Specific Involvement of G Proteins in Regulation of Serum Response Factor-mediated Gene Transcription by Different Receptors*

Regulation of serum response factor (SRF)-mediated gene transcription by G protein subunits and G protein-coupled receptors was investigated in transfected NIH3T3 cells and in a cell line that was derived from mice lacking Gαq and Gα11. We found that the constitutively active forms of the α subunits of the Gqand G12 class of G proteins, including Gαq, Gα11, Gα14, Gα16, Gα12, and Gα13, can activate SRF in NIH3T3 cells. We also found that the type 1 muscarinic receptor (m1R) and α1-adrenergic receptor (AR)-mediated SRF activation is exclusively dependent on Gαq/11, while the receptors for thrombin, lysophosphatidic acid (LPA), thromboxane A2, and endothelin can activate SRF in the absence of Gαq/11. Moreover, RGS12 but not RGS2, RGS4, or Axin was able to inhibit Gα12 and Gα13-mediated SRF activation. And RGS12, but not other RGS proteins, blocked thrombin- and LPA-mediated SRF activation in the Gαq/11-deficient cells. Therefore, the thrombin, LPA, thromboxane A2, and endothelin receptors may be able to couple to Gα12/13. On the contrary, receptors including β2- and α2-ARs, m2R, the dopamine receptors type 1 and 2, angiotensin receptors types 1 and 2, and interleukin-8 receptor could not activate SRF in the presence or absence of Gαq/11, suggesting that these receptors cannot couple to endogenous G proteins of the G12 or Gqclasses.

Hormones, neurotransmitters, and many other biologically active molecules, such as lysophosphatidic acid (LPA), 1 thrombin, catecholamines, endothelin, etc., transduce their signals through heterotrimeric G proteins (1,2). Molecular cloning has revealed at least four classes of G protein ␣ subunits: G␣ s , G␣ i , G␣ q , and G␣ 12 (3). The G␣ s subunits and G␣ i subunits regulate adenylyl cyclase activities, while the G␣ q subunits regulate phospholipase C activities. However, the function of the G␣ 12 class of G proteins, which includes G␣ 12 and G␣ 13 , remains to be elucidated. Activated forms of G␣ 12 and G␣ 13 , when transfected into fibroblast cells, were shown to induce transforma-tion phenotypes (4 -6), suggesting that this class of G proteins may be involved in cell growth regulation. Moreover, G␣ 12 and G␣ 13 were shown to induce formation of stress fibers in fibroblast cells through small G protein RhoA (7). This observation was supported by the report that G␣ 12 activated serum response factor (SRF) through RhoA (8). The in vivo function of G␣ 13 was also investigated using the gene-targeting technique in mice. Mice lacking G␣ 13 are embryonic lethal apparently due to the failure to develop vasculature structures, indicating that G␣ 13 may be involved in the function of endothelial cells (9). In the same study, thrombin-mediated chemotaxis of fibroblasts lacking G␣ 13 was blocked, indicating that the thrombin receptor couples to G␣ 13 . This is consistent with the observation that thrombin as well as a thromboxane A2 receptor agonist could stimulate the binding of a photo-affinity GTP analog to G␣ 13 (10). However, there were contradictory reports with regard to the involvement of the G q class of G proteins in RhoA and SRF activation (7,8,16).
RGS (regulator of G protein signaling) proteins belong to a growing family of proteins that contains homologous RGS domains (11,12). Some of these proteins such as GIAP and RGS4 were shown to inhibit G protein-mediated signaling by interacting with the G␣ i and G␣ o subunits and stimulating their GTPase activities (13), which are also referred to as GTPaseactivating protein activities. RGS2 and RGS4 were found to inhibit G␣ q -mediated activation of phospholipase C␤ (14 -16), implying that they may function as GTPase-activating protein for G␣ q . Interestingly, most of these RGS proteins were unable to inhibit G s -mediated signaling, and they could not stimulate the GTPase activity of the G␣ s proteins (13,14). Regulation of the G 12/13 proteins by RGS proteins has not been investigated.
In this report, we characterized the abilities of the G protein subunits and of a number of GPCRs to stimulate SRF-mediated gene transcription using a cotransfection system. We found that the ␣ subunits of the G q and G 12 classes of G proteins can induce SRF activation in a C3-dependent manner. We also found that the activation of SRF by some of GPCRs depends exclusively on G␣ q/11 , while others do not. Those that activate SRF independently of G␣ q/11 may act through G␣ 12/13 . In addition, we, for the first time, identified a RGS protein, RGS12, that can inhibit the G␣ 12/13 function.

EXPERIMENTAL PROCEDURES
Cell Culture, Transfection, and Luciferase Assay--NIH3T3 and the G␣ q/11 -deficient cells were maintained in Dulbecco's modified Eagle's medium containing 10% fetal calf serum at 37°C under 5% CO 2 . The G␣ q/11 -deficient cell line was established from mice lacking both G␣ q and G␣ 11 (17). For transfection, cells (5 ϫ 10 4 cells/well) were seeded into 24-well plates the day before transfection. Cells were transfected with 0.5 g of DNA/well using LipofectAMINE Plus (Life Technologies, Inc.), as suggested by the manufacturer. The transfection was stopped after 3 h by switching to culture medium containing 0.5% fetal bovine serum. Cell extracts were collected 24 h later for luciferase assays.
Luciferase assays were performed using Boehringer Mannheim Constant Light Luciferase Assay Kit as instructed. Cell lysates were first taken for determining in a Wallac multicounter the fluorescence intensity emitted by GFP proteins, which are cotransfected with the luciferase reporter gene plasmid and used to normalize the transfection efficiency. The Wallac counter (Wallac AG&G, Finland) is capable of counting both fluorescence and luminescence. Then, the luciferase substrate was added to the cell lysates, and luciferase activities were determined by measuring luminescence intensity using the same counter. Luminescence intensities were normalized against fluorescence intensities. DNA concentrations were adjusted if transfection of any of the cDNAs resulted in significant differences between normalized and non-normalized data.
Construction of Expression Plasmids-All of the G protein subunits and GPCRs were in pCMV expression vectors as described previously (18 -20). The SRE.L-luciferase reporter plasmid was constructed as described in Ref. 7, except the luciferase gene was used as the reporter instead of the chloramphenicol acetyltransferase gene. The CRE-luciferase reporter gene plasmid was purchased from Strategene, La Jolla, CA. RGS2, RGS4, and RGS12 (kindly provided by Sheng-Cai Lin) were also in the pCMV vector. Axin (kindly provided by F. Costantini) was in the pcDNA3 vector (Invitrogen). C3 was kindly provided by Alan Hall.

RESULTS
A cotransfection system was used to characterize signal transduction pathways mediated by G proteins that lead to the regulation of SRF-dependent gene transcription. SRF-mediated gene transcription was evaluated by determining the activity of luciferase, the production of which is regulated by a transcription regulatory sequence element, called SRE.L. SRE.L is a derivative of c-Fos serum response element (SRE), to which SRF but not tertiary complex factor binds (21). Thus, SRE.L-mediated production of luciferase mainly depends on the activity of SRF. The abilities of various G protein subunits to regulate SRE.L-mediated gene transcription were determined by cotransfecting NIH3T3 cells with the reporter gene plasmid and cDNA encoding one of constitutively active G␣ subunits. We found that cells expressing activated ␣ subunit of G q , G 11 , G 14 , G 16 , G 12 , or G 13 , produced markedly higher levels of luciferase than those expressing the control ␤-galactosidase (LacZ), whereas expression of activated G␣ i or G␣ o did not (Fig.  1). This indicates that the ␣ subunits of the G q and G 12 classes of G proteins can lead to SRF activation. The finding that C3 blocked SRF activation by the G protein ␣ subunits suggests that the small GTP-binding protein RhoA (Fig. 1) may mediate the SRF activation. C3 (Clostridium butulinum C3 transferase) is a specific RhoA inactivator, which ADP-ribosylates RhoA (21). In our transfection system, C3 inhibited only RhoA-induced but not Cdc42- (Fig. 1) or Rac1-(data not shown) induced SRF activation, indicating that C3 acted specifically.
Activation of SRF by the G q and G 12 family of G proteins allows us to determine which receptors can couple to these G proteins to activate SRF. Many cells, including fibroblasts, contain endogenous receptors for thrombin and LPA, which belong to a superfamily of GPCR. In addition, the thrombin receptor was shown previously to couple to the G 12/13 proteins (10) so that it may function as a positive control. We found that both thrombin and LPA were able to stimulate SRE.L-mediated gene transcription in NIH3T3 cells transfected with the reporter gene plasmids ( Fig. 2A). This result suggests that NIH3T3 cells contain endogenous receptors for thrombin and LPA. A number of other GPCRs were also tested for their abilities to stimulate SRE.L-mediated gene transcription in transfected 3T3 cells. Cells expressing m1R and ␣ 1 -AR showed marked increases in luciferase activities in response to carbachol and norepinephrine, respectively (Fig. 1B). Neither carbachol nor norepinephrine elicited any change in the luciferase activity in cells transfected with the reporter gene plasmid alone (Fig. 1A), indicating that there are no endogenous receptors for carbachol and norepinephrine in NIH3T3 cells. A mutant of ␣ 1 -AR, ␣ 1 -AR⌬2, was also tested in the same cotransfection system. ␣1-AR⌬2, which was unable to couple to G␣ q/11 to activate phospholipase C (22), lost the ability to activate SRF in the presence of ligand norepinephrine. This suggests that ␣ 1 -AR-mediated SRF activation appears to depend on G␣ q/11 in fibroblasts. The fact that the expression of C3 was able to abolish thrombin-, LPA-, norepinephrine-, and carbachol-induced SRF activation suggests that the SRF activation by these ligands be mediated by RhoA (Fig. 2, A and B). Furthermore, we tested G i -coupling m2R and IL-8 receptor and G s -coupling ␤ 2 -AR. None of these receptors was able to stimulate SRF activity in response to their ligands (Fig. 2B). These results are consistent with the observation that the G i and G o are not involved in regulation of SRF. It has been demonstrated by various approaches that m1R and ␣ 1 -AR couple to the G q proteins (18,23). Although the receptors for thrombin and LPA have not been rigorously tested for their abilities to couple to the G proteins of the G q family, they were shown previously to stimulate inositol phosphate accumulation (24), suggesting that they may couple to the G q proteins. To test the roles of G␣ q/11 in SRF activation by these receptors, a fibroblast cell line derived from mice lacking G q/11 was used. Thrombin, LPA, carbachol, and norepinephrine were added to the G␣ q/11 -deficient cells that were transfected with the SRE.L-luciferase reporter gene plasmid. Both thrombin and LPA were able to stimulate SRE.L-mediated gene transcription, which can be blocked by C3 (Fig. 3A). Thrombin and LPA showed EC 50 values of about 5 units and 4 M in stimulation of luciferase activities, respectively (Fig. 4, A and  B). Thus, this cell line, like NIH3T3 cells, also contains endogenous thrombin and LPA receptors, and the receptors for thrombin and LPA are able to activate SRF independent of G␣ q/11 . The G␣ q/11 -deficient cells were also cotransfected with the cDNA encoding the thrombin receptor and the reporter gene plasmid, and cells expressing the recombinant thrombin receptors showed higher thrombin-induced production of luciferase than those transfected with the control plasmid (Fig. 3A). Therefore, both endogenous and recombinant thrombin receptors are able to activate SRF in a G␣ q/11 -independent way.
The inability of norepinephrine and carbachol to elicit reporter gene transcription (data not shown) may be due to the lack of endogenous receptors. Therefore, these two ligands were tested in the G␣ q/11 -deficient cells transiently expressing ␣ 1 -AR and m1R, respectively. The ligands were still unable to elicit responses, even in the presence of the recombinant receptors (Fig. 3B). However, the responses of cells to norepinephrine were restored when G␣ q was reintroduced back into the G␣ q/11 -deficient cells by cotransfection with ␣ 1 -AR and the re-porter gene (Fig. 3B). The same result was also observed for carbachol when m1R and G␣ q were coexpressed in the G␣ q/11deficient cells (Fig. 3B). Therefore, we conclude that ␣ 1 -AR and m1R are dependent exclusively on G␣ q/11 in SRF activation. This conclusion further strengthens the idea that the G q proteins are capable of activating RhoA and SRF. Since ␣ 1 -AR is able to couple to all the members of the G q class of G proteins, including G␣ 14 and G␣ 16 (19), the inability of ␣ 1 -AR to activate SRF in the G␣ q/11 -deficient cell line suggests that there are not sufficient levels of endogenous G␣ 14 and G␣ 16 in this cell line. Thus, if any receptor can induce SRF activation in this cell line, it would suggest that this receptor is able to couple to G proteins other than the G q class, which would be the G 12/13 proteins. Thus, the G␣ q/11 -deficient cell line may be used for testing the coupling of receptors to the G proteins of the G 12 class. We tested a number of receptors in this G␣ q/11 -deficient cell line, including the endothelin receptors 1a and 1b, thromboxane A2 receptor, angiotensin (AT) receptors type 1a and type 2, ␣ 2 -and ␤ 2 -AR, and dopamine receptors type 1 and 2. Endothelin was able to activate the SRF.L-mediated gene transcription B and C, the G q/11 -deficient cells were transfected the same as in A in addition to various GPCRs (0.2 g). The next day cells were lysed 6 h after the addition of ligands as described in Fig. 2, except 10 nM U46619, 100 nM angiotensin, and 1 M dopamine were added into cells expressing the thromboxane A2 receptor (Tx2), angiotensin receptors (AT), and dopamine receptor (DR) type 1, respectively. Data are processed and presented as described in the legend to Fig. 1. in cells expressing endothelin receptor 1a (Fig. 3C) or 1b (data not shown) with EC 50 values of about 0.25 nM (Fig. 4C and data not shown), while the thromboxane A2 receptor agonist U46619 gave an EC 50 value of 1 nM (Figs. 3C and 4D). Angiotensin, norepinephrine, isoprenaline, and dopamine failed to elicit any SRF activation in cells expressing AT receptors, ␣ 1 -AR, ␤ 2 -AR, and dopamine receptors, respectively (Fig. 3C). These results suggest that endothelin receptors and the thromboxane A2 receptor may be able to couple to G 12/13 .
The effects of G␤␥ subunits on SRF activation were also tested in the same cotransfection system. Cells coexpressing ␤ 1 and ␥ 1 or ␤ 1 and ␥ 5 showed approximately 1-fold more luciferase activity than those expressing the control LacZ (Fig. 5A). This result suggests that G␤␥ subunits may also be involved in regulation of SRF, but with less potency than the G␣ q and G␣ 12 subunits. To further investigate the role of G␤␥ in SRF activation, we tested if G␤␥ has any effect on G␣-mediated SRF activation by coexpressing activated G␣ 13 or G␣ q with G␤␥. Interestingly, cells coexpressing G␣ 13 QL and G␤ 1 ␥ 1 or G␤ 1 ␥ 5 showed higher luciferase activities than the sum of the activities shown by cells expressed G␤␥ and G␣ 13 alone. This suggests that G␣ 13 may work synergistically with G␤␥ in activation of SRF. Similar synergistic effects were also observed with G␣ q and G␤␥. To confirm the above observation, the G i -coupled m2-muscarinic receptor (m2R) was expressed in NIH3T3 cells, and carbachol-induced SRF-mediated transcription was determined. As expected, cells expressing m2R showed marginal carbachol-mediated increases in luciferase activities (Fig. 5A), which may be attributed to G␤␥ released from the G i proteins, since G␣ i does not activate SRF (Fig. 1). However, when m2R was coexpressed with m1R in NIH3T3 cells, there was a synergistic response to the muscarinic ligand carbachol (Fig. 5B), i.e. carbachol induced more luciferase activities in cells coexpressing m1R and m2R than those expressing m1R or m2R alone. Furthermore, Pertussis toxin (Ptx), which did not inhibit the response to carbachol in cells expressing m1R, inhibited the response to carbachol in cells coexpressing m1R and m2R (Fig.  5B). It appears that Ptx blocked only the part contributed by m2R. Ptx is a bacteria toxin that ADP-ribosylates the ␣ subunits of the G i , G o , and G t proteins and blocks the activation of these G proteins by receptors. Therefore, it is reasonable to conclude that G␤␥ works synergistically with G␣ q and G␣ 13 in regulation of SRF.
The wild-type G␣ 12 and G␣ 13 , when expressed in NIH3T3 cells, also showed significant stimulation of SRF-mediated transcription (Fig. 6A), although the activity is usually onefifth of that of the QL mutant when the same amount of DNA is used in transfection (data not shown). The relative high activities of the wild-type G␣ 12 and G␣ 13 may be due to the slow intrinsic GTPase activities of these G␣ subunits (10,25,26). The SRF activation by the wild-type G␣ 12 and G␣ 13 allowed us to test if the RGS proteins could inhibit G␣ 12/13 -mediated effects. We tested RGS2, RGS4, RGS12, and Axin. RGS2 and RGS4 are among the well characterized RGS proteins, which show GTPase-activating protein activities for the G i and/or G q families of ␣ subunits (13,14,27,28). RGS12 (29) and Axin (30) are two recently cloned proteins that contain the RGS domains but have not been tested for their abilities to regulate G protein-mediated signaling. Coexpression of RGS12 significantly inhibited G␣ 12 -and G␣ 13 -induced SRF activation (Fig. 6A), whereas RGS2, Axin (Fig. 6A), and RGS4 (data not shown) showed little effects. Moreover, expression of RGS12 did not inhibit activated RhoA-mediated SRF activation (data not shown), indicating that inhibition of G␣ 12/13 -mediated effects by RGS12 is not due to nonspecific inhibition of downstream proteins. The inhibition by RGS12 is also unlikely to be the result of the changes in the expression levels of cotransfected G proteins, since coexpression of RGS12, Axin, or RGS2 did not alter the expression levels of G␣ 12 or G␣ 13 (Fig. 6E). Therefore, RGS12 is likely to affect the function of G␣ 12/13 directly. The effects of the RGS proteins on LPA and thrombin-mediated SRF activation were also tested in the G␣ q/11 -deficient fibroblast cell line, where LPA and thrombin-mediated SRF activation is presumably mediated by the G 12 family of G proteins. The G␣ q/11 -deficient cells were cotransfected with SRE.L re- Lluciferase reporter plasmid, 0.15 g of GFP expression construct, and 0.2 g of LacZ or 0.1 g of G␤ and G␥ in the presence or absence of 0.01 g of activated G protein ␣ subunits. One day later cells were lysed, and GFP levels and luciferase activity were determined as described in the legend to Fig. 1. B, NIH3T3 cells were cotransfected with 0.15 g of SRE.L-luciferase reporter plasmid, 0.15 g of GFP expression construct, 0.1 g of LacZ, 0.05 g of m1R, and/or 0.15 g of m2R expression plasmid as indicated in the figure. One day later, 1 M carbachol (Car) was added to cells that were treated with or without 200 ng/ml of Ptx for 2 h. Six hours after the addition of the ligand, cells were lysed, and GFP levels and luciferase activity were determined as described in the legend to Fig. 1. porter plasmids and cDNA encoding RGS2, RGS4, RGS12, or Axin, and LPA-induced production of luciferase was determined. Cells expressing RGS12 showed consistently lower luciferase activities than those expressing the control LacZ, RGS2, RGS4, or Axin (Fig. 6B). RGS12 was also able to inhibit thrombin-mediated SRF activation in the G␣ q/11 -deficient cells (data not shown). Since RGS12 was able to inhibit G␣ 12/13 (Fig.  6A), RGS12-mediated attenuation of LPA and thrombin-induced SRF activation may result from the inhibition of endogenous G␣ 12/13 proteins by RGS12 in the G␣ q/11 -deficient cells.
The effects of RGS12 on other G protein-mediated signaling pathways were also investigated. RGS12 as well as RGS2, RGS4, or Axin did not affect ␤-adrenergic agonist isoprenaline-induced CRE-mediated gene transcription (Fig. 6C), suggesting that none of the RGS proteins could inhibit the G s function. However, all the RGS proteins except Axin were also able to inhibit norepinephrine-induced SRF activation in 3T3 cells expressing ␣ 1 -AR (Fig. 6D). As demonstrated earlier, ␣ 1 -ARmediated SRF activation is dependent on G␣ q/11 proteins. Thus, inhibition of ␣ 1 -AR-mediated SRF by the RGS proteins suggests that these RGS proteins may inhibit G␣ q/11 . Both RGS2 and RGS4 were also able to inhibit norepinephrineinduced phospholipase C activation in COS-7 cells coexpressing ␣ 1 -AR. 2 Thus, the action of RGS12 may not be specific to the G 12 class of G proteins. DISCUSSION In this report, we have characterized the involvement of G protein subunits in activation of SRF by a number of GPCRs. Our findings that G␣ 12 and G␣ 13 can activate SRF are consistent with previous reports that G␣ 12 activates SRF through RhoA (8) and that G␣ 12/13 induces formation of stress fibers via RhoA (7). The literature, however, appears to be inconsistent with regard to the involvement of G␣ q in SRF activation (8,31). Our findings described in this report demonstrate that not only G␣ q , but also all other members of this class of G proteins can activate SRF probably through RhoA. The fact that m1R and ␣ 1 -AR use G␣ q/11 exclusively in activation of SRF strongly supports the involvement of G q in SRF activation. Not all G q -coupled receptors, however, activate SRF in G␣ q/11 -dependent pathways. Receptors, including the endothelin receptors and thromboxane receptor A2, both of which are known to couple to G q to activate phospholipase C, can activate SRF in a G␣ q/11 -independent way (32). Moreover, receptors for thrombin and LPA can also lead to RhoA and SRF activation independently of G␣ q/11 , although these two receptors may also be able to couple to the G i proteins in addition to the G q proteins.
Interestingly, among those we tested none of the receptors that are previously known to predominantly couple to G i or G s can induce activation of SRF in the presence or absence of G␣ q/11 . Unlike the G␣ subunits of the G q and G 12 classes, the expression of activated G␣ o or G␣ i does not activate SRF. In addition, G s does not appear to activate SRF because the ␤-AR agonist isoprenaline did not stimulate SRF-mediated transcription in cells expressing ␤ 2 -AR, while isoprenaline was able to stimulate CRE-mediated transcription. The inability of ␤-AR to activate SRF also suggests that ␤ 2 -AR cannot couple to endogenous G 12/13 in the fibroblasts. It is also apparent that m1R and ␣ 1 -AR as well as ␣ 2 -AR, ␤ 1 -AR, the IL-8 receptor, and D1 and D2 receptors are unable to couple to G␣ 12 or G␣ 13 , because these receptors could not activate SRF in the G␣ q/11deficient cells. The thromboxane A2 and thrombin receptors were shown previously to couple to G␣ 12 and G␣ 13 using different methods (9,10,33). Thus, SRF activation by these receptors should at least in part be mediated by G␣ 12/13 . The LPA receptor and endothelin receptors may also be able to couple to G␣ 12/13 unless there exist yet-to-be identified G protein ␣ subunits that can also activate SRF in the G␣ q/11 -deficient mouse fibroblast cells. Under the same premise, the inhibition of LPA and thrombin-induced SRF activation by RGS12 may be attributed to the inhibition of G␣ 12/13 . In fact, RGS12 can inhibit recombinant G␣ 12 and G␣ 13 -mediated SRF activation (Fig. 6A). The inability of RGS2 and RGS4 to inhibit G␣ 12/13 -, LPA-, or thrombin-mediated SRF activation indicates that these two RGS proteins do not act on G␣ 12/13 . Therefore, there is apparent specificity in interactions between RGS proteins and G proteins. Although Axin contains a RGS domain, it is unable to regulate any of the known G proteins. L-luciferase reporter plasmid, 0.15 g of GFP expression construct, and 0.2 g of LacZ or RGS in the presence or absence of 0.05 g of G␣ 12 or G␣ 13 subunits. One day later cells were lysed, and GFP levels and luciferase activity were determined as described in the legend to Fig. 1. B, the G␣ q/11 -deficient cells were cotransfected with 0.15 g of SRE.L-luciferase reporter plasmid, 0.15 g of GFP expression construct, and 0.2 g of LacZ or various RGS expression plasmids. Varying amounts of LPA were added the next day for 6 h. Then GFP levels and luciferase activity were determined as described in the legend to Fig. 1. C, NIH3T3 cells were cotransfected with 0.15 g of CRE-luciferase reporter plasmid, 0.15 g of GFP expression construct, and 0.2 g of LacZ or RGS expression plasmids. Isoprenaline (10 M) was added the next day for 6 h. Then GFP levels and luciferase activity were determined as described in the legend to Fig. 1. D, NIH3T3 cells were cotransfected with 0.15 g of SRE.L-luciferase reporter plasmid, 0.15 g of GFP expression construct, 0.1 g of ␣ 1 -AR, and 0.1 g of LacZ or RGS. 1 M norepinephrine was added the next day for 6 h. Then GFP levels and luciferase activity were determined as described in the legend to Fig. 1. G␤␥ subunits are also involved in regulation of SRF, especially together with G␣ q and G␣ 13 . Since SRF is downstream of the signal cascade, we do not know where the G␣-mediated pathways interact with the G␤␥-mediated ones. It appears that the interactions are upstream of RhoA because C3 was able to completely abolish carbachol-mediated effects in cells coexpressing m1R and m2R. It is also not clear how G␣ q and G␣ 13 regulate RhoA and SRF. Recent studies suggested that the Tec family of nonreceptor tyrosine kinases may be regulated by G␣ q (34) and G␣ 12/13 and that Tec kinases regulate RhoA in a C3-dependent manner. 3 In addition, G␤␥ was shown to regulate Btk, a member of the Tec family, via phosphatidylinositide 3-kinase (35). Thus, the synergistic activation of SRF by G␣ and G␤␥ may lie in the kinases. Further studies are needed to better understand these questions.