|Home | About | Journals | Submit | Contact Us | Français|
Direct generation of a homogeneous population of skeletal myoblasts from human embryonic stem cells (hESCs) and formation of tri-dimensional contractile structures for in dish disease modeling is a current challenge in regenerative medicine. Previous studies reported on the generation of myoblasts from ESC-derived embryoid bodies (EB), but not from undifferentiated ESCs, indicating the requirement for mesodermal transition to promote skeletal myogenesis. Here we show that selective absence of the SWI/SNF component BAF60C (encoded by SMARCD3) confers on hESCs resistance to MyoD-mediated activation of skeletal myogenesis. Forced expression of BAF60C enables MyoD to directly activate skeletal myogenesis in hESCs, by instructing MyoD nuclear positioning and allowing chromatin remodelling at target genes. BAF60C/MyoD-expressing hESCs are epigenetically committed myogenic progenitors, which bypass the mesodermal requirement and, when cultured as floating clusters, give rise to contractile tri-dimensional myospheres composed of skeletal myotubes. These results identify BAF60C as key epigenetic determinant of hESC commitment to the myogenic lineage, and establish the molecular basis for the unprecedented generation of hESC-derived myospheres exploitable for “in dish models” of muscular diseases.
Generation of tri-dimensional (3D) structures that recapitulate histological and functional properties of adult organs and tissues is a current challenge in regenerative medicine, as it requires abundant and homogeneous population of committed progenitors from hESCs. For instance, formation of hESC-derived contractile 3D skeletal muscles has never been reported, reflecting the unsuccessful attempts to directly generate homogeneous populations of skeletal muscle progenitors from undifferentiated hESCs. Previous works demonstrated that generation of skeletal myoblasts could be achieved from EB-derived mesodermal cells (Darabi et al., 2012; Darabi et al., 2008; Iacovino et al., 2011) or from mesenchymal derivatives of ESCs (Barberi et al., 2007; Goudenege et al. 2012), indicating the requirement for transition through the mesodermal stage to activate skeletal myogenesis. This evidence suggests an “intrinsic” resistance to the activation of the myogenic program in ESCs, prior to mesoderm formation. Indeed, no evidence of direct generation of skeletal muscle progenitors from undifferentiated hESCs has been reported so far, raising the question of whether hESCs can be directly reprogrammed into skeletal muscle cells. Direct myogenic conversion of somatic non-muscle cells upon introduction of MyoD has provided the first evidence that a single tissue-specific transcriptional activator is able to reprogram the nucleus of host cells into the skeletal muscle phenotype (Weintraub et al., 1989; Gerber et al., 1997; Chambers and Studer, 2011). However, it is currently unknown whether ESCs can be reprogrammed into skeletal myoblasts by ectopic expression of MyoD.
We tested the ability of hESCs to undergo direct myogenic conversion in response to the ectopic expression of MyoD, as compared to human fibroblasts that typically convert into skeletal muscle cells upon MyoD expression (Weintraub et al., 1989). Surprisingly, expression of MyoD in hESCs failed to activate skeletal myogenesis under the same culture conditions in which massive myogenic conversion was observed upon expression of MyoD in H27 human fibroblasts (Fig. 1A and Fig. S1A). Resistance to MyoD-induced myogenic conversion was observed in different hESC lines tested (H9 and H1) (Fig. S1B). Figure 1A shows that ectopic MyoD was expressed at comparable levels and was properly localized in the nuclei of hESCs as well as fibroblasts, but failed to activate muscle gene expression only in hESCs (Fig. 1A and Fig. S1), suggesting that hESCs might have a “nuclear landscape” not permissive for activation of the myogenic program. We reasoned that this could be due to a deficiency in key component(s) of the chromatin-modifying machinery that enables MyoD to activate transcription from target genes. We therefore analyzed the integrity of the chromatin-modifying complexes that are known to enable MyoD-activated gene expression (de la Serna et al., 2006; Guasconi and Puri, 2009; Sartorelli and Juan, 2011). We measured by quantitative RT-PCR (Q-PCR) the expression levels of key components of these complexes in distinct hESC lines (H1, H7 and H9) and human H27 fibroblasts, and found that the large majority of them were expressed in all cell lines analyzed, with the notable exception of the structural subunits of the SWI/SNF chromatin remodelling complex, BAF60C (encoded by SMARCD3) and the alternative SWI/SNF ATPase BRM (Fig. S2). Lack of Baf60c and Brm proteins in mouse ESC (mESCs) was previously observed by Crabtree and colleagues (Ho et al., 2009). However, while Baf60c is absolutely required for the activation of skeletal myogenesis (Forcales et al. 2011; Lickert et al. 2004), Brm appears dispensable (Reyes et al., 1998), with Brg1 being the essential ATPase of the SWI/SNF complex that remodels the chromatin at MyoD-target genes (Ohkawa et al. 2006; Forcales et al. 2011). Three alternative variants of the BAF60 subunits – BAF60A, B and C – are expressed in somatic cells, with mutually exclusive presence in distinct SWI/SNF complexes (Debril et al., 2004; Wang et al., 1996). Of note, while all three variants are expressed in human skeletal myoblasts and fibroblasts, which are competent to activate the myogenic program, expression of BAF60A and B, but not BAF60C, was detected in hESCs, which are resistant to the activation of the myogenic program (Fig. 1B). Interestingly, BAF60C was induced upon EB formation (Fig. 1C) – a stage permissive for the activation of skeletal myogenesis (Iacovino et al., 2011). Two BAF60C isoforms are expressed in human somatic cells – BAF60C1 and BAF60C2. However, only BAF60C2 was significantly upregulated during EB formation (Fig. 1C). Thus, we hypothesized that BAF60C2 could be a limiting factor for the direct activation of the myogenic program in hESCs, and tested whether forced expression of BAF60C2 in hESCs could restore MyoD-mediated activation of the myogenic program. Fig. 1D and E show that ectopic expression of BAF60C2 enabled MyoD-directed conversion of hESCs into myosin heavy chain (MyHC)-positive muscle cells.
Because the efficiency of myogenic conversion in BAF60C2/MyoD-expressing hESCs was reduced, as compared to the myogenic conversion of MyoD-expressing fibroblasts (Fig. 1E), we considered the possibility that hESCs could receive optimal differentiation cues from their physiological context - that is, the network of signals generated by cell aggregation, such that occurring within EB (Keller, 1995). It is well established that cell-to-cell interactions promote intracellular pro-myogenic signalling (Krauss 2010). To this purpose, we devised a protocol for generation of skeletal muscle cells from hESCs that included the transition through an aggregation stage (EB-like clusters) equivalent to the standard culture conditions that typically promote EB from hESCs (see scheme in Fig. 2A and Experimental Procedures for details). We generated hESCs (H9 line) stably expressing BAF60C2 and MyoD (hereby referred to as hESCBAF60C2/MyoD) and cultured them in conditions permissive for aggregation into floating clusters for 5 days (pre-EB conditions), followed by dissociation into single cells that were subsequently exposed to standard myogenic differentiation medium (post-EB condition) - see scheme in Fig. 2A. Of note, we tested various combinations of expression timing of MyoD and BAF60C2 in different hESC lines, and observed the activation of the myogenic program only when BAF60C2 was expressed prior to MyoD (data not shown). This protocol led to a dramatic enhancement in the activation of the myogenic program in hESCBAF60C2/MyoD, with the generation of a homogeneous population of committed cells (Fig. 2B - pre-EB conditions) that became competent to activate skeletal myogenesis upon aggregation into EB-like clusters, leading to massive formation of myogenin/MyHC-positive myotubes after dissociation and exposure to DM conditions (Fig. 2B – post-EB conditions – and Fig. 2C). Under these conditions the percentage of myogenic conversion of hESCBAF60C2/MyoD was much higher (60.1±5% MyHC-positive cells) (Fig. 2B) than that observed in standard conditions for myogenic conversion shown in Fig. 1D (4.67%±0.8% MyHC-positive cells). Sporadic formation of myogenin-positive cells was observed in MyoD-expressing H9 hESCs (hESCMyoD) cultured in EB medium (Fig. 2B and C), possibly due to localized upregulation of endogenous BAF60C2, or by the compensatory activity of BAF60B, as previously described for cardiomyogenesis (Takeuchi and Bruneau, 2009). However, only hESCBAF60C2/MyoD differentiated into MyHC-positive myotubes with high efficiency after culture in EB conditions (60.1±5% MyHC-positive cells from hESCBAF60C2/MyoD versus 2.5±0.5% from hESCMyoD). Interestingly, forced expression of BAF60C2 alone in hESCs (hESCBAF60C2) could not activate the skeletal myogenesis in the absence of MyoD (Fig. 2B), indicating that BAF60C2 requires the presence of MyoD to activate muscle gene expression. These results indicate a multi- step progression of hESCBAF60C2/MyoD through sequential stages of cellular differentiation, including lineage determination by MyoD and BAF60C2 (pre-EB stage), followed by competence to respond to differentiation signals and formation of terminally differentiated myotubes (post-EB stages). Previous works demonstrated the key role of BAF60C2 in at least two essential epigenetic events underlying myogenic differentiation, such as MyoD binding to the chromatin of target genes and the signal dependent recruitment of BRG1-based SWI/SNF chromatin remodeling complex (Forcales et al. 2011). We used chromatin immunoprecipitation (ChIP) to investigate the sequential recruitment of MyoD, BAF60C2 and BRG1 to MyoD target genes in hESCBAF60C2/MyoD or hESCMyoD or control PGK infected hESCs (hESCPgk) at pre-EB and post-EB stages. We focused on human MYOGENIN gene, since activation of MYOGENIN transcription is invariably required for the execution of the myogenic program during development and post-natal life (Hasty et al., 1993; Ohkawa et al., 2006). Moreover, the sequential recruitment of MyoD/BAF60C complex, followed by BRG1-based SWI/SNF complex, has been demonstrated on myogenin promoter in skeletal myoblasts (Forcales et al. 2011). In the absence of BAF60C2 (hESCMyoD), MyoD could not bind the MYOGENIN promoter either in pre-EB conditions or after exposure to the differentiation signals within EB-like clusters (post-EB conditions) (Fig. 2D). Forced expression of BAF60C2 (hESCBAF60C2/MyoD) enabled MyoD recruitment to MYOGENIN promoter, together with BAF60C2, but not BRG1 in pre-EB conditions (Fig. 2D). Formaldehyde-Assisted Isolation of Regulatory Elements (FAIRE) showed that BAF60C2 and MyoD binding to MYOGENIN promoter correlated with incipient chromatin accessibility in correspondence of the PBX-binding site (Berkes et al., 2004) in hESCBAF60C2/MyoD, as compared to control lines in pre-EB conditions (Fig. 2E). This evidence indicates an instructive role of BAF60C2 in the early recognition of MyoD target sequences. Moreover, only in hESCBAF60C2/MyoD BRG1 recruitment to MYOGENIN promoter was detected in response to differentiation signals (post-EB conditions) (Fig. 2D). The recruitment of BRG1-based SWI/SNF complex correlated with a dramatic enhancement in chromatin accessibility that was extended to the MEF2/Ebox sites (Fig. 2E) and the engagement of the elongation-competent (serine 2 phosphorylated) form of Polymerase II (Pol II ser2P) (Fig. 2F), leading to activation of transcription (Fig. 2C). Interestingly, serine 5 phosphorylated Pol II (Pol II ser5P) was detected at the transcription start site (TSS) of MYOGENIN promoter in hESCs regardless the presence of MyoD and/or BAF60C2 (Fig. 2F), according to the “poised” conformation of the chromatin at tissue-specific genes previously described in ESCs (Azuara et al., 2006; Boyer et al., 2006; Lee et al., 2006; Mikkelsen et al., 2007). Of note, BAF60C2 and BRG1 were not detected on the enhancer of the cardiac-specific gene NKX2.5 in hESCBAF60C2/MyoD (Fig. 2d). NKX2.5 is induced by BAF60C2 and the transcriptional activators GATA4 and TBX5 in cardiac progenitors (Lickert et al., 2004; Takeuchi and Bruneau, 2009) and was expressed in control hESCPgk and hESCMyoD, but not in hESCBAF60C2/MyoD (see Fig. 3A). Consistently, NKX2.5 enhancer showed a closed chromatin conformation in hESCBAF60C2/MyoD, while in hESCPgk and hESCMyoD an increased accessibility was detected at GATA4 binding sites (Fig. 2E). This evidence further supports the conclusion that BAF60C2 enables selective activation of skeletal muscle genes by MyoD. The sequential and dynamic chromatin modifications shown in Figure 2D-F define two distinct stages that reflect a “silent” epigenetic commitment to the myogenic lineage (pre-EB conditions) followed by the transcriptional activation of muscle genes in post-EB conditions.
The direct activation of muscle gene expression by MyoD in hESCBAF60C2/MyoD is consistent with the nuclear reprogramming toward the skeletal muscle lineage, which is typically accompanied by the silencing of pluripotency genes and repressing alternative cell fates. Indeed, pluripotency genes were readily repressed in hESCs upon MyoD and BAF60C2 expression (data not shown), likely reflecting the functional antagonism between tissue-specific transcriptional activators and pluripotency. Similarly, markers of the alternative lineages within the three germ layers were repressed in hESCBAF60C2/MyoD as compared to hESCPgk (Fig. 3A). Most of these markers showed a peak of expression between days 3 and 5 in hESCPgk-derived EBs, but not in hESCBAF60C2/MyoD-derived EBs (Fig. 3A). Interestingly, while the expression pattern of many of these genes was partly overlapping in hESCMyoD- and hESCBAF60C2/MyoD-derived EB-like clusters during the first days of cultures in EB conditions, a late increase (between days 4 and 5) in the expression of lineage- specific markers, such as NKX2.5, CERBERUS1 and NESTIN, was observed in hESCMyoD-derived aggregates (Fig. 3A). This might reflect that ability of MyoD to repress pluripotency (Watanabe et al., 2011), via a BAF60C2-independent mechanism that leads to spontaneous commitment to various lineages by default (S.A. and P.L.P. unpublished data). Importantly, primitive markers of mesoderm (BRACHYURY T, MESOGENIN and MESP1) were specifically repressed in hESCBAF60C2/MyoD-derived EBs (Fig. 3A). The only mesodermal gene that showed an earlier and more robust pattern of expression in hESCBAF60C2/MyoD-derived EB-like clusters, as compared to hESCPgk- and hESCMyoD-derived clusters, was the skeletal muscle progenitor marker PAX3, which establishes the myogenic identity within the paraxial mesoderm (Buckingham and Relaix, 2007) (Fig. 3A). This is consistent with the selective activation of the skeletal muscle program in hESCBAF60C2/MyoD. Figure 3B shows that hESCBAF60C2/MyoD-derived EBs uniformly express PAX3 (78.5±6%), while the primitive mesoderm marker BRACHYURY T could not be detected. By contrast, PAX3-positive cells were detected at much lower frequency in hESCPgk-and hESCMyoD-derived EB- like clusters (10.0±2.4% and 22.2±3%, respectively), most of which co-expressed BRACHYURY T (Fig. 3B), possibly reflecting the myogenic commitment of a subset of mesodermal derivatives. Thus, BAF60C2 and MyoD appear to generate an “epigenetic landscape” that imposes the selective activation of muscle gene expression in hESCs, leading to their direct conversion into skeletal myoblasts without the transition through the mesodermal stage. This evidence is well supported by the recruitment of BAF60C2 and BRG1 and activation of the MYOGENIN promoter, but not the NKX2.5 enhancer, in hESCBAF60C2/MyoD (Fig. 2D-F).
To further demonstrate the homogeneous composition of hESCBAF60C2/MyoD by skeletal muscle progenitors, we dissociated hESCBAF60C2/MyoD-derived EB-like clusters (as compared to the equivalent population from hESCPgk and hESCMyoD) and sorted them by FACS, using the surface marker CD56 (NCAM1), which has previously been used to isolate human skeletal myoblasts (Zheng et al., 2007)28. Surprisingly, we found that the large majority of cells from dissociated hESCBAF60C2/MyoD–derived floating aggregates, as well as hESCPgk and hESCMyoD were NCAM1 positive. This result reveals that NCAM1 expression cannot define by itself a population of skeletal muscle progenitors from hESCs, and indicates that NCAM1 is probably expressed in cells undergoing transition toward a variety of lineages (Evseenko et al., 2010). However, when we measured the relative expression levels of NCAM1 in our hESC populations as compared to those of human primary skeletal myoblasts (HSkM), we found that hESCBAF60C2/MyoD were enriched in cells with expression levels (73%) of NCAM1 that were higher than of hESCPgk (22%) and hESCMyoD (54%) and similar to those detected in HSkM (data not shown). We therefore gated the high NCAM1-expressing cells among the population of NCAM1-positive FACS-sorted hESCs (Fig. 3C) and measured their myogenic potential in vitro, as percentage of myogenin/MyHC double positive multinucleated cells formed among the same number of cells cultured in differentiation medium (Fig. 3D). Figure 3D shows that the high NCAM positive cells sorted from hESCBAF60C2/MyoD–derived EB-like clusters evenly differentiated in myogenin/MYHC double positive multinucleated myotubes (81.7±3%), while only a minority of cells from hESCMyoD–derived aggregates (4.3±0.7%) and none of the cells from hESCPgk–derived bodies could differentiate into myotubes under the same conditions. This result demonstrates that hESCBAF60C2/MyoD–derived EB-like clusters are composed of a homogeneous population of skeletal myoblasts that retain the ability to differentiate into skeletal myotubes.
We therefore hypothesized that such a homogeneous population of hESC-derived myoblasts could bias the composition of EB-like clusters toward the formation of tri-dimensional (3D) contractile structures enriched in skeletal myofibers (skeletal myospheres). We devised a specific protocol by culturing hESCBAF60C2/MyoD (or hESCPgk and hESCMyoD) in EB medium for 5 days, followed by the exposure of derived EB-like clusters to myogenic differentiation medium for additional 14 days (see scheme in Fig. 4A). This protocol differs from the one shown in Fig. 2A, as it does not include dissociation of floating EB-like clusters and is therefore permissive for the formation of 3D structures. Only hESCBAF60C2/MyoD gave rise to EB-like structures that were fully enriched in myogenin/MyHC positive myofibers (Fig. 4B and C) and expressed typical skeletal muscle markers, such as myogenin, MyH3 and MyH8, but not cardiac-specific genes, such as GATA4 (Fig. 4D). hESCMyoD–derived EBs showed rare areas, rather than diffuse formation, of myogenin/MYHC positive cells (Fig. 4B and C), possibly reflecting localized concentrations of EB-derived signals, such as morphogenes, that might induce endogenous BAF60C2 expression in MyoD-expressing cells. Importantly, hESCBAF60C2/MyoD–derived EB-like structures underwent spontaneous contraction at very high frequency, as it was detected in about 20% of the EBs during 60 second-observation of each plate (Suppl. Movie 1 – snapshot in Fig. 4E - and Fig. S3A). We therefore defined them as contractile “myospheres”. Dissociation of hESCBAF60C2/MyoD–derived myospheres resulted in the formation of myogenin/MyHC positive myofibers (Fig. S3B). We investigated the contractile activity of hESCBAF60c2/MyoD–derived myospheres. We used dynamic distribution of the cytosolic calcium in live cells to discriminate skeletal muscle type contraction from cardiac type contraction behavior. Fig. 4F, Suppl. Movie 2 (snapshot in Fig. 4G), show that hESCBAF60c2/MyoD–derived myospheres are exhibiting the slow-rising long-lasting increase in local calcium concentration typical of skeletal muscle type calcium artifacts (Capes et al., 2011). Individual fluorescent Fluo-4 signals from either intact myospheres derived from hESCBAF60C2/MyoD or dissociated cells from myospheres during 20 sec recordings show intracellular calcium dynamics typically reflecting skeletal muscle contractions (Fig. 4F – compare blue and red lines) that are clearly distinguished from cardiac contraction contractions (Fig. 4F – black line) (see also Movie 2 in Supplemental Materials). The finding that hESCBAF60C2/MyoD–derived myospheres either intact or dissociated showed a pattern of calcium transient similar to control skeletal myotubes demonstrates that the vast majority of contracting cells in hESCBAF60c2/MyoD–derived myospheres exibit skeletal muscle contractile activity (Fig. 4F-G). This evidence, together with the lack of activation (Fig. 2) and expression (Fig. 3 and and4)4) of cardiac markers, conclusively demonstrates that BAF60C2 and MyoD are sufficient to convert hESCs into a homogeneous population of committed myoblasts that, when cultured in EB conditions, generate 3D contractile myospheres. These data emphasize the importance of imposing an epigenetic landscape that commits hESCs to myogenic progenitors and make them competent to respond to the signals derived from EB-like structures (myobodies), ultimately leading to the formation of spontaneously contracting myospheres (Fig. 4H).
Collectively, these data demonstrate that the requirement for transition through the mesodermal stage to activate myogenesis in hESCs can be bypassed by forced expression of BAF60C2 and MyoD. The identification of BAF60C2 as a limiting factor for the epigenetic reprogramming of hESCs to the myogenic lineage is of particular relevance, as it provides the molecular explanation of the previously reported requirement of mesodermal transition to activate skeletal myogenesis in ESCs (Darabi et al., 2012; Darabi et al., 2008; Iacovino et al., 2011). The evidence that BAF60C2 instructs tissue-specific transcriptional activators (such as MyoD) to activate lineage specific gene expression suggests that repression of BAF60C2 in undifferentiated hESCs is a necessary event to maintain pluripotency, presumably to prevent the formation of an epigenetic landscape permissive for commitment to the myogenic lineage. Along this line, we speculate that BAF60C2 de-repression in hEBs (mimicked in our experimental conditions by hESCBAF60c2/MyoD) might confer on hESCs the competence to properly respond to complex intracellular events elicited by external signals (e.g. activation of p38 and AKT pathways) (Serra et al. 2007) and developmental cues, such as TGFbeta-SMAD2/3 signalling (Capes et al., 2011) that enable MyoD to activate muscle gene expression. Overall, the ability to epigenetically reprogram hESCs into a homogenous population of skeletal muscle progenitors that can generate 3D contractile myospheres provides an unprecedented opportunity to establish “in dish” models of study of skeletal muscle physiology and mechanism of diseases.
Undifferentiated hESCs, H9 and H1 (passages 35-45), were cultured and manipulated as indicated in Supplemental Experimental Procedures.
Calcium transients hESC cell clusters were recorded using the IC100 High Content Imaging system equipped with the KIC (Kinetic Image Cytometer) electrical field stimulation (EFS) module. For labeling and recording see Supplemental Experimental Procedures.
See Supplemental Experimental procedure for the details and antibody used. EBs were prepared according the protocol described in Gomes et al., JOVE
Total RNA was isolated with Trizol and retro-transcribed using Reverse transcription reagent (Applied Biosystems). qRT-PCR was performed according the manufacturer’s instructions on Mx3000P (Stratagene) using SYBR Green Master Mix. Data were normalized to the expression of GAPDH gene (for SYBR Green) and relative quantification was calculated by the Comparative Ct method. Primers used are listed in Table S1.
Chromatin for FAIRE (formaldehyde-assisted isolation of regulatory elements) analysis was prepared as previously described in Simon et al., 2012. 10 ng of FAIRE purified DNA or 10 ng of purified input DNA was analyzed by qPCR. The signal of FAIRE purified chromatin is presented as a fraction of input purified chromatin. See details in Supplemental Experimental Procedures. Primers used are listed in Table S1.
ChIP was performed as described in Forcales et al., 2011 with some modifications (see Supplemental Experimental Procedures for details). Chromatin extracts were immunoprecipitated overnight on rotating platform at 4°C with the following antibodies: anti-MyoD (Santa Cruz, sc-760) anti-Brg1 (Santa Cruz, sc-17796X), custom made anti-BAF60c antibody13, anti-Pol II ser5P (39233, active motif) and anti-Pol II ser2P (abcam, ab24758) with IgG-IgM linker antibody (Abcam, ab9175) or normal IgG as control. Primers used are listed in Table S1.
Data are presented as mean ± SEM (standard error of the mean) unless differently indicated. Differences between groups were analyzed for statistical significance using unpaired the Student’s t-test with significance being defined as * p<0.05; ** p<0.01; *** p<0.001.
PLP is an Associate Investigator of Sanford Children’s Health Research Center. This work has been supported by the following grants to PLP: R01AR056712, R01AR052779 and P30 AR061303 from the National Institute of Health/National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS), MDA and Sanford Children Health Research award. This work has partly benefited from research funding from the European Community’s Seventh Framework Programme in the project FP7-Health – 2009 ENDOSTEM 241440 (Activation of vasculature associated stem cells and muscle stem cells for the repair and maintenance of muscle tissue). SA was supported by CIRM fellowship. We thank Natalie Prigozhina for helping with EB recording.
AUTHOR CONTRIBUTIONS SA performed most of the experiments of the manuscripts with the help of PC; LG performed the FACS-sorting on NCAM1-pos hESCs; BM performed FAIRE assay; AS performed assay of intracellular calcium recording in living cells. SF provided the first evidence that BAF60 was necessary for MyoD-mediated activation of gene expression in embryonic stem cells. PLP conceived the project and the experiments, analyzed the data and wrote the manuscript. All the authors discussed, commented the results and read the manuscript.
CONFLICT OF INTEREST The authors declare that they have no conflict of interests.