Selective Uncoupling of RGS Action by a Single Point Mutation in the G Protein α-Subunit*

Heterotrimeric G proteins function as molecular relays, shuttling between cell surface receptors and intracellular effectors that propagate a signal. G protein signaling is governed by the rates of GTP binding (catalyzed by the receptor) and GTP hydrolysis. RGS proteins (regulators of G protein signaling) were identified as potent negative regulators of G protein signaling pathways in simple eukaryotes and are now known to act as GTPase-activating proteins (GAPs) for G protein α-subunits in vitro. It is not known, however, if Gα GAP activity is responsible for the regulatory action of RGS proteins in vivo. We describe here a Gα mutant in yeast (gpa1 sst ) that phenotypically mimics the loss of its cognate RGS protein (SST2). Thegpa1 sst mutant is resistant to an activated allele of SST2 in vivo and is unresponsive to RGS GAP activityin vitro. The analogous mutation in a mammalian Gqα is also resistant to RGS action in transfected cells. These mutants demonstrate that RGS proteins act through Gα and that RGS-GAP activity is responsible for their desensitizing activity in cells. The Gαsst mutant will be useful for uncoupling RGS-mediated regulation from other modes of signal regulation in whole cells and animals.

A wide variety of cellular signals (hormones, neurotransmitters, light, odors) act through a three component system composed of cell surface receptors, heterotrimeric G proteins, and effector proteins (1). The mating pheromones in yeast Saccharomyces cerevisiae act through receptors (STE2, STE3 gene products), a G protein ␣␤␥ heterotrimer (GPA1, STE4, STE18), and a mitogen-activated protein kinase signaling cascade that promotes cell division arrest and fusion (2). If mating is unsuccessful, however, the cells become refractory to pheromone stimulation and will eventually resume normal growth. RGS 1 proteins have recently been identified as a fourth component of the G protein signaling pathway (2,3). The founding member of the RGS family, called SST2, was identified in a genetic screen for negative regulators of the pheromone response pathway in yeast (4). Loss of function sst2 mutants render cells supersensitive to a pheromone stimulus and unable to recover from pheromone-induced growth arrest. Dominant gain-of-function alleles of SST2 have the opposite effect, rendering cells insensitive to pheromone stimulation (5). Further genetic and biochemical experiments revealed that Sst2 interacts directly with the G protein ␣-subunit, Gpa1 (6).
Behavioral genetic analyses in C. elegans uncovered a homologue of Sst2, called EGL-10 (7). egl-10 was shown to negatively regulate goa-1, which encodes the G␣ that mediates serotonindependent egg laying behavior. Two mammalian homologues, GAIP and RGS10, were identified by their interaction with G␣-subunits in a two-hybrid screen (8,9). An additional 15 mammalian members of the family were found by expression cloning, degenerate polymerase chain reaction, low stringency hybridization, and as expressed sequence tags (7)(8)(9)(10)(11). All of the RGS proteins share a conserved "RGS core domain" of ϳ120 amino acids, with Ͼ20% sequence identity across all species. Several RGS proteins have also been shown to attenuate G protein signaling in cultured cells (12)(13)(14)(15) and to partially substitute for the loss of SST2 expression in yeast (10,12,15).
RGS proteins were later shown to function as GTPase-activating proteins (GAPs) for G␣-subunits in vitro (9, 11, 16 -22). 2 These findings suggest that RGS proteins negatively regulate signaling via their physical association with G␣-subunits. By enhancing the rate of G␣ GTP hydrolysis, RGS proteins would shorten the lifetime of the active G protein species and arrest signaling.
Does RGS GAP activity account for the negative regulatory properties of these proteins in vivo? Proving this model would require that RGS knockout mutants, and G␣ mutants that disrupt RGS interaction, exhibit the same phenotype. RGS mutations have been obtained in yeast and nematodes, but not in mammals. Indeed, constructing knockout mutants in mammals will be complicated by the fact that there are so many closely related (and possibly redundant) RGS isoforms. An RGS-insensitive G␣ mutant has not been reported in any system.
Here, we report the identification and characterization of a yeast G␣ mutation that specifically disrupts Sst2 regulation in * This work was supported in part by Grant GM 55316 from the National Institutes of Health and by a grant from the American Cyanamid Company (to H. G. D.). 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 U.S.C. Section 1734 solely to indicate this fact.
‡ These authors contributed equally to this work. § Supported by Grant 95009550 from the American Heart Association.
¶ vivo and in vitro. An analogous mutation in G q ␣ is similarly insensitive to RGS action in cultured cells. An RGS-uncoupled G␣ mutation proves that G␣ is the primary target of RGS in cells. G␣ mutants of this type will be extremely useful for determining the overall contribution of RGS proteins to signal regulation in vivo.
Mutagenesis Screening and Pheromone Assays-YHD436 cells were mutagenized with ethyl methanesulfonate (Sigma) as described (28) and assayed for ␣-factor supersensitivity using a reporter gene (␤galactosidase) assay on nitrocellulose colony lifts (29). A genomic DNA library in YCp50 was used for complementation cloning (ATCC 77162). The halo assay was performed as described (29).
Sst2-GST/Gpa1 Binding, GTPase Assays, and Mammalian Cell Culture Assays-Sst2-GST binding experiments were performed as described (6)  (gpa1⌬, sst2⌬) cells expressing either GPA1 or gpa1 sst , and either wild type SST2 or vector alone, were plated and exposed to filter discs containing 10 g ␣-factor. Cells were allow to grow for 48 h on selective medium before being photographed. B, YGS5 (gpa1⌬) cells expressing either GPA1 or gpa1 sst , and either the dominant gain-of-function SST2-1 mutant or vector alone was plated as described in A. and hydrolysis assays, were performed as described previously (19). 2 Mammalian cell culture, transfections, and second messenger assays are described elsewhere (38). Data were analyzed using analysis of variance for a randomized block design, with log transformation, followed by pairwise comparisons employing the least significant difference method.

RESULTS AND DISCUSSION
In the course of a large scale genetic screen designed to identify new desensitization components in yeast, we isolated a novel allele of GPA1, designated gpa1 sst (for supersensitive allele of gpa1). Sequencing revealed a single missense mutation resulting in a Gly-to-Ser substitution at position 302. This glycine is conserved among G␣-subunit family members and is located in the first of three switch regions known to undergo a conformational change upon GTP hydrolysis (30 -36).
Since the gpa1 sst mutant mimics the loss of SST2, we investigated its ability to be regulated by SST2. First we compared the pheromone response in cells expressing GPA1 or gpa1 sst , using the growth inhibition ("halo") assay (29). In SST2 ϩ cells, the pheromone response through gpa1 sst was potentiated compared with GPA1. In the absence of SST2, however, the wild type and mutant forms of Gpa1 responded equally (Fig. 1A). In cells expressing SST2-1 (a dominant mutant that promotes pheromone desensitization), there was a striking difference between wild type and mutant GPA1 (Fig. 1B). Cells expressing GPA1 and SST2-1 exhibited a greatly attenuated response to pheromone, resulting in a "filled in" halo. However, cells containing SST2-1 and gpa1 sst responded no differently than cells containing gpa1 sst alone. Clearly, the Gly-to-Ser mutation blocks the negative regulatory effect of Sst2 in vivo. Since there is no functional difference between Gpa1 and Gpa1 sst in the absence of Sst2 expression, we conclude that Gpa1 sst is fully competent to transmit a pheromone signal and interacts normally with G␤␥ and the receptor.
Sst2 is thought to act by binding to Gpa1 and stimulating its GTPase activity. Therefore, genetic uncoupling of SST2 and gpa1 sst should accompany a physical and/or functional uncoupling of the two proteins. To test this, we purified each of the proteins and compared the ability of Gpa1 and Gpa1 sst to bind and hydrolyze GTP. We first measured the rate of [ 35 S]GTP␥S binding and steady state [␥-32 P]GTP hydrolysis and found no difference between the wild type and mutant forms of Gpa1 (Fig. 2, A and B). We then compared the ability of each protein to catalyze the rate-limiting hydrolytic step of the reaction, using a single turnover assay in the absence or presence of a purified RGS protein (GAIP, Fig. 2C). GAIP is functionally equivalent to Sst2, but is more stabile and can be purified in much larger quantities. 2 In the absence of RGS, the initial k cat of hydrolysis was ϳ0.006 min Ϫ1 for both wild type and mutant. With the addition of RGS, however, the rate of hydrolysis was greatly accelerated (at least 20-fold) for Gpa1 but not at all for Gpa1 sst . A more accurate determination of the RGS-stimulated GTPase rate could not be made, as the reaction was essentially complete at the first time point. Thus Gpa1 sst can bind and hydrolyze GTP normally, but is completely unresponsive to RGS GAP activity. These results are consistent with the in vivo experiments where, in the absence of SST2 expression, Gpa1 sst can signal as well as the wild type.
Despite the complete loss of GAP activity in vitro, Gpa1 sst does not equal the loss of SST2 in vivo. There are at least two possible explanations for this difference. First, Sst2 could regulate proteins other than Gpa1. Such interactions could involve the N-terminal 300 amino acid region of Sst2, a domain that is not found in any other RGS protein. A less likely alternative is that Gpa1 sst is weakly activated by RGS in vivo but not in vitro.
The x-ray crystal structure has recently been solved for RGS4 complexed with G i1 ␣ and GDP-AlF 4 Ϫ , a transition state mimic (37). The structure suggests a mechanism in which RGS promotes hydrolysis by stabilizing the transition state conformation of G␣. A key prediction of this model is that loss of GAP activity should accompany a loss of RGS binding. Accordingly, we compared the ability of mutant and wild type Gpa1 to bind Sst2 in vitro. Yeast lysates containing an Sst2-GST fusion protein were adsorbed onto glutathione-Sepharose and mixed with similarly prepared lysates containing Gpa1 or Gpa1 sst . The resin was washed at various NaCl concentrations, in the presence of either GDP or GDP-AlF 4 Ϫ (Fig. 3). At 50 mM NaCl, both the wild type and mutant were retained by Sst2-GST, but binding was not AlF 4 Ϫ -dependent and is therefore likely to be nonspecific or nonfunctional. At higher salt concentrations (150 -250 mM) Gpa1 was selectively retained when GDP-AlF 4 Ϫ was present, but Gpa1 sst did not bind at all. At very high concentrations (350 mM), neither protein was retained. Thus the inability of the RGS protein to act as a GAP for Gpa1 sst appears to be due to a weakened protein-protein interaction.
Since we have shown that Gpa1 sst can specifically block RGS action both in vivo and in vitro, we examined if a similar mutation in a mammalian G␣ would also block RGS action in cultured cells. We created a Gly 188 3 Ser mutation in G q ␣ (G q ␣ sst ) and examined its sensitivity to RGS7 in cells co-transfected with the 5HT 2c receptor. This particular combination of signaling proteins was chosen because they have overlapping expression patterns in the brain and are known to interact in cells (38). In cells expressing wild type G q ␣, 5-HT stimulation resulted in a typical calcium response (using Fura-2), which was attenuated ϳ40% by RGS7 expression (Fig. 4A). In cells expressing G q ␣ sst , however, the response was completely refractory to RGS7 (Fig. 4B). To confirm the results obtained by calcium release, measurements of agonist-induced [ 3 H]inositol FIG. 3. Sst2-GST binding to Gpa1 and Gpa1 sst . Yeast-expressed Sst2-GST was immobilized on glutathione-Sepharose, incubated with Gpa1 or Gpa1 sst containing lysates, and washed in NaCl (50 -350 mM) buffer as indicated. The presence of AlF 4 Ϫ in the binding and wash solutions is indicated (ϩ). Eluted protein was detected by immunoblotting with antibodies against Gpa1 or GST, as indicated. trisphosphate (IP 3 ) production were performed on the same cells. Co-expression of RGS7 with wild type G q ␣ reduced maximal IP 3 generation by ϳ30%, while co-expression with G q ␣ sst had no effect (Fig. 4C). Thus, like Gpa1 sst , G q ␣ sst effectively couples to receptor and G␤␥, yet is resistant to the effects of RGS regulation. Since signaling in this case is mediated by G␣, rather than G␤␥ (as it is in yeast), we can also conclude that effector coupling is unaltered by the Gly 188 3 Ser mutation.
To determine how the Gly to Ser substitution can disrupt G␣-RGS interactions in such a selective manner, we employed molecular modeling using the coordinates of the RGS4-G i1 ␣ crystal structure (37). The conserved Gly in G i1 ␣ (Gly 183 ) is located directly opposite Glu 83 in RGS4 at the binding interface (Fig. 5). Buried surface area in this region accounts for 120 Å 2 , or 22%, of the G␣ binding site. Substituting Gly with Ser would introduce a hydroxyl group less than 1 Å from the backbone FIG. 4. G q ␣-mediated calcium mobilization and IP 3 generation in transfected CHO K1 cells. A, cells were transfected with the 5-HT 2C receptor, wild-type G q ␣, and either RGS7 or vector alone. 5-HT stimulation of calcium mobilization was monitored using Fura-2 indicator dye. The results shown are the mean of four independent experiments. Bar graphs represent maximal 5-HT-induced calcium concentrations with (ϩ) or without (Ϫ) RGS7, after subtraction of values for mock-transfected cells. B, identical to A, except that G q ␣ sst was used in place of wild type G q ␣. C, 5-HT-dependent increase of IP 3 generation over unstimulated cells. The expression level of each protein was equivalent, as determined by immunoblotting and radioligand binding. * denotes a significant difference (p Ͻ 0.05). carbonyl of Glu 83 , an energetically unfavorable position both electrostatically and sterically (compare Fig. 5, B versus C). When mapped onto the crystal structure of G i1 ␣-␤␥ or G t ␣-␤␥, however, the same substitution shows no crowding at the ␤␥ binding interface, no interference with guanine nucleotide binding, and no effect on the conformational changes that occur during GTP hydrolysis. In a direct test of this model, the corresponding mutation in G i1 ␣ was shown to cause a Ͼ100fold reduction in RGS4 binding (by flow cytometry measurements) and Ͼ1000-fold reduction in GAP activity. 3 RGS proteins were first identified through genetic studies carried out in yeast (4,5). G proteins were identified as potential targets of RGS regulation, through enzymological studies carried out in mammals (9,19,21). With this first description of a mutation that selectively blocks G␣ interaction with RGS, their cellular target and mechanism of action are now firmly established. The next major challenge will be to disrupt G␣-RGS interactions in animals. Knockout mutations of the Ͼ20 RGS isoforms may be impractical, so G␣ sst mutants could be used instead to determine which signaling pathways are subject to RGS regulation and how desensitization of G proteins compares with desensitization of receptors. Thus we believe that G␣ sst mutants will prove useful for determining how RGS proteins regulate signaling, not just in cultured cells but also in whole animals.