G protein gamma subunit interaction with a receptor regulates receptor-stimulated nucleotide exchange.

The surfaces of heterotrimeric G proteins (alphabetagamma) in contact with receptors and the molecular events at these sites, which lead to G protein activation, are largely unknown. We show here that a peptide from the C terminus of a G protein gamma subunit blocks muscarinic receptor-stimulated G protein activation in a sequence-dependent fashion. A G protein mutated at the same site on the gamma subunit shows enhanced receptor stimulated nucleotide exchange without affecting G protein heterotrimerization. Ineffective contact between the gamma subunit and receptor increases the rate of receptor-stimulated nucleotide exchange. Specific interaction of the G protein gamma subunit with the receptor thus helps the betagamma complex to act at a distance and control guanine nucleotide exchange in the alpha subunit.

The surfaces of heterotrimeric G proteins (␣␤␥) in contact with receptors and the molecular events at these sites, which lead to G protein activation, are largely unknown. We show here that a peptide from the C terminus of a G protein ␥ subunit blocks muscarinic receptor-stimulated G protein activation in a sequencedependent fashion. A G protein mutated at the same site on the ␥ subunit shows enhanced receptor stimulated nucleotide exchange without affecting G protein heterotrimerization. Ineffective contact between the ␥ subunit and receptor increases the rate of receptor-stimulated nucleotide exchange. Specific interaction of the G protein ␥ subunit with the receptor thus helps the ␤␥ complex to act at a distance and control guanine nucleotide exchange in the ␣ subunit.
Although G protein signaling is central to a vast majority of pathways that control the physiology of mammalian cells, the coordinated changes in the conformations of the agonist bound receptor and the G protein subunits that trigger nucleotide release are not known. Peptide and mutant studies using biochemical assays that measure receptor-G protein coupling have implicated three protein domains, the N-and C-terminal domains of the ␣ subunit and the C-terminal domain of the ␥ subunit in receptor interaction (1)(2)(3). Functional contact between the ␥ subunit and a receptor is one mechanism that can explain the universal requirement of the ␤␥ complex for receptor-mediated nucleotide exchange in the ␣ subunit. However, the particular role that the ␥ subunit plays at the receptor surface during G protein activation has not been clear. The crystal structure of the G protein heterotrimer indicates that the C termini of the ␥ and ␣ subunits are a considerable distance from the nucleotide-binding site in the ␣ subunit (4,5). The mechanisms that help the receptor regulate nucleotide exchange by contacting these domains are therefore an outstanding puzzle.
Because a peptide specific to the C terminus of the ␥1 subunit type stabilizes activated rhodopsin in a sequence-dependent manner (6) and a homologous ␥5 peptide specifically inhibits muscarinic receptor modulation of a Ca 2ϩ current in intact neurons (3), we examined the effect of a geranylgeranylated ␥5 peptide on the M2 muscarinic receptor activation of a G protein. The peptide specific to the C-terminal 14 residues of the ␥5 subunit was prenylated because the native ␥5 subunit is modified with geranylgeranyl at the C-terminal Cys residue that is part of a CAAX motif (2). This peptide inhibited activation of G proteins by reconstituted M2 receptor. A peptide with the same sequence scrambled was inactive. These studies indicated that the ␥ subunit peptide interacts with the receptor in a sequence-specific fashion. To test the effect of mutations in this C-terminal domain of ␥5 on G protein activation by M2, C-terminal residues of ␥5 corresponding to the peptides were scrambled. The scrambled sequence was identical to the sequence of the scrambled peptide used in the earlier experiments. Wild type and mutant ␤1␥5 complexes were purified and bound to ␣o, and the ability of the M2 receptor to activate these heterotrimers was measured in a reconstituted system containing purified proteins. A significant difference in the receptor-stimulated GTPase activity between the wild type and mutant G o was detected in this system. A phospholipase C enzyme-based assay indicated that efficacy of interaction with the ␣ subunit was unaltered by the mutant ␥ subunit. Together these results indicate that the ␥ subunit interaction with the receptor regulates nucleotide exchange in the ␣ subunit by affecting the positioning of the ␤␥ complex with reference to the ␣ subunit. The results thus identify a mechanism that allows a receptor to regulate nucleotide exchange at a distance in the G protein.

Synthesis and Prenylation of Peptides-
The amino acid sequences of the peptides were as follows: ␥5pep-gg (wild type):VSSSTNPFRPQKVC and ␥5pep-gg-scr (scrambled): PSRTPVNFSQVSKC. Cys residues were retained at the C terminus for chemical geranylgeranylation using geranylgeranyl bromide. Prenylation and purification using fast protein liquid chromatography has been described (7). The integrity of the modified peptides were checked by mass spectrometry and chromatography, and peptide concentrations were estimated by amino acid analysis. Except where specified, all chemicals were from Sigma.
Purification and Reconstitution of M2 Receptors-Detailed description of M2 purification, reconstitution, and measurement of G protein stimulation is published elsewhere (7). Briefly, baculoviruses containing His-tagged M2 receptor cDNA (kind gift from Dr. E. M. Ross) were expressed in insect cells. M2 was purified according to a previously published protocol (8). Sf9 cell membranes containing M2 were solubilized in 50 mM Hepes, pH 7, 50 mM NaCl, digitonin (Calbiochem)/ sodium cholate added to 1:0.5% final concentration. All procedures were performed at 4°C. Solubilized receptor was bound to cobalt-chelate beads that were prepared using iminodiacetic acid beads and cobalt chloride. His-M2 eluted from these columns with 200 mM imidazole retain 50% of the N-[methyl-3 H]scopolamine binding activity in Sf9 cell membranes. Using modifications of a previous method (8), purified M2 was reconstituted into brain lipids (Folch Fraction VII). The composition of the brain lipid mixture is sphingomyelin (20%), phosphatidylethanolamine (30%), phosphatidyl-serine (20%), and other lipids according to the distributor (Sigma). Typically, 100 l (1 mg lipid) of the aqueous lipid suspension is added to 18 l of 10% sodium deoxycholate and 4 l of 10% sodium cholate. 100 -500 l of pure receptor is mixed with solubilized lipids (ϳ122 l from above). The mixture is applied to a 10-12-ml column of Sephadex G-50 (fine), equilibrated with solubilization buffer. Vesicle fractions were collected and concentrated. N-[methyl-3 H]Scopolamine binding in a standard filter binding assay was used to estimate receptor concentration. The yields were 20 -30% relative to purified solubilized receptors.
M2 Stimulation of G Protein Activity-␣o subunit protein was expressed in bacteria with yeast myristoyl-transferase and purified as described before (9). ␤1␥5 and ␤1␥5scr complexes were produced in Sf9 insect cells by triple infection of His-␣i2, ␤1 (kind gifts from Dr. T. Kozasa), and ␥ subunit viruses using a previously published procedure with minor modifications (10). The two ␤␥ complexes were separately purified in complex with His-tagged ␣i2 subunit by binding the complex to nickel-nitrilotriacetic acid resin and eluted the ␤␥ complexes with aluminum fluoride. Yields of the ␤1␥5 and ␤1␥5scr were similar. The ␥5Al and ␥5⌬ mutants were synthesized as His-tagged proteins, expressed in complex with ␤1 subunit in Sf9 cells, and purified by directly binding to nickel-nitrilotriacetic acid resin. As a control a wild type ␤1-His-tagged ␥5 complex was expressed and purified. Although the yield of ␤1His-␥5 was comparable with or higher than that of ␤1␥5, the yields of the ␤1His␥5Ala/⌬ were relatively low. The final purity was examined by gel electrophoresis and Coomassie Blue staining where no significant levels of other proteins were detected. Protein concentration was determined by laser scanning densitometry using standards. G protein heterotrimer was formed by preincubating ␣o with ␤␥ complexes for 30 min at 4°C. Lipid-reconstituted M2 was incubated with ␣o subunit with or without ␤␥ complexes for 30 min at 4°C in 20 mM Hepes buffer, pH 8, containing 2 mM MgCl 2 , 100 mM NaCl, 10 M GDP, and 1 mM dithiothreitol. GTP␥S 1 binding to ␣o was determined using previously published methods (11). In peptide assays, the incubation was performed by mixing with dried peptide or peptide vehicle (3 M CHAPS) as a control. The final reaction mix contained 1 nM M2, 100 nM ␣o, and 0 -10 nM ␤1␥5 in the buffer above and was equilibrated at 23°C. The binding reaction was started by addition of 0.2 M [ 35 S]GTP␥S and 1 mM carbachol, 1 mM atropine or vehicle. The reaction was stopped with ice-cold reaction buffer containing 200 M GTP␥S and 1 mM atropine. [ 35 S]GTP␥S bound to ␣o was measured in a filter binding assay. For GTPase assays, the conditions were the same as above. ␣o GTPase activity was measured essentially as described (11). The reaction was started with 0.2 M [␥-32 P]GTP, 1 mM carbachol, or vehicle and stopped with ice-cold 5% charcoal in 50 mM sodium phosphate, pH 7.0. The samples were centrifuged, and the radioactivity in the supernatant was estimated using scintillation counting. Purified RGS4 was kindly provided by Dr. Maurine Linder.
Inhibition of ␤␥-stimulated PLC-␤ 3 Activity by ␣o-␣o inhibition of ␤␥ complex-stimulated PLC-␤ activity was measured as described before (12). ␣o and ␤1␥5 or ␤1␥5scr complexes were preincubated for 30 min at 4°C. The reaction mixture contained G protein, PLC-␤, and lipids. Freeze dried lipid mixtures were ultrasonicated before use. Final concentrations were 150 M phosphatidyl ethanolamine, 50 M [ 3 H]phosphatidyl inositol diphosphate, 1.25 nM PLC-␤3, and 4 nM ␤1␥5 (wild type or mutant). Ca 2ϩ was added to initiate the reaction. The reaction was performed at 30°C for 30 min. The reaction was stopped with trichloroacetic acid with bovine serum albumin, and the radioactivity in the supernatant was measured. No more than 10 -15% of the substrate, [ 3 H]phosphatidyl inositol diphosphate, was used during the reactions. Purified recombinant PLC-␤ was a kind gift from Dr. A.

Smrcka.
Construction of ␥5 Mutants-The rat ␥5 cDNA was mutated by replacing the 3Ј end of the ␥5 cDNA (beginning from a BsmI site at base 156) with a double-stranded DNA cassette encoding the scrambled sequence. Alanine substitutions and the deletion of 10 residues upstream of the codon encoding the C-terminal Cys were performed using polymerase chain reaction based methods in a ␥5 cDNA that was His-tagged. Mutant cDNAs were transferred to the baculovirus using the pFastBac system (Invitrogen-Life Technologies, Inc.).

Effect of ␥5 Subunit Peptide on M2-dependent G Protein
Activation-A peptide specific to the C-terminal 14 residues of ␥5 was geranylgeranylated (␥5pep-gg), and the ability of the peptide to compete with a G protein (␣o␤1␥5) for interaction with M2 was examined. A reconstituted system containing purified M2 and G protein subunits was set up (described under "Materials and Methods"; Fig. 1, A and B). Purified reconstituted M2 had a K d for N-methyl-scopolamine of 250 pM (12) and activated G o in an agonist- (Fig. 1C) and ␤␥-dependent fashion (data not shown). M2-stimulated G o activation was significantly inhibited by ␥5pep-gg but not by a peptide with the same amino acids sequence scrambled, ␥5pep-gg-scr (Fig. 1C). Because these peptides have no effect on the G protein heterotrimer (3), these results indicate that ␥5pep-gg competes with the G protein for a site on the M2 receptor in a sequence-specific manner.
Effect of Mutating the C-terminal Domain of the ␥5 Subunit on G o Heterotrimer Activation by the M2 Receptor-Competition of the ␥5 peptide with the G protein indicated that the G protein ␥5 C-terminal domain interacts with the M2 receptor.
To obtain further evidence to support this mechanism, mutant forms of the ␥ subunit were synthesized with altered C-terminal sequences. If effective interaction of the ␥5 subunit C terminus with the M2 receptor is a requirement for activation of a G protein, one or more of these mutant forms of ␥5 were expected to alter the activation properties of the G protein.
Three different mutants of the ␥5 subunit were expressed and purified from insect cells in complex with the ␤1 subunit type ( Fig. 2A). These mutants were as follows: (i) ␥5scr: the last 13 residues upstream of the Cys in the CAAX box were scrambled identical to the sequence of ␥5pep-gg-scr; (ii) ␥5⌬: 10 residues upstream of the Cys residue were deleted; and (iii) ␥5Ala: a short sequence of three residues, NPFR, which is conserved in all G protein ␥ subunits, was substituted with Ala residues.
His-tagged forms of ␥5Ala and ␥5⌬ mutants were co-expressed with the ␤1 subunit and purified as ␤1␥5⌬ and ␤1␥5Ala complexes. A wild type ␤1His-␥5 complex was also expressed and purified as a control. Gel exclusion chromatography indicated that the two mutants formed effective complexes with the ␤1 subunit. However, in contrast to the wild type ␤1His-␥5, the mutant ␤1His-␥5Ala and ␤1His-␥5⌬ did not activate PLC-␤3 and did not form a heterotrimer with the ␣o subunit effectively (measured using the ability of pertussis toxin to ADP-ribosylate the ␣o subunit in the presence of the ␤␥ complex (data not shown)). They were therefore not used in the experiments described below.
The ␥5scr mutant was co-expressed in the native state with ␤1 and His tagged ␣i2 subunits. The heterotrimer was bound to a nickel-nitrilotriacetic acid column. The ␤1␥5scr complex was eluted using imidazole. As a control wild type ␤1␥5 was synthesized using a similar approach (Fig. 1B). The purified ␤1␥5scr activated PLC-␤3 and formed heterotrimers effectively similar to wild type ␤1␥5 (described below). All experiments were therefore performed with this mutant.
The mutant ␤1␥5scr and wild type ␤1␥5 in complex with ␣o were first compared for their relative levels of activation by the M2 receptor. At a receptor to ␣ subunit ratio of (1:100), no significant difference in receptor-stimulated GTP␥S binding was detected between the wild type and mutant heterotrimers at various concentrations of the ␤␥ complex (Fig. 2B). Assaying M2 activation of ␣o␤1␥5 and ␣o␤1␥5scr at various concentrations of carbachol also did not indicate differential activation (Fig. 2C).
In contrast to M2-stimulated GTP␥S binding where each ␣ subunit can bind utmost one molecule of GTP␥S, M2-stimulated GTPase activity can result in several cycles of activation of a G protein ␣ subunit resulting in the formation of many molecules of P i for each G protein. Because of this amplification, GTPase assays could be performed with significantly lower concentrations of the ␣o subunit compared with the GTP␥S binding assay, thus approaching a ratio of [receptor] to [G protein] of 1:1. Although the K d for G o binding to M2 is not known, the K d for G t binding to rhodopsin is 1 nM in the absence of nucleotide (13). The conditions in the GTPase assay were thus potentially closer to the K d for G o binding to M2 and likely to reveal differences between wild type and mutant ␤1␥5 complexes during M2 activation. To enhance sensitivity, we examined the receptor-stimulated GTPase activity in the presence of saturating concentrations of RGS protein, RGS4 (7). RGS4 acts as a GTPase-activating protein for the G o/i family (14). RGS4 potentiated M2-stimulated GTPase activity over 10-fold (7). When the time course of M2-stimulated GTPase activity was measured in the presence of RGS4, the M2 receptor activated ␣o␤1␥5scr significantly more than the wild type (Fig. 3). In the absence of the agonist or the ␤␥ complex, M2 did not effectively stimulate G o GTPase activity (footnote a in Table I). G o containing the ␥5scr mutant was also more active at several ratios of G o :M2 (Table I). Under these conditions the GTPase activity is a measure of the rate of receptor-stimulated nucleotide exchange (15). The results thus indicate that the mutant has a higher rate of receptor-stimulated nucleotide exchange compared with the wild type.
In the absence of the RGS protein, the mutant and wild type G o proteins still showed differential receptor-stimulated GTPase activities (Table I). This result indicates that differential receptor-stimulated GTPase activities of ␣o␤1␥5scr mutant and wild type are not due to differential effects of the RGS4 protein on receptor activation of the mutant and wild type.
Alternative explanations were excluded for this differential activity: (i) Both ␤␥ complexes were not contaminated with ␣

FIG. 2. Effect of ␥ subunit mutation on G o activation by M2.
A, amino acid sequences at the C terminus of the mutant forms of the ␥5 subunit. ␥5WT, wild type; ␥5scr, wild type amino acid sequence scrambled; ␥5Ala, the wild type sequence NPFK has been substituted with Ala residues; ␥5⌬, deletion of 10 residues upstream of the C-terminal Cys. B and C, M2-stimulated GTP␥S binding to ␣o␤1␥5 or ␣o␤1␥5scr. GTP␥S binding with 1 nM M2 and 100 nM ␣o was assayed for 2 min. The reaction is linear at this time point as in Fig. 1  subunit because M2 stimulated GTPase activity of the ␤␥ complexes alone was not detectable. (ii) The difference in activity was not due to variation in protein concentrations. When M2stimulated GTPase activity was measured at various concentrations of ␤1␥5 (but constant concentrations of ␣o and M2), a 2-3-fold increase in ␤␥ complex elicited the magnitude of increase in activity seen between the ␤␥ mutant and wild type. Thus a 2-3-fold difference in concentration between ␤1␥5 and ␤1␥5scr must remain undetected. However, we could clearly detect 2-fold differences in the concentration of ␤␥ subunits by densitometry of Coomassie Blue-stained proteins in SDS-PAGE gels (Fig. 1B). (iii) The difference was not due to differences in functional proportion of ␤␥ complexes because in the GTP␥S binding assay, a 2-fold difference in ␤␥ concentration elicits an equivalent increase in GTP␥S binding ( Fig. 2A). (iv) Thin layer chromatography of ␤␥ complex samples indicated that both samples contained the same concentrations of detergent. (v) Buffer components in the ␤␥ complexes did not contribute to the differential activity because the addition of heatdenatured mutant sample to the wild type and vice versa had no effect. (vi) The differences in M2 activation of wild type and mutant ␣o␤1␥5 and ␣o␤1␥5scr also cannot be due to differences in the proportion of prenylated ␥5 subunit because the ␤␥5 and ␤␥5scr proteins were purified using a His-tagged ␣i subunit. Prenylation is essential for ␤␥ complex interaction with the ␣ subunit. (vii) New stocks of ␤␥ complexes that were independently expressed, purified, and assayed again showed the same difference in M2-stimulated ␣o GTPase activity.
Wild Type and Mutant ␤1␥5 Have Similar Affinities for ␣o-To examine the affinities of ␤1␥5 and ␤1␥5scr for ␣o in the 1-10 nM concentration range used in the GTPase assays, we used a recently developed assay (12) that relies on the overlap in binding sites for ␣ subunits and PLC-␤2/3 on the ␤␥ complex. Thus binding of ␣o to the ␤␥ complex inhibits PLC-␤3 stimulation by the ␤␥ complex. There was no significant difference in the inhibition of ␤1␥5 and ␤1␥5scr, indicating that ␣o affinity for both is the same (Fig. 4). This result also further confirmed that both the wild type and mutant ␥ subunits are prenylated to the same extent because prenylation is essential for ␤␥ complex activation of PLC-␤. The difference in GTPase activity between mutant and wild type thus arises from differential receptor interaction.
The results from the experiment above (Fig. 1C), where ␥5 peptide interaction with M2 was tested, indicated that the ␥5 scrambled peptide does not effectively interact with the receptor. An earlier report also indicated that in contrast to the wild type, the scrambled ␥5 peptide had little effect on a muscarinic receptor modulation of a Ca 2ϩ current or on a the muscarinic receptor modulation of an excitatory postsynaptic current (3). Viewed in this context, the higher M2-stimulated GTPase ac-tivity in the ␣o␤1␥5scr mutant indicates that weak interaction of the mutant ␥ subunit C terminus with M2 leads to a higher nucleotide exchange rate.
The receptor-G protein activation cycle involves three broad steps: (i) G protein binding to receptor; (ii) receptor-initiated nucleotide exchange in the G protein and dissociation of the ternary complex; and (iii) deactivation of G␣ by RGS protein and reassociation with the ␤␥ complex. The difference in receptor-stimulated GTPase rates between the G o wild type and mutant must arise at one of these steps. We infer that differences between mutant and wild type proteins occurs during receptor-initiated nucleotide exchange in the G protein. Other steps in the receptor-G protein activation cycle cannot be affected because the results presented indicate that heterotrimer formation of the mutant is unaltered and that RGS4 has no influence on the differential activation of the mutant. The affected step cannot be initial binding of G protein with receptor because this would lead to lower GTPase rates in the mutant compared with wild type.
A model that explains the results here must take into account the effect of the ␥ subunit mutation on receptor-stimulated nucleotide exchange. It should also take into account (i) the inability of the scrambled ␥5 peptide to interact with the receptor and (ii) the higher receptor-stimulated GTPase in mutant G o containing ␥5scr.
The crystal structure of the G protein heterotrimer (G t ) when compared with the structures of G␣t bound to GTP or GDP indicates that the domains on the ␣ subunit that undergo the most significant changes in conformation during activation are the same domains that contact the ␤ subunit (4). It was suggested based on this that the ␤␥ complex occludes nucleotide release from the ␣ subunit when the G protein is bound to the receptor (4). Receptor-stimulated nucleotide exchange will  a M2-stimulated GTPase activity was measured at the following time points for various molar ratios: 0.3 and 0.6, 20 min; 1 and 2, 10 min; 4 -5 min in the presence of 1 mM. In all cases reaction rates were linear at these time points. The turnover numbers in the absence of agonist were 0.03-0.06 at all molar ratios tested both in the presence and the absence of RGS4. b WT, ␣o␤1␥5; SCR, ␣o␤1␥5scr. c Means Ϯ S.E. from at least four independent experiments performed in duplicate. All differences are significant at p Ͻ 0.05 except the reaction without RGS4 at a molar ratio of 4. therefore require the receptor to shift the ␤␥ complex away from the ␣ subunit (Fig. 5A). This was later detailed as a model for receptor activation of a G protein (16). In this model it was proposed that two mechanisms account for the ability of the receptor to stimulate nucleotide exchange in the G protein ␣ subunit: (i) the ␣ subunit C terminus contacts a receptor, and receptor activation leads to conformational changes being triggered through the C-terminal domain to the ␤6/␣5 loop of the ␣ subunit, which is in contact with GDP; (ii) receptor loop(s) enter the cavity between the ␣ subunit and ␤␥ complex, prising them apart and creating an opening through which GDP leaks out. It was proposed in this model that only the ␣ subunit C terminus contacts the receptor and that the C terminus of the ␥ subunit interacts with membranes, despite evidence to the contrary (6,17). Recently, this model has been modified to include the evidence indicating ␥ subunit interaction with a receptor (3,6,12,17) to propose that the ␥ subunit C terminus also interacts with the receptor and that the receptor loops, instead of entering the cavity between the ␣ subunit and ␤␥ complex, actually prise the subunits apart through their interaction with the ␣ and ␥ subunit C termini (18). This model is now similar to our earlier proposal that interaction with the ␣ subunit and ␥ subunit C termini with a receptor is a requirement for G protein activation (2,12). Such a model would predict that mutating the ␥ subunit C terminus would result in weaker activation of the G protein by the receptor. However, earlier results indicate that this may be a simplistic expectation. Although a ␥5 but not a ␥7 subunit-specific peptide inhibits muscarinic receptor-stimulated signaling (3), the M2 mus-carinic receptor stimulates higher nucleotide exchange in G o containing ␥7 compared with ␥5 (12). These results suggested that contrary to expectations, poor interaction of the ␥ subunit with a receptor can result in more robust G protein activation. This result is not surprising if potential mechanisms underlying nucleotide exchange are considered based on the crystal structures of a G protein. Crystal structures of the heterotrimer and active and inactive ␣ subunits indicate that guanine nucleotide release requires the ␤␥ complex to move away from the ␣ subunit (4). Any orientation of the ␤␥ complex that results in exposing the bound GDP will therefore result in more rapid nucleotide release. If the ␥ subunit does not effectively interact with the receptor and anchor the ␤␥ complex, the ␤␥ complex may be inappropriately oriented with reference to the ␣ subunit. The inappropriately oriented ␤␥ complex will enhance nucleotide exchange by exposing the bound nucleotide in the ␣ subunit. Fig. 5B shows one such potential orientation of a ␤␥ complex where the ␥ subunit C terminus does not effectively anchor the ␤␥ complex by interacting with the receptor. As shown in Fig. 5B, nucleotide exchange in the ␣ subunit will be facilitated by this orientation of the ␤␥ complex.
This model predicts that a mutant ␥ subunit that interacts weakly with a receptor will encourage higher nucleotide exchange in the associated ␣ subunit compared with a wild type ␥ subunit that interacts strongly with the receptor. This prediction is borne out by the results presented here. Peptide evidence indicates that the ␥5scr mutant interacts poorly with the receptor compared with the wild type. However, G o containing the ␥5scr mutant shows a higher rate of receptorstimulated nucleotide exchange compared with the wild type. The model also predicts that a ␥ subunit type with a lower affinity for the receptor will allow higher rates of receptorstimulated nucleotide exchange in the associated ␣ subunit compared with a different ␥ subunit type with a higher affinity for the receptor. This prediction is also supported by previous results. Only the C-terminal peptide specific to ␥5 but not ␥7 disrupts muscarinic receptor regulation of Ca 2ϩ current in neurons (3). This result indicated that the ␥5 subunit type but not ␥7 interacts with the M2/M4 receptor types. However, ␣o␤1␥7 shows significantly higher M2-stimulated nucleotide exchange compared with ␣o␤1␥5 (12). Thus, consistent with the results obtained with the ␥5scr mutant, a G protein containing a ␥ subunit type that does not interact effectively with a receptor shows enhanced receptor-stimulated GTPase activity in comparison with the wild type.
Although ␣o␤1␥5scr contains a mutant ␥ subunit that does not effectively interact with the receptor, it still possesses higher receptor-stimulated nucleotide exchange compared with wild type. This result does not imply that a model for receptor activation of a G protein that invokes receptor loops contacting the ␣ and ␥ subunit C termini and prising the ␤␥ complex away from the ␣ subunit is incorrect. It does suggest that mutational approaches with purified proteins may not provide direct evidence for such a model because any mutant that disturbs the orientation of the ␤␥ complex with reference to the ␣ subunit can potentially encourage nucleotide exchange.
As mentioned before, there is evidence for the ␣ subunit N and C termini interacting with the receptor. Inspection of the crystal structure for the heterotrimeric G protein indicates that the ␣ subunit N terminus, C terminus and the ␥ subunit C terminus lie roughly along the same axis and can be oriented toward the plane of the membrane (Fig. 5). The distance between the C termini of ␣ and ␥ subunits cannot be estimated precisely in the crystal structure because at least seven residues at the ␣ subunit C terminus are not resolved, and the ␥ subunit is devoid of the prenyl moiety as well as three residues FIG. 5. Model of G protein interaction with receptor. A, The ␥ subunit C terminus (␥ C terminus) interaction with the receptor (gray arrow) allows the ␤␥ complex to orient itself appropriately with reference to the ␣ subunit. Receptor-initiated nucleotide exchange from the ␣ subunit (GDP and GTP arrows) is regulated by the appropriate orientation of the ␤␥ complex with reference to the ␣ subunit. ␣ subunit C terminus (␣ C terminus) contact with the receptor is denoted with a gray arrow. The space-filling model of GDP bound to the ␣ subunit is shown. B, ineffective interaction of the ␥ subunit C terminus with the receptor affects orientation of the ␤␥ complex with reference to the ␣ subunit and increases the rate of nucleotide exchange during activation (gray arrows). G protein structure is from Lambright et al. (4). at its C terminus (4). The NMR structure of an 11-amino acid peptide specific to the ␣ subunit C terminus indicates that this domain forms a constrained structure in the presence of a receptor (19). It is thus likely that the distance between the ␥ subunit C terminus that includes the prenyl group and the C terminus of the ␣ subunit is less than 40 Å. The crystal structure of inactive rhodopsin that has been determined recently indicates that the intracellular portions of rhodopsin are folded such that the longest distance across the intracellular domains is more than 40 Å (20). This indicates that the exposed surface of the receptor will be sufficient for both the ␣ and ␥ subunit C termini to contact the receptor simultaneously. However, this may not be a requirement if the two domains contact the receptor in a temporal sequence. We propose that by interacting with the receptor, the ␥ subunit appropriately positions the ␤␥ complex with reference to the ␣ subunit. This can allow the receptor to regulate nucleotide exchange at a site in the ␣ subunit that does not contact the receptor directly.