directed mutagenesis using pFastBac HTc-p115RhoGEF as the template and the primers 5Ј-CACTGAGACTGCCGGATAACTGAAAGTCCCTGCCC and 5Ј-GGGCAGGGACTTTCAGTTATCCGGCAGTCTC-AGTG. Mutated bases are underlined.
Full-length G␣ 13 -His 6 -G␣ i/13 FL was purified from Sf9 cells in the same fashion as His 6 -G␣ i/13 . After elution from nickel-NTA resin (Qiagen), fractions containing His 6 -G␣ i/13 FL were pooled and incubated with 2% (w/w) His 6 -TEV protease overnight at 4°C to remove the His 6 tag and ␣ 1 helix of G␣ i1 . The protein was concentrated and buffer-exchanged to imidazolefree buffer using an Amicon Ultra 30k centrifugal filter device (Millipore Corp.). The protein was aliquoted, snap-frozen in liquid nitrogen, and stored at Ϫ80°C until use.
p115RhoGEF⌬C-His 6 -p115RhoGEF⌬C(1-762) was purified from Sf9 cells in the same fashion as full-length p115RhoGEF except that the final round of size exclusion chromatography utilized a HiLoad Superdex 200 16/60 size exclusion column (GE Healthcare), and the protein was eluted in buffer containing 20 mM HEPES, pH 8.0, 100 mM NaCl, 1 mM EDTA, 10% glycerol, and 2 mM DTT. The protein was aliquoted, snap-frozen in liquid nitrogen, and stored at Ϫ80°C until use.
GST-p115RhoGEF RH Domain-The GST-tagged RH domain of p115RhoGEF was purified from Escherichia coli as described previously (22).
RhoA-GST-RhoA lacking the C-terminal CAAX motif (RhoA(1-181)) was expressed from the plasmid pGEX-6P (GE Healthcare) in the E. coli strain BL21-CodonPlus (DE3)-RP (Stratagene). Cells were grown at 30°C to an A 600 of 0.6 -0.8, and protein expression was induced with 200 M isopropyl ␤-D-1-thiogalactopyranoside for 6 h. Cells were resuspended in lysis buffer (50 mM HEPES, pH 7.5, 1 mM EDTA, 1 mM DTT, 200 mM NaCl, 5 mM MgCl 2 , 10 M GDP, 10% glycerol) and lysed with lysozyme and sonication. The lysate was clarified by centrifugation at 100,000 ϫ g for 30 min at 4°C. The soluble portion of the lysate was applied to a column of glutathione-Sepharose 4B beads (GE Healthcare) equilibrated with lysis buffer. The column was washed with 10 bed volumes of lysis buffer, followed by 20 bed volumes of cleavage buffer (50 mM Tris, pH 7.5, 1 mM EDTA, 1 mM DTT, 5 mM MgCl 2 , 10 M GDP). On-column cleavage of the GST tag was achieved using PreScission protease according to the manufacturer's protocol (GE Healthcare). Cleaved RhoA was eluted from the column using lysis buffer. Fractions containing RhoA were pooled and concentrated using an Amicon Ultra 15k centrifugal filter device. The protein was aliquoted, snap-frozen in liquid nitrogen, and stored at Ϫ80°C until use. To form nucleotide-free RhoA for crystallography, the protein was incubated with an excess of EDTA on ice.

Biochemical Experiments
SRE-luciferase Assays-Assays were performed as described (23), with the following modification. Each well was co-transfected with 200 ng of pGL3-SRE.L, 100 ng of pCMV5-LacZ, and 10 ng of empty pCMV5 vector or the indicated G␣ 13 construct using Lipofectamine2000 (Invitrogen). For immunoblotting of cell lysates, samples were harvested directly into SDS-PAGE sample buffer and boiled. Cell lysate was separated by SDS-PAGE, and G␣ 13 was detected by Western blotting with the B860 antibody (24). GAPDH was detected using a monoclonal antibody (clone 6C5, Ambion).
In Vitro Trypsin Protection Assays-3 g of G␣ i/13 protein (wild type or mutant) was diluted into reaction buffer (20 mM HEPES, pH 8.0, 5 mM MgCl 2 , 100 M GDP) and incubated on ice for 45 min in the presence or absence of AlF 4 Ϫ (10 mM NaF, 30 M AlCl 3 ). Trypsin was added to a final concentration of 13.3% (w/w), and the samples were incubated at 30°C for 15 min. The reaction was terminated by the addition of SDS-PAGE sample buffer and boiling.
GST Pull-down Assays-Assays were preformed as described previously (22), with the exception that bound proteins were released from the beads by the addition of SDS-PAGE sample buffer and boiling.
Single Turnover GAP Assays-GTPase assays were performed as described previously (12), except that reactions were incubated on ice. [␥-32 P]GTP (specific activity ϭ 6,000 Ci/mmol) was obtained from PerkinElmer Life Sciences.
Gel Filtration of Full-length p115RhoGEF with His 6 -G␣ i/13 or His 6  In Vitro RhoGEF Assays-Assays were performed as described previously (25). [ 35 S]GTP␥S (specific activity ϭ 1,250 Ci/mmol) was obtained from PerkinElmer Life Sciences.

Crystallization and Data Collection
Crystallization-To form the G␣ i/13 -p115⌬C-RhoA complex, G␣ i/13 was activated with 8ϫ AlF 4 Ϫ (80 mM NaF, 160 M AlCl 3 ) on ice for 15 min and mixed with an equimolar amount of p115RhoGEF⌬C and nucleotide-free RhoA. The complex was subjected to size exclusion chromatography on a HiLoad Superdex 200 10/16 gel filtration column, and fractions containing the stoichiometric complex were pooled and concentrated using an Amicon4 10k centrifugal filter device. Screening of the G␣ i/13 -p115RhoGEF⌬C-RhoA complex was carried out using vapor diffusion at 20°C. Diffraction quality crystals of the G␣ i/13 -p115RhoGEF RH domain complex were grown against a reservoir solution containing 15% PEG 4000 and 0.1 M HEPES (pH 7.0). Single crystals were coated with the reservoir solution containing 20% PEG 400 as a cryoprotectant, mounted using a nylon loop, and flash-cooled in the cold stream of the goniometer. Data Collection-Diffraction data were collected at 100 K with a wavelength of 1.0 Å at SPring-8 beam line BL41XU (Harima, Japan). Data were processed with the HKL2000 program (26). The structure of the G␣ i/13 -p115RhoGEF RH domain complex was determined by molecular replacement with the program MOLREP (CCP4), using the RH domain of p115RhoGEF (Protein Data Bank entry 1IAP, chain A) and G␣ i/13 (Protein Data Bank entry 1ZCB) as search models. The model was corrected iteratively using O (27), and structure refinement was carried out using CNS (Crystallography and NMR System) (28). The final refinement statistics are shown in Table 1. The quality of the model was inspected by the program PROCHECK in the CCP4 suite (29). Structural similarities were calculated with DALI (30). Solvent-accessible surface area was calculated with the program AREALMOL in the CCP4 suite (29). Graphic figures were created with the program PyMOL (31).

RESULTS
Previous studies have suggested that the region of G␣ 13 responsible for stimulating the guanine nucleotide exchange activity of p115RhoGEF is located C-terminal to the three switch regions (23,32). These results are supported by crystallographic data, which demonstrate that the ␣ 3 helix and ␣ 3 -␤ 5 loop in G␣ 13 form a putative effector binding site with the RH domain of p115RhoGEF (17). However, the capacity of residues in this region to stimulate GEF activity has not been directly demonstrated. We utilized these data along with the interaction between G␣ q and its effectors PLC-␤ and GRK2 as models in order to identify specific amino acid residues responsible for activating p115RhoGEF. Recent work has identified several residues that are critical for the interaction between G␣ q and these effectors, namely Ala 253 , Thr 257 , and Tyr 261 (20,33). 3 The G␣ q -PLC-␤ interaction was selected as a model because, like RH-RhoGEFs, PLC-␤ functions both as a GAP (34) and effector (35) for G␣ q . Based on these data, Asn 270 , Thr 274 , and Asn 278 in G␣ 13 , which correspond to Ala 253 , Thr 257 , and Tyr 261 in G␣ q (Fig. 1A), were targeted for site-directed mutagenesis, and the effects of these mutations on the ability of G␣ 13 to regulate p115RhoGEF activity were examined.
The T274E and N278A Mutations Impair Rho Activation in Cells-A constitutively active mutant of G␣ 13 , G␣ 13 Q226L (hereafter referred to as G␣ 13 QL), has been shown to stimulate transcription from the SRE in a Rho-dependent manner when overexpressed in cells (6). Therefore, to test the ability of the N270A, T274E, N278A, and T274E/N278A mutants to stimulate Rho activation, HeLa cells, which endogenously express RH-RhoGEFs and RhoA, were transiently co-transfected with plasmids encoding each mutant in the background of G␣ 13 QL and a construct encoding the luciferase reporter gene under the control of the SRE. The activation status of Rho in cells was indirectly measured as luciferase activity of cell lysates. In this system, the N270A mutation in G␣ 13 QL did not affect Rho activation compared with G␣ 13 QL (Fig. 1B). However, both the T274E and N278A mutations decreased Rho activation by ϳ50%, and Rho activity could be reduced to nearly basal levels by the T274E/N278A double mutation. Immunoblotting of cell lysates demonstrated equal expression levels of the constructs. These data suggested that residues Thr 274 and Asn 278 , but not Asn 270 , are important for the ability of G␣ 13 to stimulate Rho activation in cells. 3 K. Tsuboi and T. Kozasa, unpublished data. FIGURE 1. Characterization of G␣ i/13 mutants. A, a primary sequence alignment of murine G␣ 12 , G␣ 13 , and G␣ q was generated using the program T-Coffee. Residues in G␣ 13 analyzed by site-directed mutagenesis in this study are underlined. Secondary structure was assigned based on the crystal structure of the G␣ 13/i1 -p115 RH domain complex (Protein Data Bank entry 1SHZ) (17). B, the T274E and N278A mutations in G␣ 13 impair Rho activation in cells. HeLa cells were transiently transfected with empty vector or the indicated G␣ 13 QL construct. The luciferase activity of cell lysates was determined as described under "Experimental Procedures." Total cell lysate was immunoblotted for either G␣ 13 or GAPDH. Data are presented as the mean Ϯ S.E. (error bars) of triplicate determinations from a single experiment, representative of three independent experiments with similar results. Data were analyzed by oneway analysis of variance followed by Dunnett's post-test. Statistically significant difference from G␣ 13 QL is shown as follows. ns, not significant; **, p Ͻ 0.01. C, G␣ i/13 mutants can undergo activation-dependent conformational changes. G␣ i/13 (wild type or mutant) was subjected to limited trypsin digestion in the presence or absence of AlF 4 Ϫ . After proteolysis, proteins were separated by SDS-PAGE and stained with Coomassie Brilliant Blue. The primary protected species is indicated with an arrow. The positions of molecular mass standards, in kDa, are shown on the left. G␣ i/13 N270A, T274E, N278A, and T274E/N278A Are Functional G␣ Subunits-To analyze these mutants in more detail, each mutation was introduced into a construct harboring amino acid residues 1-28 of G␣ i1 and 47-377 of G␣ 13 (hereafter referred to as G␣ i/13 ). The resulting protein was expressed and purified from Sf9 cells for biochemical reconstitution assays. The use of this G␣ i/13 chimera facilitates the production of recombinant protein by increasing the yield from Sf9 cells, and it retains all of the biochemical properties characteristic of native G␣ 13 , including the ability to regulate RH-RhoGEFs (22). In order to confirm that the mutants were functional G␣ subunits, limited trypsin digestion assays were performed. In their inactive, or GDP-bound, form, G␣ subunits adopt a conformation that renders them susceptible to digestion by the serine protease trypsin (36). However, in the active form, which can be induced by binding to the reversible activator AlF 4 Ϫ , the conformation changes such that the G␣ subunit is protected from digestion, largely due to ordering of the ␣ 2 helix in switch II, although trypsin still cleaves the N terminus of the G␣ subunit. Thus, the protected species runs at a smaller molecular weight than the native protein. Trypsin digestion assays can therefore be used to determine if a G␣ subunit is capable of adopting an activated conformation. As shown in Fig. 1C, each G␣ i/13 mutant was protected from tryptic digestion in the presence of AlF 4 Ϫ , as evidenced by the existence of the lower molecular weight species, indicating that these mutants were capable of undergoing the activation-dependent conformational changes indicative of a functional G␣ subunit.
The T274E and N278A Mutations Disrupt Binding to the RH Domain of p115RhoGEF and Impair GAP Activity-Because these mutations were known to be in the G␣ 13 -p115RhoGEF RH domain interface (17), the ability of these mutants to interact with the RH domain was assessed using GST pull-down assays, and their capacity to respond to the GAP activity of this domain was determined using single turnover GTPase assays. The activation-dependent interaction between each mutant and the RH domain of p115RhoGEF is shown in Fig. 2. Although the N270A mutation had no apparent effect on the ability of G␣ i/13 to interact with GST-p115 RH, the G␣ i/13 T274E mutant had significantly reduced binding, whereas the N278A mutation produced an intermediate effect. The G␣ i/13 T274E/N278A double mutant had no detectable binding to the RH domain under these conditions. Previous work has established that the isolated RH domain (amino acids 1-252) of p115RhoGEF is as effective a GAP for G␣ 13 as the full-length protein (15). Therefore, GST-p115 RH was also used to examine the effects of these mutations on p115 RH-stimulated GAP activity in vitro (Fig. 3). Wild type G␣ i/13 has a measurable rate of basal GTP hydrolysis, which can be markedly increased by the presence of GST-p115 RH (Fig. 3A). At 0°C, 10 nM GST-p115 RH increased the initial velocity of the hydrolysis reaction (V o ) from ϳ0.08 to ϳ1.8 fmol/s, or ϳ22fold (Fig. 3F). G␣ i/13 N270A displayed a very similar rate of RH domain-stimulated hydrolysis (Fig. 3F). Consistent with the fact that G␣ i/13 T274E binds the RH domain of p115RhoGEF poorly (Fig. 2), its response to the GAP activity of the RH domain was also significantly impaired. In the presence of the RH domain, GTP hydrolysis by G␣ i/13 T274E was reduced to nearly basal levels (Fig. 3, C and F). In the case of G␣ i/13 N278A, the reduction in GAP activity in response to the RH domain was ϳ50% (Fig. 3, D and F). Like G␣ i/13 T274E, the double mutant, G␣ i/13 T274E/N278A, had a rate of RH-stimulated hydrolysis that was reduced to essentially basal levels (Fig. 3, E and F). Therefore, the T274E mutation strongly affects the ability of G␣ i/13 to bind the RH domain of p115RhoGEF and respond to its GAP activity, whereas the effect of the N278A mutation was also clear but less pronounced. The N270A mutation had no detectable effect on RH domain binding or GAP activity under the assay conditions used in Figs. 2 and 3.
G␣ i/13 T274E/N278A Cannot Bind to Full-length p115RhoGEF-Next, we assessed the ability of the T274E/ N278A double mutant to bind to full-length p115RhoGEF by size exclusion chromatography. As shown in Fig. 4, upon activation with AlF 4 Ϫ , G␣ i/13 forms a stable complex with p115RhoGEF, as evidenced by its elution in earlier, p115RhoGEF-containing fractions. In contrast, binding of the double mutant to full-length p115RhoGEF in the presence of AlF 4 Ϫ was not detectable at the level of Coomassie Brilliant Blue staining. Given that micromolar concentrations of p115RhoGEF and G␣ i/13 T274E/N278A were insufficient to form a stable complex during the course of gel filtration, the affinity of G␣ i/13 T274E/N278A for full-length p115RhoGEF is likely to be quite low, suggesting that the primary binding site for G␣ 13 is in the RH domain.
G␣ 13 T274E/N278A Fails to Activate p115RhoGEF in Vitro-We next examined whether G␣ i/13 T274E/N278A could regulate p115RhoGEF activity directly by measuring GTP␥S binding to RhoA in vitro. Although G␣ i/13 was able to stimulate RhoA activation through p115RhoGEF (Fig. 5A), G␣ i/13 T274E/ N278A failed to increase GTP␥S binding to RhoA over the p115RhoGEF only condition. These results demonstrate that the T274E/N278A double mutation inhibits the ability to G␣ i/13 to stimulate the nucleotide exchange activity of p115RhoGEF.
Recently, we successfully generated recombinant, full-length G␣ 13 comprising amino acid residues 2-377. In the course of analyzing this protein, we found that it is a more efficient activator of RhoA in vitro than a functionally equivalent amount of G␣ i/13 (Fig. 5B). This suggests that the N-terminal region of Ϫ . GST-p115 RH was pulled down (PD) with glutathione-Sepharose beads. Bound proteins were released by boiling and separated by SDS-PAGE. G␣ i/13 was detected by immunoblotting (IB). GST-p115 RH was stained with Coomassie Brilliant Blue. Data presented are from one experiment, representative of three independent experiments with similar results. G␣ 13 may play a role in efficient stimulation of RhoGEF activity. Due to the higher activity of G␣ 13 in the in vitro RhoGEF assay, we chose to confirm the effect of the T274E/N278A mutations in the background of this full-length protein and examined its ability to stimulate RhoGEF activity. 100 nM full-length G␣ 13 was able to significantly increase the amount of GTP␥S bound to RhoA compared with the p115RhoGEF only condition (Fig.  5, B and C). In contrast, RhoA activation stimulated by a functionally equivalent amount of G␣ 13 T274E/N278A was significantly reduced compared with wild type G␣ 13 (Fig. 5C). The effects of the T274E and N278A mutations on the ability of G␣ 13 to stimulate p115RhoGEF activity in vitro are shown in supplemental Fig. 1. We note that in contrast to the SRE assay, in which the activity of both G␣ 13 QL T274E and G␣ 13 QL N278A was reduced ϳ50% (Fig. 1B), G␣ 13 T274E failed to stimulate RhoA activation in vitro, whereas G␣ 13 N278A had no apparent defect. However, in vitro, the functional concentra-tion of G␣ 13 and its mutants was quantitated by [ 35 S]GTP␥S binding, which allows for a comparison between functionally equivalent amounts of protein. In contrast, the SRE assay is a cell-based overexpression system that precludes the tight control over the concentration of G␣ 13 afforded by our reconstitution assay. Thus, it is difficult to make a direct comparison between the behavior of the mutants in vitro and in cell-based experiments. Taken together, the results using both G␣ i/13 and G␣ 13 are consistent and in agreement with the results of the cell-based assays, thus providing strong evidence that residues in the ␣ 3 helix of G␣ 13 regulate the GEF activity of p115RhoGEF.
X-ray Crystallography of the G␣ i/13 -p115RhoGEF RH Domain Complex-To gain more insight into the interaction between G␣ i/13 and p115RhoGEF, we purified the ternary complex of G␣ i/13 , a deletion mutant of p115RhoGEF lacking the region C-terminal to the PH domain (p115RhoGEF⌬C), and

RhoA by size exclusion chromatography in the presence of AlF 4
Ϫ and subjected the complex to crystallization screening. Platelike crystals formed at 20°C after ϳ6 weeks and contained a complex of G␣ i/13 and the RH domain of p115RhoGEF. However, the region C-terminal to the RH domain, including the DH and PH domains and RhoA, was not present in the crystal. The x-ray crystal structure of the p115RhoGEF RH domain in complex with G␣ i/13 , GDP, Mg 2ϩ , and AlF 4 Ϫ was determined at 2.4 Å resolution ( Table 1). The structure contained two pairs of the G␣ i/13 -p115RhoGEF RH domain complex in the asymmetric unit (A-B and C-D; molecules A and C designate G␣ i/13 , and molecules C and D designate the p115RhoGEF RH domain (see supplemental Tables 1 and 2 Ϫ ; and 87 water molecules. For the purpose of discussing the crystal structure, we hereafter refer to the complex as G␣ 13 -p115 RH because the N terminus of G␣ i/13 containing the residues from G␣ i1 was disordered, and thus, only residues native to G␣ 13 were present. Overall Structure-In general, the structure of the G␣ 13 -p115 RH complex is similar to the previously reported crystal structure of the G␣ 13/i1 chimera-p115RhoGEF RH domain complex (17). Both G␣ 13/i1 and G␣ 13 interact with the RH domain of p115RhoGEF through two distinct surfaces (Fig. 6). The N-terminal extension of the RH domain, the ␤N-␣N hairpin, makes extensive contacts with the helical domain and switch regions I and II of G␣ 13/i1 . This interface is also present in the current structure; however, the ␤N strand is disordered (Fig. 6). Additionally, the RGS box binds to an effector-like interaction surface consisting of the ␣ 2 and ␣ 3 helices and ␣ 3 -␤ 5 loop of both G␣ 13/i1 and G␣ i/13 (Fig. 6). However, unlike the G␣ i/13 used in the current study, G␣ 13/i1 contains several resi-dues from G␣ i1 in its effector interface, and its ability to stimulate RhoGEF activity has not been clearly demonstrated. The result of these substitutions is a significant difference in the effector interfaces between the two structures at the atomic level.
The Effector Interface-Compared with G␣ 13/i1 , G␣ 13 is shifted 3 Å closer to the RH domain of p115RhoGEF. This is probably a reflection of the higher affinity of this domain for G␣ 13 versus G␣ 13/i1 . Additionally, the bulky side chain of Trp 280 in G␣ 13/i1 , which corresponds to Val 280 in G␣ 13 , may sterically  Ϫ -activated G␣ subunit. Samples were incubated at 30°C and quenched in ice-cold buffer containing 10 mM MgSO 4 at the indicated time. A, G␣ i/13 T274E/N278A fails to stimulate p115RhoGEF activity in vitro. Samples are as follows: RhoA only (F), p115RhoGEF (f), p115RhoGEF plus G␣ i/13 T274E/N278A (OE), and p115RhoGEF plus G␣ i/13 (ࡗ). Data are presented as the mean of single determinations pooled from two independent experiments. B, full-length G␣ 13 is a more efficacious activator of p115RhoGEF in vitro than G␣ i/13 . GTP␥S binding to RhoA was determined after a 20-min incubation as described above. Data are presented as the mean Ϯ S.E. (error bars) of single determinations pooled from three independent experiments. Data were analyzed by one-way analysis of variance, followed by Dunnett's post-test. Statistically significant difference from the p115RhoGEF condition is shown as follows. ns, not significant; **, p Ͻ 0.01. C, the T274E/N278A double mutation significantly reduces p115RhoGEF activation by full-length G␣ 13 . Samples are as follows: RhoA only (F), p115RhoGEF (f), p115 RhoGEF plus G␣ 13 T274E/N278A (), p115RhoGEF plus G␣ 13 (OE), and G␣ 13 alone (Ⅺ). Data are presented as the mean Ϯ S.E. of single determinations pooled from three independent experiments.
inhibit the formation of a tight complex. G␣ 13 forms more hydrogen bonds with p115 RH as compared with G␣ 13/i1 (9 versus 3 possible hydrogen bonds) (Fig. 6). In G␣ 13 , Glu 273 forms an ion pair with Lys 160 in the RH domain of p115RhoGEF. In contrast, Glu 273 in G␣ 13/i1 makes no contact with the RH domain, whereas Lys 160 participates in a van der Waals interaction with Ser 274 of G␣ 13/i1 . The backbone of Thr 274 in G␣ 13 (Ser 274 in G␣ 13/i1 ) engages the side chain of Gln 69 of the RH domain through hydrogen bonding, whereas G␣ 13/i1 interacts with Gln 69 via a hydrogen bond with the side chain of Asn 278 . Given that this is one of the three possible hydrogen bonds between G␣ 13/i1 and the RH domain of p115RhoGEF, perturbing this interaction might be expected to result in a large decrease in binding. However, the results of our experiments with the N278A mutant of G␣ 13 suggest that this residue is not as critical for the interaction with the RH domain of p115RhoGEF as Thr 274 . Consistent with this notion, in G␣ 13 , Asn 278 does not participate in hydrogen bonding to p115 RH but rather is in hydrophobic contact with Leu 68 . Arg 279 of G␣ 13 makes extensive contacts with p115 RH, including a hydrogen bond to the backbone carboxyl group of Phe 70 , a residue that has previously been implicated in binding to G␣ 13 (14). In contrast, Phe 70 makes no contact with G␣ 13/i1 . Additionally, Arg 279 hydrogen-bonds to two additional residues in the RH domain, Ala 67 and Leu 68 , through a main chain-side chain bond and a main chain-main chain bond, respectively. As a result of these amino acid differences, the calculated surface area buried in the effector interface of the G␣ 13 -p115 RH complex (1,600 Å 2 ) is significantly larger than that of the G␣ 13/i1 -p115 RH complex (1,200 Å 2 ).
The GAP Interface-Although there are significant differences in the effector interface between G␣ 13 and G␣ 13/i1 , both interact with the GAP surface of the RH domain in an essentially identical manner (supplemental Fig. 2), which is consis-tent with the fact that the ␣ helical domain and switch regions of G␣ 13/i1 are largely derived from G␣ 13 . The interaction between the catalytic residue Arg 200 of G␣ 13 and Glu 27 of p115 RH is found in both structures. Arg 260 in G␣ 13/i1 , which hydrogenbonds with the main chain carbonyl of Ile 23 and forms an ion pair with Asp 28 in p115 RH, participates in the same bonding interactions in the G␣ 13 -p115 RH structure. Likewise, the interaction between Met 257 from G␣ 13 and Phe 31 of p115 RH, which in the G␣ 13/i1 -p115 RH domain structure is a van der Waals contact, is also present in the current structure (supplemental Table 1).
Structural Aspects of the N270A, T274E, and N278A Mutations-In the G␣ 13 -p115 RH domain complex, Asn 270 of G␣ 13 interacts with p115 RH primarily via hydrophobic contact with Met 163 . However, this interaction is well outside of the effector interface, supporting the results of cell-based experiments demonstrating that the N270A mutation does not impact GEF activity. In contrast, the T274E mutation resulted in reduced Rho activation in cells and severely disrupted binding to p115 RH and the GAP response. These observations are consistent with the fact that Thr 274 hydrogen-bonds, via its backbone carbonyl, to Gln 69 in the p115 RH domain, a residue known to be important for G␣ 13 binding (14). It is likely that the introduction of a longer, charged side chain in place of Thr 274 , as is the case with the T274E mutation, disrupts the interaction with Gln 69 and that this reduced binding results in impaired GAP and GEF activity. Additionally, in molecules C and D, Thr 274 in G␣ 13 makes additional contacts with p115 RH via hydrogen bonding of its side chain to the side chain of Lys 160 and hydrophobic interaction with Met 165 . Thus, the effects of the T274E mutation may also be manifest in the form of steric clashing and perturbation of hydrophobic contacts. To probe the role of the side chain of Thr 274 in the interaction with the RH domain of p115RhoGEF, we made several additional point mutations at this position in G␣ 13 and assessed the activity of these mutants in the SRE assay. As shown in Fig. 7A, neither the T274A nor the T274S mutation affected Rho activation by G␣ 13 QL in this system. Although the T274V mutant had slightly reduced activity, the effect was not nearly as severe as the T274E mutation. Thus, these data strongly suggest that Thr 274 interacts with the RH domain primarily through its backbone. Asn 278 interacts with p115 RH solely through a hydrophobic interaction with residue Leu 68 . Thus, the modest decreases in binding to the RH domain and GAP activity caused by the N278A mutation are consistent with the fact that Asn 278 does not contribute substantially to the interface with p115 RH. In addition, other residues in G␣ 13 that we did not initially target for site-directed mutagenesis also contribute to the interface with the RH domain of p115RhoGEF, namely Glu 273 and Arg 279 . In order to determine if these residues contribute to the ability of G␣ 13 to stimulate GEF activity, the E273K and R279E mutants were generated and analyzed by an SRE assay. As shown in Fig. 7B, mutation of these residues impairs the ability of G␣ 13 to stimulate Rho activation in cells, consistent with the notion that the G␣ 13 -p115 RH interface has the capacity to regulate effector activity. Thus, the results of the current biochemical analysis as well as previously published mutational analysis of p115RhoGEF (14) are more consistent with the G␣ 13 -p115 RH interface presented here than with the interface between the RH domain and G␣ 13/i1 . Comparison with PDZ-RhoGEF-Interestingly, the effector interface between G␣ 13 and the RH domain of p115RhoGEF is largely conserved in PDZ-RhoGEF ( Fig. 8 and supplemental Table 3). Although Asn 270 of G␣ 13 participates in hydrophobic interaction with p115 RH, mutation of this residue has no apparent effect on binding, GAP, or GEF activity in cells. There is no interaction between this residue and PDZ-RhoGEF. In contrast, residue Thr 274 of G␣ 13 , due to rotation of the side chain, makes more extensive contact with PDZ-RhoGEF RH than with p115RhoGEF RH. In addition to the interaction with Gln 351 , which is analogous to Gln 69 in p115RhoGEF, via its backbone carbonyl group, Thr 274 is also involved in hydrophobic interactions with Lys 439 and Leu 444 . Residue Asn 278 in G␣ 13 also interacts with Gln 351 of PDZ-RhoGEF via hydrophobic contact. Arg 279 of G␣ 13 , which has extensive interaction with the RH domain of p115RhoGEF, makes fewer contacts with PDZ-RhoGEF. However, the main chain-main chain interaction between Arg 279 and Leu 68 of p115RhoGEF is conserved in the analogous residue of PDZ-RhoGEF, Ser 350 . Thus, the prominent features of the effector interface between G␣ 13 and p115RhoGEF RH are conserved in the G␣ 13 -PDZ-RhoGEF RH crystal structure as well, suggesting that this interface may also play a role in regulating the GEF activity of PDZ-RhoGEF.

DISCUSSION
The RH domains found in RH-RhoGEFs, such as p115RhoGEF, are characterized by low sequence identity to classical RGS domain-containing proteins, such as RGS2 and Ϫ are shown as space-filling spheres. Carbon, oxygen, nitrogen, phosphate, magnesium, aluminum, and fluoride atoms are colored white, red, dark blue, orange, purple, light gray, and light blue, respectively. The disordered region in p115RhoGEF between the GAP interface and the RGS box is shown as a dashed yellow line. Bottom, a detailed view of the G␣ 13 -p115 RH domain effector interface. The G␣ 13 -p115 RH complex is shown on the left, and for comparison, the G␣ 13/i1 -p115 RH complex (Protein Data Bank entry 1SHZ) is shown on the right (17). The former is colored as in the top, whereas in the latter structure, G␣ 13/i1 is colored blue, and the p115 RH domain is purple. Residues that contribute to the G␣-p115 RH interface are labeled. In G␣ 13/i1 , residues that are not native to G␣ 13 are indicated in red. Hydrogen bonds are depicted as dashed lines. Oxygen and nitrogen atoms are colored red and blue, respectively.
RGS4 (12,37). Additionally, unlike these RGS proteins, the RGS box of p115RhoGEF does not contain the residues necessary to accelerate GTP hydrolysis by G␣ 13 ; these catalytic residues are found in the acidic ␤N-␣N hairpin N-terminal to the RGS box (14,17). The RH domain of p115RhoGEF has also been implicated in the regulation of nucleotide exchange activity because deletion of the first 288 amino acids of p115RhoGEF impairs both basal and G␣ 13 -stimulated GEF activity (15). Given that removal of the first 42 amino acids of p115RhoGEF, which are required for GAP activity, has no effect on the ability of G␣ 13 to stimulate GEF activity (15), the GAP and GEF functions are clearly distinct properties of the RH domain. However, specific residues in G␣ 13 responsible for stimulating p115RhoGEF activity have not been identified. Although x-ray crystallography has demonstrated that a G␣ 13/i1 chimera forms a putative effector-like interface with the RH domain of p115RhoGEF (17), the capacity of residues in this interface, many of which are derived from G␣ i1 , to regulate GEF activity is unknown.
Our results demonstrate that G␣ 13 regulates p115RhoGEF activity via the classical effector interface it forms with the RH domain. Specifically, mutation of Thr 274 and Asn 278 in the ␣ 3 helix of G␣ 13 impairs the ability of G␣ 13 to stimulate Rho activation in cells, impairs binding to p115RhoGEF, and reduces the capacity of G␣ 13 to stimulate the guanine nucleotide exchange activity of p115RhoGEF in vitro. We also determined the structure of the G␣ 13 -p115RhoGEF RH complex and found that this effector interface differs in several important respects from the interface formed with G␣ 13/i1 . Importantly, these new structural data are strongly supported by the results of our biochemical experiments.
In the context of previously reported data, our results are consistent with earlier studies that broadly defined the Rho-GEF-activating surface of G␣ 13 as the region C-terminal to the Ras-like domain (32), more specifically the last 100 amino acid residues (23). However, given that we have demonstrated that the interaction between the RH domain and G␣ 13 also regulates the effector activity of p115RhoGEF, this suggests that the pre-viously described interface between G␣ 13 and the DH/PH domains of p115RhoGEF may not contribute significantly to stimulation of GEF activity (15). If the RH domain represents the sole binding site for G␣ 13 , how might this interaction regulate the activity of the DH/PH domains? One study using PDZ-RhoGEF has suggested that the RH domain may work in concert with elements found in the linker region between the RH and DH domains to regulate GEF activity (38). Specifically, an acidic cluster of residues located in the linker may interact directly with residues in the DH domain to autoinhibit PDZ-RhoGEF activity. Mutation of these residues enhances GEF activity in vitro but only in the absence of the RH domain. Recently, the linker between the RH and DH domains of p115RhoGEF has also been implicated in regulating basal GEF activity, although the mechanism of autoinhibition appears to be distinct from that employed by PDZ-RhoGEF (39). A stretch of 40 amino acids located in the linker region of p115RhoGEF has been proposed to disrupt the interaction between the DH domain and residues in the "GEF switch" immediately N-terminal to it, an interaction that is critical for basal GEF activity. Thus, one consequence of G␣ 13 binding to the RH domain may be to reorient elements within the linker region of these RH-RhoGEFs, allowing for enhanced exchange activity.
A comparison of the structures of G␣ 13 bound to the RH domains of p115RhoGEF and PDZ-RhoGEF reveals that it binds this domain in a nearly identical manner. Additionally, the T274E and N278A mutations in G␣ 13 also impair the interaction with the RH domain of LARG in vitro (data not shown). Thus, it is likely that G␣ 13 engages the RH domains of all three RH-RhoGEFs in the same manner and stimulates RhoGEF activity through a similar mechanism. Although LARG is clearly regulated directly by G␣ 13 in vitro (40), stimulation of the RhoGEF activity of PDZ-RhoGEF by G␣ 13 in vitro has not been detectable under the conditions tested (15). However, PDZ-RhoGEF has been linked to G␣ 13 -mediated Rho activation in cells (10). Thus, although the effector interface formed between G␣ 13 and the RH domain of p115RhoGEF bears strik- FIGURE 7. Mutational analysis of the effector interface between G␣ 13 and p115 RH. A, the T274A and T274S mutations in G␣ 13 fail to impair Rho activation in cells. HeLa cells were transiently transfected with empty vector or the indicated G␣ 13 QL construct. The luciferase activity of cell lysates was determined as described under "Experimental Procedures." Total cell lysate was immunoblotted for either G␣ 13 or GAPDH. Data are presented as the mean Ϯ S.E. (error bars) of triplicate determinations from a single experiment, representative of three independent experiments with similar results. Data were analyzed by one-way analysis of variance followed by Dunnett's post-test. Statistically significant difference from G␣ 13 QL is shown as follows. **, p Ͻ 0.01; ***, p Ͻ 0.001. B, the E273K and R279E mutations in G␣ 13 impair Rho activation in cells. Assays were performed as in A. Statistically significant difference from G␣ 13 QL is shown as follows. **, p Ͻ 0.01.
ing resemblance to the G␣ 13 -PDZ-RhoGEF RH interface, in the latter case, this interaction is not sufficient to regulate RhoGEF activity in vitro. Post-translational modification, such as phosphorylation, or interaction with a co-activating protein may be required to render PDZ-RhoGEF susceptible to regulation by G␣ 13 .
It is also possible that activation of p115RhoGEF by G␣ 13 requires multiple intermolecular interfaces. Binding of G␣ 13 to the RH domain may result in a conformational change that brings additional regions of p115RhoGEF, perhaps the catalytic DH/PH domains, into contact with G␣ 13 , allowing for efficient activation of exchange activity. Although activated G␣ 13 clearly fails to stimulate the GEF activity of p115RhoGEF fragments lacking the RH domain, it has been reported to bind to the isolated DH/PH domains of p115RhoGEF in vitro (15). Additionally, the N terminus of G␣ 13 , which is replaced by the ␣N helix of G␣ i1 in G␣ i/13 and is disordered in the present crystal structure, may contribute to the binding interface with p115RhoGEF. This hypothesis is supported by the fact that G␣ 13 exhibits more robust activity in the RhoGEF assay compared with G␣ i/13 in vitro (Fig. 5B). The results of our gel filtration analysis suggest that G␣ i/13 T274E/N278A cannot bind to p115RhoGEF and are consistent with the in vitro RhoGEF assays demonstrating that G␣ i/13 T274E/N278A fails to stimulate RhoA activation over the p115RhoGEF condition. These data argue in favor of a model in which G␣ 13 binds exclusively to the RH domain; however, it is difficult to rule out the possibility of additional low affinity binding sites outside of the RH domain. Given that the RH domain occupies the classical effector interface on G␣ 13 , any additional interface would probably be a novel surface that has not been observed in other G␣-effector pairs. Aside from the residues reported here, however, screening of additional G␣ 13 mutants has so far failed to identify other residues that affect its capacity to stimulate Rho activation in cells. 4 Little is known about the mechanism by which the other G 12 family member, G␣ 12 , stimulates RH-RhoGEF activity. p115RhoGEF also functions as a GAP for G␣ 12 and serves as an effector in cells; however, G␣ 12 is incapable of activating p115RhoGEF in vitro (9). Because G␣ 12 and G␣ 13 have a high degree of identity at the primary sequence level (ϳ69%) and have nearly identical ␣ 3 helices and ␣ 3 -␤ 5 loops, it is worthwhile to consider biochemically whether G␣ 12 and G␣ 13 bind to the RH domain of p115RhoGEF using the same interface. When mutations corresponding to the G␣ 13 mutants described here were introduced into G␣ 12 , the pattern of Rho activation in SRE assays and binding to the RH domain of p115RhoGEF was similar to that observed with the G␣ 13 mutants (supplemental Fig.  3). This suggests that the binding site for G␣ 12 overlaps with that of G␣ 13 in the RH domain of p115RhoGEF and probably explains the observation that activated G␣ 12 can inhibit G␣ 13mediated activation of p115RhoGEF despite the fact that G␣ 12 cannot directly activate p115RhoGEF in vitro (9). However, these data imply that the inability of G␣ 12 to stimulate p115RhoGEF is not due to a difference in primary sequence, as had previously been suggested (23). Thus, the reasons behind this lack of activity remain unknown. However, as another RH-RhoGEF, LARG, needs to be tyrosine-phosphorylated in order to be responsive to G␣ 12 in vitro (40), this may be the case with p115RhoGEF as well.
Although the precise activation mechanism of p115RhoGEF remains to be fully elucidated, we have identified residues in G␣ 13 that play a role in regulating the nucleotide exchange 4 N. Hajicek and T. Kozasa, unpublished data. Oxygen and nitrogen atoms are colored red and blue, respectively. Hydrogen bonds are depicted as dashed lines. Residues that contribute to the GAP interface are labeled in gray, whereas hydrophobic residues that contribute to the effector interface by the RH domain are labeled in light blue. Residues in G␣ 13 that affect its capacity to stimulate the GEF activity of p115RhoGEF are labeled in red, and their bonding partners in the RH domain are labeled in purple. The conformationally flexible residue Phe 234 in G␣ 13 is labeled in black. Bottom, the effector interface formed by G␣ 13 and the RH domain of PDZ-RhoGEF (Protein Data Bank entry 3CX7) (45). The interface is depicted as a ribbon diagram, with G␣ 13 colored in olive green and the RH domain in light blue. Atoms and amino acid residues are labeled as in the top. activity of p115RhoGEF. However, given that p115RhoGEF activity is also reported to be regulated by subcellular localization (41)(42)(43) and post-translational modification (44), the activation mechanism in cells is likely to be complex. Ongoing efforts to crystallize G␣ 13 and G␣ 12 with p115RhoGEF constructs containing the RH domain and DH/PH domains will aid in directly addressing the presence of additional binding sites and will help to clarify the molecular mechanism by which heterotrimeric G proteins stimulate activation of this RhoGEF.