RGS4 inhibits Gq-mediated activation of mitogen-activated protein kinase and phosphoinositide synthesis.

Recombinant regulators of G protein-signaling (RGS) proteins stimulate hydrolysis of GTP by alpha subunits of the Gi family but have not been reported to regulate other G protein alpha subunits. Expression of recombinant RGS proteins in cultured cells inhibits Gi-mediated hormonal signals probably by acting as GTPase-activating proteins for Galphai subunits. To ask whether an RGS protein can also regulate cellular responses mediated by G proteins in the Gq/11 family, we compared activation of mitogen-activated protein kinase (MAPK) by a Gq/11-coupled receptor, the bombesin receptor (BR), and a Gi-coupled receptor, the D2 dopamine receptor, transiently co-expressed with or without recombinant RGS4 in COS-7 cells. Pertussis toxin, which uncouples Gi from receptors, blocked MAPK activation by the D2 dopamine receptor but not by the BR. Co-expression of RGS4, however, inhibited activation of MAPK by both receptors causing a rightward shift of the concentration-effect curve for both receptor agonists. RGS4 also inhibited BR-stimulated synthesis of inositol phosphates by an effector target of Gq/11, phospholipase C. Moreover, RGS4 inhibited inositol phosphate synthesis activated by addition of AlF4- to cells overexpressing recombinant alphaq, probably by binding to alphaq.GDP.AlF4-. These results demonstrate that RGS4 can regulate Gq/11-mediated cellular signals by competing for effector binding as well as by acting as a GTPase-activating protein.

Heterotrimeric G proteins transduce extracellular signals detected by transmembrane receptors into appropriate cellular responses (1,2). The intensity and duration of these responses depend on the relative rates of biochemical reactions that turn G proteins on and off. The G protein switch turns on when receptors promote replacement of GTP for GDP bound by ␣ subunits of ␣␤␥ trimers, leading to dissociation of active G␣⅐GTP from the ␤␥ dimer and consequent regulation of downstream effectors. A GTPase activity intrinsic to ␣ subunits turns off signals by converting ␣⅐GTP to inactive G␣⅐GDP, which then binds to and inactivates ␤␥. For pure G␣ subunits in vitro the turnoff reaction is slow, Յ4 min Ϫ1 (2). In contrast, many G protein-mediated physiological responses must turn off much more rapidly, in fractions of a second.
Two classes of GTPase-activating protein (GAP) 1 have been reported to accelerate deactivation of trimeric G proteins. One class includes G protein effectors, such as phospholipase C (PLC) and the cGMP phosphodiesterase ␥ subunit, which stimulate GTP hydrolysis by ␣ q and ␣ t , respectively (3,4). Recent investigations have discovered and characterized a second class of G␣-GAPs, the RGS (regulators of G protein signaling) proteins. Pure recombinant RGS proteins display GAP activities for certain G protein ␣ subunits (5)(6)(7)(8)(9). RGS proteins of mammals (8 -12), yeast (13,14), and Caenorhabditis elegans (11) share a conserved RGS domain and apparently share similar mechanisms of action. Indeed, a mammalian RGS can partially complement yeast mutations that inactivate Sst2p, the RGS of Saccharomyces cerevisiae (12).
Recent experiments indicate that RGS4 can interact with ␣ q/11 proteins albeit less efficiently than with ␣ i . A high concentration of ␣ q ⅐GDP bound to AlF 4 Ϫ can inhibit the GAP activity of RGS4 for ␣ o ⅐GTP, presumably because ␣ q ⅐GDP⅐AlF 4 Ϫ competes against ␣ o ⅐GTP for binding the RGS protein (6). Moreover, RGS4 can stimulate the GTPase activity of ␣ q in reconstituted vesicles (15). It is not known, however, whether RGS proteins can serve in intact cells as ␣ q -GAPs and inhibitors of G q -mediated cellular signals. Here we use expression of recombinant RGS4 in COS-7 cells to show that RGS4 can inhibit cellular signals mediated by G q/11 .

DNA Constructs and Transfection of COS-7 Cells-FLAG-tagged
human RGS4 cDNA was a generous gift from John H. Kehrl at the Laboratory of Immunoregulation, NIAID, National Institutes of Health, Bethesda, MD. cDNA constructs for the bombesin receptor (BR), D 2 dopamine receptor (D 2 R), and the ␤ 2 -adrenoreceptor were as described (16). Chinese hamster cDNA encoding an HA-tagged p44 MAPK was a gift from J. Pouysségur, Nice, France. pcDNAI and pCR3 were from Invitrogen, San Diego. COS-7 cells were maintained in Dulbecco's modified Eagle's H21 medium with 10% calf serum. DNA was transfected with adenovirus and DEAE-dextran as described (17). Transfection efficiencies were determined by co-transfection of the plasmid pON249 encoding ␤-galactosidase and assayed as described (17). Expression was consistently * This work was supported by National Institutes of Health Grants CA54427 and GM27800 (to H. R. B.), NIH National Research Service Award Postdoctoral Fellowship GM17533 (to Y. Y.), and HHMI Research Training Fellowship for Medical Students (to P. P. C.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Measurement of p44 HA-MAPK Activity-HA-MAPK activity was assayed by a procedure modified from that described by Faure et al. (18). COS-7 cells were transfected in 6-well plates at 0.8 ϫ 10 6 cells/well and placed in serum-free medium containing 0.1% bovine serum albumin after incubating for 24 h in medium containing 10% calf serum. MAPK activity was measured 48 h after transfection. After PTX pretreatment (100 ng/ml for 4 h), where indicated, cells were stimulated for 10 min with appropriate agonists. HA-MAPK immunoprecipitated from cell lysates was incubated with bovine myelin basic protein (MBP) (Sigma) as a substrate in the presence of [␥-32 P]ATP (DuPont NEN). 32 P-Phosphorylated MBP was quantitated with a PhosphorImager (Molecular Dynamics, Sunnyvale, CA) after resolution on a 14% polyacrylamide gel.
Inositol Phosphate Accumulation-Total cellular inositol phosphates (IP) was measured according to Conklin et al. (19). 24 h after transfection cells were replated in a 24-well plate and labeled for 24 h with myo-[ 3 H]inositol (4 Ci/ml, Amersham Corp.). After washing with medium containing 5 mM LiCl for 10 min, cells were incubated for 45 min at 37°C with the appropriate agonist in the same medium containing LiCl. IP and total inositol fractions were resolved on a Dowex AG 1-X8 formate column (Bio-Rad) (12), and cellular IP content was expressed as the ratio of IP radioactivity to the sum of IP plus inositol radioactivity.
cAMP Assay-Intracellular [ 3 H]cAMP accumulation was estimated by determining the ratio of cAMP to the cellular pool of ATP plus ADP as described (16). 24 h after transfection each 60-mm dish of 1 ϫ 10 6 cells was split into 9 wells in a 24-well plate and incubated in medium containing [ 3 H]adenine (2 Ci/ml, Amersham). 18 -24 h later, cells were washed once with 1 ml of assay medium and incubated in 1 ml of assay medium containing 1 mM 1-methyl-3-isobutylxanthine and agonist (isoproterenol) for 30 min as indicated.

RESULTS AND DISCUSSION
RGS4 Inhibits G q -dependent MAPK Activation by Bombesin-To assess effects of an RGS protein on cellular responses mediated by G proteins in the G q/11 family, we compared MAPK activation by a G q/11 -coupled receptor, BR, and a G i -coupled receptor, D 2 R, that co-expressed with or without recombinant RGS4 in COS-7 cells (Figs. 1 and 2). Expression of RGS4 blocked MAPK activation by the G q/11 -coupled receptor agonist, bombesin (1 nM); PTX, which specifically blocks signaling by receptors that activate G i proteins, did not inhibit the effect of bombesin (Fig. 1A). These results suggest that RGS4 inhibited bombesin signaling to MAPK by inhibiting the action of a G protein other than G i , probably G q/11 . This inference was supported by additional experiments described below. Confirming a previous report (12) that recombinant RGS4 can inhibit G imediated signals in an intact cultured cell, RGS4 markedly inhibited G i -dependent MAPK activation by the D 2 R agonist, quinpirole (10 nM; Fig. 1B); PTX also blocked the D 2 R effect on MAPK (Fig. 1B). RGS4 did not alter the expression of HA-MAPK in these experiments (data not shown).
To further assess the effectiveness of RGS4 in blocking cellular responses mediated by G q/11 and G i , we measured MAPK activation by a range of concentrations of both agonists (Fig. 2). In both cases, a higher concentration of each agonist was required to produce equivalent MAPK activation in cells overexpressing RGS4, that is expression of RGS4 shifted the agonist concentration curve to the right. In RGS4-expressing cells, the apparent EC 50 of bombesin was increased ϳ5-fold ( Fig. 2A). The quinpirole concentration-effect curve was shifted to the right ϳ10-fold (Fig. 2B) indicating a relatively greater effectiveness of RGS4 for inhibiting the G i -mediated effect in comparison with that mediated by G q/11 . The apparent difference in potency could reflect different mechanisms by which RGS4 inhibits the two effects, but it is also consistent with a simpler interpretation that the RGS protein catalyzes GTP hydrolysis less efficiently with ␣ q/11 than with ␣ i proteins.
The RGS-induced rightward shift of signaling concentrationeffect curves (Fig. 2) is predictable from the GAP mechanism of RGS action. When a GAP increases the rate of GTP hydrolysis, equivalent steady-state concentrations of G␣⅐GTP can only be achieved by an increased rate of receptor-catalyzed GTP-for-GDP exchange and thus by higher concentrations of agonistoccupied receptor. The extent of an RGS-induced rightward shift would be limited by the k d of the agonist ligand for the relevant receptor. Although a GAP-induced shift in concentration-response curves has not been previously documented, it would be an attractive way to fine tune responsiveness of cells to extracellular stimuli. In phospholipid vesicles reconstituted with a M 1 -muscarinic acetylcholine receptor and G q , the EC 50 of carbachol for stimulating GTP hydrolysis was increased by addition of purified PLC␤1, an ␣ q -GAP; in this case, addition of the GAP also markedly increased the maximal rate of GTP hydrolysis (3,20).
To determine whether similar concentrations of cellular RGS4 are required to inhibit G i -and G q -dependent hormonal signals, we transfected cells with graded amounts of RGS4 plasmid (Fig. 3), which produced graded cellular amounts of RGS4 protein (Fig. 3C). RGS4 inhibited G i -and G q -mediated elevation of MAPK activity with similar dose-effect curves over a 16-fold range of transfected DNA (Fig. 3, A and B). Although the endogenous amounts of cellular RGS proteins are unknown, this result argues that similar amounts of RGS4 pro-tein are required to produce both inhibitory effects. It is unlikely that all RGS proteins exhibit quantitatively similar abilities to inhibit G i -and G q -dependent hormonal signals. Indeed, another member of the RGS protein family, GAIP, inhibited G i -mediated activation of MAPK much more effectively than that mediated by G q/11 . 2 RGS4 Reduces Accumulation of Inositol Phosphates Stimulated by Bombesin-If RGS4 inhibits BR stimulation of MAPK activity by inactivating G q/11 , the RGS protein should also reduce BR-stimulated synthesis of IP by PLC, the principal effector of G q/11 . Indeed, expression of RGS4 reduced bombesininduced IP accumulation by about 50% (Fig. 4A). The inhibitory effect of RGS4 was probably exerted on G␣ q/11 rather than on G␣ i because PTX failed to inhibit BR-induced IP accumulation (not shown). The BR is likely to stimulate IP accumulation via the ␣ subunit of G q/11 rather than via its ␤␥ subunit because PLC␤1 and PLC␤3, the G protein-responsive PLC isozymes of COS cells, are sensitive to ␣ q/11 stimulation but relatively insensitive to stimulation by G␤␥; COS cells lack PLC␤2, the PLC isozyme that is most sensitive to ␤␥ (21).
RGS4 reduced maximal stimulation of IP accumulation by BR stimulation but did not alter the EC 50 for bombesin (Fig.  4A). In contrast, RGS4 expression did not affect maximal activation of MAPK by bombesin but did cause a rightward shift of the bombesin concentration-effect curve ( Fig. 2A). How could the relations between agonist concentration and response be different, if as seems likely, both BR responses are mediated by G q/11 and stimulation of PLC? Although we do not know the reason for this discrepancy, the two assays were performed 2 P. P. Chi, unpublished result. A and B, cells were transfected with plasmids encoding HA-MAPK (1 g), 1 g each of BR and D 2 R, and the indicated amounts of RGS4. Vector plasmid pCR3 was added to keep the total amount of DNA constant. Cells were exposed for 10 min to 1 nM bombesin or 10 nM quinpirole, and HA-MAPK activities were determined. HA-MAPK activities are expressed in arbitrary units of MBP phosphofluorescence (see "Experimental Procedures"). Data represent the mean Ϯ S.D. of triplicate determinations; two additional experiments gave similar results. Panel C, immunoblots of FLAGtagged RGS4 and HA-tagged MAPK expressed in cells used in the MAPK assays shown in panels A and B. Total proteins from cell lysates were resolved in 14% polyacryamide gels and transferred to nitrocellulose membranes. After blotting with M 2 FLAG antibody, the membranes were stripped and probed a second time with the 12CA5 antibody, directed against the HA tag. Blots were developed with an ECL kit (Amersham). under different conditions and reflect activation of G q/11 and PLC in different ways. BR-mediated elevation of MAPK activity measured 10 min after addition of agonist probably results from some (undefined) combination of signals triggered by diacylglycerol activating protein kinase C isozymes and by inositol trisphosphate (InsP 3 ) elevating cytoplasmic Ca 2ϩ . The IP measurements, in contrast, assessed accumulation at 45 min of total radioactive inositol phosphates in cells labeled with radioactive inositol and exposed to LiCl, which inhibits IP degradation. InsP 3 constitutes only a fraction of the total IP pool, and LiCl may not alter concentrations of InsP 3 and total inositol phosphates in the same way. Consequently, the extent and time course of bombesin-induced changes in InsP 3 under conditions used in the MAPK experiments need not parallel bombesin-induced changes in total IP accumulation.

FIG. 3. RGS4 inhibits both G i and G q -mediated activation of MAPK in a dose-dependent manner. Panels
As expected from the reported (5-8) inability of RGS4 to stimulate GTP hydrolysis by the ␣ subunit of G s , the RGS protein had no effect on cAMP production stimulated by the ␤ 2 -adrenoreceptor agonist, isoproterenol (Fig. 4B) or on the cAMP-mediated activation of MAPK by isoproterenol (not shown).
RGS4 Interacts with ␣ q ⅐GDP⅐AlF 4 Ϫ in Intact Cells-To support the idea that RGS4 inhibits PLC stimulation by an effect on ␣ q/11 , we took advantage of a recently discovered property of RGS proteins in vitro, their ability to bind G␣ proteins whose nucleotide binding pockets contain GDP complexed with AlF 4 Ϫ (6,8,9). This property is thought to reflect enhanced affinity of RGS proteins for a G␣ conformation that mimics the transition state of GTP hydrolysis; it was useful for our purposes because the conformation of G␣⅐GDP⅐AlF 4 Ϫ also allows it to regulate activity of the appropriate effector. Although RGS4 reportedly (6) binds G␣ q ⅐GDP⅐AlF 4 Ϫ less tightly than G␣ i ⅐GDP⅐AlF 4 Ϫ , we imagined that an RGS4-␣ q interaction would inhibit stimulation of IP accumulation in cells transfected with recombinant G␣ q and exposed to AlF 4 Ϫ . This turned out to be the case (Fig. 5). AlF 4 Ϫ elevated cellular IP accumulation in cells expressing recombinant G␣ q but had no effect in untransfected cells; by itself, recombinant G␣ q produced a smaller but reproducible elevation of cellular IP content. These results suggest that increased abundance of G␣ q caused a modest elevation in the cellular concentration of its GTP-bound form, and that addition of AlF 4 Ϫ activated PLC still further by binding to the GDP-bound form of transfected G␣ q , rendering it capable of activating the effector enzyme. Co-expression of RGS4 with G␣ q substantially inhibited IP accumulation in response to AlF 4 Ϫ (Fig. 5). RGS4 presumably inhibited effector stimulation in this case not by accelerating GTP hydrolysis but by binding to and sequestering G␣ q ⅐GDP⅐AlF 4 Ϫ . RGS4 also decreased the elevation of MAPK activity seen in untreated cells transfected with G␣ q (not shown), an effect that probably reflects acceleration of GTP hydrolysis by G␣ q .
In summary, we present two new sets of observations. While RGS proteins are known to inhibit signals mediated by G i , we show for the first time that an RGS protein can interact with G␣ q/11 and inhibit signals transduced by G␣ q/11 in intact cells. RGS4 probably inhibits bombesin responses by acting as a GAP, that is, by stimulating the intrinsic GTPase activity of G␣ q/11 . Our experiments also raise the possibility that RGS4 inhibits bombesin responses by sequestering the GTP-bound active conformation of G␣ q/11 (that is, by the mechanism that probably inhibits the AlF 4 Ϫ response), in addition to stimulating GTP hydrolysis.
Second, we found that RGS4 induces rightward shifts in concentration-effect curves for agonists acting on receptors coupled to either G i or G q/11 . It is likely that other RGS proteins modulate hormonal signals mediated by G i and G q/11 in much the same way. For each response, the extent of the rightward shift will depend on the local concentration of RGS protein and its relative affinity for the G protein involved. Thus different complements of RGS proteins could allow two cells to mount quantitatively different responses to the same concentration of a physiological agonist even when both cells use the same receptors and G proteins. If the relevant receptor couples to two distinct G proteins (for instance, to G q and G s or to G i and G q ), differing cellular complements of RGS proteins with distinct G␣ selectivities could even produce qualitatively different responses of two cells to the same agonist.