The C-terminal tail preceding the CAAX box of a yeast G protein gamma subunit is dispensable for receptor-mediated G protein activation in vivo.

The gamma subunits of heterotrimeric G proteins are required for receptor-G protein coupling. The C-terminal domains of Ggamma subunits can contact receptors and influence the efficiency of receptor-G protein coupling in vitro. However, it is unknown whether receptor interaction with the C terminus of Ggamma is required for signaling in vivo. To address this question, we cloned Ggamma homologs with diverged C-terminal sequences from five species of budding yeast. Each Ggamma homolog functionally replaced the Ggamma subunit of the yeast Saccharomyces cerevisiae (STE18 gene product). Mutagenesis of S. cerevisiae Ste18 likewise indicated that specific C-terminal sequence motifs are not required for signaling. Strikingly, an internal in-frame deletion removing sequences preceding the C-terminal CAAX box of Ste18 did not impair signaling by either of its cognate G protein-coupled pheromone receptors. Therefore, receptor interaction with the C-terminal domain of yeast Ggamma is not required for receptor-mediated G protein activation in vivo. Because the mechanism of G protein activation by receptors is conserved from yeast to humans, mammalian receptors may not require interaction with the tail of Ggamma for G protein activation in vivo. However, receptor-Ggamma interaction may modulate the efficiency of receptor-G protein coupling or promote activation of Gbetagamma effectors that co-cluster with receptors.

inducing GDP release and subsequent GTP loading (4). Consistent with this mechanism, a mutant of G s ␣ has been identified that requires association with G␤ to trigger a signal independently of a receptor (5). G␤ may also provide a receptor binding site that is important for G protein activation (9 -12).
G␥ subunits function indirectly to promote G protein activation, because they do not contact G␣ subunits (6 -8). G␥ subunits can bind receptors, as indicated by studies of rhodopsin and transducin, which show that ␤␥ binds activated rhodopsin and promotes activation of the ␣ subunit (13,14). Furthermore, rhodopsin preferentially activates transducin containing its cognate ␥ subunit (␥1; see Ref. 15); other receptors likewise couple preferentially with G proteins containing certain G␥ subunit isoforms in vitro (16 -18). Selective coupling appears to be mediated by the diverged C termini of G␥ subunits, because peptides corresponding to the last 14 residues of ␥1 stabilize the active conformation of rhodopsin (metaII) and competitively inhibit rhodopsin-mediated activation of transducin (19). These and other observations have suggested that receptors possess independent binding sites for the C-terminal domains of G␣ and G␥ subunits (13,14,19). Thus, one variation of the "lever hypothesis" of G protein activation requires simultaneous binding of the C termini of G␣ and G␥ to independent sites on a receptor (5), although other variations of this model do not (2,3).
In certain signaling pathways, receptor interaction with the C terminus of G␥ may have a non-essential role in G protein activation. Studies have shown that ␥7, which interacts poorly with M2 or M4 muscarinic acetylcholine receptors in vitro, supports 2-fold more efficient coupling with these receptors than does a G o heterotrimer containing ␥5, whose tail binds these receptors in vitro (16). Furthermore, muscarinic receptor coupling to G o in vitro is enhanced when the heterotrimer contains ␥5 with a scrambled C-terminal sequence that does not bind muscarinic receptors (20). However, it remains to be established whether receptor binding to the C termini of G␥ subunits is an obligate or modulatory event for G protein activation in vivo.
We established previously that the C-terminal domain of G␥ promotes receptor-G protein coupling by analyzing mutations affecting the G␥ subunit (STE18 gene product) of the yeast Saccharomyces cerevisiae. In yeast, a single type of G protein heterotrimer transduces signals from mating pheromone receptors (21), providing a simple system to evaluate G␥ function genetically and biochemically. We found that G␥ truncated at position 94 (13 residues preceding the CAAX box) associates with G␤ but strongly impairs receptor-G protein coupling in vitro (agonist binding was low affinity and insensitive to GTP␥S) 1 (22). This receptor coupling defect was not because of lack of C-terminal prenylation or palmitoylation, because G␥ bearing a substitution (C107S) inactivating its CAAX box associated with G␣ and G␤ and supported efficient receptor-G protein coupling in vitro (agonist binding was high affinity and sensitive to GTP␥S). Therefore, peptide sequences between residue 94 and the CAAX box of yeast G␥, equivalent to the receptor binding domains of mammalian G␥ subunits (14,15,18,19,23,24), were proposed to promote receptor-G protein coupling.
Here we have investigated whether the C-terminal domain of yeast G␥ is required for receptor-mediated signaling in vivo. By analyzing the function of a diverged family of Ste18 homologs from several species of budding yeast and amino acid substitutions affecting the C terminus of Ste18 we provide evidence indicating that the C-terminal domain preceding the CAAX box of G␥ is dispensable for receptor-mediated signaling in vivo.

EXPERIMENTAL PROCEDURES
Yeast Strains and Media-The S. cerevisiae strains used in these studies are listed in Table I. S. cerevisiae cells were grown in standard synthetic medium containing adenine and amino acid supplements but lacking appropriate nutrients to select for cells containing plasmids.
Plasmids, Oligonucleotide-directed Mutagenesis, and Sequence Analysis-STE18 was expressed from the PGK promoter, and terminator sequences were cloned into the yeast high copy plasmid pRS425 as follows. A fragment containing the PGK promoter and terminator separated by a multicloning site containing restriction sites for EcoRI, HindIII, and BamHI was excised from the pPGK vector (25) by digesting with XhoI and SalI and inserted into pRS425 (26) cut with XhoI and SalI. The resultant plasmid was digested with SalI and NotI and circularized to produce pRS425PGK, which has unique EcoRI, HindIII, and BamHI sites for expression cloning. Wild-type and mutant forms of the STE18 coding region were cloned as HindIII-BamHI fragments into pRS425PGK, which allowed expression of wild-type and mutant forms of HA-tagged Ste18. The N-terminal HA tag fully preserves Ste18 protein function (22). All point mutations made in the STE18 gene expressed from the ADH promoter were constructed in pVT-HA-Ste18 (22) using the QuikChange TM site-directed mutagenesis method (Stratagene). The STE18 internal deletion mutant removing sequences encoding amino acids 94 -105 was constructed using the ExSite TM mutagenesis method (Stratagene), which resulted in the additional substitutions N92T and A93R. All mutations were confirmed by DNA sequencing using appropriate primers.
Cloning and Expression of STE18 Homologs from Five Divergent Saccharomyces Species-The genome sequences of five species of yeast of varying relatedness to S. cerevisiae (Saccharomyces bayanus (strain NRRL Y-11845), Saccharomyces castellii (NRRL Y-12630), Saccharomyces kluyveri (NRRL Y-12651), Saccharomyces kudriavzevii (IFO 1802), and Saccharomyces mikatae (IFO 1815); provided by M. Johnston, Washington University Genome Sequencing Center) were searched by BLAST for STE18 homologs. As is the case in S. cerevisiae, the genome of each of these species of yeast encoded a single STE18 homolog (data not shown). Contig sequences encoding each STE18 homolog were provided by P. Cliften and M. Johnston (Washington University Genome Sequencing Center) and will be deposited in Gen-Bank TM by those investigators. Primers used to amplify only the coding region of each STE18 homolog were designed such that the resultant amplification products could be cloned as HindIII-BamHI fragments into pRS425PGK as follows. Genomic DNA from each species of yeast (obtained from M. Johnston, Washington University School of Medicine) and the following primers were used: forward S. bayanus, 5Ј-GCCCCCAAGCTTATGTCTGCAGTTCAGAACTCG-3Ј and reverse S. bayanus, 5Ј-CGCGGATCCTTACATAAGCGTACAACAAGC-3Ј; forward S. kudriavzevii, 5Ј-GCCCCCAAGCTTATGTCTGCTATTCAGAA-CTCG-3Ј and reverse S. kudriavzevii, 5Ј-CGCGGATCCTTACATAAGC-GTACAACAAGC-3Ј; forward S. mikatae, 5Ј-GCCCCCAAGCTTATGT-CTGTACTTCAAGATTCACC-3Ј and reverse S. mikatae, 5Ј-CGCGGA-TCCTTACATAATCGTACAGCAAAC-3Ј; forward S. kluyveri, 5Ј-GCCC-CCAAGCTTATGTCTTCAGAAGAGCAGCAACC-3Ј and reverse S. kluyveri, 5Ј-CGCGGATCCTTACATAATTGCACAGCAGTTG-3Ј; forward S. castellii, 5Ј-GCCCCCAAGCTTATGTCACAACAGATAAAA-ACACC-3Ј and reverse S. castellii, 5Ј-CGCGGATCCTTACATTATAGC-ACAGCAACC-3Ј. The amplification conditions using Taq polymerase were as follows: 96°C for 5 min; 30 cycles of 96°C for 30 s, 43°C for 2 min, and 72°C for 2 min; 72°C for 10 min. The resultant PCR fragments were sequentially digested with BamHI and HindIII and cloned into pRS425PGK to express the native, untagged forms of these Ste18 homologs in S. cerevisiae. The correct nucleotide sequence of each construct was confirmed by sequencing.
Random Mutagenesis of S. cerevisiae STE18 and Identification of STE18 Dominant-negative Mutants-A plasmid suitable for performing random mutagenesis of codons 79 -110 in the STE18 gene was constructed by introducing a silent KspI site into sequences encoding amino acids 77-79 and a KpnI site immediately downstream of the STE18 stop codon in plasmid pVT-HA-STE18, resulting in pVT-HA-STE18-KK. Oligonucleotides containing 1% of the three incorrect bases at each position spanning the coding region for amino acids 79 -110 of Ste18 were synthesized and inserted into pVT-HA-STE18-KK cleaved with KspI and KpnI. Introduction of the library into Escherichia coli produced a pool of 1500 bacterial transformants that were used in the following ways. Sequencing of 16 random plasmids revealed that one was wild-type, and the remaining contained mutations affecting various codons throughout the targeted region, as expected. To select for dominant-negative STE18 mutations, we introduced the pool of 1500 plasmids into the wild-type S. cerevisiae strain RK511-6B. Portions of the transformation mixture were plated on selective media that lacked or contained a dose of ␣-factor (1 M) sufficient to arrest growth of the parent strain expressing wild-type Ste18. Colonies resistant to ␣-factor were recovered at a frequency of 3%. Plasmid dependence of the ␣-factor-resistant phenotype was demonstrated by recovery of ␣-factor sensitivity following plasmid loss on media containing 5-fluoroorotic acid. Plasmids conferring ␣-factor resistance were isolated from yeast cells and sequenced. The same pool of mutagenized DNA was also subjected to a screen for STE18 dominant-negative mutations in which wild-type cells (RK5111-6B) were transformed with the library, and random colonies were picked and screened for impaired response to ␣-factor in growth arrest (halo) assays.
Pheromone Response Assays-Long term (48 h) pheromone-induced growth arrest (halo) assays were used to determine the ability of cells to respond to varying concentrations of agonist (␣-factor) as described previously (27). Sterile paper disks containing various amounts of synthetic agonist (␣-factor; 15 pmol, 50 pmol, 150 pmol, 500 pmol, 1.5 nmol) were applied to lawns of cells embedded in soft agar. After incubation at 30°C for 2 days, zones of growth inhibition were measured, and plates were scanned electronically to record images. Unless indicated otherwise, halo assays were performed with ste18⌬ cells expressing various wild-type or mutant STE18 alleles from the ADH or PGK promoter on high copy plasmids. Although various STE18 homologs and S. cerevisiae STE18 alleles were expressed from strong promoters on high copy plasmids, this would not result in overexpression of G␣␤␥ complexes, because G␣ and G␤ subunits were expressed from their normal chromosomal loci and therefore were limiting. Quantitative Mating Assays-Cells were grown to mid-log phase in synthetic media lacking leucine to select for plasmids. Approximately 1 ϫ 10 7 cells of the tester strain of the appropriate mating type and 2 ϫ 10 6 cells of the experimental strain (a ste18⌬ mutant of either mating type that expressed various STE18 alleles from plasmids) were mixed, collected on 0.45-m nitrocellulose filters, incubated for 3 h at 30°C on YPD media, suspended, diluted, and plated in triplicate on appropriate media to select for diploids. The number of viable haploids of the experimental strains present before mating was quantified by plating on appropriate selective media. Mating efficiency (%) was calculated by dividing the number of diploids produced by the number of haploids of the limiting strain and multiplying by 100.

Ste18
Homologs from Divergent Yeast Species Can Functionally Replace S. cerevisiae Ste18 -As an initial means of investigating whether the C-terminal domain preceding the CAAX box of Ste18 is required for receptor-G protein coupling and signaling in vivo, we determined whether the complete signaling defect of cells lacking Ste18 could be corrected by plasmidmediated expression of Ste18 homologs encoded by five diverged species of budding yeast whose genomes have been sequenced completely (29). 2 Two factors indicated that this would be a useful approach. First, because Ste18 is the only G␥ subunit in S. cerevisiae, the function of Ste18 homologs from other yeast species could be examined in the absence of S. cerevisiae Ste18 because of deletion of the chromosomal STE18 locus. Second, BLASTX searches indicated that the genome of each yeast species encoded a single Ste18 homolog with a relatively conserved G␤ binding domain (central two-thirds of the molecule) and a more highly diverged C-terminal sequence in the putative receptor coupling domain immediately preceding the CAAX box (Fig. 1). These analyses indicated that the Ste18 homologs of S. bayanus, S. kudriavzevii, and S. mikatae are most similar to S. cerevisiae Ste18 (ϳ85% identity overall), whereas the Ste18 homologs of S. castellii and S. kluyveri are more highly diverged (57 and 50% identity, respectively).
Despite exhibiting varying degrees of sequence conservation in their putative receptor coupling domains, all of the Ste18 homologs could correct the signaling defect of S. cerevisiae cells carrying a disruption of the chromosomal STE18 gene. As indicated by quantitative assays of agonist (␣-factor)-induced growth arrest (Fig. 2), expression of Ste18 homologs most similar to S. cerevisiae Ste18 (S. bayanus, S. mikatae, or S. kudriavzevii) fully restored agonist responsiveness to the S. cerevisiae ste18⌬ mutant (zones of ␣-factor-induced growth inhibition were equivalent to those of ste18⌬ cells expressing the S. cerevisiae STE18 gene from a plasmid). More strikingly, expression of either of the more highly diverged Ste18 homologs (S. kluyveri and S. castellii) nearly completely restored agonist responsiveness to the ste18⌬ mutant (zones of growth inhibition were slightly smaller or turbid). Because the C-terminal domains of these latter two Ste18 homologs are quite diverged relative to S. cerevisiae Ste18, these results suggested that specific amino acid sequences or motifs preceding the CAAX box of Ste18 may not be critical for receptor-G protein coupling and signaling.
Missense tide sequences preceding the CAAX box of S. cerevisiae Ste18 are important for receptor-mediated signaling, we constructed and analyzed several types of missense mutations affecting this domain. We first used alanine scanning mutagenesis targeted to residues 99 -105, just upstream of the CCAAX box (residues 106 -110), which is prenylated and palmitoylated and required for activation of the downstream mitogen-activated protein kinase cascade (30 -32). Pairs of residues were substituted with an Ala-Ala dipeptide to create six mutants (M99A,S100A; S100A,N101A; N101A,S102A; S102A,N103A; N103A,S104A; S104A,V105A) that were analyzed for their ability to function when expressed from a plasmid in a ste18⌬ mutant. As indicated by quantitative assays of agonist-induced growth arrest, expression of each of these alleles in a ste18⌬ mutant resulted in wild-type or nearly wild-type response to agonist (Fig. 3A), suggesting that the side chains of residues 99 -105 are dispensable for receptor-mediated signaling.
As an alternative to alanine scanning mutagenesis, we altered the charge distribution of selected residues within the C-terminal domain of Ste18. In one mutant, lysine 95 was changed to aspartic acid, and in another mutant three uncharged residues (Met-99, Ser-100, Asn-101) were all changed to aspartic acid. Expression of either of these mutants from a plasmid in cells carrying a deletion of the chromosomal STE18 locus fully restored agonist-dependent signaling (Fig. 3B), further suggesting that specific amino acid sequences in the Cterminal domain of Ste18 are not essential for receptor coupling and signaling in vivo.
Although suggestive, the preceding results could not exclude the possibility that the C-terminal domain of Ste18 contains several sequences or motifs that mediate receptor coupling in a functionally redundant manner. To address this possibility, we introduced a string of four alanine residues at positions 94 -97 and 102-105 or eight alanine residues at positions 98 -105. Strikingly, expression of any of these STE18 alleles from a plasmid fully rescued the ability of a ste18⌬ mutant to respond to agonist (Fig. 3C).
We also considered that Ste18 containing a string of eight alanine residues (positions 98 -105) preceding the CAAX box could cause a quantitative defect in receptor-mediated signaling that was difficult to detect by in vivo signaling assays. Therefore, to increase the sensitivity of these assays we expressed plasmid-borne STE18 alleles in a ste18⌬ mutant that expressed ␣-factor receptors lacking their C-terminal domains.
Receptors lacking their C-terminal tail were used, because they are partially impaired for G protein coupling in vitro (33), which is manifested in vivo by zones of agonist-induced growth inhibition that are slightly turbid and have less distinct margins (Fig. 3D). Despite this partial impairment in function, truncated ␣-factor receptors signaled with similar efficiency whether wild-type Ste18 or mutant Ste18 containing eight alanine residues was expressed from a plasmid in ste18⌬ cells (Fig. 3D). Therefore, this mutant form of Ste18 appeared to be highly functional even when receptor-G protein coupling was partially impaired by other mechanisms.
Mutations throughout Sequences Encoding the C-terminal Domain Preserve Ste18 Function-To address whether receptor coupling information is present elsewhere within the C-terminal domain of Ste18, we performed random oligonucleotide mutagenesis of sequences encoding residues 79 -110. Sequences further upstream were spared, because they probably are required for association with G␤, as suggested by structural studies of mammalian G␤␥ complexes (34). A pool of plasmids created by random oligonucleotide mutagenesis of this region contained point and frameshift mutations as indicated by sequencing random clones (data not shown). This plasmid pool was subjected to two types of analyses to identify mutations that potentially impair receptor coupling. In the first, we selected for plasmids that encode dominant-negative G␥ subunits. Dominant-negative G␥ mutants were selected by their ability to allow wild-type cells to form colonies on medium containing a high dose of agonist (␣-factor; 1 M) sufficient to arrest the growth of control cells. This approach yielded only mutations that inactivated or removed the CAAX box of Ste18 (truncation, frameshift, or missense mutants; see Table II), which is required for membrane targeting (31).
Because the previous selection method might have been biased for very strong dominant-negative STE18 mutations that nearly completely block signaling, our second approach used a screen that is capable of identifying plasmids in the mutagenized library that encode weaker dominant-negative G␥ subunits. The screen was performed by introducing the library of mutagenized plasmids into wild-type cells, picking colonies at random, and assaying for impaired agonist-induced signaling, as indicated by formation of smaller or turbid zones of agonist-induced growth inhibition. Despite the ability of this assay to detect partial loss of function mutants (35), the only mutations that caused a detectable dominant-negative phenotype affected the CAAX box (truncation, frameshift, missense, or point mutations; see Fig. 4 and Table II). Therefore, these results suggested that a functionally critical receptor-coupling domain may not be present within residues 79 -107 of Ste18.
Ste18 Deleted for Residues Preceding Its CAAX Box Is Functional-The preceding results suggested that sequences preceding the CCAAX box of Ste18 are not required for receptormediated signaling. As a further test of this hypothesis, we constructed an internal deletion mutant (⌬94 -105) of Ste18 that lacks the region of the C-terminal tail implicated biochemically in receptor coupling (22) but preserves the CCAAX box. The CCAAX box was retained, because palmitoylation and prenylation of Ste18 is required for downstream signaling (30 -32), although these modifications are not required for recep-tor-G protein coupling in vitro (22). Strikingly, expression of this STE18 internal deletion allele (⌬94 -105) from a plasmid fully restored the ability of a ste18⌬ mutant to respond to agonist (␣-factor; see Fig. 5). Expression of an N-terminally HA-tagged version of this internal deletion mutant protein, as well as two other mutants analyzed in this study (98 -105Ala; C107S, a CAAX mutant) was confirmed by immunoblotting (Fig. 6). Therefore, the C-terminal region of Ste18 immediately preceding the CCAAX box is dispensable for ␣-factor receptormediated G protein activation and signaling.
Signaling by Neither the a-Factor Nor the ␣-Factor Receptor Requires the Domain of Ste18 Preceding the CCAAX Box-The final possibility we considered was that the C-terminal tail of Ste18 may be required for coupling to one type of mating pheromone receptor but not the other. Indeed, previous studies have indicated that the yeast G␣ subunit uses different regions of its C terminus to couple with ␣-factor versus a-factor receptors (36), the only two receptors in yeast that couple with the G protein containing Ste18 (37). Accordingly, because the preceding experiments examined the effects of Ste18 mutations on ␣-factor receptor signaling, we subsequently determined whether the tail of Ste18 is required for a-factor receptor signaling. These experiments could not use assays of a-factorinduced growth arrest, because fully processed and prenylated a-factor is unavailable in sufficient quantity. As an alternative, we compared the mating efficiencies, a sensitive and quantitative assay of receptor-dependent signaling, of cells expressing a-factor receptors and either wild-type Ste18 or mutant Ste18 lacking residues 94 -105. As a control, we also quantified the mating efficiencies of cells expressing ␣-factor receptors and either wild-type Ste18 or mutant Ste18 lacking residues 94 -105. Remarkably, cells expressing a-factor receptors and mutant Ste18 mated as efficiently as controls expressing wild-type Ste18 (Table III). Cells lacking a STE18 plasmid did not mate (data not shown). Similarly, cells expressing ␣-factor receptors mated with equivalent efficiency whether they expressed wild- a Codon changed to nonsense mutation. b Codon affected by frameshift mutation.
FIG. 4. Inhibition of wild-type Ste18 function by overexpressed dominant-negative STE18 alleles. The indicated wild-type (WT) and dominant-negative STE18 alleles were overexpressed from the ADH promoter on a high copy plasmid (pVT-HA-STE18) in a strain (RK511-6B) carrying a wild-type STE18 gene on the chromosome. The results of agonist (␣-factor; 1.5 nmol/disk)-induced growth arrest assays are shown. Codons affected by nonsense and frameshift mutations are indicated by * and **, respectively. type or mutant Ste18 (Table III), and the ability of these cells to mate required the presence of a STE18 plasmid (data not shown). Therefore, we conclude that signaling by neither the ␣-factor receptor nor the a-factor receptor in vivo requires the C-terminal domain preceding the CAAX box of Ste18.

DISCUSSION
The primary conclusion of this study is that the C-terminal domain preceding the CAAX box of the G␥ subunit encoded by the STE18 gene of the yeast S. cerevisiae is dispensable for signaling by its two cognate G protein-coupled receptors in vivo. Furthermore, we obtained no evidence indicating that alteration of the C-terminal domain of yeast G␥ has a strong modulatory effect on the efficiency of receptor-mediated signaling in vivo. These findings are in contrast to the well established and evolutionarily conserved requirement for receptor contact with the C-terminal domain of the G␣ subunit in the process of G protein activation (reviewed in Refs. 1, 2, 4, 21, and 38).
The results presented here and our previous studies suggest that caution must be exercised when assessing the potential importance of certain receptor-G protein contacts solely by biochemical approaches. In the in vitro system used to detect receptor-G protein coupling in yeast plasma membrane fractions, agonist binding affinity of the ␣-factor receptor is sensitive to GTP␥S (i.e. becomes low affinity) only under conditions of high salt concentration and moderately elevated pH (40). At physiological salt concentration and pH, in contrast, agonist binding to ␣-factor receptors in plasma membrane fractions is GTP␥S-insensitive (i.e. remains high affinity; see Ref. 40). Therefore, only at high salt concentration and elevated pH is it possible to show that ␣-factor receptor-G protein coupling in vitro requires the G␥ tail (22). However, these reaction conditions may decrease the stability of receptor-G protein complexes such that a functional interaction between the receptor and the G␥ tail becomes evident. In contrast, the ␣-factor receptor-G protein interface may be more stable in vivo and therefore tolerates loss of sequences preceding the CAAX box of G␥.
How do our results impact current models of receptor-mediated G protein activation in which receptors are hypothesized to act as a lever to open the nucleotide binding pocket of G␣ (3,5)? One variation of this model suggests that simultaneous contact between the receptor and the C-terminal domains of G␣ and G␥ is required for receptors to open the nucleotide binding site of G␣ and catalyze nucleotide exchange; however, our results exclude this model, at least in yeast. Instead, they support mechanisms in which the receptor contacts G␣, G␤, and/or other regions of G␥ to activate the G protein. They are also consistent with a model in which receptor interaction with G␣ is responsible for perturbing the ␣5/␤6 loop via movement of the ␣5 helix (2, 3), which does not invoke an essential role for receptor-G␥ interaction.
Are the C-terminal domains of G␥ subunits dispensable for receptor-mediated G protein activation in mammalian cells? Although this question requires further investigation, several considerations suggest that contacts between the tails of G␥ subunits and mammalian G protein-coupled receptors are unlikely to be essential for G protein activation in vivo. First, the mechanism of receptor-mediated G protein activation is conserved from yeast to humans, because similar domains of yeast and mammalian receptors and G proteins are required for receptor-G protein coupling (22, 35, 36, 40 -43). Second, sequences preceding the CAAX boxes of mammalian and yeast G␥ subunits are not conserved, yet several types of mammalian receptors expressed in yeast can activate a G protein consisting of a yeast/mammalian G␣ chimera, yeast G␤ (Ste4), and yeast G␥ (Ste18) (44 -46). Third, G␥5 carrying a scrambled amino acid sequence preceding its CAAX box does not bind M2 muscarinic acetylcholine receptors but can support efficient coupling of G o ␣ and G␤ 1 with these receptors in vitro (20). Fourth, G␥ 7 , which does not appear to interact with M2 or M4 muscarinic acetylcholine receptors in vitro, supports more efficient coupling with these receptors than does a G protein containing G␥ 5 , whose C-terminal tail can bind these receptors in vitro (16).
In apparent contrast, however, there is evidence suggesting that the G␥ tail is important for coupling between muscarinic receptors and G proteins in vivo (24). Cytoplasmic injection of a geranylgeranylated peptide corresponding to the wild-type C terminus of ␥5 strongly impairs the ability of muscarinic receptors to inhibit N-type Ca 2ϩ channels in primary sympathetic neurons, whereas a scrambled version of this peptide or wild-type peptides corresponding to the C termini of ␥7 or ␥12 have no effect (24). These observations are consistent with the hypothesis that muscarinic receptors have a functionally important binding site for the C terminus of ␥5. However, they are equally likely to suggest that the ␥5 peptide blocks the ability of G␤␥ subunits to inhibit N-type Ca 2ϩ channels by interfering with G␤␥ binding to the channel.
Because the C-terminal domains of G␥ subunits clearly have the ability to bind receptors (14 -20, 22, 24), this interaction

TABLE III
Mating efficiencies Efficiency of mating (%, calculated as described under "Experimental Procedures") Ϯ S.E. (n ϭ 4) using cells expressing the indicated receptors and STE18 alleles. YML80 (expressing ␣-factor receptors and the indicated STE18 alleles) was mated with DC17; YML81 (expressing a-factor receptors and the indicated STE18 alleles) was mated with DC14. may have novel functions in yeast or mammalian cells. For example, this interaction might allow G␤␥ subunits to remain bound to receptors after G␣ subunits are activated, which could promote downstream signaling via G␤␥ effectors that cluster with receptors. Indeed, the blocking effect of the ␥5 C-terminal peptide on muscarinic receptor-mediated inhibition of N-type Ca 2ϩ channels could be because of release of G␤␥ subunits from receptors that potentially cluster with N-type channels, decreasing the local concentration of G␤␥ required to maximally inhibit these channels. Tethering of G␤␥ subunits to receptors also has the potential to promote resetting of the system to the inactive state by facilitating capture of G␣ subunits after GTP hydrolysis. Therefore, further studies of receptor-G␥ interaction may reveal new insights into mechanisms that control the efficiency, fidelity, and kinetic control of G protein signaling.