Mechanisms for Reversible Regulation between G 13 and Rho Exchange Factors*

The heterotrimeric G proteins, G 12 and G 13 , mediate signaling between G protein-coupled receptors and the monomeric GTPase, RhoA. One pathway for this modu-lation is direct stimulation by G (cid:1) 13 of p115 RhoGEF, an exchange factor for RhoA. The GTPase activity of both G (cid:1) 12 and G (cid:1) 13 is increased by the N terminus of p115 Rho guanine nucleotide exchange factor (GEF). This region has weak homology to the RGS box sequence of the classic regulators of G protein signaling (RGS), which act as GTPase-activating proteins (GAP) for G i and G q . Here, the RGS region of p115 RhoGEF is shown to be distinctly different in that sequences flanking the predicted “RGS box” region are required for both stable expression and GAP activity. Deletions in the N terminus of the protein eliminate GAP activity but retain substantial binding to G (cid:1) 13 and activation of RhoA ex- change activity by G (cid:1) 13 . In contrast, GTRAP48, a homo-log of p115 p115 baculovirus EE-tagged

Heterotrimeric G proteins mediate signals from seven transmembrane receptors to a wide array of effectors, including adenylyl cyclase, ion channels, phospholipases, and the exchange factor p115 RhoGEF (1). Every G protein is composed of a heterotrimer made up of ␣, ␤, and ␥ subunits. The four G protein families, G i , G q , G s , and G 12 , have been categorized by their sequence identity and the functional similarity of their ␣ subunits (2,3). The G 12 family contains two members, ␣ 12 and ␣ 13 (4), which have been implicated in cellular transformation (5), gastrulation of Drosophila melanogaster (6), vascular development in mice (7), and actin re-arrangement (8 -10). The cytoskeletal changes mediated by G 13 have been shown in several studies to require the monomeric Rho GTPases (10).
Both the heterotrimeric G proteins and monomeric Rho GT-Pases utilize the same basic cycle of regulation. The inactive proteins contain bound GDP. Activation is facilitated by guanine nucleotide exchange factors (GEFs) 1 that promote dissociation of GDP and subsequent binding of GTP to the G protein.
GTPase-activating proteins (GAPs) return the GTPase to the inactive state by accelerating hydrolysis of the terminal phosphate of the bound GTP (2,3,11).
Members of the RGS (regulators of G protein signaling) family of proteins can act as GAPs for heterotrimeric G proteins (12)(13)(14). RGS proteins that have been characterized to date function by increasing the intrinsic rate of GTP hydrolysis of G␣ subunits through allosteric binding (15). This mechanism was first worked out for RGS4, which acts upon G␣ i and G␣ q (16,17). The smallest region capable of accelerating GTPase activity is called the RGS box. This domain is characterized by strong primary sequence identity and structural similarity among the four RGS family members for which structures are known: RGS4 (18), GAIP (19), axin/conductin (20), and RGS9 (21).
LARG (22), PDZ RhoGEF (23), p115 RhoGEF (24), and GTRAP48 (25) are guanine nucleotide exchange factors for Rho; all share a highly conserved region that interacts specifically with G␣ 12 and/or G␣ 13 (22,23,25,26). This region invariantly lies N-terminal to the tandem Dbl homology (DH) (27) and Pleckstrin homology (PH) (28) domains found in all four proteins. Even though these N-terminal regions contain only weak sequence identity to a stereotypical RGS box, the structure of this region in p115 RhoGEF (aa 42-252) is similar to the folding pattern of other RGS boxes (29). The high sequence identity among these regions in the four RhoGEFs suggests structural identity, and this clear subfamily of RGS domains is subsequently referred to as the rgRGS (RhoGEF RGS) domain.
The best-studied member of this group, p115 RhoGEF, was initially isolated as a protein that tightly bound the nucleotide free form of RhoA and increased the nucleotide exchange rate of RhoA (24). The rgRGS domain within the N terminus of p115 RhoGEF was subsequently established as a GAP for G␣ 12 and G␣ 13 (26). Most interestingly, p115 RhoGEF is also an effector of G␣ 13 , which increases the activity of p115 RhoGEF as a guanine nucleotide exchange factor (GEF) for RhoA (1). Recently, GTRAP48 was found to be a GEF for RhoA that also binds G␣ 13 , but the functional implications of this interaction are not known (25).
In this study, the interaction of p115 RhoGEF and GTRAP48 with the G 12 family of heterotrimeric G proteins is more precisely defined. Deletion analysis of p115 RhoGEF provides evidence that regions outside of the apparent classic RGS box are required for accelerating the GTPase activity of G␣ 13 . A mechanism for stimulation of p115 RhoGEF by G␣ 13 is suggested by the determination of a second binding site for G␣ 13 in the tandem DH/PH domains of p115 RhoGEF. Finally, GTRAP48 was found to bind G␣ 13 and act as a weak GAP on the ␣ subunit, but its nucleotide exchange activity was not stimulated by G␣ 13 .

Plasmids and Viruses for Expression of Protein-
The cDNA encoding full-length p115 RhoGEF (24) was used for amplification of fragments of p115 RhoGEF by the polymerase chain reaction. All fragments were amplified with an N-terminal EcoRI site and a C-terminal HindIII site for cloning into pCMV5-myc (provided by Melanie Cobb), pGEX-KG (30), and a modified pTrc D expression vector as described previously (31). Briefly, the intervening sequence between the hexa-histidine tag (HIS) and the EcoRI restriction site of pTrc C (Invitrogen) was replaced with the amino acids Met-Gly-Ala. Fragments were transferred via EcoRI/XbaI sites from pCMV5 into pVL1392-EE (31), which contains an N-terminal EE tag (EYMPME) (24). Baculoviruses were produced through co-transfection of SF9 cells with pVL1392-EE vectors and BakPak6 that was digested with Bsu36I (CLONTECH). The N-terminal DNA fragments of p115 RhoGEF were amplified by PCR and cloned into pGEX-KG and pTrc D. The different pieces of p115 RhoGEF are named by the primers used for their amplification. All cDNA constructs were sequenced to confirm correct amplification and construction.
Expression and Purification of Proteins-All GTRAP48 and p115 RhoGEF proteins were expressed via baculovirus in cultured Spodoptera frugiperda (SF9) cells or in the transformed BL21(DE3) strain of Escherichia coli. Recombinant EE-tagged proteins were expressed in SF9 cells after infection with baculovirus. The expressed proteins were purified from lysates by affinity chromatography with anti-EE coupled Sepharose (BAbCO) as described elsewhere (32). N-terminal fragments of p115 RhoGEF and GTRAP48 were produced in E. coli as either HIS-tagged fusion proteins or as chimeras with GST (glutathione Stransferase). GST-tagged proteins were purified by chromatography using glutathione-Sepharose (Amersham Biosciences, Inc.) and solution A (25 mM NaHEPES, pH 7.5, 1 mM dithiothreitol, 50 mM NaCl) containing the protease inhibitors (2.5 g/ml leupeptin, 1 g/ml pepstatin A, 21 g/ml phenylmethylsulfonyl fluoride, 21 g/ml N ␣ -p-tosyl-L-lysine chloromethyl ketone, 21 g/ml tosylphenylalanyl chloromethyl ketone, and 21 g/ml N ␣ -p-tosyl-L-arginine methyl ester). HIS-tagged proteins were purified by isolation with Ni 2ϩ -nitrilotriacetic acid resin (Qiagen) in 25 mM NaHEPES, pH 7.5, 2.5 mM ␤-mercaptoethanol, 50 mM NaCl, and the protease inhibitors. G␣ 13 was prepared as described by Singer et al (33). Prenyl RhoA was co-expressed with GST-tagged guanine nucleotide dissociation inhibitor via Baculovirus in SF9 cells (32). Lysates were prepared by freezethawing cell pellets three times in Solution A containing protease inhibitors. After removing particulate material by centrifugation at 100,000 ϫ g, the cytosol was passed over a glutathione-Sepharose column and eluted with the same solution containing 1% cholate to separate RhoA from GST-guanine nucleotide dissociation inhibitor. Proteins were concentrated, and cholate was removed by dilution of samples with solution A containing protease inhibitors and final concentration via pressure filtration through an Amicon PM10 membrane.
Binding of Purified EE-tagged p115 RhoGEF and EE-tagged Pieces of p115 RhoGEF to G␣ 13 -G␣ 13 was activated by incubation with solu-tion B (25 mM NaHEPES, pH 7.5, 100 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol) containing AMF (30 M AlCl 3 , 5 mM MgCl 2 , and 10 mM NaF) and protease inhibitors for 10 min. Activated G␣ 13 (5 pmol) was mixed with the indicated concentrations of exchange factor, which had been pre-bound to anti-EE-Sepharose, and the final volume was adjusted to 200 l with solution B and AMF if indicated. Samples were incubated for 1 h on a rocking platform at 4°C. The beads were then pelleted in a microcentrifuge for 1 min and washed twice with solution B. The p115 RhoGEF proteins and G␣ 13 were eluted for 15 min at room temperature with 50 l of solution B containing EE peptide (EYMPME) at a final concentration of 100 g/ml. G␣ 13 in the supernatant was visualized by SDS-PAGE and by immunoblot analysis using the B860 antibody.
Binding of GST-tagged rgRGS Domains of p115 RhoGEF and GTRAP48 to G␣ 13 -The purified GST-tagged domains were bound to 20 l of glutathione-Sepharose (packed beads) and washed in solution B to remove unbound protein. Activated G␣ 13 (5 pmol, activated as described above) was added to the indicated amount of immobilized rgRGS domain in 200 l of solution B. After incubation on a rocking platform for 1 h at 4°C, the samples were washed three times with solution B. The beads were then boiled in SDS sample buffer, and the eluted G␣ 13 was visualized by SDS-PAGE and by immunoblot analysis using the B860 antibody.
Co-immunoprecipitations of G␣ 12 or G␣ 13 with myc-tagged p115 Rho-GEF or GTRAP48 Pieces-COS cells were grown to 80% confluency in 60-mm dishes. The cells were then transfected using the FuGENE transfection reagent (Hoffman-LaRoche) with pCMV5-myc vectors that expressed the different exchange factors. Cells were also co-transfected with pCMV5 plasmids expressing either G␣ 13 or G␣ 12 . Transfected cells were incubated at 37°C in 5% CO 2 for 24 h. Cells were then lysed in solution B containing 0.1% Triton X-100. Non-extracted material was removed from the lysates by centrifugation at 13,600 ϫ g for 5 min. The cleared lysates were then incubated with protein G-Sepharose in solution B for 30 min at 4°C to remove proteins that bound nonspecifically to the Sepharose resin. After removal of the resin, the samples were added to protein G-Sepharose coupled to a monoclonal antibody directed against the myc tag and further incubated for 1 h at 4°C on a rocking platform. The Sepharose beads were pelleted and washed three times with solution B. Proteins were eluted by boiling in SDS sample buffer and visualized by immunoblot analysis with specific antibodies against either the myc-tag, G␣ 12 , or G␣ 13 .
Binding of GTP␥S to RhoA-Binding of GTP␥S to 2 M RhoA was assayed at 30°C in 20 l of solution containing 20 M GTP␥S, [ 35 S]GTP␥S (200,000 cpm ), 50 mM NaHEPES, pH 7.5, 1 mM EDTA, 50 mM NaCl, 1 mM dithiothreitol, and 5 mM MgCl 2 . G␣ 13 was added to a final concentration of 100 nM after being activated for 10 min in binding buffer containing AMF. After mixing, assays were stopped at various times by the addition of 2 ml of filtration buffer (20 mM TrisCl, pH 8.0, 25 mM MgCl 2 , 100 mM NaCl), and the proteins were immediately collected by filtration through BA-85 filters (Intermountain Scientific). The amount of [ 35 S]GTP␥S bound to RhoA on each dried filter was determined by liquid scintillation spectrometry. The apparent rates of guanine nucleotide exchange for RhoA were examined at three to four concentrations to determine the average turnover rate for each p115 RhoGEF piece.

RESULTS
Mutants of p115 RhoGEF and GTRAP48 were used to identify the domains responsible for binding to G␣ 13 , stimulation of RhoA exchange activity by G␣ 13 , and acceleration of GTPase activity of G␣ 12 and G␣ 13  Expression of the p115 RhoGEF rgRGS Domains-Stable expression of the N terminus of p115 RhoGEF (aa 1-248) as a fusion protein with GST has been shown previously, and aa 45-161 were predicted to form an RGS box (26). However, subsequent deletion mutants of this N terminus identified two areas outside of the predicted RGS box that were necessary for stable expression of protein. Because p115 RhoGEF was successively deleted from the N terminus by 7 or 13 amino acids, the level of expression remained stable. Further deletion up to aa 17 or 21 resulted in fragments that did not express intact protein. This N-terminal region, which apparently interferes with protein stability, ends at or before amino acid 25, because a fragment of p115 RhoGEF encoded by aa 25-252 can be expressed at 100 -200 g/gram of E. coli (wet weight). The deletion of the first 41 amino acids resulted in a domain, p115 RhoGEF (aa 42-252), which was expressed at levels comparable to domains with intact N termini.
Disruption of the region lying C-terminal to the predicted RGS box that lies between amino acids 161 and 252 also affected expression of protein. Protein fragments starting with aa 1 and ending at aa 248 or 252 express extremely well in E. coli (3-4 mg/g of packed wet cells). However, p115 RhoGEF (aa 1-215) could only be expressed poorly at 30 -40 g/g (wet weight), and this was only possible in SF9 cells. Fragments of p115 RhoGEF with shorter C termini, which encoded either aa  or aa 42-170, showed little to no detectable expression in either bacteria or SF9 cells (data not shown).
Definition of the Minimal p115 RhoGEF rgRGS Domain Required for Acceleration of G␣ 13 GTPase-Several N-terminal fragments of p115 RhoGEF, which included aa 1-252, 6 -252, 13-252, 25-252, 42-252, and 1-215 (described in Figs. 1A and 3A), were tested for their ability to stimulate the GTPase activity of G␣ 13 . Single-turnover assays were utilized, which measure the release of [ 32 P]P i upon hydrolysis of [␥-32 P]GTP that had been pre-bound to G␣ 13 . Assays were performed as described by Singer et al (33), with the modifications outlined under "Materials and Methods." As shown previously for a GST fusion protein containing the first 246 amino acids of p115 RhoGEF (26), the p115 rgRGS (aa 1-252) stimulated the GTPase activity of G␣ 13 as well as the full-length exchange factor (Fig. 3B). However, a protein with further truncation at its C terminus, p115 rgRGS (aa 1-215), was a less effective stimulator of the GTPase activity of G␣ 13 . Removal of additional C-terminal residues resulted in unstable proteins that had no detectable stimulation of the GTPase activity of G␣ 13 (data not shown).
The effects of N-terminal deletion are shown in Fig. 3 (C-F). Removal of the first 5 or 12 N-terminal amino acids, p115 rgRGS (aa 6 -252) or p115 rgRGS (aa 13-252), respectively, did not alter GAP activities toward G␣ 13 (Fig. 3C). Removal of the first 17 or 21 N-terminal amino acids did not allow expression of protein domains as discussed previously. P115 rgRGS (aa 25-252) was expressed modestly as a viable domain. Although an active GAP, the potency of this construct was only about 0.1% that of the full-length rgRGS domain encoded within aa 1-252 (Fig. 3, D-F). Finally, deletion of the N terminus up to the predicted RGS box, p115 RhoGEF (aa 42-252), produced a fragment that expressed well (see above). Despite binding strongly to G␣ 13 (about 10 -20% as well as p115 rgRGS (aa 1-252), see Fig. 2B), this fragment had essentially no GAP activity toward G␣ 13 (Fig. 3, C-E). The hint of activity observed at 10 M (Fig. 3E) is similar to nonspecific effects of adding other control proteins and is not increased at higher concentrations of p115 rgRGS (aa 42-252).
Binding of GTRAP48 to G␣ 13 and G␣ 12 -GTRAP48 has been shown to bind to G␣ 13 (25), but the functional consequences and potential interaction with G␣ 12 are unknown. The domain arrangements of GTRAP48 and p115 RhoGEF are compared in Fig. 4A, and schematic descriptions of various constructs are shown. In a chimeric protein, N48C115, the N-terminal rgRGS region of p115 RhoGEF is replaced with the homologous region from GTRAP48. Examples of purified proteins that were expressed via these constructs with EE or GST tags are shown in Fig. 4B.
The ability of GTRAP48 to bind to G␣ 12 as well as G␣ 13 was assessed by immunoprecipitation after transient expression in COS cells. COS cells were transfected with either myc-tagged GTRAP48 or myc-tagged p115 RhoGEF and either constitutively active G␣ 13 (Q226L) or constitutively active G␣ 12 (Q229L). Both p115 RhoGEF and GTRAP48 bound to G␣ 13 in the presence of AlF 4 Ϫ . Although p115 RhoGEF also bound well to G␣ 12 , the interaction of this ␣ subunit with GTRAP48 is apparently weaker and was hard to detect (Fig. 5A, see "Discussion"). The dependence of association of these GTPase-deficient forms of G␣ 12 and G␣ 13 on AlF 4 Ϫ may seem surprising. However, this reflects both the multiple states of the ␣ subunits upon lysis of cells in GDP and slow conversion of G␣⅐GTP to G␣⅐GDP through slow hydrolysis or nucleotide exchange over the extensive timeframe required for the immunoprecipitation.
To define the regions in GTRAP48 responsible for binding G␣ 13 , two segments, which included sequences homologous to the rgRGS domain of p115 RhoGEF, were expressed and examined. The first segment encodes aa 1-539, which includes the PDZ-and proline-rich domains that precede the rgRGS domain (Fig. 4). The second construct encodes aa 285-495, which contains amino acids homologous to the N-terminal region of p115 RhoGEF (Fig. 4). Both constructs of GTRAP48 were found to preferentially bind the activated form of G␣ 13 (Fig. 5B).
GTRAP48 Is a Poor GAP for G␣ 13 -The ability of GTRAP48 to stimulate the GTPase activity of G␣ 13 in a single-turnover assay is shown in Fig. 6A. Both GTRAP48 and N48C115 (a chimera derived from GTRAP48 and p115 RhoGEF; Fig. 4A) displayed low but significant stimulation of the GTPase activity of G␣ 13 . However, their activities were substantially less than the stimulation obtained with an equivalent amount of p115 RhoGEF. The N-terminal fragments of GTRAP48, which contain its rgRGS domain, failed to stimulate the GTPase activity of G␣ 13 at the concentrations tested (Fig. 6B). Thus, by this in vitro measure, the rgRGS domain is only a poor GAP for G␣ 13 at best. This contrasts with the robust GAP activity of p115 rgRGS.
Identification of a Second Binding Site for G␣ 13 within p115 RhoGEF-The N-terminal 248 amino acids in p115 RhoGEF are known to interact with G␣ 12 and G␣ 13 and assumed to be important for stimulation of exchange activity by G␣ 13 (1,26). Interestingly, fragments of p115 RhoGEF that did not contain the rgRGS region, but did contain the DH and PH domains, also bound G␣ 12 and G␣ 13 (Fig. 7A). As expected, all pieces of p115 RhoGEF that contained the rgRGS domain, p115 wild type, p115 RhoGEF (aa 1-760), and p115 RhoGEF (aa 1-637) bound the activated form of G␣ 13 in these pull-down assays. Fragments representing the C-terminal tail (aa 760 -912) and the DH domain (aa 288 -637) did not bind G␣ 13 . A fragment encoding the PH domain of p115 RhoGEF (aa 637-760) also did not bind G␣ 13 (data not shown). However, a fragment of p115 FIG. 5. Binding of GTRAP48 to G␣ 13 and G␣ 12 . A, myc-tagged GTRAP48 or p115 RhoGEF were co-transfected into COS cells with either constitutively active G␣ 13 (Q226L) or G␣ 12 (Q229L). After expression and lysis of cells, the exchange factors were immunoprecipitated with antibodies specific for the myc tag in the presence or absence of aluminum fluoride as described under "Materials and Methods." Immunoprecipitates were separated by SDS-PAGE, and the relative amount of exchange factor was visualized by immunoblot (IB) analysis using an anti-myc antibody (top panel). The amount of G␣ 12 or G␣ 13 that co-immunoprecipitated with exchange factor was visualized with their respective specific polyclonal antibodies (bottom panel). B, binding of constructs encoding the rgRGS domain of GTRAP48 to G␣ 13 . 70 pmol of GST-tagged GTRAP48 (aa 289 -539), GTRAP48 rgRGS (aa 289 -495), or p115 rgRGS (aa 1-252) were immobilized on glutathione-Sepharose and incubated with 50 pmol of purified G␣ 13 in the presence or absence of aluminum fluoride. The relative amount of G␣ 13 bound to the rgRGS constructs after washing the Sepharose beads is shown by immunoblot analysis using B860 antisera. C, the relative binding affinities of p115 RhoGEF, p115 rgRGS (aa 1-252), GTRAP48, and GTRAP48 (aa 289 -539) for G␣ 13 were compared by incubating increasing concentrations of each piece with 20 pmol of immobilized GST-tagged p115 rgRGS (aa 1-252) and 10 pmol of G␣ 13 . The relative amounts of G␣ 13 bound to GST p115 RhoGEF (aa 1-252) after washing were detected by immunoblot analysis using B860 antisera. RhoGEF that consists essentially of the DH and PH domains (aa 288 -760), bound G␣ 13 almost as well as the full-length protein, and this binding showed a marked dependence on the activated state (AlF 4 Ϫ ) of G␣ 13 . In contrast, p115 RhoGEF (aa 288 -912), which contains the C terminus of p115 RhoGEF as well as the DH and PH domains, bound G␣ 13 with low avidity, and this association was not dependent on AlF 4 Ϫ . This could explain the failure to see binding of the p115 RhoGEF (aa 246 -912) protein to G␣ 13 in the previous study by Hart et al (1).
Binding of p115 RhoGEF pieces to G␣ 12 or G␣ 13 was also tested by transient co-expression in COS cells of the G protein and myc-tagged p115 RhoGEF wild type, p115 RhoGEF (aa 288 -912), p115 RhoGEF (aa 288 -760), or p115 RhoGEF (aa 288 -637) as indicated (Fig. 7B). Co-immunoprecipitations of the ␣ subunits with specific myc-tagged proteins support the in vitro data with purified proteins. G␣ 13 bound to proteins that contained the DH and PH domains, but lacked the rgRGS region. The same pattern was observed with binding of G␣ 12 . A weaker interaction with the G␣ subunits was also observed when the C terminus was present. In contrast to the in vitro measurement, there was little dependence on activation of the G proteins with AlF 4 Ϫ (Fig. 7B). 13 -The ability of different truncated p115 RhoGEF proteins to effect exchange of guanine nucleotide on RhoA was determined either in the presence or absence of G␣ 13 that had been activated with AlF 4

Functional Mapping of the Domains in p115 RhoGEF Required for Stimulation of Exchange Activity by G␣
Ϫ (Table I). Average rates of turnover for stimulation of RhoA were determined at a fixed concentration of RhoA and at various concentrations of exchange factor as described under "Materials and Methods." All of the constructs that contained the rgRGS region could be stimulated by G␣ 13 . Thus, removal of the first 41 N-terminal amino acids of p115 RhoGEF in p115 RhoGEF (aa 42-912) (described in Fig. 1A) or of the C terminus in p115 RhoGEF (aa 1-760) did not affect activation of RhoA exchange activity by G␣ 13 . P115 RhoGEF (aa 1-637), which encodes the rgRGS domain and the DH domain but lacks the PH domain and C terminus, has greatly reduced basal activity, but this activity was still stimulated by G␣ 13 . Removal of 32 amino acids (aa 252-288) that lie between the rgRGS domain and the DH domain and have a marked effect on basal activity (31) also did not reduce stimulation by G␣ 13 . In contrast, removal of the entire rgRGS domain in p115 RhoGEF (aa 288 -760) eliminated sensitivity to G␣ 13 (Table I).
GTRAP48 is also an exchange factor with specificity for RhoA (25). However, G␣ 13 did not stimulate RhoA exchange mediated by GTRAP48 (Fig. 8A). To test whether the rgRGS domain of GTRAP48 was capable of mediating stimulation by G␣ 13 , it was used to replace the native rgRGS domain of p115 RhoGEF (see N48C115 in Fig. 4A for details). Interestingly, the RhoA nucleotide exchange activity of the N48C115 chimera was activated 3-to 4-fold by G␣ 13 (Fig. 8B). This indicates that the GTRAP48 rgRGS domain can mimic this function of the p115 rgRGS domain in the context of the rest of the p115 RhoGEF molecule.

DISCUSSION
The rgRGS Domain of p115 RhoGEF Is Unique from the Classic RGS Proteins-RGS proteins were originally identified in genetic screens as negative regulators of G protein signaling (35)(36)(37). The majority of RGS proteins were subsequently cloned by degenerate PCR using primers based on the RGS boxes of these founder members (15). The lack of sequence identity between the RGS and rgRGS domains would explain why the latter proteins were not identified by strategies using PCR or homology searching of data bases. The elucidation of the N terminus of p115 RhoGEF as a GAP for G␣ 12 and G␣ 13 (26) led to the suggestion of potential structural relationships with the RGS family and initiated identification of the subfamily of highly homologous rgRGS domains in PDZ RhoGEF (KIAA0380) (23), LARG (KIAA0382) (22), and GTRAP48 (25).
A region, designated the RGS box (about 110 amino acids), has been shown to be sufficient for the GAP activity (38) of several members of the RGS family, including RGS4, GAIP, and RGS10. The studies reported here demonstrate that this is not the case with the p115 rgRGS domain. Secondary structure analysis of p115 RhoGEF suggested that amino acids 45-161 would likely comprise an RGS box (26). Initial observations FIG. 7. The DH/PH domain of p115 RhoGEF contains a second binding site for G␣ 13 . A, purified EE-tagged p115 RhoGEF and truncated pieces of p115 RhoGEF were bound to Sepharose beads that had been conjugated to anti-EE antibodies. After incubation with purified G␣ 13 (5 pmol), either in the presence or absence of aluminum fluoride (as indicated), the beads were washed and the relative amount of G␣ 13 that remained bound to immobilized p115 RhoGEF constructs was determined by immunoblot (IB) with B860 antisera after separation by SDS-PAGE. B, COS cells were transfected with myc-tagged p115 Rho-GEF pieces and G␣ 13 or G␣ 12 . The exchange factors were then immunoprecipitated using an antibody directed against the myc tag. The amount of exchange factor and any associated G␣ 12 or G␣ 13 13 The apparent turnover rate for each p115 RhoGEF construct is based on the measured number of moles of GTP␥S bound to RhoA. This was determined by measuring the amount of RhoA that bound [ 35 S]GTP␥S over time and at multiple concentrations of exchange factor as described under "Materials and Methods." The rates calculated from each time course were plotted against the amount of exchange factor used, and an average apparent turnover rate was determined by linear regression analysis. The rate measurements were made either in the presence or absence of 200 nM G␣ 13 , which was activated with aluminum fluoride. The last column indicates the -fold activation over basal activity affected by G␣ 13 . The basal rates of turnover for these proteins were reported previously (31). A coefficient of determination, R 2 , that measured the degree to which the derivatives of the apparent rates fit a linear regression model was 0.99 or greater except for the ␣ 13 -stimulated rate of p115 RhoGEF (aa 288 -760), which was 0.98. showed that aa 1-246 of p115 RhoGEF possessed GAP activity equivalent to the full-length enzyme (26). Reduction of the C-terminal end of this piece by 31 amino acids, p115 RhoGEF (aa 1-215), results in poor expression and reduced GAP activity. Further truncation of C-terminal residues results in proteins that express very poorly and have no measurable GAP activity (data not shown). Deletion of the N-terminal 25 amino acids of p115 RhoGEF reduced GAP activity over 99%. Removal of the N-terminal 41 amino acids still allowed expression of this domain but reduced binding to G␣ 13 by 80 -90% (Fig. 2B) and completely eliminated GAP activity (Fig. 3, C-F). Thus, an RGS region of p115 RhoGEF that retains some catalytic function requires ϳ200 residues (aa 26 -216); full function requires more.
These data indicate significant differences between this rgRGS domain and the classic RGS domains. First, the rgRGS domain requires an extended C terminus for stability and, perhaps, function. This is supported by the recent elucidation of crystallographic structures for the rgRGS domain of p115 RhoGEF (29) and the rgRGS domain of PDZ RhoGEF (39). Both of these rgRGS domains show similarity to the RGS box structure (15, 19 -21) in their core regions (aa 45-161 of p115 Rho-GEF) but also show close association of their core regions with C-terminal residues that form three ␣-helices and fold back onto the core. The capability of p115 RhoGEF (aa 42-252) to bind, but not act as a GAP on G␣ 13 , offers a clear dissociation of these two activities. In contrast, mutational analysis of classic RGS proteins indicates that reductions in GAP activity correlate much more strongly with decreases in binding avidity between the RGS and the targeted G protein ␣ subunit (40). Because RGS4 binds and allosterically stabilizes the transition state of the switch 1 and switch 2 domains of G␣ i , it is hypothesized that any reduction in binding energy would also reduce the degree of stabilization of the transitions state (15,40,41). The clear dissociation of binding and GAP activity in the p115 rgRGS indicates that it has a novel mechanism of accelerating the GTPase activity of G␣ 12 and G␣ 13 .
Comparison of the rgRGS Domains of GTRAP48 and p115 RhoGEF-GTRAP48, a recently characterized protein with an apparent rgRGS domain, was previously shown to bind G␣ 13 (25). Binding of GTRAP48 to G␣ 13 was confirmed in the current studies, but binding to G␣ 12 was not readily detected (Fig. 5). GTRAP48 could act as a very poor GAP for G␣ 12 (data not shown). Thus, GTRAP48 can interact with G␣ 12 , albeit weakly. The GAP activity of GTRAP48 for G␣ 13 was also poor and truncated proteins that contained its rgRGS region and could also bind G␣ 13 had little to no GAP activity (Fig. 6). Similar to GTRAP48, the rgRGS domain of PDZ RhoGEF also interacted with G␣ 12 and G␣ 13 and was a very poor GAP for G␣ 13 . 2 This latter phenotype contrasts with the rgRGS region of p115 Rho-GEF encoded within aa 1-252, which is as good as the fulllength p115 RhoGEF at activating the GTPase activity of G␣ 13 . The lack of identity between the highly negatively charged N-terminal region of p115 rgRGS (aa 26 -41) and the matching region of GTRAP48 (aa 314 -326) may provide one explanation for the observed low activity of GTRAP48. Removal of these amino acids eradicates the GAP activity in p115 RhoGEF. These findings then suggest that GTRAP48 is not a major GAP for the G 12 family. Alternatively, it is possible that another factor, yet to be identified, provides the functional equivalent of the N-terminal residues (aa 26 -41) in the p115 rgRGS to effect stimulation of GAP activity by GTRAP48.
The Mechanisms of p115 RhoGEF for GAP Activity on G␣ 13 and Mediation of Rho Exchange Activity by G␣ 13 Are Different-Initial experiments indicated that the N-terminal region of p115 RhoGEF, which encompasses the rgRGS domain, was sufficient for GAP activity (26). The inability of p115 RhoGEF (aa 288 -760) but not of p115 RhoGEF (aa 1-760) to be stimulated by G␣ 13 indicates that the first N-terminal 288 amino acids of p115 RhoGEF (which contains the rgRGS domain) are important for this process. What is the role of the GAP activity in this regulation? The observation that removal of the Nterminal 41 amino acids in p115 RhoGEF eradicates GAP activity, but has no effect on the ability of activated G␣ 13 to stimulate exchange activity, indicates both structural and functional divergence in these activities. Thus, the stimulation of GTPase activity has no impact on the mechanism for stimulation of exchange activity and should only impact rates of inactivation.
Interactions between G␣ 13 and Regions Outside the rgRGS Domain of p115 RhoGEF Play a Role in Activation of Exchange Activity by G␣ 12 -The abrogation of regulation upon removal of its rgRGS domain suggests three mechanisms by which the exchange activity of p115 RhoGEF is stimulated by G␣ 13 . One mechanism would utilize the interaction of G␣ 13 with the rgRGS region to alleviate an autoinhibitory action of the domain on exchange activity. This is most unlikely, because truncation of this rgRGS actually caused a reduction in basal exchange activity (31) rather than an increase that would be expected from removing an autoinhibitory constraint. The more rigorous characterization reported here contrasts with an initial observation that the activity of a p115 RhoGEF fragment lacking the N terminus (aa 246 -912) was greater than that of the full-length protein. One reason for this discrepancy is the use of prenylated RhoA in the current studies. Preny-2 T. Kozasa and P. Sternweis, unpublished data. lated RhoA is a much more potent substrate for p115 RhoGEF (32) than the non-prenylated GTPase used previously. P115 RhoGEF (aa 246 -912) exhibits less of an increase in activity toward the prenylated form of RhoA than wild type p115 Rho-GEF and, therefore, is less active than p115 RhoGEF in this context (data not shown). A second mechanism would use binding of G␣ 13 to the rgRGS domain to induce an allosteric mechanism for activation by which the rgRGS region causes higher nucleotide exchange activity of the DH domain on RhoA. In a third scenario, binding of the rgRGS domain to G␣ 13 helps stabilize interaction of G␣ 13 with another part of p115 RhoGEF to promote a higher activity state of the DH domain.
The discovery that G␣ 13 binds to a second region of p115 RhoGEF outside of the rgRGS domain suggests that the third mechanism is most likely. Because the rgRGS domain alone provides the same GAP activity as the full-length protein (Fig.  3), this second site of interaction is clearly not needed for this function and is more likely to play a key role in the stimulation of exchange activity. The location of this second site appears to be in the area of the DH domain. A construct composed essentially of the DH and PH domains, p115 RhoGEF (aa 288 -760), binds G␣ 13 better than to a comparable construct missing the PH domain, p115 RhoGEF (aa 288 -637) (Fig. 7, A and B), or to the PH domain alone (data no shown). However, the ability of G␣ 13 to effectively stimulate a truncated protein that contained the rgRGS and DH domains but lacked the PH domain, p115 RhoGEF (aa 1-637), suggests that this second site of interaction is still present. The higher affinity observed when the PH domain is present may be due to stabilization of the DH domain and preservation of higher affinity for G␣ 13 . The drastically reduced basal exchange activity of p115 RhoGEF (aa 1-637) shown previously (31) (and in Fig. 8A) and the apparent low affinity of constructs lacking the PH domain for RhoA (31) are consistent with this interpretation.
Although GTRAP48 has rgRGS and DH domains that are similar in sequence and arrangement to the corresponding domains in p115 RhoGEF, G␣ 13 does not stimulate the exchange activity of GTRAP48, in vitro. The physiological implication of this difference is not known. It is possible that the interaction of G␣ 13 with GTRAP48 could be stimulatory in vivo by mediating localization of the exchange factor or that another factor may be required to mediate a regulatory effect of G␣ 13 on GTRAP48. Alternatively, it is possible that G 13 does not regulate the activity of GTRAP48 in the cellular milieu. Attempts to directly answer this question by overexpression of the exchange factor and G␣ 13 have not yet yielded definitive results for two major reasons. Expression of GTRAP48 has been highly variable, especially when co-expressed with G␣ 13 (e.g. Fig. 5A,  compare lanes 1 and 2 with lanes 3 and 4). Furthermore, the expression of G␣ 13 alone gives robust activation of Rho and downstream events. The use of alternative and better controlled expression systems may eventually allow better analysis of this putative regulation.
The differential response of GTRAP48 and p115 RhoGEF to G␣ 13 may be explained by dissimilarities between the rgRGS or DH domains of these two exchange factors, because these domains in p115 RhoGEF are sufficient for activation of exchange activity by G␣ 13 . To address this question, the rgRGS domain of p115 RhoGEF was replaced with the rgRGS domain of GTRAP48 to make the N48C115 chimera. The ability of this chimera to be stimulated by G␣ 13 just like wild type p115 RhoGEF strongly argues that the rgRGS domains of these two proteins are functionally equivalent when mediating activation of exchange activity by G␣ 13 . Thus, differences within GTRAP48 and p115 RhoGEF that lie outside the rgRGS domain must explain their disparate responses to G␣ 13 . The rgRGS domain of p115 RhoGEF may, therefore, function to tether and/or position G␣ 13 so that it can directly interact with the DH domain of p115 RhoGEF. Alternatively, interaction with the rgRGS may induce conformational changes in G␣ 13 that yield an ability to stimulate exchange.