|Home | About | Journals | Submit | Contact Us | Français|
During muscle regeneration, the mechanism integrating environmental cues at the chromatin of muscle progenitors is unknown. We show that inflammation-activated MKK6-p38 and IGF1-induced Pi3K/AKT pathways converge on the chromatin of muscle genes to target distinct components of the muscle transcriptosome. p38 α/β kinases recruit the SWI/SNF chromatin-remodeling complex; AKT 1 and 2 promote the association of MyoD with p300 and PCAF acetyltransferases, via direct phosphorylation of p300. Pharmacological or genetic interference with either pathway led to partial assembly of discrete chromatin-bound complexes, which reflected two reversible and distinct cellular phenotypes. Remarkably, Pi3K/AKT blockade was permissive for chromatin recruitment of MEF2-SWI/SNF complex, whose remodeling activity was compromised in the absence of MyoD and acetyltransferases. The functional interdependence between p38 and IGF1/Pi3K/AKT pathways was further established by the evidence that blockade of AKT chromatin targets was sufficient to prevent the activation of the myogenic program triggered by deliberate activation of p38 signaling
Muscle regeneration occurs at the expense of myogenic progenitors (e.g. satellite cells), which are activated in response to myofiber injury (Wagers and Conboy 2005). The extensive genomic re-programming that occurs in the nuclei of satellite cells during the transition from quiescence to terminal differentiation is reflected in the dramatic changes of the chromatin at specific loci. For instance, the chromatin conformation at muscle-specific loci is repressive in undifferentiated, proliferating myoblasts, but becomes permissive for transcription at the onset of the differentiation program (Sartorelli and Caretti 2005; Palacios and Puri 2006). In contrast, the conformation of chromatin at the regulatory regions of proliferation genes is permissive for transcription in myoblasts, but precludes their expression in terminally differentiated myotubes (Ait-Si-Ali et al. 2004).
Satellite cell-mediated muscle regeneration is accompanied by the local release of several paracrine substances, e.g. cytokines, growth factors and hormones, as well as by cell-to-cell interactions, which are triggered upon muscle injury and the ensuing inflammatory response (Charge and Rudnicki 2004). These environmental cues govern satellite cell transition from quiescence to terminal differentiation by imparting to the chromatin of muscle loci the changes underlying this progression (Forcales and Puri 2006; Berkes and Tapscott 2005). Despite extensive knowledge of the intracellular cascades that transmit external cues to the nucleus, the molecular mechanism by which they are converted into chromatin modification at discrete loci remains largely unknown.
Recent studies have begun to analyze the composition of the transcriptosome that is assembled on the chromatin of muscle genes in response to the activation of the p38 kinases – the effectors of a pathway elicited in satellite cells by regeneration cues (Keren et al. 2006; Lluis et al. 2006). These studies have revealed that the p38 pathway promotes the assembly of a transcription-competent transcriptosome by recruiting the chromatin remodeling SWI/SNF complex to the regulatory regions of muscle genes (Simone et al. 2004a). Moreover, the p38 pathway regulates additional and related events, such a MyoD/E47 interactions (Lluis et al. 2005), MEF2 phosphorylation and activity (Zhao et al. 1999; Zetser et al. 1999; Ornatsky et al. 1999; Wu et al. 2000) and the RNA stability of selected myogenic transcripts (Briata et al. 2005).
Other signaling pathways elicited by regeneration cues cooperate with the p38 pathway in regulating the expression of genes implicated in the control of satellite cell differentiation. Among them, Pi3K/AKT signaling mediates satellite cell response to growth factors (e.g. IGF1) that promote critical events in the regeneration process, such as proliferation, muscle gene expression, myoblast fusion, survival and post-mitotic growth of myotubes (Musaro et al. 1999; Lawlor and Rotwein 2000; Rommel et al. 2001). Evidence of the functional impact of the IGF1-Pi3K-AKT pathway on muscle regeneration is also provided by in vivo studies (Musaro et al. 2001; Barton et al. 2002).
A number of downstream targets of the IGF1-Pi3K-AKT pathway have been identified (Sartorelli and Fulco 2004). However, the mechanism by which the IGF1-Pi3K-AKT pathway influences chromatin structure and chromatin-bound complexes of target genes in myoblasts is unknown.
Our previous studies showed that p38 and IGF1-Pi3K-AKT pathways proceed as two parallel promyogenic cascades in myoblasts induced to differentiate (Wu et al. 2000). Here we show that these two pathways converge at the chromatin level to control the assembly of the myogenic transcriptosome, by targeting two pharmacologically separable, yet functionally interdependent, events.
We used satellite cells derived from intact myofibers to investigate the impact of pharmacological blockade of p38 and IGF1/Pi3K on muscle regeneration. Upon culturing intact myofibers in mitogen rich medium (growth medium – GM), satellite cells delaminate and proliferate. Subsequent cell-to-cell contact together with mitogen withdrawal (differentiation medium – DM) promotes their differentiation into multinucleated myotubes. Pharmacological blockade of the p38 kinases α and β, by SB203580 (SB), or the IGF-1/Pi3K/AKT pathway, by the Pi3K inhibitor LY294002 (LY), was achieved by the continuous exposure to the inhibitors throughout the transition from myofiber-associated satellite cells to myoblasts cultured in DM. Inhibition of either pathway prevented the expression of muscle-specific genes and precluded satellite cell fusion into myotubes with comparable efficiency (Fig. 1A and B). The same effects were also observed in established muscle cell lines, such as C2C12 cells (Suppl. Fig. 1A and B) and in primary cultures of human skeletal myoblasts (data not shown). However, the impact of SB and LY on the cell cycle machinery was different; only p38 blockade prevented the down-regulation of cyclin A that typically occurs during myotube formation (Fig. 1B). Furthermore, the undifferentiated myoblasts derived from the exposure to SB or LY during the incubation in DM showed striking morphological differences; SB-treated satellite cells appeared as elongated myoblasts, while LY-treated cells displayed rounded morphology (Fig. 1A) Both phenotypes could be readily distinguished from the typical morphology of normal, satellite cells. Comparable morphological effects were observed after simultaneous downregulation of individual p38 α and β or AKT 1 and 2 kinases, respectively, by RNA interference (RNAi) (Suppl. Fig. 1C, D, E and F). Thus, despite the shared ability to prevent the initiation of the differentiation program, pharmacological or genetic blockade of p38 and Pi3K resulted in biochemical and morphological differences, which likely reflect discrete chromatin signatures at specific loci. Consistent with this notion, we detected two unique patterns of histone acetylation at muscle loci in SB- vs LY-treated satellite cells (Fig. 1C) that reflect the assembly of two distinct chromatin-bound complexes with different enzymatic activity. Only Pi3K blockade prevented the local increase in histone acetylation on the regulatory elements of muscle genes (Fig. 1C), despite the global histone acetylation did not change in LY-treated satellite cells (Fig. 1D).
The results presented in Fig 1 suggest that the Pi3K/AKT and p38 pathways target different components of the muscle transcriptosome, and that the selective inhibition of either cascade can generate two populations of undifferentiated myoblasts showing distinct cellular phenotypes and discrete patterns of chromatin modification at muscle loci.
Based on this hypothesis, we carried out an extensive chromatin immunoprecipitation (ChIP) analysis to compare the assembly of the myogenic trancriptosome in myoblasts induced to differentiate in the presence of SB vs LY. We employed C2C12 myoblasts for this analysis, as they provided ample material necessary for ChIP. The chromatin of myoblasts in GM or DM, with and without SB or LY, was immunoprecipitated with antibodies against different components of the muscle transcriptosome and specific histone modifications. Consistent with Pi3K/AKT and p38 being parallel cascades delivering distinct external cues to discrete components of the myogenic transcriptosome, the composition of chromatin-bound complexes on myogenin promoter and muscle creatine kinase (MCK) enhancer was different in myoblasts exposed to SB or LY. SB selectively prevented the recruitment of the SWI/SNF complex, as previously reported (Simone et al 2004a). LY precluded the recruitment of p300 and PCAF acetyltransferases, and the consequent hyperacetylation on muscle promoters/enhancers (Fig. 2A). As hyperacetylation of MyoD increases its DNA binding affinity (Sartorelli et al. 1999), it was not surprising that in the absence of acetyltransferases, the majority of MyoD was hypoacetylated and hence unable to bind chromatin (Fig. 2A). By contrast, chromatin binding of MEF2A/C and SWI/SNF was not affected by LY treatment (Fig. 2A). The relative occupancy of Brg1, p300, PCAF, MyoD and the acetylation status of histone H3 were also quantified by real time PCR in independent experiments, which confirmed the differences reported above (Suppl. Fig. 2A and B). Interestingly, p38 α kinase was detected on the chromatin of target genes only in conditions permissive for differentiation (DM), regardless of the presence of SB or LY (Fig. 2A). In these conditions the large majority of p38 α is phosphorylated (Wu et al. 2000), suggesting the MKK3/6-dependent phosphorylation of p38 could be an important signal for chromatin recruitment. The presence of p38 on the chromatin of muscle genes in SB-treated cells is consistent with this scenario, as SB blocks the catalytic activity of p38α, without altering the phosphorylation status of this kinase. Notably, the effect of Pi3K blockade on the composition of chromatin-bound complexes appeared to be restricted to those muscle genes, whose expression was inhibited by LY. Indeed, p300 recruitment and histone hyperacetylation on the promoter of Atrogin-1, a catabolic gene repressed by IGF1 pathway (Stitt et al. 2004; Sandri et al. 2004), was not reduced, but rather increased in LY-treated myoblasts (Suppl. Fig. 2C and D).
The comparison of the effects produced by SB vs LY emphasizes the notion that these two cascades converge on the chromatin of muscle genes to target distinct components of the myogenic transcriptosome. The presence of the SWI/SNF complex in association with a transcriptional activator (MEF2) on the chromatin of muscle genes in LY-treated myoblasts raises the question of whether this complex is competent to remodel the chromatin at these loci. To address this issue we performed an endo-nuclease accessibility assay and found that, despite the different composition of chromatin-bound complexes in SB- and LY-treated myoblasts, both treatments precluded chromatin remodeling (Fig. 2B). This evidence demonstrates that the MEF2-SWI/SNF complex detected on the regulatory elements of muscle genes in LY-treated cells is not competent to remodel the chromatin.
Taken together these data reveal a functional interdependence between the p38 and Pi3K pathways at the chromatin level. The p38 pathway directs the recruitment of the SWI/SNF complex, but the ability of this complex to remodel the chromatin is conferred by the IGF-1/Pi3K pathway, via engagement of MyoD/acetyltransferases and consequent local hyperacetylation. In this regard, it is difficult to discriminate the role of MyoD as an essential factor for the recruitment of acetyltransferases (Yuan et al. 1996; Eckner et al. 1996; Puri et al. 1997a and b; Sartorelli et al. 1997) or as an independent regulator of SWI/SNF remodeling activity (Gerber et al. 1997; de la Serna et al. 2001).
Previous reports have documented the ability of MEF2 to form distinct complexes with transcriptional repressors (Miska et al. 1999; Lu et al. 2000). Consistently, immunoprecipitated MEF2A/C co-purified significantly higher deacetylase activity in LY-treated myoblasts, as compared to untreated or SB-treated myoblasts (Fig. 2C), and HDAC4 was found to be associated to MEF2A/C and SWI/SNF on the chromatin of muscle genes in LY-treated myoblasts (Fig. 2A). The reduced histone acetylation observed at the regulatory elements of muscle genes following Pi3K inhibition raised the possibility that these lysines were instead methylated. Recruitment of Polycomb-associated methyltransferase Ezh2 was previously reported to mediate silencing of muscle genes in undifferentiated myoblasts (Caretti et al. 2005). Indeed, the presence of the methyltransferase Ezh2 and its enzymatic effect - tri-methylation of H3 lysine K27 - was detected on the promoter of myogenin and the MCK enhancer in LY-treated myoblasts (Suppl. Fig. 3A). This finding suggests that the pharmacological blockade of Pi3K converts the conformation of the chromatin at muscle loci from permissive to repressive. The drastic changes in the components of the chromatin-bound complexes generated by LY treatment, which lead to histone lysine methylation, rather than acetylation and chromatin remodeling, also suggest that, in the absence of the IGF1 signaling, the MEF2-SWI/SNF complex associated with co-repressors could impose repressive chromatin modifications. We next tested whether the inhibition of the myogenic program by IGF1-Pi3K or p38 blockade is a stable or reversible epigenetic event. Suppl. figure 3B shows that the effects of SB and LY on muscle differentiation were readily reversible. These data indicate that is possible to generate a dynamic population of myoblasts with different chromatin conformation at muscle loci by pharmacological manipulation of signaling pathways.
The data presented in Fig. 2 suggest that the IGF1/Pi3K signaling affects the composition of the complexes bound to the chromatin of muscle genes both by promoting the recruitment of positive regulators (MyoD and acetyltransferases) and by preventing the engagement of co-repressors (histone deacetylases and methyltransferases). Given the importance of recruiting MyoD and acetyltransferases to enable the chromatin remodeling activity of SWI/SNF, we investigated the mechanism by which IGF1/Pi3K signaling promotes formation of the MyoD/acetyltransferase complex on the chromatin of muscle genes.
Interactions between MyoD and p300 are regulated at multiple levels (Puri and Sartorelli 2000). We first investigated the role of the IGF-1/Pi3K signaling in promoting physical interactions between MyoD and acetyltransferases in muscle cells, by co-immunoprecipitation studies with endogenous proteins. An association of MyoD with p300 and PCAF was detected in myoblasts induced to differentiate, and this complex was specifically disrupted by Pi3K inhibition (Fig. 3A). Simultaneous downregulation of indivudual kinases p38 α and β or AKT 1 and 2, respectively, by RNAi was also exploited to definitively assess the contribution of these kinases in regulating MyoD-p300 interaction. Although downregulation of either p38 α and β or AKT 1 and 2 (Fig. 3B) impaired the expression of muscle genes in differentiating myoblasts (Fig. 3C), only donwregulation of AKT1 and 2 prevented the association between endogenous MyoD and p300 (Fig. 3D). The interaction between MyoD and p300 was further measured by a mammalian two-hybrid system, by using a Gal4-luciferase reporter co-transfected with Gal4MyoD and p300 fused with the acidic activation domain VP16 (VP16p300). Exposure to IGF-1 in a serum-free medium stimulated the interaction between Gal4MyoD and VP16p300, and this effect was prevented by LY (Fig. 3E). By using two truncated versions of p300 (N-terminal aa1-744 and C-terminal aa871-2377), we established that the IGF-1 responsive region maps to the C-terminus (Fig. 3E). This finding is in agreement with the reported observation that the IGF1/Pi3K/AKT pathway promotes interaction between bHLH transcription factors and p300 during neurogenesis (Vojtek et al. 2003). Other works have reported on the ability of AKT to phosphorylate p300 (Huang and Cheng 2005). Indeed, two putative AKT consensus sites are present at the C-terminus of p300, within the MyoD interaction domain - the CH3 region. One consensus site (RRLS) maps to serine 1734 and is preferential for AKT1. The second phospho-acceptor site maps to serine 1834 (RRRMASM) and can be phosphorylated both by AKT1 and 2 (Fig. 3F). These sites are conserved in both p300 and CBP in different species. However, whether MyoD-p300 interaction is regulated via direct phosphorylation of p300 by AKT or by any other downstream kinase(s) has not been addressed by previous studies.
We monitored the phosphorylation pattern of endogenous p300 in myoblasts after metabolic labeling with 32P orthophosphate and immunoprecipitation with anti-p300 antibodies. This analysis showed minimal fluctuations of global p300 phosphorylation along the transition from myoblasts to myotubes. However, Pi3K inhibition by LY significantly reduced the phosphorylation of p300 in myoblasts incubated in differentiation medium (DM) (Fig. 4A). This evidence further indicates the contribution of the IGF1/Pi3K pathway to p300 phosphorylation during myogenic differentiation. To precisely map the AKT phosphorylation sites in the p300 C-terminal, we expressed a HA-tagged truncated version of p300 (HA-1640-1840) in 293 cells along with constitutively active, myristoylated (myr) AKT1 or with inactive kinase death (kd) AKT1. After metabolic labeling with 32P orthophosphate, the p300 HA-1640-1840 fragment was immunoprecipitated with anti-HA antibodies and subjected to phospho-peptide mapping (Fig. 4B). The two prominent spots detected in cells transfected with AKT1myr, but not in cells expressing AKT1 kd, were further subjected to phospho-aminoacid analysis, which revealed the presence of phospho-serines (Fig. 4C). The migration pattern of the phospho-peptides (Fig. 4B) and their confirmed identity by phospho-aminoacid analysis (Fig. 4C) are consistent with AKT-dependent phosphorylation of p300 on serines 1734 and 1834. Note that ectopic expression of AKT1myr correlated with the activation of endogenous AKT2 (data not shown), suggesting a reciprocal activation and functional synergy between these two kinases. To confirm that these serines are phosphorylated in vivo during myoblast differentiation, we transfected the p300 C/H3 (p300 HA-1640-1840) fragment in C2C12 cells and monitored its phosphorylation after metabolic labeling with 32P orthophosphate. An increased phosphorylation was observed in myoblasts induced to differentiate (Fig. 4D, compare lane 1 and 2, and Fig. 4E, compare lane 1 and 2). The 32P incorporation was abrogated by either Pi3K inhibition with LY (Fig. 4D) or by replacement of the two serines with non-phosphorylatable alanines (p300 CH3 S1734/1834A) (Fig. 4E).
The phosphorylation of serine 1834 of endogenous p300 was further studied in human skeletal myoblasts (HSKM) using ser1834 phospho-specific antibodies (Liu et al. 2006). Immunofluorescence studies showed an increased nuclear staining in multinucleated, differentiated myotubes, as compared to undifferentiated myoblasts. Treatment with LY, but not SB, drastically reduced the nuclear accumulation of phosphoserine 1834 p300, although both compounds inhibited the formation of myotubes (Suppl. Fig. 4A). The changes in phosphorylation of serine 1834 of p300 in the conditions described above were confirmed by western blot on nuclear extracts (Suppl. Fig. 4B).
We next addressed the functional impact of AKT-mediated phosphorylation of p300 C-terminal serines 1734 and 1834 on p300-MyoD interaction, by using the mammalian two-hybrid-based assay shown in Fig. 3E. We compared the interaction of Gal4MyoD and the p300 C/H3 fragment, either wild type (VP16p300 C/H3 wt) or phospho-mutant (VP16p300 C/H3 S1734/1834A), fused to VP16. An interaction between Gal4MyoD and the VP16p300 C/H3 wt was observed in response to differentiation cues and was inhibited by LY. In the same conditions, the VP16-p300C/H3 S1734/1834A double mutant failed to interact with Gal4MyoD (Fig. 5A). Consistent with the role of AKT in promoting the association between MyoD and p300 via phosphorylation of the C/H3 domain of p300, co-transfection of the constitutively active form of AKT1myr, but not the AKT1 kd, was sufficient to promote interaction between Gal4MyoD and the VP16-p300 C/H3 wt, but not the VP16-p300C/H3 S1734/1834A mutant (Fig. 5B). The impact of each phosphoacceptor site on MyoD-p300 interaction was further analyzed by comparing the ability of full length Flag-tagged p300 wild type (wt) and single mutants (S1734A and S1834) to interact with endogenous MyoD, upon transient transfection in C2C12 myoblasts. Figure 5C shows that although these mutants are expressed at comparable levels (bottom panel), only p300 wt was found associated with MyoD (top panel). The relative impact of AKT1 and 2 on p300-MyoD interactions was also addressed by using mouse embryonic fibroblasts (MEFs) derived from AKT1 or AKT2 null mice (Liu et al. 2006). Employment of AKT 1 or 2 null cells eliminates the caveat intrinsic to the RNAi downregulation, consisting in the presence of residual, minimal expression level of the targeted isoform. We first monitored the binding of Gal4MyoD and VP16-p300 C/H3 wt in AKT1 and AKT2 deficient MEFs, as compared to wt MEFs. Absence of either AKT1 or AKT2 drastically impaired Gal4MyoD-VP16p300 interactions (Fig. 5D), further indicating that phosphorylation of p300 by both AKT1 and AKT2 is required for complex formation. In keeping with this conclusion, the interaction between endogenous p300 and ectopically expressed MyoD was detected in wt MEFs and in MEFs from MKK3/6 null mice, in which endogenous p38 cannot be activated (Brancho et al. 2003), but not in AKT1 null or AKT2 null MEFs (Suppl. fig. 5A). Consistently, the recruitment of MyoD and p300 to the chromatin of muscle genes was selectively impaired in MyoD-converted AKT1 or AKT2 null MEFs (Fig. 5E), while Brg1 was detected on the chromatin of these cells, but not in MyoD-converted MKK3/6 null MEFs (Fig. 5E).
Cooperation at the chromatin level between the Pi3K/AKT-mediated recruitment of MyoD and acetyltransferases and the p38-directed recruitment of SWI/SNF on the regulatory sequences of muscle genes was further explored by functional assays that combined deliberate activation of p38 with inactivation of AKT chromatin effectors. The p38 pathway was activated in C2C12 myoblasts by the ectopic expression of the constitutively active form of its upstream kinase, MKK6EE-HA, via adenoviral delivery (Fig. 6A and B). These conditions resulted in the forced expression of muscle genes despite culture conditions non-permissive for differentiation (GM), and either p38 or Pi3K blockade was sufficient to prevent MKK6EE-induced muscle differentiation (Fig. 6A and B). While the effect of SB is expected, as this compounds blocks the downstream MKK6EE effector p38, the effect of LY was independent on the activation status of p38 (Fig. 6A and B). Consistent with the notion that the p38 and the IGF1/Pi3K/AKT pathways proceed as parallel cascades, which converge on different chromatin-bound complexes, the ability of LY to prevent MKK6EE-mediated activation of the myogenic program correlates with the local hypoacetylation and impaired chromatin remodeling at muscle loci (Fig. 6C and D). We next performed experiments aimed at establishing a direct relationship between MKK6EE-mediated activation of muscle gene expression and Pi3K/AKT-directed recruitment of p300 to the chromatin of muscle genes. In one experimental setting we compared the effect of the cytoplasmic inhibitor of the Pi3K/AKT pathway (LY) with that of the inhibitor of the Pi3K/AKT nuclear target p300 (the LysCoA) (Lau et al. 2000) on MKK6EE-dependent activation of the differentiation program in myoblasts. Figure 6E shows that microinjection of LysCoA at concentrations previously reported to inhibit p300 enzymatic activity (Lau et al. 2000; Polesskaya et al. 2002) was as efficient as LY in preventing the expression of myosin heavy chain (MyHC) in myoblasts injected with MKK6EE. These results establish a functional link between the p38 pathway and the integrity of the enzymatic activity of the Pi3K/AKT chromatin sensor p300. However, they do not address the specific role of AKT-mediated phosphorylation of p300 in the functional cooperation between the p38 pathway and p300. To this aim, we have compared the effect of over-expression of the truncated form of the C/H3 p300 (1640-1840), either wild type (wt) or the phospho-mutant (mt) ser1734/1834ala, on the ability of MKK6EE to promote the myogenic conversion in MyoD-expressing fibroblasts cultured in GM (Figure 6F). Increasing concentrations of C/H3 (1640-1840) wild type countered the ability of MKK6EE to promote myogenic conversion, but the effect was minimal with the phospho-mutant (Figure 6F). This result suggests that the dominant negative effect of C/H3 (1640-1840) fragment on muscle differentiation relies on the presence of AKT phosphorylation sites. Increasing concentrations of C/H3 (1640-1840) likely buffer the kinase availability of AKT toward endogenous p300. This evidence further supports the functional interdependence at the chromatin level between different enzymes, which are recruited into the myogenic transcriptosome by different signal-activated kinases, via direct phosphorylation.
Our results illustrate a model of cooperation between insulin/IGF1-activated signaling and the p38 pathway during skeletal myogenesis. In this model, IGF1-activated Pi3K/AKT signaling promotes the recruitment of MyoD and acetyltransferases to the chromatin of muscle genes, an event necessary to enable the chromatin remodeling activity of the p38-recruited SWI/SNF complex. Furthermore, our data suggest another, apparently distinct activity of IGF1 signaling, which is to displace or prevent the association of histone deacetylases and methyltransferases within the muscle transcriptosome. Selective interruption of IGF1/Pi3K/AKT signaling in conditions permissive for differentiation (DM), in which the p38 pathway is active, results in the formation of a complex containing MEF2 associated with co-repressors (i.e HDAC4, Ezh2), instead of co-activators (acetyltransferases) (depicted in Fig. 7). Indeed, MEF2-associated class II HDACs (Lu et al. 2000) and Ezh2-directed lysine methylation (Caretti et al. 2004) were reported to contribute to muscle gene silencing in myoblasts. Interestingly, the SWI/SNF complex can mediate repression of transcription, instead of activation, when associated to co-repressors (de la Serna et al. 2006; Marenda et al. 2004; Martens and Winston 2002). Further studies should establish whether p38-directed phosphorylation is involved in recruitment of SWI/SNF into repressive complexes. Likewise, it will be important to elucidate the role of IGF1 signaling in the chromatin re-distribution of histone deacetylases and methyltransferases.
The terminal effectors of IGF1 signaling to the chromatin of muscle genes are AKT 1 and 2 kinases, which phosphorylate the C-terminal region of the acetyltransferase p300 at the onset of differentiation. The individual contribution of AKT 1 and 2 to the myogenic program is currently unclear, with works reporting on the essential role of AKT2, which is upregulated in differentiating myoblasts (Kaneko et al. 2002), and works reporting on the central role of AKT 1 in skeletal myogenesis (Wilson and Rotwein 2007). Our data indicate that AKT 1 and 2 are both necessary to phosphorylate p300 and promote interaction with MyoD and possibly other muscle specific transcription factors. Although the amino-acidic sequence of one p300 phospho-acceptor site (serine 1734) appears to be a preferential consensus for AKT 1 (Fig. 3F), our results indicate the requirement of both kinases for optimal phosphorylation of p300 C-terminal and the consequent interaction with MyoD. Genetic studies in MEFs deficient for either AKT 1 or 2 show that one kinase cannot compensate for the absence of the other (Fig. 5D, E and Suppl. Fig. 5A). Likewise, elimination of either phosphoacceptor site impairs the ability of p300 to interact with MyoD, as revealed by complementary assays (Fig. 5A, B and C). Conversely, expression of the constitutively active AKT 1 (AKT 1 myr) alone led to the phosphorylation of both serines 1734 and 1834, and this correlates with the activation of endogenous AKT 2 (data not shown). These results suggest that cross-activation between these two kinases might take place to achieve a cooperative effect on phosphorylation of common substrates (i.e. p300), although the biochemical basis for such synergy needs to be elucidated by future studies.
How does p300 C-terminal phosphorylation regulate the interaction with MyoD? Several mechanisms can cooperatively mediate this effect. The C/H3 domain is the region of p300 that mediates the interaction with MyoD (Yuan 1996, Eckner et al. 1996; Sartorelli et al. 1997). AKT-mediated C/H3 phosphorylation can induce/stabilize this interaction. MyoD and histone acetylation are additional events that promote MyoD/p300 binding and chromatin recruitment (Sartorelli et al. 1999; Polesskaya et al. 2001). As AKT-mediated phosphorylation was reported to stimulate the enzymatic function of p300 (Huang and Chen 2005), it is possible that an increased acetyltrasferase activity toward MyoD contributes to AKT-dependent interaction between MyoD and p300. Additionally, the C/H3 domain mediates interactions between p300 and pCAF (Yang et al. 1996), and AKT-mediated phosphorylation of p300 appears to contribute to the recruitment of pCAF, which in turn acetylates MyoD, p300 and histones to further stabilize the p300/MyoD/pCAF complex and activate transcription (Puri et al. 1997b; Sartorelli et al. 1999; Dilworth et al. 2004). Consistent with our results, others showed that inhibition of IGFII production in MyoD-converted fibroblasts precluded the engagement of p300 and PCAF on myogenin promoter (Wilson and Rotwein 2006). However, in their experimental setting chromatin recruitment of MyoD was not affected and the consequence on SWI/SNF enzymatic activity was not addressed.
The importance of AKT-mediated phosporylation of p300 extends to the regulation of SWI/SNF chromatin remodeling activity, as in the absence of MyoD/acetyltransferases the chromatin of muscle genes was hypoacetylated and un-remodeled, despite the presence of Brg1 (Fig. 2A and B). It is possible that histone acetylation is an event required for SWI/SNF to remodel chromatin, as nucleosomes containing acetylated histones are better predisposed for the enzymatic activity of of SWI/SNF (Chandy et al. 2006). The presence of SWI/SNF on hypoacetylated chromatin is apparently in conflict with previous works reporting on SWI/SNF chromatin binding via interactions of the bromodomain with acetylated histones (Hassan et al. 2001). However, according to a two-step model of SWI/SNF chromatin recruitment and activation, SWI/SNF can first be recruited by transcriptional activators, and then stabilized by interactions with acetylated histones. Local hyperacetylation (Chandy et al. 2006) and the association with transcriptional activators (Gutierrez et al. 2007) appear to further promote the chromatin remodeling activity of SWI/SNF. Thus, in conditions in which acetyltransferases and MyoD are not recruited to the chromatin of muscle genes (i.e. interruption of Pi3K/AKT signaling), SWI/SNF recruitment by MEF2 (Fig. 2A and Suppl. Fig. 5B), in complex with deacetylases and methyltransferases (Fig. 2A, C and Suppl. Fig. 3A), could result in an impaired remodeling activity of SWI/SNF (Fig. 2B).
The interplay between two parallel signaling pathways at the chromatin levels may have a remarkable impact during the regeneration process. Both pathways are activated by regeneration cues and influence the ability of muscle progenitors to execute different stages of the regeneration program. The Pi3K/AKT pathway is elicited by locally released IGFs and promotes satellite cell proliferation, survival and differentiation (Mourkioti and Rosenthal 2005). The p38 pathway is activated in satellite cells, presumably in response to locally released inflammatory soluble factors or cell-to-cell interactions, and promotes cell cycle arrest (Puri et al. 2000; Perdiguero et al. 2007) and terminal differentiation (Zetser et al. 1999; Wu et al. 2000). The convergence of these two pathways at the chromatin level provides a mechanism for integration of regeneration cues to coordinate gene expression during cellular differentiation.
The different chromatin modifications observed in response to pharmacological inhibition of the p38 or the IGF1-activated pathways correlate with distinct cellular phenotypes in treated myoblasts. Pi3K blockade prevents the hyperacetylation of the chromatin of muscle genes, and leads to quiescence. p38 blockade promotes proliferation, which correlates with hypearcetylation at muscle genes (see illustration in Fig. 7). The acetylation status of the chromatin at muscle loci has a relevant biological impact during regeneration. For instance, MyoD acetylation regulates satellite cell capability to regenerate injuried muscles (Dunquet et al. 2006) and agents that increase histone hyperacetylation at muscle loci, such as deacetylase inhibitors, promote myogenesis in vitro and muscle regeneration in vivo, and revealed to be effective in the treatment of muscular dystrophies (Iezzi 2002; Iezzi 2004; Minetti 2006). Furthemore, deacetylase inhibitors can induce AKT activation in satellite cells (CM and PLP unpublished observation) and promote AKT-mediated phosphorylation of p300 in other cell types (Liu et al. 2006). Thus, the AKT signaling to chromatin-bound proteins, such as p300, could also contribute to the beneficial effects of deacetylase inhibitors on muscles regeneration.
Our results illustrate an example of convergence of distinct signaling pathways at the chromatin level, to coordinate the expression of genes implicated in satellite cell transition from quiescence to terminal differentiation. These data suggest potential pharmacological avenues for selective control of gene expression to manipulate muscle regeneration.
C2C12 mouse myoblasts, 10T1/2 mouse fibroblasts converted by ectopic expression of MyoD were cultured in DMEM supplemented with 20% FBS (GM -growth medium) and switched to differentiation medium (DM - DMEM supplemented with 2% horse serum) or serum free medium (0.1 % FBS) supplemented or not with insulin growth factor 1 (IGF1) or insulin/transferrin when indicated. 293 cells were cultured in 10% FBS. Human skeletal myoblasts were purchased from Clonetics. Mouse embryonic fibroblasts (MEFs) were cultured in 10% FBS or switched to DM, when indicated.
Single muscle fibers with associated satellite cells were isolated as described (Minetti et al. 2006).
SB203580 and LY294002 were purchased from Calbiochem.
The p300 S1734A point mutant was made by quikchange using p300 WT pCI as a template. The Quikchange II XL kit from Stratagene was used with the following primers:
1734 Fwd - CGATTCTCGCCGCCTGGCTATCCAGCGCTGCATCCAG
1734 Rev – CTGGATGCAGCGCTGGATAGCCAGGCGGCGAGAATCG.
The p300 S1834A point mutant was described in (Liu et al. 2006)
Transfections were carried out with the Lipofectamine 2000 reagent and luciferase activity was determined with a Promega Luciferase kit and normalized by either β-Gal values or using the Dual–Luciferase reporter assay system (Promega). Mammalian two-hybrid system was performed as previously described (Sartorelli et al 1997), using p300VP16 constructs received from V. Sartorelli.
Simultaneous downregulation of AKT 1 and 2 or p38 α and β was achieved by RNAi using two distinct oligonucleotides specific to each isoform (Silencer Pre-designed siRNA from Ambion), as indicated in supplementare information. Oligonucleotide transfection of C2C12 was performed by using DharmaFETC 3 protocol and resulted in significant reduction (ranging from 80 to 90%) of targeted gene products.
Adenovirus expressing MKK6EE from the CMV promoter was described in (Simone et al. 2004a). C2C12 and MEFs were infected with pAdMKK6 for 2 hours in serum free medium and then placed in GM for additional 18 hours.
Semi-quantitative RT-PCR was performed on RNA extracted from cells with Trizol (Invitrogen). The RNA was reverse transcribed and amplified by using the SuperScript One-Step RT-PCR (Invitrogen), according to standard procedures. For real time PCR, cDNA synthesis was performed from two independent RNA preparations using SuperScript III Platinum Two-Step qRT-PCR kit (Invitrogen) and High Capacity cDNA Archive Kit (Applied Biosystems) and quantitive PCR was performed with the ABI Prism 7500 PCR instruments (Applied Biosystems) to amplify samples in triplicate. PCR reactions were performed using the primers described in supplementary information.
Western blot was performed on cell extracts according to standard procedures. The following antibodies were used: anti-Ha (F7 from Santa Cruz), anti-total p38 and phospho-p38 (Cell Signaling), anti-myogenin (FD5 hybridoma), anti-acetylH3 (K9 and K14) (Upstate Biotechnology), anti-phospho 1834 p300 (Liu et al. 2006), anti total p300 (Santa Cruz n15) and anti-tubulin (Santa Cruz).
Microinjection of C2C12 cells was performed as described in (Simone et al. 204a). Immunofluorescence studies on satellite muscle cells and human skeletal muscle cells were performed in paraformaldeide-fixed cells with the indicated antibodies.
Nuclear cell extracts and immunoprecipitations were performed as described in (Simone et al 2004). Briefly, nuclear extracts were pre-cleared with an A+G sepharose mix for 30 min at 4C and immunoprecipitated with agarose-conjugated anti-MyoD antibodies (M-318 Ac, Santa Cruz and RGB7 from Dr. S. Alemà). Precipitated proteins were revealed by western blot with anti-p300 (N15, Santa Cruz), anti-PCAF (H-369, Santa Cruz), and anti-MyoD (M-318, Santa Cruz) antibodies.
293T cells were transfected with HA tagged wt and mutant p300-CH3 (aa 1640-1840) constructs in the presence of kinase-dead of myristoylated Akt expression vectors. After 36 h, transfected cells were incubated with phosphate-free Dulbecco’s modified Eagle medium containing 10% dialyzed serum and 1 mCi of [32P]orthophosphate/ml. Following a 4hr incubation, the cells were washed with cold phosphate-buffered saline, and harvested in sodium dodecyl sulfate (SDS) lysis buffer (0.5% SDS, 50 mM Tris [pH 8.0], 1 mM EDTA) containing a cocktail of phosphatase inhibitors. The lysate was precleared with of pansorbin and supernatants were immunoprecipitated with agarose-conjugated anti-HA antibodies (Santa Cruz). Immunoprecipitates were resolved by SDS- gel electrophoresis, and transferred to nitrocellulose membranes. Following western blotting using HRP-conjugated (Santa Cruz) or AP-conjugated (Sigma) anti-HA antibodies, the HA-CH3 bands were excised, and subjected to tryptic digestion followed by 2-D phosphopeptide maps. Phosphopeptide maps were visualized by autoradiography. For phospho-aminoacid analysis, about 500 cpm of labeled peptides were oxidized in 6 N HCl, and heated to 110°C and the individual phosphoaminoacids were separated by two-dimensional high-voltage electrophoresis and visualized by ninhydrin staining and autoradiography.
We used a Histone Deacetylase Assay Kit purchased from Upstate Biotechnology, as previously described (Simone et al. 2004a). C2C12 nuclear extracts were prepared as described and endogenous MEF2 A/D was immunoprecipitated by anti-MEF2 antibodies (Simone et al. 2004a). The MEF2-associated deacetylase activity was counted in a deacetylation assay by incubation with 60000cpm 3H-acetylated H4 in 200ml 1x Assay buffer for 3h at 30C. 100nM trichostatin A (TSA) was used to inhibit deacetylase activity. Free 3H-Acetyl was determined with liquid scintillation counting and results were adjusted to background.
A ChIP assay was performed as previously described (Simone et al. 2004a). The following antibodies were used: anti-MyoD (M-318 Ac, Santa Cruz), anti-MEF2c, anti-acetylated histone H4 (Upstate Biotechnology), anti-acetylated histone H3 (Upstate Biotechnology), anti-acetylated MyoD, anti-p300 (N-15 Ac, Santa Cruz), anti-PCAF (H-369, Santa Cruz), anti-p38 (received from Dr Sudo), anti-Brg1 (polyclonal N-15 and H-88, Santa Cruz), anti-HDAC4 (Santa Cruz). PCR was performed on immuno-precipitated DNA with the primers indicated in supplementary information.
Real-time PCR was performed using the SyberGreen Master Mix at least three times using three independent samples. Data reported in figures are values normalized to input DNA and to values from normal rabbit IgG, which were set as the background (one unit) in each calculation. The primers used are indicated in supplementary information.
P.L.P is an associate Telethon scientist at Dulbecco Telethon Institute (DTI). We thank M. Birnbaum for providing AKT1 and 2 null MEFs, R. Davis for MKK3/6 double null MEFs, V. Sartorelli for p300VP16 constructs, P. Tsichlis for AKT constructs, P. Cole for LysCoA, and S. Alema for anti MyoD antibodies (RGB7). This work was supported by Telethon grant, RO1 NIH grant, Muscular Dystrophy Association grant (MDA), Parent Project and Compagnia San Paolo di Torino to PLP. CM is supported by Parent Project fellowship. UJ was supported by grants from JDRF and Hillblom foundation
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.