MAPPING THE G α 13 BINDING INTERFACE OF THE rgRGS DOMAIN OF p115RhoGEF*

focused on both the globular helical domain of the rgRGS and the N-terminal residues that precede the RGS-box. Results indicate that there is global resemblance between the interaction surface of the rgRGS domain with G α 13 and the interactions of and RGS9 for their G α substrates. However, there are distinct differences in the distribution of functionally critical residues between these structurally similar surfaces and an additional essential requirement for a cluster of negatively charged residues at the N-terminus of rgRGS. Placement of the same mutations in the context of the p115RhoGEF molecule produces the same deficiencies in GAP activity as observed with the rgRGS domain alone but show no attenuation of the regulation of Rho exchange activity by G α 13.


Introduction
The intrinsic guanosine triphosphatase (GTPase 1 ) activity of the α subunits of certain heterotrimeric guanine nucleotide-binding proteins (G proteins) can be stimulated by members of the Regulators of G protein Signaling (RGS) family (1)(2)(3). The catalytic activity of these GTPase Activating Proteins (GAPs) resides in an ~120-residue α−helical domain that corresponds to a consensus "RGS-box" defined by amino acid sequence similarity. RGS-box sequences show considerable diversity and can be subdivided into distinct homologous clusters (4), but all share conserved features recognizable in the three-dimensional structures of representative family members: RGS4 (5), GAIP (6) and RGS9 (7). p115RhoGEF, a Guanine nucleotide Exchange Factor (GEF) for the monomeric G protein RhoA (8,9), is the prototype of a family that includes LARG (10), Lsc (11), PDZRhoGEF (12) and GTRAP48 (13). These proteins contain regions in their amino-terminal halves with low sequence identity to RGS domains (14).
In p115RhoGEF, this region extends from residue 42 to 163. Recombinant p115RhoGEF and fusion proteins containing the first 246 (14) or 252 (15) amino terminal residues of 5 overexpressed as a soluble protein in Escherichia coli although as little as 50 additional C-terminal residues allowed expression of a functional domain in eukaryotic cells. In addition to an extended C-terminus, 25 residues that precede the RGS-box were also required for full GAP activity (15). We refer to the N-, and C-terminally extended RGSbox segment of p115RhoGEF and its homologs as the rgRGS (RhoGEF RGS) domain.
Two members of the p115RhoGEF family, GTRAP48 (13,15) and PDZRhoGEF 2 , bind to Gα13 but have little or no GAP activity; these form a distinct sequence subset with respect to both the RGS-box domain and the N-terminal segment.
Structural (5,7) and mutagenesis (16)(17)(18)(19) studies have defined three distinct regions of the RGS domain that interact with Gα and convey GAP activity. These correspond to two surface polypeptide turns that join helical segments, together with the surface of the C-terminal α helix of the RGS domain. These RGS elements directly contact the catalytic site of Gα, principally, the Switch I and Switch II segments that undergo conformational rearrangement upon hydrolysis of GTP (20). Together, these two segments in Gα subunits contain residues that participate directly in the catalytic mechanism of GTP hydrolysis, or bind the magnesium ion cofactor (21,22). The complex formed by GDP, Mg 2+ and AlF4 -, which promotes an activated state of Gα (23), also mimics the penta-coordinate transition state for GTP hydrolysis (21,22). RGS proteins have been shown to bind more strongly to the GDP•Mg 2+ •AlF4 -complexes of Gα subunits than to those formed with GTP analogs (24). This, together with structural by guest on March 24, 2020 http://www.jbc.org/ Downloaded from evidence, indicates that RGS proteins accelerate GTP hydrolysis by stabilizing a transition state-like conformation of Gα or destabilize the GTP-bound ground state (25).
The crystal structures of the amino-terminally truncated rgRGS domain of p115RhoGEF (26) and that of its homolog, PDZRhoGEF (27), were recently determined.
These domains were comprised of eleven α helices; the N-terminal seven helices form a bilobal fold similar to that in classic RGS domains. The remaining four helices pack against the latter to generate a single globular domain. The core of the rgRGS domain from p115RhoGEF, which encompasses two of the three segments involved in Gα binding, shows high structural similarity to the corresponding segments in RGS proteins of known structure. Accordingly, mutation of two p115RhoGEF residues that correspond, in RGS4, to side chains that interact with Switch I and Switch II, were found to diminish GAP activity (26). These observations led to the inference that the general features of the protein-protein interface observed in the published structures of RGS:Gα complexes are likely be preserved in the interaction of rgRGS with Gα13.
Although the N-terminus of p115RhoGEF is not present in the crystal structure, it is required for GAP activity (15), and structural modeling of the rgRGS:Gα13 complex suggests that amino terminal residues of the rgRGS domain might interact with the helical domain and switch regions of Gα13 (26).
Here, we describe an extensive mutagenic analysis performed to define the residues in the rgRGS domain that interact with Gα13 and convey GAP activity. These 7 studies focused on both the globular helical domain of the rgRGS and the N-terminal residues that precede the RGS-box. Results indicate that there is global resemblance between the interaction surface of the rgRGS domain with Gα13 and the interactions of RGS4 and RGS9 for their Gα substrates. However, there are distinct differences in the distribution of functionally critical residues between these structurally similar surfaces and an additional essential requirement for a cluster of negatively charged residues at the N-terminus of rgRGS. Placement of the same mutations in the context of the p115RhoGEF molecule produces the same deficiencies in GAP activity as observed with the rgRGS domain alone but show no attenuation of the regulation of Rho exchange activity by Gα13.

Materials and Methods
Expression Plasmids--Complementary DNA oligomers encoding single or double amino acid changes from the wild type p115RhoGEF gene were used to generate point mutants by the polymerase chain reaction (PCR). The mutagenesis template consisted of amino acids 1-252 of human p115RhoGEF (rgRGS domain) subcloned into the pGEX-KG vector. The sequences of primers used are available upon request. The QuickChange Kit from Stratagene was used in a four-step procedure to generate mutant cDNA. Plasmids encoding the desired mutations were selected and verified by sequencing. Selected mutated rgRGS DNAs were also subcloned into the full length p115RhoGEF gene in the pFastbac1 vector (Gibco), which has been modified to provide an N-terminal EE tag (EYMPME) (8) for production of baculovirus and subsequent expression in Spodoptera frugiperda (Sf9) cells.

Mutations of the rgRGS helical domain that affect GAP activity towards Gα13 --
Structural modeling of the rgRGS:Gα13 complex predicted that residues of loops L3, L5, helix α8 and loop L11 interact with the switch regions of Gα13 (Fig. 1A) (26). To test this hypothesis, we mutated residues located within this predicted interaction surface of the p115RhoGEF rgRGS domain and measured the GAP activity of the mutant domains towards Gα13. Unless stated otherwise, "p115rgRGS" refers to the protein fragment encompassing residues 1-252 of p115RhoGEF. Mutations are identified by the one letter code for the residue in wild type p115rgRGS, followed by the position of the residue in the amino acid sequence and the one letter code for the residue to which it was mutated (e.g., E71A has glutamic acid in position 71 mutated to an alanine). Residues chosen for mutagenesis ( Fig. 1B) fall into three categories. The first group includes residues predicted to form direct and specific contacts with the switch regions of Gα13, according to structural modeling studies of the rgRGS:Gα13 complex (Glu 71, Arg 111, 12 sequence between p115RhoGEF and two rgRGS-containing proteins with little or no GAP activity, GTRAP48 and PDZRhoGEF. None of the residues chosen for mutagenesis are involved in extensive packing interactions. Hence, alterations of these residues are therefore not expected to disrupt the tertiary structure of p115rgRGS.

Insert Figure 1
Initial experiments were conducted with constructs encompassing only the

Insert Figure 2
A summary of the results with the 21 mutant domains tested under the conditions described above is shown in Fig. 2C. Several mutations, including R111A,

13
E155K, E212A, E213A, K214A and K214E, had no effect on GAP activity. In contrast, three mutations, Q69A, F70A and D156A, appear to largely abolish GAP activity while others mutations had more modest effects (apparent decreases in GAP activity of 20 -60 % of the wild type domain). GAP activities of mutants designed to target specific Gα13 interaction sites are consistent with the involvement of the structurally conserved L3 and L5 loops, and also the less well-conserved helix, α8. However, the lack of effects by mutations in L11 from the C-terminal helical region does not support a role for this structural element in the GAP activity of the rgRGS domain.
The L3 loop of the rgRGS domain is a likely site for interaction with Gα13, yet its conformation differs from that of RGS4. The group 1 residue in this loop, Glu 71, is proposed to engage in electrostatic and van der Waals interactions with Lys 204 of Gα13 and is conserved among rgRGS proteins (Fig. 1B). Accordingly, mutants in which Glu 71 was substituted with either alanine or lysine have 50 % of the GAP activity towards Gα13 compared to the wild type rgRGS domain (Fig. 2C). Non-conservative mutations of the group 2 and group 3 residues in L3, Q69A and F70A, respectively, caused large reductions in GAP activity. Both residues are exposed to solvent, and are in close proximity to Switch I of Gα13 in the model of the rgRGS: Gα13 complex, although their interaction partners cannot be predicted. Gln 69 is absolutely conserved among known rgRGS proteins (Fig. 1B) whereas Phe 70 is conserved only in p115RhoGEF, Lsc and Structural and mutagenic studies of RGS proteins indicate that the residue corresponding to Asn 128 in RGS4 is important for GAP activity. A proline residue occupies the analagous position (residue 113 in p115RhoGEF) in the rgRGS domains with strong GAP activity, but is replaced by lysine in GTRAP48 (Fig. 1B). Mutation of Pro 113 in L5 to either an alanine or a lysine reduces GAP activity (Fig. 2C), but the effect is not as severe as that caused by mutations in the L3 or α8 regions. Other mutations of group 1 residues in L5, including R111A and P115G/P116G, had little effect on GAP activity (Fig. 2C).
Residues outside the α3-L5 core region of the rgRGS domain align poorly in tertiary structure with RGS4. Therefore, although α8 is a probable interaction partner for Switch I of Gα13, only one residue in this helix, Glu 155, could be included in group 1. Mutation of this residue had little effect on GAP activity. In contrast, mutation of one of the two neighboring group 2 residues, Arg 152, reduced GAP activity by 50 %. Arg 152 is replaced by glutamate in PDZRhoGEF and GTRAP48, two rgRGS proteins with very weak GAP activity towards Gα13. Substitution of Arg 152 with glutamate in p115rgRGS resulted in only a modest reduction in activity similar to replacement with alanine. Three charged group 3 residues were selected for mutagenesis in α8. Of these, substitution of Asp 156, a residue in α8 that is conserved among rgRGS proteins (Fig.   1B), severely reduced GAP activity. However, structure-based modeling reveals no residues in Gα13 poised to interact directly with Asp 156 in the rgRGS domain.
Mutation of other charged residues in α8, Glu 155, Ser 159 and Arg 161, had little effect.
To elucidate the extent to which certain mutations affect GAP activity of the rgRGS domain, we measured the quantity of GTP hydrolyzed by Gα13 within a two minute period as a function of rgRGS concentration (Fig. 2D). At sufficiently high concentration, all of the mutated rgRGS domains were capable of stimulating Gα13 to the extent observed for the wild type domain. Thus, the D156A, Q69A and F70A mutations do not substantially alter the efficacy of rgRGS as a GAP for Gα13, but strongly reduce the potency of the domain for this function. Even the most debilitated mutant, p115rgRGS (D156A), exhibited GAP activity that was comparable to that of the wild type protein at concentrations that were 100-1000 fold greater than Gα13 in the assay. % that of wild type, whereas mutations of residues 40-45 had no effect. Substitution of Ser 14 and Arg 15 did not reduce GAP activity appreciably. The concentrationdependence of GAP activity was also measured for p115rgRGS bearing single mutations in residues 27-31 (Fig. 4D). In contrast to mutations within the RGS-box region, mutation of these residues appeared to severely reduce efficacy as well as potency.

Insert Figure 4.
Mutations of residues required for rgRGS GAP activity also reduce affinity for Gα13 -- http://www.jbc.org/ Downloaded from that reduced the efficacy, but not the potency of p115rgRGS in the GAP assay, bound to Gα13•GDP•Mg 2+ •AlF4 -, albeit to a lesser extent than wild type p115rgRGS. Other mutants tested in the pull down assay, which showed only modest reduction in the GAP activity, retain their ability to bind Gα13•GDP•Mg 2+ •AlF4 -. Mutations that severely impair the GAP activity of rgRGS also strongly diminish its ability to bind to activated Gα13; in no case do we observe an inactive rgRGS mutant that retains its full ability to bind Gα13.

Activities of p115RhoGEF containing point mutations in the rgRGS domain --Several
of the single residue mutations characterized for the rgRGS domain were inserted into the holo-p115RhoGEF protein to determine their effect in this context. As shown in Fig.   6A, these mutations are as debilitating to the GAP activity of the holo-protein as they are to that of the rgRGS domain alone. Thus, mutations F31A and E27K caused almost total loss of GAP activity while three other mutations (D156A, D28A, and E29K) showed impaired potency but substantial efficacy at higher concentrations. As with the rgRGS domain, these mutations also reduced the affinity of the holo-protein for activated Gα13 (Fig. 6B). This indicates that the rgRGS domain confers the dominant elements required for binding of p115RhoGEF to Gα13.
In contrast to the impairment of binding and GAP activities, the point mutations described above did not affect the ability of activated Gα13 to stimulate the Rho by guest on March 24, 2020 http://www.jbc.org/ Downloaded from 19 exchange activity of p115RhoGEF (see Fig. 7 for examples). Among the mutants tested, stimulation of exchange activity in either the wild type or mutant proteins ranged from 3-5 fold with an EC50 for Gα13 between 5-15 nM. There is no apparent decline in potency that correlates with the observed effects on activities of the rgRGS domain.
Thus, while the rgRGS domain is required for regulation of Rho exchange by Gα13 (15), binding of the domain to the α subunit does not appear to be coupled to this effect.

Discussion
The rgRGS domains form a unique and distinct subgroup of the RGS family.
Although the rgRGS domains are, in three-dimensional structure, apparent homologs of the RGS domains, their sequences bear little similarity to the RGS consensus sequence (3). The mutagenesis experiments that we have conducted are consistent with the hypothesis that the Gα13 interaction surface of the RGS-box region in p115RhoGEF is roughly similar to that observed in the structures of RGS4 and RGS9 bound to their Gα substrates (5,7). However, the set of rgRGS residues that emerges from the analysis as "hot spots" for GAP activity are not all cognates of the functionally critical residues in RGS4 (16).
The segment extending from L3 to L5 is the most highly conserved structural feature common to the rgRGS and RGS folds. Less well conserved is the segment corresponding to α7-α8 in RGS4, which folds as a single helix, α8 in rgRGS. Residues The N-terminal 41 residues of p115RhoGEF include a hydrophobic and prolinerich sequence (residues 11-26) followed by a fifteen-residue segment (aa [27][28][29][30][31][32][33][34][35][36][37][38][39][40][41] containing nine acidic residues (Fig. 8A). Although the first 12 residues of p115RhoGEF were shown to be dispensable for GAP activity, deletion of the first 41 residues abolished this function (15). Here, we show that the electronegative cluster encompassed by residues 27-30 is crucial to GAP function. Indeed, substitution of any of the first three acidic residues in the cluster caused a greater than 99 % loss of GAP potency and efficacy. Mutation of the aromatic residue, Phe 31, which is adjacent to the acidic sequence, is equally deleterious. The only mutation within the RGS domain of rgRGS that causes an equivalent degree of impairment is that of Asp 156 in α8, which, of the residues mutated, has the least solvent-accessible (~25 Å 2 ) surface area in the structure of p115rgRGS. Thus, residues within the RGS-box, and a small cluster in the preceding N-terminal are both necessary for the GAP activity of rgRGS.

Insert Figure 8.
In the absence of structural information, we speculate that residues N-terminal to The rgRGS subfamily represented by GTRAP48 and PDZRhoGEF have little or no GAP activity towards Gα13. The absence of GAP activity for these two rgRGS proteins may stem from specific sequence differences in either the RGS-box or the preceding N-terminal sequence. The overall sequence similarity between p115RhoGEF and GTRAP48 is high (57 %) in the RGS-box region (Fig. 1B) and many of the substitutions are conservative. Three mutations, F70A, P113K and R152E, were designed to reflect non-conserved differences in sequence between p115RhoGEF and GTRAP48 in regions hypothesized to form contact sites with Gα13. Of these, F70A is sufficient to severely reduce catalytic efficiency and binding affinity of the rgRGS domain towards Gα13. In contrast, differences between the N-terminal sequences of p115RhoGEF and GTRAP48 are so great that it is difficult to align them. Although both active and inactive RhoGEFs possess a sequence consisting of three or four acidic residues preceding a phenylalanine (residues 27-31 in p115RhoGEF) or tyrosine, the position of this sequence relative to the N-terminus differ. Therefore, spatial mismatch of the N-terminal acidic cluster with respect to its potential binding site on Gα13, might explain the reduced GAP activity of GTRAP48.
All of the mutations that diminish the GAP activity of p115rgRGS also reduce its affinity for Gα13. A quantitative relationship between affinity and activity cannot be deduced from the data presented here, as is the case for RGS4 (18), however, no mutants capable of binding Gα13 are devoid of GAP activity. It therefore appears that the GAP activity of p115rgRGS arises from its ability to bind and stabilize a conformational state of Gα13 that is conducive to transition state formation. The mechanism by which rgRGS acts in this role may be in part elucidated by structural studies now in progress. The requirement for specific N-terminal residues outside of the The rgRGS domain contributes to the intrinsic nucleotide exchange activity of p115RhoGEF as demonstrated by the 60 % reduction of this activity when the domain is deleted (30). However the most notable effect that results from this truncation is a total loss of the ability of p115RhoGEF to be stimulated by Gα13 (15). One hypothesis is that it is the binding of the rgRGS to Gα13 that facilitates this regulation. This is consistent with the observation that the regulation of exchange activity by Gα13 was retained when the rgRGS region of GTRAP48 was substituted for the endogenous rgRGS domain of p115RhoGEF (15). Yet, as we show here, mutations that compromise the ability of the rgRGS domain to bind to Gα13 do not affect the susceptibility of p115RhoGEF to activation by Gα13. This finding is consistent with a model in which the rgRGS domain plays a structural or allosteric role, for example, by conferring stability upon conformational states of other segments of p115RhoGEF that are required for Gα13-mediated stimulation of GEF activity. The reduction of inherent GEF activity upon deletion of the rgRGS domain (30) is an evidence for direct intramolecular coupling between the rgRGS and DH/PH domains. This raises the intriguing possibility that the ability of the rgRGS domain to affect the efficacy of Gα13 might itself be subject to regulatory mechanisms such as covalent modification, which have yet to be discovered.
The data presented here also suggest that stimulation of p115RhoGEF by Gα13 results from interaction of the α subunit at a surface on p115RhoGEF different from the 25 rgRGS domain. Previous binding studies of Gα13 to p115RhoGEF detected a second site for interaction between residues 288 and 760, which includes the DH and PH domains (15). It is most likely the surface of Gα13 that binds the rgRGS domain and that which engages the second site are distinct. This could resemble the interaction demonstrated by Slep, et al. (7), that Gαt binds the γ subunit of cyclic GMP phosphodiesterase and RGS9 simultaneously at separate, yet structurally interacting interfaces (7). The mechanisms by which Gα13 interacts with regions beyond the rgRGS domain of p115RhoGEF, and exploitation the coupling between this and other domains of the molecule, remain to be explored.    (32). The surface of the protein presumed to interact with Gα 13 faces the viewer. Residues that were mutated are labeled. The surface corresponding to each solvent accessible residue is color coded, from light pink to deep red, in proportion to the loss of GAP activity as measured by %V 0 WT (Fig. 2C) upon mutation. The surfaces of residues that were not mutated, or where mutation had no significant affect on GAP activity (%V 0 WT > 80 %), are rendered in light gray. B, the corresponding surface representation of RGS4, color-coded according to GAP activity as in panel A, using relative activity data obtained for mutants of RGS4 reported by Srinivasa, et al. , Table 1 (16). The activity data obtained in the presence of 200 nM RGS4 was used to color-code this surface diagram. Only those solvent accessible residues mutated to alanine were color-coded in for this diagram. In both panels, circles labeled and color-coded according to the scheme used in Fig. 1A, enclose structural elements that form the RGS4:Gα i1 binding site, or the proposed p115rgRGS:Gα 13 binding site.  with Gα13 (gray except for switch segments in purple) was modeled as described (26).