The myogenic basic helix-loop-helix family of transcription factors shows similar requirements for SWI/SNF chromatin remodeling enzymes during muscle differentiation in culture.

The myogenic basic helix-loop-helix family of transcription factors, MyoD, Myf5, myogenin, and MRF4, can each activate the muscle differentiation program when ectopically expressed in non-muscle cells. SWI/SNF complexes are ATP-dependent chromatin remodeling enzymes. We demonstrated previously that SWI/SNF enzymes promote MyoD-mediated muscle differentiation. To ascertain the requirement for SWI/SNF enzymes in muscle differentiation mediated by different MyoD family members, we examined MyoD, Myf5, MRF4, and myogenin-mediated induction of muscle differentiation in cells expressing dominant negative versions of BRG1 or BRM-based SWI/SNF enzymes. We demonstrated that expression of dominant negative BRG1 or BRM inhibited the induction of muscle-specific gene expression by Myf5 and MRF4; however, myogenin failed to induce measurable quantities of muscle-specific mRNAs, even in cells not expressing dominant negative SWI/SNF. In contrast, all four myogenic regulators induced expression of the cell cycle regulators p21, Rb, and cyclin D3 and promoted cell cycle arrest independently of the SWI/SNF enzymes. We proposed that SWI/SNF enzymes are required for the induction of all muscle-specific gene expression by MyoD, Myf5, and MRF4, whereas induction of the cell cycle regulators, p21, Rb, and cyclin D3 occurred independently of SWI/SNF function.

Mammalian SWI/SNF enzymes were first described as multiprotein complexes that could alter nucleosome structure in an ATP-dependent manner and facilitate the binding of transcriptional activators as well as the TATA-binding protein to nucleosomal DNA (1)(2)(3). Subsequent characterization of the mechanisms that generate changes in nucleosome and chromatin structure has revealed that these enzymes are capable of altering histone-DNA contacts within the nucleosome to generate an altered structure in which the DNA component shows increased accessibility to nucleases and DNA-binding proteins (1)(2)(3)(4)(5)(6). Additionally, SWI/SNF enzymes have the ability to mobilize nucleosomes such that the histone octamer "slides" along the DNA, in effect permitting re-positioning of the nucleosome (7,8).
At least two distinct SWI/SNF enzymes exist in mammalian cells. They contain 8 -12 subunits, many of which are shared by the different forms of the enzyme; however, they differ in the ATPase that acts as the catalytic core (3). Two highly related ATPases, BRG1 and BRM, have been identified (9 -11), and it has been shown that the chromatin remodeling activities of the enzymes can be accomplished, albeit less efficiently, by recombinant BRG1 or BRM alone (6). The functions of the remaining subunits are not well defined. Interestingly, both BRG1 and a 46-kDa subunit termed ini1 are missing in some human tumors and tumor cell lines (12)(13)(14)(15)(16)(17)(18)(19). Inactivation of the Brg1 and Ini1 genes in mice confirmed that these subunits act as tumor suppressors, as mice heterozygous for each gene developed a range of tumors (20 -23). Further insight into the role of these SWI/SNF subunits in growth and development was precluded, however, as nullizygous embryos die peri-implantation (20 -23).
Cellular responses to diverse signaling events often require the activation of genes that were previously maintained in an inactive state. Factors that can rearrange or alter chromatin structure may therefore be required for such activation events. In fact, recent work has indicated that SWI/SNF enzymes or individual subunits of these enzymes contribute to the activation of new programs of gene expression during the induction of cellular differentiation pathways. To date, a role for SWI/SNF has been implicated in the initiation of erythroid (24 -28), myeloid (29), macrophage (30), myogenic (31,32), and adipogenic (33,34) specific gene expression. In the case of muscle differentiation, MyoD-mediated conversion of fibroblasts to musclelike cells and induction of muscle-specific gene expression was abrogated by prior expression of ATPase-deficient alleles of BRG1 or BRM, which resulted in the formation of dominant negative SWI/SNF enzymes (31,35). Lack of induction of muscle-specific genes was correlated with a significant reduction in chromatin remodeling at an endogenous muscle-specific promoter in cells expressing dominant negative SWI/SNF enzymes (31). Thus SWI/SNF enzymes are essential for MyoD-mediated muscle differentiation.
Initiation of muscle-specific gene expression during differentiation of muscle cells is accompanied by withdrawal of the cells from the cell cycle in a coordinated but temporally separable manner (36,37). Despite the requirement for SWI/SNF enzymes in activation of muscle-specific gene expression, MyoDmediated induction of several cell cycle regulators was unaffected by the absence of functional SWI/SNF enzymes (32). This suggested that the chromatin remodeling activities of the SWI/SNF enzymes are differentially required during muscle differentiation and that the mechanisms by which MyoD activates muscle-specific genes and up-regulates cell cycle regulatory proteins are distinct.
MyoD is a basic helix-loop-helix transcription factor that interacts with the ubiquitous E box-binding proteins E12, E47, and HEB to form functional heterodimers (38 -42). These bind to E box sequences frequently found in promoters induced during myogenesis and promote gene expression (40 -42). Three other MyoD-related mygenic regulatory factors (MRFs) 1 have been identified as follows: Myf5, MRF4, and myogenin (43)(44)(45)(46)(47)(48). Each of these factors has highly related DNA-binding and helix-loop-helix regions and each is capable of initiating muscle differentiation when ectopically expressed in other cell types. The impact of each of the four MRFs on muscle-specific gene expression in culture and on muscle differentiation during mouse development has been examined. Clearly, some functional redundancies exist; for example, deletion of either the MyoD or Myf5 gene in mice has no notable defect in skeletal muscle (49,50), whereas deletion of both genes results in embryos lacking myoblasts and differentiated skeletal muscle (51). However, during development, each MRF displays temporal and spatial differences in gene expression, and myogenindeficient embryos form myoblasts but are unable to differentiate them into muscle (52,53). In culture, MyoD and Myf5 are expressed in undifferentiated myoblasts, whereas myogenin expression occurs early during differentiation, and MRF4 is expressed during late differentiation or post-differentiation (reviewed in Refs. 54 -56). In addition, MRF4 shows a more restricted ability to trans-activate muscle-specific genes in culture (57,58). Comparison of the ability of MyoD, Myf5, and myogenin to induce the activation of endogenous muscle-specific genes from chromatin revealed that myogenin was much less efficient at gene activation than were MyoD or Myf5 (44,59). This difference was recently mapped to specific residues in the basic helix-loop-helix domains of these proteins (60). Thus it is likely that the four MRFs have distinct functions during muscle differentiation.
In light of the differences between individual MRFs, we introduced each MRF into fibroblasts that inducibly express dominant negative versions of the SWI/SNF ATPases BRG1 or BRM. We then determined whether activation of muscle-specific and cell cycle-regulated genes by each of the four MRFs required SWI/SNF chromatin remodeling enzymes. The results demonstrate that activation of muscle-specific genes uniformly required SWI/SNF enzymes, whereas activation of cell cycle regulators and induction of cell cycle arrest uniformly did not.
Cell Lines and Cell Differentiation-Cell lines that inducibly express the Tet-VP16 regulator (Tet-VP16), dominant negative BRM (H17), or dominant negative BRG1 (B22) were maintained and induced by removal of tetracycline (Sigma) as described previously (31,35). To generate retrovirus, BOSC23 cells were cultured in 100-mm dishes and transfected at 80% confluence with FuGENE (Roche Molecular Biochemicals) with 10 g of pBABE-MyoD, pBABE-mouse Myf5, pBABE-MRF4, pBABE-Myogenin, or the empty vector (pBABE). Viral supernatants were harvested 48 h after transfection. Dishes (60 mm) of B22, H17, and Tet-VP16 cells grown in the presence of tetracycline at 50% confluence were infected individually with equal volumes of each different virus in DMEM containing 10% calf serum, 4 g/ml of Polybrene, and 2 g/ml tetracycline in a final volume of 5 ml. Each dish was split 1:3 48 h after infection, and the media were replaced with DMEM containing 10% calf serum, 2 g/ml puromycin, and 2 g/ml tetracycline. Following drug selection, each infected cell line was split 1:4 into DMEM plus 10% calf serum containing or lacking tetracycline and differentiated as described (31).
Protein Extracts and Western Analysis-Isolation of proteins and Western blotting were as described (35). Antibodies against the FLAG epitope (M2) and Rel B were from Sigma and Santa Cruz Biotechnology, respectively. The anti-cyclin D3 antibody was obtained from Signal Transduction Laboratories.
Expression Assays-Total cellular RNA was isolated by Trizol (Invitrogen) as described by the manufacturer. Northern analysis was performed as described (32). Probes for myogenin, myosin heavy chain, p21, and troponin T were described (32). The entire cDNAs encoding MyoD, MRF4, and Myf5 were excised from pEMSV-MyoD, pEMSV-MRF4, and pIC 19H and used as probes. RT-PCR procedures and Rb and hypoxanthine-guanine phosphoribosyltransferase primers were described previously (32). For the transfection experiments, the 4RtkCAT reporter gene plasmid and a control CMV-␤gal plasmid were transfected using Superfect (Qiagen) 24 h prior to the onset of differentiation. CAT and ␤-galactosidase assays were performed as described (31,64,65). CAT signal was normalized to ␤-galactosidase signal.
FACS Analysis-Cells were grown and differentiated as described above and then fixed as described (66). Incorporation of propidium iodide was determined by flow cytometry.

RESULTS
We previously created NIH 3T3-based cell lines that stably incorporate inducible, ATPase-deficient alleles of either the BRG1 or BRM catalytic subunit of mammalian SWI/SNF chromatin remodeling enzymes. In the absence of tetracycline, these cells express the mutant ATPases, which associate with other SWI/SNF subunits and generate non-functional SWI/ SNF complexes that act as dominant negatives (35). Previous studies utilizing these cell lines revealed a requirement for SWI/SNF complexes for full induction of the endogenous Hsp70 gene in response to some but not all environmental stresses and demonstrated that the expression of several constitutively expressed genes were unaffected by the dominant negative complexes (35). Subsequent studies demonstrated that SWI/ SNF enzymes were absolutely required for induction of the myogenic differentiation pathway mediated by introduction of the MyoD regulator (31,32). To further understand the initiation of myogenesis and the myogenic regulatory proteins that mediate this process in this in vitro differentiation model, we sought to compare the requirement for SWI/SNF enzymes when differentiation was induced by each of the four myogenic regulators MyoD, Myf5, MRF4, and myogenin.
Our previous studies examining MyoD-induced differentiation utilized a retroviral vector (pBABE (67)) to introduce MyoD. For these comparative studies, we cloned the Myf5, MRF4, and myogenin cDNAs into this vector and used the derived retroviruses to initiate the myogenic program in cell lines expressing or not expressing dominant negative BRG1 (B22 cells) or dominant negative hBRM (H17 cells). The parental cell line tet-VP16, which encodes the tetracycline-regulated transactivator but no target gene under tetracycline operator control, was used as a control. Cells infected with the empty pBABE vector served as an additional control. Fig. 1 demonstrates that FLAG-tagged mutant BRG1 was inducibly expressed in the differentiated B22 cells, and FLAG-tagged mutant hBRM was inducibly expressed in the differentiated H17 cells. Expression levels of the mutant ATPase alleles were similar for each cell line regardless of the retroviral vector utilized. As expected, no FLAG immunoreactivity was observed in the tet-VP16 cells. Because these cell lines are derived from NIH 3T3 cells, they do not fuse upon differentiation to form myotubes as do many other fibroblasts, thus our analysis involved molecular detection of myogenic markers.
Among the four myogenic regulators, myogenin can be in-duced by the other three factors; however, MyoD, Myf5, and MRF4 typically do not induce the expression of each other in cell culture models of muscle differentiation. Initial experiments therefore analyzed levels of MyoD, Myf5, and MRF4 in tet-VP16 control and B22 and H17 dominant negative cell lines that were infected with retroviruses encoding each of the four MRFs. Tet-VP16 cells infected with the Myf5, MyoD, or MRF4 retrovirus produced only mRNA for the gene encoded by the retrovirus (Fig. 2, A and B). Identical results were obtained upon infection of the dominant negative cell lines (Fig. 2, A and  B). The presence or absence of tetracycline did not affect the level of MRF mRNA, indicating that the dominant negative SWI/SNF enzymes did not affect retroviral integration or expression from the viral promoter (Fig. 2, A and B; see also Ref. 31). As expected, infection with the empty pBABE vector did not result in expression of any of these MRFs. In each of these blots we observed only one band for each MRF mRNA, and each was slightly larger than the size expected for mRNA transcribed from the endogenous gene (data not shown). Because the retrovirus generates a transcriptional fusion between retroviral gag gene and the MRF cDNA, this indicates that the ectopic expression of Myf5, MyoD, or MRF4 did not activate transcription from the respective endogenous loci. Previous work (31,68) had demonstrated that ectopic expression of MyoD in NIH 3T3 cells does not activate the endogenous MyoD gene; similar conclusions can now be made for Myf5 and MRF4.
Infection of the three cell lines with a myogenin-encoding retrovirus or with the empty pBABE vector did not induce expression of Myf5, MyoD, or MRF4 (Fig. 2B). We next examined the expression of myogenin and two downstream myogenic markers, myosin heavy chain (MHC) and troponin T. We previously reported that expression of dominant negative BRG1-or hBRM-based SWI/SNF complexes totally inhibited expression of these markers during MyoDinduced differentiation (31,32). These results were confirmed in Fig. 3. When differentiation was induced by Myf5 or MRF4, a similar inhibition of muscle-specific gene expression was observed in B22-and H17-infected cells, although MRF4-infected samples reproducibly showed low levels of marker gene expression in the presence of the dominant negative enzymes (Figs. 3 and 4). This suggests that a small amount of MRF4-stimulated transcription may be able to occur in the absence of functional SWI/SNF enzymes. The absence of tetracycline in the tet-VP16 cell samples had no effect on marker gene expression (Figs. 3 and  4). These data indicate that expression of dominant negative SWI/SNF enzymes inhibited the ability of MyoD, Myf5, and MRF4 to activate the expression of muscle-specific genes.
As in previous studies (31,32,35) with these cell lines that express dominant negative SWI/SNF enzymes, the results do not specifically indicate whether it is the BRG1-or BRM-based SWI/SNF complexes (or both) that are required for these regulatory events. Because the ATPases form distinct complexes with many of the same subunits, induction of one dominant negative ATPase could titrate subunits away from the other, wild type ATPase (see Ref. 35), effectively crippling both BRG1 and BRM complexes at the same time.
Infection of the three cell lines with myogenin generated equivalent levels of the retrovirally encoded myogenin mRNA in the presence and absence of tetracycline (Fig. 4). The endogenous myogenin gene was not activated by the retrovirally encoded myogenin, as was observed for MyoD, Myf5, and MRF4. Despite the fact that ectopically expressed myogenin can induce the differentiation of fibroblast lines, including NIH 3T3 fibroblasts (44), we observed no bulk mRNA accumulation for the downstream MHC or troponin T marker genes. Previous studies have documented that myogenin is less efficient at activating endogenous muscle-specific genes than are MyoD or Myf5 (44,59,60), and a recent study demonstrated that the levels of muscle-specific mRNAs from cells ectopically expressing myogenin were severely reduced or not detectable (60). To confirm that the myogenin introduced into our cells was transcriptionally competent, we transiently transfected the infected cells with the plasmid 4RtkCAT (62), which encodes a bacterial chloramphenicol acetyltransferase (CAT) gene under the control of four tandem E box sites placed upstream of the herpes simplex virus thymidine kinase promoter. Analysis of CAT activity showed that myogenin activated the reporter gene 3-5-fold, which was 2-3-fold less efficient than activation by MyoD (Fig. 5). The expression of dominant negative BRG1 or hBRM had no effect on myogenin activity and little effect on MyoD activity (Fig. 5), as we reported previously (31). Thus the retrovirally produced myogenin was transcriptionally active and could activate a transfected reporter gene; however, it was not capable of activating transcription of endogenous musclespecific genes.
Cell cycle arrest is a requirement for muscle differentiation, and introduction of MyoD activates the expression of several cell cycle regulators that help promote withdrawal from the cell cycle. In contrast to the requirement for SWI/SNF enzymes in the MyoD-mediated activation of muscle-specific genes, stimulation of the cell cycle regulators cyclin D3, Rb, and the p21 cyclin-dependent kinase (cdk) inhibitor by MyoD occurred normally in the presence of dominant negative SWI/SNF enzymes, and cell cycle arrest was achieved (32). These data indicated a differential requirement for SWI/SNF enzymes during muscle differentiation and suggest that the mechanism by which MyoD promotes activation of cell cycle regulators is distinct from that by which MyoD activates muscle-specific genes. To determine the requirement for SWI/SNF enzymes in the activation of cell cycle regulators and in cell cycle arrest induced by Myf5, MRF4, and myogenin, we first examined p21 mRNA levels. p21 is a cdk inhibitor of the CIP1/KIP1 class that can be induced by MyoD to promote cell cycle arrest (69 -71). As shown in Fig. 6, all four of the myogenic regulators were able to robustly induce p21 levels. Thus the dominant negative SWI/ SNF enzymes do not interfere with p21 induction, regardless of the MRF used to initiate differentiation. Samples were also analyzed by Western blotting to detect levels of cyclin D3. Cyclin D3 is transcriptionally stimulated during muscle differentiation and promotes withdrawal from the cell cycle via its interactions with Rb, cdk4, and other cell cycle regulators (72). We observed that MyoD, Myf5, MRF4, and myogenin increased cyclin D3 levels by roughly equivalent amounts and that the presence of dominant negative SWI/SNF enzymes had no effect on cyclin D3 induction (Fig. 7A). Finally, RNA samples were analyzed by RT-PCR for Rb mRNA levels. Rb is up-regulated by MyoD and is required both for activation of MEF2 transcription factors that cooperate with the MRFs to activate musclespecific genes and for repressing E2F in order to prevent cell cycle progression (73). All four MRFs generated a modest induction of Rb mRNA compared with cells infected with the empty pBABE vector (Fig. 7B). These data indicate that the up-regulation of cell cycle regulatory gene expression that occurs during muscle differentiation in culture can occur, despite the presence of mutant SWI/SNF enzymes at levels that inhibit activation of muscle-specific genes.
The observations showing that all four MRFs can up-regulate the expression of cell cycle regulators that function in cell cycle arrest, even in the presence of dominant negative SWI/ SNF enzymes, strongly support the prediction that withdrawal from the cell cycle occurred normally in these cells. To confirm this prediction, B22 and H17 cells were individually infected with each of the four MRFs, and fluorescence-activated cell sorter analysis was performed on propidium iodide-stained cells to measure DNA content. The percentage of cells showing Ͼ2n DNA content was defined as the population in S phase.

FIG. 4. Expression of dominant negative SWI/SNF enzymes inhibits MRF4-induced activation of endogenous, muscle-specific loci.
tet-VP16, B22, and H17 cells were infected with virus encoding MRF4 or myogenin or with the empty vector and differentiated in the presence or absence of tetracycline. Total cellular RNA was isolated and used for Northern blotting. Ethidium bromide staining of the rRNA photographed before transfer is shown as a loading control. Exog., exogenous; Endog., endogenous. Fig. 8 shows the results of this experiment in graph format. In subconfluent cultures of uninfected, proliferating cells, the percentage of cells in S phase ranged from 23 to 44% (Fig. 8 and data not shown). Cells infected with any of the four MRFs and subjected to in vitro differentiation generally showed 3-9% of the cells in S phase, and no effect, either positive or negative, of the dominant negative SWI/SNF enzymes was observed.
Our analysis of the four MRFs indicates that each is capable of inducing the expression of cell cycle regulators and promoting withdrawal from the cell cycle, regardless of the presence or absence of dominant negative SWI/SNF enzymes. The data confirm earlier observations that there is a differential requirement for SWI/SNF enzymes during muscle differentiation in culture and suggest that the molecular distinctions between activation of muscle-specific genes and cell cycle regulators are largely independent of the MRF utilized to initiate the differentiation program. DISCUSSION We examined the requirement for SWI/SNF chromatin remodeling enzymes during the initiation of muscle-specific gene expression and during withdrawal from the cell cycle when muscle differentiation is induced in culture by each of the members of the MyoD-related family of myogenic regulatory factors. In vivo analyses of these four factors, MyoD, Myf5, MRF4, and myogenin, have demonstrated distinct patterns of gene expression for each in the developing embryo, and the combination of single and multiple knockout studies have indicated that MyoD and Myf5 are determinants of the muscle lineage, whereas myogenin is most likely contributing to terminal differentiation (reviewed in Refs. 54 -56 and 74). We therefore considered that the different MRFs might show different requirements for SWI/SNF chromatin remodeling enzymes during gene activation. Whereas this remains a possibility in the developing embryo, it appears that in the cellular model for muscle differentiation utilized, SWI/SNF enzymes are similarly required to initiate muscle-specific gene expression by the three MRFs (MyoD, Myf5, and MRF4) that generated measurable levels of the muscle-specific gene products. Additionally, each of the four MRFs induced the expression of cell cycle regulators, indicating that all four MRF4 do not require the presence of functional SWI/SNF enzymes to activate genes that promote cell cycle arrest.
Despite the fact that myogenin was cloned as a factor that could convert NIH 3T3 fibroblasts to cells expressing musclespecific proteins, we were unable to detect measurable quantities of muscle-specific mRNAs in our NIH 3T3-derived dominant negative cell lines. The original observations indicating that myogenin could convert fibroblasts to muscle-like cells were based on immunofluorescence studies that reported relatively low rates of conversion based on the presence of MHC and other muscle-specific markers (44,59). Although there is no question that myogenin can initiate the myogenic pathway in other cell types, it may be rather inefficient in this process compared with the ability of other MRFs to do so. In fact, previous reports (44, 59) have used immunofluorescence to indicate that myogenin is less efficient than MyoD and Myf5 at inducing endogenous MHC and desmin expression. In addition, Bergstrom and Tapscott (60) recently reported that introduction of myogenin into NIH 3T3 cells resulted in little or no accumulation of mRNA from the endogenous myogenin, MHC, and desmin loci in these cells; our results indicating a failure of myogenin to cause significant accumulation of mRNA from the endogenous myogenin, MHC, and troponin T genes are consistent with this previous study. Perhaps the inefficiency of myogenin to induce muscle-specific gene expression is mediated at the level of individual cells, resulting in a small percentage of cells that fully induce expression of the muscle-specific mRNAs, whereas most do not. Such a scenario could account for the failure to visualize differentiation by Northern or other assays that measure stably accumulated transcripts.
The three MRFs that can efficiently drive expression of muscle-specific genes share a conserved amphipathic ␣-helix in their respective C termini. The sequence of the corresponding residues in myogenin is distinct, and Bergstrom and Tapscott (60) have performed detailed analysis of this region to demonstrate that this sequence confers the ability to activate endogenous, muscle-specific gene expression. Substitution of the MyoD C-terminal ␣-helix into myogenin increased the ability of the protein to activate endogenous muscle-specific gene expression, and replacement of a nearby histidine-rich region with the corresponding MyoD sequences enhanced the activity of the chimeric molecule such that it could activate endogenous gene expression nearly as well as MyoD (60). The histidine-rich region had been shown previously (59) to be important for promoting chromatin remodeling at endogenous loci, thus suggesting that the conserved ␣-helix may in some way also be contributing to chromatin remodeling (60). However, the mechanism by which the conserved ␣-helix promotes transcription from endogenous genes has not been defined. Given the correlation between conservation of the ␣-helix among MyoD, Myf5, and MRF4 and the ability of these activators to activate endogenous muscle-specific genes in a SWI/SNF-dependent manner, it is tempting to speculate that the conserved ␣-helix, alone or in combination with the histidine-rich region, directly contacts one or more components of the SWI/SNF enzymes. One could then attribute the muscle-specifying functions of MyoD (and Myf5 and MRF4) to the ability to interact with and potentially target the SWI/SNF chromatin remodeling enzymes to endogenous loci that need to undergo chromatin remodeling to facilitate transcription. However, to date there have been no reports documenting interaction between SWI/SNF components and MyoD or myogenin in extracts from differentiating cells. Moreover, other chromatin remodeling activities, in particular, histone acetylation, have been observed to occur on muscle-specific promoters (75); direct interactions between MRFs and components of histone acetyltransferase enzyme complexes or other chromatin-modifying machines could account for the observed specificity of the MyoD/Myf5/MRF4 C-terminal ␣-helix.
In contrast to the observed SWI/SNF dependence for activation of muscle-specific genes by the three MRFs that could activate endogenous genes, all four MRFs were able to upregulate the expression of cell cycle regulators and induce withdrawal from the cell cycle in an SWI/SNF-independent manner. These data reinforce the conclusion that SWI/SNF enzymes are not universally required for cell cycle withdrawal (32). Other studies have demonstrated that introduction of a constitutively active allele of the Rb tumor suppressor into proliferating cells induced cell cycle arrest, except in cells lacking BRG1 and BRM or in cells expressing dominant negative BRG1 (76 -78). Thus there is a requirement for BRG1 or BRM containing SWI/SNF enzymes during cell cycle arrest mediated by Rb. However, when cell cycle arrest is promoted by expression of one of the MRFs and the cells are subjected to differentiation (low serum) conditions, it appears that lack of functional SWI/SNF complexes is not sufficient to prevent cell cycle withdrawal. In fact, under the differentiation conditions used, even cells not expressing an MRF arrested in the presence of dominant negative SWI/SNF enzymes. We propose that the convergence of signals directed by the MRF-mediated up-regulation of cell cycle regulators, the reduction in serum concentration, and the confluence of the cells must be enough to overcome whatever function SWI/SNF enzymes perform during Rb-mediated cell cycle arrest, and that the contribution of the MRF under these conditions is not absolutely required for arrest to occur.
Work from Bergstrom and Tapscott (60) also revealed that myogenin could fully induce the p21 cdk inhibitor, indicating that the myogenin C-terminal ␣-helix was sufficient to induce activation from the endogenous p21 locus despite the fact that endogenous muscle-specific gene products were only poorly induced. Our current data and our previous study indicating that induction of p21 and other cell cycle regulators is independent of SWI/SNF enzyme function strongly suggest a distinction between the mechanism by which the cell cycle regulators and the muscle-specific gene products are transcriptionally activated. We propose that the three MRFs capable of activating endogenous muscle-specific genes (MyoD, Myf5, and MRF4) do so via a mechanism that requires SWI/SNF-mediated chromatin remodeling at the promoters of these loci. Whereas alteration of chromatin structure by SWI/SNF enzymes is required, it is likely that the activation of these loci will be accomplished from the cumulative effects of the MRFs, FIG. 8. Cell cycle arrest occurs in the presence of dominant negative SWI/SNF enzymes. B22 and H17 cells were infected with virus encoding one of the four MRFs or the empty vector and differentiated in the presence or absence of tetracycline. Cells were fixed, stained with propidium iodide, and subjected to FACS analysis. A bar graph presenting the percentage of cells in S phase from a representative experiment is shown. The proliferating cell control represents uninfected, subconfluent B22 or H17 cells growing in the presence of tetracycline. SWI/SNF enzymes, and other chromatin-modifying enzymes. Clearly the specific sequence of the C-terminal ␣-helix found in these three MRFs contributes to this activation process as well. In contrast, MRF-mediated activation of p21 and other cell cycle regulators is independent of SWI/SNF function and also does not require the C-terminal ␣-helix sequence found in MyoD, Myf5, and MRF4. We propose that the distinction between activation of cell cycle regulators and muscle-specific genes stems from the relative state of the promoters at the onset of differentiation. Cell cycle regulators are expressed constitutively or at specific stages of the cell cycle during proliferative growth and some are up-regulated during differentiation. In contrast, during proliferative growth, the musclespecific loci are inactive and likely packaged into a repressive chromatin environment. Activation of loci packaged into inactive chromatin would therefore require chromatin remodeling by SWI/SNF enzymes in a manner that is in some way facilitated by the MyoD C-terminal ␣-helix. Analysis of in vivo promoter structure and promoter occupancy by the different MRFs and chromatin remodeling enzymes on muscle-specific and cell cycle regulator promoters should identify differences in the modes of activation of both classes of MRF-inducible genes.