RhoA Signaling via Serum Response Factor Plays an Obligatory Role in Myogenic Differentiation*

Serum response factor (SRF) plays a central role during myogenesis, being required for the expression of striated α-actin genes. As shown here, the small GTPase RhoA-dependent activation of SRF results in the expression of muscle-specific genes, thereby promoting myogenic differentiation in myoblast cell lines. Co-expression of activated V14-RhoA and SRF results in an approximately 10-fold activation of the skeletal α-actin promoter in replicating myoblasts, while SRFpm1, a dominant negative SRF mutant, blocks RhoA dependent skeletal α-actin promoter activity. Serum withdrawal further potentiates RhoA- and SRF-mediated activation of α-actin promoter to about 30-fold in differentiated myotubes. In addition, the proximal SRE1 in the skeletal α-actin promoter is sufficient to mediate RhoA signaling via SRF. Furthermore, SRFpm1 and to a lesser extent dominant negative N19-RhoA inhibit myoblast fusion, postreplicative myogenic differentiation, and expression of direct SRF targets such as skeletal α-actin and indirect targets such as myogenin and α-myosin heavy chain. Moreover, RhoA also stimulates the autoregulatable murine SRF gene promoter in myoblasts, and the expression level of SRF is reduced in myoblasts overexpressing N19-RhoA. Our study supports the concept that RhoA signaling via SRF serves as an obligatory muscle differentiation regulatory pathway.

Serum response factor (SRF), 1 a homodimeric DNA-binding protein, is generally presumed to be a ubiquitous transcription factor (1,2). SRF is a member of an ancient DNA-binding protein family, whose relatives share a highly conserved DNAbinding/dimerization domain of 90 amino acids, termed the MADS box (3). SRF acts through binding to a consensus DNA sequence, the serum response element (SRE), which is a 22-bp element of dyad symmetry containing an inner core sequence, CC(A/T) 6 GG, also known as the CArG box or CBARs (4 -6).
Functional SREs are found in the promoters of genes involved in the early mitogenic response (4 -6) and also in the promoters of several muscle-specific genes such as striated and smooth muscle ␣-actins (7)(8)(9). A CArG box factor reported to regulate the transcription of ␣-actin genes has been shown to be indistinguishable from SRF that binds to the c-fos SRE (6,7,10). These ␣-striated actins are also among the earliest gene products that mark the commitment of embryonic cell types to cardiac, skeletal, and smooth muscle cell lineages. Paradoxically, how could SRF function as a mediator of intracellular signals leading to cell proliferation when it appears to also have a primary role in mediating expression of muscle-specific genes in postreplicative myocytes?
Considerable amounts of evidence support an obligatory role for SRF in regulating muscle gene expression. Like the related MADS box-containing MEF2 genes (11), SRF is preferentially expressed in embryonic heart, skeletal, and smooth muscle tissues (12,13). We have observed increased SRF gene expression during avian embryogenesis in precardiac splanchnic mesoderm and dorsal somitic mesoderm. SRF protein is also expressed selectively in the myocardium during heart morphogenesis and in the myotomal segment of anterior somites in avian embryos. In fact, the expression levels of SRF in embryonic and adult cardiac tissues are at least 2 orders of magnitude greater than those detected in the liver and other endodermal derived tissues (12). Thus, high SRF protein mass coincides with terminal differentiation of striated and smooth muscle (14,15). Moreover, SRF has been shown to transactivate skeletal ␣-actin gene transcription under conditions that block myogenic differentiation (16). Further evidence for its role as a positive regulator of ␣-actin genes comes from studies in which neutralization of SRF activity by microinjection of SRF antibodies (17) or blocking of SRF gene expression through application of SRF antisense oligonucleotides (18) prevented myogenic dependent gene expression and late stage terminal differentiation in C2 myoblasts. In addition, a dominant negative mutant of SRF, SRFpm1, blocked transcriptional activation of the skeletal ␣-actin promoter in primary chick myoblast cultures (13). This SRF mutant, containing triple point mutations within the MADS box, is defective in DNA binding but capable of heterodimerizing with wild type SRF monomere, thereby inhibiting the transcriptional activity of SRF (19).
We wanted to determine if cell replication-and differentiation-associated signals might differentially regulate SRF-dependent gene activity during myogenesis. Recent studies indicate that at least two distinct signaling pathways can activate SRF-dependent c-fos transcription in response to growth factor stimulation. One pathway requires the ternary complex factor (TCF) (e.g. Elk-1, Sap-1), which contacts an Ets recognition motif adjoining the SRF-binding site on the c-fos promoter (reviewed in Ref. 20). Activation of the Ras/Raf/mitogen-activated protein kinase kinase/extracellular signal-regulated kinase signaling cascade phosphorylates TCF (21), causing it to bind to the SRF⅐SRE complex to form a ternary complex, which in turn activates c-fos gene transcription. Ras regulates the expression of specific genes associated with cellular proliferation and transformation. Ras is also known to inhibit myogenesis in a manner that involves suppression of the expression of key myogenic factors such as MyoD and myogenin (22,23).
Via a second pathway, elucidated by Hill et al. (24), Rho GTPases activate SRF-dependent but TCF-independent c-fos transcription. Constitutively active RhoA, Rac1, and Cdc42 activate c-fos transcription by regulating the SRF-SRE interaction directly in the absence of TCF binding. RhoA is involved in many actin-based processes including cell adhesion to the extracellular matrix, cytokinesis, and smooth muscle contraction (reviewed in Refs. 25 and 26). The ability of RhoA to orchestrate actin filament rearrangement and to regulate SRF transcriptional activity raises the possibility that RhoA might have an important role in cell morphology changes and inactivation of muscle-specific gene expression occurred during muscle differentiation. Consistent with this hypothesis, a recent report has shown that inhibition of Rho family proteins by the GDP dissociation inhibitor RhoGDI suppresses myogenesis as well as expression of muscle-specific genes in C 2 C 12 myoblasts (27). Concurrently, we also asked whether RhoA signaling was dependent upon SRF during muscle differentiation.
In the present study, we found that unlike Ras, which mediates suppression of myogenic gene activity, RhoA strongly activates skeletal ␣-actin promoter activity in passaged myoblast cell lines. RhoA and an activated V14-RhoA mutant potentiate skeletal ␣-actin promoter activity in synergy with SRF. Expression of dominant-negative SRFpm1 inhibits RhoAmediated skeletal ␣-actin promoter activation. In addition, RhoA signaling through SRF appears to be necessary for myogenic differentiation, since overexpression of SRFpm1 and to a smaller extent N19-RhoA, the dominant-negative RhoA mutant, inhibits the appearance of myogenin and contractile proteins. RhoA also stimulates the autoregulatable murine SRF promoter in myoblasts, and the expression level of SRF is reduced in myoblasts overexpressing N19-RhoA. This study supports the concept that RhoA signaling via SRF is an integral part of a regulatory pathway during muscle terminal differentiation.

MATERIALS AND METHODS
Plasmid Constructs-The reporter plasmid SK-luc contains the avian skeletal ␣-actin promoter from Ϫ398 to ϩ25 bp followed by firefly luciferase (28). The reporter plasmids SK-SRE1-luc and c-fos-luc contain the SRE closest to the transcription initiation site of the skeletal ␣-actin promoter (Ϫ100 to Ϫ73 bp) and the c-fos SRE region (Ϫ318 to Ϫ291 bp), respectively (28). The reporter plasmid SRF-luc contains the mouse SRF promoter region from Ϫ456 to Ϫ23 bp (12). A cytomegalovirus promoter-driven expression vector (pCGN), which contains an N-terminal influenza hemagglutinin (HA) epitope, was used to express full-length SRF (pCGN-SRF), and SRFpm1 protein containing triple point mutants, which converted Arg 143 , Lys 145 , and Leu 146 to Ile 143 , Ala 145 , and Gly 146 within the MADS box (19). RhoA cDNA was cloned from HeLa cells by reverse transcription-polymerase chain reaction, and cDNA sequence specifying the entire human RhoA coding region (152-733 bp) was cloned into the XbaI and BamHI sites of the expression vector pCGN, immediately downstream of and in frame with the HA epitope (29). The expression plasmid pCGN-V14 was also subcloned into pCGN by polymerase chain reaction from pEBG-V14-RhoA (kindly provided by Dr. Joe Avruch). HA-N19-RhoA inserted into the pCMV5 vector (pCMV5-N19-RhoA) was kindly provided by Dr. Gary Bokoch.
Tissue Culture, Plasmid DNA Transfection, and Reporter Gene Assays-Sol8 (30) or C 2 C 12 mouse myoblasts (31) were maintained in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) with 10% fetal bovine serum. Cells were plated at a density of 2 ϫ 10 5 cells in 60-mm tissue culture plates and were transfected after 24 h. Cells were transfected with approximately 1 g of total plasmid DNA containing the indicated reporter plasmid (SK-luc, c-fos-SRE-luc, SRF-luc, or G5-TATA-luc) and the indicated expression plasmids balanced with parental expression vector. Transfections were performed using Lipo-fectAMINE (Life Technologies) according to the manufacturer's instructions. Cells were placed in Dulbecco's modified Eagle's medium with 10% fetal bovine serum and harvested 48 h post-transfection. Luciferase activity and protein content were measured as described previously (28). Luciferase activity was normalized to the total protein, and data were expressed as luciferase activity normalized to base-line reporter gene activity. For experiments utilizing differentiated myotubes, transfection and initial culture conditions were as described above, but the growth medium (Dulbecco's modified Eagle's medium with 10% fetal bovine serum) was removed 14 -18 h post-transfection and replaced with the differentiation medium (Dulbecco's modified Eagle's medium with 2% heat-inactivated horse serum). Cells were cultured in this medium until harvested 48 h later. All experiments were performed in duplicate and were repeated two or three times.
Isolation of Stably Transfected SRFpm1 and N19-RhoA Murine Myoblasts-Sol8 or C 2 C 12 myoblasts plated at a density of 1 ϫ 10 6 cells/ 100-mm tissue culture dish in growth medium were co-transfected with 20 g of pCGN-SRFpm1 or pCMV5-N19-RhoA and 2 g of pSVneo selection plasmid in a DNA calcium phosphate mixture (10). After overnight incubation, myoblasts were allowed to recover, split 1:20 with fresh growth media, and then challenged in selection media (400 g/ml G418). Individual colonies were selected approximately 10 -14 days following the addition of selection media. Whole cell protein extracts were collected from expanded individual colonies and evaluated for expression of the HA-SRFpm1 or HA-N19-RhoA. To initiate terminal differentiation assay, cells were plated at density of 5 ϫ 10 5 onto 60-mm dishes. After 24 h, the cultures were switched to differentiation medium. At the time points indicated, cultures were harvested and examined for myogenin and contractile protein expression.
Protein Expression Analysis-Expression levels of cDNA constructs were examined by Western blot analysis. Myoblasts on 60-mm plates were transiently transfected with 2 g of each expression plasmid using LipofectAMINE. After 48 h, cells were solubilized with lysis buffer (20 mM Tris, pH 7.5, 1% Triton X-100, 100 mM NaCl, 5 mM EDTA, 2 mM phenylmethylsulfonyl fluoride, 2 g/ml aprotinin, 1 g/ml leupeptin, 1 g/ml pepstatin, 50 g/ml antipain) for 30 min with shaking at 4°C. Cell lysates were then cleared by centrifugation. In the case of detecting HA epitope-tagged SRF and RhoA, whole cell protein extracts (50 g) were run on a 10% SDS-polyacrylamide gel electrophoresis gel, transferred to Immobilon membrane (Millipore), and probed with anti-HA monoclonal antibody 12CA5 (Boehringer Mannheim). In the case of detecting myogenin and ␣-myosin heavy chain (MHC), cell extracts (20 g) were fractionated on 12 and 7% SDS-polyacrylamide gel electrophoresis gels, respectively, and were probed with M-225 polyclonal antibody (Santa Cruz Biotechnology, Inc, Santa Cruz, CA) recognizing myogenin and with MF-20 antibody detecting ␣-MHC (Developmental Studies Hybridoma Bank).
Electrophoretic Mobility Shift Assays-Electrophoretic mobility shift assays (EMSAs) were performed using ␥-32 P-end-labeled doublestranded oligonucleotide probe corresponding to the proximal SRE1 of the skeletal ␣-actin promoter ( Ϫ96 GACACCCAAATATGGCGACG Ϫ77 ). Whole cell protein extracts were collected from cell cultures according to the protocols of Lee et al. (16). EMSA reactions were performed as described in Lee et al. (16) with the inclusion of 4 g of poly(dG-dC) as the nonspecific competitor. SRF binding activity was estimated by autoradiography on x-ray film (Eastman Kodak Co.) and by Betagen scans.
RNA Hybridization Assays-Total RNA was obtained from myoblast cultures according to Chomczynski and Sacchi (32). Total (20 g) RNA was electrophoresed on a 1% formaldehyde agarose slab gel and transferred onto GeneScreen membranes. All blots, with the exception of the cardiac ␣-actin probe, were prehybridized for 2 h at 42°C in hybridization solution (50% formamide, 5ϫ SSC, 5ϫ Denhardt's, 1% SDS, 10% dextran sulfate, and 200 g/ml salmon sperm DNA) and hybridized with 2 ϫ 10 6 counts/ml of probe. Blots for cardiac ␣-actin were hybridized overnight with 2 ϫ 10 5 counts/ml at 65°C in the hybridization solution. The skeletal ␣-actin and myogenin cDNA probes were labeled by random priming using [␣-32 P]dCTP. Cardiac ␣-actin riboprobe was 32 P-labeled using an RNA transcription kit (Stratagene). Blots were washed at 68°C under high stringency conditions, 0.1ϫ SSC, and 0.1% SDS, and radioactivity was determined by phosphor-imaging analysis and visualized by autoradiography on x-ray film (Kodak).

RhoA Activates both Skeletal ␣-Actin and c-fos Promoters in
Muscle Cells-SRF has been shown to play a primary role in regulating the activity of the skeletal ␣-actin promoter, which contains three SREs at Ϫ90, Ϫ135, and Ϫ180 upstream of the transcriptional start site. Hill et al. (24) have shown that Rho GTPases regulate transcriptional activity of SRF in the context of the c-fos promoter in a manner that is independent of TCF in fibroblasts. To examine whether Rho GTPases are also involved in regulating skeletal ␣-actin gene transcription via SRF, we compared the effects of wild-type RhoA and constitutively active V14-RhoA on the SK-luc reporter plasmid with their effects on the c-fos-luc reporter plasmid in replicating C 2 C 12 myoblasts (Fig. 1). The V14 mutation in RhoA decreases the intrinsic GTPase activity and makes it unresponsive to GTPase activating proteins.
RhoA and V14-RhoA activated both the skeletal ␣-actin promoter and the c-fos promoter (Fig. 1, A and B). At the optimal transfection dose of 0.5 g, wild type RhoA activated both the skeletal ␣-actin promoter and the c-fos promoter to a similar extent, i.e. 6 -6.5-fold over basal level (as measured in cells transfected with parental vector), while V14-RhoA had a 7-fold activation effect on the skeletal ␣-actin promoter and a more modest 4.5-fold effect on the c-fos promoter. Western analysis showed that both RhoA and V14-RhoA proteins were expressed at a similar level in transfected myoblasts (Fig. 1C).
To investigate whether RhoA is required for skeletal ␣-actin promoter activity, we examined the effects of a RhoA dominant negative mutant (N19-RhoA) on skeletal ␣-actin promoter reporter activity. N19-RhoA acts by competitively inhibiting the interaction of endogenous RhoA with its exchange factors. Expression of N19-RhoA in passaged muscle cells resulted in repression of basal activity of the skeletal ␣-actin promoter by more than 40% at the transfection dose of 0.5 g (Fig. 1A). Western analysis showed that expression of N19-RhoA through the pCMV5 expression vector was similar to that of RhoA or V14-RhoA through pCGN vector (Fig. 1C). Together, these data indicate that RhoA activates both c-fos and skeletal ␣-actin promoters to a similar extent in myoblasts and that efficient activation of the skeletal ␣-actin promoter requires functional RhoA.
N19 myogenic differentiation. To answer this question, we stably transfected C 2 C 12 myoblasts with pSV2neo plus a 10-fold excess of either the empty vector pCMV5 or pCMV5-N19-RhoA. G418resistant colonies were pooled, switched into differentiation medium for 3 days, and examined for the morphological changes as well as biochemical markers of terminal differentiation.
The results shown in Fig. 2 are representative of several stable transfection experiments. C 2 C 12 cells transfected with pCMV-N19-RhoA stably expressed HA-N19-RhoA ( Fig. 2A). The morphological phenotypes of both control and C 2 C 12 -N19-RhoA cells were identical when the cells were allowed to replicate in the presence of serum (data not shown). However, 3 days after serum withdrawal, while the control cells fused and formed multinucleated myotubes, C 2 C 12 -N19-RhoA cells displayed significantly fewer and smaller myotubes (Fig. 2B). On the other hand, myoblast stably transfected with wild type RhoA were morphological indistinguishable from myoblasts transfected with the control vector (Fig. 2B). Thus, stable overexpression of a dominant negative form of RhoA appears to inhibit morphological differentiation of passaged muscle cells.
We then examined the effects of N19-RhoA on the expression of endogenous muscle-specific genes. Western analysis showed the induction of myogenin and ␣-MHC in the fusing control myoblasts following the withdrawal of growth media. In contrast, in C 2 C 12 -N19-RhoA cells, both myogenin and MHC protein levels were significantly reduced by more than 80% (Fig.  2C). In addition, Northern analysis revealed that the myogenic up-regulation of skeletal ␣-actin expression following serum withdrawal was also significantly reduced by more than 50% (Fig. 2D). Western analysis also showed that the SRF expres-sion level was reduced about 2-fold compared with control cells but to a lesser extent than other myogenic markers (Fig. 2C). Similar results were obtained for SRF binding activity to skeletal ␣-actin SRE1 oligonucleotide (data not shown). Thus, stable overexpression of a dominant negative form of RhoA obstructs biochemical differentiation of passaged myoblasts.
SRF Mediates RhoA Activation of the Skeletal ␣-Actin Promoter-To investigate how transcription factors are involved in mediating RhoA signaling on muscle-specific gene expression, we focused on SRF's role in mediating activation of the skeletal ␣-actin promoter by RhoA. We tested SRFpm1, a dominant negative SRF mutant, to inhibit RhoA-mediated skeletal ␣-actin promoter activation. The SRFpm1 mutant dimerizes with wild type SRF and interferes with the function of wild type SRF homodimers by forming DNA binding-defective heterodimers (19). Introduction of SRFpm1 by itself suppressed basal transcription of SK-luc, while RhoA-or V14-RhoA-mediated promoter activation was blocked by co-transfected SRFpm1 (Fig.  3A). On the other hand, overexpression of exogenous SRF in replicating C 2 C 12 myoblasts induced a modest activation of the skeletal ␣-actin promoter. Co-transfection of SRF with wild type RhoA or V14-RhoA appeared to have an additive effect (about 3-or 4-fold activation by SRF or RhoA and about 10-fold activation by co-transfection of SRF and RhoA) (Fig. 3B). Furthermore, co-transfection of N19-RhoA with SRF also resulted in inhibition of the effect of SRF in activating SK-luc (Fig. 3C). These results indicate that RhoA requires functional SRF to transduce its signal to the skeletal ␣-actin promoter.
All of the preceding experiments were performed in replicating myoblasts, in which skeletal ␣-actin gene expression is repressed. The expression of the skeletal ␣-actin, as well as an array of other myogenic factors, is markedly up-regulated when myoblasts undergo terminal differentiation into myotubes. Since RhoA activates the skeletal ␣-actin promoter through SRF even under replicating conditions, we asked if the effect of RhoA on the skeletal ␣-actin promoter would be enhanced in terminally differentiated myotubes. Compared with the 3-5fold activation levels of the skeletal ␣-actin promoter by wild type RhoA in replicating myoblasts, there was a 6 -10-fold activation of skeletal ␣-actin promoter activity in differentiated muscle cells (Fig. 3B). Interestingly, co-expression of exogenous SRF with RhoA had a marked synergistic effect on the skeletal ␣-actin promoter in differentiated myotubes, indicated by a 25-30-fold increase over basal levels. Thus, differentiation condition appears to potentiate activation of the skeletal ␣-actin promoter by RhoA activation in part through an increase in SRF expression in differentiated myotubes.
Deletion analysis of the skeletal ␣-actin promoter from Ϫ398 to ϩ25 bp has revealed that although there are three SREs in this region, the SRE closest to the transcriptional start region (SRE1) is necessary for activation of the skeletal ␣-actin promoter in response to extracellular stimuli (28). To test whether SRE1 is also sufficient to mediate the effect of RhoA, transfections were performed in replicating myoblasts using the SRE1 minimal promoter reporter plasmid, SK-SRE1-luc. This reporter was activated by RhoA, V14-RhoA, and SRF and was inhibited by SRFpm1 to a similar extent as with SK-luc (Fig.  4A), suggesting that SRE-1 is sufficient for activation of the skeletal ␣-actin promoter by RhoA.
In an attempt to determine which regions of SRF are involved in RhoA-dependent activation of SRF, we asked if RhoA activates Gal4-SRF-(1-508) and Gal4-SRF-(266 -508) fusion proteins. On these constructs, the DNA binding site of SRF has been replaced by that of Gal4 by fusing the DNA binding domain of Gal4 to different portion of the coding region of SRF (19). Gal4-SRF-(266 -508) contains the transcriptional activation domain of SRF and has been shown to be constitutively active (19). We found that Gal4-SRF-(1-508) and Gal4-SRF-(266 -508) activated the Gal4 reporter plasmid, G5-TATA-luc, about 2-and 30-fold, respectively, compared with the control plasmid containing only the Gal4 binding domain (data not shown). However, co-transfection with RhoA failed to further increase the activity of the reporter, suggesting that direct SRF FIG. 3. SRF mediates RhoA activation of the skeletal ␣-actin. A, C 2 C 12 myoblasts were transfected with 0.2 g of SK-luc together with 0.5 g of control vector or with 0.25 g each of the expression plasmids SRFpm1, RhoA, and V14-RhoA, alone or in combination as indicated. B, transfection conditions were the same as in A, except that SRF expression plasmid was substituted for SRFpm1 expression plasmid, and one set of the transfected cells were placed in differentiation medium instead of growth medium. C, transfection conditions were the same as in A, except that N19-RhoA expression plasmid was substituted for RhoA. Data were presented as described in the legend for Fig. 1.   FIG. 4. RhoA activates the SRE1 minimal promoter and the murine SRF promoter. A, C 2 C 12 cells were transfected with 0.2 g of SRE1-luc reporter plasmid together with 0.5 g of control vector plasmid or with 0.25 g each of the expression plasmids, SRF, SRFpm1, RhoA, and V14-RhoA, alone or in combination as indicated. B, C 2 C 12 cells were transfected with 0.5 g of SRF-luc reporter plasmid together with 0.5 g of control vector plasmid or varying amounts of RhoA, V14-RhoA, or N19-RhoA expression plasmids as indicated. Data were presented as described in the legend for Fig. 1. binding to DNA is required for its activation by RhoA and also that the C-terminal activation domain of SRF is not a direct target for the RhoA signal pathway.
RhoA Is Required for SRF Promoter Activity-We have recently described a positive role for SRF in regulating the activity of its own promoter, which contains two SREs (12). Since RhoA regulates activity of the skeletal ␣-actin promoter through SRF, we hypothesized that RhoA is also involved in regulation of SRF promoter activity. We compared the effects of RhoA, V14-RhoA, and N19-RhoA on the SRF-luc reporter plasmid, which contains the mouse SRF promoter region from Ϫ456 to Ϫ23 bp (12). As shown in Fig. 4B, RhoA and V14-RhoA activated, while N19-RhoA inhibited, SRF promoter activity to a similar extent as observed with the skeletal ␣-actin promoter. These results indicate that RhoA regulates the autoregulatory loop of SRF expression. Thus, the RhoA signaling pathway may regulate SRF-dependent transcription directly by regulating SRF activity and indirectly by increasing SRF expression.
SRFpm1 Blocked SRF DNA Binding Activity and Myogenesis-To determine the functional role of the RhoA target, SRF, during myogenic differentiation, the SRFpm1 mutant was stably transfected into the Sol8 and C 2 C 12 muscle cell lines. Under high serum conditions, the Sol8-SRPpm1a and C 2 C 12 -SRFpm1 cell lines appeared morphologically similar to proliferating wild type cells. However, under low serum conditions, both Sol8-SRFpm1a and C 2 C 12 -SRFpm1 cell lines were unable to fuse into multinucleated myotubes (Fig. 5A). Two individual clones, Sol8-SRFpm1a and Sol8-SRFpm1b, which showed the highest levels of SRFpm1 expression, were further examined (Fig. 5B). EMSAs were performed to determine if the expression of the SRFpm1 mutant inhibited the DNA binding activity of the endogenous SRF to a double-stranded skeletal ␣-actin SRE1 probe. EMSAs with increasing doses of whole cell extracts indicated that myoblasts expressing the SRFpm1 mutant showed at least a 60% reduction of SRF DNA binding activity over wild type cells (Fig. 5C).
We then examined the effects of the SRFpm1 mutant on the expression of endogenous skeletal and cardiac ␣-actin genes. SRFpm1a showed a 10-fold, and Sol8-SRFpm1b showed a 20fold, inhibition of endogenous cardiac ␣-actin mRNA as compared with wild type cells 2 days after the introduction of differentiation medium (Fig. 5D). In addition, the induction of ␣-MHC and myogenin genes that do not directly bind SRF was also markedly suppressed in SRFpm1 mutant cells (Fig. 5, D  and E). These results show that SRFpm1 obstructed a fundamental myogenic process, suggesting that SRF might be a critical early factor required for initiating myogenic differentiation and that RhoA signaling via SRF plays an important role in the control of myogenesis.  5. SRFpm1 inhibited myogenic  differentiation. A, Sol8 or C 2 C 12 myoblasts were co-transfected with SRFpm1 and a pSVneo vector. Cells were photographed 3 days following the addition of differentiation medium. B, expressed mutant SRF proteins from stable cell lines were detected by Western blotting with anti-HA antibody. C, endogenous SRF DNA binding activity was assayed by an EMSA, in which (0, 0.5, 1.0, 1.5, or 2.0 g/lane) cell extracts were incubated with the 32 P-labeled skeletal ␣-actin SRE1 double-stranded oligonucleotide DNA probe. Whole cell extracts were isolated 1 day after the switch to differentiation medium from wild type and Sol8-SRFpm1a cell lines. D, RNA was collected from wild type and mutant cells in growth (D0) and differentiation (D2) conditions. Expression of skeletal ␣-actin (␣Sk), cardiac ␣-actin (␣Ca), and myogenin (MyoG) was examined by Northern blot. 28 and 18 S ribosomal bands were visualized by ethidium bromide staining to show equal RNA loading. E, expression of ␣-MHC was detected by Western blot using the MF-20 monoclonal antibody.

DISCUSSION
The present study shows that RhoA is involved in the regulation of the promoter activity of the skeletal ␣-actin gene, since overexpression of wild type or activated RhoA activates the skeletal ␣-actin promoter in passaged replicating and differentiating murine myogenic cell lines. In addition, dominant negative RhoA reduces the activity of the skeletal ␣-actin promoter, and its stable overexpression in myoblasts hinders the formation of myotubes and expression of key muscle-specific proteins including skeletal ␣-actin, ␣-MHC, and regulatory factors such as the basic helix-loop-helix factor, myogenin, and SRF. Consistent with a recent report showing that inhibition of Rho family proteins, consisting of RhoA, Rac1, and Cdc42, by the GDP dissociation inhibitor, RhoGDI, suppresses myogenesis including expression of muscle specific genes in C 2 C 12 myoblast (27), our results further demonstrate that expression of RhoA potentiates muscle differentiation.
The present study provides several lines of evidence for a critical role of SRF in mediating RhoA activation of the skeletal ␣-actin promoter. RhoA-stimulated ␣-actin promoter activity is blocked by a dominant negative SRF mutant, SRFpm1, and is potentiated by the addition of wild type SRF. In addition, the proximal SRE1 in the skeletal ␣-actin promoter is sufficient to mediate the effect of RhoA and SRF. Moreover, RhoA activation of the skeletal ␣-actin promoter is increased under differentiation condition, which is associated with an increase in SRF expression level. Furthermore, RhoA also activates the SRF promoter, which contains two SRE sequences, thereby regulating levels of SRF transcripts. These observations indicate that RhoA regulates SRF-dependent transcription directly by regulating SRF activity and indirectly by increasing SRF expression.
Our study shows that RhoA is involved not only in expression of muscle-specific genes that require direct interaction with SRF on the SREs in their promoter region such as skeletal and cardiac ␣-actins but also in the expression of myogenin and ␣-MHC that do not directly bind SRF. Similarly, dominant negative mutant SRFpm1 inhibits both SRE-dependent and SRE-independent muscle-specific gene expressions, supporting a general role of SRF in mediating effects of RhoA in muscle differentiation. The inhibition of the endogenous expression of skeletal and cardiac ␣-actins by SRFpm1 was consistent with the hypothesis that SRF is an obligatory factor required for muscle-specific expression of the striated ␣-actin genes. Expression of these genes appears to require direct interaction with SRF on the SREs in the promoter region. SRF is preferentially expressed in embryonic heart, skeletal, and smooth muscle tissues, and increased SRF levels are detected during myogenic differentiation (12,13,15). The increased levels of SRF during myogenesis might saturate multiple SREs in the promoters of muscle-specific genes and displace negative acting factors such as YY1 (16). Also, multifactor protein complexes may enable SRF to induce the expression of other musclespecific genes that do not directly bind SRF. Recent experimental evidence has suggested that SRF can physically associate with myogenic basic helix-loop-helix proteins in vivo and in vitro (33). Association occurred at the respective MADS box and basic helix-loop-helix domains of the protein. The apparent necessity of SRF for expression of ␣-MHC and myogenin might be due to indirect mechanisms, resulting from the requirement for SRF protein-protein association during the myogenic differentiation process. Our observation that SRF is an important target of RhoA does not rule out the involvement of other transcription factors in mediating myogenic signaling of RhoA. Takano et al. (27) have suggested that MEF2 could mediate effects of Rho family proteins in muscle differentiation, since expression of MEF2 is inhibited in myoblasts overexpressing RhoGDI. Whether RhoA and/or SRF directly regulate transcriptional activity of MEF2 remains to be determined.
In contrast to the well documented Ras-dependent signal pathway leading to SRF activation, the precise mechanism by which RhoA regulates SRF transcriptional activity remains unclear. Rho GTPases are known to be involved in regulation of cytoskeletal organization in response to extracellular signals (34) and also in activation of the c-Jun N-terminal kinase and p38 kinase cascades (35)(36)(37). These two pathways play important roles in the regulation of gene transcription by Rho GT-Pases. However, in fibroblasts, RhoA-dependent activation of SRF appears to act through other mechanisms (24). Over the past few years, considerable progress has been made in identifying downstream effector molecules of RhoA. Of particular interest is a family of Rho-activated serine/threonine kinases (Rho kinase and its close relative p160ROCK) (38 -41). These kinases have been shown to be involved in RhoA-dependent stress fiber formation and focal adhesion formation (39,42). Chihara et al. (43) have recently reported that constitutively active Rho kinase stimulates the transcriptional activity of c-fos SRE. However, a recent study using RhoA effector loop mutants has shown that signaling to SRF by RhoA does not correlate with binding to any of the effectors tested including Rho kinases (44). We have recently found that myotonic dystrophy kinase, which has a significant homology to the Rho kinases, enhances the binding of SRF to the skeletal ␣-actin promoter in vitro and also activates skeletal ␣-actin transcription via SRF in synergy with RhoA, 2 suggesting that myotonic dystrophy kinase might be a downstream mediator of RhoA in SRF activation.
Upstream activators of RhoA include serum, lysophosphatidic acid, thrombin, and bombesin (34,45,46), which act through cell surface receptors coupled to heterotrimeric Gproteins. It is unlikely that either serum or lysophosphatidic acid would initiate RhoA-dependent myogenic signal pathways, since serum withdrawal potentiates RhoA-dependent activation of the skeletal ␣-actin promoter. One potential signaling pathway interacting with RhoA during muscle differentiation might be through the switching of integrin expression. It has been shown that ␤ 1 -integrin is required for terminal muscle differentiation (47) and that expression of ␤ 1 A and ␤ 1 D isoforms is developmentally regulated in skeletal muscle and heart, since the muscle-specific ␤ 1 D isoform appears following myoblast fusion, replacing the common ␤ 1 A isoform (48,49). In addition, RhoA appears to function both upstream and downstream of integrins in the context of focal adhesion formation and stress fiber assembly (reviewed in Ref. 50). In particular, activated RhoA has been shown to stimulate the activation of focal adhesion kinase, an important mediator of integrin signaling, which links to a number of signaling molecules including phosphatidylinositol 3-kinase (reviewed in Ref. 51). Recent studies have shown that phosphatidylinositol 3-kinase plays an important role in muscle differentiation (52) and is required for insulin-like growth factor-induced muscle differentiation (53). Moreover, phosphatidylinositol 3-kinase has been found to function downstream from RhoA in platelet and Swiss 3T3 cells (54,55). Further study will investigate the possible interactions between signal pathways mediated through RhoA, integrins, focal adhesion kinase, and phosphatidylinositol 3-kinase during muscle differentiation.
While the members of the myogenic basic helix-loop-helix family of transcription factors have long been regarded as critical regulators of myogenic induction, other factors, such as the MADS box proteins in the MEF2 family, have been shown to play key roles for full myogenic differentiation. A complex network regulated by members of different protein families has been hypothesized to induce the expression of E-box-dependent and -independent genes during myogenesis. The observations that expression of SRFpm1, as shown here, and injection of SRF antibodies into muscle cells (17) and SRF antisense oligonucleotides (18) block myogenic differentiation strongly suggest that SRF is an obligatory regulator of the myogenic process. Given the important role of SRF in the transcriptional regulation of muscle-specific genes and in terminal differentiation (12,13,17,18), the present study suggests that the effects of RhoA on myogenic differentiation are mediated in part through its direct effect on SRF transcriptional activity. By virtue of the inhibitory activity of the dominant negative RhoA, which was similar to the myogenic blocking activity of SR-Fpm1, our study supports the concept that the RhoA signaling pathway via SRF has a central obligatory role in muscle differentiation.