Serum Response Factor Activation by Muscarinic Receptors via RhoA

The muscarinic cholinergic receptor (mAChR) subtypes share high sequence similarity except in their third intracellular loop and COOH terminus, domains thought to be involved in signal transduction. Subtypes M1, M3, and M5 couple mainly through Gαq/11, and M2 and M4 couple mainly through Gαi/o. Whether subtypes within each of these two groups differ in their signaling pathways remains to be resolved. This study focused on nuclear signaling pathways leading to activation of the transcription factor, serum response factor (SRF). Genes encoding M1, M2, and M3 were co-expressed in Jurkat T lymphocytes with a reporter gene driven by a mutant serum response element, SRE.L, which responds to SRF activation. We show that only M1 mAChR activated SRF through a pathway involving the small GTPase RhoA, with no response observed for M2 and M3. Transfection of GTPase-deficient Gα subunits (GαQL; constitutively active form) demonstrated that SRF was activated by Gα13QL but only marginally by GαqQL and Gα12QL in Jurkat cells. Yet transfection of regulator of G protein-signaling protein, RGS2 and RGS4, which inhibit Gαq/11 activity, indicated that Gαq/11 and Ca2+ mobilization were required for SRF activation by M1. Calmodulin inhibitors suppressed the M1 and the Gα13QL pathways, acting both upstream and downstream of RhoA. However, calcineurin inhibitors and the tyrosine kinase inhibitor genistein selectively suppressed SRF activation by M1, but not by Gα13QL, indicating the presence of separate pathways. The calmodulin-dependent tyrosine kinase Pyk2 was also activated by M1 but not M3, and Pyk2 appears also to play a role in M1-SRF activation, as judged by experiments with two dominant-negative Pyk2 mutants. These results reveal a novel calmodulin-dependent RhoA-SRF signaling pathway unique to the M1 mAChR subtype.

sitol phosphates, or diacylglycerol. Long term GPCR effects on the other hand depend on changes in gene expression (1) via multiple pathways involving the Ras (2) and Rho families of small GTPases (3). Several mechanisms by which GPCRs and heterotrimeric G proteins couple to these pathways have been proposed (4), including calmodulin-dependent transactivation of tyrosine kinase receptors (5)(6)(7). Moreover, receptor activation of G␣ 12 and G␣ 13 was shown to activate Rho via Rho-GEF (guanine-nucleotide exchange factor) proteins, which enhance both the GTPase activity of G␣ proteins and the GDP/GTP exchange rate of Rho (8), or via tyrosine kinases such as Pyk2 (9). Nuclear signaling pathways may vary for each GPCR and in each tissue and, therefore, often remain unresolved.
Muscarinic cholinergic receptors (mAChR) also affect gene expression such as transcription of the immediate early genes (IEG), c-fos and c-jun in neuronal cells (10 -14). Multiple mechanisms appear to contribute to nuclear signaling pathways of mAChR (10,12,15). The mAChR subtypes M1-5 share high sequence similarity with each other except in their large third intracellular (i3) loop and the COOH terminus. Subtypes M1, M3, and M5 form a group of receptors mainly coupled through G␣ q/11 , whereas M2 and M4 couple mainly through G␣ i/o (16). However, each GPCR subtype typically interacts with multiple G proteins (17). Patterns of signaling pathways for the M1, M3, M5, and the M2, M4 subtypes are thought to be similar within each group even though the i3 loops and COOH termini of each receptor subtype, the main domains involved in receptor signaling, differ substantially (16). Moreover, i3 loops and COOH termini are highly conserved for each mAChR subtype across species, suggesting distinct conserved functions for each subtype. Recent results with M1 receptor knock-out mice suggest differences in signaling between M1 and other mAChR subtypes coupled to G␣ q/11 (18). Whereas mAChR-dependent G␣ q/11 coupling in brain tissue was reduced by only ϳ50% in the M1 knock-out mice compared with the wild type, suggesting the presence of residual activity by M3 and M5, activation of the mitogen-activated protein kinase/extracellular signalregulated kinase pathway was abolished (18). Because M1 receptors represent a major target for Alzheimer's therapy (19), it is important to identify signaling pathways unique to the M1 subtype.
The present work focuses on the transcriptional regulation of IEGs mediated by mAChRs in an attempt to identify novel pathways that differ among mAChR subtypes. IEGs regulate the expression of a variety of proteins involved in mitogenesis and neuronal differentiation (20,21). Serum response factor (SRF), a key regulator of IEG transcription acting at serum response elements (SRE) (22), is a ubiquitous transcription factor that mediates serum-and growth factor-induced activation of IEGs via mitogen-activated protein kinases Ras and Raf (23). Another SRF activation pathway involves GTPases of the Rho family (24), which also affects cytoskeletal dynamics and converges on Ras/Raf (25). Recent evidence indicates that heterotrimeric G proteins (G␣ q/11 , G␣ 12 , G␣ 13 ) can mediate SRF activation through RhoA (7,9,26,27). mAChRs were also found to activate RhoA via several mechanisms (4,9).
In this study, lymphocyte Jurkat T cells were transfected with genes encoding M1, M2, and M3 mAChR to dissect the signaling pathways from receptor to SRF using a luciferase reporter assay. We show that in this cell line only M1 mAChR, but not M2 and M3, activates SRF-mediated gene transcription via a novel Rho pathway involving calmodulin and calcineurin and the calmodulin-dependent tyrosine kinase Pyk2. These findings demonstrate significant signaling differences between the closely related M1 and M3 receptor subtypes and, moreover, the M2 subtype in a pathway known to affect cytoskeleton dynamics and cell survival.

MATERIALS AND METHODS
Cell Culture and Transfection-Jurkat leukemic T cells were maintained in RPMI 1640 medium containing 10% fetal bovine serum, 50 units/ml penicillin, and 50 g/ml streptomycin in 5% CO 2 incubator at 37°C. Transfections were performed using LipofectAMINE Plus (Invitrogen) according to manufacturer's instructions. Jurkat T cells (1 ϫ 10 7 ) were co-transfected with 0.5 g of SRE.L-luciferase reporter gene plasmid and 2 g of either empty pSG5 vector or mAChRs cDNA in pSG5 vector. The transfection was stopped after 4 h by adding medium containing 5% fetal bovine serum.
The genes encoding the human mAChR subtypes (M1, M2, and M3) in pSG5 vector were obtained from a human placental genomic library as previously described (33,34). The gene encoding G␣ i/q chimera (G␣ iq5 ) was constructed by exchanging of the COOH termini between G␣ i and G␣ q , which switches the specificity of G␣ i -coupled receptor to G␣ q (35) (obtained from Dr. Bruce Conklin, University of California, San Francisco, California).
Reporter Gene Assay-Eighteen hours after transfection, cells were replated in 96-well plates and stimulated with or without the mAChR agonist carbachol at 37°C for 4 h as indicated. Cells were then lysed by reporter lysis buffer (Promega, Madison, WI), and luciferase activities were measured with a luminometer (EG&G). For inhibitor experiments, cells were preincubated with inhibitors for 30 min before adding carbachol.
The Dual-Luciferase reporter assay system (Promega) was used to normalize the transfection efficiency where indicated. pRL-CMV vector yielding constitutive expression of Renilla luciferase was co-transfected with the SRE.L reporter gene vector, and dual luciferase activities were measured according to the manufacturer's protocol.
Receptor Binding Assay-Expression of M1, M2, and M3 mAChR receptors was determined by measuring the binding of N-[ 3 H]methylscopolamine on the surface of intact cells as previously described (36). Briefly, transfected cells were incubated with 1.5-2.0 nM N-[ 3 H]methylscopolamine in phosphate-buffered saline at 12°C for 90 min. Nonspecific binding was determined in the presence of 10 M atropine. After labeling, cells were placed on ice, filtered, and rinsed with ice-cold phosphate-buffered saline three times (S&S No. 32 glass fiber filter). The radioactivity on the filters was determined by liquid scintillation counting.
Intracellular Ca 2ϩ Measurement-Measurement of free intracellular calcium was determined as described (37). Briefly, 18 h after transfection, cells were washed with Krebs-HEPES buffer and loaded with 3 M Oregon Green 488 fluorescent dye at room temperature for 30 min. After loading, cells were washed 3 times with Krebs-HEPES buffer containing 0.5% bovine serum albumin, diluted to ϳ2 ϫ 10 6 cells/ml, and distributed evenly (3 ϫ 10 5 cells/well) into an opaque white 96-well plate (Corning Costar, Cambridge, MA). Buffer control (4 samples/data point) or buffer containing test compounds (4 samples/data point measured immediately before the 4 control samples) was injected sequentially into separate wells, and the fluorescence intensity was monitored at 1-s intervals using an excitation wavelength of 485 nm and emission of 538 nm.
Immunoprecipitation and Immunoblotting of Phosphorylated Pyk2-10 7 Jurkat cells were transfected with 2 g of hM1-pGS5 or hM3-pSG5 together with 0.5 g of Pyk2 plasmid. 24 h after transfection, cells were washed with serum-free medium then resuspended in 0.3 ml of Hanks' solution and incubated at 37°C for 30 min. After treatment with different reagents, cells were placed on ice, and 0.3 ml of 2ϫ lysis buffer was added (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 2 mM EDTA, 0.5% sodium deoxycholate, 2 mM phenylmethylsulfonyl fluoride, 2 mM NaF, 2% Nonidet P-40, 10 mM NaVO 3 , 20 g/ml each of leupeptin and aprotinin). After incubation on ice for 10 min, cell lysates were centrifuged at 13,000 ϫ g for 10 min at 4°C. A suspension of 60 l of 50% protein G-Sepharose beads was added to the supernatants (Amersham Biosciences) and incubated at 4°C for 30 min with rotation. The beads were centrifuged at 13,000 ϫ g for 2 min. The precleared lysates were incubated with anti-Pyk2 monoclonal antibody (Upstate Biotechnology, Lake Placid, NY) at 4°C for 24 h. Then 60 l of 50% protein G beads was added and incubated for another 2 h at 4°C with rotation. The beads were washed 3 times with 1ϫ lysis buffer, and bound proteins were eluted with 30 l of sample loading buffer. Proteins were resolved by SDS-PAGE and transferred to polyvinylidene difluoride membranes, and phosphorylated Pyk2 was detected by anti-phosphotyrosine monoclonal antibody (4G10) (Upstate, Lake Placid, NY).

Activation of SRF by mAChR Receptors-
We tested the ability of transfected M1, M2, and M3 mAChRs to regulate SRFmediated gene transcription using a luciferase reporter gene.
The reporter construct used, SRE.L, is a derivative of c-Fos SRE, which is activated by SRF acting alone and does not require formation of tertiary complex (29). The plasmids expressing different mAChRs were co-transfected with the SRE.L-luciferase reporter gene into Jurkat cells. Treatment with the mAChR agonist carbachol (1 mM) increased SRF activity only in M1 mAChR-transfected cells ( Fig. 1a) even though all three transfected cell lines expressed comparable levels of receptor on the surface, measured with the membraneimpermeable tracer N-[ 3 H]methylscopolamine (300 -400 fmol receptor/mg of protein). SRF activation induced by M1 mAChR was antagonized by 1 M atropine (Fig. 1a), indicating that activation was a mAChR-mediated process.
SRF-mediated gene transcription was activated in M1-transfected cells within 1 h and reached a maximum after 4 h of carbachol treatment. Thus, 4 h of treatment was used in all follow-up experiments. SRF was activated in a dose-dependent manner, with a carbachol EC 50 of 10 M, which is comparable with the EC 50 values obtained for inositol 1,4,5-trisphosphate production and Ca 2ϩ release by M1.
Role of G␣ q/11 -mediated Ca 2ϩ Increase in SRF Activation by M1 mAChR-M1 mAChR has been shown to activate SRF in several cell lines via multiple heterotrimeric G proteins, including G␣ q/11 and G␣ 12/13 (4,39). In contrast to the results with M1, the closely related G␣ q/11 -coupled M3 failed to activate SRF upon stimulation with carbachol in Jurkat T cells. To determine whether M1 and M3 mAChR couple efficiently to G proteins in Jurkat cells, intracellular Ca 2ϩ release by carbachol was measured. Both receptors independently elicited Ca 2ϩ mobilization (Fig. 2, a and b) with similar intensity and duration, indicating effective coupling to G␣ q/11 in this system. To test whether differences in receptor expression could have contributed to the discrepancy in SRF activation between M1 and M3 mAChR, experiments were carried out with different amounts of DNA (ranging from 1 to 10 g of DNA) used for transfection. This did not affect SRF activation by M1, whereas M3 mAChR was inactive under all conditions tested (data not shown).
To test whether G␣ i/o proteins are involved in M1-mediated SRF activation, we treated Jurkat cells with pertussis toxin to inactivate G␣ i/o . Shown in Fig. 1b, pertussis toxin had no significant effect on M1-mediated SRF activation. The G␣ i/o -coupled M2 receptor alone also did not mobilize Ca 2ϩ nor SRF activation in Jurkat T cells ( Fig. 1a and 2c). To further test the involvement of G␣ q/11 signaling in SRF activation, a G␣ i /G␣ q chimeric construct (G␣ iq5 ) was transfected together with M2 receptor. Substituting the COOH terminus of G␣ q with that of G␣ i , G␣ iq5 confers G␣ q/11 -coupling specificity to G␣ i -coupled receptors (35). After co-transfection of M2 receptor with G␣ iq5 , carbachol stimulated a robust Ca 2ϩ release in M2 mAChRexpressing cells comparable with M1-transfected cells (Fig. 2c). However, the same pool of cells failed to stimulate SRF activation (Fig. 1a). This result suggests the hypothesis that G␣ q/11 itself is either not required or insufficient for SRF activation by M1 in Jurkat T cells. G␣ iq5 transfection alone had no detectable effect on SRF activation.
The ability of various G␣ protein subunits to regulate SRF gene transcription was determined by co-transfecting Jurkat cells with the reporter gene plasmid and cDNA encoding various G␣ subunits and their constitutively active mutants, G␣ q Q209L, G␣ 11 Q209L, G␣ 12 Q231L, G␣ 13 Q226L, G␣ 14 Q205L, and G␣ 15 Q212L. To account for differences in expression efficiency, we normalized the data using a dual luciferase plasmid strategy (Promega). Our results showed a similar pattern with or without normalization. Cells overexpressing normal G␣ subunits (G␣ q , G␣ 11 , G␣ 12 , G␣ 13 , G␣ 14 , and G␣ 15 ) did not activate SRF nor did co-transfection of these G␣ subunits enable M3 to activate SRF or enhance M1-mediated SRF activation (data not shown). This result suggests that the inability of M3 to activate SRF does not appear to result from a lack of these G␣ subunits. Of the constitutively active G␣ subunit tested, only G␣ 13 QL strongly activated SRF (Fig. 3a), whereas G␣ q QL and G␣ 12 QL were less effective and in some experiments did not exceed 10 -20% of the level achieved with G␣ 13 QL. Other constructs had no significant effects on SRF activation (Fig. 3a).
Regulators of G protein signaling (RGS) are GTPase-activating proteins capable of attenuating heterotrimeric G protein signaling. RGS2 and RGS4 proteins function as GTPase-activating proteins for G␣ q/11 (40), and RGS4 functions addition- ally for G␣ i/o (41) but not for G␣ 12/13 (4). In contrast, RGS12 is a selective inhibitor of G␣ 12/13 signaling (4,40). To test the involvement of G␣ q/11 and G␣ 12/13 in M1-mediated SRF activation, we co-transfected cells with cDNAs encoding M1, the reporter gene, and RGS2, RGS4, or RGS12. Shown in Fig. 3b, RGS2 and RGS4 proteins blocked M1-mediated SRF activation. This result suggests that G␣ q/11 does play a role in M1mediated SRF activation. In contrast, RGS12 had little effect on M1-stimulated SRF activation (Fig. 3b). This result is in contrast to a previous report that RGS12 inhibited M1-mediated SRF activation in NIH 3T3 cells (4), and it suggests that M1-mediated SRF activation does not depend on G␣ 12/13 activation in Jurkat cells.
Involvement of Rho in SRF Activation by M1 mAChR-Small GTPases such as RhoA and Cdc42 are involved in SRF activation. In the present study, two different approaches were utilized to test the involvement RhoA in SRF activation. First, C3 toxin (C. botulinum C3 transferase), which specifically inhibits RhoA by ADP-ribosylation (42), was expressed along with M1 mAChR. Second, the dominant negative RhoA construct T19N-RhoA, which acts to block the upstream activation of endogenous Rho, was used. In this transfection system, both C3 and dominant negative Rho (Fig. 3c) blocked SRF activation.
Effects of Different Inhibitors on M1-mediated SRF Activation-Multiple signaling proteins can regulate the SRF pathway. To test the involvement of kinases and phosphatases in SRF activation, a panel of inhibitors was used in M1-transfected Jurkat cells. Inhibitors of mitogen-activated protein kinase (PD98059), protein kinase C (K252a), calmodulin kinase (KN62), and protein kinase A (H89) had no effect on SRF activation by M1 at concentrations selected for activity against the respective target enzymes (Fig. 4a). In contrast, the calmodulin (CaM) inhibitor W7 and the calcineurin inhibitor cyclosporin A dose-dependently inhibited SRF-dependent gene transcription mediated by M1 (Fig. 4, a and b). Higher concentrations of W7 additionally lowered basal SRF activity, observed in the absence of carbachol. The calcineurin inhibitor failed to fully inhibit M1-mediated SRF activation in numerous experiments, suggesting the presence of a minor, calcineurinindependent pathway. M1-mediated SRF activation was also inhibited by the intracellular calcium chelator BAPTA/AM (Fig. 4a), consistent with a Ca 2ϩ -dependent pathway. Moreover, M1-mediated SRF activation was inhibited by the general tyrosine kinase inhibitor, genistein, but not by the epidermal growth factor receptor inhibitor AG1478 (Fig. 4a).
To further test the involvement of CaM and calcineurin in SRF activation mediated by M1, we used another potent CaM inhibitor, ophiobolin A, and a plasmid encoding endogenous calcineurin inhibitor (CainC; Cain 2078 -2173), which contains a calcineurin binding domain (28). Shown in Fig. 4, b and c, ophiobolin A and CainC inhibited carbachol-induced SRF activation in M1-transfected cells. These results further support the involvement of Ca 2ϩ /CaM/calcineurin in M1-mediated SRF activation.
Effect of Inhibitors on SRF Activation by Constitutively Active GTPase-deficient G␣ 13  mediated SRF activation was also insensitive to inhibitors of mitogen-activated protein kinase, protein kinase A, protein kinase C, and calmodulin kinases (data not shown) but was inhibited by the CaM inhibitor W7 and by BAPTA/AM (Fig. 5a). However, in contrast to carbachol/M1-induced SRF activation, G␣ 13 QL-mediated SRF activation was not inhibited by cyclosporin A and genistein (Fig. 5a). This result suggests the presence of a distinct pathway involving Ca 2ϩ /CaM/calcineurin and a tyrosine kinase in M1-mediated SRF activation, independent of G␣ 13 QL. The inhibition of G␣ 13 QL-mediated SRF activation by dominant-negative RhoA (T19N-RhoA) and C3 toxin (Fig. 5a) indicates the involvement of RhoA in SRF activation. Similar to G␣ 13 QL-mediated SRF activation, SRF activation mediated by constitutively active RhoA, G14V-RhoA, was also inhibited by W7 and BAPTA/AM but not by cyclosporin A and genistein (Fig. 5b). This result indicates that Ca 2ϩ /CaM acts downstream of RhoA, but it leaves the possibility open that Ca 2ϩ /CaM also acts upstream of RhoA.
We also tested the effects of inhibitors on G␣ q QL-induced SRF activation. Similar to G␣ 13 QL, G␣ q QL-induced SRF activation was inhibited by W7 but not by cyclosporin A and genistein (data not shown).
Role of Pyk2 in SRF Activation-Pyk2 is a Ca 2ϩ /CaMdependent tyrosine kinase that can be activated by stimulation of several G protein-coupled receptors (43,44) including M1 mAChR (45). Moreover, Pyk2 is involved in G␣ 13 -mediated serum response element-dependent transcription (9). Because M1-induced SRF activation is genistein-sensitive (Fig. 4a), indicating the involvement of a tyrosine kinase, we tested whether Pyk2 is involved in M1-mediated SRF activation. Shown in Fig. 5d, Pyk2 was phosphorylated upon M1 but not M3 stimulation, and genistein abolished both basal and M1activated phosphorylation of Pyk2. Moreover, transfection of Pyk2 activated SRF (Fig. 5c). This effect was inhibited by W7, BAPTA/AM, T19N-RhoA, and C3 toxin, indicating the involvement of Ca 2ϩ /CaM-and RhoA-dependent pathways. However, Pyk2-mediated SRF activation was not inhibited by genistein, suggesting that yet another tyrosine kinase sensitive to genistein could be involved in M1-mediated SRF activation and specifically in Pyk2 phosphorylation.
To test whether Pyk2 contributes to M1-mediated SRF activation, we used two distinct Pyk2 dominant-negative mutants, K457A-Pyk2 and Y402F-Pyk2 (31). These constructs inhibited the Pyk2-mediated activation of SRF by ϳ80% when cotransfected with the Pyk2 wild type (not shown). Shown in Fig. 5e, both Pyk2 mutants dose dependently inhibited M1-mediated SRF activation, indicating that this pathway involves Pyk2 phosphorylation and kinase activity in Jurkat cells. The only partial inhibition seen in these experiments could have resulted from gene dosage effects or from the presence of alternative parallel pathways.
Sensitivity of M1-mediated SRF activation to CaM/calcineurin inhibitors suggested the presence of a distinct pathway from that activated by G␣ 13 QL. When M1 mAChR was co-transfected with G␣ 13 QL, SRF-mediated luciferase expression was considerably greater than for either M1 or G␣ 13 QL alone (Fig. 6), suggesting the presence of two pathways possibly exhibiting synergism. Co-transfection of either Pyk2 or constitutively active RhoA with M1 displayed much less additive effects (considering the elevated basal levels in the absence of carbachol) (Fig. 6). This suggests that Pyk2 and RhoA participate in the same pathway. Co-transfection of G␣ 13 QL, constitutive active RhoA, and Pyk2 did not enable M3 to further activate SRF (Fig. 6).
We also performed experiments with cotransfection of M1 and G␣ 12 QL. Because the response to G␣ 12 QL was considerably less than that with G␣ 13 QL both with and without M1 stimulation (data not shown), G␣ 12 QL was not considered to play a major role in SRF signaling in Jurkat T cells.
RhoA Activity Assay-Stimulation of M1 receptors led to a rapid RhoA activation (Fig. 7a), which was detectable as early as 1 min. Surprisingly, stimulation of M3 receptors also caused RhoA activation (Fig. 7a), although M3 stimulation did not activate SRF. This indicates that RhoA activation alone was insufficient or that there are critical differences in the time course of RhoA activation between M1 and M3. M1-mediated RhoA activation was inhibited by W7, cyclosporin A, BAPTA/ AM, dominant-negative RhoA, and partially by genistein (Fig. 7b). This is consistent with the pattern seen for M1mediated SRF activation. Transfection of G␣ 13 QL and Pyk2 also increased RhoA activity (Fig. 7, c and d). Similar to G␣ 13 QL and Pyk2-mediated SRF activation, RhoA activation by G␣ 13 QL and Pyk2 was also sensitive to W7 and BAPTA/AM but not to cyclosporin A and genistein. Taken together, these results indicate that cabarchol/M1 activated SRF through RhoA by a unique pathway involving CaM, calcineurin, Pyk2, and a genistein-sensitive tyrosine kinase, which is separate from the RhoA-SRF pathway activated by G␣ 13 QL.

DISCUSSION
The ability of mAChR to activate SRF-mediated gene transcription and the involvement of G protein subunits in SRF activation were investigated in Jurkat T cells. The finding that G␣ q -coupled M1 mAChR, but not G␣ i/o -coupled M2 mAChR, activated SRF through a RhoA-mediated pathway is consistent with previous reports (4,46,47). In contrast, failure of M3 mAChR to activate SRF was unexpected because M1 and M3 are thought to have similar signaling pathways primarily involving G␣ q/11 . Yet G␣ q coupling of both M1 and M3 was intact as seen by a robust calcium response in Jurkat T cells. Moreover, use of the chimeric G␣ protein construct G␣ iq5 , which conferred the ability of the M2 receptor to signal along G␣ q/11mediated pathways, restored the Ca 2ϩ response for M2 but not SRF activation. These results suggested that M1 activation of SRF involves a G␣ q/11 -independent pathway or that G␣ q/11 is insufficient in Jurkat T cells. Inhibition of M1-SRF signaling by co-transfection of RGS2 and RGS4 (which suppress G␣ q/11 activity) indicated that G␣ q/11 did play a role in M1-SRF activation, but it appeared to be insufficient per se. Moreover, the suppression of M1-SRF signaling by the calcium chelator BAPTA suggested a role for intracellular calcium, which is released by G␣ q/11 activation.
The discrepancy between the present results and those of Mao et al. (4) showing that SRF signaling depends upon G␣ q can be accounted for by the differences among cell lines used. Rho pathways involve numerous protein factors, and different cell lines express different complements of G proteins and other regulatory proteins. Signaling by a single subfamily of G proteins can be regulated distinctly in different cell systems (48 -50). Use of the Jurkat T cell line thus revealed a pathway of SRF activation that distinguishes the closely related M1 and M3 receptors.
Fromm et al. (46) propose a G␣ 12/13 -dependent pathway in a similar cell system. Constitutively active G␣ subunits were used to determine the involvement of G␣ subunits in Jurkat cells. Only constitutively active G␣ 13 QL activated SRF strongly, whereas G␣ 12 QL and G␣ q QL were poorly effective, and G␣ 14 QL and G␣ 15 QL were inactive. A differential role of G␣ proteins of the G␣ 12/13 family has been demonstrated in stress fiber and focal adhesion formation (51). Our results show that G␣ 13 QL and M1 receptors additively or synergistically activated SRF in Jurkat cells, which suggests that alternative pathways exist in this signaling event (52). Co-transfection with RGS12 (to suppress G␣ 12/13 QL signaling) failed to affect M1-SRF signaling, supporting the notion that G␣ 12/13 proteins are not involved.
Protein kinase C and mitogen-activated protein kinase can activate SRE-mediated transcription through ternary complex factor (TCF) pathways (53). To delineate alternative signaling pathways leading to transcriptional activation, we tested the effect of a panel of kinase and phosphatase inhibitors on M1 mAChR-mediated SRF activation. Protein kinase C and pro-tein kinase A inhibitors (K252a and H89), the mitogen-activated protein kinase inhibitor PD98059, and the calmodulin kinase II/IV inhibitor KN62 failed to block carbachol-M1 induced activation of SRF. In addition, carbachol did not stimulate mitogen-activated protein kinase activity measured by an extracellular signal-regulated kinase reporter gene assay in M1-transfected Jurkat cells (data not shown).
Although SRF activation was insensitive to the calmodulin kinase inhibitor KN62, calmodulin inhibitors reduced SRF activation by M1 stimulation. This suggests a role for CaM in mediating SRF activation by M1 mAChR. We have recently found that activation of M1 causes significant transfer of CaM from plasma membranes to the cytosol in M1-transfected HEK293 cells. 2 Moreover, CaM directly interacts with the opioid receptor at a sequence motif in the i3 loop (54) and appears to serve as a signaling factor for opioid receptors (55). Because muscarinic receptors have a similar sequence motif in their i3 loop, it is possible that CaM interacts directly with M1, thereby serving as a second messenger per se. The use of constitutively active RhoA revealed that CaM is also required for the RhoA-SRF downstream pathway. Thus, CaM plays a pervasive role in M1-SRF signaling. A role for direct CaM-M1 interactions in these pathways is under investigation. SRF activation by M1, but not by G␣ 13 QL, was also sensitive to cyclosporin A and CainC, inhibitors of the calcium/CaMregulated phosphatase 2B, calcineurin. Calcineurin-dependent SRF activation had not been reported previously, whereas reporter gene expression driven by multiple AP-1 sites had been shown to be selectively sensitive to cyclosporin A and FK506 in Jurkat T cells (56,57). Whether this new M1-RhoA-SRF pathway is specific to Jurkat T cells or more broadly distributed remains to be established. Calcineurin could either interact directly with RhoA or, more likely, activate any of the numerous RhoA GDP/GTP exchange factors, inactivate inhibitors of GDP/GTP exchange, or last, inactivate RhoA GTPase activating proteins. Direct assay of RhoA activation by measuring GTP loading onto RhoA in a pull-down assay revealed that calcineurin appears to work upstream of RhoA. In agreement, SRF activation by constitutively active RhoAG14V was insensitive to cyclosporin, indicating that this calcineurin inhibitor acted only upstream of RhoA. The parallel G␣ 13 QL-RhoA pathway was also suppressed by a CaM inhibitor but not the calcineurin inhibitor, indicating distinct pathways.
Further separation between SRF activation pathways mediated by M1 and G␣ 13 QL emerged with the use of the tyrosine kinase inhibitor genistein. Only the M1 pathway to RhoA was sensitive to genistein, suggesting the involvement of a tyrosine kinase upstream of RhoA. Stimulation of RhoA and SRF by G␣ 13 QL-RhoA was unaffected by genistein. Because the tyrosine kinase Pyk2 is activated by GPCRs and requires CaM, we tested its role in M1 and G␣ 13 QL signaling. Transfection of Pyk2 also enhanced SRF activation in a Ca 2ϩ /CaM-dependent fashion, but this was insensitive to the kinase inhibitor genistein. Yet the Pyk2 dominant-negative mutants K457A-Pyk2 and Y402F-Pyk2 partially inhibited carbachol-induced SRF activation, suggesting that Pyk2 activation does contribute to M1-RhoA-SRF signaling. Moreover, carbachol-stimulation of M1, but not M3, activated Pyk2 phosphorylation, measured with phosphotyrosine antiserum, which parallels the selective activation of SRF by M1 but not M3. These results indicate that Pyk2 alone cannot account for the observed genistein-mediated inhibition of M1-RhoA-SRF signaling but, rather, that another genistein-sensitive kinase is (additionally) involved. Specifically, Pyk2 phosphorylation by M1 activation was inhibited by genistein, indicating that a Src-like tyrosine kinase is involved upstream of Pyk2. Recent studies show that Pyk2 forms physical complexes with Src-like tyrosine kinases (45). It is possible that Pyk2 activation involves such complex formation, triggered by M1 activation.
In conclusion, RhoA and SRF activation by M1 involves a unique pathway requiring calcium, CaM, calcineurin, and the tyrosine kinase Pyk2 (Fig. 8). These studies demonstrate that multiple independent pathways are involved in the signaling of mAChRs but more importantly reveal an M1 pathway not shared by other muscarinic receptors tested, even the very closely related M3 subtype. The physiological significance of this new CaM-dependent pathway from M1 to RhoA and SRF activation remains to be established, particularly in the central nervous system where the M1 subtype is thought to play a key role in memory functions and the pathophysiology of Alzheimer's disease.