Differential Activity of the G Protein b 5 g 2 Subunit at Receptors and Effectors*

The G protein b 5 subunit differs substantially in amino acid sequence from the other known b subunits suggesting that bg dimers containing this protein may play specialized roles in cell signaling. To examine the functional properties of the b 5 subunit, recombinant b 5 g 2 dimers were purified from baculovirus-infected Sf9 insect cells using a strategy based on two affinity tags (hexahistidine and FLAG) engineered into the N terminus of the g 2 subunit ( g 2HF ). The function of the pure b 5 g 2HF dimers was examined in three assays: activation of pure phospholipase C- b in lipid vesicles; activation of recombinant, type II adenylyl cyclase expressed in Sf9 cell membranes; and coupling of a subunits to the endothelin B (ET B ) and M 1 muscarinic receptors. In each case, the efficacy of the b 5 g 2HF dimer was compared with that of the b 1 g 2HF dimer, which has demonstrated activ- ity in these assays. The b 5 g 2HF dimer activated phospholipase C- b with a potency and efficacy similar to that of b 1 g 2 or b 1 g 2HF ; however, it was markedly less effective than the b 1 g 2HF or b 1 g 2 dimer in its ability to activate type II adenylyl cyclase (EC 50 of ; 700 n M versus 25 n M ). Both the b 5 g 2HF and the b 1 g 2HF dimers supported cou- pling of M 1 muscarinic receptors to the G q a subunit. The ET B receptor coupled effectively to both the G i and G q a subunits in data role for 2HF determine of carboxyl g deconvoluted of g the two molecular One molecular of 10,014 Da, with full processing of the g 2HF subunit: re- moval of the three carboxyl-terminal amino acids (-AIL), addition of a geranylgeranyl lipid to the carboxyl-terminal cysteine, and of a carboxylmethyl group to the carboxyl terminus. The other species, with a molecular mass of 10,024 Da, corresponded to a g 2HF subunit that had undergone no carboxyl-terminal processing. Both species had molecular weights consistent with the removal of the amino-terminal methionine and acetylation of resulting amino-terminal alanine.

Complex biochemical mechanisms exist to discriminate, integrate, and modulate a cell's response to the hormones, autacoids, neurotransmitters, and growth factors in their environment. One of the best characterized signal transduction systems is the pathway used by receptors coupled to heterotrimeric G proteins 1 (1)(2)(3). Current understanding of this signal-ing pathway shows it to be surprisingly complex with large families of proteins comprising the receptors, G proteins, and effectors (1,4,5) and with most cell types expressing multiple isoforms of each category. Many receptors are known to signal through several G proteins. In turn, many effectors receive signals from different isoforms of both the ␣ and ␤␥ subunits of G proteins (5,6). Thus, an important issue in cell signaling is identification of the protein-protein interactions underlying the cellular responses to a particular stimulus.
It is clear that a cellular response is the summation of many molecular interactions. A primary determinant of the outcome is selective expression of signaling molecules, as occurs in the rod outer segment of the eye where the visual receptor, rhodopsin, and its G protein transducin are the major signaling proteins in the membrane (7). In less specialized cells, compartmentation or targeting of the participants in a signaling cascade may determine the specificity observed in vivo (2,8). Indeed, several lines of investigation have suggested specific localization of receptors, G proteins, and effectors within the cell that may lead to selective and efficient interactions between members of a signaling cascade (9). Moreover, critical steps in the activation of G proteins, such as the rate of exchange of GTP for GDP on the ␣ subunit, are highly regulated processes wherein receptors, effectors, and accessory proteins such as RGS molecules are all involved in determining the kinetics of GTP/GDP exchange (10 -12). At the G protein-effector level, it has long been recognized that certain ␣ subunits selectively regulate specific effectors (2)(3)(4), and recently, defined isoforms of ␤␥ dimers have been reported to differentially activate effectors (13,14). Finally, the large diversity in the G protein ␣ and ␤␥ subunits suggests that there are important protein-protein interactions involving the ␣ and ␤␥ subunits of the heterotrimer that determine specificity (4,9). Indeed, selective interaction of ␣ and ␤␥ isoforms has been observed in vitro between the ␤ 5 subunit and ␣ subunits in the G q family (15). As receptors coupled to the ␣ q subunit regulate important functions in most all tissues, this latter finding makes it important to determine the functional properties of ␤␥ dimers containing the ␤ 5 subunit. To date, there is no information about the receptors that couple to dimers containing the ␤ 5 subunit and release this potentially unique signal.
Only limited data are available about the functionality of ␤␥ dimers containing the ␤ 5 subunit. However, both the ␤ 5 subunit and its splice variant, ␤ 5L , have been shown to stimulate PLC-␤ 2 activity when transfected into COS-7 cells with the ␥ 2 subunit (16,17). Interestingly, the ␤ 1 ␥ 2 dimer markedly activates the mitogen-activated protein kinase pathway in COS cells, whereas the ␤ 5 ␥ 2 dimer does not (13). Moreover, the ␤ 1 ␥ 2 dimer stimulates type II adenylyl cyclase and inhibits type I adenylyl cyclase when co-transfected into these cells, but the ␤ 5 ␥ 2 subunit appears to inhibit both type I and type II cyclase activity (14). These observations suggest that dimers containing the ␤ 5 subunit may have different functions in cell signaling from those containing other ␤ subunits.
In the experiments reported here, we have taken two approaches toward examining the function of the ␤ 5 ␥ 2 dimer. Recombinant dimers were purified using two affinity tags (hexahistidine and FLAG) engineered into the N terminus of the ␥ 2 subunit (␥ 2HF ). In one approach, we examined the ability of the purified dimer to activate two effectors in vitro, PLC-␤ and type II adenylyl cyclase. In a second approach, we compared the ability of ␤ 5 ␥ 2HF and ␤ 1 ␥ 2HF dimers to support coupling of ␣ subunits with the M 1 muscarinic and ET B receptors. The M 1 muscarinic receptor was chosen because it couples selectively to G q ␣ subunits (18,19) and the ET B receptor because it couples to both the G i and G q ␣ subunits (20,21). The results show that the ␤ 5 ␥ 2HF dimer is able to fully activate PLC-␤ but is much less effective at activating type II adenylyl cyclase. The ␤ 5 ␥ 2HF dimer supports coupling of G q ␣ subunits to M 1 muscarinic and ET B receptors with an efficacy similar to that of the ␤ 1 ␥ 2HF dimer. However, in keeping with its apparent lower affinity for the G i ␣ subunit (15), the ␤ 5 ␥ 2HF dimer does not support coupling of the G i ␣ subunit to the ET B receptor. Overall, the data suggest that the ␤ 5 ␥ 2 dimer couples selectively to ␣ q -linked receptors and may only interact with a limited set of effectors. Thus, in tissues such as brain where it is highly expressed, this dimer may indeed have specialized functions in cellular signaling.

Construction of Recombinant Baculoviruses for Endothelin
Receptors-A clone encoding the human ET B receptor (GenBank TM accession no. L06623) was the kind gift of Dr. Ponal Nambi at SmithKline Beecham Pharmaceuticals (21). The receptor cDNA was excised from pBluescript (Stratagene) with SacI and KpnI, further digested with MspI and subcloned into the pCNTR shuttle vector. The ET B coding sequence was excised with BamHI and subcloned into the pVL1393 baculovirus transfer vector. A recombinant baculovirus encoding the receptor was obtained by co-transfecting the transfer vector with linearized viral DNA into Sf9 cells using the Pharmingen BaculoGold ® kit (15). The recombinant baculoviruses were purified by one round of plaque purification (22). The construction of recombinant baculoviruses coding for the G i1 , G q , G s ␣ subunits, the ␤ 1 , ␤ 1HF , ␤ 5 , ␥ 2 and ␥ 2HF subunits have been described (15,(23)(24)(25). The baculovirus for the M 1 muscarinic receptor was the kind gift of Dr. Elliott M. Ross.
Expression and Preparation of Plasma Membranes Containing Recombinant Receptors-Sf9 cells were infected with recombinant baculovirus encoding the ET B or M 1 muscarinic receptor at a multiplicity of infection of 3 and the incubation continued for 48 -60 h (26). Washed cell pellets were resuspended in either ET B membrane homogenization buffer (25 mM Hepes, pH 7.5, 1% (w/v) glycerol, 100 mM NaCl, 17 g/ml PMSF, 20 g/ml benzamidine, and 2 g/ml of aprotinin, leupeptin, and pepstatin A) or M 1 membrane homogenization buffer (20 mM Hepes, pH 8.0, 2 mM MgCl 2 , 1 mM EDTA, 17 g/ml PMSF, 2 g/ml aprotinin, and 10 g/ml leupeptin) (27) and stored frozen at Ϫ70°C. All subsequent steps were at 4°C or on ice. Cell pellets were thawed in ice-cold membrane homogenization buffer and burst by N 2 cavitation (600 psi, 20 min). The cell lysate was centrifuged for 10 min at 750 ϫ g. The low speed supernatant was centrifuged for 30 min at 28,000 ϫ g. To reduce the level of endogenous Sf9 G proteins, the pelleted membranes were resuspended in membrane resuspension buffer (50 mM Hepes, pH 7.5, 1 mM EDTA, 3 mM MgSO 4 ) containing 7 M urea and incubated for 30 min (28 -30). The suspension was then centrifuged at 142,000 ϫ g for 30 min. The stripped membranes were washed twice with membrane resuspension buffer and resuspended in membrane homogenization buffer at 5-15 mg of protein/ml, frozen, and stored at Ϫ70°C. The B max and K d values of the receptors were measured before and after urea treatment. The binding parameters for recombinant ET B receptors were determined using 125 I-endothelin-1 in a buffer containing 10 mM Hepes, pH 7.4, 5 mM MgCl 2 , 1 mM EDTA, and 0.1% bovine serum albumin using our published techniques (31) and the conditions described pre-viously (32). Nonspecific binding was determined using 10 M unlabeled endothelin-1 and varied from 13% at 0.01 nM to 90% at 100 nM endothelin-1. Binding of the antagonist ligand [ 3 H]QNB to the M 1 muscarinic receptor was performed as described (33) using a concentration range of 0.5-50 nM [ 3 H]QNB. Nonspecific binding was determined using 10 M atropine and was less than 10% at 50 nM [ 3 H]QNB. The B max and K d were determined by fitting the data to a rectangular hyperbola using the nonlinear least squares routines in the GraphPad Prizm ® software. The binding parameters of the membranes as determined before and after urea treatment were as follows: ET B receptors (before B max , 27 pmol/mg membrane protein; K d 0.05 nM; after B max , 100 pmol/mg membrane protein; K d 0.1 nM); M 1 muscarinic receptors (before B max , 2.6 pmol/mg membrane protein; K d 0.15 nM; after B max , 2 pmol/mg membrane protein; K d 0.4 nM). These data indicate that urea treatment did not greatly change the affinity for ligand.
Expression and Purification of Recombinant G Protein ␣ and ␤␥ Subunits-The recombinant G i1 ␣ subunit and ␤ 1 ␥ 2 and ␤ 1 ␥ 2HF dimers were purified from baculovirus-infected Sf9 cells as described previously (26,34). The purification of G q ␣ was based on the method of Biddlecome et al. (35). The G q ␣ subunit was expressed in Sf9 cells in combination with the ␤ 1HF ␥ 2HF dimer at a multiplicity of infection of 3 each. The cells were harvested 48 h after infection, washed three times with insect phosphate-buffered saline, resuspended in a few milliliters of Q lysis buffer (20 mM Tris, pH 8.0, 10 M GDP, 17 g/ml PMSF, and 2 g/ml pepstatin, leupeptin, and aprotinin), frozen in liquid nitrogen and stored at Ϫ70°C. All extractions and centrifugation steps were performed at 4°C or on ice. Cell pellets were thawed in half the original culture volume of Q lysis buffer and lysed by nitrogen cavitation (26). The lysate was centrifuged (24,000 ϫ g, 20 min) and the pellet resuspended in one-fourth the previous volume of Q lysis buffer containing 10 g/ml DNase and 5 mM MgCl 2 with a Dounce homogenizer. After a 15-min incubation period, the suspension was rehomogenized and repelleted. The washed pellet was resuspended in Q extraction buffer (1% cholate, 20 mM Tris, pH 8.0, 100 mM NaCl, 5 mM ␤-mercaptoethanol, 10 M GDP, PMSF, and 2 g/ml pepstatin, leupeptin, and aprotinin) at 5 mg/ml total protein. The suspension was stirred for 1 h and then centrifuged (142,000 ϫ g, 45 min). The cholate extract was frozen and stored at Ϫ70°C. The Q chromatography buffer used to purify the G q ␣ subunit contained 20 mM Hepes, pH 8.0, 100 mM NaCl, 1 mM MgCl 2 , 5 mM ␤-mercaptoethanol, 17 g/ml PMSF, and 2 g/ml each leupeptin and aprotinin. The cholate extract (ϳ100 ml) from 1.5 ϫ 10 9 Sf9 cells was diluted with 400 ml of loading buffer (Q chromatography buffer containing 0.5% (v/v) Genapol C-100 and 10 M GDP) and applied to a 15-ml Ni 2ϩ -NTA superflow agarose resin column equilibrated in loading buffer. The loaded column was washed sequentially with 75-ml volumes of loading buffer, loading buffer containing 0.5 M NaCl, loading buffer containing 0.5 M NaCl and 10 mM imidazole, loading buffer containing 1 M NaCl, and loading buffer. The column was warmed to room temperature and subsequent wash and elution buffers were applied at room temperature. The column was washed with 75 ml of loading buffer containing 0.2% cholate and 3 M GTP␥S, 75 ml of Q chromatography buffer containing 0.3% cholate and finally eluted with Q chromatography buffer containing 1% cholate, 10 M GDP, 20 M AlCl 3 , 10 mM MgCl 2 , and 10 mM NaF. Fractions containing the G q ␣ subunit were pooled and concentrated on ice using an Amicon ultrafiltration apparatus and a PM10 membrane. The concentrated protein was diluted 10-fold with Q chromatography buffer containing 0.4% (w/v) cholate with 1 mM dithiothreitol substituted for 5 mM ␤-mercaptoethanol and re-concentrated. This procedure was repeated three times. The retentate (ϳ1 ml) was further concentrated to approximately 100 l in a spin concentrator (Microcon 10, Amicon). The spin concentrator was pretreated with 0.1% bovine serum albumin in phosphatebuffered saline overnight at room temperature and rinsed twice by centrifugation with deionized water. Approximately 10 g of highly purified ␣ q was recovered from 1.5 ϫ 10 9 cells.
The ␤ 5 ␥ 2HF subunit was generated by infecting Sf9 cells with the ␤ 5 and ␥ 2HF viruses at a multiplicity of infection of 3 each (23). The washed cell pellet was resuspended in a few milliliters of histag lysis buffer (20 mM Hepes, pH 7.4, 150 mM NaCl, 1 mM ␤-mercaptoethanol, 3 mM MgCl 2 , 17 g/ml PMSF, 20 g/ml benzamidine, and 2 g/ml aprotinin, leupeptin, and pepstatin), frozen, and stored at Ϫ70°C. The cell pellet was thawed and lysed by nitrogen cavitation in ice-cold histag lysis buffer (26). The cell lysate was centrifuged (10 min, 750 ϫ g) and membranes were pelleted (100,000 ϫ g, 30 min) from the low speed supernatant. The membrane pellet was resuspended in Genapol extraction buffer (histag lysis buffer containing 0.1% (v/v) Genapol C-100) at a ratio of 1 ml of buffer/20 mg of wet pellet weight and stirred on ice for 1 h. The detergent extract was centrifuged (100,000 ϫ g, 60 min), the supernatant decanted, frozen in liquid nitrogen and stored at Ϫ70°C. Typically, 50 ml of Genapol extract (prepared from 5 gm of cell pellet) was applied to a 5-ml FLAG M2 column equilibrated in Genapol extraction buffer, washed with 30 ml of Genapol extraction buffer, and eluted with 30 ml of 400 M FLAG peptide in Genapol extraction buffer. This step was repeated with a second aliquot of Genapol extract. The elution fractions were pooled, the Genapol concentration adjusted to 0.5% (v/v) by addition of histag lysis buffer containing 2% Genapol C-100, and the pool applied to a 2.5-ml Ni 2ϩ -NTA agarose column equilibrated in histag-loading buffer (histag lysis buffer containing 0.5% (v/v) Genapol). The column was washed with the following buffers: 12.5 ml of histag loading buffer; 12.5 ml of histag-loading buffer plus 350 mM NaCl; 25 ml of histag-loading buffer plus 350 mM NaCl and 10 mM imidazole; 10 ml of histag-loading buffer plus 1 M NaCl; and 10 ml of histag-loading buffer. The bound ␤ 5 ␥ 2HF was eluted with 10 ml of histag-loading buffer containing 250 mM imidazole. The eluted dimers were diluted onefourth with CHAPS dilution buffer (20 mM Hepes, pH 7.5, 1 mM EDTA, 3 mM MgCl 2 , 0.1% (w/v) CHAPS, 1 mM dithiothreitol), applied to a 1-ml HiTrap Q column (Amersham Pharmacia Biotech) equilibrated in the same buffer, washed with 10 ml of buffer and eluted with CHAPS dilution buffer containing 500 mM NaCl. The highly purified ␤ 5 ␥ 2HF was stored at Ϫ70°C. Approximately 100 g of the pure ␤ 5 ␥ 2HF dimer was obtained from a 10-gm Sf9 cell pellet.
Analysis of the Post-translational Processing of the ␥ Subunit by Mass Spectrometry-The purified ␤ 5 ␥ 2HF dimer was analyzed by electrospray mass spectrometry to determine the extent of modification of the carboxyl terminus of its ␥ subunit (36). The deconvoluted mass spectrum of the ␥ 2HF derived from the dimer indicated that the sample contained two molecular weight species. One species, with a molecular mass of 10,014 Da, is consistent with full processing of the ␥ 2HF subunit: removal of the three carboxyl-terminal amino acids (-AIL), addition of a geranylgeranyl lipid to the carboxyl-terminal cysteine, and addition of a carboxylmethyl group to the carboxyl terminus. The other species, with a molecular mass of 10,024 Da, corresponded to a ␥ 2HF subunit that had undergone no carboxyl-terminal processing. Both species had molecular weights consistent with the removal of the amino-terminal methionine and acetylation of the resulting amino-terminal alanine. Based on analysis of two different ␤ 5 ␥ 2HF preparations, approximately half of the ␥ 2HF subunits in the sample had a fully processed carboxyl terminus and were therefore capable of high-affinity interactions with ␣ subunits, receptors, and effectors.

Reconstitution of Recombinant Receptors with Purified G Protein ␣ and ␤␥ Subunits and Measurement of GTP␥S Binding-Urea-treated
Sf9 cell membranes expressing recombinant receptors were washed with reconstitution buffer (25 mM Hepes, pH 7.4, 100 mM NaCl, 5 mM MgCl 2 , 1 mM EDTA, 0.1% bovine serum albumin), and resuspended at 0.5-1 mg of protein/ml in reconstitution buffer. Typically, 50 l of membrane suspension containing 2.7 pmol of recombinant endothelin receptors (measured as 125 I-endothelin binding sites) or 120 fmol of recombinant M 1 muscarinic receptors (measured as [ 3 H]QNB binding sites) was mixed with 400 l of reconstitution buffer containing 1 M AMP-PNP and incubated on ice for 15 min. The ␣ subunit was diluted into the mix before addition of the ␤␥ dimer, then concentrated GDP was added to give a final concentration of 50 nM. The final concentration of G␣ and ␤␥ added to the membranes was 5-14 nM and 5-47 nM, respectively, and the receptor:G protein ratio was approximately 1:2 for the ET B receptor and 1:40 for the M 1 muscarinic receptor. The final concentration of CHAPS was held to less than 0.006% (w/v), and in the case of the ␣ q experiments, the final cholate concentration was held to less than 0.001% (w/v). The receptor:G protein mixture was incubated on ice for 45 min followed by a 10 min equilibration at 25°C. The assay was begun by the addition of [ 35 S]GTP␥S to a final concentration of 7 nM. The ligand was added 8 min later, and membrane aliquots removed for filtration through nitrocellulose filters (Millipore, HAWP-025) every 60 s.
Measurement of Phospholipase C-␤ and Adenylyl Cyclase Activity-The ability of ␤␥ dimers to stimulate PLC-␤ was measured exactly as described (25). The ability of the ␤␥ subunits to activate adenylyl cyclase was measured using Sf9 insect cell membranes overexpressing recombinant, rat type II adenylyl cyclase (37). The assay was performed as described (15,25) except that it was incubated 7 min at 30°C.
Calculations and Expression of Results-Experiments presented under "Results" are representative of 3 or more similar experiments. Data expressed as dose-response curves were fit to sigmoid curves using the fitting routines in the GraphPad Prizm ® software. Curves were fit to the equation representing a single binding site. Statistical differences between the curves were determined using all the individual data points from multiple experiments to calculate the F statistic as described (38). The initial slope of agonist induced GTP␥S binding in Figs. 4 -6 was calculated by analyzing the first 5 min of data using a linear regression routine.
Materials-All reagents used for the culture of Sf9 cells, the construction of baculovirus and the expression and purification of G proteins have been described in detail (15,23,34

RESULTS
We previously reported that the ␣ q subunit bound selectively to immobilized ␤ 5 ␥ 2HF dimers (15). The object of the current study was to examine the functional consequences of this selectivity using assays of ␤␥ interaction with ␣ subunits and receptors or directly with effectors. To achieve this goal, we required pure, recombinant ␣ q subunits and ␤ 5 ␥ 2 dimers. As our earlier work determined that the ␤ 5 ␥ 2 dimer has a very low affinity for the ␣ i1 subunit (15), our usual ␤␥ purification strategy (DEAE chromatography followed by ␣ i -agarose affinity chromatography) (34) could not be used to purify the ␤ 5 ␥ 2 dimer. Therefore, the ␤ 5 subunit was paired with a ␥ 2HF subunit to provide a rapid and efficient means for purification of functional ␤ 5 ␥ 2HF dimers (see Fig. 1). The ␥ 2 subunit was chosen because other dimers containing this isoform are active in most assays of ␤␥ function (13,14,16,17). Moreover, previous experiments have shown that affinity tags on the N terminus of either ␤ or ␥ subunits do not inhibit the activity of ␤␥ dimers (40 -42).
The data in Fig. 1 provide information about the purity of the ␣ and ␤␥ subunits used in this study. Fig. 1A presents a silver-stained, SDS-polyacrylamide gel resolving the proteins from the final three steps in the purification of the ␤ 5 ␥ 2HF dimer. The arrows on the left side of the figure indicate the migration positions of the ␣ and ␤ subunits of the pure brain heterotrimer used as a molecular weight standard. The arrows on the right side of the figure indicate the migration position of the purified ␤ 5 and ␥ 2HF subunits. Note that the ␤ 5 subunit migrates with most ␣ subunits because of the additional amino acids in its sequence (16). The second lane shows the proteins eluted from the FLAG antibody column with 400 M FLAG peptide. The product contains about four contaminating proteins, which are removed by the Ni 2ϩ -NTA column. The final step in the purification used anion exchange chromatography in 0.1% CHAPS to: (a) remove the imidazole used to elute the dimer from the Ni 2ϩ -NTA column; (b) to substitute a low concentration of CHAPS for Genapol C-100; and (c) to concentrate the final product. Because the ␤ 5 ␥ 2HF dimer is sensitive to detergents (15), the final product was examined for both subunits via immunoblotting. The immunoblot shown in Fig. 1B clearly demonstrates that the purified ␤␥ dimer contains both the ␤ 5 and ␥ 2HF subunits. As prenylation of the ␥ subunit is essential for proper function of the ␤␥ dimer (36,43), the purified product was analyzed by electrospray mass spectrometry (see "Experimental Procedures"). About half of the ␥ subunits in the dimer were fully processed. The fraction of ␥ subunits that are fully processed in the ␤ 5 ␥ 2HF preparation is less than that found in ␥ subunits in dimers purified over ␣-agarose affinity columns (36). However, this might be expected as the ␣ subunit selects for prenylated ␤␥ dimers (36,43). However, the overall quality of the ␤ 5 ␥ 2HF preparation obtained using the affinity tags is comparable with ␤␥ dimers purified from ␣-agarose columns. Finally, the silver-stained, SDS-polyacrylamide gel presented in Fig. 1C shows the purity of the recombinant ␣ q subunit prepared via the method of Biddlecome et al. (35) as described under "Experimental Procedures." The purified ␤ 5 ␥ 2HF dimers were tested for their ability to activate two effectors, PLC-␤ and type II adenylyl cyclase. The effect of both the ␣q subunit (also used in the receptor-coupling experiments) and the ␤ 5 ␥ 2HF dimer on the activity of PLC-␤ is shown in Fig. 2, A and B. Recombinant turkey PLC-␤ was used in these studies because it is highly sensitive to the ␤␥ subunit and has a low basal activity (44,45). The data in Fig. 2A show that, as expected, recombinant ␣ q activated PLC-␤ about 11fold in the presence of AIF 4 Ϫ . This ␣ q preparation also activated recombinant, human PLC-␤1 3-fold in the presence of AIF 4 Ϫ (data not shown). The data in Fig. 2B indicate that all three of the ␤␥ dimers tested could activate PLC-␤ about 20-fold. It is important to note that the ␤ 5 ␥ 2HF dimer was as effective and almost as potent as either the ␤ 1 ␥ 2 or the ␤ 1 ␥ 2HF dimers. The V max values estimated for these activators were about 3 mol/mg PLC/min and the EC 50 values ranged from 5.1 to 13.3 nM (see the legend to Fig. 2). The observation that the ␤ 5 ␥ 2HF dimer is fully effective in the PLC assay provides strong evidence that the two recombinant proteins are properly folded and modified.
The ␤␥ dimer causes a synergistic activation of type II adenylyl cyclase in the presence of GTP␥S activated G s ␣ subunit (46). Therefore, we examined the ability of the ␤ 1 ␥ 2 , ␤ 1 ␥ 2HF and ␤ 5 ␥ 2HF dimers to activate recombinant, type II adenylyl cyclase overexpressed in Sf9 cell membranes. Fig. 3 shows that, in agreement with our previous observations (25), the ␤ 1 ␥ 2 and ␤ 1 ␥ 2HF dimers activated type II adenylyl cyclase about 12-fold with an estimated EC 50 of 25 nM. In contrast, the ␤ 5 ␥ 2HF subunit did not activate adenylyl cyclase until concentrations greater than 100 nM were reconstituted into the assay. A full response curve could not be performed because of limitations on the concentration of the ␤ 5 ␥ 2HF preparation. Thus, the EC 50 for ␤ 5 ␥ 2HF activation can only be estimated. From the data available, the EC 50 appears to be over 700 nM, making the ␤ 1 ␥ 2HF dimer about 30-fold more potent at stimulating type II adenylyl cyclase than the ␤ 5 ␥ 2HF dimer. The observation that FIG. 2. Effect of the G q ␣ and the ␤ 5 ␥ 2HF dimer on the activity of PLC-␤. A, the indicated concentrations of pure, recombinant G q ␣ subunits were reconstituted into phospholipid vesicles containing [ 3 H]phosphatidylinositol 4,5-bisphosphate and their ability to activate turkey PLC-␤ (tPLC-␤) in the presence (Ⅺ) or absence (OE) of AIF 4 Ϫ was measured as described under "Experimental Procedures." B, comparison of the ability of ␤ 1 ␥ 2 (E), ␤ 1 ␥ 2HF (OE) and ␤ 5 ␥ 2HF (Ⅺ) to activate turkey PLC-␤. Data are representative of three independent experiments, each performed in duplicate. Fits of the averaged data to sigmoid curves give the following EC 50 (NM) and V max (mol/mg PLC/min) values: ␤ 1 ␥ 2 , 5.1 and 3.4; ␤ 1 ␥ 2HF , 6.5 and 3.1; and ␤ 5 ␥ 2HF , 13.3 and 3.4. Differences between the ␤ 1 ␥ 2 and ␤ 1 ␥ 2HF curves were not statistically significant. Differences between the ␤ 1 ␥ 2 and ␤ 5 ␥ 2HF curves were statistically significant (p Ͼ 0.01).
FIG. 1. Silver-stained gels and immunoblots depicting the purification of ␤ 5 ␥ 2HF and the G q ␣ subunit from Sf9 cells infected with recombinant baculoviruses. A, proteins in the elution fractions from sequential affinity chromatography on the anti-FLAG antibody column, the Ni 2ϩ -NTA column, and the HiTrap anion exchange column were resolved on SDS-polyacrylamide gel electrophoresis gels and stained with silver. B, the ␤ 5 ␥ 2HF eluted from the HiTrap column with 500 mM NaCl was immunoblotted with three antibodies to demonstrate the presence of both the ␤ 5 (␤ 5 ) and ␥ 2HF subunits (␥ 2 and FLAG) in the final product. C, silver-stained SDS-polyacrylamide gel electrophoresis gel of recombinant ␣ q (50 ng) as purified from Sf9 cells infected with baculoviruses encoding the G q ␣ subunit, ␤ 1HF and ␥ 2HF subunits, using the method of Biddlecome et al. (35) with slight modifications. See "Experimental Procedures" for details. ␤ 1 ␥ 2 and ␤ 1 ␥ 2HF stimulate adenylyl cyclase with equivalent potency implies that the reduced potency of ␤ 5 ␥ 2HF is not simply because of the presence of the affinity tags.
Although the ␤ 5 ␥ 2 subunit has been tested for its ability to activate PLC-␤, adenylyl cyclase and mitogen-activated protein kinase in transient transfection assays (13,14,16), to date, there is no data to indicate which receptors and ␣ subunits couple to the ␤ 5 subunit. The observation that the ␤ 5 ␥ 2 dimer has a selective affinity for the ␣ q subunit (15) suggests that the dimer might be released from G q -linked receptors. Accordingly, we compared the ability of the ␤ 1 ␥ 2HF and ␤ 5 ␥ 2HF dimers to interact with two types of G q -linked receptors, the M 1 muscarinic receptor, which couples specifically with ␣ q (18,19) and the ET B receptor, which couples to both ␣ q and ␣ i (20,21). For these experiments, the receptor was overexpressed in the Sf9 cell membrane, the pure ␣ q and ␤␥ dimers reconstituted into purified membranes, and agonist-stimulated GTP␥S binding measured. The data in Fig. 4 illustrate the ability of the ␤ 5 ␥ 2HF and ␤ 1 ␥ 2HF dimers to support coupling of the ␣ q subunit to the M 1 muscarinic receptor. Fig. 4A shows that neither the ␣ q (triangles) nor ␤ 1 ␥ 2HF (diamonds) subunits alone support receptor coupling, indicating that endogenous Sf9 cell G proteins are not effective in the assay. However, adding both ␣ q and the ␤ 1 ␥ 2HF dimer together (squares) effectively supports receptor coupling as indicated by the marked change in the rate of GTP␥S bound after addition of 1 mM carbachol.
Note that the rate of agonist-stimulated GTP binding to ␣ q tends to decline after 5-6 min of stimulation. Therefore, the initial rates of the reaction were estimated from a linear regression analysis of the first 5 min of data. This rate is represented by the line drawn through the squares to the point of carbachol addition. Calculation of the stoichiometry of GTP binding indicates about 110 fmol of GTP␥S was bound in the first 5 min after agonist addition, representing 5% of the GTP␥S added to the assay mix. If all of the GTP was bound to the ␣ q reconstituted into the membrane, about 4% of the ␣ subunit bound nucleotide. These values are in an appropriate range to make these measurements kinetically valid. Accordingly, the initial rates of agonist-stimulated GTP binding shown in Figs. 5 and 6 were calculated using the same procedure. The basal rates of GTP␥S binding to the membranes for all data sets were also calculated using a linear regression analysis. Fig. 4B shows that the ␤ 5 ␥ 2HF dimer also supports coupling of ␣ q to the M 1 muscarinic receptor. These experiments were performed with 5 nM ␣ q and 20 nM ␤␥ dimers. The carbachol stimulated rate of GTP␥S binding supported by ␤ 5 ␥ 2HF was slightly lower than observed with ␤ 1 ␥ 2HF (135 versus 188 fmol/mg of protein/min). In other experiments, the amount of ␣ q was held constant at 5 nM and the concentration of ␤␥ dimers varied between 5 and 20 nM. Varying the concentration of ␤ 1 ␥ 2HF dimer from 5 to 20 nM did not change the rate of GTP␥S binding, suggesting that a 1:1 ratio of ␣ to ␤␥ was able to provide maximal coupling. However, increasing the ratio of ␣ q :␤ 5 ␥ 2HF from 1:1 to 1:4 did increase the rate of GTP␥S binding from 60 to 135 fmol/mg of protein/min, perhaps because only about half of the ␥ subunits in this preparation are prenylated.
The data in Fig. 5 present similar experiments performed with the ET B receptor reconstituted with the ␣ q subunit and the two ␤␥ dimers. In this experiment, the ␣ q concentration added to the membranes was 14 nM, the ␤ 1 ␥ 2HF concentration was 20 nM and the ␤ 5 ␥ 2HF concentration 47 nM. Note that addition of 20 nM endothelin-1 caused a marked increase in the rate of GTP␥S binding in the presence of either the ␤ 5 ␥ 2HF or the ␤ 1 ␥ 2HF dimer. As was the case with the M 1 muscarinic receptor, the rates of GTP␥S binding induced by agonist are lower in the presence of the ␤ 5 ␥ 2HF dimer (314 fmol/mg of FIG. 3. Comparison of the ability of the ␤ 1 ␥ 2 , ␤ 1 ␥ 2HF and ␤ 5 ␥ 2HF dimers to stimulate type II adenylyl cyclase. Sf9 cells were infected with a recombinant baculovirus encoding the type II adenylyl cyclase, membranes prepared, and the cyclase reaction performed with the indicated concentrations of ␤␥ dimers as described under "Experimental Procedures." Differences between the ␤ 1 ␥ 2 and ␤ 1 ␥ 2HF curves were not statistically significant. Differences between the ␤ 1 ␥ 2HF and ␤ 5 ␥ 2HF curves were statistically significant (p Ͼ 0.001). Data shown is the average of three determinations, each performed in duplicate.

FIG. 4.
Comparison of the ability of the ␤ 1 ␥ 2HF and ␤ 5 ␥ 2HF dimers to support coupling of the G q ␣ subunit to the M 1 muscarinic receptor. A, Sf9 cells were infected with a recombinant baculovirus encoding the M 1 muscarinic receptor, membranes prepared and reconstituted with the indicated ␣ and ␤␥ subunits. The reconstituted membranes were incubated with GTP␥S and stimulated with 1 mM carbachol as described under "Experimental Procedures." The ␣ q concentration was 10 nM and the ␤ 1 ␥ 2HF concentration was 10 nM. B, effect of 20 nM ␤ 5 ␥ 2HF or ␤ 1 ␥ 2HF dimers on coupling the M 1 receptor to the ␣ q subunit. The ␣ q concentration was 5 nM. The solid lines estimate the initial rate of agonist-stimulated GTP␥S binding calculated using linear regression over the first 5 min. Basal rates were determined using linear regression on all data points. Data shown is representative of three to five similar experiments.
protein/min versus 750 with ␤ 1 ␥ 2HF ). In other experiments, the stoichiometry of the ␣:␤␥ ratio was varied from 1:1 to 1:4 to determine whether higher concentrations of ␤␥ dimer supported faster rates of GTP␥S binding (data not shown). In general, increasing the ratio of ␣ q :␤ 1 ␥ 2HF above 1:1 did not increase the rate of GTP␥S binding. However, as with the M 1 muscarinic receptor, increasing the ratio of ␣ q :␤ 5 ␥ 2HF from 1:1 to 1:4 did increase the rate of GTP␥S binding from 83 to 214 fmol/mg of protein/min. Finally, we tested the ability of the two dimers to support coupling of the ET B receptor to the ␣ i1 subunit. These data are presented in Fig. 6, A and B. Initial experiments were performed with 10 nM ␣ i1 and 10 nM ␤␥ subunits. As expected, the ␤ 1 ␥ 2HF dimer effectively supported coupling of the receptor to the ␣ i1 subunit (Fig. 6A, triangles). To determine whether higher amounts of ␤␥ dimer might enhance the effect, the amount of ␣ i1 was held constant at 10 nM and the concentration of ␤ 1 ␥ 2HF increased to 20 nM (diamonds) or 40 nM (circles). Note that increasing the concentration of dimer from 10 to 40 nM had little effect, indicating that effective coupling occurs at a stoichiometry of 1:1. The rates of receptor-stimulated GTP␥S binding obtained with the ET B receptor and the ␣ i1 subunit were above 800 fmol/mg of protein/min and were consistently the highest obtained in this study. In marked contrast, even at an ␣:␤␥ ratio of 1:4 the ␤ 5 ␥ 2HF dimer was not able to couple the ␣ i1 subunit to the ET B receptor (Fig. 6B). This result is consistent with a low affinity of the ␣ i1 subunit for the ␤ 5 ␥ 2HF dimer (15). DISCUSSION This report presents the functional effects of ␤␥ dimers containing the ␤ 5 subunit using pure, recombinant ␤ 5 ␥ 2 dimers. The ability of the purified ␤ 5 ␥ 2 dimer to stimulate two effectors was studied in vitro. The purified dimers were also reconstituted with the G i and/or G q ␣ subunits and two different receptors overexpressed in Sf9 cell membranes to examine the ability of the ␤ 5 subunit to support receptor coupling. The data indicate four important functions of this ␤␥ dimer: 1) the ␤ 5 ␥ 2HF and ␤ 1 ␥ 2HF dimers are equally effective in their ability to activate PLC-␤; 2) the ␤ 5 ␥ 2HF dimer is ineffective at stimulating type II adenylyl cyclase; 3) the ␤ 5 ␥ 2HF dimer can effec-tively support coupling of the G q ␣ subunit to two G q -linked receptors; and 4) the ␤ 5 ␥ 2HF dimer is not able to support coupling of the G i ␣ subunit to the ET B receptor. Because parallel experiments showed the ␤ 1 ␥ 2HF dimer to be effective in each assay, the data strongly suggest that dimers containing the ␤ 5 subunit are released from receptors coupling to ␣ subunits in the G q family and that the ␤ 5 subunit has a more limited role in regulating cell function than does the ␤ 1 subunit. These findings have broad implications because the ␤␥ dimer has many roles in G protein-mediated signaling. In addition to being required for the ␣ subunit to couple to receptors (31,(47)(48)(49), the dimer can regulate the activity of multiple effectors including PLC-␤, K ϩ and Ca 2ϩ channels, phosphatidylinositol 3-kinase, adenylyl cyclase, and the mitogen-activated protein kinase pathway (5,50,51). The dimer also plays a role in the translocation of receptor kinases to the plasma membrane (5). Determination of those effectors regulated by dimers containing the ␤ 5 subunit will provide a much greater understanding of its function in cell signaling.
Little is known about the role of the different ␤ subunits in determining the specificity of these signaling events. The first four subunits identified, ␤ 1 -␤ 4 , are widely expressed, each contain 340 amino acids and are 80 -90% identical in sequence (4). In contrast, the ␤ 5 subunit has an 8-amino acid extension near the amino terminus and 2 short amino acid insertions within the WD repeat regions of the molecule. Overall, it is only 52% identical and 64% similar to ␤ 1 (16). Originally, Northern analysis of murine tissues showed the ␤ 5 subunit to be expressed predominantly in the brain (16). However, ␤ 5 subunit expression has also been detected in rat portal vein by using the polymerase chain reaction (52). The human ␤ 5 subunit has been cloned and analysis of the distribution of this ␤ 5 subunit in human tissues shows it to be widely expressed (53). The ␤ 5 subunits have been demonstrated to form functional dimers with the ␥ 2 , ␥ 3 , ␥ 4 , ␥ 5 , and ␥ 7 subunits (16,17), and the yeast two-hybrid technique also shows an interaction between the ␤ 5 subunit and multiple ␥ subunits (54). Overall, these data suggest that the ␤ 5 subunit may be involved in many signaling systems.
Recently, two splice variants of ␤ subunits have been discovered. The ␤ 5L subunit has an additional 42 amino acids on its amino terminus and appears to be expressed only in certain areas of the retina (17). Neither the functional properties nor reasons for the restricted distribution of this novel ␤ subunit are well understood. Interestingly, some hypertensive patients express a variant of the ␤ 3 subunit, termed ␤ 3S , that is missing one of the 7 WD repeat regions of the native protein (55). Although the functional properties of the ␤ 3S dimer have not been fully examined, initial evidence suggests that it interacts with the ␣ subunit and receptors differently, leading to a more rapid GDP/GTP exchange (55,56). Taken together, these observations indicate that the diversity in the various ␤ subunits may be more important in regulating the interaction of the ␤␥ dimer with ␣ subunits and effectors than previously anticipated.
Differences in the signaling roles of the various ␤ subunits are beginning to emerge. Studies using recombinant dimers with defined ␤ and ␥ subunits have shown that the ␤ 1 protein is effective in most assays (13,14,16). However, potential differences in the function of the two ␤ 5 subunits have been demonstrated using transfected COS cells. Both the ␤ 5 and ␤ 5L subunits can stimulate PLC-␤ 2 activity when the cDNAs for these proteins are transiently transfected with those for the ␥ 2 subunit and PLC-␤ 2 (16,17). In analogous experiments, the ␤ 5 ␥ 2 subunit was found to inhibit the activity of type I adenylyl cyclase in parallel with ␤ 1 ␥ 2 but was unable to stimulate type II adenylyl cyclase like the ␤ 1 ␥ 2 subunit. Rather, the ␤ 5 ␥ 2 subunit inhibited type II cyclase (14). Interestingly, although the ␤ 5 ␥ 2 dimer can interact with both adenylyl cyclase and PLC-␤ 2 in transfected COS cells, it does not activate the mitogen-activated protein kinase or c-Jun N-terminal kinase pathways in these cells (13).
Our data using pure ␤ 5 ␥ 2HF subunits reconstituted into defined assays provide new insight into these observations. Clearly, the observation that the ␤ 5 ␥ 2 dimer is most likely to be released from G q -linked receptors helps define the potential roles of the ␤ 5 signal. Because one major role for the ␣ q subunit is activation of PLC-␤, it is very interesting that dimers containing the ␤ 5 subunit are able to activate this effector with a potency and efficacy similar to that of dimers containing the ␤ 1 subunit. This observation suggests that cells expressing a high level of the ␤ 5 subunit and an isozyme of PLC-␤ sensitive to ␤␥ may use signals from both subunits to initiate phosphoinositide signaling. The observation that the ␤ 5 ␥ 2HF dimer does not readily activate type II adenylyl cyclase is consistent with this hypothesis as cyclic AMP opposes the actions of inositol 1,4,5trisphosphate and diacylglycerol in many tissues (57,58). Neural tissue, which expresses many ␣ q -linked receptors (59) and a high level of the ␤ 5 subunit, poses an interesting problem in this regard. The brain expresses high levels of PLC-␤ 1 , which is not activated by the ␤␥ dimer (44, 60), and type II adenylyl cyclase (6). Thus, the ␤ 5 signal released by G q -linked receptors in nervous tissue is not likely to activate these effectors and must have other important targets. The known ability of the ␤␥ dimer to regulate K ϩ and Ca 2ϩ channels (5,50,59) suggests ion channels as potential effectors for dimers containing the ␤ 5 subunit. As multiple G protein-mediated signals are often integrated by a single neuron (50,59), differential expression of effector isoforms and certain ␤ subunits may provide distinct cellular responses.
Current data regarding G protein structure and function point to possible mechanisms for the selective actions of the ␤ 5 subunit observed in our experiments. Clearly, the ET B receptor is able to couple to both the G q and G i ␣ subunits in the presence of the ␤ 1 ␥ 2HF dimer. The preference of the ␤ 5 ␥ 2HF dimer for the G q ␣ subunit in coupling to the ET B receptor seems likely to be because of a low affinity of the ␣ i1 -␤ 5 interaction. This possibility is supported by two types of evidence: examination of the interaction of different ␣ subunits using immobilized, recombinant ␤␥ dimers and modeling the ␣ q -␤ 5 interaction based on the existing crystal structures of the ␣␤␥ heterotrimer. In a previous study, we examined the ability of ␣ i1 , ␣ i2 , ␣ o , ␣ s , and ␣ q to associate with 3 different ␤ subunits, ␤ 1 , ␤ 2 , and ␤ 5 , immobilized on an anti-FLAG affinity column via the affinity-tagged ␥ 2HF subunit. Although the ␤ 1 ␥ 2HF dimer bound all 5 ␣ subunits tested, the ␤ 5 ␥ 2HF dimer bound only ␣ q (15). Although these experiments did not precisely measure the affinity of the various ␣ subunits for the different ␤ subunits, they did indicate a reduced affinity of the ␤ 5 subunit for ␣ subunits in the ␣ i and ␣ s families. It will be important to determine the affinity of the ␤ 5 subunit for the various ␣ subunits directly, because affinity differences may lead to selectivity in receptor coupling and thus to specificity in the signals generated by different receptors.
Continued refinement of the crystal structure of the ␣ i1 :␤ 1 :␥ 2 heterotrimer has identified 26 residues on the ␣ subunit and 29 residues on the ␤ subunit participating in the ␣-␤␥ interaction (61). Comparing the ␣ q -␤ 5 interaction in this context shows that there are two pairs of changes that may explain the preference of the ␤ 5 subunit for the ␣ q subunit. The first substitution occurs at the location of the Arg 15 in ␣ i where the Arg 15 -Asn 132 interaction in the ␣ i1 -␤ 1 pair becomes Ile 15 -Met 142 in the ␣ q -␤ 5 pair. The second change is at the Gly 27 -Leu 55 interaction in the ␣ i1 -␤ 1 pair, which becomes Lys 27 -Gly 63 in the ␣ q -␤ 5 pair (61). It will be interesting to determine how the differences in the charge and van der Waals interactions produced by these amino acid replacements affect the measured affinity of the ␣-␤ interaction. These changes between the ␣ i and ␣ q subunits are found in all members of the ␣ q family. Thus, ␣ 11 , ␣ 14 , ␣ 15 , and ␣ 16 may also interact selectively with the ␤ 5 subunit.
Data from studies of point mutations made in the ␤ subunit may help explain our observations that the ␤ 5 ␥ 2HF dimer was very effective at stimulating PLC-␤ but only weakly stimulated type II adenylyl cyclase. In two recent studies, the residues in the ␤ 1 subunit responsible for binding the ␣ subunit were mutated to alanine (or in certain instances, serine or arginine), the mutated ␤ subunits expressed with the ␥ 2 subunit and tested for their activity against effectors (62,63). One notable result from these studies is that mutation of a single amino acid in the ␤ 1 subunit can dramatically reduce the ability of the ␤ 1 ␥ 2 dimer to activate type II adenylyl cyclase. Examples include both the Leu 55 3 Ala 55 and the Trp 332 3 Ala 332 mutations (62). Interestingly, alignment of the ␤ 1 and ␤ 5 sequences indicate a Leu 55 (␤ 1 ) 3 Gly 63 (␤ 5 ) that might be functionally analogous to the Leu 55 3 Ala 55 mutation that blunts the action of the ␤␥ dimer on type II adenylyl cyclase. The data obtained with the various mutated ␤ subunits and recombinant PLC-␤ were more complicated in that different responses were ob-tained with PLC-␤ 2 and PLC-␤ 3 (63). In addition, in one of the studies, none of the mutations had large effects on the ability of the ␤␥ dimers to activate PLC-␤ 2 (62). This latter result is consistent with our observation that the ␤ 5 ␥ 2HF and the ␤ 1 ␥ 2HF dimers were both effective on PLC-␤. However, it is important to note that our experiments were performed with recombinant turkey PLC-␤. In terms of sequence homology, turkey PLC-␤ is most similar to human PLC-␤ 2 (71% identity). However, with respect to its broad tissue distribution and sensitivity to stimulation by the ␤␥ dimer, the turkey isoform is more similar to human PLC-␤ 3 (45). Further experiments to examine the effect of the ␤ 5 ␥ 2 dimer on other isozymes of PLC-␤ are needed.
In summary, the data in this report provide a partial understanding of the functions of the ␤ 5 subunit. It will be important to determine which other ␥ subunits dimerize with this particular subunit and to examine its functionality with other effectors. Results from such studies will indicate how the diversity in the families of proteins comprising the G protein heterotrimer contribute to the specificity of cellular signaling.