The Core Binding Factor CBF Negatively Regulates Skeletal Muscle Terminal Differentiation

Background Core Binding Factor or CBF is a transcription factor composed of two subunits, Runx1/AML-1 and CBF beta or CBFβ. CBF was originally described as a regulator of hematopoiesis. Methodology/Principal Findings Here we show that CBF is involved in the control of skeletal muscle terminal differentiation. Indeed, downregulation of either Runx1 or CBFβ protein level accelerates cell cycle exit and muscle terminal differentiation. Conversely, overexpression of CBFβ in myoblasts slows terminal differentiation. CBF interacts directly with the master myogenic transcription factor MyoD, preferentially in proliferating myoblasts, via Runx1 subunit. In addition, we show a preferential recruitment of Runx1 protein to MyoD target genes in proliferating myoblasts. The MyoD/CBF complex contains several chromatin modifying enzymes that inhibits MyoD activity, such as HDACs, Suv39h1 and HP1β. When overexpressed, CBFβ induced an inhibition of activating histone modification marks concomitant with an increase in repressive modifications at MyoD target promoters. Conclusions/Significance Taken together, our data show a new role for Runx1/CBFβ in the control of the proliferation/differentiation in skeletal myoblasts.


Introduction
Runx1 (for Runt-related transcription factor 1, also known as AML1 for Acute Myeloid Leukemia 1, CBFA2 or PEPB2aB) belongs to a family of highly homologous heterodimeric transcription factors named Core Binding Factors or CBF (reviewed in: [1]). In addition to the Runx1 subunit which binds DNA directly, CBF is composed of a non-DNA-binding subunit named CBFbeta (CBFb) [2]. Runx1 binds better DNA in the presence of CBFb. Runx1 was originally identified at a breakpoint on human chromosome 21 in the t(8;21) translocation, known as the most common target of chromosomal translocations in human leukemia [3,4]. Genetic studies showed that Runx1 is essential in the developing murine embryo for definitive hematopoiesis of all lineages [5,6].
Runx proteins have the potential to either activate or repress transcription in a context dependent manner. Runx1 seems to promote proliferation in progenitor cells, whereas in differentiating cells it cooperates with tissue-specific transcription factors to regulate tissue-specific gene expression. For example, Runx1 cooperates with C/EBPa and C/EBPb to regulate hematopoiesis and osteogenesis, respectively [17,18]. The dual role of Runx1 in regulating proliferation and differentiation could depend on differential interactions with protein partners, specific for each stage of cell development. The molecular mechanisms underlying such a switch in Runx1 function remain however to be deciphered.
Runx1 and CBFb have also been linked to skeletal muscle differentiation [19][20][21], and prevention of muscle wasting [20]. In skeletal muscle, proliferation and differentiation are mutually exclusive. Indeed, skeletal muscle terminal differentiation begins with an irreversible withdrawal from the cell cycle, followed by muscle-specific marker expression [22]. Irreversible cell cycle exit involves a definitive silencing of proliferation-associated genes (reviewed in [23] and references therein). Terminal muscle differentiation is orchestrated by myogenic bHLH transcription factors, such as MyoD and Myf5, two master myogenic determination factors. MyoD is expressed in proliferating myoblasts, but is unable to activate its target genes even when bound to their promoters [24,25]. MyoD therefore may have a repressive role at its target genes prior to initiating chromatin remodeling in differentiating cells [24,26,27]. In proliferating myoblasts, MyoD is associated with histone deacetylases (HDACs), the histone methyltransferase Suv39h1 and heterochromatin protein HP1, and might actively inhibit expression of its target genes by inducing a local repressive chromatin structure [24,28,29].
Here we show that CBF associates with MyoD preferentially in proliferating myoblasts, and knockdown of Runx1 or CBFb accelerates cell cycle exit and terminal differentiation. Conversely, overexpression of CBF slows cell cycle exit and delays muscle differentiation. In proliferating myoblasts, the MyoD/CBF complex contains several chromatin modifying enzymes such as HDACs. In agreement with this, when overexpressed, CBFb maintains histone H3 hypoacetylated, hypomethylated on lysine 4 and hypermethylated on lysine 9, on MyoD target promoters, along with HDAC1 recruitment, even in differentiation conditions. Finally, Runx1 is recruited to MyoD target genes preferentially in proliferating myoblasts, when these genes are repressed. Altogether, our data suggest that CBF transcription factor plays a pivotal role as a negative regulator of skeletal muscle terminal differentiation.

CBF Subunits, Runx1 and CBFb, Interact with MyoD in Proliferating Myoblasts
In an attempt to characterize MyoD protein partners, we carried out double-affinity purification of HA-Flag MyoD stably expressed in HeLa cells (see purification scheme on Figure S1). MyoD protein complex composition was then analyzed by mass spectrometry (MS) and western blot (WB). MS analysis of the purified protein complex revealed some already known partners of MyoD ( Figure S2), such as Id, Pbx1, PC4 and E12/E47, and partners that had never been described to interact with MyoD. Indeed, MS analysis unveiled CBFb protein within MyoD complex with a high number of peptides, covering almost 20% of its protein mass and amino acids content ( Figure 1A). WB analyses confirmed that result and showed that Runx1 also copurified with MyoD ( Figure 1B).
To further investigate the CBF/MyoD interaction, we turned to myogenic cells: the murine myoblastic cell line C2C12. Both Western blot analysis of double-purified Flag-HA-MyoD (MyoD), or eluate from HeLa control cells (Mock) with the indicated antibodies (WB Ab). S: soluble; C: chromatin associated. C. HeLa cells stably expressing Flag-HA-CBFb (eCBFb) or HeLa control cells stably transfected with the empty vector (Ctr) were transiently transfected with expressing vectors for HA-tagged MyoD and/or HA-tagged Runx1. 24 hours posttransfection, cells were harvested and lysates were used for immunoprecipitation (IP) using Flag resin to precipitate Flag-HA-CBFb. Precipitates were then subjected to western blot using HA Ab (WB a-HA) to simultaneously detect HA-Runx1, HA-MyoD, and Flag-HA-CBFb (discriminated on the gel by their molecular weight). However, we have checked the identity of each HA-revealed band by using antibodies recognizing the native proteins ( Figure S8). D. Nuclear extracts from proliferating (prolif.) or differentiating myoblasts (48 h, indicated as Diff.) were used for immunoprecipitation (IP) with antibodies (Ab) raised against MyoD (lanes 4 and 9) and Myf5 (lanes 2 and 7), with control beads (lanes 3 and 8) or with normal rabbit IgG (lanes 5 and 10) as negative controls. The resulting precipitates were then subjected to western blot analysis (WB) for the presence of MyoD, Myf5 and CBFb. Input extracts were loaded to show endogenous protein levels (lanes 1 and 6). E. Runx1, CBFb, or luciferase (Luc) were in vitro translated in the presence of 35 S-Methionine (inputs on lanes 1-3, respectively) and incubated with equivalent amounts of GST-MyoD beads (lanes 4-7) or GST beads (lanes [8][9][10][11]. GST pull-down was then conducted as described in the Material and Methods section, and the radiolabeled proteins were detected by autoradiography. *: CBFb. doi:10.1371/journal.pone.0009425.g001 CBFb and Runx1 are expressed in the C2C12 myoblasts and their protein level do not vary significantly during differentiation ( Figure  S3). We found that MyoD, and the other myogenic determination factor Myf5, co-precipitated with CBFb preferentially in proliferating compared to differentiating C2C12 myoblasts ( Figure 1D).
To assay whether MyoD has the ability to interact directly with CBF, we performed GST pull-down experiments, which showed that GST-MyoD strongly interacts with Runx1 ( Figure 1E, lane 4), but not with CBFb ( Figure 1E, lane 5). The interaction of MyoD with Runx1 was specific; we did not detect any Runx1 signal in the presence of GST protein alone ( Figure 1E, lane 8) nor any luciferase signal with GST-MyoD ( Figure 1E, lane 7). Interestingly, GST-MyoD interacts with CBFb only in the presence of Runx1 ( Figure 1E, compare lanes 5 and 6), in agreement with our previous findings ( Figure 1C). These results show that MyoD interacts directly with heterodimeric transcription factor CBF, via the Runx1 subunit.
The bHLH and the Transactivating Domains of MyoD, and the Transcription Regulation Domain of Runx1 Are Required for Their Interaction In an attempt to delimit the domain of MyoD responsible for interaction with Runx1, we used HA-tagged mutants of MyoD transfected into HEK 293 cells. Anti-HA immunoprecipitation revealed an interaction of Runx1 and CBFb with wild-type MyoD as expected, and only with a MyoD 82-318 mutant, which retains the bHLH and the C-terminal transactivating domains (Figure 2A). Runx1 and CBFb failed to interact with truncated MyoD versions lacking either the bHLH, i.e., mutants Nter and Cter, or the Cterminal domain, i.e., mutants DCter, Nter and bHLH ( Figure 2A). These experiments clearly show that the bHLH and the Cterminal transactivating domains are required for interaction with Runx1. Notably, MyoD deletion mutants that do not interact with Runx1 do not interact with CBFb ( Figure 2A). This result, combined with the results presented on Figure 1E, indicate that Runx1 is most likely the subunit that directly interact with MyoD.
To delimit the Runx1 domain involved in the interaction with MyoD, we performed a GST-pull down experiment ( Figure 2B). Our results show that the transcription regulation domain located in the C-terminal part of Runx1 is required for the interaction with MyoD ( Figure 2B). The interaction of MyoD with Runx1 was specific; we did not detect any Runx1 signal in the presence of GST protein alone nor any luciferase signal with GST-MyoD ( Figure 2B).

CBF Negatively Regulates Cell Cycle Exit and Terminal Differentiation in Skeletal Myoblasts
We used siRNAs to decrease Runx1 level in order to investigate its role in differentiating myoblasts ( Figure 3A, see Runx1 quantification). Downregulation of Runx1 resulted in a more efficient differentiation ( Figure 3A); both the expression of muscle markers and the proportion of multinucleated cells were higher in Runx1depleted cells ( Figure 3A and Figure S4A). In particular, myogenin and MCK (Muscle Creatine Kinase) were expressed at higher levels in Runx1-depleted cells ( Figure 3A). Interestingly, cyclin D1 level decreased more rapidly when Runx1 is downregulated ( Figure 3A). Similarly, CBFb downregulation induced an accelerated differentiation ( Figure 3B, top panel) and a more rapid decrease in cyclin D1 ( Figure 3B, lower panel). Indeed, we could detect MCK expression in CBFb-depleted myoblasts as soon as 24 h after the induction of differentiation ( Figure 3B, top panel). These cells moreover exhibited larger myotubes ( Figure S4B). We confirmed these results in primary myoblasts ( Figure 3C). In agreement with our previous results, we have found that Runx1 or CBFb downregulation led to a significant decrease of Sphase cells proportion concomitant with an increase in G1-phase cells ( Figure 3D). This suggests that CBF positively regulates myoblasts proliferation.
As for the specificity of the siRNAs, we obtained the same phenotype with three different siRNAs that target CBF: the two targeting CBFb subunit and the one targeting Runx1 subunit isoforms. Thus, the observed effects are unlikely due to any offtarget effect.
To complete our analysis, we studied the effect of CBFb overexpression on terminal differentiation. Overexpression of CBFb in C2C12 myoblasts (C2C12-CBFb) was well tolerated and did not lead to morphological abnormality ( Figure S5). However, in differentiation conditions, C2C12-CBFb cells showed a delay in cell cycle exit, as measured by early-G1 phase cyclin D1 level that decreased with a 24 to 48 h delay compared to control cells ( Figure 4A), but not that of late-G1 cyclins A and E ( Figure 4C). Note that proliferating C2C12-CBFb cells contain more cyclin D1 protein then the control cells ( Figure 4A), but not cyclins A2 and E ( Figure 4C). The delayed cell cycle exit correlated with delayed expression of muscle markers such as myogenin (48 h delay), MCK (24 h delay), and MHC (Myosin Heavy Chain, not detected at 120 h) ( Figure 4A). In contrast to control cells, C2C12-CBFb cells exhibited smaller and mainly mononucleated myotubes ( Figure 4B and Figure S5) with low expression of MCK and MHC ( Figure 4B) in differentiation conditions. This suggests that differentiation kinetics was not completely impaired but greatly delayed when CBFb was overexpressed. Further analysis of cell cycle regulators expression showed that, in addition to a delayed decrease in cyclin D1 ( Figure 4A and C), cyclin D3 and p21 expression is delayed in C2C12-CBFb cells compared to C2C12 control cells ( Figure 4C).
Altogether, these results suggest that CBF plays a dual role during skeletal muscle differentiation by regulating cell cycle withdrawal and expression of muscle markers.

CBF Transcription Factor Is Located to MyoD Early Target Genes and Regulates Negatively Their Expression
The fact that CBF is a transcription factor which interacts with MyoD preferentially in proliferation conditions, led us to investigate whether it would be targeted to MyoD target genes to repress their transcription. Using an in silico approach, we first found that on early target gene promoters of MyoD, Runx1 and MyoD binding sites were adjacent ( Figure S6A). In order to test the effective recruitment of Runx1 onto MyoD target promoters, we performed ChIP experiments. Our results showed a preferential enrichment in Runx1 on myogenin, p21 and cycD3 promoters in proliferating compared to differentiating myoblasts ( Figure 5A). These MyoD target genes are expressed early in differentiating but not in proliferating myoblasts ( Figure S6B).
Our ChIP assays showed that Runx1 was not located on late target genes of MyoD, such as Desmin, MHC and MCK (data not shown), while it was on early muscle differentiation genes myogenin, p21 and cycD3. In addition, we did not find Runx1 binding sites adjacent to E-boxes on late MyoD target genes' promoters. These findings suggest that Runx1 would essentially regulate early events of skeletal muscle terminal differentiation. Taken together, our results strongly suggest that Runx1 could be recruited onto MyoD early target genes to regulate negatively their expression in proliferating myoblasts.
To gain insights into the mechanism of action of CBF on MyoD target genes, we purified CBFb protein complex from proliferating C2C12-CBFb cells via its Flag tag. As expected, CBFb co-purified with the myogenic factor MyoD and with its dimerization partner Runx1 ( Figure 5B). The other partners that co-purified specifically with CBFb are proteins known to be involved in transcriptional repression: the histone 3 lysine 9 (H3K9) methyltransferase Suv39h1, the heterochromatin protein HP1b, and the histone deacetylases HDACs 1, 2 and 3 ( Figure 5B). These interactions could be mediated by Runx1. Indeed, these proteins are already known partners of Runx1 on the one hand [30], and repressors of MyoD activity on the other hand [31][32][33].
Given the association of CBFb with chromatin-modifying enzymes, we studied the chromatin status of three target gene promoters of MyoD in differentiating C2C12-CBFb using chromatin immunoprecipitation (ChIP). Our results showed that, in contrast to control cells in which histone H3 acetylation (a mark associated with transcription activation) on myogenin, cyclin D3 and p21 promoters increased in differentiation compared with proliferation conditions, histone H3 acetylation levels at these promoters did not vary in C2C12-CBFb cells ( Figure 5C). More generally, in differentiation conditions, we found that activating marks (histone H3 acetylation, histone H3 lysine 4 tri-methylation) are abnormally lower on myogenin and cyclin D3 promoters in C2C12-CBFb cells compared to control cells ( Figure 5D). Concomitantly, repressive marks (histone H3 lysine 9 trimethylation, presence of HDAC1) are higher ( Figure 5D).
These results are in agreement with our findings showing that CBF associates with chromatin modifying enzymes, such as HDACs, Suv39h1 and HP1b ( Figure 5B), which are known to repress MyoD activity in proliferating myoblasts. Thus, CBF has an effect on the chromatin structure of three early target genes of MyoD and contributes to maintain a repressive chromatin state.

Runx1 Represses MyoD Transcriptional Activity
To investigate the effects of Runx1 on MyoD transcriptional activity, we used a luciferase reporter gene under the control of the myogenin promoter, which is a direct target promoter of MyoD that harbors a Runx-binding site adjacent to MyoD-binding site ( Figure S6A). Co-transfection experiments were performed in HeLa cells line that does not express MyoD endogenously. We observed that the co-transfection of Runx1-expressing plasmid together with MyoD expression vector resulted in the inhibition of myogenin promoter activity in a dose-dependent manner ( Figure 6). The inhibitory effect of Runx1 is specific and is MyoD-dependent. Indeed, it was not seen with Renilla-luciferase expression under a CMV promoter (used as a normalization control for transfection efficiency). Thus, Runx1 inhibits MyoD activity.

Discussion
Here we show that CBF transcription factor, composed of Runx1 and CBFb subunits, is expressed in proliferating myoblasts where it interacts with MyoD. Modulating the expression levels of either Runx1 or CBFb impaired cell cycle exit and terminal myogenic differentiation.

Runx1/CBFb Interact with MyoD
Mass spectrometry analysis revealed that CBFb is part of MyoD complex. Further western blot analyses of the same MyoD complex revealed the presence of Runx1 subunit. We have narrowed our study to Runx1 because of all the three Runx proteins, to our knowledge, Runx1 is the only one linked to skeletal muscle [20]. In addition, Runx1 is one of the MyoD target genes in myoblasts [34]. Among the other Runx members, Runx2 is mainly involved in osteogenesis, and it has already been shown that Runx2 expression is not detectable in myoblasts [35]. Moreover, ectopic expression of Runx2 in myoblasts triggers osteogenic transdifferentiation [35][36][37]. Finally, concerning Runx3, it has been shown that it is not expressed in skeletal muscle and is mainly linked to neurogenesis [38][39][40][41][42]. Thus, we explored the role of Runx1/CBFb during muscle skeletal terminal differentiation. However, Runx1 has been reported to play an important role in protecting denervated, fully differentiated, myofibers from atrophy and autophagy [20]. In this paper [20], Runx1 has been assigned a role of a growth-promoting factor of muscle cells to limit muscle wasting. In this very elegant study, the authors generated knock-in homozygous mice carrying inactive Runx1 in skeletal muscle cells. However, Runx1 inactivation was conducted under the indirect control of MCK promoter, which is active only at late stages of terminal differentiation. Thus, with this system, we cannot see the effect of Runx1 inactivation on early stages of terminal differentiation, where MCK promoter is not yet active.

Runx1 and CBFb Regulate Cell Cycle Exit and Terminal Differentiation
Runx1 or CBFb downregulation in myoblasts induced an accelerated cell cycle exit. Indeed, cyclin D1 protein disappears  We quantified copy numbers of the myogenin, p21 and cyclin D3 promoter regions harboring the MyoD and Runx1 target sequences, compared to 36B4 gene, which was used as a reference gene. Results are the mean of 3 independent experiments. B. Western blot analysis with the indicated antibodies (WB Ab) of Flag-purified Flag-HA-CBFb stably expressed in C2C12 cells (CBFb), or from C2C12 control cells (Ctr). Both inputs are probed in lanes 1-2 and eluates are shown in lanes 3-4. C. Chromatin immunoprecipitation (ChIP) experiments using anti-acetyl H3 antibody were performed from either proliferating (grey bars) or differentiating (black bars) C2C12 control (left) and C2C12-CBFb (right) cells. We quantified copy numbers of the myogenin (myog.), cyclin D3 and p21 promoter regions harboring the MyoD target sequences. Results are the mean of three measurements. D. ChIP experiments using antibodies against acetyl histone H3 (AceH3), trimethylated histone 3 lysine 4 (me3H3K4), trimethylated histone 3 lysine 9 (me3H3K9), Histone Deacetylase 1 (HDAC1) were performed from differentiating C2C12 control (grey bars) or C2C12-CBFb (black bars) cells (48 h differentiation time). We quantified copy numbers of the myogenin and cyclin D3 promoter regions harboring the MyoD target sequences. Results with the transcription activating marks (AceH3 and me3H3K4) were normalized using the expressed housekeeping gene 36B4, while the repressive marks (me3H3K9 and HDAC1) were normalized using the repressed major satellite repeats. Results are the mean of three measurements. doi:10.1371/journal.pone.0009425.g005 24 h earlier in differentiation conditions. In addition, Runx1 or CBFb downregulation led to a an increase in G1-phase cells, suggesting that CBF promotes myoblasts proliferation. Conversely, CBFb overexpression delayed the early-G1 phase cyclin D1 disappearance in differentiation conditions (cyclin D1 is still detectable 72 hours after the induction of differentiation), but not that of late-G1 and early-S phase cyclins A2 and E1. Note that proliferating C2C12-CBFb cells contain more cyclin D1 protein, but not cyclins A2 and E, than the control cells. Together, these results suggest that CBFb overexpression impacts on early G1 markers. In addition, expression of the cell cycle exit regulators p21 and cyclin D3 [43,44] is delayed in differentiating CBFb-overexpressing cells, suggesting a delayed or impaired cell cycle exit and terminal differentiation entry. In agreement with this, our ChIP results showed that in proliferating myoblasts, CBF (via Runx1) is recruited to repressed p21 and cyclin D3, which encode cell cycle exit regulators [43,44]. Together, these results suggest that CBF regulates positively proliferation, and negatively terminal differentiation, of skeletal myoblasts. Thus, CBF impacts on the proliferation/differentiation switch in myoblasts (see model on Figure S7).
In differentiation conditions, C2C12-CBFb cells showed delayed molecular differentiation (expression of muscle markers), and delayed appearance of myotubes that are abnormally small and mainly mono-nucleated. One alternative explanation is that these cells managed to differentiate correctly, although in a delayed manner, but exhibited a specific block in fusion and multinucleation. Indeed, CBFb subunit is retained in the cytoplasm by cortical filamins [45]. In muscle cells notably, structural proteins and cell adhesion proteins are required for the reorganization of the cell cytoskeleton during cell fusion to form myotubes [46]. Some studies do suspect a cytoplasmic role for CBFb [47]. Thus, CBFb overexpression in myoblasts, especially in the cytoplasm, could have a role in impairing the correct cytoskeleton reorganization during fusion. This could explain the phenomenon we observe in differentiating CBFb-overexpressing myoblasts. Moreover, we observed a very low expression of myogenin transcription factor in myoblasts overexpressing CBFb and since that myogenin is involved in cell fusion, this is an alternative explanation of the observed the mono-or di-nucleated myotubes.

CBF Regulates Muscle Differentiation via a Direct Interaction between MyoD and the Runx1 Subunit
Our results revealed that the role of CBF in myoblasts is likely to be partly mediated through direct interaction with MyoD. GST pull-down experiments showed that MyoD interacts directly with Runx1 subunit, but not with CBFb. This is confirmed by the use of MyoD deletion mutants in living cells. Indeed, MyoD deletion mutants that fail to interact with Runx1 do not interact neither with CBFb. MyoD/Runx1 interaction implicates the bHLH and the C-terminal transactivating domains of MyoD, and the transcription regulation domain of Runx1.
The preferential association of MyoD and CBF in proliferating myoblasts could mean that CBF might be acting as a negative cofactor of MyoD. Indeed, we provide evidence that CBF is recruited to early MyoD target genes, via Runx1, in proliferating myoblasts, where MyoD is mainly associated with transcriptional repressors [24,26,31]. This suggests that CBF may serve for assembly of a transcription repression complex at early MyoD target genes such as myogenin, p21 and cyclin D3 (see our model on Figure S7). As for example, such a mechanism could be involved in the repression of the skeletal muscle acetylcholine receptor gene, which contains a repressive E-box that mediates its repression in proliferating myoblasts [48]. In agreement with this, we found that in proliferating myoblasts, CBF associates with many chromatin modifying enzymes, such as Histone Deacetylases (HDACs 1, 2 and 3), the histone H3 lysine 9 (H3K9) methylase Suv39h1, and Heterochromatin Protein beta (HP1b), which are known to repress MyoD activity in proliferating myoblasts [24,31,49,50]; and already known to interact with Runx1 [30,51,52]. In agreement with this, in myoblasts overexpressing CBFb, histone H3 acetylation and its trimethylation on lysine 4 (marks of active transcription) are delayed on early MyoD target genes in differentiation conditions. Concomitantly, these genes remain abnormally marked by histone repressive marks (me3H3K9, HDAC1).
Interestingly, although MyoD is expressed both in proliferating and differentiating cells, we found that the interaction between MyoD and CBF was lost in differentiating cells. In addition, we showed that overexpression of CBFb in myoblasts led to stabilization of Runx1 subunit that could more efficiently repress MyoD transactivating activity, which induces a delay in terminal differentiation. In agreement with this, we have shown that Runx1 represses MyoD activity in a gene reporter assay.
We did not succeed to show the concomitant presence of Runx1 and MyoD on MyoD target genes, given that it has previously been demonstrated that in proliferating conditions, only a small fraction of MyoD contributes to the repressive remodeling of its target genes, prior differentiation. Alternatively, Runx1 could prevent the proper binding of MyoD and the recruitment of the transcriptional machinery. Notably, the displacement of Runx1 in differentiating conditions is concomitant with a strong binding of MyoD to its target promoters (data not shown).
MyoD and Runx1 are both subject to post-translational modifications. Notably, MyoD is phosphorylated during the cell cycle while it becomes acetylated during differentiation. Runx1 can also be phosphorylated, acetylated or methylated, while these modifications still need to be characterized in muscle cells and during muscle differentiation. We propose that these modifications could favor or impair MyoD interaction with Runx1, respectively.

Conclusion
Our findings concerning the role of CBF in the regulation of the proliferation/differentiation balance are in agreement with several reports. CBF was indeed implicated in skin epidermis and hair follicle differentiation [11,12], as well as in neuronal differentiation [8,9]. These data support an emerging role for Runx proteins in cell fate regulation in many cell lineages. Furthermore, MyoD was also shown to regulate osteogenic differentiation [53]. In addition, it has been shown that muscle satellite cells can differentiate into osteocytes or adipocytes under some conditions [54,55]. Thus, our results point to a model in which CBF and myogenic bHLH protein families could act in concert to induce cell-lineage-specific gene expression, dependent on the extra-cellular stimuli.
In summary, we propose that CBF transcription factor might participate in recruiting chromatin modifying enzymes to repress MyoD early target genes by locally inducing a repressive chromatin structure. Our data reveal a new critical role of CBF in the regulation of the balance between proliferation and differentiation in skeletal muscle cells. They also demonstrate a new mechanism of repression of differentiation genes in proliferating myoblasts.

Cell Culture
C2C12, HEK 293 and HeLa-S3 cells were cultured under standard conditions. C2C12 cells and mouse primary myoblasts were cultured and differentiated as described in: [56].

Stable Cell Lines Establishment and Plasmid Construction
A HeLa cell line stably expressing MyoD was established with a transgene encoding for full-length MyoD; and HeLa and C2C12 cell lines expressing CBFb were established with a transgene encoding for full-length CBFb. The transgenes were tagged with double-HA (Haemagglutinin) and double-FLAG epitopes at the N-terminus as described in [57].
Control cell lines transduced with the empty vector were also established. Murine CBFb cDNA (a kind gift from Dr Nancy A. Speck) was amplified by PCR with specific primers with protruding restriction sites (fw-Pspx1: CCGCTCGAGCCG-CGCGTCGTCCCGGG, rev-Not1: ATTCTATATGCGG-CCGCTAACGAAGTTTGAGATCATCG, and sub-cloned into the XhoI-NotI sites in the pREV retroviral vector after Pspx1 and Not1 digestion (Pspx1 is compatible with Xho cloning site in the pRev vector) [57,58].

Protein Complex Purification
Flag-HA-MyoD complex purification from HeLa-MyoD cells was performed as described in: [31]. Briefly, we used retroviral transduction strategy to establish HeLa-S3 cell lines expressing double tagged Flag-HA-MyoD [31,57,58], or a control cell line transduced with the empty pREV vector has been established. We carried out double-affinity purification of Flag-HA-MyoD from HeLa cells ( Figure S1), using either nuclear soluble or chromatin fractions. For this, cells were resuspended in a hypotonic buffer (10 mM Tris-HCl pH 7.65; 1.5 mM MgCl 2 ; 10 mM KCl) and disrupted with 20 strokes of a tight-fitting Dounce homogenizer. The cytosolic fraction was separated from nuclei by 7 min centrifugation at 4uC at 9000 rpm. The nuclear soluble fraction was obtained by incubation of the nuclear pellet in a high salt buffer (900 mM NaCl, 20 mM Tris pH 7.65, 25% glycerol, 1.5 mM MgCl 2 , 0.2 mM EDTA), to get 300 mM NaCl, for 30 min at 4uC and centrifugation at 10,000 rpm. The resulting pellet, which corresponds to chromatin fraction, was resuspended and digested with micrococcal nuclease (Sigma, Saint-Quentin Fallavier, France), until it consisted primarily of mononucleosomes [57]. Nuclear soluble and chromatin fractions were then ultracentrifugated at 32000 rpm for 1 h at 4uC. Tagged-MyoD complex were then purified using anti-FLAG antibody immobilized on agarose beads (Sigma). After elution with the FLAG peptide (Ansynth, The Netherlands), the bound complexes containing nucleosomes were further affinity-purified on anti-HA antibody-conjugated agarose (Sigma) and eluted with the HA peptide (Ansynth, The Netherlands). Double-immunopurified complexes were resolved on 4-12% SDS-PAGE bis-Tris acrylamide gradient gel in MOPS buffer (Invitrogen), and stained using either the SilverQuest kit (Invitrogen, Cergy-Pontoise, France) [31], or with Colloidal blue (Invitrogen) for mass spectrometry (MS) analyses. In the latest, bands corresponding to proteins were cut from the gel, trypsin-digested using 0.4 mg of sequencinggrade trypsin (Promega, Charbonnières, France), and identified by MS analysis.
To purify CBFb complex from C2C12 cells, 3 grams of C2C12-CBFb cell pellet were used to purify tagged CBFb using a simpleaffinity purification method using Flag resin.  Figure 1C). Each IP was performed with 1.5 mg of total nuclear extracts and with 25 mL stock of ssDNA and BSA-pre-blocked Agarose Flag M2 resin from Sigma. IP was performed on wheel overnight at 4uC. Resin was then washed 5 times with TEGN buffer (20 mM Tris pH 7.65; 0.1 mM EDTA; 10% glycerol; 150 mM NaCl; 0.5% NP-40) and eluted by competition with high-purity Flag peptide at a final concentration of 0.2 mg/ml. The resin-free eluate was retrieved using Clean-up Post reaction columns (Sigma).

Immunoprecipitation of Endogenous Proteins
For IP, we usually use 2 mg of antibodies, 10 mL protein A/G Sepharose beads from Perbio and 1.2 mg of nuclear extracts from C2C12 cells, or 0.5 mg from primary myoblasts. Elution was performed with 40 ml of 0.1 M glycine pH 2.5, 15 min at 25uC, the eluate was recovered using Spin cleaning-up post-reaction column (Sigma). Acidity was neutralized with Tris pH 8.0 before adding loading buffer.

Plasmids, GST Fusions and GST Pull-Down
GST and GST-MyoD plasmid constructs were expressed in Escherichia coli strain BL21 and purified using glutathionesepharose beads according to the manufacturer (Sigma, Saint Quentin-Fallavier, France). Purified proteins were quantified by coomassie staining after SDS-PAGE separation. In vitro transcription and translation (TNT) of pcDNA3-Runx1 and its deletion mutants aa 1-189 and 1-242, pcDNA3-CBFb and luciferase were performed with Riboprobe in vitro transcription systems (Promega, Charbonnières, France) in the presence of 35 S-labelled methionine.
Agarose beads coated with equal amounts of GST or GST-MyoD (1 mg) were incubated with 10 mL of radioactive TNT reaction in reaction buffer (50 mM Tris pH 7.6, 150 mM NaCl, 0.1% Triton) during 2 h at 4uC. Beads were washed 5 times with wash buffer (50 mM Tris pH 7.6, 300 mM NaCl, 0.5% Triton 100), resuspended and proteins resolved by SDS-PAGE gel and revealed by autoradiography.

Immunofluorescence
Cells were cultured in Labtecks permanox (Falcon) and fixed briefly with 4% formaldehyde in PBS. Residual formaldehyde was neutralized with 0.1 M Glycine pH 8.0, and washed with PBS. Cells were permeabilized and blocked using 1% BSA, 1% goat serum, 0.3% Triton-X100 in PBS. Primary and secondary antibodies were diluted in the permeabilizing/blocking solution and were washed with 0.3% Triton-X100 in PBS. Nuclei are stained with DAPI and the glass lid is fixed using an anti-fading polymerizing media from DakoCytomation (Dako, Trappes, France).

FACS Analysis
C2C12 were transfected with the siRNAs as indicated in the Material and Method section. 48 hours post-transfection, cells were washed with PBS, then scraped in 500 mL of PBS. Cells were kept on ice while 4,5 ml of ethanol 70% were added. Then cells are kept at least overnight at 220uC. Propidium iodide (PI) staining proceeds as follows: cells are centrifuged and the pellet is washed with PBS. Cells are then centrifuged and resuspended in 2 mL PI solution (PI 25 ng/ml, RNase 200 ng/ml, Triton 0,1%) 30 minutes and kept in the dark. Cells are homogeneized by vortexing before analysis. We worked on a Beckman and Coulter FACS apparatus and we counted at least 3000 events for each condition.

Gene Reporter Assays
HeLa cells at 60% confluence were co-transfected by Calcium Phosphate co-precipitation with a myogenin promoter-driven Firefly luciferase reporter plasmid (kind gift of V. Sartorelli, NIH) without or with a fixed amount of MyoD expression vector (800 ng), and increasing amounts of vector expressing Runx1. The quantities of Runx1 expression vector used were: 0, 30, 60, 150, and 300 ng. The total amount of plasmid was normalized when necessary to 300 ng with the empty vector. A Renilla luciferase expression under a CMV (cytomegalovirus) promoter was used as a normalization control for the transfection. 24 h post-transfection, cells were lysed in a reporter lysis buffer (Promega, Charbonnières, France). Luciferase activity was determined using Dual-Luciferase Reporter Assay System (Promega). Firefly luciferase activity was then normalized to the level of Renilla luciferase and to the total protein amount.  Figure S7 Proposed model of skeletal muscle terminal differentiation regulation by CBF and MyoD. Note that in muscle system, proliferation inhibition or cell cycle exit is a pre-requisite to terminal differentiation. Thus, cell cycle exit regulators, such as p21 and cycD3, are activated early during terminal differentiation. Muscle specific early markers, such as myogenin, are also activated before late muscle markers. In proliferating myoblasts, Runx1/ CBFb (CBF dimer) proteins repress MyoD target genes, possibly via a direct interaction with MyoD. Thus, in proliferating myoblasts, CBF binds early MyoD target genes via Runx1 subunit and recruits chromatin modifying enzymes such as HDAC1, Suv39h1 and HP1. Upon triggering of terminal differentiation, CBF dissociates from the promoters and MyoD recruits activating chromatin modifying enzymes, such as HATs. HDAC1: Histone Deacetylase 1; HAT: Histone acetyltransferase. Found at: doi:10.1371/journal.pone.0009425.s007 (0.54 MB EPS) Figure S8 Control of the anti-HA western blot results presented on Figure 1C using antibodies against native proteins: MyoD, CBFb and Runx1. Found at: doi:10.1371/journal.pone.0009425.s008 (0.53 MB EPS)