Role of Isoprenoid Lipids on the Heterotrimeric G Protein g Subunit in Determining Effector Activation*

Post-translational prenylation of heterotrimeric G protein g subunits is essential for high affinity a - bg and a - bg -receptor interactions, suggesting that the prenyl group is an important domain in the bg dimer. To determine the role of the prenyl modification in the interaction of bg dimers with effectors, the C AAX (where A indicates alipathic amino acid) motifs in the g 1 , g 2 , and g 11 subunits were altered to direct modification with different prenyl groups. Six recombinant bg dimers were overexpressed in baculovirus-infected Sf9 insect cells, purified, and examined for their ability to stimulate three phospholipase C- b isozymes and type II adenylyl cyclase. The native b 1 g 2 dimer ( g subunit modi- fied with geranylgeranyl) is more potent and effective in activating phospholipase C- b than either the b 1 g 1 (far- nesyl) or the b 1 g 11 (farnesyl) dimers. However, farnesyl modification of the g subunit in the b 1 g 2 dimer ( b 1 g 2-L71S ) caused a decrement in its ability to activate phos- pholipase C- b . In contrast, both the b 1 g 1-S74L (gera-nylgeranyl) and the b 1 g 11-S73L (geranylgeranyl) dimers were more active than the native forms. The b the amount of dimer inserted into the vesicles. amount of phospholipid eluted bg dimers bound vesicles quantitated described under “Experimental Procedures.” of three )

Heterotrimeric G proteins 1 are transducers of numerous ex-tracellular signals from seven transmembrane receptors to intracellular effectors (1)(2)(3)(4). G proteins are composed of ␣, ␤, and ␥ subunits and associate with the inner side of the plasma membrane. Receptor activation catalyzes the exchange of GDP for GTP on the ␣ subunit, resulting in dissociation of the GTP-liganded ␣ subunit from the ␤␥ dimer (1). Both the GTPbound form of the ␣ subunit and the released ␤␥ dimer regulate a variety of effectors, including PLC-␤ (5, 6) and adenylyl cyclases (7,8). To date, 7 ␤ subunits and 11 ␥ subunits have been identified in mammalian systems (9 -15). Selective assembly of ␤␥ heterodimers from these proteins may produce a large number of unique complexes that differ in their interactions with ␣ subunits, receptors, and effectors. While the first four ␤ subunits identified, ␤ 1 -␤ 4 , are 85-90% identical in primary amino acid sequence (9), the ␥ subunits are much more divergent. For example, the ␥ 1 and ␥ 5 subunits are only 25% identical. Thus, the ␥ subunits may impart some specificity to the ␤␥ signal.
The G protein ␥ subunits are subject to post-translational modification by the addition of isoprenoid lipids to an invariant cysteine residue in the CAAX motif at their C terminus (15)(16)(17). The last amino acid (X) of the CAAX motif is an important determinant for modification by one of two distinct isoprenoids, the 15-carbon farnesyl group or the 20-carbon geranylgeranyl group. Among the 11 known ␥ subunits, the ␥ 1 , ␥ 8 , and ␥ 11 subunits are thought to be modified with the addition of a 15-carbon farnesyl group, whereas the other ␥ subunits contain a 20-carbon geranylgeranyl group (10). The ␥ 11 subunit was recently identified as a widely expressed ␥ subunit that is 76% identical to the ␥ 1 subunit (10,18). However, the function of ␤␥ dimers containing the ␥ 11 subunit has not been studied. Moreover, the significance of farnesyl versus geranylgeranyl modification of the ␥ subunit in ␤␥ subunit-mediated activation of effectors has not been established.
Lipid modification is important for anchoring the ␤␥ subunit to the membrane (17,19,20), but the prenyl group may also play a major role in determining functional interaction with other proteins (15,17,21). For instance, prenylation of the ␥ subunit is necessary for formation of an active transducin ␣-␤␥ complex (22,23), for translocation of the ␤-adrenergic receptor kinase to the plasma membrane (24), for high affinity interactions with either ␣ subunits or adenylyl cyclases (25), and for stimulation of PLC-␤ by the ␤ 1 ␥ 1 dimer (26). Since the ␥ subunits may have different primary amino acid sequences and different prenyl groups at their C terminus, investigators have attempted to determine whether the type of prenyl group is an important determinant of the activity of the ␤␥ dimer. Some studies suggest that ␤␥ dimers containing the farnesylated ␥ 1 subunit couple rhodopsin to the G t ␣ subunit more effectively (27,28), while others show a small increase in the activity of ␤␥ dimers containing a geranylgeranylated ␥ 1 subunit in receptorcoupling assays (29). We have determined that ␤␥ dimers containing a geranylgeranyl group on the ␥ subunit couple the G i ␣ subunit to the A1 adenosine receptor more effectively than those containing a farnesyl group (30). Overall, these results suggest that the type of prenyl group on the ␥ subunit does play an important role in determining the activity of ␤␥ dimers.
Most investigators have considered that the prenyl group is involved in tethering the ␤␥ subunit to the membrane (17,19,20). However, the recent x-ray crystal structure of a phosducin-␤␥ complex, which was crystallized with an intact ␥ subunit containing a farnesyl group, shows that the isoprenoid group folds into a hydrophobic pocket formed by blades 6 and 7 of the ␤ propeller (31). Interestingly, the conformation of the ␤ subunit in this complex is different from the conformation of the free ␤␥ dimer (32). Thus, it is possible that there is an active conformation of the ␤␥ dimer, which occurs upon binding to effectors, and that the prenyl group participates in establishing this conformation. These observations provide a compelling reason to examine the ability of ␤␥ dimers in which the ␥ subunit contains different isoprenoid lipids in regulating effectors.
We have reported previously that the prenyl group on the ␥ 1 and ␥ 2 subunits can be altered by mutation of the last amino acid (X) in the CAAX motif at their C terminus (30,33). Expression of the mutant ␥ subunits in Sf9 cells with a ␤ subunit allows formation of functional ␤␥ dimers in which the lipid on the ␥ subunit has been changed from farnesyl to geranylgeranyl or vice versa (30,33). Using this approach, we have purified six recombinant ␤␥ dimers containing ␥ subunits with native or altered prenyl groups and verified that the altered CAAX sequences result in the expected post-translational modifications using mass spectrometry. We examined the effects of the ␤ 1 ␥ 2 , ␤ 1 ␥ 1 , and ␤ 1 ␥ 11 dimers and their prenyl mutants on the activity of PLC-␤ and type II adenylyl cyclase. The results indicate that both the primary amino acid sequence of the ␥ subunits and the type of prenyl group on the ␥ subunit are important determinants of the activation of PLC-␤ and type II adenylyl cyclase.

EXPERIMENTAL PROCEDURES
Construction of Recombinant Baculoviruses-The recombinant baculoviruses encoding the ␣ S , ␤ 1 , ␥ 1 , ␥ 2 subunits (34,35), and the ␥ 1 and ␥ 2 subunits with altered prenylation sequences in the CAAX motifs, ␥ 1-S74L and ␥ 2-L71S , were prepared as described (33). A full-length cDNA encoding the human ␥ 11 protein was identified in the expressed sequence tag data base and obtained from Research Genetics, Inc. (␥ 11 , GenBank TM accession number H00850). To minimize the length of the construct 5Ј from the ATG start codon, the polymerase chain reaction was used to add a SmaI restriction site to the 5Ј end of ␥ 11 . For the ␥ 11 cDNA, the primers used were: sense primer, 5Ј-TCCCGGGCGAAAATGCCTGCC-3Ј; antisense primer, 5Ј-CCACTTAGGATGCAGTTTCTCCC-3Ј. The polymerase chain reaction products were subcloned into the pCNTR shuttle vector, and the ␥ 11 coding sequence excised from pCNTR with SmaI and XbaI was ligated into these sites in the baculovirus transfer vector, pVL1393. The cDNA encoding a ␥ 11 with the altered prenylation sequence in the CAAX motif, ␥ 11-S73L , was constructed with the same strategy. Modification of the C-terminal CAAX motif of ␥ 11 was performed using polymerase chain reaction to introduce a CVIS 3 CVIL into the CAAX sequence. To produce the ␥ 11-S73L cDNA, the primers used were: sense primer, 5Ј-TCCCGGGCGAAAATGCCTGCC-3Ј; antisense primer, 5Ј-TTTATAAAATAACACAGCTGCC-3Ј. The polymerase chain reaction products were subcloned into the pCNTR shuttle vector, digested with SmaI and XbaI, and ligated into these sites in the pVL1393 transfer vector. The completed constructs of both ␥ 11 and ␥ 11-S73L in the pVL1393 transfer vector were sequenced to confirm the fidelity of the ␥ sequences. Recombinant baculoviruses were produced by co-transfecting each recombinant plasmid DNA with linear wild type BaculoGold® viral DNA (PharMingen) into Sf9 cells as described (36). The recombinant baculoviruses were purified by one round of plaque purification (37).
Expression and Purification of Recombinant G Protein ␤␥ Subunits-G protein ␣ and ␤␥ subunits were overexpressed by infecting Sf9 insect cells with recombinant baculoviruses as described (34,35). Sf9 cells were co-infected at a multiplicity of infection of 3 with the appropriate ␤ and ␥ recombinant baculoviruses and harvested 48 h after infection. The ␤␥ subunits were extracted from frozen cell pellets with 0.1% Genapol C-100 and purified on a DEAE column followed by affinity chromatography on a G i1 -␣-agarose column as described (35). The G s ␣ subunit used in the adenylyl cyclase assays was prepared from a 0.1% (w/v) CHAPS extract of crude cell lysates as described (36).
Analysis of the Post-translational Processing of ␥ Subunits by Mass Spectrometry-Full processing of the ␥ subunit requires the addition of either a farnesyl or a geranylgeranyl group to the C-terminal cysteine in the CAAX motif, the removal of the three C-terminal amino acids (AAX), and the addition of a carboxymethyl group to the C terminus (16,33). To determine the extent of the post-translational lipid modification of the ␥ subunits, the six purified ␤␥ dimers used in this study were analyzed by electrospray ionization mass spectrometry as described previously (33). The result of this analysis showed that 96% of the native ␥ 1 subunit contained the expected farnesyl group and had a completely processed C terminus. Similarly, 87% of the prenyl mutant, ␥ 1-S74L , contained the geranylgeranyl group and was fully processed at its C terminus. Ninety-one percent of the ␥ 2 subunit and 70% of the ␥ 11 subunit were found to have the expected geranylgeranyl and farnesyl groups, respectively, and a completely processed C terminus. The prenyl mutants of these two ␥ subunits, ␥ 2-L71S and ␥ 11-S73L , were fully processed with expected isoprenoid lipids to the extent of 95 and 87%, respectively. Therefore, all ␥ subunits used in this study contain the expected isoprenoid lipid at the C terminus and are capable of highaffinity interactions with ␣ subunits, receptors and effectors.
Preparation of Large Unilamellar Vesicles-Phospholipid vesicles were prepared as described (40). Briefly, phospholipids were mixed at a molar ratio of 4:1 of phosphatidylethanolamine to PIP 2 with [inositol-2-3 H]PIP 2 in a buffer containing 50 mM HEPES, pH 8.0, 3 mM EGTA, 80 mM KCl, 1 mM dithiothreitol and the mixture dried under argon. The dried lipid was suspended with 3.0 ml of the above buffer to give final concentrations of 1 mM phosphatidylethanolamine, 250 M PIP 2 , and 800 cpm/l [ 3 H]PIP 2 and hydrated in the dark at room temperature. Extruded large unilamellar vesicles were formed from multilamellar vesicles by vortexing the lipid solution followed by 10 extrusion cycles through a stack of two polycarbonate filters using a mini-extruder (Avanti Polar Lipids, Inc.) (41). After extrusion, the final lipid concentration (phosphatidylethanolamine and PIP 2 ) was about 1100 M. Extruded large unilamellar vesicles were stored under argon at 4°C until use.
Reconstitution of ␤␥ Dimers to Large Unilamellar Vesicles by Gel Filtration-CHAPS-solubilized ␤␥ subunits (ranging from 10 ng to 10 g) were mixed with vesicles in a volume of 200 l. The final CHAPS concentration in the mixture was held to 0.01%. After reconstituting the mixture on ice for 30 min, the vesicles were applied to a 2-ml gel filtration column prepared with Ultrogel AcA34 as described previously (40,42). The column was pre-equilibrated at 4°C with buffer containing 20 mM HEPES, pH 8.0, 2 mM MgCl 2 , 100 mM NaCl, and 1 mM EDTA (450 l/min). The vesicles reconstituted with ␤␥ dimers were eluted at the void volume of the column in a total volume of about 900 l. This protocol has been demonstrated to remove more than 98% of the detergent from the vesicles (42,43) and to resolve the vesicles from the free ␤␥ dimers (40). The concentration of lipid in each fraction was monitored by counting the incorporated [ 3 H]PIP 2 , and the amount of ␤␥ protein inserted into the vesicles was quantified by silver staining as described below. The second fraction from the AcA34 column contained vesicles with the least amount of free ␤␥ dimer and were used in all experiments (40).
Measurement of Phospholipase C-␤ Activity-Reaction buffer containing 50 mM HEPES, pH 8.0, 17 mM NaCl, 67 mM KCl, 0.83 mM MgCl 2 , 0.17 mM EDTA, 3 mM EGTA, 1 mM dithiothreitol, and 1 mg/ml BSA was added to 70 l of gel-filtered vesicles (fraction 2) in a total volume of 80 l. Controls were performed using column-reconstituted vesicles without ␤␥ dimers. The reactions were initiated by adding 10 ng of PLC-␤ and 3 M free Ca 2ϩ and placing the mixture in a 30°C water bath. After a 15-min incubation, the reactions were terminated by transferring each assay tube to a 4°C ice bath and adding 200 l of ice-cold 10% trichloroacetic acid followed by the addition of 100 l of 10 mg/ml BSA (40). The assay mixtures were centrifuged to remove precipitated protein and intact PIP 2 and the [ 3 H]inositol 1,4,5-trisphosphate released into the supernatant quantitated by liquid scintillation spectrometry (44,45).
Measurement of Adenylyl Cyclase Activity-Adenylyl cyclase activity was measured as described previously (36,46). Briefly, Sf9 insect cells were infected with recombinant baculovirus encoding rat type II adenylyl cyclase (47) and harvested 48 h after infection. Sf9 membranes containing type II adenylyl cyclase (5 g of protein/assay tube) were reconstituted with GTP␥S-activated G s ␣ subunit (48) and varying concentrations of ␤␥ dimers on ice for 30 min. The reaction was initiated by addition of the reconstituted membranes to the reaction buffer containing 25 mM HEPES, pH 8.0, 10 mM phosphocreatine, 10 units/ml of creatine phosphokinase, 0.4 mM 3-isobutyl-1-methylxanthine, 10 mM MgCl 2 , 0.5 mM ATP, and 0.1 mg/ml BSA and incubated for 7 min at 30°C. The assay was stopped by addition of 0.1 N HCl and cyclic AMP measured using an automated radioimmunoassay (49).
Electrophoresis-For quantitation of the ␤␥ concentration in vesicles, the ␤␥ dimers reconstituted into phospholipid vesicles were electrophoresed on a 12% acrylamide, sodium dodecyl sulfate-polyacrylamide gel and stained with silver according to the method of Bloom et al. (50). The ␤␥ concentrations in the gel were estimated using the amount of stained ␤ subunit protein as compared with ovalbumin standards run in the same gel. The stained proteins were quantitated using a Bio-Image scanning densitometer and the Whole Band® software (Bio-Image, Ann Arbor, MI) as described (36,40). The accuracy of this procedure was verified by comparison with the BCA (Pierce) protein assay (40). To better display the mobility of the ␥ subunits, Tricine/ SDS-polyacrylamide gels were run according to the procedure of Schagger and von Jagow (51). The separating gel contained 16.5% total acrylamide, 0.4% bisacrylamide, and 10% (v/v) glycerol. The stacking gel contained 4% total acrylamide and 0.1% bisacrylamide. Gels were run at constant voltage (ϳ100 volts) at 10°C for 4 -5 h. Resolved proteins were stained with silver according to the method of Morrissey (52) with the modification that the dithiothreitol incubation was reduced to 15 min.
Calculation and Expression of Results-Experiments presented under "Results" are representative of three or more similar experiments. Data expressed as concentration-response curves were fit to sigmoid curves using the fitting routines in the GraphPad Prism™ software. Statistical differences between the curves were determined using all the individual data points from multiple experiments to calculate the F statistic as described (53).

RESULTS
To determine the role of the prenyl modification of the ␥ subunit in the interaction of ␤␥ dimers with effectors, the CAAX motifs in the ␥ 1 , ␥ 2 , and ␥ 11 subunits were altered to direct the addition of different prenyl isoprenoids. Each ␥ subunit was co-expressed with the ␤ 1 subunit in Sf9 insect cells, and the six ␤␥ dimers, and ␤ 1 ␥ 11-S73L , were purified using ␣ i1 -agarose affinity chromatography and tested for their ability to activate two effectors, PLC-␤ and type II adenylyl cyclase. All six ␥ subunits were analyzed by mass spectrometry to ensure that their CAAX motifs were appropriately processed. The results of the analysis showed that each ␥ subunit had the expected isoprenoid lipids at its C terminus with the extent of processing ranging from 70 to 95% (see "Experimental Procedures"). Therefore, all ␤␥ dimers used in this study are capable of high-affinity interactions with ␣ subunits, receptors, and effectors. The purity of these dimers is shown in Fig. 1A. The purity of the three recombinant PLC-␤ isoforms used in this study is shown in Fig.  1B.
Differential Effect of Three Different ␤␥ Dimers on PLC-␤ and Adenylyl Cyclase Activity-We compared the ability of ␤␥ dimers with ␥ subunits containing either a farnesyl or a geranylgeranyl moiety to stimulate PLC-␤. An assay system was used in which purified ␤␥ dimers were reconstituted into extruded phospholipid vesicles and the vesicles separated from mono-dispersed ␤␥ dimers and the detergent used to solubilize G proteins on an Ultrogel AcA34 column (40). The data in Fig.  2A illustrate the ability of three ␤␥ dimers to activate recombinant turkey PLC-␤. Avian PLC-␤ was used initially, because it has a lower basal activity and higher ␤␥ subunit-stimulated activity than either the mammalian PLC-␤1 or PLC-␤2 isozymes (39) (see also Fig. 3). The ␤ 1 ␥ 2 dimer activated turkey PLC-␤ about 10-fold with an estimated EC 50 value of 0.7 nM (refer to Table II). The ␤ 1 ␥ 11 and ␤ 1 ␥ 1 subunits activated turkey PLC-␤ with estimated EC 50 values of 4.1 and 1.9 nM, respectively. The V max values were 50 -60% of those observed with the ␤ 1 ␥ 2 dimer (refer to Table II). Therefore, the ␤ 1 ␥ 2 dimer is more potent and effective in activating PLC-␤ than either ␤ 1 ␥ 1 or ␤ 1 ␥ 11 , and the potency of ␤ 1 ␥ 11 is less than that of ␤ 1 ␥ 1 . The data in Fig. 2B compare the ability of the three ␤␥ dimers to activate type II adenylyl cyclase. The ␤ 1 ␥ 2 dimer activated type II adenylyl cyclase about 12-fold with an estimated EC 50 value of 14 nM. However, neither ␤ 1 ␥ 1 (54) nor ␤ 1 ␥ 11 effectively activated type II adenylyl cyclase. Since the ␥ subunits were combined with the same ␤ 1 subunit, these results indicate that differences in the activation of either PLC-␤ or type II adenylyl cyclase are due to differences in the ␥ subunit. The major differences between the known ␥ subunits are their divergent amino acid sequences and their lipid modifications. Interestingly, the data in Fig. 2A show that ␤␥ dimers containing the ␥ 2 subunit modified with a geranylgeranyl group are more potent and effective in activating both effectors than those containing either the ␥ 1 or the ␥ 11 subunit modified with a farnesyl group. Therefore, to address the possibility that the identity of the prenyl group on the ␥ subunit is an important determinant of the dimer's activity, the CAAX motif in the ␥ 2 subunit was altered to direct the addition of the farnesyl group and the ability of the ␤ 1 ␥ 2-L71S dimer to activate three PLC-␤ isoforms was examined. Effect of ␤ 1 ␥ 2 and ␤ 1 ␥ 2-L71S on PLC-␤ Activity-The data in Fig. 3 show that the ␤ 1 ␥ 2 dimer stimulated PLC-␤ with a subnanomolar potency and that exchanging the prenyl group on the ␥ 2 subunit from geranylgeranyl to farnesyl (␤ 1 ␥ 2-L71S ) caused a diminution in the ability of the ␤ 1 ␥ 2 dimer to activate all three isoforms of PLC-␤. The farnesyl-modified ␤ 1 ␥ 2-L71S dimer was less potent and effective in activating turkey PLC-␤ than the native ␤ 1 ␥ 2 dimer (Fig. 3A). Fitting the data to sigmoid curves estimated that the maximal stimulation with the ␤ 1 ␥ 2-L71S dimer decreased by 35% and that the EC 50 value increased from 0.7 to 1.2 nM (refer to Table II). Similarly, the ␤ 1 ␥ 2-L71S dimer was less potent and effective than the native ␤ 1 ␥ 2 dimer in the activation of human PLC-␤2 (Fig. 3B). The data in Fig. 3C indicate that the ␤␥ subunits produced little effect on the activity of PLC-␤1 as demonstrated previously (5,44,45,55,56). Thus, ␤␥ subunits stimulated activity in the order of tPLC-␤ Ն PLC-␤2 Ͼ Ͼ Ͼ PLC-␤1. Overall, these results indicate that the prenyl group on the ␥ subunit is an important determinant of the interaction between PLC-␤ and ␤␥ dimers.
Partition of Different Recombinant ␤␥ Dimers into Phospho-lipid Vesicles-Since most studies of G protein activation of PLC-␤ have used the reconstitution of pre-aggregated lipid with ␣ or ␤␥ subunits to activate PLC-␤, a potential problem might arise if ␤␥ dimers containing different isoprenoid lipids did not partition equally into pre-aggregated lipid layers. This is especially important considering the different chain lengths of the lipids associated with the native ␥ 1 , ␥ 2 , or ␥ 11 subunits.
To determine the partitioning into vesicles of the six different ␤␥ dimers used in this study, fractions 2 and 3 from the AcA34 column were collected and the amount of protein inserted into the vesicles was measured as described (40).  2. Differential abilities of three ␤␥ dimers to regulate PLC-␤ and adenylyl cyclase. A, the ability of the ␤ 1 ␥ 1 and ␤ 1 ␥ 11 dimers to activate phospholipase C-␤ as compared with that of the ␤ 1 ␥ 2 dimer. Extruded phospholipid vesicles containing phosphatidylinositol 4,5-bisphosphate were reconstituted with 0.01, 0.03, 0.1, 0.3, 1.0, 3.0, or 10.0 g of each ␤␥ dimer on ice for 30 min. The mixture was applied to a 2-ml Ultrogel AcA34 column, fraction 2 was collected, and the ability of ␤ 1 ␥ 1 (open circles), ␤ 1 ␥ 2 (closed squares), and ␤ 1 ␥ 11 (closed triangles) in fraction 2 to stimulate turkey PLC-␤ activity was measured as described under "Experimental Procedures." The difference between the effect of ␤ 1 ␥ 2 and either ␤ 1 ␥ 1 or ␤ 1 ␥ 11 was statistically significant (p Ͻ 0.0001). B, comparison of the ability of three ␤␥ dimers to stimulate type II adenylyl cyclase. Sf9 cells were infected with a recombinant baculovirus encoding type II adenylyl cyclase, membranes prepared, and the cyclase reaction performed with the indicated concentrations of three recombinant ␤␥ dimers as described under "Experimental Procedures." The results are representative of six similar experiments, each performed in duplicate. Sigmoid curves were fit to the data and statistical differences between the curves determined as described under "Experimental Procedures." The difference between the effect of ␤ 1 ␥ 2 and either ␤ 1 ␥ 1 or ␤ 1 ␥ 11 was statistically significant (p Ͻ 0.0001).
Effect of ␤ 1 ␥ 1 , ␤ 1 ␥ 1-S74L , ␤ 1 ␥ 11 , and ␤ 1 ␥ 11-S73L on PLC-␤ Activity-It was important to determine whether switching the farnesyl group to geranylgeranyl improves the ability of either the ␤ 1 ␥ 1 or the ␤ 1 ␥ 11 dimer to activate PLC-␤. Thus, the CAAX motifs in the ␥ 1 and ␥ 11 subunits were altered to direct the addition of a geranylgeranyl group, and the ability of both the ␤ 1 ␥ 1-S74L and ␤ 1 ␥ 11-S73L dimers to activate either turkey PLC-␤ or human PLC-␤2 was examined. The data in Fig. 4A show that the geranylgeranyl-modified ␤ 1 ␥ 1-S74L dimer exhibited an increased potency and maximal activity compared with the native ␤ 1 ␥ 1 dimer. Similarly, the geranylgeranyl-modified ␤ 1 ␥ 11-S73L dimer was more potent and effective in activating turkey PLC-␤ than the native ␤ 1 ␥ 11 dimer. The ␤ 1 ␥ 1-S74L and ␤ 1 ␥ 11-S73L dimers also exhibited increased capacity to stimulate human PLC-␤2 activity (Fig. 5). These four ␤␥ dimers (␤ 1 ␥ 1 , ␤ 1 ␥ 1-S74L , ␤ 1 ␥ 11 , and ␤ 1 ␥ 11-S73L ) showed little effect on the activity of PLC-␤1 (data not shown, but see Fig. 3C). The data in Table II summarize the potency and efficacy of the six ␤␥ dimers used in this study on the activation of these two PLC-␤ isoforms. The ␤ 1 ␥ 1-S74L dimer exhibited an about 1.6-fold increase in maximal activity as compared with the native ␤ 1 ␥ 1 dimer and the EC 50 value decreased from 1.9 to 1.1 nM. The estimated maximal activity of the ␤ 1 ␥ 11-S73L dimer increased about 1.3-fold and the EC 50 value decreased from 4.1 to 1.6 nM. Similar changes were observed in the ability of the dimers to activate human PLC-␤2. Taken together, the data indicate that ␤␥ dimers with a ␥ subunit containing a geranylgeranyl group interact with higher affinity with PLC-␤ than ␤␥ dimers with a ␥ subunit containing a farnesyl group.
Effect of the Prenyl Group on the Activity of Type II Adenylyl Cyclase-To determine the role of the isoprenoid group in the interaction with an effector which is membrane-associated, the capacity of the six ␤␥ dimers to activate type II adenylyl cyclase was also examined. The data in Fig. 6A indicate that switching the prenyl group on the ␥ 2 subunit from geranylgeranyl to farnesyl caused a significant decrement in the ability of the ␤ 1 ␥ 2 dimer to activate type II adenylyl cyclase. Our previous experiments showed that about the same amount of the ␤ 1 ␥ 2 and ␤ 1 ␥ 2-L71S (about 25% of the ␤␥ subunits added to a reconstitution mixture) are incorporated into Sf9 cell membrane, indicating regardless of the prenyl group on the ␥ subunit, the dimers partition equally into the Sf9 cell membrane (30). Thus, the result from Fig. 6A suggests that the type of prenyl group is important for the interaction of the ␤␥ dimer with adenylyl cyclase. As expected, the ␤ 1 ␥ 1 dimer did not activate type II adenylyl cyclase (54). Interestingly, the ␤ 1 ␥ 11 dimer was completely ineffective on type II adenylyl cyclase. Surprisingly, neither the ␤ 1 ␥ 1-S74L nor the ␤ 1 ␥ 11-S73L dimer containing a geranylgeranyl group on the ␥ subunit were able to significantly stimulate type II adenylyl cyclase (Fig. 6, B and C). The data in Table III summarize the potency and efficacy of the six ␤␥ dimers on the activation of type II adenylyl cyclase. The farnesyl-modified ␤ 1 ␥ 2-L71S dimer was about 60 -65% less active, and its EC 50 value was increased from 14 to 76 nM. This change in potency and maximal activity is larger than that observed with PLC-␤ (see Fig. 3). Thus, ␤␥ dimers with a ␥ subunit containing a geranylgeranyl group interact with higher affinity with type II adenylyl cyclase than those with a FIG. 4. Comparison of the activity of native and altered ␤␥ dimers to stimulate recombinant turkey PLC-␤. A, the ability of ␤ 1 ␥ 1 (modified with farnesyl; open circles) and ␤ 1 ␥ 1-S74L (modified with geranylgeranyl; closed circles) dimers to stimulate recombinant turkey PLC-␤ (tPLC-␤) was measured as described under "Experimental Procedures." The difference between the effect of ␤ 1 ␥ 1 and ␤ 1 ␥ 1-S74L was statistically significant (p Ͻ 0.001). B, analogous experiments were performed with ␤ 1 ␥ 11 (modified with farnesyl; closed triangles) and ␤ 1 ␥ 11-S73L (modified with geranylgeranyl; open triangles) dimers. The difference between the effect of ␤ 1 ␥ 11 and ␤ 1 ␥ 11-S73L was statistically significant (p Ͻ 0.001). ␥ subunit containing a farnesyl group. While these results are not surprising, the observation that neither the ␥ 1 or the ␥ 11 subunit modified with a geranylgeranyl group have any activity on type II cyclase was unexpected. These observations suggest that the differences in amino acid sequence between ␥ 1 and ␥ 2 greatly affect the ability of the dimers to activate adenylyl cyclase. DISCUSSION This study demonstrates that ␤␥ dimers with ␥ subunits containing different isoprenoids have distinct abilities to activate PLC-␤ and type II adenylyl cyclase. The two major findings of this study are (a) the isoprenoid lipid added to the FIG. 5. Comparison of the activity of native and altered ␤␥ dimers to stimulate recombinant human PLC-␤2. A, phospholipid vesicles reconstituted with ␤ 1 ␥ 1 (modified with farnesyl; open circles) or ␤ 1 ␥ 1-S74L (modified with geranylgeranyl; closed circles) dimer were incubated with recombinant human PLC-␤2 (hPLC-␤2) and PLC-␤ activity measured as described under "Experimental Procedures." The data were fit to sigmoid curves. The difference between the effect of ␤ 1 ␥ 1 and ␤ 1 ␥ 1-S74L was statistically significant (p Ͻ 0.001). B, analogous experiments were performed with ␤ 1 ␥ 11 (modified with farnesyl; closed triangles) and ␤ 1 ␥ 11-S73L (modified with geranylgeranyl; open triangles) dimers. The difference between the effect of ␤ 1 ␥ 11 and ␤ 1 ␥ 11-S73L was statistically significant (p Ͻ 0.001).

TABLE II
Comparison of EC 50 and V max of six ␤␥ dimers on the activation of two PLC-␤ isoforms Phospholipid vesicles containing phosphatidylinositol 4,5-bisphosphate were reconstituted with different recombinant ␤␥ dimers on ice for 30 min. After the incubation, the mixture was applied to a 2-ml Ultrogel AcA34 column at 4°C and fraction 2 eluting at the void volume used for the measurement of PLC-␤ activity. Fraction 2 was incubated for 15 min at 30°C in the presence of purified recombinant turkey PLC-␤ (tPLC-␤) or human PLC-␤2 (hPLC-␤2) as described under "Experimental Procedures." The data are expressed as mean Ϯ S.E. and the average of three determinations, each performed in duplicate. The EC 50 (nM) and V max (mol/mg of PLC/min) values were estimated by fitting of the averaged data to sigmoid curves.  6. Comparison of the activity of native and altered ␤␥ dimers to stimulate type II adenylyl cyclase. A, 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 ␤ 1 ␥ 2 (modified with geranylgeranyl; closed squares) and ␤ 1 ␥ 2-L71S (modified with farnesyl; open squares) as described under "Experimental Procedures." The difference between the effect of ␤ 1 ␥ 2 and ␤ 1 ␥ 2-L71S was statistically significant (p Ͻ 0.0001). B, the cyclase reaction was performed with the indicated concentrations of ␤ 1 ␥ 1 (modified with farnesyl; open circles) and ␤ 1 ␥ 1-S74L (modified with geranylgeranyl; closed circles) and compared with the effect of ␤ 1 ␥ 2 (dotted lines). The difference between the effect of ␤ 1 ␥ 2 and ␤ 1 ␥ 1 was statistically significant (p Ͻ 0.0001), but the difference between ␤ 1 ␥ 1 and ␤ 1 ␥ 1-S74L was not statistically significant. C, analogous experiments were performed with ␤ 1 ␥ 11 (modified with farnesyl; closed triangles) and ␤ 1 ␥ 11-S73L (modified with geranylgeranyl; open triangles) dimers. The difference between the effect of ␤ 1 ␥ 2 and ␤ 1 ␥ 11 was statistically significant (p Ͻ 0.0001), but the difference between ␤ 1 ␥ 11 and ␤ 1 ␥ 11-S73L was not significant. cysteine residue in the CAAX motif of the ␥ subunit plays a role in determining the activity of the ␤␥ dimer at both effectors and (b) ␤␥ dimers with a ␥ subunit containing a geranylgeranyl group are more potent and effective in activating either PLC-␤ or type II adenylyl cyclase than those with a ␥ subunit containing a farnesyl group. This study also presents the first demonstration of the functional activity of a ␤␥ dimer containing the ␥ 11 subunit. The ␤ 1 ␥ 11 dimer is about six times less potent and about 50% less effective in activating PLC-␤ than the ␤ 1 ␥ 2 dimer and has little or no effect on the type II adenylyl cyclase. Overall, the data presented here indicate that in addition to the primary amino acid sequence of the ␥ subunits, the prenyl group on the ␥ subunit is an important determinant of the activation of PLC-␤ and type II adenylyl cyclase.
The recent x-ray crystal structure of a complex of bovine retinal phosducin with the transducin ␤␥ subunit in which the ␥ subunit contained a farnesyl group provides a conceptual background for interpreting the results obtained with PLC-␤ and adenylyl cyclase (31). Remarkably, the interaction between phosducin and the ␤␥ dimer appears to result in a significant conformational change in the ␤ subunit as compared with the conformation seen in the free ␤␥ dimer (32). This event also appears to cause the insertion of the prenyl group into a hydrophobic pocket in the ␤ subunit formed between blades 6 and 7 (31). Interestingly, this conformational change occurs on the outer strands of blades 6 and 7 of the ␤ propeller and recent mutagenesis studies show that these regions are critical for the activation of PLC-␤ (57). If the hydrophobic pocket within the ␤ subunit that binds the prenyl group is able to discriminate between the different length of the farnesyl and geranylgeranyl lipids, different conformational changes might be induced by the ␥ 1 and ␥ 2 subunits (or their prenyl mutants). This possibility might explain differences in the ability of the various ␤␥ dimers to stimulate the activity of PLC-␤ and adenylyl cyclase. Overall, these observations suggest that insertion of a prenyl group into the hydrophobic pocket formed in the ␤ subunit may be an important part of the mechanism by which the ␤␥ dimers activate effectors when released from the ␣ subunit.
The interaction of protein prenyltransferases with their isoprenoid substrates provides another example of a hydrophobic binding pocket determining isoprenoid specificity. Protein prenyltransferases are responsible for the addition of isoprenoid lipid to proteins and are classified by their lipid substrate: protein farnesyltransferase (FTase), protein geranylgeranyltransferase (GGTase-I), and the Rab geranylgeranyltransferase (GGTase-II) (58). G protein ␥ subunits are targets for either FTase or GGTase-I. These two prenyltransferases are heterodimers composed of the common 48-kDa ␣ subunit and distinct ␤ subunits: 46-kDa ␤ subunit for FTase (␤ F ) and 43-kDa ␤ subunit for GGTase-I (␤ GGI ) (59 -62). Thus, the binding sites for different lipid substrates on these two prenyltransferases are thought to reside in their ␤ subunits (58). The x-ray crystal structure determined by Park et al. (63) illustrates that farnesyl diphosphate occupies a hydrophobic pocket in the center of the ␤ subunit barrel. Recently, it has been demonstrated that the isoprenoid moiety of farnesyl diphosphate binds in an extended conformation within the hydrophobic cavity of the ␤ subunit of the FTase, and the diphosphate moiety binds to a positively charged cleft at the top of this cavity near the subunit interface (64). These findings suggest that the isoprenoid substrate specificity of the hydrophobic binding cavity is determined by the depth of the cavity and that the pocket acts as a ruler discriminating between isoprenoids of different length (64).
The three classes of PLC isozymes, PLC-␤, PLC-␥, and PLC-␦, contain two highly conserved catalytic domains, X and Y, but only the PLC-␤ isoforms are regulated by G proteins (65). These isoforms are distinguished from the other families by an extended C terminus. Deletion of the C-terminal amino acids of PLC-␤1 and PLC-␤2 eliminated G q ␣ subunit-mediated PLC-␤ activation (66 -68). Peptides corresponding to a portion of the C-terminal region of PLC-␤1 inhibited activation by G q ␣ subunit in vitro (67), suggesting that the C-terminal region might be important for the ␣ subunit-induced PLC-␤ activation. Since removal of the C-terminal region of PLC-␤1 and PLC-␤2 did not impair ␤␥ subunit-induced enzyme activation, the remaining domains of the enzyme must provide the ␤␥ subunit interaction sites. A PLC-␤2 fragment corresponding to a region between the X and Y domains (Glu 435 to Val 641 ) inhibited ␤␥ subunit-mediated activation of PLC-␤2 in transfected COS-7 cells (69). Peptides from this sequence expressed as a series of fusion proteins bound to purified ␤␥ subunits in vitro (69). Overexpression of this region also blocked ␤␥ subunitinduced activation of co-transfected PLC-␤2 in COS cells (69). It has been reported recently that two overlapping PLC-␤2 fragments corresponding to Asn 564 -Lys 583 and Glu 574 -Lys 593 of PLC-␤2 inhibited activation of PLC-␤2 by ␤␥ subunits, inhibited ␤␥ subunit-dependent ADP-ribosylation of G i1 ␣ subunit by pertussis toxin, and inhibited ␤␥ subunit-dependent activation of phosphoinositide 3-kinase (70). These observations suggest that the region between Glu 574 and Lys 583 in PLC-␤2 may be the site that interacts with ␤␥ subunits (70).
A number of studies have attempted to determine the sites at which the ␤␥ dimer interacts with adenylyl cyclase. Type I adenylyl cyclase is inhibited by G i ␣ subunits and ␤␥ subunits, whereas type II cyclase is activated by the ␤␥ subunits in the presence of G s ␣ subunits (7,71). In an attempt to identify the sites in adenylyl cyclase that interact with the ␤␥ subunits, a chimeric soluble adenylyl cyclase was engineered by linking the type I adenylyl cyclase C1a domain (the N-terminal part of the first cytoplasmic domain) with the type II adenylyl cyclase C2 domain (the second cytoplasmic domain) (72,73). The chimeric soluble enzyme was unresponsive to the G i ␣ subunit but was inhibited almost completely by the ␤ 1 ␥ 2 dimer, suggesting that the C1a domain is critical for ␤␥ subunit-induced inhibition of type I adenylyl cyclase (72,73). In contrast, a 27-amino acid peptide (956 -982) derived from the C2 domain of type II adenylyl cyclase blocked the interaction of ␤␥ dimers with several effectors, including PLC-␤, type II adenylyl cyclase, and K ϩ channels, but did not block ␤␥ interaction with ␣ subunits (74). This result suggests that ␤␥ dimers can also interact with a region on the C2 domain of type II adenylyl cyclase (74).
Whereas the G protein ␤ 1 -␤ 4 subunits are highly conserved in primary amino acid sequence, the ␥ subunits are much more divergent, suggesting that the functional specificity of different ␤␥ dimer combinations may be due to the differences in ␥ subunits. The results presented here support this possibility. Combination of different ␥ subunits with the same ␤ subunit resulted in ␤␥ dimers exhibiting very different ability to activate PLC-␤ and adenylyl cyclase. The differences are due both to the primary amino acid sequences of the ␥ subunits and to their lipid modifications. For example, all ␤␥ dimers containing a geranylgeranyl group on the ␥ subunit were more potent and effective in activating the two effectors than those containing a farnesyl group. Thus, the type of lipid at the C terminus of the ␥ subunit is important to the ability of ␤␥ dimers to activate effectors. Other groups have also obtained data suggesting that the isoprenoid lipid is important in the interaction of ␤␥ dimers with effectors. Matsuda et al. (29) have shown that the ␤ 1 ␥ 1-S74L dimer containing a mixture of farnesyl and geranylgeranyl groups had a greater ability to inhibit Ca 2ϩ /calmodulin-stimulated adenylyl cyclase and stimulate PLC-␤ than did the native farnesylated ␤ 1 ␥ 1 dimer. However, our data suggest that the amino acid sequence of the ␥ subunit is also an important determinant of the activity of the ␤␥ dimer. For example, neither the ␤ 1 ␥ 1 nor the ␤ 1 ␥ 11 dimer are able to activate type II adenylyl cyclase. In contrast to the results obtained with PLC-␤, switching the prenyl group in the ␤ 1 ␥ 1 and ␤ 1 ␥ 11 dimers to geranylgeranyl did not increase their ability to activate type II adenylyl cyclase. Overall, these results indicate that both the primary amino acid sequence and the prenyl group of the ␥ subunits are important determinants for the activation of effectors. For the activation of PLC-␤, the type of prenyl group appears to be a major determinant of the activity of ␤␥ dimer.
In the case of type II adenylyl cyclase, the prenyl group also appears to be important, but the amino acid sequences of certain ␥ subunits, such as ␥ 1 and ␥ 11 , appear to be incapable of activating the enzyme regardless of prenyl group at their C terminus.
Our previous data supports the possibility that charge differences in the ␥ subunit can have a large effect on the ability of ␤␥ dimers to activate type II adenylyl cyclase. Phosphorylation of the ␥ 12 subunit in the ␤ 1 ␥ 12 dimer significantly inhibits the ability of the dimer to stimulate type II adenylyl cyclase (46), and the phosphorylation site has been determined to be at Ser 1 in the N terminus of the molecule (11). This result suggests that introduction of negative charges at the N-terminal region of the ␥ subunit inhibits the interaction of the ␤␥ dimer with the type II adenylyl cyclase. Interestingly, comparison of the N-terminal 20 amino acids of the ␥ 2 with the ␥ 1 or ␥ 11 subunits shows that ␥ 1 and ␥ 11 have six negatively charged amino acids whereas ␥ 2 has only one. Thus, the negatively charged N terminus of the ␥ 1 and ␥ 11 subunits may explain the inability of dimers containing these ␥ subunits to activate type II adenylyl cyclase. Preliminary experiments with chimeric ␥ subunits in which the N-terminal 20 amino acids of ␥ 2 were replaced with the corresponding amino acids from ␥ 1 support this explanation. When expressed with the ␤ 1 subunit, this dimer has a greatly reduced ability to stimulate type II adenylyl cyclase. 2 Thus, multiple domains of the ␥ subunit may be involved in the interaction with different effectors.