Regulators of G Protein Signaling 6 and 7

Regulators of G protein signaling (RGS) proteins that contain DEP (disheveled, EGL-10,pleckstrin) and GGL (G protein γsubunit-like) domains form a subfamily that includes the mammalian RGS proteins RGS6, RGS7, RGS9, and RGS11. We describe the cloning of RGS6 cDNA, the specificity of interaction of RGS6 and RGS7 with G protein β subunits, and certain biochemical properties of RGS6/β5 and RGS7/β5 complexes. After expression in Sf9 cells, complexes of both RGS6 and RGS7 with the Gβ5 subunit (but not Gβs 1–4) are found in the cytosol. When purified, these complexes are similar to RGS11/β5 in that they act as GTPase-activating proteins specifically toward Gαo. Unlike conventional Gβγ complexes, RGS6/β5 and RGS7/β5 do not form heterotrimeric complexes with either Gαo-GDP or Gαq-GDP. Neither RGS6/β5 nor RGS7/β5 altered the activity of adenylyl cyclases types I, II, or V, nor were they able to activate either phospholipase C-β1 or -β2. However, the RGS/β5 complexes inhibited β1γ2-mediated activation of phospholipase C-β2. RGS/β5 complexes may contribute to the selectivity of signal transduction initiated by receptors coupled to Gi and Go by binding to phospholipase C and stimulating the GTPase activity of Gαo.

Initial studies of proteins that belong to the regulators of G protein signaling (RGS) 1 family demonstrated that they are GTPase-activating proteins (GAPs) that accelerate hydrolysis of GTP bound to the ␣ subunits of certain heterotrimeric G proteins (1)(2)(3). As such, RGS proteins can function as negative regulators of G protein-mediated signal transduction by speeding deactivation of the active form of G ␣ subunits, thereby promoting formation of inactive G protein heterotrimers (G␣ GDP ␤␥). It is now clear that some proteins that contain an RGS domain have more complex functions. For example, p115 RhoGEF, which contains an RGS domain near its amino terminus, is an effector for G protein action, catalyzing guanine nucleotide exchange on the monomeric G protein Rho more efficiently when stimulated by G␣ 13 -GTP. The capacity of G␣ 13 to activate p115 RhoGEF is dependent on the RGS domain of p115, which also accelerates hydrolysis of GTP by G␣ 13 (4,5). p115 RhoGEF thus resembles phospholipase C␤, a well characterized effector for G protein action that also deactivates its regulator (G␣ q ) by acting as a GAP (6 -8).
A subfamily of RGS proteins has been identified in which each member possesses so-called DEP (disheveled, EGL-10, pleckstrin) and GGL (G protein ␥ subunit-like) domains in addition to an RGS domain (9,10). Members of this group include mammalian RGS proteins (RGS6, RGS7, RGS9, and RGS11), a Drosophila RGS protein (dRGS7), and EGL-10, an RGS protein found in Caenorhabditis elegans (10 -12) Functionally, the GGL domain was shown to specify binding of RGS11 or a fragment of RGS7 to the G protein ␤5 subunit (10). It has also been shown that both RGS7 and RGS9 can be isolated from brain and retina as a complex with G␤5 (13)(14)(15) 2 and that the distribution of mRNA for the GGL-containing RGS proteins and G␤5 overlap in these tissues (10, 11, 16 -20). In addition, Snow and co-workers (10) demonstrated that the RGS11/␤5 complex accelerates hydrolysis of GTP bound to the ␣ subunit of the G protein G o . The functional significance of the DEP domains of these proteins remains unknown. The mammalian members of this subfamily of RGS proteins are found predominantly in the central nervous system (10,11,16,17), and to date, their role in G protein-mediated signal transduction is unknown.
The superficial resemblance of the complex formed by G␤5 and a GGL-containing RGS protein to a G protein ␤␥ subunit complex suggests potential targets for modulation of G protein signaling. G protein ␤␥ subunits regulate the activity of a diverse array of effectors (adenylyl cyclases, ion channels, and phospholipase C-␤ isoforms, among others), play a significant role in interactions of G proteins and G protein-coupled receptor kinases with receptors, and facilitate signal transfer from G protein-coupled receptors to mitogen-activated protein kinase cascades (21). However, relatively little is known about the signaling properties of G␤5. Of the five known G␤ subunits, G␤5 is the clear outlier. Although G␤s 1-4 are 80 -90% identical to each other, G␤5 is only about 50% homologous to its relatives (18). Functionally, the recombinant ␤5␥2 dimer appears capable of stimulating the phospholipase activity of PLC-␤2 (18,22,23) and coupling G␣ q with ET B or m1 muscarinic receptors in crude reconstituted systems (23). However, this complex was unable to stimulate either type II adenylyl cyclase in vitro (23) or the mitogen-activated protein kinase pathway in COS cells (22).
We describe herein the cloning of RGS6 and explore the biochemical properties of RGS6 and RGS7, the most closely related member of this group of RGS proteins. We demonstrate that both RGS6 and RGS7 associate strongly and specifically with G␤5. These RGS/␤5 complexes have been purified following expression directed by recombinant baculoviruses and tested for interactions with certain recombinant G protein ␣ subunits and known effectors for G protein action. 35 S]GTP␥S and [␥-32 P]GTP were obtained from NEN Life Science Products. Spodoptera frugiperda (Sf9) cells were maintained and recombinant baculoviruses were amplified as described previously (24). Baculoviruses encoding amino-terminally hexahistidinetagged G␤1 and 5 have been described previously (10,25). Viruses encoding hexahistidine-tagged G␤3 and 4 were prepared similarly. G␣ q , G␣ z , G␣ 12 , and p115 RhoGEF were generously provided by Dr. Tohru Kozasa (University of Texas Southwestern Medical Center); PLC-␤1 and PLC-␤2 by Dr. Paul Sternweis (University of Texas Southwestern Medical Center); antisera against RGS6 (U1895) and RGS7 (U1480) by Dr. Andrejs Krumins (University of Texas Southwestern Medical Center); antiserum against G␤5 (SGS-1) by Dr. William Simonds (National Institutes of Health); an antiserum that reacts with G␤1-4 subunits (B600) by Dr. Susanne Mumby (University of Texas Southwestern Medical Center); and a recombinant baculovirus encoding RGS7 by Dr. Sheu-Fen Lee (University of Texas Southwestern Medical Center). Synthetic oligonucleotides were obtained from commercial sources or prepared by the Pharmacology Core Facility of the University of Texas Southwestern Medical Center.

Materials-[␥-
Cloning of Human RGS6 cDNA-A partial rat RGS6 cDNA sequence (GenBank: RNU32436), identified by homology to the C. elegans EGL-10 protein, was used with the BLAST algorithm to identify a human clone (GenBank: HUMORFE) containing sequences homologous to the consensus RGS domain found in all members of the RGS protein family. Further BLAST alignments with a human expressed sequence tag data base identified g874443 as identical to HUMORFE except with 11 additional nucleotides and an EcoRI site at its 5Ј end. This incomplete RGS6 cDNA clone was obtained from the Lawrence Livermore National Laboratory collection of expressed sequence tags through Genome Systems (St. Louis, MO), and sequences derived from its 5Ј region were used to prepare oligonucleotide primers for 5Ј-RACE (rapid amplification of cDNA ends) cloning of the missing DNA sequences.
Marathon-Ready™ (CLONTECH Laboratories, Palo Alto, CA) cDNA from human brain was used for 5Ј-RACE. Oligonucleotide primer B413 (5Ј-CCACATCCTGTAGGGGTTGTTTC-3Ј) derived from RGS6 sequence downstream of the EcoRI site at nucleotide position 1117 ( Fig. 1) was used with oligonucleotide primer AP1 (supplied by CLONTECH) to produce double-stranded cDNA according to a polymerase chain reaction protocol provided by the supplier. Following removal of oligonucleotide primers by gel filtration, the enriched cDNA was used to prepare a specific 1153-base pair human RGS6 5Ј-RACE product using primers B413 and B415 (5Ј-TCTAGACTGCAGGATGGCTCAAGGATC-CGGGGATC-3Ј). The latter primer was derived from a human RGS6 cDNA sequence obtained from GenBank with the accession number AF073920. The human RGS6 5Ј-RACE cDNA was joined to the g874443 partial RGS6 cDNA at their common EcoRI restriction site at nucleotide position 1117 by a sequential polymerase chain reaction protocol. The amplified cDNA from this procedure (1730 base pairs) was ultimately cloned into baculovirus transfer vector pFastBac HTb for preparation of a human RGS6 protein with a hexahistidine tag near its amino terminus.
Construction of a Truncated RGS6 cDNA-A modified RGS6 cDNA lacking the sequences coding for the consensus RGS box domain was constructed using the polymerase chain reaction, and the amplified product (997 base pairs) was cloned into pFastBac HTb. The truncated RGS6 protein (RGS6⌬R, 41.4 kDa) includes amino acid residues 1-324 of human RGS6 in addition to 25 residues of amino-terminal sequences containing the hexahistidine tag and 8 residues of carboxyl-terminal sequence derived from the pFastBac HTb vector.
Interaction of RGS/G␤5 Complexes with G␣-GDP Proteins-The ability of purified, myristoylated G␣ o and G␣ q to form heterotrimeric complexes with ␤1␥2 and RGS6 or 7/G␤5 complexes was assessed as follows. To 100 l of buffer A (20 mM Na Hepes (pH 8), 20 mM ␤-mercaptoethanol, 2 mM MgSO 4 , 2 M GDP, 100 mM NaCl, and 0.005% C 12 E 10 ) was added 10 pmol of myristoylated G␣ o or 10 pmol of G␣ q and 25 pmol of hexahistidine-tagged RGS6/␤5, RGS6⌬R/␤5, RGS7/␤5, or ␤1␥2. Following a 30-min incubation on ice, 25 l of Ni-NTA resin (Qiagen) equilibrated in buffer A was added and incubated for an additional 30 min. The Ni-NTA resin was collected by centrifugation and washed three times with 150 l of buffer B (buffer A supplemented with 400 mM NaCl and 10 mM imidazole). Proteins bound to the Ni-NTA resin were eluted with 50 l of buffer C (buffer A supplemented with 150 mM imidazole). The identity of proteins eluted from Ni-NTA resin was determined by immunoblotting after separation through 11% sodium dodecyl sulfate-polyacrylamide gels.
Interaction of RGS6 and RGS7 with G␤ Subunits-Cultures of Sf9 cells (50 ml, 1.0 ϫ 10 6 cells/ml) were infected with RGS6 or RGS7 baculoviruses in addition to baculoviruses encoding one of each of the five G␤ subunits. In these experiments, the G␤ proteins were synthesized with a His 6 tag at their amino termini; the RGS proteins were not tagged. Infections were carried out at a multiplicity of infection of 1 for each virus. Infected cells were harvested by centrifugation, suspended to a density of 2.5 ϫ 10 7 cells/ml in lysis buffer, subjected to nitrogen cavitation, and clarified by centrifugation as described above. Supernatant fractions were applied to Ni-NTA resin in small columns. The resin was washed with 10 column volumes each of buffer D, buffer E, and buffer J (buffer A plus 150 mM NaCl and 20 mM imidazole). Each sample was eluted with buffer H. Column fractions were analyzed by immunoblotting using the following antibodies: U1895 (anti-RGS6 diluted 1:10,000), U1480 (anti-RGS7 diluted 1:4000), SGS-1 (anti-␤5 diluted 1:2000), and B600 (anti-G␤1-4 diluted 1:10,000). Particulate fractions of the cells were washed once with lysis buffer and then resuspended in lysis buffer, flash-frozen in liquid nitrogen, and stored at Ϫ80°C.
Miscellaneous Procedures-G ␣ subunits were purified from Escherichia coli using the methods of Lee et al. (29). Hexahistidine-tagged RGS4 was purified from E. coli as described by Berman et al. (27). R183C G␣ q was prepared from Sf9 cells as described previously (28), as was G␣ o -GTP␥S (30). Measurements of phospholipase C activity and adenylyl cyclase activity were conducted according to published protocols (31,32).

RESULTS
Cloning of Human RGS6 cDNA-A full-length coding sequence for RGS6 was obtained by joining a human expressed sequence tag clone (g874443) encoding all of the 3Ј region to a primer-extended and polymerase chain reaction-amplified cDNA encoding the complete 5Ј region of RGS6 (Fig. 1). The full-length clone contains 2946 nucleotides and has an open reading frame encoding a protein of 472 amino acid residues with a deduced molecular weight of 54,422.
There are three regions of particular interest in the RGS6 protein. The DEP domain comprises amino acid residues 42-121. It is a highly conserved hydrophobic domain found in RGS6, 7, 9, and 11, in addition to a wide variety of other signaling proteins (9). The second region of interest is the GGL domain, defined by Snow et al. (10) as a domain that contains the required elements for interaction with G␤5. Within the full-length RGS6 protein, the GGL domain is found between amino acid residues 261 and 309 (Fig. 1). The third recognizable motif, the RGS domain, is found between amino acid residues 325 and 451 of RGS6. A series of critical amino acid residues in the RGS domain that contact G ␣ was identified from the crystal structure of the complex formed between RGS4 and G␣ i1 (33). Although there is a substantial tolerance for amino acid substitutions within the consensus RGS domain of the family, there is an expected conservation of those residues that are critical for the direct interaction with the switch domains of the G ␣ subunit (at least for those RGS proteins that interact with members of the G␣ i subfamily). Within RGS6, those critical residues are: Glu 357 -Phe-Ser 359 , Glu 361 -Asn 362 , Asn 401 , Asp 403 , Leu 432 , Asp 436 -Ser 437 and Arg 440 . Based on the conservation of these critical residues, we predicted that RGS6 would function as a GTPase-activating protein.
Association of RGS6 or RGS7 with G Protein ␤ Subunits-In a previous study with RGS11 (10), Snow and co-workers demonstrated that full-length RGS11 or a fragment of RGS7 formed specific and high affinity complexes with G␤5. This association required the GGL domain present in RGS11 and RGS7; this domain can also be recognized in RGS6 and RGS9. Although several studies have now shown that RGS7 and RGS9 form heterodimers with G␤5 (10,(13)(14)(15), the specificity of this interaction with G␤5 has yet to be generalized for the entire subfamily of RGS proteins.
We have addressed this question for full-length RGS6 and RGS7 with a recombinant baculovirus expression system. In these experiments, Sf9 cells were infected with an RGS virus either alone or in combination with one of five viruses encoding known G␤ subunits (Fig. 2). When expressed alone or with an RGS protein, significant amounts of G␤5 were found in the supernatant (cytosolic fraction) produced by high speed centrifugation of cell lysates ( Fig. 2 and data not shown). With longer exposure times, it was possible to detect the presence of the other ␤ subunits in the cytosolic fraction (Fig. 2). When expressed alone or with G␤s 1-4, RGS6 and RGS7 are found predominantly in the particulate fraction. However, concurrent expression of RGS6 or RGS7 with G␤5 alters this distribution dramatically, and a significant amount of the RGS protein is found in the cytosolic fraction (Fig. 2). Although attempts to purify an RGS/G␤ complex from the cytosol were unsuccessful in experiments with G␤s 1-4, both RGS6 and RGS7 co-purified with G␤5 reproducibly (see below). Gel filtration of crude cytosolic fractions also revealed that only G␤5 formed a heterodimer with these RGS proteins (data not shown). In summary, these data demonstrate strong and specific association of the mammalian GGL-containing RGS family members with G␤5.
Purification of the RGS6/␤5 and RGS7/␤5 Complexes-The distribution of the RGS6/␤5 and RGS7/␤5 complexes between the soluble and particulate fractions is similar to that observed for G␤5 in brain homogenates (18). Both complexes can be purified to homogeneity from the soluble fraction (Fig. 3) and remain intact when passed through anion and cation exchange resins. Gel filtration analyses indicate a 1:1 complex of RGS7 with G␤5 (data not shown). Two protein bands are visualized with the RGS7 antibody and migrate electrophoretically near the expected position for the protein (Fig. 3, left panel). The upper band in these preparations disappears after treatment with phosphoprotein phosphatases (data not shown). RGS6 migrates as a single protein species (Fig. 3, right panel).
Efforts to purify these RGS/␤5 complexes from the particulate fraction have not been successful. This may be due to several factors, including an instability of the complex in the presence of detergent and/or the presence of a large amount of poorly folded or aggregated recombinant protein in the particulate fraction.
Stimulation of G␣ GTPase Activity by RGS6/␤5 and RGS7/ ␤5-Although GAP activity that is largely specific for G␣ o -GTP was observed previously with the RGS11/␤5 complex (10), Levay et al. (14) claim that the nucleotide-dependent interaction of RGS7 with G␣ o is abolished in the presence of G␤5. Despite this claim, GAP activity of purified RGS6/␤5 or RGS7/␤5 complexes was detected readily. Of interest, both of these complexes showed the same remarkable specificity for G␣ o observed previously with RGS11/␤5 (Fig. 4). The GTPase activity of G␣ o was clearly enhanced by both RGS6/␤5 and RGS7/␤5, but the activities of G␣ i1 , G␣ i2 (data not shown), G␣ i3 (data not shown), G␣ z , G␣ s , G␣ q , and G␣ 12 were not.
Substrate specificity was not altered when crude membranes containing RGS6/␤5 or RGS7/␤5 were tested for GAP activity (data not shown). Interestingly, RGS6 and RGS7 in crude membrane fractions exhibited modest GAP activity toward G␣ o -GTP in the absence of recombinant G␤5 (data not shown). Although we cannot rule out the presence of a S. frugiperda ortholog of G␤5 in these experiments, it is certainly possible (or likely) that association with G␤5 is not a prerequisite for the GAP activity of these proteins. The GAP activity of membraneassociated RGS7 was consistently greater in the presence of recombinant G␤5 but was not augmented by coexpression of RGS7 with other G␤ subunits. However, there may be several explanations for this difference, including activation of RGS7 by G␤5 and/or greater stability or proper folding of RGS7 protein in the presence of G␤5.
Interactions of RGS6/␤5 or RGS7/␤5 with G␣-GDP-We have also assessed the capacity of RGS6 or 7/␤5 complexes to form heterotrimers with G␣ q or myristoylated G␣ o proteins (Fig. 5). Purified (His 6 )RGS6/␤5, RGS7/(His 6 )␤5, or a truncated RGS6 protein lacking the RGS domain ((His 6 )RGS6⌬R/␤5) were incubated on ice for 30 min with GDP and purified G ␣ proteins. We then attempted to detect heterotrimeric complexes by adsorption to and elution from Ni-NTA resin. Both G␣ q and myristoylated G␣ o readily formed stable heterotrimers with ␤1/(His 6 )␥2. By contrast, heterotrimeric complexes were not detected when RGS6, RGS7, or RGS6⌬R/␤5 complexes were tested. In at least this sense, these proteins do not appear to function as G protein ␤␥ subunit-like complexes.
Interactions with Adenylyl Cyclase and Phospholipase C-␤-Although previous studies have demonstrated several roles for ␤␥ complexes in G protein-mediated signaling (21), the G␤5 subunit appears to be functionally restricted when compared with G␤s 1-4 (18, 22, 23). Although the ␤5␥2 dimer can stimulate PLC-␤2, it does not activate type II adenylyl cyclase. ␤5/␥2 also associates preferentially with members of the G q subfamily of G ␣ proteins in vitro and may be selectively released by G q -linked receptors. Nevertheless, G␤5 forms heterodimers with ␥3, ␥4, ␥5, and ␥7 subunits, as well as the GGL-containing RGS proteins (10,18). Thus, the role of G␤5 in signaling may be quite extensive.
We have considered the potential of the RGS6/␤5 and RGS7/␤5 complexes to interact with three isoforms of adenylyl cyclase and two isoforms of PLC-␤. Several ␤␥ dimers inhibit the G␣ s -stimulated activity of type I adenylyl cyclase and activate (conditionally with G␣ s ) type II adenylyl cyclase (34); ␤␥ subunits do not appear to interact with type V adenylyl cyclase. Interestingly, neither RGS6/␤5 nor RGS7/␤5 was able to modulate the activity of any of these adenylyl cyclases in the presence or absence of activated G␣ s (Fig. 6). In addition, neither complex was able to interfere with the capacity of ␤1␥2 to activate type II adenylyl cyclase (data not shown). We considered the possibility that the RGS domain might influence interactions of the complex with effectors. However, neither deletion of the RGS domain of RGS6 nor inclusion of activated G␣ o in these assays failed to alter the earlier results.
G protein ␤␥ subunits are also capable of stimulating the activity of selected isoforms of PLC-␤. PLC-␤2 can be stimulated up to 20-fold by ␤␥ subunits, while the activity of PLC-␤1 is relatively insensitive to the ␤␥ dimer (32,35). However, both RGS/␤5 complexes failed to influence the activity of PLC-␤1 or PLC-␤2 in the presence or absence of activated G␣ o (Fig. 7, A  and B); removal of the RGS domain did not alter these results (Fig. 7B). We did observe a moderate, concentration-dependent inhibition by RGS/␤5 complexes of the ability of ␤1␥2 to activate PLC-␤2 (Fig. 7C). At the highest concentration of RGS/␤5 tested, ␤␥-stimulated phospholipase activity was reduced roughly 25% and 40% by RGS6/␤5 and RGS7/␤5, respectively. The inclusion of activated G␣ o did not influence these results (data not shown). DISCUSSION We describe herein the cloning of RGS6, the specificity of interaction of RGS6 and RGS7 with the G protein ␤5 subunit, and biochemical properties of the RGS6/␤5 and RGS7/␤5 complexes. Sequence alignments of RGS6 with known RGS proteins reveal high sequence homology to a subfamily of RGS proteins that includes RGS7, dRGS7, RGS9, RGS11, and EGL-10. The members of this subfamily each contain a DEP domain near the amino terminus, a GGL domain roughly in the middle of the protein, and an RGS domain near the carboxyl terminus. The mRNA transcripts of the mammalian members of this subfamily are found primarily in the central nervous system, including the retina (10,11,16,17,19,36). Among the subfamily members, RGS7 is most similar in sequence to RGS6. It thus seemed likely that RGS6 and RGS7 would share other important properties. Interestingly, their distribution in the central nervous system is somewhat distinct, which may suggest subtle differences in their roles in G protein-mediated signaling (17).
Both RGS6 and RGS7 can be expressed in Sf9 cells and purified to homogeneity from cytosolic extracts as complexes with G␤5. Like RGS11 (10), the GAP activities of the RGS6/␤5 and RGS7/␤5 complexes in solution (in vitro) appear specific for G␣ o . Previous studies conducted with the RGS homology domain of RGS7 fused to glutathione S-transferase suggested a The concentration of G␣ s -GTP␥S (100 nM) was constant in each assay. RGS6/␤5 and RGS7/␤5 also failed to modulate cyclase activity when G␣ s was omitted from the assay (data not shown). broader substrate specificity (G␣ o , G␣ i1 , G␣ i3 , and G␣ z ) (20,37,38). Interaction of full-length RGS7 or RGS6 with these ␣ subunits may be precluded by the presence of the GGL domain, the DEP domain, other elements of the protein, and/or G␤5.
The GAP activity of RGS6/␤5 and RGS7/␤5 is a property of the complexes themselves. We have not been able to dissociate the RGS7/␤5 complex by incubation with G␣ o -GDP-AlF 4 Ϫ . For example, RGS7/G␣ oa -GDP-AlF 4 Ϫ complexes were not observed when a mixture of RGS7/␤5 and an 8-fold molar excess of G␣ o -GDP-AlF 4 Ϫ were gel-filtered. These observations contradict a previous assertion that G␤5 prevents the interaction of RGS7 and other GGL-containing RGS proteins with G␣ o (14). This assertion was based on binding assays that failed to reveal complexes of G␣ o with RGS7/␤5; GAP assays were not performed. In addition, the authors were incorrect in their statement that Snow et al. (10) tested only fragments of RGS11 (associated with G␤5) for GAP activity. Compared with RGS proteins such as RGS4 and GAIP (39), the affinity of the RGS7/␤5 complex for G␣ o -GDP-AlF 4 Ϫ is modest, 3 and interaction between these proteins would likely go undetected in assays designed to measure binding rather than catalytic activity. It has also been suggested that RGS7 may be important in Gq-mediated signaling (38). To date, we have failed to detect GAP activity for these RGS/␤5 complexes with G␣ q in single turnover assays (Fig. 4) or in steady-state GTPase assays in which the m1 muscarinic cholinergic receptor and heterotrimeric G q were reconstituted in phospholipid vesicles (data not shown).
The specific and high affinity association of the GGL domains of RGS6, RGS7, and RGS11 with G␤5 appears to be a general property of this subfamily of RGS proteins. RGS9 has been isolated from rod outer segment membranes as a complex with G␤5L, a long splice variant of G␤5 found primarily in retina (18). In addition, RGS9 can also form a complex with G␤5 when the two genes are coexpressed in Sf9 cells using a recombinant baculovirus system 4 ; however, the specificity of the RGS9 GGL domain for G␤5 has not yet been tested. There is an apparent ortholog of G␤5 in C. elegans (Q206636), but its association with EGL-10 has not been described (14).
Many of the experiments that we have performed to date have been guided by the hypothesis that RGS/␤5 complexes may play roles analogous to more conventional G protein ␤␥ subunits. Many of these experiments have proven negative, particularly including those designed to detect interactions between RGS/␤5 complexes and GDP-bound G protein ␣ subunits. Similarly, these RGS/␤5 complexes by themselves appear unable to modulate the activities of at least certain effectors for G␤␥ proteins. However, both complexes are capable of inhibiting stimulation of PLC-␤2 by ␤ 1 ␥ 2 (Fig. 7C). Although the magnitude of the inhibition appears modest, the affinity of the interaction may be sufficient to target the RGS domain of these proteins to PLC-␤2. Deactivation of G␣ o in the vicinity of PLC-␤2 by the RGS domain would ensure rapid sequestration of ␤␥ and termination of phosphoinositide-mediated signaling from the relevant subset of receptors. Importantly, these RGS/␤5 complexes do not inhibit ␤ 1 ␥ 2 stimulation of type II adenylyl cyclase, and they do not display significant GAP activity (at least in vitro) for the three isoforms of G␣ i . This constellation of effects raises the possibility of a role of RGS6/␤5 and RGS7/␤5 as enforcers of selectivity in signaling initiated from receptors that can activate G i and G o .