Transforming Growth Factor 1 Is Up-regulated by Activated Raf in Skeletal Myoblasts but Does Not Contribute to the Differentiation-defective Phenotype*

The Raf/MEK/MAPK signaling module elicits a strong negative impact on skeletal myogenesis that is reflected by a complete loss of muscle gene transcription and differentiation in multinucleated myocytes. Recent evidence indicates that Raf signaling also may contribute to myoblast cell cycle exit and cytoprotection. To further define the mechanisms by which Raf participates in cellular responses, a stable line of myoblasts expressing an estrogen receptor-Raf chimeric protein was created. The cells (23A2RafER) demonstrate a strict concentration-dependent increase in chimeric Raf protein synthesis and downstream phosphoMAPK activation. Initiation of low-level Raf activity in these cells augments contractile protein expression and myocyte fusion. By contrast, induction of high level Raf activity in 23A2RafER myoblasts inhibits the formation of myocytes and muscle reporter gene expression. Interestingly, treatment of myoblasts with conditioned medium isolated from Raf-repressive cells inhibits all of the aspects of myogenesis. Closer examination indicates that the transforming growth factor1 (TGF1) gene is up-regulated in Raf-repressive myoblasts. The cells also direct elevated levels of Smad transcriptional activity, suggesting the existence of a TGF1 autocrine loop. However, extinguishing the biological activity of TGF1 does not restore the myogenic program. Our results provide evidence for the involvement of Raf signal transmission during myocyte formation as well as during inhibition of myogenesis.

The Raf/MEK/MAPK signaling module elicits a strong negative impact on skeletal myogenesis that is reflected by a complete loss of muscle gene transcription and differentiation in multinucleated myocytes. Recent evidence indicates that Raf signaling also may contribute to myoblast cell cycle exit and cytoprotection. To further define the mechanisms by which Raf participates in cellular responses, a stable line of myoblasts expressing an estrogen receptor-Raf chimeric protein was created. The cells (23A2RafER DD ) demonstrate a strict concentration-dependent increase in chimeric Raf protein synthesis and downstream phos-phoMAPK activation. Initiation of low-level Raf activity in these cells augments contractile protein expression and myocyte fusion. By contrast, induction of high level Raf activity in 23A2RafER DD myoblasts inhibits the formation of myocytes and muscle reporter gene expression. Interestingly, treatment of myoblasts with conditioned medium isolated from Raf-repressive cells inhibits all of the aspects of myogenesis. Closer examination indicates that the transforming growth factor-␤ 1 (TGF-␤ 1 ) gene is up-regulated in Raf-repressive myoblasts. The cells also direct elevated levels of Smad transcriptional activity, suggesting the existence of a TGF-␤ 1 autocrine loop. However, extinguishing the biological activity of TGF-␤ 1 does not restore the myogenic program. Our results provide evidence for the involvement of Raf signal transmission during myocyte formation as well as during inhibition of myogenesis.
Myogenesis, the formation of contractile-competent skeletal muscle cells, is tightly regulated by the presence of several critical growth factors in the extracellular environment. As an example, local production of insulin-like growth factor-I (IGF-I) 1 is a strong permissive growth factor for both morphological and biochemical differentiation of myoblasts (1)(2)(3)(4). Ectopic IGF-I delivery to skeletal muscle cells increases muscle mass and morphometric measures of strength in birds and mice (5,6). On the contrary, several laboratories have documented the repressive effects of transforming growth factor ␤ 1 (TGF-␤ 1 ) and members of the fibroblast growth factor family, chiefly FGF2 (for review see Refs. 7 and 8). The mechanism by which these soluble inhibitory proteins impede skeletal myogenesis involves the activation of specific intracellular signaling cascades. In the case of TGF-␤ 1 , a requisite induction of Smad3 protein phosphorylation and nuclear translocation occurs that interferes with muscle gene transcription (9). The precise mechanism for FGF2-mediated inhibition of myogenesis is less understood but may involve downstream signaling events that are controlled through G-proteins (10).
Overexpression studies and application of specific kinase inhibitors has led to the identification of several essential signaling pathways in skeletal myoblasts. It is firmly established that phosphatidylinositol 3-kinase and p38, a stress-activated mitogen-activated protein kinase (MAPK), both are necessary for the formation of mature myocytes (11)(12)(13)(14)(15). Removal of downstream signaling events by the respective kinases through the use of chemical inhibitors abolishes myogenesis. By contrast, initiation of downstream signaling modules by constitutively active Ras, a membrane-localized GTPase that participates in numerous receptor tyrosine kinase-initiated signaling events, results in a severe reduction in muscle gene transcription and myoblast fusion (16,17). Activation of Raf, the downstream target of Ras, also abolishes the myogenic program (18 -21). Interestingly, the means by which Ras and Raf inhibit muscle gene expression does not appear to be directly dependent upon the subsequent phosphorylation and activation of extracellular signal-regulated kinase 1 and 2 (ERK1/2). Inhibition of MEK/ERK signal transduction does not reinstate the full complement of muscle gene transcription and myoblast fusion to myoblasts expressing constitutively active Ras or Raf (17,22).
Raf kinase is a serine/threonine kinase whose activation precludes ERK1/2 (MAPK) phosphorylation and a concomitant alteration in gene transcription (23,24). Signaling through the Raf/MEK/ERK pathway is a common mitogenic response for many cell types. Recently, several groups have reported differential cellular responses as a consequence of Raf/MAPK signal intensity. At lower levels of MAPK activity, epithelial and fibroblasts demonstrate increased proliferative rates (25,26). By contrast, high levels of Raf/MAPK result in cell cycle arrest and senescence (26,27). The Raf-mediated responses are attributed to signaling through MEK/MAPK as well as non-MEKdependent signaling events. Regulation of Raf activity indicates that the protein physically associates with other kinases, regulatory proteins, and scaffolding proteins (for review see Ref. 28). Indeed, it has become increasingly apparent that Raf kinase can control cellular transcriptional responses through mechanisms that are independent of the archetypical MEK/ ERK module. A Raf kinase allele that fails to interact with MEK and cause ERK1/2 phosphorylation retained the ability to activate NFB-directed transcription and promote neuronal differentiation (29). These results argue that Raf participates in several downstream signaling cascades. Differential utilization of these pathways may be reflected in the decision of the cell to undergo proliferation, differentiation, apoptosis, and senescence.
Because many growth factors exhibit contrasting effects on skeletal myogenesis yet utilize many of the same intracellular signaling pathways, it is likely that signal intensity plays a critical role in the decision to complete terminal differentiation. To this end, myoblasts that express an inducible activated Raf allele were created. Our results indicate that low level Raf activity promotes myogenesis, whereas high level Raf activity inhibits muscle formation. Coincident with repression of differentiation by activated Raf is an increase in TGF-␤ 1 gene expression and Smad-directed transcriptional activity. However, the removal of TGF-␤ 1 from the extracellular environment of the Raf-repressive myoblasts does not reinstate the differentiation program. In summary, Raf signaling modules are both positive and negative mediators of myogenesis that is a direct reflection of signal strength.

MATERIALS AND METHODS
Cell Culture, Plasmids, and Transfections-23A2RafER DD myoblasts were created by transduction of 23A2 myoblasts with a retrovirus encoding the fusion protein RafER DD (30). RafER DD is comprised of the kinase domain of human c-Raf-1 fused in-frame with the estrogen receptor ligand binding domain. Following infection, the cells were selected in puromycin and clones were isolated by limiting dilution. 23A2 and 23A2RafER DD myoblasts were cultured on gelatinized tissue culture grade plasticware in Dulbecco's modified Eagle medium containing 15% fetal bovine serum (BioWhittaker), 1% penicillin-streptomycin, and 0.5% Geneticin (Invitrogen). Differentiation was induced in confluent cultures by continuous culture in Dulbecco's modified Eagle medium supplemented with 2% horse serum, 1% penicillin-streptomycin, and 0.5% Geneticin. For the measurement of muscle-specific reporter gene activity, myogenic cells cultured in 6-well tissue plasticware were transiently transfected with 1 g of troponin I luciferase (TnI-Luc), the minimal E-box reporter plasmid, 4Rtk-Luc, or a multimerized AP-1 binding site reporter (AP-1-Luc) and 50 ng of pRL-tk, a Renilla luciferase plasmid as a monitor of transfection efficiency, using standard calcium phosphate methods (16,22). The cells were maintained in differentiation media for 48 h in the presence or absence of varying concentrations of 4-hydroxytamoxifen (4HT, Sigma) prior to lysis and measurement of luciferase activities (Dual-luciferase, Promega, Madison, WI). Inhibition of MEK activity was achieved by supplementation of differentiation medium with 20 M PD98059. The amount of corrected luciferase activity generated by 23A2 or 23A2RafER DD myoblasts treated with vehicle was set to 100%. Each experiment was replicated a minimum of three times. For conditioned medium experiments, the culture medium was removed from myoblasts treated for 48 h with Me 2 SO or varying amounts of 4HT. The medium was centrifuged to remove debris and brought to a final concentration of 2% horse serum. The resulting medium was applied to confluent 23A2 myoblasts, and the cells were maintained for an additional 48 h prior to fixation. Heat inactivation of the conditioned medium was performed as described previously (31).
RNA Isolation, Northern Blots, and RT-PCR-Total RNA was isolated with STAT60-denaturing solution per the manufacturer's recommendations (Tel-Test, Friendswood, TX). Twenty micrograms of total RNA were separated through 1% agarose gels containing 2% formaldehyde, transferred to nitrocellulose membrane, and irreversibly crosslinked to the medium by UV. The blots were hybridized with 32 P-labeled cDNA probes generated by random hexamer priming (DecaPrime, Ambion, Houston, TX). The probes corresponded to regions contained within the TGF-␤ 1 and glyceraldehydes 3-phosphate dehydrogenase (GAPDH) gene-coding regions. Northern blot analysis was performed using Ultra-Hyb (Ambion) at 42°C overnight according to the manufacturer's recommendations. Blots were washed with 2ϫ SSC, 0.1% SDS once at room temperature and twice with 1ϫ SSC, 0.1% SDS at 42°C. Blots were exposed to phosphorimaging screens for the visualization and quantification of message levels. For the semi-quantitative assessment of mRNA levels, 1 g of total RNA was reverse-transcribed with Moloney murine leukemia virus reverse transcriptase. Amplification of GAPDH and TGF-␤ 1 was performed using Taq polymerase (Fisher Scientific, Pittsburgh, PA), gene-specific primers, and thermocycle conditions of 95°C for 45 s, 55°C for 45 s, and 72°C for 90 s for a total of 35 cycles. To ensure that quantification was accomplished in the linear range of amplification, an aliquot was removed from each reaction after 25, 30, and 35 cycles. GDF-8 was amplified as described previously (32). Amplicons were separated through 2% agarose gels containing ethidium bromide, visualized under UV, and photographed.
Western Blot and Scanning Densitometry-23A2RafER DD myoblasts were differentiated in the presence or absence of 4HT. After 48 h, the cells were lysed in 4ϫ SDS-PAGE sample buffer and an aliquot was removed for protein quantification. Equal amounts of protein were electrophoretically separated through denaturing gels and transferred to nitrocellulose membrane. The blots were incubated with 5% nonfat dry milk in TBST (10 mM Tris, pH 8.0, 150 mM NaCl, 0.1% Tween 20) to remove nonspecific binding sites. Primary antibodies were diluted in blocking buffer, and the blots were incubated overnight at 4°C with shaking. Antibodies and dilutions included the following: anti-myogenin (F5D ascites, Developmental Hybridoma Bank, University of Iowa, 1:5000); anti-myosin heavy chain (MF20 hybridoma supernatant, Developmental Hybridoma Bank, 1:5); anti-ERK1/2 and anti-phos-phoERK1/2 (Cell Signaling); and anti-estrogen receptor (Santa Cruz Biotechnology, Santa Cruz, CA, 1:300). After extensive washing with TBST, the blots were reacted with the appropriate peroxidase secondary antibody for 45 min at room temperature. Visualization of protein bands was accomplished by chemiluminescence and autoradiography. Multiple exposures to x-ray film were used to ensure that the linear range of densitometry was maintained. Autoradiograms were scanned on a Storm 860 phosphorimaging system (Molecular Dynamics, Amersham Biosciences).
Immunocytochemistry-23A2 and 23A2RafER DD myocytes were fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS) for 10 min at room temperature. The cells were washed with PBS, and nonspecific antigen sites were blocked by incubation with 5% horse serum in PBS containing 0.1% Triton X-100 for 20 min at room temperature. Cultures were incubated with anti-myosin heavy chain (MF20) for 60 min at room temperature. After exhaustive washing with PBS, the cells were reacted with donkey anti-mouse fluorescein isothiocyanate (Vector Laboratories, Burlingame, CA, 1:200) for 45 min at room temperature. Cultures were washed with PBS and counterstained with 4,6-diamidino-2-phenylindole (DAPI). Immunofluorescent detection was accomplished using a Nikon TE200 inverted phase microscope equipped with epifluorescence. Representative photomicrographs were captured to slide film and assembled in Adobe Photoshop.

Creation and Characterization of an Inducible Raf Myogenic
Cell Line-Previous work has clearly demonstrated that overexpression of activated alleles of Raf are inhibitory to myocyte formation and muscle gene expression (17,18,21). More importantly, MAPK-dependent repression appears to be a function of Raf signaling intensity in avian myoblasts (18). To extend these observations, an activated Raf kinase that is responsive to the estrogen analog, 4-hydroxytamoxifen (4HT), was stably expressed in 23A2 mouse myoblasts (23A2RafER DD ). Confluent cultures of 23A2RafER DD myoblasts were differentiated in the presence or absence of increasing concentrations of 4HT. After 48 h, the cells were lysed for Western analysis of chimeric Raf (␣-estrogen receptor), activated MAPK (␣-phosphoMAPK), and total MAPK (␣-MAPK) expression. As shown in Fig. 1, a dosedependent increase in the relative amount of RafER DD protein is found with amounts 4HT from 0.25 nM to 2.5 M (lanes 2-6). Coincident with the increase in chimeric Raf protein is an elevation in the amounts of phosphorylated MAPK. Phosphorylated ERK2 is preferentially activated at 4HT concentrations as low as 0.25 nM, whereas activated ERK1 is first apparent at 4HT concentrations of 25 nM or greater. No differences in the amounts of total MAPK (ERK1/2) were evident.
Activation of ERK1/2 signal transmission leads to an increase in AP-1-directed transcriptional activity in many cell types including skeletal myoblasts. To ensure that the Raf/ ERK pathway was functional in 23A2RafER DD myoblasts, semiconfluent cells were transiently transfected with a multimerized AP-1 reporter plasmid and pRL-tk, a plasmid encoding Renilla luciferase as a marker of transfection efficiency. Subsequently, the cells were treated with increasing amounts of 4HT for 48 h prior to lysis and measurement of luciferase activities. Control 23A2 myoblasts were treated in an analogous fashion. AP-1-Luc reporter activity was normalized to Renilla luciferase activity. Results demonstrate a dose-dependent increase in AP-1-Luc activity with increasing amounts of 4HT (Fig. 2). At concentrations of 4HT greater than 250 nM, a decline in AP-1 luciferase activity was evident. Moreover, the response is directly attributed to the induction of RafER activity as 23A2 myoblasts failed to initiate AP-1-transcriptional activity upon treatment with 4HT. These results indicate that the Raf/ERK signaling axis is intact and functional in the myogenic cells.
As final confirmation of the integrity of the Raf signaling system in 23A2RafER DD myoblasts, confluent cultures of cells were treated for 48 h in differentiation medium containing vehicle only (Me 2 SO) or 1 M 4HT. Subsequently, cells were fixed and immunostained for myosin heavy chain (␣-MyHC). Control 23A2RafER DD myocytes treated with vehicle readily differentiate into large multinucleated myofibers that express copious amounts of the contractile protein, MyHC (Fig. 3). By contrast, activation of the Raf/ERK signaling pathway leads to a severe reduction in both myofiber number and MyHC protein expression. These results directly reflect published reports of the effects of activated Raf on skeletal myogenesis. Thus, 23A2RafER DD myoblasts represent a myogenic cell line that retains its muscle features in the absence of Raf activity and is differentiationdefective in the presence of Raf signal transmission.
Contrasting Effects of Raf Signal Transduction on Skeletal Myoblasts-To understand the effects of signal intensity of myogenesis, 23A2RafER DD myoblasts were differentiated in the presence of increasing amount of 4HT. After 48 h, the cells were lysed and equal amounts of total cellular proteins were analyzed by Western blot for the expression of muscle proteins (anti-MyHC, anti-myogenin). Scanning densitometry was performed on the chemiluminescent autoradiograms. Muscle protein expression was normalized to total ERK2 expression, an internal marker for protein loading, transfer, and detection efficiency. In the absence of 4HT, 23A2RafER DD myoblasts differentiate into myocytes that express both MyHC and myo-genin (Fig. 4). Unexpectedly, low level induction of Raf activity using concentrations of 4HT (0.25 or 2.5 nM) causes an increase in the amount of contractile and regulatory protein expression. Supplementation of differentiation medium with 25 nM 4HT or greater causes the predicted loss of MyHC and myogenin protein expression. The biphasic response of 23A2RafER DD myoblasts to increasing 4HT concentrations is further reflected in muscle-specific reporter gene activity. 23A2 and 23A2RafER DD myoblasts were transiently transfected with TnI-Luc and pRLtk. After 48 h in differentiation medium supplemented with 4HT, the cells were lysed and luciferase activities were measured. TnI luciferase enzymatic activity did not vary as a function of 4HT treatment in control 23A2 myoblasts, indicating that the estrogen analog does not significantly alter basal muscle gene expression (Fig. 5). The levels of Raf activity directed in response to 25 nM 4HT or greater are sufficient for a reduction in TnI-Luc activity. However, low level Raf activity (0.25-2.5 nM 4HT) directs only a slight increase in muscle reporter gene activity. It is likely that a larger portion of the TnI regulatory region is necessary to obtain a substantial increase in reporter gene transcription. Our results clearly demonstrate that Raf has contrasting effects on muscle formation that are intensity-dependent.
Previously, we have documented the existence of MEK-independent signaling in response to activated Raf (18,33). To determine the role of MEK in the RafER-imposed block to myogenesis, confluent 23A2RafER DD myoblasts were transiently transfected with TnI-Luc and pRL-tk. The myoblasts were treated for 48 h with increasing amounts of 4HT and 20 M PD98059 or vehicle only. Cell lysates were collected, and luciferase activities were measured. The amount of muscle reporter gene activity directed in the absence of 4HT and the chemical MEK inhibitor was set to 1. As expected, the selected concentrations of 4HT caused a reduction in the amount of TnI-Luc reporter activation (Fig. 6). Suppression of MEK function restored muscle gene transcription as evidenced by an increase in TnI-Luc activity to levels comparable to controls. However, strong stimulation of Raf signaling initiated by 2.5 M 4HT was refractile to the effects of the MEK inhibitor. Western analysis of replica samples demonstrates that treatment with PD98059 reinstates myosin heavy chain expression to levels Ͻ10% of controls (data not shown). These results duplicate previous observations using constitutive Raf alleles of various signaling intensities (21,33).

Raf Inhibition of Myogenesis Involves Secretion of a Soluble
Factor-It has been reported that the inability of myoblasts to differentiate in the presence of activated Raf is a product of sequestration of MEF2 in the cytoplasm (21). However, forced expression of the transcription factor in avian myoblasts does not reinstate the muscle gene program, arguing that additional factors are involved (33). Moreover, rhabdomyosarcoma cells secrete bioactive TGF-␤ 1 that acts in an autocrine manner to suppress differentiation (34). To further characterize the differentiation-defective phenotype of myoblasts directing extreme levels of Raf/ERK signal, conditioned medium was collected from 23A2RafER DD myoblasts that were treated with Me 2 SO, 2.5 nM 4HT, or 2.5 M 4HT in differentiation medium. Cell debris was removed by centrifugation, and the supernatants were supplemented with horse serum to a final concentration of 2%. An aliquot of conditioned medium isolated from the 2.5 M of treatment group was heat-inactivated prior to supplementation with horse serum (31). Confluent cultures of 23A2 myoblasts were allowed to differentiate for 48 h in the conditioned medium followed by fixation and immunostaining for MyHC. Control cultures were treated with Me 2 SO, 2.5 nM 4HT, or 2.5 M 4HT and analyzed as described. Conditioned medium from 23A2RafER DD myoblasts differentiated in the presence or absence of 2.5 nM 4HT did not alter the ability of myoblasts to fuse into MyHC-positive myofibers (Fig. 7). By contrast, medium harvested from 23A2RafER DD myoblasts treated with 2.5 M 4HT dramatically inhibited myocyte formation. Ͻ1% of the myoblasts were able to fuse into myosinexpressing myofibers. Control myoblasts were not affected by 2.5 M 4HT to the same extent. The complete absence of differentiated cells strongly argues that a soluble factor is secreted from Raf-expressing myoblasts that contributes to the loss of myofiber formation. Moreover, the secreted factor probably is a protein because heat-denatured conditioned medium is unable to inhibit 23A2 myocyte formation.
TGF-␤ 1 Does Not Contribute to the Raf-imposed Block to Myogenesis-Differentiation-defective rhabdomyosarcoma cells secrete TGF-␤ 1 into the culture medium that contributes to the loss of myogenic potential. To determine whether a TGF-␤-like factor may be responsible for the inhibition of muscle formation in myoblasts treated with Raf-repressive conditioned medium, 23A2 myoblasts were transiently transfected with 3TP-Lux, a reporter plasmid whose activity is modulated by TGF-␤ proteins, and pRL-tk. Cells were treated for 48 h in conditioned medium harvested from 23A2RafER DD myoblasts that were treated with Me 2 SO, 2.5 nM 4HT, or 2.5 M 4HT. Subsequently, the cells were lysed and luciferase activities were measured. Myoblasts that are capable of fusing and expressing contractile proteins (Me 2 SO, 2.5 nM 4HT) do not direct a significant amount of 3TP-Lux activity (Fig. 8). However, myoblasts that are unable to express the differentiated phenotype demonstrate a 6-fold increase in TGF-␤-responsive re- FIG. 4. Low-level Raf activity enhances skeletal myogenesis. 23A2RafER DD myoblasts were cultured in differentiation medium supplemented with increasing amounts of 4HT. After 48 h, total cellular lysates were prepared and analyzed by Western blot for MyHC, myogenin, and ERK1/2 protein expression. Relative amounts of proteins were quantified by scanning densitometry. MyHC and myogenin protein amounts were normalized to the amount of ERK2 protein. Mean Ϯ S.E. were Ͻ5% for each measurement.

FIG. 5. Raf activation alters two distinct phases of myogenesis.
23A2 and 23A2RafER DD myoblasts transiently transfected with TnI-Luc and pRL-tk were cultured in differentiation medium supplemented with increasing concentrations of 4HT. TnI-Luc was normalized to Renilla luciferase activity. The amount of reporter activity directed by 23A2 myocytes in the absence of 4HT was set to 100%. Low level Raf activity causes a slight increase in muscle reporter gene activity, whereas high level Raf activity inhibits myogenesis. Data represent the mean Ϯ S.E. of at least three independent experiments. RLU, relative light units. porter gene function. Therefore, we conclude that a TGF-␤-like factor is present in the medium harvested from 23A2RafER DD myoblasts transmitting high level Raf signals.
To determine whether the soluble factor is TGF-␤ 1 , genespecific primers for the growth factor were designed for the amplification of messages by RT-PCR. In brief, total RNA was harvested from 23A2 and 23A2RafER DD myoblasts treated for 48 h in differentiation supplemented with 2.5 M 4HT. Equal amounts of RNA were reversed-transcribed and amplified using primers specific for the TGF-␤ 1 and GAPDH. The cDNAs were amplified with Taq polymerase, and an aliquot of each reaction was removed after 25, 30, and 35 cycles (see "Materials and Methods" for conditions). Reaction products were separated through 2% agarose gels containing ethidium bromide. By semi-quantitative PCR, a cycle-dependent increase in GAPDH was evident in both control and Raf-repressive samples (Fig. 9). No amplicons for TGF-␤ 1 were detected in RNA isolated from control myocytes, but gene products for the growth factor were evident after 30 and 35 cycles in 23A2RafER DD myoblasts treated with 4HT (Raf-repressive) RNA samples. Northern blot analysis confirmed the increase in TGF-␤ 1 mRNA in Raf-expressing myoblasts (Fig. 9B).
Our results indicate that TGF-␤ 1 , a potent inhibitor of muscle differentiation, is produced by 23A2RafER DD myoblasts exhibiting extreme levels of Raf signaling and propose that this protein may be responsible for the block to muscle formation. Therefore, removal of secreted TGF-␤ may reverse the inhibi-tion of myogenesis in myoblasts with high level Raf signal intensity. To this end, cultures of 23A2RafER DD myoblasts treated with Me 2 SO or 500 nM 4HT were incubated for 48 h in differentiation medium supplemented with a soluble inhibitor of TGF-␤. The TGF-␤ inhibitor (␤SRI) is a chimera of the TGF-␤ receptor ligand binding domain fused to human Fc portion of IgG that prevents TGF-␤ 1 , TGF-␤ 2 , and TGF-␤ 3 from interacting with cell surface receptors (36,37). After 48 h, the cells were fixed and immunostained for MyHC and numbers of immunopositive cells were measured. 23A2RafER DD myoblasts in the absence of 4HT readily fuse into large multinucleated, MyHC-immunopositive fibers (Fig. 10). Treatment of the cells with TGF-␤ 1 (2 ng/ml) suppressed myocyte formation to Ͻ25% of control, indicating that the cells respond to the growth factor in a manner analogous to parental 23A2 skeletal myocytes. This level of inhibition also is attained by treatment of 23A2RafER DD myoblasts with 500 nM 4HT. However, supplementation of the culture medium with ␤SRI at a concentration that effectively restores myogenesis to TGF-␤ 1 -inhibited myoblasts does not reinstate the differentiation program to 23A2RafER DD myoblasts directing elevated Raf signaling. Thus, we conclude that TGF-␤ 1 inhibits the differentiation of 23A2RafER DD myoblasts and that the cells produce TGF-␤ 1 in response to elevated Raf but the establishment of an autocrine TGF-␤ 1 loop in these cells does not significantly contribute to the differentiation-defective phenotype-imposed Raf signaling.
GDF-8 Is Up-regulated in Response to Activated Raf and Can Inhibit Myogenesis-GDF-8, a member of the TGF-␤ superfamily, inhibits myogenesis through a Smad-dependent signaling pathway (38,39). Total RNA was isolated from 23A2 and 23A2RafER DD myoblasts treated for 48 h in differentiation media supplemented with 1 M 4HT and analyzed by RT-PCR for GDF-8, GAPDH, and RafER DD mRNA expression. As shown in Fig. 11, 23A2 myofibers do not contain detectable amounts of GDF-8 mRNA. By contrast, message levels for the gene are up-regulated in differentiation-defective 23A2RafER DD myoblasts, which also express abundant levels of RafER DD mRNA. No differences were found in GAPDH mRNA expression between control and Raf-repressive myoblasts. Thus, the increase in GDF-8 gene expression may contribute to the Raf-imposed block to myogenesis.

DISCUSSION
Activated Raf signal transmission represents a powerful disruptor of normal myocyte formation and muscle gene expression. The exact mechanism by which activated Raf alleles accomplish this feat remains largely unknown. Several contributing factors to the loss of myogenic capacity include inhibition of myogenin gene expression and cytoplasmic sequestration of MEF2 protein (21,33). Recently, we have demonstrated that forms of activated Raf that initiate activation of differential downstream signaling intermediates cause varying responses on myogenin gene expression that may be correlated to overall signal strength (33). 23A2RafER DD myoblasts were created to provide additional insight into the effects of Raf signal intensity on myocyte formation. In the absence of Raf activity, the cells readily form muscle in a manner analogous to parental 23A2 myoblasts. Interestingly, the introduction of a low level Raf-mediated signal stimulates an increase in the numbers of myocytes formed and enhances their ability to transcribe muscle genes. This is an intriguing finding in light of the fact that medium and high level Raf activities lead to a reduction in markers of muscle differentiation. Using stable clones of mouse myoblasts that direct differing levels of Raf kinase activity, DeChant et al. (40) reported that suboptimal Raf signal transmission leads to an increase in muscle reporter gene transcription and myocyte formation. The increase in myogenic capacity is not a reflection of diminished apoptosis because the low level Raf clone and parental myoblasts did not differ in the extent of DNA fragmentation. Therefore, it is likely that enhanced muscle formation in response to nominal Rafmediated signaling occurs independent of enhanced cell survival. Others have suggested that Raf inhibits myogenesis by maintaining the cell in a proliferative state (20). L6 myoblasts expressing an estrogen-inducible allele of activated Raf fail to express muscle genes or fuse into myocytes upon initiation of Raf signaling. In addition, these cells demonstrate an increased mitotic index in response to the kinase. 23A2RafER DD myoblasts and primary avian myoblasts expressing activated Raf are refractile to the mitogenic effects of Raf/MAPK (data not shown) (18). These contrasting results may reflect differences in the absolute amounts of Raf signaling. More importantly, these differences in myogenic responses directed by Raf signaling may explain the opposing effects of various growth factors on skeletal myogenesis. For example, both IGF-I and FGF2 initiate signaling through the Raf/MEK/ERK pathway, yet IGF-I promotes differentiation and FGF2 inhibits myocyte formation. Its possible that low level stimulation of ERK via Raf leads to the positive effects of IGF-I while a strong sustained ERK function may contribute to repression of myogenesis.
The ability of Raf signaling to inhibit myogenesis occurs through both MEK-dependent and independent signaling events. Low to moderate activation of Raf suppresses muscle gene transcription and myocyte formation, but these repressive effects are countered by inclusion of a chemical MEK inhibitor. Interestingly, severe amounts of Raf activity produce a condition that is refractile to the MEK inhibitor, suggesting the existence of a MEK-independent function. The MEK-independent signaling probably involves downstream kinases or phosphatases and is not a product of altered gene transcription because the addition of PD98059 was coincident with 4HT. Hence, Raf signaling was initiated simultaneously with MEK inhibition. The existence of Raf-controlled, MEK-independent signaling pathways have been described in fibroblasts recently (29). A constitutive Raf mutant that fails to associate with MEK and activate ERK1/2 can duplicate some of the effects of activated Raf including NFB transcriptional activation. Preliminary studies in our laboratory demonstrate that this Raf mutant abolishes the transcriptional activation properties of MyoD. 2 Thus, Raf-initiated, MEK-independent signaling pathways are present in myogenic cells and probably participate in the suppression of skeletal myogenesis.
High intensity Raf signaling causes cell cycle arrest and senescence in epithelial and fibroblast cells (25,27,30,41). Coincident with mitotic arrest, Raf initiates an increase in TGF-␤ 1 gene expression, leading to the creation of an autocrine loop that directly participates in the growth inhibition (42). Similarly, TGF-␤ I is up-regulated in rhabdomyosarcoma cells and may play a role in the block to myogenic differentiation (34). Our results indicate that elevated Raf signaling does elicit an increase in TGF-␤ 1 gene expression; however, this negative regulator of myogenesis does not contribute significantly to the differentiation-defective phenotype of Raf-expressing myoblasts. 23A2RafER DD myoblasts are inhibited by TGF-␤ 1 to the same extent as parental cells, arguing that the Raf-expressing cells are not refractile to the inhibitory actions of the growth factor and that the cells synthesize a functional TGF-␤ receptor. It is possible that TGF-␤ is secreted and retained in the extracellular matrix of the muscle cells, thus leaving it unavailable for sequestration by the soluble TGF-␤ inhibitor. This scenario is doubtful as retention of the TGF-␤ 1 by the extracel-2 X. Wang and S. E. Johnson, unpublished data. Total RNA was isolated and reverse-transcribed. PCR amplification was performed using RafER DD , GAPDH, and GDF-8 gene-specific primers. Five microliters of the amplification reaction (50 l total) was separated through a 2% agarose gel impregnated with ethidium bromide. lular matrix would inactivate the growth factor and would lend itself to the promotion of myogenesis, a condition contradictory to those observed. Thus, it is most likely that TGF-␤ 1 production in response to activated Raf does not play a significant role in the suppression of muscle formation.
Treatment of parental myoblasts with conditioned medium from Raf-repressive myoblasts causes a significant increase in the amount of Smad-driven reporter gene transcription. This finding endorses the synthesis and release of a TGF-␤-like soluble factor by Raf-expressing cells. The identity of the factor remains unknown. Of the TGF-␤ superfamily members, GDF-8 (also referred to as myostatin) represents a strong candidate for the putative secreted factor. GDF-8 myoblast differentiation in vitro and animals lacking or carrying a dysfunctional GDF-8 gene exhibit myocyte hypertrophy (38,(43)(44)(45)(46). The growth factor signals through the activin receptor, a membrane-bound receptor that transmits information through the Smad proteins (39,47). Previously, our group reported an inhibition of DNA synthesis in avian myoblasts transduced with a highly active allele of Raf (18). GDF-8 not only inhibits myogenesis, but the growth factor also causes a noticeable reduction in proliferation rates in myoblasts (35,48). These two pieces of information lend support to the notion that GDF-8 acts through an autocrine loop to synergize with Raf signaling and more effectively inhibit myogenesis. However, 23A2RafER DD myoblasts treated with 1 M 4HT fail to display a mitotic index that differs significantly from control myoblasts (8 versus 13%, respectively). Moreover, the concentrations of GDF-8 required to inhibit differentiation of 23A2 myoblasts is ϳ10-fold higher than the obligatory concentration of TGF-␤ 1 (data not shown). These results suggest that GDF-8 is not as effective as TGF-␤ 1 for suppression of the myogenic program. Alternatively, it is possible that a novel TGF-␤ family member contributes to the Raf-induced block to muscle formation.
In summary, Raf signal transmission directs contrasting effects on skeletal myoblasts that is commensurate with intensity level. Low level Raf/MAPK signaling elicits a positive effect on myogenesis that is reflected by enhanced regulatory and contractile protein synthesis. By contrast, sustained and extreme levels of Raf signaling leads to the prototypical reduction in myocyte formation and muscle protein expression. Coincident with repression of myogenesis by activated Raf is an increased expression of TGF-␤ 1 and GDF-8 and release of a soluble TGF-␤-like factor into the medium. Although the identity of the secreted protein remains unknown, it likely is not TGF-␤ 1 because incubation of Raf-repressive myoblasts with an inhibitor of the growth factor does not reinstate the differentiation program.