Mechanism of RGS4, a GTPase-activating Protein for G Protein α Subunits*

GTP hydrolysis by guanine nucleotide-binding proteins, an essential step in many biological processes, is stimulated by GTPase-activating proteins (GAPs). The mechanisms whereby GAPs stimulate GTP hydrolysis are unknown. We have used mutational, biochemical, and structural data to investigate how RGS4, a GAP for heterotrimeric G protein α subunits, stimulates GTP hydrolysis. Many of the residues of RGS4 that interact with Giα1 are important for GAP activity. Furthermore, optimal GAP activity appears to require the additive effects of interactions along the RGS4-Gα interface. GAP-defective RGS4 mutants invariably were defective in binding Gα subunits in their transition state; furthermore, the apparent strengths of GAP and binding defects were correlated. Thus, none of these residues of RGS4, including asparagine 128, the only residue positioned at the active site of Giα1, is required exclusively for catalyzing GTP hydrolysis. These results and structural data (Tesmer, J. G. G., Berman, D. M., Gilman, A. G., and Sprang, S. R. (1997) Cell 89, 251–261) indicate that RGS4 stimulates GTP hydrolysis primarily by stabilizing the transition state conformation of the switch regions of the G protein, favoring the transition state of the reactants. Therefore, although monomeric and heterotrimeric G proteins are related, their GAPs have evolved distinct mechanisms of action.

By cycling between active GTP-bound and inactive GDPbound states, guanine nucleotide-binding proteins control many biological processes, including translation, vesicular trafficking, cytoskeletal organization, and signal transduction (1)(2)(3). GTP hydrolysis is required to deactivate guanine nucleotide-binding proteins, as shown by the effects of hydrolysisresistant GTP analogs, or mutations and toxins that block this reaction. Because guanine nucleotide-binding proteins have slow intrinsic rates of GTP hydrolysis, they are acted upon by GTPase-activating proteins (GAPs), 1 achieving rates of GTP hydrolysis in the physiological range (4).
RGS (regulators of G protein signaling) proteins are GAPs for ␣ subunits of heterotrimeric guanine nucleotide-binding proteins (G proteins), principally G q and members of the G i family (5,6). RGS proteins, which are unrelated in primary sequence to GAPs that act on monomeric guanine nucleotidebinding proteins such as Ras, have been identified in the yeast Saccharomyces cerevisiae (Sst2p) (7)(8)(9), Aspergillus nidulans (FlbA) (10), Caenorhabditis elegans (Egl-10) (11) and in mammalian cells (at least 16 family members) (reviewed in Refs. 5 and 6). In addition to acting as GAPs, certain RGS family members are capable of inhibiting signaling by binding activated (GTP-bound) G ␣ subunits, antagonizing effector binding (12). Therefore, RGS family members are thought to govern the strength and duration of physiological responses triggered by an array of G protein-dependent signaling pathways.
The mechanisms whereby RGS proteins, or other GAPs, stimulate GTP hydrolysis are unknown. They could introduce residues into the active sites of guanine nucleotide-binding proteins, participating directly in catalyzing GTP hydrolysis as has been suggested by mutational studies of p120 GAP (13)(14)(15); however, proof of this mechanism requires determining the structure of Ras-p120 GAP complexes. Alternatively, RGS proteins or other GAPs could act allosterically by stabilizing the transition state structures of the "switch" regions of guanine nucleotide-binding proteins that change conformation during the GTPase cycle, stimulating intrinsic GTPase activity. Consistent with this mechanism, p120 GAP and certain RGS family members bind preferentially to the transition state conformations of Ras and G ␣ subunits, respectively (16 -18).
The structure of RGS4-G i␣1 complexes in the transition state has recently been solved to a resolution of 2.8 Å (19). The structure suggests that RGS4 could act allosterically and/or catalytically to stimulate GTP hydrolysis. An allosteric mechanism is suggested because binding of RGS4 stabilizes the switch regions of G i␣1 . However, a catalytic role is possible because an asparagine residue (Asn-128) of RGS4 interacts with an active-site glutamine residue (Gln-204) of G i␣1 that is thought to bind or polarize the attacking water molecule in the GTPase reaction. Furthermore, modeling studies suggest that asparagine 128 of RGS4 potentially binds a water molecule when it associates with GTP-bound G i␣1 , earlier in the GTPase reaction mechanism (19). Therefore, determining whether RGS4 acts allosterically or catalytically requires mutational and biochemical data that reveal which residues of RGS4 are required specifically to bind G ␣ subunits and/or catalyze GTP hydrolysis.

EXPERIMENTAL PROCEDURES
Mutagenesis of RGS4 -RGS4 was expressed in yeast from the ADH promoter using the polymerase chain reaction to clone the coding region of an RGS4 cDNA (rat (20); rat RGS4 was used for structural studies (19)) into pVT102U cut with BamHI and XbaI, creating pADH-RGS4. A C-terminally Myc-tagged form of RGS4 (which did not affect RGS4 function in yeast; data not shown) was generated by ligating an XbaI fragment encoding three in-frame copies of the c-Myc epitope (in pUC119; gift of D. Pellman, Whitehead Institute) with XbaI-cut pADH-RGS4, creating pADH-RGS4 -3xMyc. N-terminally His-tagged RGS4 * This work was supported by United States Public Health Service Training Grant AR07279-18 (to N. W.) and by National Institutes of Health Grant GM44592 and a grant from Monsanto (to K. J. B). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  was expressed in Escherichia coli using plasmids described previously (17). Point mutations in the RGS4 coding regions of yeast and E. coli expression plasmids were generated using the QuickChange mutagenesis kit (Stratagene). N-and C-terminal truncation mutations of RGS4 were generated by the polymerase chain reaction and cloned into pET15B (Novagen). The RGS4 coding regions of all constructs were sequenced to verify that only the desired mutations had been introduced. Pheromone response of yeast cells (BC180, an sst2 mutant) expressing wild-type and mutant forms of RGS4 was determined by performing quantitative growth arrest (halo) assays (20). Expression of Myc-tagged RGS4 in yeast cells was examined by performing immunoblot experiments using 9E10 monoclonal antibodies and enhanced chemiluminescence detection (Amersham Corp.).
Purification of Proteins, GTPase Measurements, and G Protein Binding Assays-N-terminally His-tagged forms of RGS4 (rat) and G o␣ were expressed in and purified from E. coli (BL21(DE3)) by Ni 2ϩ -nitrilotriacetic acid chromatography as described previously (17). Recombinant myristoylated G i␣2 (provided by M. Linder) was purified from E. coli as described previously (21). The activities of G proteins were assessed by determining the stoichiometry of [ 35 S]GTP␥S binding, which was Ͼ70%.
GTP hydrolysis by G ␣ subunits during a single catalytic turnover was determined as described previously (17). Briefly, His-tagged G o␣ (100 nM in 400 l) was incubated with [␥-32 P]GTP (0.1 M, 20,000 -30,000 cpm/pmol) in the absence of Mg 2ϩ ; stoichiometries of nucleotide binding were Ͼ70%. Aliquots (50 l) were removed 15 s before and 45 s after hydrolysis of GTP was initiated at 5°C by adding MgSO 4 (10 mM final concentration), unlabeled GTP (100 M final concentration), and either a buffer control or various amounts of wild-type or mutant forms of recombinant His-tagged RGS4. The amount of 32 P i released was determined by liquid scintillation spectrometry.
Binding of RGS4 to G i␣2 subunits was determined as described previously (17). Briefly, purified myristoylated G i␣2 (5 g) was incubated with GDP, GTP␥S, or GDP and AlF 4 Ϫ . Purified His-tagged wild-type or mutant forms of RGS4 (10 g) were added. After a 30-min incubation on ice, Ni 2ϩ -nitrilotriacetic acid beads were added, incubated for 30 min with agitation at 4°C, washed three times, and boiled in Laemmli sample buffer to elute bound proteins. Eluted proteins resolved by SDS-polyacrylamide gel electrophoresis were transferred to nitrocellulose; G i␣2 subunits were detected by immunoblotting with antiserum P960 (G ␣ -common antiserum) and peroxidase-labeled secondary antibodies and by enhanced chemiluminescence.
Graphics-Images were produced using RasMol and the coordinates of the transition state structure of RGS4-G i␣1 complexes (19).

RESULTS AND DISCUSSION
Identification of Mutations That Disrupt the GAP Activity of RGS4 -To determine the mechanism whereby RGS4 stimulates GTP hydrolysis by G protein ␣ subunits, we analyzed the effects of mutations affecting RGS4. If certain residues of RGS4 are required exclusively for catalysis, they should be dispensable for binding G ␣ subunits, similar to what has been shown for p120 GAP (13)(14)(15). Alternatively, if RGS4 catalyzes GTP hydrolysis strictly by binding and stabilizing the transition state of G ␣ subunits, then mutations that disrupt GAP activity should invariably cause a corresponding defect in G ␣ binding.
Because these experiments were initiated before the structure of RGS4-G i␣1 complexes was reported, our first objective was to define the minimal region of RGS4 (a 205-residue polypeptide) that possesses normal GAP activity in vitro. This domain could then be subjected to extensive point mutagenesis.
Analysis of purified forms of several truncation mutants indicated that the domain characteristic of RGS family members (RGS homology domain, residues 58 -177 of RGS4) had normal GAP activity toward G o␣ (Table I), similar to results obtained using ret-RGS-d (22). Truncations extending into the RGS homology domain resulted in the production of RGS4 in an insoluble form (Table I), indicating a defect in protein folding or stability. The RGS homology domain of RGS4 therefore appears to be a single functional domain.
To identify amino acids in RGS4 that are functionally important, we constructed a set of 43 point mutations (most of which were alanine substitutions) that affect many of the charged or conserved residues of its RGS homology domain (Fig. 1). As a means of screening rapidly for loss-of-function mutations, Cterminally Myc-tagged forms of the RGS4 mutants were tested for their ability to inhibit G protein-dependent signaling (mating pheromone response) when expressed in yeast (20) (see "Experimental Procedures"; results are summarized in Fig. 1). The results indicated that many of the mutants were as functional as wild-type RGS4, whereas other mutants were nonfunctional. Mutants showing an intermediate activity in this in vivo assay were not obtained, even though subsequent experiments (see below) revealed that many of the nonfunctional mutants were partially defective in GAP activity in vitro. This was expected because the in vitro assay is much more sensitive. None of the loss-of-function mutations significantly decreased expression of RGS4 in yeast, as indicated by immunoblotting (data not shown).
We subsequently determined whether these loss-of-function mutations affected the stability or GAP activity of recombinant RGS4 expressed in and purified from E. coli (Table I); as controls, several mutations that did not affect RGS4 function in yeast were also analyzed. Equivalent results were obtained using either G o␣ or G i␣2 as a substrate (data not shown). The results are interpreted below in light of the structure of RGS4-G i␣1 complexes (19). His-tagged forms of the indicated RGS4 mutants were assayed for GAP activity as described under "Experimental Procedures." Data are expressed relative to the values obtained using an excess (200 nM) of wild-type RGS4-(1-205) (0.6 -0.9 pmol P i released). Each RGS4 mutant was analyzed at least three times; standard errors are indicated in parentheses. Deletion mutants are indicated by the residues of RGS4 remaining. Point mutations were made in RGS4- . Those mutations that did not affect GAP activity were among those that did not affect RGS4 function in yeast. Mutations That Apparently Affect the Stability of the RGS Domain-The RGS homology domain of RGS4 forms a bundle of nine ␣-helices. The binding site for G i␣1 is a cleft consisting of conserved amino acids at the ends of helices 4, 7, and 8 and loops between helices 3 and 4 and helices 5 and 6. However, many other conserved amino acids in RGS4 are located distal to the G i␣1 -binding site. Some of these residues are important for the folding and/or stability of the RGS domain. These include a pair of phenylalanine residues (Phe-79 and Phe-168) apposed at the interface of helices 3 and 8; substituting either with alanine resulted in an insoluble protein when expressed in E. coli (data not shown). Other interacting pairs of residues may also maintain the stability or rigidity of the RGS fold because mutations affecting them resulted in soluble proteins with reduced GAP activity (Table I). These interacting residues include an isoleucine-phenylalanine pair (Ile-114 and Phe-149) apposed between helices 5 and 7, a pair of phenylalanine residues (Phe-91 and Phe-118) apposed between helices 4 and 5, and an isoleucine-tryptophan pair (Ile-67 and Trp-92) between helices 2 and 4. In contrast, two highly conserved serine residues (Ser-164 and Ser-171), which were predicted to stabilize the RGS fold by acting as helix breakers between helices 7, 8, and 9 (19), were not required for RGS4 function in vivo (Fig. 1).
Mutations Affecting Residues on the Surface of RGS4 That Interact with G i␣1 -In remarkable agreement with structural data, the remaining loss-of-function mutations affect residues on the surface of RGS4 that interact with G i␣1 ( Fig. 2A and Table I). As discussed below, this indicated that many of the interactions between RGS4 and G i␣1 are important for GAP activity ( Fig. 2A and Table I). Furthermore, this allowed us to investigate the relative functional importance of specific protein-protein contacts and the mechanism whereby RGS4 stimulates GTP hydrolysis. In describing these results, we have used the convention of Tesmer et al. (19), in which residues of RGS4 are preceded with "r" and those of G i␣1 with "a." Interactions occurring at the edge of the RGS4-G i␣1 interface are functionally important (Fig. 2B), as indicated by the effects of various alanine substitutions in RGS4. These include hydrophobic interactions between the side chains of r-Tyr-84 and a-His-213 and a salt bridge between r-Glu-87 and a-Lys-210. r-Glu-83 is also functionally important, probably because it interacts with the side chain of r-Arg-167 ( Fig. 2A). Potentially, this interaction pulls r-Glu-83 and the adjacent residue, r-Tyr-84, into place such that r-Tyr-84 interacts effectively with a-His-213. Alternatively, the side chain of r-Glu-83 may be important for GAP activity because it contacts the side chain of a-Val-185 (Fig. 2B).
In the center of the RGS4-G i␣1 interaction footprint, a-Thr-182 in switch I of the G protein binds a pocket in RGS4 consisting of several conserved residues, including r-Ser-85, r-Asn-88, r-Leu-159, r-Asp-163, r-Ser-164, and r-Arg-167 ( Fig. 2B) (19). With the exception of r-Ser-85 and r-Ser-164, the side chains of residues forming the pocket are functionally important because an alanine substitution at each site diminished the GAP activity of RGS4 (Table I).
Based on these results and the structural data of Tesmer et al. (19), we propose the following roles for residues forming the pocket that binds a-Thr-182. Three residues of the pocket are important because they interact directly with a-Thr-182. The bulky side chain of r-Leu-159 may determine the size or shape of the pocket because it interacts extensively with the side chain of a-Thr-182 (Fig. 2, B and C) and because the r-L159A substitution causes a severe defect in GAP activity. Furthermore, the side chain of r-Leu-159 may be of critical importance because it also participates in a hydrophobic interaction with the side chain of a-Lys-180 (Fig. 2, B and C). The side chains of r-Asn-88 and r-Asp-163 appear to be important because they form hydrogen bonds with the side chain hydroxyl and backbone nitrogen of a-Thr-182, respectively. Although r-Ser-85 and r-Ser-164 form part of the pocket that binds a-Thr-182, their side chain hydroxyl groups are unimportant because alanine substitutions at these sites did not affect RGS4 function in vivo (Fig. 1). However, this does not exclude the possibility that the ␤-carbon atoms of r-Ser-85 and/or r-Ser-164 that do interact with a-Thr-182 are functionally important.
Residues forming the pocket, but which do not interact extensively with a-Thr-182, are also functionally important, but probably for different reasons. r-Met-160, which lies at the floor of the pocket but does not contact a-Thr-182, may be important because its side chain interacts with that of r-Asn-88 (Fig. 2D), potentially orienting it toward a-Thr-182. Similarly, because r-Arg-167 does not contact a-Thr-182 extensively (their side chains are a minimum of 3.9 Å apart), its principal function may be to stabilize two parts of the RGS4 surface such that each can interact with G i␣1 . This is suggested because 1) the side chain of r-Arg-167 interacts with side chain carboxyl groups of r-Glu-83 and r-Asp-163 (Fig. 2B), both of which are important for GAP activity; 2) an alanine substitution of r-Arg-167 leads to a stronger defect than an alanine substitution affecting either r-Glu-83 or r-Asp-163 (Table I); and 3) changing r-Arg-167 to alanine, lysine, or histidine disrupted GAP activity (Table I).
There also was evidence that normal function of RGS4 requires the additive effects of two sites that bind different regions of G i␣1 . r-Glu-87 and r-Asn-88, which are adjacent to one another, are likely to function cooperatively because they in- teract with a-Lys-210 of switch II and a-Thr-182 of switch I, respectively (Fig. 2B). Consistent with this hypothesis, a double mutant in which r-Glu-87 and r-Asn-88 are changed to alanine had a much stronger defect in GAP activity than either single mutation (Table I).
Asparagine 128 of RGS4 Is Critical for GAP Activity-A cluster of residues in G i␣1 (including a-Lys-180 of switch I and a-Gln-204, a-Ser-206, and a-Glu-207 of switch II) cradles r-Asn-128 (Fig. 2, E and F). This structural arrangement has suggested that r-Asn-128 does the following: 1) facilitates GTP hydrolysis by orienting and/or polarizing the side chain of a-Gln-204, which interacts with the attacking water molecule in the transition state; 2) binds a water molecule early in the GTPase mechanism, before the transition state is reached; and/or 3) binds and stabilizes switches I and II (19). Indeed, we found that r-Asn-128 has a pivotal role because replacing it with alanine resulted in essentially a complete loss of GAP activity (Table I).
Effects of Mutations on Binding of RGS4 and G ␣ Subunits-To investigate the mechanism of RGS4-stimulated GTPase activity, we determined whether the panel of GAPdefective RGS4 mutants can bind G ␣ subunits in their active (GTP-bound) or transition state (GDP ϩ AlF 4 Ϫ ) conformations. Strikingly, we found that all of the GAP-defective RGS4 mu-tants were impaired in their ability to bind G ␣ subunits. RGS4 mutants lacking appreciable GAP activity, including r-N128A, were unable to bind activated (GTP-bound) G o␣ subunits because at high concentration (3 M), they failed to block the ability of wild-type RGS4 (10 nM) to stimulate GTP hydrolysis (Fig. 3A).
Similarly, RGS4 mutants displaying various defects in GAP activity were also defective in binding G i␣2 in its transition state (GDP ϩ AlF 4 Ϫ ) (Fig. 3B). The results of the binding assays, although somewhat qualitative, suggested that the severity of the defects in G ␣ binding and GAP activity were correlated. Mutations that severely reduced or eliminated GAP activity (E87A,N88A double mutant, I114D, N128A, L159A, and R167A) appeared to cause severe defects in binding (binding was not detected) (Fig. 3B, third row), whereas those that partially disrupted GAP activity caused less severe binding defects (binding was less efficient than with wild-type RGS4) (Fig. 3B, first and second rows; 10% of the bound material was analyzed for each mutant in the first row and 20% for those in the second and third rows). Because all of these mutations disrupted or impaired the binding of RGS4 to G ␣ subunits, none of the residues they affect, including r-Asn-128, is required exclusively to catalyze GTP hydrolysis.

FIG. 2. Functional map of the surface of RGS4 that binds G i␣1 .
A, effects of alanine substitutions affecting residues of RGS4 that interact with G i␣1 . Part of the surface of RGS4 is shown. Residues of RGS4 (spacefill) located within 4.5 Å of G i␣1 are colored, each according to the magnitude of the defect in GAP activity (see Table I and/or Fig. 1) caused by an alanine substitution: no effect (green), moderate defect (yellow), severe defect (orange), nearly complete defect (red), and not mutated (cyan) because the residue is not highly conserved or because an alanine substitution would be a conservative replacement. B, residues of G i␣1 that interact with the functionally important surface of RGS4. The orientation of RGS4 and color coding of its functionally important amino acids (spacefill) are the same as in A. Residues of G i␣1 (sticks) that interact with this surface of RGS4 are indicated (carbon atoms are gray; nitrogen, blue; and oxygen, red). C, interaction of leucine 159 of RGS4 with threonine 182 and lysine 180 of G i␣1 (spacefill; color coding of atoms is as in B). Switch I of G i␣1 (cyan ribbon) and helix 7 of RGS4 (white ribbon) are indicated. D, interaction of methionine 160 of RGS4 with asparagine 88 of RGS4, which interacts with threonine 182 of G i␣1 . Color coding of atoms (spacefill) is as in B; in addition, sulfur atoms are yellow. Switch I of G i␣1 (cyan ribbon) and helices 4 and 7 of RGS4 (white ribbons) are indicated. E, surface of G i␣1 (spacefill; the indicated residues are color-coded arbitrarily) that interacts with asparagine 128 of RGS4 (sticks; atoms are color-coded as in B). F, position of asparagine 128 of RGS4 near active-site residues of G i␣1 . Atoms of asparagine 128 of RGS4 (spacefill) are color-coded as in B. Residues of G i␣1 , GDP, and AlF 4 Ϫ are indicated (sticks; atoms are color-coded as in B, except that phosphorus is orange, aluminum is yellow, and fluorine is blue-gray). The backbones of RGS4 (white ribbon) and G i␣1 (cyan ribbon) are indicated. Nearly all the residues of G i␣1 that interact with RGS4 are conserved in the G ␣ subunits we have used for in vivo (Gpa1) and in vitro (G o␣ and G i␣2 ) assays of RGS4 function; an exception is that in Gpa1, a threonine residue occurs at the position equivalent to valine 185 of G i␣1 .
Our functional analysis of RGS4 mutants and the structural data of Tesmer et al. (19) lead to the following conclusions regarding the mechanism whereby RGS4 stimulates GTP hydrolysis by G ␣ subunits. First, a principal function of asparagine 128 of RGS4 is to bind and stabilize switches I and II of G ␣ subunits, contributing to the overall stability of the RGS4-G ␣ transition state complex. High affinity binding of RGS4 and G i␣1 may not require a hydrogen bond formed by the amide nitrogen of r-Asn-128 and the carbonyl oxygen of the side chain of a-Gln-204 because in several other RGS proteins, including GAIP, r-Asn-128 is replaced by a serine residue. Therefore, interactions between r-Asn-128 and a-Lys-180, a-Ser-206, or a-Glu-207 may be important for stabilizing RGS4-G ␣ complexes. Indeed, a-Glu-207 of G i␣1 , which is highly conserved in G ␣ subunits, is critical for RGS binding and RGS4-stimulated GTP hydrolysis (18,23), but it is dispensable for intrinsic GTPase activity and effector recognition.
Second, our finding that GAP-defective RGS4 mutants are defective in binding G ␣ subunits provides direct evidence in support of the hypothesis of Tesmer et al. (19) that RGS4 stimulates GTP hydrolysis primarily, if not exclusively, by binding and stabilizing the transition state conformation of G ␣ subunits. Apparently, the rigidity or stability of RGS4 is important for the mechanism because we found that pairs of interacting conserved residues distal to the G i␣1 -binding site of RGS4 are important for GAP activity. Furthermore, the mechanism appears to involve the additive effects of several binding interactions along the RGS4-G ␣ interface, as indicated by the analysis of double mutants. Thus, structural and functional data indicate that RGS4 stimulates GTP hydrolysis by binding the switch regions of G ␣ subunits, creating a rigid environment that favors the transition state of the reactants.
Finally, the mechanism used by RGS4 to stimulate GTP hydrolysis is different in certain respects from that used by p120 GAP . Binding of p120 GAP is believed to introduce one or two invariant arginine residues into the active site of Ras (13)(14)(15). One of these conserved arginine residues, which are required for GAP activity but are dispensable for binding of p120 GAP to Ras, is thought to stabilize the developing negative charge of the ␥-phosphate during GTP hydrolysis, a role subserved by a-Arg-178 in G i␣1 (24). Direct support of this conclusion has been provided recently from the crystal structure of p120 GAP bound to Ras (25). However, a common feature of p120 GAP and RGS4 is that they both stabilize the structures of the switch regions of their cognate GTP-binding proteins in the transition state (19,25,26). Therefore, although monomeric and heterotrimeric GTP-binding proteins are evolutionarily related, their attendant GAPs are distinct in terms of their sequences, structures, and mechanisms of stimulating GTP hydrolysis.