β-catenin promotes self-renewal of skeletal-muscle satellite cells Free

Homeostasis, hypertrophy and repair of adult skeletal muscle are carried out by resident stem cells – called satellite cells – located on the surface of the myofibre, below the surrounding basal lamina (Mauro, 1961) (reviewed in Zammit et al., 2006a). Satellite cells are normally mitotically quiescent in the adult and, therefore, must first be activated and must undergo extensive proliferation to generate myoblasts that eventually differentiate to repair or replace myofibres. We have recently shown that satellite-cell self-renewal is the primary mechanism responsible for maintaining a viable satellite-cell pool (Zammit et al., 2004; Collins et al., 2005). When transplanted into muscle in association with a myofibre, satellite cells proliferate extensively and give rise to both a substantial amount of donor-derived muscle and many new satellite cells (Collins et al., 2005). This process can be modelled in culture where satellite-cell progeny adopt divergent fates. Quiescent satellite cells express the paired-box 7 (Pax7) transcription factor (Seale et al., 2000) and, when activated, they express Pax7 together with MyoD (Grounds et al., 1992; Yablonka-Reuveni and Rivera, 1994) – a member of the myogenic regulatory factor family (comprising MyoD, Myf5, myogenin and Mrf4). After proliferation, most satellite cells downregulate Pax7 and differentiate. Other satellite-cell progeny, however, maintain Pax7 expression but lose that of MyoD, withdraw from both cell cycle and immediate myogenic differentiation, and return to a quiescent-like state (Zammit et al., 2004). Similar observations have also been made during muscle growth in chicken (Halevy et al., 2004) and rat (Schultz et al., 2006).

What controls whether a satellite cell either self-renews or differentiates is only beginning to be understood and has mainly centred on Pax and Notch genes. Pax7 is able to activate transcription in both quiescent satellite cells and those that adopt the self-renewal pathway, and constitutive Pax7 expression can delay myogenic differentiation (Zammit et al., 2006b). Recently Pax7 has been shown to inhibit MyoD function, whereas myogenin can inhibit Pax7, providing a possible mechanistic insight (Olguin et al., 2007). Satellite-cell activation is accompanied by activation of Notch1, which leads to cell proliferation (Conboy and Rando, 2002). Inhibiting the Notch pathway results in a shift to a Pax7-negative–MyoD-positive (Pax7^–MyoD⁺) phenotype, indicating that Notch signalling is also involved in self-renewal (Kuang et al., 2007). Moreover, there is evidence that this may be controlled by Numb, which can be differentially segregated to one daughter after an asymmetric satellite cell division (Conboy and Rando, 2002; Shinin et al., 2006).

Wnt signalling also has a vital role in myogenesis, as clearly demonstrated during development (reviewed in Parker et al., 2003), where it has been shown to contribute to induction of the myogenic lineage (Munsterberg et al., 1995; Tajbakhsh et al., 1998). In adult, Wnt signalling may be actively suppressed during the early stages of muscle regeneration (Zhao and Hoffman, 2004). Wnts act through distinct canonical and non-canonical signalling pathways. The canonical pathway involves the stabilisation of β-catenin that can then translocate to the nucleus and control transcription via the lymphoid enhancer-binding factor 1 (LEF1, also known as TCFα) TCF/LEF family of transcription factors (reviewed in Willert and Jones, 2006). β-catenin however, not only influences cellular events as a necessary transcriptional co-activator, but also has an important role in cell adhesion complexes. It binds members of the cadherin family of adhesion molecules at the cell membrane (Kuch et al., 1997; Ozawa et al., 1989) that is involved in myogenic differentiation, particularly myoblast fusion into multinucleated myotubes (Goichberg et al., 2001). Indeed, the importance of β-catenin for adult muscle function has been demonstrated by the recent observation that it is essential for muscle hypertrophy in response to overload (Armstrong et al., 2006).

Here, we sought to examine whether β-catenin is involved in determining satellite-cell fate. We found that β-catenin is expressed in satellite cells and their progeny. Constitutive β-catenin expression or blocked degradation of endogenous β-catenin in satellite cells affects cell-fate choice, with more satellite cells remaining Pax7-positive (Pax7⁺) and fewer undergoing myogenic differentiation. Similarly, infection of satellite cells with β-catenin-encoding retroviral constructs resulted in an increased proportion of reserve cells in C2 cultures, and inhibited differentiation. By contrast, if levels of β-catenin are reduced by using small interfering RNA (siRNA) or if its transcriptional targets are repressed by a dominant-negative version, myogenic differentiation of satellite-cell progeny is enhanced. Together, these observations show that the role of β-catenin is bimodal: in addition to its acknowledged role in myoblast fusion, β-catenin also directs satellite cells to the self-renewal pathway and away from immediate myogenic differentiation.

Immunostaining was first used to determine whether satellite cells express β-catenin during myogenic progression. Satellite cells associated with freshly isolated extensor digitorum longus (EDL) myofibres weakly expressed β-catenin. At this time, before MyoD is detectable, most satellite cells exhibiting a peri-nuclear or cytoplasmic distribution of β-catenin (Fig. 1a). β-catenin expression is upregulated in activated satellite cells after 24 hours of culture in their niche on the myofibre (Fig. 1b). When analysed after the first division (at 48 hours) β-catenin showed a nuclear localisation in many cells (Fig. 1c,d). Satellite cells adopt divergent fates after ∼72 hours in culture (Zammit et al., 2004) and, at this time, β-catenin expression was clearly evident in both Pax7⁺ and Pax7-negative (Pax7^–) cells (Fig. 1e,f). Nuclear β-catenin levels were decreased as satellite cells differentiated or underwent self-renewal, with β-catenin located at the cell surface or peri-nuclearly in most cells (Fig. 1d,e). Therefore, mouse satellite cells in culture express β-catenin during myogenic progression, as has been shown for adult rat myogenic cells (Ishido et al., 2006; Wrobel et al., 2007).

Having shown that β-catenin is present in satellite cells, we next examined the effects of consitutive β-catenin expression on satellite-cell-fate choice. We made two retroviral expression constructs (pMSCV–β-catenin–IRES–eGFP and pMSCV–ST-β-catenin–IRES–eGFP), which encode wild-type β-catenin and stabilised β-catenin (ST-β-catenin), respectively. Stabilised β-catenin contains amino acid substitutions to prevent phosphorylation and subsequent degradation (Barth et al., 1999). Myofibre-associated satellite-cell progeny were exposed to either pMSCV–β-catenin–IRES–eGFP, pMSCV–ST-β-catenin–IRES–eGFP or control pMSCV-IRES-eGFP and fixed 48 hours later. Immunostaining for eGFP and β-catenin confirmed that satellite cells infected with pMSCV–β-catenin–IRES–eGFP or pMSCV–ST-β-catenin–IRES–eGFP robustly co-expressed these proteins (Fig. 1g-i). Co-immunostaining for eGFP and Pax7 showed that significantly more (P<0.05) cells that had been infected with constructs encoding wild-type or stabilised β-catenin contained Pax7 protein, compared with those infected with control retrovirus (71.5±0.7% or 70.6±0.9%, respectively, versus 31.6±0.4%; see Fig. 2, and compare a,j and b,j, with c,j).

Since more satellite-cell progeny that expressed wild-type or stabilised β-catenin contained Pax7 than controls, we next examined MyoD and myogenin protein levels to determine whether myogenic differentiation is compromised (Zammit et al., 2004). By co-immunostaining for eGFP and MyoD, we found that MyoD expression was significantly lower (P<0.05) in satellite-cell progeny constitutively expressing wild-type β-catenin or stabilised-β-catenin compared with control-infected satellite-cell progeny (18.5±0.4% or 21.2±0.5%, respectively, versus 54.05±0.9%; compare Fig. 2, d,j or e,j with f,j). The predominance of the Pax7⁺MyoD^– phenotype in cells that constitutively express β-catenin indicates that β-catenin drives satellite cells towards the self-renewal pathway (Zammit et al., 2004). Co-immunostaining for eGFP and myogenin revealed significantly fewer (P<0.05) eGFP-positive (eGFP⁺) satellite-cell progeny that also express myogenin in cells infected with pMSCV–β-catenin–IRES–eGFP or pMSCV–ST-β-catenin–IRES–eGFP compared with parallel cultures that had been infected with control pMSCV-IRES-eGFP (10.1±0.4% or 8.9±0.4%, respectively, versus 73.8±0.7%; compare Fig. 2, g,j or h,j with 2i,j). This indicates that fewer satellite cells were committing to myogenic differentiation in the presence of constitutively expressed β-catenin.

Culture of satellite cells associated with a myofibre is a useful model to examine the early events of myogenic progression in satellite cells, but is less-well suited for studying the later events of differentiation – such as myoblast fusion. Culture of myofibres on Matrigel allows satellite cells to migrate from the myofibre onto the tissue culture substrate and to proliferate, before differentiating and fusing into multi-nucleated myotubes or opting out of immediate differentiation. However, because this process is asynchronous, switching culture conditions to mitogen-poor medium can be used to coordinate this cell-fate choice in primary myogenic cells (Kitzmann et al., 1998). Plated satellite-cell-derived myoblast cultures were infected with pMSCV–β-catenin–IRES–eGFP, pMSCV–ST-β-catenin–IRES–eGFP or control pMSCV-IRES-eGFP; 24 hours later culture medium was changed to mitogen-poor medium and cells were cultured for at least another 4 days. Compared with control-infected cells, myogenic differentiation and fusion were clearly compromised in the presence of β-catenin. Control-infected cells demonstrated robust differentiation and myotube formation, as shown by immunostaining for MyoD (Fig. 3a-c), myogenin (Fig. 3d-f) and myosin heavy chain (MyHC) (Fig. 3g-i; all quantified in m).

Whereas the vast majority of plated primary myogenic cells challenged with serum withdrawal respond with differentiation, others maintain Pax7, downregulate MyoD and exit the cell cycle, entering a quiescent-like state (Kitzmann et al., 1998; Collins et al., 2007). The presence of constitutively expressed wild-type β-catenin or stabilised β-catenin increased the proportion of Pax7⁺ cells in mitogen-poor medium (Fig. 3j-l) to 68.5±6.9% or 78.3±4.2% respectively, compared with control (3.96±0.83%; quantified in Fig. 3m). The percentage of cells expressing MyoD was drastically reduced (from 95.1±0.86% with control pMSCV-IRES-eGFP, to 16±1.7% with wild-type β-catenin and 20.3±1.6% for stabilised β-catenin; see Fig. 3m), and myogenin or MyHC failed to be induced throughout (Fig. 3m). Therefore, β-catenin influences cell fate choice towards a Pax7⁺MyoD^– phenotype.

We also tested the effects of β-catenin by using the reserve-cell model of myogenic quiescence (Yoshida et al., 1998) and the immortalised adult post-injury-derived C2 cell line (Yaffe and Saxel, 1977). Levels of Pax7 vary between C2 clones, with some entirely lacking expression of Pax7 as proliferating myoblasts; however, this does not affect their myogenicity (Olguin and Olwin, 2004; Zammit et al., 2006b). Among the proliferating C2 cells of the clone used here, a mean of 17±5% expressed Pax7. Since these cells are immortalised, they must be switched to mitogen-poor medium in order to force a cell-fate choice. Under these culture conditions, the vast majority of cells differentiated (Fig. 4), with few Pax7-expressing reserve cells present (1.1±0.09%; Fig. 4m), consistent with previous observations (Olguin and Olwin, 2004; Zammit et al., 2006b). Crucially, retroviral infection with wild-type or stabilised β-catenin, followed later by culture in mitogen-poor medium, resulted in a considerable increase in the number of cells with Pax7 protein (70.9±2.3% or 62.7±4.2%, respectively, compared with control 1.1±0.09%; Fig. 4m). Sister cultures immunostained for MyoD showed that the proportion of positive cells fell from 86.2±3.5% in controls to 8.4±0.8% for wild-type β-catenin and 11.3±0.5% for stabilised β-catenin, whereas myogenin and MyHC were not induced in the vast majority of cells (Fig. 4m). Since the retroviral infection efficiency in C2 cells was ∼80%, most Pax7⁺ cells were MyoD-negative (MyoD^–) and, consequently, reserve cells. Therefore, consistent with observations in primary satellite cells, the presence of β-catenin also effectively inhibited myogenic differentiation and promoted self-renewal in C2 cells.

To determine whether constitutively expressed β-catenin perturbed the cell cycle, we infected satellite-cell progeny with either pMSCV–β-catenin–IRES–eGFP, pMSCV–ST-β-catenin–IRES–eGFP or control pMSCV-IRES-eGFP. After 72 hours, cells were pulsed with BrdU for 3 hours and then fixed. Co-immunostaining for eGFP and BrdU showed that ∼60% fewer satellite cell progeny containing wild-type or stabilised β-catenin had incorporated BrdU when maintained on the myofibre compared with control infected cells (Fig. 5). Satellite-cell-derived myoblasts and C2 cells were also infected and 24 hours later switched to mitogen-poor medium for several days and fixed. We found that wild-type or stabilised β-catenin significantly reduced BrdU incorporation from the already low levels in reserve cells (Fig. 5).

β-catenin is phosphorylated by GSK3β at its N-terminus and tagged for ubiquitylation and subsequent proteosomal degradation (Aberle et al., 1997). Therefore, when GSK3β is inhibited, β-catenin is stabilised, leading to its accumulation in the cytoplasm, translocation to the nucleus, interaction with TCF/LEF and to the activation of transcriptional targets. Satellite cells associated with a myofibre were exposed to 10 μM of the GSK3β inhibitor SB216763 (Coghlan et al., 2000) in plating medium for 3 days (Fig. 6a). We found that the numbers of Pax7⁺MyoD^– satellite-cell progeny were significantly increased (29.9±0.6%) compared with myofibres cultured in plating medium alone (9.5±0.33%), demonstrating that the stabilised endogenous protein was driving satellite cells towards self-renewal (Fig. 6, compare a,c and b,c). Immunostaining for Pax7 and myogenin (Fig. 6d-f) showed that fewer satellite-cell progeny were Pax7^– and myogenin-positive (Myog⁺) when GSK3β was inhibited, so fewer cells were committing to differentiation (P<0.05, Fig. 6f). Thus, the effect of drug-induced GSK3β inhibition in order to raise endogenous β-catenin levels is the same as constitutively expressing wild-type or stabilised β-catenin (Fig. 2).

Since increased levels of β-catenin promote the Pax7⁺MyoD^– phenotype and inhibit myogenic differentiation, we next asked whether decreasing levels of β-catenin have the opposite effect. β-catenin levels were reduced using siRNA-mediated gene silencing. Transfection with β-catenin siRNA caused a significant decrease in β-catenin levels in both proliferating C2 cells (GM), which remained low after three days in low-mitogen medium (DM), as shown by western blot analysis (Fig. 7a). Immunostaining showed that reduced levels of β-catenin resulted in enhanced differentiation and an increased number and size of C2-derived myotubes (Fig. 7b,c). This is consistent with a previous observation by Gavard et al. that myogenin expression is promoted when β-catenin is silenced (Gavard et al., 2004). Having established that the siRNA targeting β-catenin was effective, satellite-cell-derived myoblasts were transfected and analysed by immunostaining. As expected, in the presence of β-catenin siRNA, few satellite-cell-derived myoblasts were found with significant β-catenin immunosignal compared with those transfected using control siRNA (Fig. 7d,e). The reduction of β-catenin levels using RNA interference (RNAi), followed by culturing the cells in mitogen-poor medium for 3 days, enhanced differentiation and fusion compared with cells transfected with control siRNA (Fig. 7f-h). Significantly, when cultured in low-mitogen medium, fewer Pax7⁺ cells were observed in cells transfected with β-catenin siRNA compared with those transfected with control siRNA (6.2±1.1% compared with 21.1±0.7%, respectively; Fig. 7k; P<0.05).

To examine whether the enhanced differentiation observed with decreased β-catenin levels was due to transcriptional activity, we employed a dominant-negative version of β-catenin (Montross et al., 2000). In the β-catenin–ERD construct, the activation domain of β-catenin is replaced by the Drosophila engrailed repressor domain (ERD) that blocks the activation of target gene transcription (Han and Manley, 1993). Importantly, β-catenin–ERD is still able to associate and function with the cadherin complex, therefore allowing examination of its direct role in transcriptional regulation without perturbing its role in cadherin-mediated events (Montross et al., 2000). Myofibres were infected with pMSCV–β-catenin–ERD–IRES–eGFP, and 48 hours later fixed and immunostained (Fig. 8). The presence of β-catenin–ERD, as shown by expression of eGFP, significantly increased the number of satellite cells containing myogenin (Fig. 8a) as compared with those infected with control pMSCV-IRES-eGFP (Fig. 8b). Also, fewer satellite cells that expressed β-catenin–ERD contained Pax7 and MyoD protein (Fig. 8c). Plated satellite-cell-derived myoblasts that were infected with β-catenin–ERD-encoding retrovirus and cultured in low-mitogen medium for at least 5 days (Fig. 8d) had a higher fusion index compared with cells infected with control pMSCV-IRES-eGFP (Fig. 8e, quantified in f). These experiments show that transcriptional repression of β-catenin targets enhances myogenic differentiation.

One of the fundamental questions in stem-cell biology is: what drives the choice between self-renewal or differentiation? We have previously shown that satellite cells adopt alternative fates in culture. Whereas most cells differentiate, others maintain Pax7 protein but lose MyoD and escape immediate differentiation (Zammit et al., 2004). This provides an accessible model with which to examine factors that influence satellite-cell fate. Using this system, we show here that β-catenin drives satellite cells towards the self-renewal pathway and away from immediate myogenic differentiation.

β-catenin has been shown to control stem-cell fate; promoting self-renewal of haematopoietic stem cells (Reya et al., 2003) and neuronal precursor cells in early development (Hirabayashi et al., 2004). In myogenesis, β-catenin can activate Myf5 in somites, as part of the commitment of multi-potent stem cells to the myogenic lineage (Borello et al., 2006) and induce myogenesis in pluripotent embryonal carcinoma P19 cells (Petropoulos and Skerjanc, 2002).

α-catenin and β-catenin are present in skeletal muscle cells (Kuch et al., 1997) and we show here that mouse satellite cells in culture express β-catenin during myogenic progression, as has been shown for adult rat-derived myogenic cells (Ishido et al., 2006; Wrobel et al., 2007) and C2 cells (e.g. Goichberg et al., 2001). We found that constitutive expression of wild-type or stabilised β-catenin, or inhibiting degradation of endogenous β-catenin results in MyoD downregulation and no induction of myogenin in satellite cells, consistent with observations in C2 cells (Goichberg et al., 2001). These experimental interventions direct satellite-cell-derived myoblasts towards the self-renewal pathway, whereby Pax7 expression is maintained and cell division is reduced. Similarly, constitutively expressed β-catenin promoted the adoption of the C2-reserve-cell phenotype (Pax7⁺MyoD^–), consistent with a role in promoting self-renewal in primary satellite cells. β-catenin has been shown to induce Pax gene expression in somitic mesoderm during development (Capdevila et al., 1998) and in P19 cells (Petropoulos and Skerjanc, 2002). By contrast, when β-catenin is downregulated using RNAi or has its transcriptional targets repressed using a dominant-negative β-catenin–ERD construct, fewer Pax7⁺MyoD^– self-renewing satellite cells or C2 reserve cells are found. Inhibiting the function of β-catenin actually promotes myogenic differentiation, consistent with β-catenin silencing having a marked stimulatory effect on myogenin expression in C2 cells (Gavard et al., 2004).

β-catenin can affect cell function as part of canonical Wnt signalling. Wnt proteins comprise a large family of secreted signalling molecules that bind different Frizzled receptors and low-density lipoprotein receptor-related proteins, acting through distinct canonical and non-canonical signalling pathways. The canonical Wnt pathway involves the stabilisation of β-catenin, which can then translocate to the nucleus and control transcription as a necessary coactivator of TCF/LEF transcription factors (reviewed in Willert and Jones, 2006).

The role of Wnt proteins in developmental myogenesis is well established. Wnt1 is expressed in the neural tube and has been shown to activate expression of Myf5 in epaxial myotome, whereas Wnt7a is expressed in dorsal ectoderm and activates expression of MyoD in hypaxial myotome (Munsterberg et al., 1995; Tajbakhsh et al., 1998); Wnt3a is also able to activate MyoD (reviewed in Cossu et al., 1996; Parker et al., 2003). Recently, it has been shown that Wnt activation of Myf5 in somites involves this canonical β-catenin pathway (Borello et al., 2006). Members of the Wnt family are also implicated in regenerative myogenesis in adult, where there is evidence that Wnt signalling may be involved in myogenic activation (Rochat et al., 2004) and in recruitment of non-satellite cells to the myogenic lineage (Polesskaya et al., 2003). Interestingly, microarray studies of muscle regeneration have shown that the mRNA for secreted Frizzled-related protein increases in the early stages, indicating that Wnt signalling is actively suppressed (Zhao and Hoffman, 2004). Speculatively, this might allow sufficient satellite-cell proliferation to occur in order to generate enough myoblasts for effective muscle repair before cell-fate decisions need be made using Wnt/β-catenin signalling. Interestingly, if culturing satellite cells in their niche on an isolated myofibre together with Wnt1-producing cells, the self-renewing Pax7⁺MyoD^– phenotype is promoted (our unpublished observations), consistent with our findings on the role of β-catenin at this stage of lineage progression.

β-catenin also functions in association with the cadherin complex at the cell membrane. The C-termini of cadherins associate with β-catenin, which, in turn, can recruit the actin-binding α-catenin, thus linking adherence junctions to the actin cytoskeleton. In muscle cells, β-catenin associates with both N-cadherin and M-cadherin (Kuch et al., 1997). This interaction between cadherins and β-catenin is involved in the formation of functional adherens junctions, a process central to cell fusion (e.g. Goichberg et al., 2001). In proliferating myoblasts though, β-catenin is often found in the nucleus (Goichberg et al., 2001), and we show here that constitutive expression of wild-type or stabilised β-catenin at this time – before induction of myogenic differentiation – actually inhibits differentiation. β-catenin–ERD associates and functions normally with the cadherin complex (Montross et al., 2000), yet β-catenin and β-catenin–ERD have opposite effects on satellite-cell-fate choice. This indicates that repression of transcriptional targets elicited by β-catenin–ERD, rather than through cadherin interaction, is more likely to be the mode of action.

β-catenin–TCF/LEF and β-catenin–cadherin interactions might be antagonistic. Both nuclear localisation and LEF1-responsive reporter activation in cells expressing high levels of β-catenin can be inhibited by overexpressing N-cadherin or α-catenin (Sadot et al., 1998; Simcha et al., 1998). Stimulation of cadherin-mediated adhesion in proliferating myoblasts induces cell cycle arrest and myogenic differentiation (Gavard et al., 2004; Goichberg and Geiger, 1998). Our results suggest that accumulation of β-catenin inhibits myogenic differentiation through direct transcriptional control, rather than a mechanism that depends on N-cadherin adhesion.

In conclusion, our study shows that β-catenin influences satellite-cell-fate choice in favour of self-renewal and away from immediate myogenic differentiation. It so contributes to the maintenance of a viable stem-cell pool in adult skeletal muscle. What controls β-catenin in this role to drive satellite-cell self-renewal is currently unknown, but it is probable that it is as a component of the canonical Wnt signalling pathway.

Mice were killed by cervical dislocation; the EDL muscle was removed and myofibres were isolated as described in detail elsewhere (Rosenblatt et al., 1995). For suspension culture, myofibres were incubated in plating medium [DMEM with 10% (v/v) horse serum (PAA Laboratories) and 0.5% (v/v) chick embryo extract (ICN Flow)] at 37°C in 5% CO₂. For adherent cultures, myofibres were placed in LAB-TEK® eight-well chamber slides (Nunc) coated with 1 mg/ml Matrigel (Collaborative Research Inc.) in plating medium and maintained at 37°C in 5% CO₂. Satellite cells were then passaged and re-plated at high density and switched to low-mitogen medium [DMEM with 2% (v/v) horse serum] 24 hours later, and cultured for several days to examine self-renewal and myogenic differentiation. C2 myogenic cells (Yaffe and Saxel, 1977) were cultured in DMEM with 10% (v/v) foetal calf serum or in DMEM with 2% (v/v) horse serum. Where used, BrdU was added to the medium at a final concentration of 10 μM. GSK3β was inhibited by exposing cells to 10 μM SB216763 (Coghlan et al., 2000), obtained from TOCRIS bioscience. Cultures were fixed in 4% paraformaldehyde in PBS for 12 minutes and then rinsed several times in PBS.

Fixed myofibres were permeabilised with 0.5% (v/v) Triton X-100 in PBS and non-specific antibody binding was blocked using 20% (v/v) goat serum in PBS (Beauchamp et al., 2000). Primary antibodies were applied overnight at 4°C and included monoclonal anti-BrdU (clone BU1/75: Abcam), anti-β-catenin (Cell Signalling), anti-myogenin (clone F5D: DakoCytomation or DSHB), anti-MyoD1 (clone 5.8a: DakoCytomation), anti-Pax7 (DSHB), anti-MyHC (MF20), rabbit polyclonal anti-MyoD (Santa-Cruz), anti-myogenin (Santa-Cruz), anti-eGFP (Molecular Probes) and anti-β-catenin (Abcam). Primary antibodies were visualised with fluorochrome-conjugated secondary antibodies (Molecular Probes) before mounting in DakoCytomation Faramount fluorescent mounting medium containing 100 ng/ml 4,6-diamidino-2-phenylindole (DAPI).

The retroviral backbone pMSCV-puro (Clontech) was modified to replace the puromycin selection gene with an IRES-eGFP in order to create pMSCV-IRES-eGFP, which served as the control (Zammit et al., 2006b). Wild-type β-catenin cDNA was then cloned into pMSCV-IRES-eGFP to produce pMSCV–β-catenin–IRES-eGFP, producing β-catenin as a bi-cistronic message with eGFP. We also used a stabilised version of β-catenin (ST-β-catenin), containing mutations of amino acids Ser33, Ser37, Thr41, and Ser45 to Ala, to prevent phosphorylation and degradation (Barth et al., 1999), in order to produce pMSCV–ST-β-catenin–IRES–eGFP. In the β-catenin dominant-negative construct the C-terminal transactivation domain of β-catenin was replaced with the active repression domain of Drosophila Engrailed (Montross et al., 2000). β-catenin–ERD was cloned in pMSCV-IRES-eGFP, generating pMSCV-β-catenin ERD-IRES-eGFP, which was used to repress β-catenin transcriptional targets. Retroviruses were then packaged in 293T cells using standard methods.

5×10³ satellite cells or 2×10³ C2 cells were plated in LAB-TEK® eight-well chamber slides (Nunc). The following day medium was replaced with undiluted 293T supernatant supplemented with 4 μg/ml polybrene and left at 37°C for 3 hours, before the cells were rinsed and placed in fresh mitogen-rich medium. Infection rate in C2 cells was calculated to be usually ∼80% (e.g. control-R, 91.8±0.76%; pMSCV–β-catenin–IRES–eGFP, 76.3±1.03%; pMSCV-ST-β-catenin–IRES–eGFP, 82.5± 1.96%). Values are given as percentage of eGFP⁺ cells per total (DAPI⁺) cells derived from five random fields per condition (total ∼300 cells) from each of the three independent experiments. To infect satellite cells associated with myofibres, a dilution (1:10) of the retroviral containing supernatant – without polybrene or medium change – was used.

Cells were plated in six-well plates and siRNA transfection was performed at 30% confluence. siRNA duplexes (Stealth siRNA; Invitrogen) were diluted in OptiMEM (Invitrogen) to 20 pmol per well and incubated with Lipofectamine 2000 (Invitrogen) diluted in OptiMEM, according to the manufacturer's instructions. A second transfection was carried out 24 after the first. To induce myogenic differentiation, transfected cells were transferred to LAB-TEK® eight-well chamber slides (NUNC) and incubated in low-mitogen medium for 72 hours. The sequence for β-catenin siRNA was 5′-GGACGTTCACAACCGGATTGTAAT-3′ with a control siRNA selected by Invitrogen.

Immunostained myofibres and plated cells were viewed on a Zeiss Axiophot 200M microscope, and digital images acquired with a Charge-Coupled Device (Zeiss AxioCam HRm) using AxioVision software version 4.4 (Zeiss). Images were optimised globally and assembled into figures using Adobe Photoshop.

Where satellite cells associated with an isolated myofibre were infected and analysed, eGFP⁺ cells were scored for protein content from multiple fibres until n was ∼400-500 cells per mouse; at least three mice were analysed. Results are expressed as the percentage of total eGFP⁺ cells with a given antigen. For plated satellite cells, ten random fields were examined, which usually equated to ∼400-500 eGFP⁺ cells in total per mouse; at least three mice were analysed. For C2 cells, ∼400-500 cells were examined from each experiment. In all cases, data were pooled and cell population is expressed as the mean ± s.e.m.; P<0.05 was considered significantly different between conditions, calculated by using Student's t-test.

We thank Silvia Brunelli for much help; Angela Barth, Ilona Skerjanc and Helen Petropoulus for generously sharing reagents; and the colleagues who made their antibodies available through the Developmental Studies Hybridoma Bank. A.P.R. was funded by a Fundación Ramón Areces post-doctoral fellowship and briefly by the Muscular Dystrophy Campaign, V.F.G. received support from Italian MIUR and EMBO fellowships and is now supported by The Medical Research Council (grant number G0700307), while Y.O. is funded by the Muscular Dystrophy Campaign. The laboratory of P.S.Z. is supported by The Medical Research Council UK, The Muscular Dystrophy Campaign and the Association of International Cancer Research. We acknowledge the support of the MYORES Network of Excellence, contract 511978, from the European Commission 6th Framework Programme.