cGMP-dependent Protein Kinase Inhibits Serum-response Element-dependent Transcription by Inhibiting Rho Activation and Functions*

RhoA, in its active GTP-bound form, stimulates transcription through activation of the serum-response factor (SRF). We found that cGMP inhibited serum-induced Rho·GTP loading and transcriptional activation of SRF-dependent reporter genes in smooth muscle and glial cells in a cGMP-dependent protein kinase (G-kinase)-dependent fashion. Serum stimulation of the SRF target gene vinculin was also blocked by cGMP/G-kinase. G-kinase activation inhibited SRF-dependent transcription induced by upstream RhoA activators including Gα13 and p115RhoGEF, with Gα13-induced Rho·GTP loading inhibited by G-kinase. G-kinase had no effect on the high activation levels of RhoA(63L) or the double mutant RhoA(63L,188A) but inhibited transcriptional activation by these two RhoA mutants to a similar extent, suggesting an effect downstream of RhoA and independent of RhoA Ser188phosphorylation. Constitutively active forms of the Rho effectors Rho kinase (ROK), PKN, and PRK-2 induced SRF-dependent transcription in a cell type-specific fashion with ROK being the most efficient; G-kinase inhibited transcription induced by all three effectors without affecting ROK catalytic activity. G-kinase had no effect on RhoA(63L)-induced morphological changes in glial cells, suggesting distinct transcriptional and cytoskeletal effectors of RhoA. We conclude that G-kinase inhibits SRF-dependent transcription by interfering with RhoA signaling; G-kinase acts both upstream of RhoA, inhibiting serum- or Gα13-induced Rho activation, and downstream of RhoA, inhibiting steps distal to the Rho targets ROK, PKN, and PRK-2.

Rho proteins are small GTPases of the Ras family that cycle between an active, GTP-bound form and an inactive, GDPbound form and regulate diverse cellular functions such as cytoskeletal organization, smooth muscle contraction, muscle and neuronal differentiation, cell cycle progression, and gene expression (1)(2)(3)(4)(5). RhoA is activated downstream of multiple membrane receptors, for example the thrombin and lysophos-phatidic acid receptors, which couple to the heterotrimeric Gproteins G␣ 12 and G␣ 13 and activate RhoA-specific guanine nucleotide exchange factors, e.g. p115RhoGEF (1,6). RhoAinduced cytoskeletal changes include formation of F-actin stress fibers and focal adhesion complexes, and RhoA-mediated smooth muscle contraction involves calcium sensitization of the contractile apparatus (1,2,6). Downstream effectors of RhoA include the Rho kinase (ROK) 1 family, the protein kinase Crelated kinases PRK-1 (also called PKN) and PRK-2, mDia, and rhotekin (2).
As a regulator of gene expression, RhoA modulates the activity of the ubiquitous serum-response factor (SRF), which binds to the sequence CC(A/T) 6 GG in the core of the serumresponse element (SRE) (3,7). Induction of immediate early genes, e.g. c-fos, by serum and growth factors is mediated by SRF in cooperation with a ternary complex factor (TCF) that is controlled by the Ras-Raf-MEK-Erk pathway (8). SRF target genes also include vinculin and many muscle-specific genes, but promoters of these genes lack TCF-binding sites, and SRF appears to cooperate with other cell type-specific transcription factors (4,5,8,9). Activated RhoA increases transcription of smooth muscle-specific promoters and SRF-dependent reporter genes, whereas RhoA inactivation by C3 exoenzyme abolishes their transcription. RhoA activation has little effect on the c-fos promoter, but RhoA inactivation inhibits c-fos induction by various growth factors, indicating that at least basal RhoA activity is required for c-fos induction (8,9). Although the mechanism(s) by which RhoA regulates SRF activity are incompletely defined, recent studies (8 -10) suggest that depletion of the cellular G-actin pool by RhoA-induced actin polymerization is necessary and sufficient for stimulation of SRF-dependent transcription. Others (11) have suggested that SRF activation by RhoA requires the cooperation of SRF with NF-6B and C/EBP␤. RhoA signaling to SRF cannot be attributed to a single known downstream effector, and RhoA effector mutants suggest distinct but partially overlapping effector pathways for cytoskeletal reorganization and SRF activation (12,13). RhoA prenylation may not be required for induction of SRF-dependent transcription but is required for the cytoskeletal effects of RhoA, also suggesting different pathways (14).
The nitric oxide (NO)/cGMP/cGMP-dependent protein kinase I (G-kinase I) signal transduction pathway plays an important role in vascular biology, regulating smooth muscle tone as well as SMC proliferation and differentiation (15,16); G-kinase I also regulates endothelial cell permeability and motility and platelet aggregation (16,17). Homozygous deletion of the Gkinase I gene in mice abolishes NO/cGMP-dependent relaxation of smooth muscle, resulting in severe vascular and intestinal dysfunction with death at an early age (18). G-kinase I affects smooth muscle tone by decreasing the release of calcium from intracellular stores and by reducing calcium sensitivity of the contractile apparatus (16); the latter effect is due to a direct inhibition of RhoA-induced calcium sensitization of SMC contraction by G-kinase I (19). Additional reports have confirmed inhibition of RhoA functions by the NO/cGMP/G-kinase signaling pathway. Insulin-induced relaxation of vascular SMCs is mediated via NO/cGMP/G-kinase inhibition of RhoA (20), and SMC contraction induced by constitutively active mutants of G␣ 12 and G␣ 13 is RhoA-dependent and inhibited by cGMP analogs (21). Incubation of SMCs or G-kinase-transfected HeLa cells with NO donors or cGMP analogs prevents membrane translocation of activated RhoA and induces disassembly of the actin cytoskeleton (19,22). The effects of NO and cGMP on the actin cytoskeleton are absent in G-kinase-deficient cells and restored by transfection of cells with G-kinase I, indicating that they are mediated by G-kinase (19,22). G-kinase I phosphorylates RhoA in vitro on Ser 188 , and the cytoskeletal effects of G-kinase activation are inhibited by expression of a mutant RhoA(A188), suggesting that inhibition of RhoA cytoskeletal functions by G-kinase I is at least partially due to phosphorylation of serine 188 (19,22). G-kinase regulates gene expression, both at the transcriptional and post-transcriptional level, and has been shown to increase the expression of some genes, for example c-fos, junB, and mitogen-activated protein kinase phosphatase 1, as well as decrease expression of others, for example, thrombospondin, gonadotropin-releasing hormone, and soluble guanylate cyclase (15,23,24). The precise mechanism of G-kinase action is not known; however, in fibroblasts and neuronal/glial cells, transactivation of the c-fos promoter by G-kinase I requires an intact cAMP response element (CRE), nuclear translocation of the kinase, and phosphorylation of the CRE-binding protein (25)(26)(27). To our knowledge, the effect of G-kinase on RhoAmediated gene expression has not yet been examined.
We now show that G-kinase inhibits SRE/SRF-dependent transcription induced by serum or by constitutively active forms of G␣ 12 /G␣ 13 and p115RhoGEF in smooth muscle cells and glial cells. G-kinase inhibited the activation of endogenous, cellular RhoA by serum, had no effect on the high activation level of mutant RhoA(63L), but inhibited transcriptional activation by RhoA(63L) and RhoA(63L, 188A). These data suggest that G-kinase acts both upstream and downstream of RhoA. Of the downstream effectors studied, constitutively active ROK, PKN, and PRK-2 significantly activated the SRE-dependent reporter in a cell type-specific manner, and the transcriptional effects of all three kinases were inhibited by G-kinase activation.

EXPERIMENTAL PROCEDURES
Materials and DNA Constructs-The expression vector pGEX-2T-TRBD, which encodes glutathione S-transferase (GST) coupled to the Rho binding domain (RBD) of rhotekin, was from M. A. Schwartz, with GST-RBD affinity-purified from bacteria as described (28). Materials for measurement of GTP/GDP bound to Rho were from sources described previously (29,30). A RhoA-specific antibody (SC-179), the monoclonal anti-Myc antibody (clone 9E10, SC-40), and the anti-Erk1/2 antibody (SC-94) were from Santa Cruz Biotechnology; the phosphospecific anti-active Erk antibody (V6671) was from Promega. The anti-C-terminal G-kinase I antibody (KAP-PK005) was from StressGen; the anti-FLAG epitope antibody (F7425) and purified myosin light chain were from Sigma, and the hybridoma line producing the anti-EE epitope antibody was from G. Walter (31). The ROK inhibitor Y27632 was from Calbiochem.
Human G-kinase I␤ cDNA was provided by S. Lohmann and was expressed from the cytomegalovirus early promoter as described previously (26). Expression vectors encoding constitutively active G␣ 12 and G␣ 13 were from M. Simon (32), and the vector for C3 exoenzyme was from A. Hall (7). The vector for p115RhoGEF and the cDNAs encoding wild type RhoA and mutant RhoA(63L) were from J. H. Brown (33,34); the RhoA coding sequence was excised and placed in-frame with the N-terminal epitope tag EEEEYMPME (EE tag) in the expression vector pcDNA3. To produce double mutant RhoA(63L/188A), the coding sequence of RhoA(63L) was amplified with primers encoding the C-terminal Ser 188 3 Ala substitution and subcloned downstream of the EE epitope into pcDNA3, after sequence verification of the amplified product. Myc epitope-tagged expression vectors encoding constitutively active or kinase-dead mutant forms of the catalytic domain of ROK were from K. Kaibuchi (see Ref. 35); FLAG epitope-tagged constitutively active PKN was from Y. Ono (see Ref. 36), and FLAG epitope-tagged constitutively active PRK-2 was from J. Settleman (see Ref. 37).
The reporter constructs pSRE-Luc and pSRF-Luc were from Stratagene; they contain five copies of the full c-fos serum-response element (5Ј-TAGGATGTCCATATTAGGACATC-3Ј) or the isolated SRF-binding site (5Ј-GTCCATATTAGGAC-3Ј) upstream of the minimal thymidine kinase promoter and luciferase coding sequence, respectively.
The human vinculin promoter (38) was amplified using the primers 5Ј-GAGCTCGAGTACAAAGCGTGAGGGAACTG-3Ј and 5Ј-GGCAAGC-TTCGTATGAAACACTGGCTTCG-3Ј with genomic DNA isolated from 1321-N1 astrocytoma cells. The amplification product encoded nucleotide Ϫ1113 to ϩ17 relative to the translation start site with AAG substituted for ATG at ϩ2 (GenBank TM accession number L04933); it was verified by DNA sequencing and subcloned into pGL2-basic (Promega) upstream of the luciferase coding sequence. The vinculin cDNA probe was amplified using the primers 5Ј-CAAGTGTGACCGAGTG-GACC-3Ј and 5Ј-TTGGTATCAATGGCTTCGTC-3Ј and sequenced (38).
Cell Culture and Transfections-Bovine aortic smooth muscle cells (BoASMCs) were from Cell Applications, Inc. (San Diego, CA), and were used between passages 4 and 7; they were routinely cultured in DME medium supplemented with 10% fetal bovine serum (FBS), penicillin, streptomycin, and fungizone at 37°C in a 95% air, 5% CO 2 atmosphere. Subconfluent cells in 12-well cluster dishes were transfected with 0.6 g of DNA and 4 l of Polyfect TM (Promega) in DME supplemented with 2% FBS; 24 h later they were transferred to DME without serum for 16 h, treated with 250 M 8-chlorophenylthio-cGMP (CPT-cGMP) for 8 h, and stimulated with 2% serum during the last 4 h. Primary neonatal rat cardiomyocytes were provided by W. Dillmann (39); they were plated on gelatin-coated 12-well dishes and treated as described for BoASMCs, except that 10% FBS was used for stimulation. Rat CS54 pulmonary arterial smooth muscle cells (provided by A. Rothman (40)) and C6 glioma cells (provided by M. Ellisman) were transfected as described previously (27,41) using FuGENE TM (Roche Molecular Biochemicals) and LipofectAMINE TM Plus (Invitrogen), respectively. After transfection, cells were recovered for 2 h in DME containing 10% FBS before they were transferred to DME without serum for 24 h; 250 M CPT-cGMP was added for the last 8 h, and cells were stimulated with 10% serum for the last 4 h as indicated. To measure the effect of G-kinase I on the activity of endogenous RhoA, C6 cells were transfected with the G-kinase I␤ expression vector and selected for growth in G418; a clone was selected that stably expressed G-kinase I at an activity level comparable with that found in vascular SMCs.
Reporter Gene Assays-In BoASMCs, firefly luciferase activity was normalized to Renilla luciferase expressed from the co-transfected control vector pTK-RL (Promega); firefly and Renilla luciferase activities were measured using the Dual-luciferase Reporter Assay System TM from Promega. In CS54 and C6 cells, firefly luciferase activity was normalized to ␤-galactosidase activity expressed from pRSV-␤Gal; the reporter gene activities were measured as described before (27). Histochemical staining for ␤-galactosidase activity was performed with 5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside as described (26).
G-kinase Assay-G-kinase activity was measured using the synthetic peptide Kemptide and the specific cAMP-dependent protein kinase (protein kinase A) inhibitor PKI as described previously (26).
Western Immunoblots-Western blots were generated and developed using horseradish peroxidase-coupled secondary antibodies and enhanced chemiluminescence as described previously (26). Assays for RhoA Activity-Cells were incubated in DME medium without serum for 48 h, treated with 250 M CPT-cGMP for 2 h, and serum-stimulated for 3 min. The rhotekin pull-down assay for Rho⅐GTP was performed essentially as described (28); cells were harvested by lysis in situ, and 30 g of GST-RBD (or the same amount of GST) coupled to glutathione-Sepharose beads were added to the samples. After a 1-h incubation at 4°C with gentle shaking, beads were washed, and bound Rho was eluted in SDS sample buffer; samples were analyzed by SDS-PAGE and electroblotting with the amount of RhoA estimated by Western blotting using an RhoA-specific antibody.
To measure absolute amounts of GTP and GTP ϩ GDP bound to Rho, we modified a procedure we developed previously to measure GTP and GTP ϩ GDP bound to Ras and Rap1 (29,42); this method will be described in detail elsewhere. 2 Cells were lysed in a buffer containing 50 mM Tris-HCl, pH 7.4, 1% Nonidet P-40, 1% CHAPS, 200 mM NaCl, 1 mM MgCl 2 , 10 g/ml leupeptin, 10 g/ml aprotinin, and 0.2 mM phenylmethylsulfonyl fluoride. After centrifugation at 10,000 ϫ g for 2 min, one portion of the supernatant was added to tubes containing 10 mM MgSO 4 (to measure GTP bound to Rho and inhibit exchange activity), and another portion was added to tubes containing 2 M GTP and 10 mM EDTA (to stimulate nucleotide exchange and convert all of the Rho to the GTP-bound state, and thus measure total nucleotides bound to Rho). To both portions was added 30 g of GST-RBD bound to glutathione beads, and samples were incubated for 1 h at 4°C with gentle shaking. After washing the beads four times in 50 mM Tris-HCl, pH 7.4, 2% Nonidet P-40, 500 mM NaCl, 10 mM MgSO 4 , and two times in 20 mM TrisPO 4 , pH 7.4, 5 mM MgSO 4 , Rho-bound GTP was released by heating the beads for 3 min at 100°C in 5 mM TrisPO 4 , pH 7.4, 2 mM EDTA, 2 mM dithiothreitol. GTP eluted from both samples was measured in a coupled enzymatic assay by conversion to ATP in the presence of ADP and nucleoside diphosphate kinase (29). The resulting ATP was measured by the firefly luciferase method in a photon-counting luminometer (MGM instruments). In control experiments to be described elsewhere, 2 we showed that the nucleotide exchange on Rho is complete and that the GTP is quantitatively eluted from Rho bound to RBD under the conditions used. The assay is linear with respect to cell lysate input, and the percent Rho activation, calculated as [GTP]/[total nucleotides] bound to Rho, is constant over a wide range of lysate input.
To measure the amount of GTP and GDP bound to EE epitope-tagged RhoA, cells were extracted in situ as described above, and after centrifugation, the supernatants were split in half and added to tubes containing protein G-agarose beads coated with either an anti-EE antibody or control mouse IgG. After 1 h of gentle shaking at 4°C, the beads were washed, and GTP and GDP were released from the immunoprecipitated Rho as described above. In one aliquot of the sample, GTP was measured as described above, and in another aliquot the sum of GDP plus GTP was measured by first converting GDP to GTP using pyruvate kinase and phosphoenolpyruvate as described previously (30). Control experiments demonstrated that the amount of GTP and the sum of GTP plus GDP eluted from the immunoprecipitated RhoA was the same in the absence or presence of soluble GST-RBD and yielded an activation state of RhoA similar to that measured for endogenous Rho under various conditions. 2 Phalloidin Staining-Cells were plated on glass coverslips coated with fibronectin (10 g/ml), serum-starved for 36 h, and treated with 250 M CPT-cGMP for 2 h prior to stimulation with 10% FBS for 30 min. Cells were fixed for 10 min in freshly prepared 1% para-formaldehyde (in phosphate-buffered saline (PBS)), permeabilized for 10 min in PBS containing 0.1% Triton X-100, and blocked for 1 h in PBS containing 3% bovine serum albumin prior to staining with Texas Red-labeled phalloidin (1:50, Molecular Probes). Cells were visualized using a Zeiss fluorescence microscope.
Northern Blots-Cytoplasmic RNA was isolated and resolved by denaturing agarose gel electrophoresis, and Northern blots were hybridized with a radioactively labeled vinculin cDNA and a 28 S rRNA probe as described previously (43).
ROK Assay-C6 cells were transfected in 6-well dishes with expression vectors encoding G-kinase I␤ and either the constitutively active or inactive (kinase-dead) catalytic domain of ROK (1 g of each DNA). After 36 h of culture in serum-free DME, cells were incubated for 2 h in the presence or absence of 250 M CPT-cGMP and harvested in 0.3 ml of lysis buffer (50 mM Tris-HCl, pH 7.5, 1% Triton X-100, 1 mM EGTA, 1 mM dithiothreitol, 10 mM MgCl 2 , 50 mM ␤-glycerol phosphate, 1 mM Na 3 VO 4 , 100 M microcystin-LR, and a protease inhibitor mixture). Cleared lysates were subjected to immunoprecipitation with 5 g of 9E10 anti-Myc antibody and protein G-agarose. Washed beads were incubated in 50 l of kinase buffer (50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 10 mM MgCl 2 , 50 mM NaCl, 1 mM dithiothreitol), containing 10 M [␥- 32 PO 4 ]ATP (60 Ci/mmol) and 80 g/ml myosin light chain for 30 min at 30°C. Reaction products were analyzed by SDS-PAGE/electroblotting/autoradiography, and the amount of ROK in the immunoprecipitates was determined by Western blotting with anti-Myc antibody.
Statistical Analysis-Unless stated otherwise, the data presented in bar graphs represent the mean Ϯ S.D. of at least three independent experiments performed in duplicate. Other data showing blots or photographs of a typical experiment represent data that were reproduced at least three times in independent experiments. Statistical analyses were performed using the Student's t test. A two-tailed value of p Ͻ 0.05 was considered to indicate statistical significance.

Effects of cGMP on Serum-induced RhoA Activation, Cytoskeletal Changes, and SRE-dependent Transcription in
BoASMCs-Several studies have suggested inhibition of RhoA function by the NO/cGMP/G-kinase signal transduction pathway (19 -22, 44). This inhibition could be through the following: (i) an "upstream" mechanism, i.e. a step preceding RhoA activation; (ii) direct RhoA phosphorylation by G-kinase; (iii) a "downstream" mechanism, i.e. a step subsequent to RhoA activation; or (iv) a combination of these mechanisms. Changes in the distribution of RhoA between membrane and cytosol have been used as an indication for Rho activation with increased membrane association occurring when RhoA is activated, e.g. by the addition of GTP␥S to cell lysates (6). Activation of G-kinase prevents agonist-induced RhoA membrane translocation in several cell systems, suggesting inhibition of RhoA activation by an upstream mechanism (19,20,44). However, the only study directly measuring the effect of G-kinase activation on RhoACGTP levels found no significant effect of cGMP on thromboxane A 2 -induced, G␣ 12/13 -mediated Rho activation in platelets (45).
We therefore decided to examine the effect of cGMP on serum-induced RhoA activation in early passage BoASMCs. The specific activity of G-kinase in near-confluent cultures of these BoASMCs was 0.36 Ϯ 0.20 nmol/min/mg protein. We used two methods to assess Rho activation. In the first assay, developed by Ren et al. (28), Rho⅐GTP is isolated from cell lysates using the rhotekin Rho binding domain (RBD), and the amount of RhoA bound to the RBD is determined by semi-quantitative Western blotting (28). Serum stimulation of BoASMCs increased the amount of Rho⅐GTP, and pretreating cells with the membrane-permeable cGMP analog CPT-cGMP almost completely prevented this increase (Fig. 1A, upper panel). Control blots of whole cell lysates (10% of the assay input) demonstrated similar amounts of total RhoA and G-kinase I present under all conditions (Fig. 1A, lower two panels).
The second method allows quantitation of Rho activation by measuring the amount of GTP on RBD-bound Rho in an enzymatic assay. 2 As before, BoASMCs were stimulated with serum in the absence and presence of CPT-cGMP, and one part of the sample was subjected to the RBD pull-down to isolate Rho⅐GTP directly, whereas in another part of the sample, Rho was first completely loaded with GTP prior to the RBD pull-down to yield the total amount of Rho-bound nucleotides. Determination of GTP bound to Rho in both samples allowed us to calculate the percent Rho activation as [GTP]/[total nucleotides] bound to Rho. By using this method, we found that 4 -5% of Rho was in the GTP-bound state in serum-starved BoASMCs, and serum stimulation increased the amount of Rho⅐GTP about 2.5-fold (Fig. 1B, open bars). Treating cells with CPT-cGMP marginally decreased basal Rho activity but inhibited serum-stimulated Rho activation by more than 50% (Fig. 1B, filled bars; in this and all subsequent figures, open and filled bars represent cells cultured in the absence and presence of cGMP, respectively).
Consistent with these findings and with the findings of others in rat aortic SMCs (19), we observed that CPT-cGMP treatment inhibited serum-induced stress fiber formation in BoASMCs, a process that is well known to depend on RhoA activation (Fig. 1C). Of note, CPT-cGMP treatment also appeared to decrease the amount of stress fibers in serum-starved cells, although we did not find a statistically significant decrease in basal RhoCGTP under these conditions, suggesting that cGMP/G-kinase may inhibit stress fibers by additional mechanisms.
Next, we examined the effect of CPT-cGMP on serum-induced transcription of an SRE-dependent reporter gene. Serum stimulation of BoASMCs increased SRE-dependent transcription about 7-fold ( Fig. 1D, open bars). Treating cells with CPT-cGMP had little effect on basal transcription of the SRE-dependent reporter but inhibited serum-stimulated reporter gene activity by Ͼ50% (Fig. 1D, filled bars). At levels that did not impair cell viability, C3 exozyme, which ADP-ribosylates and inactivates RhoA (6), completely inhibited serum-induced transcription ( Fig. 1D). Thus, as described for other cell types (8,46), serum induction of the pSRE-Luc appeared to be dependent on RhoA function in BoASMCs.
Changes in Erk1/2 activity could influence SRE-dependent transcription through regulation of TCF, which cooperates with SRF (7,8), and the cGMP/G-kinase signaling pathway has been reported either to inhibit, stimulate, or not affect the Raf/MEK/Erk1/2 pathway in different cell types (43,(47)(48)(49)(50)(51)(52). We therefore examined the effect of CPT-cGMP on seruminduced Erk1/2 activation in BoASMCs, but we found no significant effect of cGMP on Erk1/2 phosphorylation in the absence or presence of serum stimulation ( Fig. 1E shows 5-and 10-min time points at which serum stimulation was maximal; similarly, there was no effect of cGMP at longer incubation times). Thus, inhibition of serum-induced transcription of pSRE-Luc by cGMP in BoASMCs is associated with inhibition of serum-stimulated RhoA activity but not with modulation of Erk1/2 activity, suggesting that cGMP regulates SRF rather than TCF activity.
Effects of cGMP on SRE/SRF-dependent Transcription in CS54 Pulmonary Artery SMCs-To examine the effect of cGMP on SRF-dependent transcription further, we used CS54 rat pulmonary artery SMCs which, unlike BoASMCs, maintain a differentiated phenotype and endogenous G-kinase I expression over multiple passages (40). The specific activity of Gkinase in near-confluent cultures of CS54 cells was 0.31 Ϯ 0.52 nmol/min/mg protein. The transfection efficiency of CS54 cells was higher than that of BoASMCs and allowed us to compare the effect of cGMP on two different reporters as follows: pSRE-Luc under control of the full fos SRE, and pSRF-Luc containing an isolated SRF-binding site without the adjacent TCF-binding site. Serum stimulation of CS54 cells increased luciferase expression from pSRE-Luc 17-fold and from pSRF-Luc 4.5-fold (Fig. 2, A and B, show results obtained with pSRE-Luc and pSRF-Luc, respectively). The effects of serum were completely inhibited in cells co-transfected with an expression vector encoding C3 exoenzyme, indicating Rho dependence (data not shown). Although treating cells with CPT-cGMP had no effect stimulated with 2% serum for 5 (lanes 3 and 4) or 10 min (lanes 5 and 6). Duplicate Western blots of whole cell lysates were probed with an antibody specific for activated, Tyr/Thr-phosphorylated Erk-1 and -2 (pErk, upper panel) and with an antibody that recognizes Erk1/2 irrespective of their phosphorylation status (Erk, lower panel).

FIG. 1. Effects of cGMP on serum-induced RhoA activation, cytoskeletal changes, and SRE-dependent transcription in BoASMCs.
A, BoASMCs were incubated in serum-free DME for 48 h, treated with 250 M CPT-cGMP (cGMP) for 2 h, and serum-stimulated for 3 min as indicated. The rhotekin pull-down assay was performed as described under "Experimental Procedures" with the amount of Rho⅐GTP bound to rhotekin estimated by Western blotting using a RhoA-specific antibody (upper panel). The total amount of RhoA present in cell lysates is shown in the middle panel (10% of input), and the lower panel shows the same blot probed with a G-kinase I-specific antibody. B, BoASMCs were serum-stimulated as described above in the absence (open bars) or presence (closed bars) of 250 M CPT-cGMP. GTP and total nucleotides bound to Rho were measured as described under "Experimental Procedures" with the percent Rho activation calculated as [GTP]/[total nucleotides] bound to Rho (*, p Ͻ 0.03 for the comparison between serum-stimulated Rho activation in the absence versus presence of cGMP). C, BoASMCs were serum-starved for 36 h and treated with 250 M CPT-cGMP for 2 h prior to serum stimulation for 30 min as indicated. F-actin was visualized by staining with Texas Red-labeled phalloidin as described under "Experimental Procedures." Magnification was ϫ40 in all four panels. D, BoASMCs were transfected with 0.2 g of pSRE-Luc and 0.2 g of pTK-RL as described under "Experimental Procedures"; some cells were co-transfected with 0.05 g of an expression vector encoding C3 exoenzyme, and the total amount of DNA was adjusted to 0.6 g. After 16 h of serum starvation, cells were incubated for 8 h in the absence (open bars) or presence (filled bars) of 250 M CPT-cGMP and stimulated as indicated with 2% serum for the last 4 h prior to harvesting. Firefly luciferase activity was normalized to Renilla luciferase activity in each sample, and the relative luciferase activity measured in untreated cells (1st column) was assigned the value of 1 (*, p Ͻ 0.01). E, BoASMCs were serum-starved for 24 h and incubated for 2 h in the absence (lanes 1, 3, and 5) or presence (lanes 2, 4, and 6) of 250 M CPT-cGMP; some cells were on basal luciferase activity, it significantly inhibited seruminduced luciferase expression from both reporters (Fig. 2, A and B, filled bars). As in BoASMCs, treating CS54 cells with CPT-cGMP largely prevented the serum-induced increase in Rho⅐GTP levels and formation of phalloidin-stained stress fibers, and CPT-cGMP had no effect on serum-induced Erk1/2 phosphorylation (data not shown). These results indicate that cGMP inhibits SRF-dependent transcription as well as decreasing serum-induced RhoA activation and stress fiber formation in smooth muscle cells.
Effect of G-kinase on Vinculin mRNA Expression and Vinculin Promoter Activity-Vinculin is an SRF target gene transcriptionally induced by serum in a RhoA-dependent fashion (8,10). In contrast to the fos promoter, where SRF and TCF bind to adjacent sites and cooperate with each other, the vinculin promoter lacks TCF binding, and serum induction of the promoter is independent of MEK/Erk/TCF signaling (8,38).
We examined the effect of serum and cGMP on vinculin mRNA expression in CS54 cells and BoASMCs. Whereas cGMP did not detectably change basal vinculin mRNA levels in serum-starved cells, serum induction of vinculin mRNA was inhibited by cGMP in both cell types (Fig. 3A shows CS54 cells, but similar results were obtained with BoASMCs; equal loading of RNA was demonstrated by reprobing the blot with an rRNA probe shown in the lower panel).
We also examined the effect of cGMP/G-kinase on a luciferase reporter construct under control of the human vinculin promoter (pVP-Luc). We found that serum stimulation increased luciferase expression from pVP-Luc 2-3-fold in CS54 cells and BoASMCs (Fig. 3B shows CS54 cells), similar to the 3-fold serum induction of the promoter reported in NIH3T3 cells (38). When CS54 or BoASMCs were pre-treated with CPT-cGMP, serum induction of pVP-Luc was significantly inhibited. Taken together, these results indicate that G-kinase activation inhibits serum induction of the SRF target gene vinculin.
Effects of cGMP on Serum-induced RhoA Activation and SRE/SRF-dependent Transcription in C6 Glial Cells-C6 glioma cells contain very little endogenous G-kinase activity and, therefore, provide the opportunity to determine whether the effects of cGMP are dependent upon the presence of G-kinase, or could be attributable to other cGMP effectors or cross-activation of cAMP-dependent protein kinase (protein kinase A). To test the effect of cGMP on Rho activity, we stably transfected C6 cells with a G-kinase I expression vector and isolated a clone (C6-GKI) that expressed G-kinase at a physiological level; the specific G-kinase activity of C6-GKI cells was 0.38 Ϯ 0.06 nmol/min/mg protein compared with 0.02 Ϯ 0.01 nmol/ min/mg protein in parental C6 cells. Fig. 4A shows that CPT-cGMP largely prevented the serum-induced increase in Rho⅐GTP without lowering basal Rho⅐GTP levels in C6-GKI cells. When we compared the effect of cGMP in parental and G-kinase-expressing C6 cells, we found that cGMP had no significant effect on serum-induced Rho⅐GTP levels in the Gkinase-deficient parental cells, whereas it lowered serum-induced Rho⅐GTP levels by about 40% in C6-GKI cells (Fig. 4B, open bars, cells stimulated with serum; filled bars, cells stimulated with serum in the presence of CPT-cGMP).
Next, we compared the effect of cGMP on SRE-dependent transcription in transiently transfected parental C6 cells cotransfected with either empty vector or G-kinase vector. The expression and specific activity of G-kinase I in transiently transfected C6 cells was slightly higher than that found in the stably transfected C6-GKI cells (41) (Fig. 4C, compare lanes 2 and 4, C6-GKI cells and C6 cells transiently transfected with G-kinase; lanes 1 and 3 show untransfected C6 cells and cells transfected with empty vector). In G-kinase-deficient C6 cells co-transfected with empty vector, cGMP had no significant effect on either basal or serum-stimulated luciferase activity, whereas in G-kinase-expressing cells, cGMP inhibited serum- stimulated luciferase expression by Ͼ50% (Fig. 4D). Thus, inhibition of SRE-dependent transcription by cGMP, like inhibition of serum-stimulated Rho activation by cGMP, was strictly G-kinase-dependent and not due to cross-activation of protein kinase A or other effects of cGMP in C6 cells.
Control experiments showed that cGMP treatment had no significant effect on serum-induced increases in Erk-1 activity in G-kinase-expressing C6 cells (data not shown). Basal and serum-induced Erk1/2 phosphorylation and luciferase expression from pSRE-Luc were inhibited by the MEK inhibitor U0126 to the same degree; cGMP treatment further inhibited luciferase expression in serum-stimulated cells, suggesting that the G-kinase-mediated inhibition was independent of the Raf/MEK/Erk/TCF pathway (data not shown). Serum stimulation increased luciferase expression from the SRF-specific reporter pSRF-Luc by 3.6 Ϯ 1.1-fold in the absence and 2.3 Ϯ 0.8-fold in the presence of CPT-cGMP in G-kinase-expressing C6 cells, confirming the inhibition of SRF function by cGMP/G-kinase.

Effect of G-kinase on SRF-dependent Transcription and Rho
Activation Induced by Constitutively Active G␣ 12 , G␣ 13 , and p115RhoGEF-Lysophosphatidic acid, the main RhoA-stimulating component of serum, activates RhoA through the heterotrimeric G-protein G␣ 13 , whereas thrombin appears to activate RhoA through G␣ 12 (6). Constitutively active forms of G␣ 12 and G␣ 13 have been shown to induce SRF-dependent transcription through activation of RhoA, which may involve activation of p115RhoGEF as an intermediate step (1,53,54). To define more clearly the mechanism(s) by which G-kinase inhibits serum-induced SRF-dependent transcription, we decided to examine the effects of G-kinase on G␣ 12 , G␣ 13 , and p115RhoGEF signaling to SRF.

FIG. 4. Effects of cGMP on serum-induced RhoA activation and SRE-dependent transcription in C6 glial cells.
A, stably transfected, G-kinase I ␤-expressing C6 cells were incubated in serum-free DME for 48 h, treated with 250 M CPT-cGMP for 2 h, and stimulated with 10% FBS for 3 min as indicated. The amount of Rho⅐GTP bound to rhotekin was estimated by pull-down assay and Western blotting with a RhoA-specific antibody as described in Fig. 1A; the total amount of RhoA present in cell lysates (10% of input) is shown in the lower panel. B, G-kinase-deficient wild type C6 cells (C6-Wt) and G-kinase-expressing stably transfected C6 cells (C6-GKI) were serum-starved and then serum-stimulated as described in A; GTP and total nucleotides bound to Rho were measured as described in dead mutant G-kinase I, but in C6 cells co-transfected with wild type G-kinase I, cGMP inhibited G␣ 13 (QL)-stimulated luciferase expression by Ͼ50% (Fig. 5A). Similar results were obtained when G␣ 12 (QL) was substituted for G␣ 13 (QL) (data not shown). Activation of protein kinase A in G-kinase-deficient C6 cells, e.g. by addition of 100 M 8-chlorophenylthio-cAMP, inhibited G␣ 13 (QL)-stimulated luciferase expression from pSRE-Luc by 63 Ϯ 2% (n ϭ 3), similar to the effect observed with cGMP-activated G-kinase. These results are in keeping with the inhibition of G␣ 13 (QL)-induced SMC contraction by cAMP as well as cGMP analogs (21).
Transfection of a constitutively active p115RhoGEF construct (⌬N-p155-RhoGEF (33)) into C6 cells increased SRE-dependent transcription about 10-fold, and this effect was inhibited by cGMP in a G-kinase-dependent fashion (Fig. 5B). For comparison, transfection of full-length, wild type p115RhoGEF increased reporter gene expression only 2.7 Ϯ 0.4-fold in the absence and 1.7 Ϯ 0.2-fold in the presence of cGMP in Gkinase-expressing C6 cells (n ϭ 3).
The high transfection efficiency of C6 cells allowed us to determine the effect of constitutively active G␣ 13 and ⌬N-p115RhoGEF on RhoA activity and examine the effect of cGMP/G-kinase. We co-transfected C6 cells with G-kinase I and EE epitope-tagged RhoA together with either empty vector, constitutively active G␣ 13 , or ⌬N-p115RhoGEF, and we measured the amount of GTP and GDP bound to RhoA in anti-EE immunoprecipitates. The results are shown in Fig. 5C; transfection of constitutively active G␣ 13 or ⌬N-p115RhoGEF increased RhoA activation from 4 to about 12%. Treating cells with CPT-cGMP significantly inhibited G␣ 13 -stimulated Rho activity but did not affect basal or ⌬N-p115RhoGEF-stimulated Rho activity. The effect of cGMP on G␣ 13 -stimulated Rho activity is in keeping with the inhibitory effect of cGMP on G␣ 13stimulated SRE-dependent transcription (Fig. 5A) and the inhibitory effect of cGMP on G␣ 13 -stimulated SMC contraction (21). However, the lack of an effect of cGMP on ⌬N-p115RhoGEF-stimulated Rho activity was surprising. For comparison, transfection of full-length, wild type p115RhoGEF increased the amount of GTP bound to Rho only 1.8-fold in the absence and 1.4-fold in the presence of cGMP in G-kinaseexpressing C6 cells (n ϭ 2; the difference between the absence and presence of cGMP did not reach statistical significance). These results suggest that cGMP/G-kinase inhibition of serumand G␣ 13 -stimulated Rho activation may involve functional inhibition of a limiting, endogenous RhoGEF, which could be overcome by overexpression of constitutively active ⌬N-p115RhoGEF. These results further suggest that the G-kinase inhibition of ⌬N-p115RhoGEF-stimulated transcription is not due to inhibition of RhoA activation but rather involves inhibition of downstream effects of RhoA.
We also examined the effect of cGMP on G␣ 12 (QL)-and ⌬N-p115RhoGEF-stimulated SRE-dependent transcription in different types of muscle cells expressing endogenous G-kinase. In BoASMCs, CS54 pulmonary artery SMCs, and primary neonatal rat cardiomyocytes, transfection of constitutively active G␣ 12 increased luciferase expression from pSRE-Luc about 7-9-fold, and treating cells with CPT-cGMP inhibited this effect by ϳ50% (Table I). Similar results were obtained with G␣ 13 (data not shown). Transfection of constitutively active ⌬N-p115RhoGEF increased SRE-dependent transcription 4 -8-fold with about 40% inhibition in cGMP-treated cells. All three cell types contain comparable amounts of G-kinase activity (described above for BoASMCs and CS54 cells, and in Ref. 55 for cardiomyocytes). Of note, the effects of cGMP on serum-and G␣ 12 -induced transcription diminished progressively with increasing passage of BoASMCs and were lost in cells cultured for Ͼ11 passages; high passage BoASMCs cease to express G-kinase I (15). Thus, the effect of cGMP on G␣ 12,13 -and ⌬N-p115RhoGEF-stimulated transcription occurs in cells at physiological levels of G-kinase I.
Transcriptional Effects of Constitutively Active RhoA Mutants Are Inhibited by G-kinase-To determine whether Gkinase may inhibit transcriptional effects downstream of RhoA, we used two GTPase-deficient, constitutively active RhoA mutants. Transfection of RhoA(63L) into G-kinase-expressing C6 cells increased pSRE-Luc transcription about 9-fold (Fig. 6A), with similar results having been reported in other cell types (7,46,56). Treating the cells with CPT-cGMP significantly inhibited RhoA(63L)-induced reporter gene activity (Fig. 6A), although it did not affect the expression level of the epitopetagged RhoA(63L) (Fig. 6B, compare lanes 3 and 4, cells transfected with EE-RhoA(63L) cultured in the absence and presence of cGMP, respectively, to lanes 1 and 2, cells transfected with empty vector). Moreover, cGMP treatment did not affect the high activation state of RhoA(63L) (Fig. 6C). In G-kinase-deficient C6 cells, CPT-cGMP treatment did not affect RhoA(63L)-induced reporter gene activity; however, activation of protein kinase A by CPT-cAMP inhibited RhoA(63L)induced luciferase expression by 55 Ϯ 4% (n ϭ 3). Results similar to those described in Fig. 6A were obtained in CS54 cells, and CPT-cGMP also inhibited the effect of RhoA(V14) on pSRE-Luc expression (data not shown). These data suggest that cGMP/G-kinase inhibit steps downstream of RhoA.
G-kinase, like protein kinase A, phosphorylates RhoA on Ser 188 in vitro, and it has been suggested that protein kinase A phosphorylation of Ser 188 may uncouple RhoA from interaction with downstream effectors such as ROK (57). To test the hypothesis that RhoA Ser 188 phosphorylation by G-kinase may mediate inhibition of RhoA-dependent transcription, we examined the effect of cGMP/G-kinase on the transcriptional effects of RhoA(63L/188A), a constitutively active RhoA mutant that cannot be phosphorylated by G-kinase. We confirmed that purified G-kinase phosphorylated bacterially expressed RhoA(63L) but not RhoA(63L/188A) in vitro (data not shown).  12 and ⌬N-p115RhoGEF in muscle cells BoASMCs, CS54 pulmonary artery SMCs, and primary cardiomyocytes (myocytes) were transfected with pSRE-Luc and either pTK-RL (for BoASMCs) or pRSV-␤Gal and were co-transfected with either empty vector or expression vectors encoding constitutively active G␣ 12 or ⌬N-p115-RhoGEF. After 24 h of serum starvation, cells were cultured for 8 h in the presence (ϩcG) or absence (ϪcG) of 250 M CPT-cGMP as indicated. Firefly luciferase activity was normalized to either Renilla luciferase activity (for BoASMCs) or ␤-galactosidase activity, and the relative luciferase activity measured in untreated cells transfected with empty vector was assigned a value of 1. As demonstrated in Fig. 6A, RhoA(63L/188A) induced SRE-dependent transcription to the same degree as RhoA(63L), and this effect was inhibited by cGMP to a similar extent as with RhoA(63L). The expression level and activation state of RhoA(63L/188A) were similar to that of RhoA(63L) and were not affected by cGMP (Fig. 6, B and C). We conclude that inhibition of RhoA-dependent transcription by cGMP/G-kinase does not require phosphorylation of RhoA on Ser 188 .

Expression of a Constitutively Active RhoA Mutant Prevents the Effects of G-kinase on Glial Cell
Morphology-In neuronal and astroglial cells, cytoskeletal changes induced by RhoA activation cause rounding of the cell body and retraction of cell processes, whereas inactivation of RhoA by C3 exoenzyme leads to extension of neurite-like processes (6, 58 -60). To examine the effect of cGMP on glial cell morphology, we transiently transfected parental C6 cells with either empty vector or G-kinase I vector and co-transfected an expression vector for ␤-galactosidase to identify transfected cells. Cells transfected with empty vector remained flat, elongated, and polygonal, in the absence and presence of cGMP (Fig. 7). In contrast, cells transfected with G-kinase vector and treated with cGMP became small and dense, with Ͼ80% of the cell population developing neurite-like extensions greater than one cell diameter in length (Fig. 7). Cells in which RhoA was inhibited by C3 exoenzyme transfection showed similar morphological changes as those induced by G-kinase activation, suggesting that some of the same effector molecules may be involved (Fig. 7, C3 exoenzyme induced neurite-like extensions in Ͼ90% of the cell population).
Next, we examined the effect of cGMP/G-kinase on RhoA(63L)-induced changes in glial cell morphology. As described for other cell types of glial/neuronal origin (58 -60), transfection of constitutively active RhoA(63L) induced cell rounding in G-kinase-expressing C6 cells, producing pancakeshaped cells without extensions (Fig. 7, upper panels, Ͻ5% of the cells showed neurite-like extensions). Treating these cells with CPT-cGMP failed to induce extension of cell protrusions in contrast to the morphological changes observed in cells expressing G-kinase alone (Fig. 7, lower panels, Ͻ5% of the cGMP-treated RhoA(63L)-and G-kinase-expressing cells demonstrated neurite-like extensions). Of note, the amounts of G-kinase I and RhoA(63L) vector DNA transfected in these experiments were the same as those used in the experiments where G-kinase activation inhibited the transcriptional effects of RhoA(63L) (Fig. 6A). Thus, the effects of RhoA(63L) on glial cell morphology are dominant over the effects of G-kinase, whereas the downstream effectors of RhoA(63L) mediating SREdependent transcription are sensitive to cGMP/G-kinase inhibition.
Transcriptional Effects of RhoA Effector Kinases Are Cell Type-specific and Inhibited by G-kinase-The RhoA effectors mediating SRF-dependent transcriptional activation have not been clearly defined, although it is likely that RhoA activates SRF through a combination of effectors and that some of the Rho effectors involved in transcription also induce cytoskeletal changes (6,10,12,13,56,(61)(62)(63). Constitutively active forms of ROK, PKN, and PRK-2 have been shown to stimulate transcription from the fos SRE in different cell types (56,61,62). We examined the effect of constitutively active ROK, PKN, and PRK-2 on pSRE-Luc in SMCs and glial cells and studied the effect of G-kinase. In G-kinase-expressing C6 cells, constitutively active ROK induced reporter gene activity about 8-fold, and this effect was inhibited Ͼ50% by cGMP (Fig. 8A). PKN and PRK-2 had no effect on pSRE-Luc in C6 cells. All Rho effector kinases were expressed at similar levels in the presence and absence of cGMP (Fig. 8B, compare lanes 2 and 3, Fig. 2. B, cells were transfected with empty vector (lanes 1  and 2), EE-RhoA(63L) (lanes 3 and 4), or EE-RhoA(63L/188A) (lanes 5 and 6) and incubated in the absence (lanes 1, 3, and 5) or presence (lanes 2, 4, and 6) of CPT-cGMP as described in A. Whole cell lysates were analyzed by SDS-PAGE/electroblotting, and blots were developed with an anti-EE antibody. C, cells were transfected with EE epitopetagged wild type RhoA (Rhowt), RhoA(63L), or RhoA(63L,188A) and treated as described in A, but cell lysates were subjected to immunoprecipitation with anti-EE antibody; GTP and GDP bound to the Rho immunoprecipitates were measured as described under "Experimental Procedures" with percent Rho activation calculated as [GTP]/[GTP ϩ GDP] bound to Rho.  1 g), or an expression vector encoding C3 exoenzyme (0.05 g) as indicated; the total amount of DNA was adjusted to 0.6 g. Cells were maintained in DME with 10% FBS for 24 h and incubated for 1 h in the absence (upper panels) or presence (lower panels) of 250 M CPT-cGMP prior to fixing and staining for ␤-galactosidase activity to visualize successfully transfected cells. All cells were photographed at ϫ40 magnification. gene activity about 4-fold with PRK-2 having a small but statistically significant effect and ROK showing a minimal effect (Fig. 8C, open bars). Both PKN-and PRK-2-induced SRE-dependent transcription was significantly inhibited by cGMP (Fig. 8C, filled bars). In CS54 cells, the transcriptional effect of constitutively active ROK reached statistical significance (2.3fold increase in luciferase expression in the absence of cGMP) and was inhibited in the presence of cGMP (1.6-fold increase in luciferase expression in the presence of cGMP, mean of two independent experiments performed in duplicate). Thus, Gkinase activation appears to interfere with the function of several downstream effectors of RhoA involved in transcriptional regulation.
Because ROK seemed to be one of the main RhoA effectors able to increase SRE-dependent transcription in C6 cells, we examined the effect of the specific ROK inhibitor Y27632 on G␣ 13 -and RhoA(63L)-induced reporter gene expression in these cells (Fig. 9A). By itself, Y27632 inhibited G␣ 13 -and RhoA(63L)-induced transcription to a similar extent as did cGMP; the effects of Y27632 and cGMP combined appeared to be additive. These results suggest that cGMP/G-kinase inhibit downstream effectors of RhoA in addition to ROK in C6 cells, as suggested by the findings in other cell types (Fig. 8C) (12,13,61,62).
The transcriptional effects of cGMP and G-kinase I on the full-length fos promoter require nuclear translocation of the kinase in C6 cells (41); therefore, we examined whether Gkinase has to translocate to the nucleus to inhibit SRF-dependent transcription. We used a G-kinase I␤ variant with two amino acid substitutions in the nuclear localization signal that prevent nuclear translocation and fos promoter transactivation (GKI␤(K407A/R409A)) without affecting kinase activity (26); we found that this G-kinase mutant mediated inhibition of RhoA(63L)-and ROK-induced pSRE-Luc expression by cGMP as efficiently as wild type G-kinase (data not shown). These results suggest that G-kinase acts on extranuclear targets to inhibit SRF-dependent transcription.
G-kinase Does Not Modulate the Activity of Constitutively Active ROK-Because G-kinase inhibited transcriptional effects of ROK in C6 cells, we examined whether G-kinase affected the enzymatic activity of the constitutively active ROK catalytic domain, which is independent of Rho⅐GTP binding (35). We transfected C6 cells with the expression vector encoding Myc epitope-tagged constitutively active ROK, and we determined ROK activity in anti-Myc immunoprecipitates using myosin light chain as a substrate. The amount of 32 PO 4 incorporation into myosin light chain was the same in ROK immunoprecipitates from G-kinase-expressing C6 cells cultured in the presence and absence of cGMP (Fig. 9B, upper panel,

DISCUSSION
The data presented in this study demonstrate that the cGMP/G-kinase signaling pathway inhibits SRE/SRF-dependent transcription through inhibition of RhoA signaling. Treating cells with cGMP inhibited SRE-dependent transcription induced by serum, by upstream activators of RhoA, and by constitutively active RhoA mutants in a G-kinase-dependent fashion; it also inhibited serum induction of the SRF target gene vinculin. G-kinase inhibited RhoA signaling by both upstream and downstream mechanisms: (i) G-kinase inhibited RhoA activation by serum and G␣ 13 ; and (ii) G-kinase inhibited transcription induced by constitutively active RhoA mutants and by the constitutively active Rho effector kinases ROK, PKN, and PRK-2. These effects were observed both in cells of smooth muscle and neuronal origin. The regulation of RhoAdependent transcription by the NO/cGMP/G-kinase pathway may play an important role during smooth muscle differentiation and cardiac hypertrophy, because recent studies have demonstrated that RhoA signaling to SRF is required for expression of smooth muscle-specific genes during muscle differentiation and is involved in transcriptional responses associated with cardiac hypertrophy (4,5,9,61,64). On the other hand, RhoA activation has little effect on the intact fos promoter, consistent with the finding that cGMP/G-kinase activates the fos promoter and increases c-fos mRNA expression in serum-starved cells through RhoA/SRE-independent mechanisms (8,23,27).
Investigations from multiple laboratories have provided evidence that activation of either the NO/cGMP/G-kinase or the cAMP/protein kinase A pathway can antagonize RhoA functions. In vascular SMCs, G-kinase inhibits RhoA-induced calcium sensitization of the contractile apparatus, which contributes to the vasodilatory effects of NO/cGMP (16, 19 -21). In several cell types including vascular SMCs, G-kinase activation induces cytoskeletal disorganization with the break down of stress fibers and disassembly of focal adhesion complexes (19,22,44,65,66). Activation of protein kinase A similarly inhibits agonist-induced smooth muscle contraction and induces stress fiber/focal adhesion breakdown in different cells; in addition, cAMP inhibits Rho-dependent leukocyte adhesion and induces morphological changes in neuronal and glial cells that are reciprocal to those induced by RhoA activation (21,57,60,(67)(68)(69).
Several observations indicate that the cGMP-or cAMP-induced effects on RhoA-dependent SMC contraction and cytoskeletal organization require phosphorylation of RhoA by G-kinase or protein kinase A. G-kinase inhibits RhoA but not RhoA(188A)-induced calcium sensitization of SMC contraction, and cells transfected with RhoA(188A) are protected from Gkinase-induced stress fiber disassembly (19,22). Similarly, transfection of RhoA(14V,188A) is more effective than transfection of RhoA(14V) in preventing cAMP-induced cytoskeletal changes in SH-EP cells (57). In vitro, Ser 188 in RhoA is efficiently phosphorylated by both G-kinase and protein kinase A (19,22,70). However, demonstration of G-kinase-induced RhoA phosphorylation in vivo has not been possible, and difficulties in detecting in vivo phosphorylation of RhoA by protein kinase A have been reported (19,71). One group (70) found that cAMP treatment induced low levels of phosphorylation of only membrane-bound RhoA in lymphoid cells and demonstrated that cAMP induced translocation of RhoA from the membrane to the cytosol. Whereas protein kinase A phosphorylation of RhoA in vitro does not affect the ability of RhoA to bind or hydrolyze GTP (70), cAMP treatment of intact leukocytes has been shown to inhibit chemoattractant-stimulated guanine nucleotide exchange on RhoA (67). These findings have led to the hypothesis that protein kinase A/G-kinase phosphorylation of RhoA may lead to membrane dissociation and interfere with Rho/Rho-GEF association leading to decreased Rho activity (67,70). Recently, one group (72) demonstrated that integrininduced RhoA activation and membrane translocation was inhibited by increased intracellular cAMP concentrations.
Activation of G-kinase has also been shown to prevent agonist-induced RhoA membrane translocation suggesting inhibition of RhoA activation, but RhoA⅐GTP levels were not measured in these studies (19,20,22,44). We found that G-kinase activation inhibited serum-induced RhoA⅐GTP loading in BoASMCs, CS54 cells, and in G-kinase-expressing but not in G-kinase-deficient C6 cells. The results with C6 cells exclude the possibility that cross-activation of protein kinase A by cGMP might be responsible for the reduction in Rho⅐GTP levels. Although serum stimulation of RhoA activity was modest in all three cell types studied, it was similar to that described in other cell types (28). Transfection of a constitutively active G␣ 13 increased RhoA⅐GTP loading more dramatically, and the effect was inhibited by G-kinase. Transfection of the constitutively active ⌬N-p115RhoGEF increased RhoA⅐GTP loading to a similar extent as that observed with G␣ 13 , but in contrast to G␣ 13 , the effect of this GEF construct was resistant to G-kinase inhibition. These results suggest that G-kinase may inhibit the function of an endogenous GEF responsible for RhoA activation by serum and G␣ 13 , e.g. by interrupting the interaction of the GEF with its upstream regulator or with RhoA. An overexpressed, constitutively active ⌬N-p115RhoGEF could be resistant to G-kinase inhibition because the GEF activity is independent of G-protein activation, the altered GEF is no longer sensitive to G-kinase, or high concentrations of active GEF may force an interaction with RhoA irrespective of G-kinase phosphorylation. Compared with ⌬N-p115RhoGEF, transfection of full-length, wild type p115RhoGEF increased RhoA⅐GTP loading significantly less, and its effect appeared to be inhibited by cGMP/G-kinase. However, activation of RhoA by wild type p115 was marginal, and the effect of cGMP did not reach statistical significance. More work is required to identify the exact mechanism(s) by which G-kinase inhibits Rho activation.
Although RhoA Ser 188 phosphorylation by G-kinase or protein kinase A may induce changes in RhoA membrane association and interfere with RhoA activation by GEFs (67,70), other workers have reported that phosphorylation of RhoA by protein kinase A in vitro decreased the binding of RhoA to ROK, leading to the notion that RhoA phosphorylation may interfere with RhoA association with downstream effectors (57). We found that G-kinase inhibited the transcriptional effects of RhoA(63L) and RhoA(63L,188A) to a similar extent without affecting the high activation state of both mutants; these results indicate that G-kinase inhibited downstream effects of RhoA independent of Rho phosphorylation. G-kinase inhibition of RhoA(63L)-dependent transcription contrasted with the finding that RhoA(63L)-transfected C6 cells were resistant to the morphological changes induced by cGMP; similarly, RhoA(14V)-transfected glial cells are resistant to the morphological changes induced by cAMP (60). Although the morphological changes induced by cGMP (or cAMP) in glial cells are similar to the changes induced by C3 exoenzymemediated Rho inactivation (60, 69) (Fig. 7), cGMP treatment did not significantly decrease basal Rho⅐GTP levels in serumstarved C6 cells, suggesting that the cGMP-induced morphological changes occur independently of Rho⅐GTP levels. The cytoskeletal effectors responsible for RhoA(63L)-induced morphology appeared to be less sensitive to G-kinase inhibition than the RhoA effectors involved in transcriptional regulation, supporting the notion that RhoA effectors involved in cytoskeletal organization only partly overlap with effectors involved in transcriptional regulation (12,13).
The effectors of RhoA that mediate transcriptional activation of SRF are incompletely defined. Confirming the observations of others (56,61,62), we found that constitutively active ROK, PKN, and PRK-2 activated the SRE-dependent reporter gene to different degrees in different cell types. All three kinases have been found to induce cytoskeletal changes (1,(35)(36)(37). Notably, the transcriptional effects of the three constitutively active kinase constructs were inhibited by cGMP/G-kinase, suggesting that G-kinase inhibited step(s) distal or parallel to these Rho effectors which is (are) required for SRF activation. We found that the catalytic activity of constitutively active ROK was not affected by G-kinase. Sandu et al. (20) reported that cGMP treatment of SMCs prevented thrombin-induced ROK activation, presumably by preventing thrombin-induced RhoA activation. Likewise, cGMP treatment of SMCs inhibited thrombin-induced RhoA activation and Rho/ROK-dependent DNA synthesis. 3 On the other hand, stress fiber induction by constitutively active ROK is resistant to inhibition by cAMP/ protein kinase A (57).
A recently developed model for SRF regulation via actin dynamics suggests that reduction of the soluble G-actin pool by RhoA-induced actin polymerization may be coupled to activation of SRF (10). G-kinase could inhibit RhoA signaling to SRF by increasing the G/F-actin ratio through phosphorylation of cytoskeletal components distal to the known RhoA effector kinases. We found that G-kinase activation reduced phalloidinstainable F-actin in serum-starved BoASMCs and CS54 cells, without significantly decreasing basal Rho⅐GTP levels. Others (19,22,44,65,73) have also observed decreased F-actin staining and/or an increase in the G-actin fraction in NO/cGMPstimulated cells. One possible mechanisms whereby G-kinase could inhibit actin polymerization and increase G-actin levels independently of RhoA⅐GTP levels is through phosphorylation of the vasodilator-stimulated phosphoprotein VASP. VASP is found in focal adhesions and along F-actin fibers, and VASP phosphorylation down-regulates its F-actin binding and actinpolymerization-promoting activity (74). Work is underway to examine the role of VASP phosphorylation in the regulation of SRF-dependent transcription.