Reconstitution of receptors and GTP-binding regulatory proteins (G proteins) in Sf9 cells. A direct evaluation of selectivity in receptor.G protein coupling.

The selectivity in coupling of various receptors to GTP-binding regulatory proteins (G proteins) was examined directly by a novel assay entailing the use of proteins overexpressed in Spodoptera frugiperda (Sf9) cells. Activation of G proteins was monitored in membranes prepared from Sf9 cells co-expressing selected pairs of receptors and G proteins (i.e. α, β1, and γ2 subunits). Membranes were incubated with [35S]guanosine 5′-(3-O-thio)triphosphate (GTPγS) ± an agonist, and the amount of radiolabel bound to the α subunit was quantitated following immunoprecipitation. When expressed without receptor (but with β1γ2), the G protein subunits αz, α12, and α13 did not bind appreciable levels of [35S]GTPγS, consistent with a minimal level of GDP/[35S]GTPγS exchange. In contrast, the subunits αs and αq bound measurable levels of the nucleotide. Co-expression of the 5-hydroxytryptamine1A (5-HT1A) receptor promoted binding of [35S]GTPγS to αz but not to α12, α13, or αs. Binding to αz was enhanced by inclusion of serotonin in the assay. Agonist activation of both thrombin and neurokinin-1 receptors promoted a modest increase in [35S]GTPγS binding to αz and more robust increases in binding to αq, α12, and α13. Binding of [35S]GTPγS to αs was strongly enhanced only by the activated β1-adrenergic receptor. Our data identify interactions of receptors and G proteins directly, without resort to measurements of effector activity, confirm the coupling of the 5-HT1A receptor to Gz and extend the list of receptors that interact with this unique G protein to the receptors for thrombin and substance P, imply constitutive activity for the 5-HT1A receptor, and demonstrate for the first time that the cloned receptors for thrombin and substance P activate G12 and G13.

The actions of numerous cell-surface receptors on enzymes and ion channels are mediated by heterotrimeric (␣␤␥) GTPbinding regulatory proteins (G proteins) 1 (1). Four families of G proteins have been defined, represented by G s , G i , G q , and G 12 (2). The activity of each G protein is tightly linked to the binding and hydrolysis of GTP (3). Agonists working through 7-transmembrane domain receptors promote the release of GDP from the ␣ subunit and thus an exchange for GTP present in the cytoplasm. Correlates of the exchange are an altered conformation of the ␣ subunit and dissociation of the subunit from ␤␥. The ␣ subunit as a monomer and ␤␥ as a heterodimer, or both operating conjointly, are responsible for the regulation of effectors. G proteins can also be activated by nonhydrolyzable analogues of GTP, for example GTP␥S, a process similarly promoted by agonists.
Considerable effort has been expended on identifying the G proteins activated by various receptors. Notwithstanding work with purified proteins, interactions have been defined mostly through inferences drawn from effectors regulated. The stimulation of adenylyl cyclase is almost always achieved through activation of G s , for example, while the inhibition is linked to members of the G i family (4). Stimulation of the phosphoinositide-specific phospholipase C-␤ can be accomplished through G i (pertussis toxin (PTX)-sensitive) or through members of the G q family (PTX-insensitive) (5). No effector has yet been identified for the G 12 family, although Na ϩ /H ϩ exchange is a tightly linked correlate (6). The use of effector regulation to deduce receptor⅐G protein linkages, however, is less than perfect. There is often no resolution among individual members of a particular G protein family, and the nature of effector regulation itself has become increasingly complicated with the appreciation that subunits released from many G proteins can modulate the activation achieved by one (4). Effectors are also subject to numerous forms of feedback or cross-regulation that further limit the extent to which they faithfully mirror receptor⅐G protein communication (4,6,7).
A number of sophisticated techniques have been used to gain either a greater degree of resolution among G proteins regulating a particular effector or a measurement of receptor⅐G protein communication directly. Antisense constructs (8,9), PTX-resistant analogues of G i and G o (10), and antibodies that disrupt interactions between receptors and G proteins (11,12) are examples of the former. Co-purification of receptor⅐G protein complexes (13,14), agonist-promoted photoaffinity labeling with azidoanilido-GTP (15,16), and agonist-promoted GTP␥S binding (17) have been used to examine interactions directly. None of these techniques, however, is used widely, and many suffer from practical limitations. Even the use of purified receptors and G proteins to assess the potential for communication has led to some debate regarding the fidelity of interactions (18).
Spodoptera frugiperda (Sf9) cells have recently been established as an intact cell setting for reconstitution of the human 5-hydroxytryptamine 1A (5-HT 1A ) receptor with mammalian G protein subunits (19). Receptors endogenous to Sf9 cells have yet to be characterized, but so far have not interfered with the analysis of numerous mammalian receptors introduced through recombinant baculoviruses. The levels of G proteins endogenous to Sf9 cells, moreover, are quite low relative to those of mammalian G proteins that can be similarly introduced. Sf9 cells also carry out processing events that support the normal targeting of receptors and G protein subunits to the cell membrane, thus providing a relatively normal phospholipid milieu for interaction. In the previous study (19), the 5-HT 1A receptor was demonstrated to couple to various members of the G i family, i.e. G i , G o , and G z , as assessed by Gpp(NH)p-sensitive increases in affinity for radiolabeled agonists.
A valuable index of coupling is the process of G protein activation itself. An assay of activation permits an evaluation of coupling without resort to radiolabeled agonists and circumvents presumptions regarding changes in agonist affinity. More importantly with respect to many other assays, the requirement for effector activity as a means of monitoring coupling is eliminated. In the present study, we have investigated the selectivity of receptor⅐G protein coupling by measuring directly the activation of G proteins introduced into Sf9 cells. The measurement is based on agonist-promoted binding of [ 35 S]GTP␥S to G protein ␣ subunits isolated subsequently by immunoprecipitation. We have used this methodology to study the coupling of four different receptors (the 5-HT 1A , ␤ 1 -adrenergic, neurokinin-1 (NK 1 ), and thrombin receptors) to members of each class of G protein. The expected selectivity in coupling, whereas only intimated previously, was confirmed, and novel interactions between receptors and G proteins were identified.

EXPERIMENTAL PROCEDURES
Baculoviruses-Recombinant baculoviruses encoding ␣ s-s , ␣ i1 , ␣ q , ␤ 1 , and ␥ 2 (20 -22) were kindly provided by Drs. T. Kozasa and A. Gilman at Southwestern Medical Center (Dallas), and those for ␣ 12 and ␣ 13 were a gift from Dr. N. Dhanasekaran at Temple University School of Medicine (Philadelphia). Baculoviruses for the rat ␤ 1 -adrenoreceptor, human NK 1 receptor (23), and human thrombin receptor were gifts from Drs. E. Ross at Southwestern Medical Center (Dallas), T. Fong at Merck & Co., and Drs. S. Seiler and P. Rose at Bristol-Myers Squibb (Princeton, NJ), respectively. Those encoding the 5-HT 1A receptor and ␣ z were constructed in this laboratory (19).
Cell Culture and Membrane Preparation-Sf9 cells were cultured as described previously (19) but with charcoal-treated serum. For infection with recombinant baculoviruses, the cells were subcultured in monolayer and infected with one or more viruses at a multiplicity of infection of at least one for each virus. The medium was replaced 16 h following infection with Sf900II optimized serum-free medium (Life Technologies, Inc.). The cells were harvested at 48 h and homogenized in ice-cold 20 mM HEPES (pH 8.0), 1 mM EDTA, 0.1 mM phenylmethysulfonyl fluoride, 10 g/ml leupeptin, and 2 g/ml aprotinin by repeated passage through a 26-gauge needle. The homogenate was centrifuged at 100 ϫ g for 5 min, and the resulting supernatant fraction was centrifuged at 16,000 ϫ g for 30 min to pellet the membranes. The membranes were washed and resuspended at ϳ3 mg/ml protein in homogenization buffer for storage at Ϫ70°C. In experiments where the thrombin receptor was expressed, the thrombin protease inhibitor D-Phe-Pro-Arg chloromethyl ketone (1 M) and N ␣ -tosyl-Lys chloromethyl ketone (100 M) were included throughout the period of infection and subsequent expression of receptor.
Assay of [ 35 S]GTP␥S Binding-Membranes (20 g of protein) from Sf9 cells expressing receptors and/or G protein subunits were resuspended in 55 l of 50 mM Tris-HCl (pH 7.4), 2 mM EDTA, 100 mM NaCl, 1 M GDP, and a concentration of MgCl 2 calculated to give the desired concentration of free Mg 2ϩ . [ 35 S]GTP␥S (1300 Ci/mmol, DuPont NEN) was added to a final concentration of 30 nM, and the incubation was allowed to proceed for 5 min at 30°C in the absence or presence of a selected agonist. The incubation was terminated by adding 600 l of ice-cold 50 mM Tris-HCl (pH 7.5), 20 mM MgCl 2 , 150 mM NaCl, 0.5% Nonidet P-40 (Calbiochem), 1% aprotinin, 100 M GDP, and 100 M GTP. After 30 min, the extract was transferred to an Eppendorf tube containing 2 l of non-immune serum pre-incubated with 150 l of a 10% suspension of Pansorbin ® cells (Calbiochem). Nonspecifically bound proteins were removed after 20 min by centrifugation. The extract was then incubated for 1 h at 4°C with 10 l of a G protein ␣ subunit-directed antiserum, pre-immune serum, or non-immune serum, all of which had been pre-incubated with 100 l of a 5% suspension of protein A-Sepharose. With the exception of the ␣ 12 -directed antiserum, generated toward the peptide QENLKDIMLQ, the antisera have been described previously (19,24). Immunoprecipitates were collected and washed three times in the extraction buffer, once in the buffer without detergent, and then boiled in 0.5 ml of 0.5% SDS followed by addition of 4 ml of Ecoliteϩ TM (ICN, Costa Mesa, CA). The samples were analyzed directly by scintillation spectrometry.
Other Immunological Procedures-Immunotransfer blotting and quantitation of ␣ z was accomplished as before (19). Immunoprecipitation of [ 35 S]methionine-labeled proteins was also accomplished as before (19) but under the conditions of extraction and immunoprecipiation used for the nucleotide-binding assays above. Efficiency of immunoprecipitation for antiserum 6354 under these conditions was calculated by comparison to the amount of [ 35 S]methionine-labeled ␣ z immunoprecipitated by antiserum 8645 following denaturation (presumed to be 20% (25)).

RESULTS
Activation of G proteins in membranes prepared from Sf9 cells was examined first for the combination of G z and the human 5-HT 1A receptor. G z is a member of the G i family and was chosen based on experiments indicating that the binding of [ 35 S]GTP␥S to ␣ z is strictly dependent on co-expression of receptor (see below). G z also exhibits the capacity to couple to the 5-HT 1A receptor, as do other members of the G i family (19). Fig.  1 represents a set of experiments in which Sf9 cell membranes containing G z (i.e. ␣ z , ␤ 1 , and ␥ 2 ) and the 5-HT 1A receptor were incubated with [ 35 S]GTP␥S Ϯ serotonin (5-HT) over a range of Mg 2ϩ concentrations. ␣ z was subsequently immunoprecipitated with antiserum 6354 (generated toward residues 24 -33), and bound [ 35 S]GTP␥S was counted directly. As shown in the figure, the binding of [ 35 S]GTP␥S to ␣ z was dependent on Mg 2ϩ and was optimum in the range of 0.5-10 mM of the divalent cation. Binding was enhanced by serotonin but also occurred in the apparent absence of agonist. The use of non-immune serum instead of antiserum 6354 confirmed specificity of binding for ␣ z , as did pretreatment of antiserum 6354 with the peptide used for immunization (not shown). Similar results were achieved with antiserum 2921, directed toward the C terminus of ␣ z (residues 346 -355).
Having established that G z communicates with the 5-HT 1A receptor as monitored by [ 35 S]GTP␥S binding, we examined further the requirements of the binding assay for receptor and G protein subunits (Fig. 2). As above, a significant degree of binding of [ 35 S]GTP␥S to ␣ z was evident when the 5-HT 1A receptor, ␣ z , and ␤ 1 ␥ 2 were co-expressed, and inclusion of serotonin in the assay increased binding further. Identical results were obtained when the 5-HT 1A receptor and ␣ z were expressed together but without addition of ␤ 1 ␥ 2 . Co-expression of the receptor and ␤ 1 ␥ 2 without ␣ z confirmed that the binding was specific for ␣ z . Omission of the receptor demonstrated that G z alone did not bind [ 35 S]GTP␥S.
The amount of ␣ z following infection of Sf9 cells with recombinant viruses encoding the 5-HT 1A receptor, ␣ z , ␤ 1 , and ␥ 2 was 15-30 pmol of subunit per mg of membrane protein, and the efficiencies of extraction and immunoprecipitation were 90 and 60%, respectively. We calculated that 0.5 pmol of [ 35 S]GTP␥S was immunoprecipitated with ␣ z per mg of membrane protein (following treatment of membranes with agonist). Thus, the amount of bound [ 35 S]GTP␥S was 5-10% of immunoprecipitated ␣ z . Binding was not enhanced by increasing the concentration of [ 35 S]GTP␥S nor by omitting GDP from the assay (GDP was used to suppress agonist-independent binding). Binding of [ 35 S]GTP␥S could be increased 2-3-fold by increasing the time of incubation to 60 min.
The assay was next extended to G proteins from other families and to other receptors. Fig. 3 illustrates the ability of the chosen antisera to immunoprecipitate the respective ␣ subunits. The antisera were generated with peptides corresponding the C-terminal 10 amino acid residues of ␣ s (1191), ␣ q (0945), ␣ 12 (121), and ␣ 13 (120). Results obtained with the two antibodies specific for ␣ z are shown for comparison. No subunit was evident when non-immune serum was substituted or when the Sf9 cells were not infected. The appearance of ␣ q as two bands has been reported previously (21). Western blots using the "consensus" antisera 8645 and 1398 reveal that expression levels of the different subunits were within severalfold of each other (␣ 12 could not be analyzed by this means). Binding of [ 35 S]GTP␥S to each of the ␣ subunits co-expressed with ␤ 1 ␥ 2 but not receptors is shown as a function of Mg 2ϩ in Fig. 4. No binding occurred for ␣ z , ␣ 12 , or ␣ 13 at any concentration of Mg 2ϩ . ␣ s and ␣ q , in contrast, bound [ 35 S]GTP␥S in a Mg 2ϩ -dependent fashion, as did ␣ i1 (not shown). Subsequent experi- Solubilization and immunoprecipitation of ␣ subunits was achieved as described under "Experimental Procedures." A, antisera were generated with peptides corresponding the C-terminal 10 amino acid residues of ␣ s (1191), ␣ q (0945), ␣ 12 (121), and ␣ 13 (120). Immunoprecipitated proteins were resolved by SDS-polyacrylamide gel electrophoresis and analyzed for incorporated label by autoradiography. The exposure time was 2-3 days. B, antisera were generated with peptides corresponding to amino acid residues 24 -33 (antiserum 6354) or the C-terminal 10 amino acid residues (2921) of ␣ z . Where indicated, the antisera were pre-incubated with 1 g of cognate peptides. Visualization of immunoprecipitated ␣ z was carried out as described for the previous panel. The potential for coupling of G z to ␤ 1 -adrenergic, thrombin, and NK 1 (substance P) receptors was next assessed (Fig. 5). The 5-HT 1A receptor, as noted previously, caused an increase in [ 35 S]GTP␥S binding in the absence of added agonist and promoted a further increase when agonist (serotonin) was added. Binding was also promoted by the thrombin and NK 1 receptors, though to a lesser extent. In the latter instances, agonists were required. The ␤ 1 -adrenergic receptor failed to activate G z .
A completely different order of selectivities was evident for G s (Fig. 6). As noted above, ␣ s binds some [ 35 S]GTP␥S regardless of receptor. Omission of ␣ s revealed that the binding was relatively specific for the introduced subunit. Activation of the ␤ 1 -adrenergic receptor with isoproterenol caused a 3-4-fold increase in binding. No increase was evident without agonist. Co-expressed 5-HT 1A , thrombin, and NK 1 receptors, with or without agonists, had no effect. The level of binding for the G s /␤ 1 -adrenergic receptor combination (ϳ90,000 cpm) was considerably higher than that achieved for G z with any receptor, despite equivalent expression of the two G proteins.
As with G s , a significant level of receptor-independent binding was observed for G q (Fig. 7). Omission of the recombinant ␣ q indicated that some of the binding could be accounted for by ␣ q (or a cross-reactive subunit) endogenous to Sf9 cells. Binding to the endogenous subunit was enhanced by activation of the NK 1 receptor. However, the signal provided by endogenous subunit was well below that achieved with the mammalian subunit. A modest degree of [ 35 S]GTP␥S binding was elicited by the activated 5-HT 1A receptor and was dependent on the mammalian subunit (not shown). A much higher level of binding was attained with activated thrombin and NK 1 receptors. The receptor-enhanced binding in all three instances was agonistdependent. The ␤ 1 -adrenergic receptor, regardless of agonist, did not promote binding.
Communication of the receptors with G 13 is shown in Fig. 8. The data for G 12 were identical (not shown). In the case of these two G proteins, only two of the receptors (the thrombin and NK 1 receptors) promoted binding of [ 35 S]GTP␥S. No binding was observed without agonists. The -fold enhancement in binding achieved by agonists, in fact, was the highest for any of the receptor/G protein combinations (greater than 6-fold). As noted previously, GDP was included in all binding assays to suppress receptor-independent binding, an action we had confirmed for G z and G s . However, GDP suppresses [␣-32 P]GTP azidoanilide incorporation into ␣ 12 and ␣ 13 in platelet membranes (16).
Consistent with this report, we found that removal of GDP, while having no effect on the negligible receptor-independent binding of [ 35 S]GTP␥S by ␣ 12 and ␣ 13 , increased binding promoted by agonists by about 50%. DISCUSSION We have examined the potential of various receptors to couple to members of each of the four families of G proteins. The method used, receptor-promoted binding of [ 35 S]GTP␥S to G protein ␣ subunits expressed in Sf9 cells, is powerful. Purification of receptors and G protein subunits is not required for reconstitution, nor is the assay compromised by the heterogeneity in both types of proteins inherent to mammalian expression systems. Importantly, receptor⅐G protein coupling can be analyzed without resort to effector activity. This latter property allows the modeling of receptor⅐G protein interactions directly (but nevertheless in a native milieu) and a definition of coupling for those G proteins whose effectors are not yet known. We have demonstrated here that the ␤ 1 -adrenergic receptor couples selectively to G s and not to other G proteins, that the 5-HT 1A receptor is selectively coupled to G z (of the G proteins tested) but shows some activity toward G q , and that the thrombin and NK 1 receptors couple to G 12 and G 13 , as they do to G q and G z , but not G s . Selectivities in interactions previously intimated are now demonstrated directly, and novel interactions are identified. G z was used as a representative of the G i family and as a prototype in the design of the assay. Among the most important traits exhibited by G z was an inability to bind [ 35 S]GTP␥S without co-expression of receptor. This property was evident at all concentrations of Mg 2ϩ and was subsequently found to be shared with G 12 and G 13 . The inability to bind [ 35 S]GTP␥S under the constraints of the assay was consistent with the low rates of exchange of GDP for GTP␥S established previously for purified ␣ z , ␣ 12 , and ␣ 13 (26 -28). G s and G q both displayed significant levels of receptor-independent binding at millimolar concentrations of Mg 2ϩ . Although GDP/GTP␥S exchange has not been fully analyzed for G q (21), purified G s exchanges GDP for GTP␥S quite rapidly at high concentrations of Mg 2ϩ (29). G z is activated to the greatest extent by the serotonin-activated 5-HT 1A receptor. The activation by this receptor was predicted based on the communication between the two proteins implied previously (19) and the fact that the 5-HT 1A receptor is coupled to members of the G i family in mammalian neurons (30,31). We found no evidence for activation of G s through the 5-HT 1A receptor, as once implied (32,33), but did document a modest activation of G q . The latter finding is consistent with activation of phosphoinositide hydrolysis in several types of cells expressing the 5-HT 1A receptor at high concentrations of agonist (34) and is also reminiscent of a somewhat paradoxical affect of ␣ q on ligand affinity previously noted in Sf9 cells (19). Activation of G z by the agonist-activated 5-HT 1A receptor was clearly greater than the activation achieved by the thrombin and NK 1 receptors, while the converse was true for activation of G q .
Only a small proportion of immunoprecipitated ␣ z from Sf9 membranes containing the 5-HT 1A receptor and exposed to serotonin was complexed with [ 35 S]GTP␥S. The low level of binding may simply be related to a normal low rate of GDP/ GTP␥S exchange. Alternatively, it may reflect some instability of the [ 35 S]GTP␥S⅐subunit complex through extraction and immunoprecipitation. We were somewhat surprised that introduction of ␤␥ had no influence on [ 35 S]GTP␥S binding to ␣ z . We had determined previously that ␣ z alone could increase the affinity of the 5-HT 1A receptor for agonist but that a greater degree of coupling was achieved upon co-expression with ␤ 1 ␥ 2 (19). To some extent, the apparent inactivity of ␤␥ might be explained by the fact that expression of the two additional subunits causes an approximately 2-fold suppression in levels of ␣ z (not shown). However, we also suspect that ␤␥ endogenous to Sf9 cells may act catalytically with respect to the activation process.
A significant activation of G z occurred in the presence of the 5-HT 1A receptor but without serotonin. On the one hand, serotonin may have been carried over from the serum in which the cells were initially cultured. However, we employed serum-free medium during the time at which 5-HT 1A receptors were expressed and washed the cells and membranes extensively. The receptor, instead, may exhibit constitutive activity. G proteincoupled receptors can convert between active and inactive conformations, a process especially evident in overexpression systems (35). Of the four types of receptors studied here, however, only the 5-HT 1A receptor exhibits a readily identified activity.
The greatest degree of selectivity in the interaction of a receptor with a G protein was encountered at the level of the ␤ 1 -adrenergic receptor and G s . G s was activated only by the ␤ 1 -adrenergic receptor, and the receptor had no action on any G protein but G s . The pairing of the ␤ 1 -adrenergic receptor and G s was obviously expected (36). Another possible interaction with G s , involving the NK 1 receptor, was also sought, since an earlier report had linked this receptor not only to the activation of phosphoinositide hydrolysis but, at very high concentrations of agonist, to the stimulation of adenylyl cyclase (37). However, we were unable to observe any interaction between the NK 1 receptor and G s . We suspect that the stimulation of adenylyl cyclase observed previously was indirect.
With respect to G q , the activation by thrombin and NK 1 receptors was anticipated, though, as for all other pairings but that of the ␤ 1 -adrenergic receptor and G s , had not been meas- ured directly in previous work. The stimulation of phosphoinositide hydrolysis by thrombin in most types of cells is largely insensitive to PTX, implying the use of a G q -like protein (38). Sensitivity to PTX, though partial, is common, however, suggesting at least some contribution by G i . That thrombin can communicate with G i is quite clear from its ability to inhibit adenylyl cyclase through a PTX-sensitive element in a large number of cells. Conjoint utilization by thrombin of G q and G i is consistent with our data, wherein G z is used as a representative of the G i family. Substance P also stimulates phosphoinositide hydrolysis in a PTX-insensitive fashion (39), and the addition of G q/11 to phospholipid vesicles containing NK 1 receptors results in conversion of the receptors to a high affinity state (40). Utilization of G i (and hence G z in our experiments here) by NK 1 receptors is less well documented. However, the NK 1 receptor appears to regulate a large conductance Cl Ϫ channel through a member of the G i family (41). The list of receptors that activate G z can now be extended to those for thrombin and substance P.
We were most interested in the potential of the different receptors to activate G 12 and G 13 . Receptors normally linked to these G proteins are poorly characterized, and effectors have not yet been identified. Our results clearly demonstrate that thrombin and NK 1 receptors link to the activation of G 12 and G 13 , while ␤ 1 -adrenergic and 5-HT 1A receptors do not. That thrombin should utilize these G proteins is consistent with its ability to support incorporation of [␣-32 P]GTP azidoanilide into ␣ 12 and ␣ 13 in platelet membranes (16) and with the block of thrombin-stimulated DNA synthesis with an antibody directed toward ␣ 12 (42). Our results substantiate the capacity of the cloned receptor for thrombin (43), as distinct from a recently deduced second receptor (44), to achieve the activation. It is intriguing that the two receptors that activate G 12 and G 13 here, the cloned thrombin and NK 1 receptors, also activate G q and G z . Whether activation of the latter two G proteins conjointly is predictive of coupling to G 12 and G 13 is worth pursuit. We were also interested to note that the ␤ 1 -adrenergic receptor does not couple to G 12 or G 13 , as several reports had indicated that ␤-adrenergic receptors activate Na ϩ /H ϩ exchange, a correlate of G 12 or G 13 activation (6), independently of G s (45,46). The fact that the 5-HT 1A receptor does not couple to G 12 and G 13 is also notable. Serotonin operating through 5-HT 1A receptors is not viewed to be a complete mitogen, but thrombin is (47). It is conceivable that the mitogenic properties of thrombin are linked to the activation of G 12 and G 13 . Overexpression of ␣ 12 or ␣ 13 , or expression of GTPase-deficient mutants, is linked to unregulated cell growth (6).
Sf9 cells constitute a well defined, intact cell setting upon which the expression of mammalian receptors and G proteins can be superimposed. Our data support the conclusion that GTP␥S binding is an effective means of monitoring activation of G proteins by receptors. Our data also define the interactions of several receptors with representatives of each family of G protein. Novel interactions have been identified, and their authenticity is supported by the selectivity in interactions otherwise predicted from measurements of second messenger regulation. We anticipate that the Sf9 reconstitution system will lend itself to the analysis of inverse agonism and the coupling achieved by orphan receptors. We also propose the use of the Sf9 reconstitution system for developing or otherwise optimiz-ing techniques to map receptor⅐G protein communication in mammalian cells.