Activation of Leukemia-associated RhoGEF by Gα13 with Significant Conformational Rearrangements in the Interface*

The transient protein-protein interactions induced by guanine nucleotide-dependent conformational changes of G proteins play central roles in G protein-coupled receptor-mediated signaling systems. Leukemia-associated RhoGEF (LARG), a guanine nucleotide exchange factor for Rho, contains an RGS homology (RH) domain and Dbl homology/pleckstrin homology (DH/PH) domains and acts both as a GTPase-activating protein (GAP) and an effector for Gα13. However, the molecular mechanism of LARG activation upon Gα13 binding is not yet well understood. In this study, we analyzed the Gα13-LARG interaction using cellular and biochemical methods, including a surface plasmon resonance (SPR) analysis. The results obtained using various LARG fragments demonstrated that active Gα13 interacts with LARG through the RH domain, DH/PH domains, and C-terminal region. However, an alanine substitution at the RH domain contact position in Gα13 resulted in a large decrease in affinity. Thermodynamic analysis revealed that binding of Gα13 proceeds with a large negative heat capacity change (ΔCp°), accompanied by a positive entropy change (ΔS°). These results likely indicate that the binding of Gα13 with the RH domain triggers conformational rearrangements between Gα13 and LARG burying an exposed hydrophobic surface to create a large complementary interface, which facilitates complex formation through both GAP and effector interfaces, and activates the RhoGEF. We propose that LARG activation is regulated by an induced-fit mechanism through the GAP interface of Gα13.

Heterotrimeric G proteins 3 serve as key molecular switches to transduce a large array of extracellular signals into cells by actively alternating their conformations between GDP-bound inactive and GTP-bound active forms. In the current model, the ligand-activated G protein-coupled receptors (GPCRs) catalyze the exchange of GDP for GTP on G␣ subunits (1). Upon activation, three switch regions in the G␣ subunit undergo significant conformational changes, followed by dissociation of the GTP-bound G␣ subunit from the G␤␥ subunits. Both G␣-GTP and free G␤␥ interact with diverse downstream effectors to transmit intracellular signals. The G␣ subunit hydrolyzes bound GTP to GDP by its intrinsic GTPase activity. This deactivation process is further accelerated by GTPase-activating proteins (GAPs) such as regulator of G protein signaling (RGS) proteins (2,3). G␣-GDP dissociates from effectors and re-associates with G␤␥ to terminate the signal.
Although this model explains the basic concept of G protein signaling, the molecular dynamics of interactions among GPCR, G protein, RGS protein, and effector during the signaling process is not well understood. It has been suggested that the GPCR signals are integrated into the intracellular signaling network at the level of G proteins (4). Accumulating evidence suggests that the G␣ subunit acts as the core of the signaling complex at the membrane, which is formed through the transient protein-protein interactions of multiple signaling components (5,6). Thus, the quantitative analysis of the dynamic molecular interactions in the GPCR signaling complex will be crucial to understanding various cellular processes. G␣ 12 and G␣ 13 subunits have been demonstrated to regulate the activity of Rho GTPase through RhoGEFs, which contain an N-terminal RGS homology domain (RH-RhoGEFs) (7)(8)(9)(10). RH-RhoGEFs, which consist of p115RhoGEF/Lsc, PDZ-Rho-GEF/GTRAP48, and LARG in mammalian species, directly link the activation of GPCRs by extracellular ligands to the regulation of Rho activity in cells (10 -14). All three RH-RhoGEFs contain an N-terminal RH domain, which specifically recognizes the active form of G␣ 12 or G␣ 13 and central DH/PH domains characteristic of GEFs for Rho GTPases. It has been demonstrated in vitro that LARG and p115RhoGEF serve as specific GAPs for G␣ 12/13 through their RH domains and also as their effectors to regulate Rho GTPase activation (11)(12)(13). A structural study has demonstrated that the interface of the RH domain of p115RhoGEFs and a G␣ 13/i1 chimera is different from that of the RGS domain of RGS4 and G␣ i1 (7). The N-terminal small element in the RH domain, which is required for GAP activity toward G␣ 13 , contacts the switch regions and the helical domain of the G␣ 13/i1 chimera. The core module of the p115RhoGEF RH domain binds to the region of G␣ 13/i1 , which is conventionally used for effector binding. These results suggest roles for the RH domain in the stimulation of GEF activity by G␣ 13 in addition to GAP activity. On the other hand, several studies have also indicated that regions outside of RH domain of RH-RhoGEFs, particularly the DH/PH domains, interact directly with activated G␣ 13 (11,14,15). In addition, we have demonstrated recently that p115RhoGEF interacts with distinct surfaces of G␣ 13 for the GAP reaction or GEF activity regulation (16). However, the molecular mechanism of LARG activation upon G␣ 13 binding is not clearly understood.
In this study, we have developed a quantitative method for the kinetic and thermodynamic analysis of G␣ 13 -effector interaction using surface plasmon resonance (SPR) with sensor chips on which G␣ 13 was immobilized. We examined the kinetics and thermodynamics of the G␣ 13 -LARG interaction and assessed LARG activation using both in vitro and cell-based approaches. We present evidence that, in addition to the interaction with the RH domain, the DH/PH domains and C-terminal region of LARG also interact with G␣ 13 to form the high affinity G␣ 13 -LARG complex and activate RhoGEF activity. We further propose that LARG adopts the active conformation using an induced-fit mechanism through association with the GAP interface of G␣ 13 . A similar mechanism may also be used with other G␣-effector interactions.
Expression and Purification of Proteins-FL-LARG, LARG-RDPC, LARG-DPC, LARG-RDP, LARG-DH/PH, LARG-RH, LARG-C, and LARG-PDZ constructs were subcloned into the pFastBacHT transfer vector with a six-histidine tag at the N terminus (Invitrogen), and their recombinant baculoviruses were generated. Recombinant G␣ 13 wild-type (G␣ 13 WT), His 6geranylgeranylated RhoA (gg-RhoA), and His 6 -LARG proteins were purified using the Sf9-baculovirus expression system as described before (17). His 6 -G␣ i/13 or -G␣ i/13 K204A chimeric protein, with the N-terminal coding region of rat G␣ i1 (residues 1-28), and the backbone of G␣ 13 WT or G␣ 13 K204A (residues 47-377) were purified from Sf9 cells as described (9). This substitution of the N-terminal helix of G␣ i1 for the corresponding region of G␣ 13 has been shown to facilitate the production of functional protein retaining the specific biochemical properties of wild-type G␣ 13 . For SPR assays, the His 6 tags of LARG, G␣ i/13 , and G␣ i/13 KA were removed by digestion with tobacco etch virus protease for 8 h at 4°C.
Kinetic binding analysis was performed at 15°C. Immobilized G␣ i/13 or G␣ i/13 KA on a biosensor chip, which was in an inactive form in the absence of AMF (30 M AlCl 3 /10 mM MgCl 2 /10 mM NaF), was activated by AMF in running buffer. Samples of serially diluted LARG proteins at five to six steps were injected over the inactivated or activated G␣ i/13 and G␣ i/13 KA surfaces using Running Buffer AMF (Ϫ) or Running Buffer AMF (ϩ) (10 mM Hepes, pH 7.4, 150 mM NaCl, 3 mM EDTA, 0.005% surfactant P20, 10 mM MgCl 2 , 10 M GDP, 1 mM dithiothreitol, 10 mM NaF, 30 M AlCl 3 ) and reference flow cells at a flow rate of 10 l/min for 1 min followed by 2-min dissociation using KINJECT mode. The sensor surfaces were regenerated by the injection of Regeneration Buffer (1 N NaCl, 1 mM MgCl 2 , 10 M GDP, 1 mM dithiothreitol) after each binding cycle. All data were corrected for nonspecific binding and buffer shifts by double subtracting binding responses collected from a blank reference cell and a buffer injection over a G␣ i/13 or G␣ i/13 KA immobilized flow cell. Data were fitted to a simul-taneous k a /k d , 1:1 binding, and Global fitting using the BIAevaluation program. Kinetic analyses were not performed when the responses were Ͻ1/10 compared with those with G␣ i/13 in the presence of AMF. The experiments were performed more than four times. Thermodynamic experiments were done at a range of temperatures from 7°C to 15°C in 2°increments. Six different concentrations of LARG proteins were used for each temperature.
Thermodynamic Data Analysis-At an equilibrium state, the thermodynamic parameters, ⌬H°(enthalpy change), ⌬S°( entropy change), ⌬G°(free energy change), and ⌬Cp°(heat capacity change) were derived from the van't Hoff equation, which states that ⌬G°ϭ ⌬H°Ϫ T⌬S°ϭ RTlnK D , by measuring the temperature dependence of K D (equilibrium constant). Such parameters determined by the van't Hoff method are demonstrated to agree with those determined directly using calorimetry (18 -20). If ⌬H and ⌬S have significant temperature dependence and ⌬Cp was assumed to be independent of temperature over a short temperature range, the relationship between the parameters becomes, with T 0 defined as the reference temperature (25°C). The thermodynamic parameters in this study were calculated by fitting the data to Equation 1 using BiacoreT100 software.
For thermodynamic analysis of a transition state, the Eyring theory was employed (21), where the ‡ symbol denotes the transition state, k is the association or dissociation constant, k B is the Boltzmann constant, and is the Planck constant. According to Equation 2, the thermodynamic parameters at association and dissociation were calculated by linear fitting by measuring k a and k d at several temperatures using BiacoreT100 software.
Binding of LARG Proteins with His-G␣ i/13 Immobilized on Ni-NTA-Agarose-COS1 cells were transfected using the Lipofectamine PLUS reagent with the indicated LARG constructs: myc-LARG-DPC (8 g) and myc-RH (4 g). Transfected cells were incubated at 37°C in 10% CO 2 for 24 h. Cells were lysed with 650 ml of 1.25 ϫ lysis buffer B (50 mM Tris-HCl, pH 7.5/ 150 mM NaCl/1% Nonidet P-40/0.5 mM EDTA/9.8 mM 2-mercaptoethanol/10 mM ␤-glycerophosphate/10 mM Na 3 VO 4 /10 mM MgCl 2 /100 M ATP/30 M GDP/protease inhibitors) and centrifuged at 100,000 ϫ g for 30 min. Equal amounts of supernatants were separated and incubated with 1.4 g of His-G␣ i/13 immobilized on 5 l of Ni-NTA-agarose beads in 1 ϫ lysis buffer B with or without AMF for 60 min at 4°C. The mixtures were washed three times 1 ϫ lysis buffer B with or without AMF. The relative amount of LARG fragments bound to the activated-or deactivated-G␣ i/13 was visualized by immunoblot analysis using anti-Myc antibody.
Statistical Analysis-Data represent the mean Ϯ S.E. of at least three independent experiments. Statistical significance of the difference was assigned based on results of Student's t test (Figs. 2C and 3D), or analysis of variance followed by Fisher's protected least significant difference test (Fig. 3b) (*, p Ͻ 0.05). Actual n and p values are provided in each figure legend.
Other Procedures-Co-immunoprecipitations, SRE-luciferase assays, subcellular localization of LARG, measurement of Rho activation in cells, cell rounding assays, and in vitro Rho-GEF assays were performed as described before (9,13). Details of these procedures are described in the supplemental information.

RESULTS
Direct Interaction of LARG with G␣ 13 through Its RH Domain, DH/PH Domains, and C-terminal Region-We first examined the ability of several deletion constructs of LARG to interact with G␣ 13 by co-immunoprecipitation using COS1 cells that were transiently transfected with myc-tagged LARG mutants and G␣ 13 (Fig. 1, A and B). As previously reported, LARG-RDPC and LARG-RH, which contain the RH domain, preferentially bound to AMF-activated G␣ 13 , whose conformation resembles that of the transition state for GTP hydrolysis (15,22). Intriguingly, G␣ 13 also co-immunoprecipitated with LARG constructs without the RH domain: LARG-DPC, LARG-DH/PH, and LARG-C. Although the amount of co-immunoprecipitated G␣ 13 was much lower than with LARG-RDPC or LARG-RH, the interaction with these LARG fragments was still dependent on activation of G␣ 13 . We reproducibly observed a small but significant increase in the amount of precipitated G␣ 13 with LARG-C in response to G␣ 13 activation. These results suggested that LARG might also interact with activated G␣ 13 through the DH/PH domains and/or C-terminal region.
We thus examined the direct interaction between LARG and G␣ 13 using an SPR method. We have recently constructed a G␣ i/13 chimera by substituting the N-terminal ␣1 helix of G␣ i1 for the corresponding region of G␣ 13 and demonstrated that this G␣ i/13 protein purified by an Sf9-baculovirus expression system retains the same biochemical properties as wild-type G␣ 13 for binding and hydrolysis of guanine nucleotides, activation of RH-RhoGEFs, and GAP response (9) (supplemental Fig. S1).
For the SPR analysis, G␣ i/13 was immobilized on the biosensor chip by amine coupling. The interaction of LARG fragments with G␣ i/13 on the sensor chip was analyzed by Biacore3000. An advantage of this method is that the affinity difference of an effector for the active or inactive G␣ 13 can be quantitatively analyzed by the addition or exclusion of AMF in buffer on the same G␣ i/13 chip. For the kinetic analysis, the concentrations of LARG fragments as analytes were different for each construct. Thus, it should be noted that we cannot evaluate the binding affinity between different LARG proteins by comparing the amplitudes of the SPR responses.
Consistent with the cellular assays, the SRP assay clearly demonstrated that LARG-RH, LARG-DH/PH, and LARG-C preferentially interact with the active form of G␣ i/13 ( Fig. 1C and Table 1). Because FL-LARG was sensitive to degradation during the assay, we exploited LARG-RDPC instead of FL-LARG for further analysis. LARG-RDPC showed kinetic parameters similar to FL-LARG in the presence of ALF 4 Ϫ . For The affinity of LARG-RDP or LARG-RDPC for activated G␣ i/13 was ϳ2-fold or 20-fold higher than the RH domain alone, respectively (Table 1). This suggests that the DH/PH domains and the C-terminal region of LARG may contribute to forming the high affinity G␣ i/13 -LARG complex together with the RH domain. In particular, the C-terminal region of LARG greatly decreased the rate of dissociation of the G␣ i/13 -LARG complex. In the presence of the C-terminal region, k d was reduced by Ͼ5-fold (Table 1).
We have previously demonstrated that the mutation of lysine 204 to alanine in the switch I region of G␣ 13 significantly reduces its affinity for LARG and its sensitivity to the GAP activity of LARG (8,9,15). To further characterize the role of the RH domain in the G␣ 13 -LARG interaction, we compared the binding kinetics of LARG fragments with G␣ i/13 KA and G␣ i/13 , which are immobilized on separate channels of the same biosensor chip ( Fig. 1C and Table 1). Consistent with our previous result, the binding of LARG-RH to G␣ i/13 was abolished by the K204A mutation. In contrast, the interaction of LARG-DH/PH or LARG-DPC with G␣ i/13 was not affected by the K204A mutation (Table 1). These results indicate the existence of an interaction between G␣ 13 and the DH/PH domains. It is also important to note that the KA mutation in G␣ 13 significantly reduced the affinity for LARG fragments containing the RH domain but did not change the affinity for LARG-DH/PH and LARG-DPC. Thus, the initial association of the RH domain with G␣ 13 might be necessary to induce the proper conformational change in LARG to form a high affinity complex with G␣ 13 .
Stimulation of the RhoGEF Activity of LARG through Direct Interaction between the DH/PH Domains and G␣ 13 -We next examined whether the interaction of G␣ 13 with the DH/PH domains of LARG induces RhoGEF activation. First, we coexpressed LARG mutants lacking the RH domain with a constitutively active, GTPase-deficient mutant of G␣ 13 (G␣ 13 Q226L) in HeLa cells and assayed SRF activation of the cell lysates as an indicator of Rho activation. The SRF activity of cell lysates expressing LARG-DH/PH or LARG-DPC was further stimulated in the presence of G␣ 13 QL similar to LARG-RDPC ( Fig. 2A). The immunoblot of the transfected cell lysates showed similar expression levels of LARG constructs (Fig. 2D). However, the co-expression of G␣ 13 WT didn't synergistically potentiate the SRF activity stimulated by LARG-DPC or LARG-RDPC (supplemental Fig. S2). We also examined the amount of GTP-bound Rho in cells using GST-RBD pulldown assays (13,23). Consistent with Fig. 2A, GTP-bound Rho in cells expressing LARG-DH/PH, LARG-DPC, or LARG-RDPC increased with the co-expression of G␣ 13 QL (Fig. 2B).
We then examined the stimulation of the GEF activity of the DH/PH domains by in vitro reconstitution assays using recombinant wild-type G␣ i/13 (G␣ 13 WT) and LARG proteins (13). AMF-activated G␣ i/13 WT stimulated the RhoGEF activity of LARG-RDPC (Fig. 2C). However, in contrast to the results from cellular assays, G␣ i/13 WT did not significantly stimulate the activity of LARG-DH/PH (Fig. 2, B and C). The lower binding affinity of LARG-DH/PH for G␣ i/13 compared with LARG-DPC (K D : 4.3 ϫ 10 Ϫ7 M versus 7.6 ϫ 10 Ϫ8 M, respectively) may be responsible for this discrepancy.
The effect of co-expression of G␣ 13 QL on the subcellular distribution of LARG mutants was also examined under the same conditions as for the SRF assays in Fig. 2 (A and D). All of the LARG constructs were distributed mainly in cytosolic fractions in the absence of G␣ 13 QL. In the presence of G␣ 13 QL, LARG-RDPC, the only construct with a RH domain, was redistributed to the membrane fraction. LARG-RH was recruited to the membrane fraction in the presence of G␣ 13 QL, but not G␣ 13 WT or G␣ 13 KA (supplemental Fig. S3). These data suggest that the interaction of LARG-RH with activated G␣ 13 is likely to be important for inducing membrane distribution of LARG similar to the case of p115RhoGEF (5). Thus, it is suggested that G␣ 13 may regulate LARG activity through two distinct mechanisms, by inducing its membrane translocation through the RH domain and by stimulating its GEF activity through the DH/PH domains.
To confirm these results in a more physiological context, we performed cell rounding assays as an indicator of Rho activa- a ND, the kinetic analysis was not done because the responses were less than 1/10 compared to those with G␣ i/13 in the presence of ALF Ϫ .

Regulation of LARG by G␣ 13
tion (24). MDCKII cells were co-transfected with myc-tagged RH-RhoGEF constructs and G␣ 13 QL. In the presence of G␣ 13 QL, LARG-DH/PH or LARG-DPC induced cell rounding to a level similar to that observed with LARG-RDPC ( Fig. 2E and supplemental Fig. S4). We further examined the role of lysine 204 in stimulating the GEF activity of LARG constructs. Unexpectedly, the GEF activity of LARG-RDPC was still stimulated by ALF 4 Ϫ -activated G␣ i/13 KA in the reconstitution assay (Fig. 2F). These results further support the existence of a direct interaction of G␣ 13 with the DH/PH domains of LARG for Rho-GEF activation.

Functional Roles of the RH Domain and C-terminal Region in RhoGEF Activation Mediated by the DH/PH Domains-We
next examined the effect of the isolated RH domain or C-terminal region on the G␣ 13 -stimulated RhoGEF activity of DH/PH domains. GST-RBD pulldown assays demonstrated that co-expression of LARG-RH, but not p115RhoGEF-RH, potentiated the RhoGEF activity of LARG-DH/PH stimulated by G␣ 13 QL (Fig. 3A). In the reconstitution assay, the G␣/13stimulated RhoGEF activity of LARG-DH/PH was also significantly potentiated by the addition of LARG-RH to the reaction (Fig. 3B). However, RhoGEF activity of LARG-DH/PH was not stimulated by G␣ 13 without the RH domain even at a higher concentration (120 nM). These results and the higher affinity of LARG-RDP versus LARG-RH or LARG-DH/PH for G␣ i/13 (Table 1) suggest that, even though the effect of these separated domains is less than that of the combined molecule, the RH-G␣ 13 interaction increases the affinity of the DH/PH domains for G␣ 13 .
We then examined the binding of Myc-tagged LARG-RH and LARG-DPC expressed in COS1 cells to His-G␣ i/13 immobilized on Ni-NTA-agarose resin. The binding of LARG-DPC to AlF 4 Ϫ -activated G␣ i/13 was dependent on the presence of LARG-RH (Fig. 3C). In addition, although the amount of bound LARG-DPC was much lower than that of LARG-RH, the presence of LARG-DPC and LARG-RH enhanced their binding to G␣ i/13 . With G␣ i/13 KA-resin, reduced binding of LARG-RH without LARG-DPC was observed, which is consistent with the results of the SPR analysis (Table 1). These results suggest that the RH domain and DH/PH domains simultaneously bind to G␣ 13 through separate interfaces and that the binding of the RH domain with G␣ 13 would enhance the affinity of the DH/PH domains for G␣ 13 .
We also investigated how the interaction of the C-terminal region of LARG with G␣ 13 affects the RhoGEF activity using in vitro reconstitution assays (Fig. 3D). The addition of LARG-C fragment did not potentiate the GEF activity of LARG-RDP stimulated by G␣ i/13 . Nevertheless, the RhoGEF activity of  FEBRUARY 20, 2009 • VOLUME 284 • NUMBER 8 LARG-RDPC was potentiated 2.5-fold compared with that of LARG-RDP, suggesting that the interaction of the C-terminal region with G␣ 13 might also be involved in the conformational change of the LARG molecule to activate the RhoGEF activity.

Regulation of LARG by G␣ 13
In summary, these results strongly indicate that the multiple interactions of LARG with G␣ 13 through its RH domain, DH/PH domains, and the C-terminal region coordinate together to form the active G␣ 13 -LARG signaling complex. The association of the RH domain with the G␣ 13 surface, including Lys-204 likely induces the conformational change of the DH/PH domains of LARG necessary to stimulate RhoGEF activity.
Thermodynamic Analysis of the Interaction between G␣ 13 and LARG-To further elucidate the dynamics of the conformational changes induced by G␣ 13 -LARG binding, especially the contribution of the RH domain or the DH/PH domains of LARG to the interaction with G␣ 13 , we determined the thermodynamic parameters of LARG-RH/G␣ 13 and LARG-RDP/G␣ 13 interactions by SPR analysis. The SPR analysis at several different temperatures from 7°C to 15°C was performed with a chip on which G␣ i/13 was immobilized using BiacoreT100 as described under "Experimental Procedures." Thermodynamic parameter (free energy G°, enthalpy H°, entropy S°, or heat capacity Cp°) changes at equilibrium state were calculated from the temperature dependence of K D using the van't Hoff equation ( Table 2). The thermodynamic parameters at equilibrium estimated from SPR analysis have been consistent with those determined directly using isothermal titration calorimetry (18 -20). The parameters for the transition state of the interactions were calculated by the Eyring theory using the temperature dependence of association rate constant (k a ) or dissociation rate constant (k d ) as described under "Experimental Procedures" (21). The temperature dependence of K D , k a , and k d from SPR analysis fit well with the theoretical equations ( Fig. 4A and supplemental  Fig. S5A).
The LARG-RDP/G␣ 13 interaction showed more negative free energy change (⌬G°) than the LARG-RH/G␣ 13 interaction,  Thermodynamic parameters of the interactions were derived from the van't Hoff plots and Eyring plots in Fig. 4 and supplemental Fig. S5.
which is consistent with the higher affinity of LARG-RDP versus LARG-RH (Table 2). In addition, thermodynamic parameters demonstrated the essential differences between the G␣ 13 -RH interaction and G␣ 13 -RDP interaction ( Table 2). The binding of LARG-RH to G␣ 13 is enthalpy-driven and entropically unfavorable. In contrast, the binding of LARG-RDP to G␣ 13 is less favorable than that of LARG-RH from the enthalpy term but is entropy-driven with a positive ⌬S°. In general, a large negative heat capacity change (⌬Cp°) in a protein-protein interaction indicates the removal of water-accessible hydrophobic surface area coupled to conformational changes (25,26). ⌬Cp°of 1 kJ mol Ϫ1 K Ϫ1 corresponds to the removal of ϳ990 Å 2 of exposed hydrophobic surface area (25,26). Thus, a large negative ⌬Cp°of Ϫ3.8 kJ mol Ϫ1 K Ϫ1 together with the positive ⌬S°(0.084 kJ mol Ϫ1 K Ϫ1 ) suggests that Ͼ3500 Å 2 of hydrophobic surface is buried in the interface between G␣ 13 and the region between the RH domain and the DH/PH domains upon complex formation. Such extensive hydrophobic interactions would certainly stabilize the G␣ 13 -LARG complex. Furthermore, the free energy changes at the association and dissociation phases calculated by the Eyring theory were suitable for the equation ⌬G°ϭ ⌬G° ‡ ass Ϫ ⌬G° ‡ diss . A free energy barrier at the transition state of association between G␣ 13 and LARG-RDP was calculated as ⌬G° ‡ ass of 36 kJ mol Ϫ1 . The interaction between G␣ 13 and LARG-RDP will be driven by this binding free energy (Fig. 4B).

Dynamics of the Interaction between G␣ 13 and LARG with
Large Conformational Rearrangements at the Interface-In this study, we have characterized the molecular mechanism of G␣ 13 -LARG interaction by cellular and biochemical assays, including quantitative SPR analysis using a biosensor chip on which G␣ 13 was immobilized. Our results from the kinetic and thermodynamic analysis demonstrate that the simultaneous interaction of G␣ 13 through the RH domain, DH/PH domains, and C-terminal region of LARG facilitates the formation of the high affinity active G␣ 13 /LARG complex. Furthermore, a large hydrophobic surface spanning Ͼ3500 Å 2 is created at the interface between G␣ 13 and the region between the RH domain and the DH/PH domains. It has been demonstrated by Spolar and Record (26) that the formation of a such large hydrophobic surface at the interface cannot be accounted for by association between pre-existing surfaces, but rather is coupled to conformational rearrangements between the proteins, most likely by induced fit. Thus, it appears unlikely that the binding of G␣ 13 to the RH domain uncovers a large pre-existing surface on the DH/PH domains that contacts G␣ 13 via hydrophobic interactions. The K204A mutation at the RH domain contact position in G␣ 13 resulted in large decreases in affinity for LARG fragments containing the RH domain ( Fig. 1C and Table 1). This fact strongly supports the notion that lysine 204 is in the interface between G␣ 13 and LARG-RH and could be a hotspot for triggering significant conformational rearrangements in the binding site, which means that the interaction is most likely driven by an induced-fit mechanism. These results indicate that the interaction between G␣ 13 and LARG is induced through the binding of G␣ 13 with the RH domain and proceeds by burying an exposed hydrophobic surface to create a large complementary interface. Thus, both GAP and effector interfaces facilitate complex formation and activation of LARG (Figs. 4 and 5). A structural study has indicated that the N-terminal small region of the DH domain of LARG is necessary to exert GEF activity for Rho (27). Interestingly, this region has been shown to act as an activation "switch" for another RhoGEF, Vav. Therefore, it is possible that the association of the RH domain with G␣ 13 driven by an induced-fit mechanism directly or indirectly affects this region, and enables LARG to adopt the active conformation. Further x-ray crystal structure analysis of LARG or its complex with G␣ 13 will be necessary to clarify these points.
We have recently reported that p115RhoGEF interacts with G␣ 13 through separate surfaces for the GAP reaction and GEF activity regulation (16). In agreement with this finding, the K204A mutation does not affect the affinity of LARG-DH/PH or -DPC for G␣ i/13 (Fig. 1C) FIGURE 4. Thermodynamic analysis of G␣ 13 -LARG interaction. A, thermodynamic analysis of the G␣ 13 -LARG complex formation and dissociation through its RH and DH domain. van't Hoff plots and Eyring plots of the experimental data are shown. The thermodynamic parameters at an equilibrium state and at a transition state were estimated from van't Hoff plots and Eyring plots as described under "Experimental Procedures" ( Table 2 and supplemental Fig. S3). B, schematic reaction profile of the thermodynamic energies at the different states of G␣ 13 -LARG interaction. activity in reconstitution assays (Fig. 2F) strongly suggesting the existence of an additional interface between G␣ 13 and the DH/PH domains. Compared with a lock and key interaction, which proceeds via pre-existing complementary surfaces such as seen in the antigen-antibody interaction, the induced-fit mechanism will provide G␣ subunits with much more flexibility to interact with various downstream effectors. Similar to this finding, it was demonstrated recently that the G␤␥ subunit interacts with a variety of downstream effectors using homologous interfaces containing hot spots (29 -32). It is thus possible that other G␣-effector interactions may be regulated by a similar mechanism.
Our result using ALF 4 Ϫ -activated G␣ 13 represents the transition state of GTP hydrolysis. We have found that LARG-DH/PH has a higher affinity for G␣ 13 QL, which mimics the GTP-bound state, than for ALF 4 Ϫ -activated G␣ 13 (data not shown). In contrast, LARG-RH demonstrated the opposite affinity as previously demonstrated. The discrepancy between activation of LARG-DH/PH by G␣ 13 in cell-based assays and reconstitution assays may result from the affinity difference of LARG-DH/PH for G␣ 13 QL and ALF 4 Ϫ -activated G␣ 13 . G␣/GAP/Effector Signaling Complex-We have demonstrated that the interaction of G␣ 13 with LARG through the RH domain (a GAP interface) and the DH/PH domains (an effector interface) could coordinate together to stimulate the RhoGEF activity of LARG (Fig. 3, A and B). The spatial and kinetic connection of GAP and GEF activities within a LARG molecule may help to regulate amplification of G protein signaling, increase the temporal resolution of the response, and increase the specificity of the signal output. Because G␣ 12/13 have characteristically slow rates of nucleotide exchange and GTP hydrolysis, this mechanism could be a rational system for G␣ 12/13 signaling to regulate multiple important cellular functions with fast responses as well as long term processes (33,34). Cooperative regulation of GAP and catalytic activities in one single effector molecule may also apply to other G␣ effectors such as PLC␤ (35).
The simultaneous interaction of GAP and effector with activated G␣ has been clearly demonstrated with the crystal structure of the G␣ t -RGS9-PDE␥ complex (36). The structure explains the mechanism by which PDE␥ potentiates the GAP activity of RGS9-1: by increasing the affinity of RGS9-1 for G␣ t to enhance the fidelity and temporal resolution of visual signal transduction (37,38). Furthermore, the crystal structures of G␣ q -p63RhoGEF-RhoA and G␣ q -GRK2-G␤␥, where the switch I region appears available for the simultaneous binding of RGS proteins, suggest that G␣ can bind an effector and a GAP at the same time (8,39). Thus, simultaneous and nonoverlapping binding of a GAP and an effector with G␣ may be a conserved mechanism to help efficiently regulate G proteinmediated signaling output.
It is also interesting to note that the G␣ 13 -RH interaction through the GAP interface may confer specificity on the G␣ 13 -LARG interaction through its effector interface (Fig. 3A). The result is analogous to the selective potentiation of the GAP activity of RGS9 for G␣ t by PDE␥ (40 -42). The interaction of the RH domain with G␣ 13 outside of the switch regions might contribute to determining the specificity of GEF activation by G␣ 13 (7,9). These results suggest that RGS proteins may play another important role in GPCR signaling pathways by specifying the effector molecule for G␣. We also demonstrated that activation of G␣ 13 contributes to the subcellular redistribution of LARG from the cytosol to the plasma membrane through the interaction with its RH domain (Fig. 2D), which is consistent with the case of p115RhoGEF (5). This translocation will help LARG closely interact with its target, RhoA at the membrane.
Role of the C-terminal Region of LARG-Our kinetic studies indicate that the C-terminal region of LARG significantly affects the interaction with G␣ i/13 , especially the dissociation of LARG from G␣ 13 (Fig. 1C and Table 1), and regulates the G␣ 13stimulated RhoGEF activity of LARG (Fig. 3D). Our preliminary thermodynamic analysis of the LARG-RDPC/G␣ 13 interaction could not produce an appropriate fitting to the theoretical curve, which suggests that the C-terminal region might be unstable, and some additional interacting protein may be required to stabilize the conformation upon binding to G␣ 13 (supplemental Fig. 3B). Recently, the presence of homo-or hetero-oligomerization of RH-RhoGEFs through their C-terminal regions has been reported (43)(44)(45). However, the physiological function of this oligomerization process in cells has not yet been clearly understood. Further study is required to understand the functional role of the C-terminal region in the regulation of G␣ 13 -LARG signaling.

CONCLUSION
In this study we have characterized the molecular dynamics of G␣ 13 -effector interaction by applying both kinetic and thermodynamic analyses. The G␣ 12/13 -RhoGEF-Rho signaling pathway has been shown to participate in a variety of disease conditions, such as leukemia, cancer invasion, and hypertension (28,46,47). Detailed characterization of the mechanism of regulation of RH-RhoGEFs by G␣ 13 as demonstrated in this study will contribute to the screening of novel drugs to control these diseases. X-ray crystallography structures of G␣ 12/13 and the complex with target molecules has made progress toward understanding G␣ 12/13 -mediated signaling pathways (7,9). However, it is essential to understand the molecular dynamics of the system to provide an active view of the protein-protein interactions during the signaling process. The combined efforts of the dynamic-interaction analysis such as SPR presented in this study together with the high resolution x-ray crystallography approach will be critical to further understand the molecular mechanism of this G␣ 12/13 -mediated signaling pathway.