Canonical Wnt signalling induces satellite-cell proliferation during adult skeletal muscle regeneration

Adult mammalian skeletal muscle displays a tremendous capacity for regeneration, a point exemplified by examining the histological characteristics of muscle following acute chemical insult when large areas of damaged tissue are totally restored to normal architecture within 10 days. The normal nuclear turnover rate of adult skeletal muscle is relatively low, and has been estimated in the rat to be approximately 1-2% of myonuclei per week (Schmalbrunch and Lewis, 2000). However, large numbers of myogenic cells are required during muscle growth and regeneration. The nuclei within a syncytially organised myofibre exist in a postmitotic state, thus rendering them incapable of providing cellular material for repairing damaged tissue or postnatal growth.

There is much controversy concerning the identity and origins of cells that promote skeletal muscle regeneration. In a historical context, the first cells thought to be responsible for skeletal muscle repair and postnatal muscle growth were described by Katz (Katz, 1961) and Mauro (Mauro, 1961), who identified a unique population situated between the basal lamina and sarcolemma of muscle fibres. Termed satellite cells because of their peripheral location relative to the muscle fibres, these cells are normally found in a quiescent state. During periods of muscle regeneration and growth, satellite cells become activated, then proliferate and fuse either with one another or with existing muscle fibres. Proliferating satellite cells can also exit the cell cycle to replenish the stem cell niche in anticipation of future rounds of regeneration.

In addition to satellite cells, numerous other cell types have been shown to be able to participate in muscle regeneration. These cell types can be split broadly into two groups: cells that are found outside skeletal muscles and those that are found within muscle. However, there is much debate concerning the efficacy and relevance of non-satellite-cell-mediated skeletal muscle regeneration and in particular, whether these cells contribute to the satellite-cell pool. The case for satellite cells being the primary mediators of skeletal muscle repair was strengthened by a study in which single genetically labelled muscle fibres were transplanted into irradiated injured muscle, which led to the regeneration of muscle solely from the marked tissue and thus excluding the contributions, to any considerable level, of other sources (Collins et al., 2005).

Very little is known about the mechanisms that control satellite-cell proliferation. We developed a two-stage programme to address this issue by first establishing the identity of molecules that control the development of the precursors of satellite cells during embryogenesis (Otto et al., 2006). We found that members of the Wnt family of signalling proteins were potent regulators of somitic expression of Pax7, which is a marker for adult satellite cells (Otto et al., 2006). We also analysed the role of Wnt proteins in mediating the adult satellite-cell response to injury.

The Wnt family of signalling molecules consists of 21 secreted glycoproteins (Cadigan and Nusse, 1997). Wnt signalling can be divided into three main branches, the canonical, the planar-cell-polarity (PCP or non-canonical) and the calcium-dependent pathway. Studies revealed that Wnt1, an archetypal canonical Wnt, is a secreted protein. However, following exit from the cell, it binds to the extracellular matrix and therefore only signals over short distances (Bradley and Brown, 1990). A wealth of data exists documenting the influence of Wnt proteins on embryonic muscle development. Their action impacts all aspects of muscle formation, from induction to commitment to the myogenic lineage and proliferation of muscle cells (Galli et al., 2004; Parr et al., 1993; Stern et al., 1995; Tajbakhsh et al., 1998; Wagner et al., 2000).

By contrast, we have only a sketchy understanding of the role that Wnt signalling plays during adult skeletal muscle regeneration. One study has shown that the side population of stem cells from muscle can be induced by Wnt proteins into the myogenic lineage, which then participates in muscle regeneration (Polesskaya et al., 2003). By contrast, another study has reported that Wnt proteins promote the fibrogenic conversion of adult muscle cells (Brack et al., 2007).

In this study, we focused on determining the role of Wnt signalling on satellite cells. We exploited three timely developments: (1) the advent of a protocol that permits the in vitro culturing of whole skeletal muscle fibres in the absence of all other tissue, thereby facilitating the study of satellite cells as the sole mononuclear cell type; (2) the development of cell lines that secrete specific Wnt proteins; (3) the development of a protocol that allows the culture of isolated skeletal muscle fibres on Wnt-expressing cells. We show that the Wnt signalling is active during satellite-cell proliferation. Furthermore, we present data demonstrating that the action of Wnt proteins depends on the identity of the ligand such that the action can either promote the proliferation of satellite cells or induce a mitotically inactive, yet reversible state. The differing outcomes of Wnt action are mirrored by the distribution of Act-β-Cat, a key component of the canonical Wnt-signalling pathway. During proliferation, this transcriptional activator becomes localised to the nucleus. By contrast, it is excluded from the nucleus when satellite cells are not dividing. We also show that the relationship between Act-β-Cat and proliferation/quiescence is maintained during both acute and chronic skeletal muscle regeneration in rodents and humans.

The role of Wnt proteins during the development of skeletal muscle is well documented. There is considerable evidence that members of the gene family are expressed in the correct temporal and spatial domain for them to play a vital role in myogenesis. Adult muscle development is not a very amenable system to investigate the expression of components of the Wnt-signalling pathway; however, the use of single muscle fibres does allow us to overcome some of the major drawbacks. Satellite-cell activation is a relatively fast process once muscle has been injured. The enzymatic removal of connective tissue and physical dispersal of a muscle into single fibres results in the activation of satellite cells. Activated satellite cells undergo proliferation, but in the floating fibre culture system, all daughter cells remain adhered to the myofibre. We carried out four sets of experiments to show the presence of Wnt-signalling components and active Wnt signalling in isolated single muscle fibres.

In the first series of experiments, we established the presence of the Wnt ligands in muscle fibres. Single muscle fibres were isolated and then cultured for 48 hours. Subsequently, they were harvested and used as a source for RT-PCR analysis. Using gene-specific primers, we were readily able to detect transcripts for Wnt1, Wnt3a, Wnt5a and Wnt11 (Fig. 1A).

Second, we analysed activated single muscle myofibres for the presence of canonical Wnt-signalling activity using a reporter cell assay that is sensitive to Wnt-mediated signalling. Rat osteogenic sarcoma (ROS) cells were transfected with the Super8XTOPFlash canonical Wnt reporter construct, which consists of eight TCF/LEF sites and a minimal promoter linked to the luciferase gene. The same ROS cells were also transfected with a constitutively active pRL-CMV Renilla luciferase reporter construct enabling levels of Super8XTOPFlash reporter activity to be normalised. As a positive control, we incubated transfected cells with lithium chloride for 30 hours. This salt is an inhibitor of GSK3β function that prevents the phosphorylation and degradation of β-catenin (official protein symbol CTNB1) (Stambolic et al., 1996; Aberle et al., 1997) and thus is a potent activator of canonical Wnt signalling. ROS cells alone were used as a negative control. As expected, transfected ROS cells stimulated with lithium chloride showed a 2.5-fold increase in TCF/LEF activation compared with nonstimulated basal rates. Intriguingly, ROS cells transfected with Super8XTOPFlash cocultured with activated myofibres for 30 hours showed a significant 2.15-fold increase in reporter activity when compared with the untreated control group (Fig. 1B, P<0.01). Therefore, the results of the TOPFlash studies suggest the presence of robust canonical Wnt signalling in the muscle fibres.

Next, we confirmed the presence of Wnt signalling in activated proliferating satellite cells by using epigallocatechin-3-gallate (EGCG), a potent canonical Wnt inhibitor that acts by inducing β-catenin breakdown (Dashwood et al., 2002). We cultured single myofibres in the presence of EGCG for 36 hours and 72 hours, and assessed the effect on satellite-cell proliferation by quantifying the number of satellite-cell progeny per myofibre. Culture of myofibres for 72 hours allows the majority of satellite cells to initiate cell proliferation.

At 36 hours under control conditions, an average of 14.7 satellite-cell progeny per myofibre was seen. Significantly, only 11.3 progeny cells per myofibre were seen in EGCG-treated cultures (n=50 and n=42 myofibres for control and EGCG-treated cultures, respectively; P=0.004). Following 72 hours of culture, satellite cells under control conditions gave an average of 75.3 progeny cells per myofibre. EGCG-treated cultures at 72 hours showed a smaller increase in progeny cells, producing just 31.9 cells per myofibre (Fig. 1C, P<0.0001). These data indicate the active presence of canonical Wnt signalling during the initial satellite-cell response to injury, and that the inhibition of Wnt signalling causes a significant decrease in satellite-cell proliferation.

We then carried out a detailed temporospatial expression profile of β-catenin, a key component of the canonical Wnt pathway. We found that satellite cells from freshly isolated fibres expressed nonactivated β-catenin (Fig. 1D,E), but importantly, did not express the activated form of the protein (Fig. 1F). We followed the activation process by determining the expression of the transcription factor MyoD, as well as the distribution of activated-β-catenin (Act-β-Cat). We found that MyoD, which serves as a marker for activated satellite cells, was expressed within 2 hours and at this time point, we were unable to detect Act-β-Cat (data not shown). Similarly, following 6 hours in culture, we found MyoD⁺ satellite cells remained negative for Act-β-Cat (Fig. 1G, red arrowhead), although interestingly, myonuclei showed weak Act-β-Cat expression that was maintained throughout the 72 hour time course (Fig. 1G, green arrowhead). However, between 24 and 48 hours, we were able to detect robust expression of Act-β-Cat in MyoD⁺ satellite cells. Furthermore, at this time, the majority of Act-β-Cat was found in the nucleus, determined by its colocalisation with MyoD (Fig. 1H,I). We determined whether nuclear Act-β-Cat was directly associated with satellite-cell proliferation by examining its localisation to the nucleus during the cell cycle; Act-β-Cat was localised to the nuclei of satellite cells as they prepared to undergo cell division, as indicated by the colocalisation of Act-β-Cat and incorporated BrdU (Fig. 1O-R). Immediately after cell division, Act-β-Cat was excluded from the nucleus and found at the interface between two daughter cells (Fig. 1J). Quantification of BrdU-positive satellite-cell progeny showed that 91.8% of BrdU⁺ cells showed nuclear Act-β-Cat expression following just 36 hours of culture (Fig. 1O,P, yellow and white arrowheads, and Fig. 1S) and 74.5% of nuclei were positive for both proteins following 72 hours of culture (Fig. 1S). Notably, at 72 hours in clonal clusters, proliferating BrdU⁺ cells were also positive for nuclear Act-β-Cat; however, BrdU^– cells showed weaker peripheral Act-β-Cat expression (Fig. 1Q,R red and white arrowheads, respectively). After 72 hours in culture, progeny of satellite cells can differentiate, indicated by the detection of myogenin. We observed that at the onset of differentiation, there was a mutually exclusive distribution of myogenin and Act-β-Cat expression. In 20% of cases (n=1232 clusters analysed), we were able to detect small clusters in which one cell expressed myogenin with no expression of Act-β-Cat, which was in direct contact with other cells that expressed Act-β-Cat but no myogenin (Fig. 1M,N). On rare occasions, we detected large clusters formed exclusively of cells that expressed Act-β-Cat that was only localised within the nucleus. In these clusters, we never detected the expression of myogenin (Fig. 1K,L), although MyoD expression was maintained (data not shown). Therefore, expression of Act-β-Cat in the nucleus labels cells that are undergoing cell division, and it becomes excluded from the nucleus after division. Furthermore, Act-β-Cat is downregulated in cells that undergo differentiation but is maintained in those that do not.

After establishing the presence of Wnt ligands in isolated fibres, as well as intracellular components of the Wnt-signalling pathway in activated satellite cells, we then determined the response of the cells to exogenous sources of Wnt proteins. In particular, we wanted to determine the response of satellite cells to Wnt proteins that activate the canonical or the noncanonical pathway, because recent studies suggest that they act in an antagonistic manner (Mickels and Nusse, 2006). Experimentally, this was achieved by coculturing freshly isolated single fibres with different Wnt-expressing cells and determining both the total progeny number and average cluster size on each muscle fibre. Quantification of satellite-cell progeny on single fibres following 36 hours of coculture was achieved through the visualisation of immunocytochemically labelled Pax7⁺/DAPI⁺ nuclei. Remarkably, the response of the satellite cells fell into two groups. One set of Wnt-expressing cells, Wnt1, Wnt3a and Wnt5a, induced a higher degree of cell division than in fibres either cultured with control cells expressing LacZ or in culture medium alone (Fig. 2A-C,F). For example, fibres exposed to ectopic Wnt1 had more Pax7-positive cells on each fibre (Fig. 2B,C) and when the size of each cluster was calculated, the cluster size was significantly larger than that in fibres exposed to LacZ-expressing cells (Fig. 2A). All cells that constituted a cluster were positive for Pax7, suggesting that these cells were indeed myogenic. Satellite cells treated with Wnt1, Wnt3a or Wnt5a, all gave significantly more satellite cells per cluster (2.6, P<0.0001; 2.3, P=0.0002 and 2.4, P=0.0001, respectively) when compared with an average of 1.35 in LacZ control cultures (Fig. 2A). Fibres exposed to Wnt1, Wnt3a or Wnt5a showed a remarkable increase in the number of satellite cells that had undergone at least one division (Fig. 2C,F) compared with control fibres, where satellite cells rarely divided more than once (Fig. 2E,H).

In contrast to the effect of Wnt1, Wnt3a or Wnt5a, we found that coculture of single fibres with Wnt4 or Wnt6, resulted in an acute reduction of satellite-cell proliferation. Total progeny numbers per myofibre were only 7.2 and 7.8, respectively, significantly lower than the control average of 10.45 (Fig. 2B,D,G; P<0.001 and P=0.015, respectively). Moreover, the number of satellite cells per cluster was also significantly lower in the noncanonical Wnt-treated cultures when compared with the controls (1.03 for Wnt4, P<0.0001 and 1.1 for Wnt6, P<0.0001, compared with 1.35 for controls, Fig. 2A). In order to assess whether the differing effects of canonical and noncanonical Wnt proteins on satellite-cell progeny numbers were due to changes in satellite-cell proliferation as opposed to activation, we quantified activated satellite cells through MyoD expression following 2 hours of culture with Wnt1 or Wnt6, or under control conditions. No statistical difference was seen between the numbers of activated MyoD⁺ satellite cells on either Wnt1- or Wnt6-treated myofibres when compared with the control (P=0.44 and P=0.2, for Wnt1 and Wnt6, respectively, Fig. 2I) suggesting that changes in satellite-cell progeny numbers were due to altered rates of proliferation.

In summary, these results show that the action of Wnt proteins can be categorised into two groups. One group, including Wnt1, Wnt3a and Wnt5a, enhanced the rate of cell division. By contrast, another group, which included Wnt4 and Wnt6, not only failed to promote satellite-cell proliferation, but repressed it and did so to a similar degree as the canonical Wnt inhibitor EGCG.

Subsequently, we addressed the question of whether satellite cells that had been inhibited from proliferating in the presence of Wnt4 or Wnt6 retained the capacity to undergo cell division, or whether they had been permanently forced into quiescence. To this end, we devised a regime in which fibres were exposed to one set of conditions and after 36 hours were transferred to another environment. For these experiments, we chose Wnt1- and Wnt6-expressing cells as representatives of the two categories of Wnt proteins defined above.

Fibres from the extensor digitorum longus (EDL) muscle cultured sequentially with Wnt6-expressing cells displayed a significantly attenuated cell division capacity compared with fibres grown with nontransformed fibroblasts (Fig. 3A,D, P<0.02). However, when fibres that had been cultured with untransfected fibroblasts were transferred to Wnt6-expressing cells, there was an increase in the number of satellite-cell progeny compared with that in fibres cultured with Wnt6 alone (Fig. 3A). Furthermore, a greater stimulation in proliferation was found when fibres were first cultured in Wnt6-expressing cells and then transferred to Wnt1-expressing cells compared with transferring them to control cells (Fig. 3A, P<0.005). We also examined the effect of Wnt1 on cells that had already started to proliferate. We found that fibres cultured with control cells for the first half of the experiment before transferring to Wnt1-expressing cells, resulted in a significantly greater number of progenitors compared with cells maintained in control conditions (Fig. 3A, P<0.001). Furthermore, we found that fibres cultured for a total period of 72 hours with control cells had numerous cells that expressed myogenin, the marker for differentiation (Fig. 3B). By contrast, fibres that that been cultured with control cells for the first 36 hour period before an additional similar time period with Wnt1-expressing cells, showed very few myogenin-positive cells (Fig. 3C). However, the cluster sizes were larger than those of the controls. Fibres that were raised on Wnt1-expressing cells were not easy to transfer to other cells because some clusters were large and did not adhere strongly to the fibre. In addition, the fibres grown in Wnt1 were not readily separated from the fibroblasts and so these experiments were impossible to quantify.

These results show that cells that have been exposed to Wnt6 can be activated once they are removed from the inhibitory effect of Wnt6. Furthermore the cells respond to the proliferation-enhancing effects of Wnt1. Wnt1 acts not only on the initial stages of satellite-cell proliferation immediately after activation from the quiescent state, but can also enhance division of cells that have been proliferating for a considerable time.

We then examined the localisation of Act-β-Cat in the presence of different Wnt proteins to determine a mechanistic explanation of the actions of the proliferation-promoting or proliferation-inhibiting Wnt molecules. During the normal programme of satellite-cell proliferation, we have shown that Act-β-Cat becomes localised to the nucleus during cell division, and at the end of division it translocates out of the nucleus to a membrane-associated region. Wnt1 and Wnt6 were used as representatives of the two classes of Wnt proteins. Mature EDL myofibres were cocultured in the presence of Wnt1- and Wnt6-expressing NIH3T3 cells, and satellite cells were analysed for the expression of nuclear Act-β-Cat. Remarkably, following just 2 hours of culture in the presence of Wnt1, satellite cells coexpressing nuclear Act-β-Cat and MyoD were observed (Fig. 4A,D,G). Under Wnt6 conditions, satellite cells were found expressing Act-β-Cat; however, this expression, where present, was cytoplasmic and never nuclear (Fig. 4B,E,H), similarly to that in control progeny at 72 hours of incubation. Control cultures predominantly showed no Act-β-Cat expression in MyoD⁺ satellite cells following 2 hours in culture (Fig. 4C,F,I). Again, as shown in Fig. 1G, weak Act-β-Cat was occasionally seen in the myonuclei of the single fibres (Fig. 4C,I).

To confirm the subcellular localisation of Act-β-Cat to the nuclei we used the expression domain of MyoD as an indicator of the position of the nucleus. We analysed confocal images generating fluorescence profiles for both Act-β-Cat and MyoD through image analysis software showing the relative fluorescence intensity between two set points on a sample. Fig. 4J shows the fluorescence profile for the satellite-cell nucleus shown in Fig. 4G between the asterisk-marked points. Clear colocalisation of Act-β-Cat and MyoD signal is shown over the profile plot. Conversely, the profile for the nucleus shown in Fig. 4H shows that in the presence of ectopic Wnt6, Act-β-Cat is maintained in a non-nuclear domain flanking either side of the nuclear MyoD (Fig. 4K). Finally, Fig. 4L shows the control profile of the satellite-cell nucleus shown in Fig. 4F,I (red arrowhead). No significant Act-β-Cat expression was detected in the satellite cell following 2 hours of culture. However, MyoD was robustly expressed in the nucleus.

To establish whether Wnt1-treated satellite cells expressing nuclear Act-β-Cat at 2 hours were undergoing proliferation, we quantified the nuclear colocalisation of Act-β-Cat with BrdU. In Wnt1-treated conditions, BrdU⁺ Act-β-Cat⁺ events were frequently detected and found at a rate of 1.52 satellite-cell nuclei per myofibre following 2 hours of culture. By contrast, BrdU⁺ Act-β-Cat⁺ satellite-cell nuclei were an incredibly rare occurrence in myofibres cultured with LacZ-expressing cells (one event in every 15-17 myofibres Fig. 4M). Furthermore, BrdU⁺ Act-β-Cat⁺ satellite-cell nuclei were never seen in Wnt6-treated cultures (Fig. 4M).

These results show that although no changes in satellite-cell activation, determined via MyoD expression, were seen under the different Wnt conditions, the presence of a Wnt protein that promotes cell division leads to increased numbers of satellite cells entering the cell cycle and the premature expression of Act-β-Cat protein, which becomes localised to the nucleus. However, in the presence of a Wnt protein that inhibits proliferation, the premature expression of Act-β-Cat is still observed, but the protein fails to move into the nucleus and the cells do not enter the cell cycle.

Our observations that the subcellular localisation of Act-β-Cat depended on whether a satellite cell is undergoing proliferation or not was investigated in an in vivo setting in two independent muscle regeneration scenarios. In the first set of experiments, we determined the localisation of Act-β-Cat in mouse skeletal muscle at differing stages of regeneration following cardiotoxin-induced damage. We found that during the first 3 days after cardiotoxin injection, the muscle had undergone severe degeneration (data not shown). However 6 days following injury, signs of robust regeneration were evident, including the appearance of large numbers of heterogeneously sized muscle fibres that contained centrally located nuclei (Fig. 5A). We examined the distribution of Act-β-Cat at this stage and found considerable expression of the protein in a nuclear domain judged by its colocalisation with DAPI (Fig. 5B,F-H). The Act-β-Cat was located within the boundaries of the muscle fibre as determined by the colocalisation with laminin (Fig. 5C). The expression of Act-β-Cat was substantially different when we examined muscle that had been allowed to regenerate for a longer period. Fifteen days after injury, the muscle architecture had returned to a more mature appearance (Fig. 5I) and the fibre size was more uniform. The expression of Act-β-Cat was considerably weaker than at 6 days. Furthermore, when present, it was no longer localised in the nucleus (Fig. 5J,K,N-P).

Then we determined the expression profile during muscle regeneration in humans. In samples showing overt muscle regeneration, we found a similar profile for Act-β-Cat distribution as the 6-day post injury mouse muscle: Act-β-Cat was located in the nucleus of cells found within the basement membrane of muscle fibres (Fig. 5Q-T). However, in a sample that lacked signs of muscle regeneration, the expression of Act-β-Cat was almost totally absent and never localised to the nucleus (Fig. 5U-X).

In order to ascertain whether nuclear Act-β-Cat localisation is present in proliferating satellite cells in vivo, antibody staining for both Pax7 and Act-β-Cat along with the nuclear antigen Ki67 (which labels cells capable of entering, or in, S phase) was carried out on human regenerating and nonregenerating muscle biopsy samples (Fig. 5Y-F1). Similarly to in vitro data where BrdU was utilised, colocalisation of Pax7 and Ki67 was apparent in regions of regenerating muscle tissue, suggesting that satellite cells were actively dividing (Fig. 5Y,Z, yellow and white arrowheads, respectively). Importantly, these Ki67⁺ satellite cells also showed robust nuclear expression of Act-β-Cat (Fig. 5A1,B1, yellow and white arrowheads, respectively) suggesting the importance of nuclear Act-β-Cat during in vivo satellite-cell proliferation. As expected, nonregenerating muscle showed no regions of damage and Pax7⁺ satellite cells were negative for Ki67 (Fig. 5C1,D1, green and white arrowheads, respectively). Moreover, Act-β-Cat expression in this tissue was localised to small membranous regions or non-nuclear pockets (Fig. 5E1,F1, green arrowheads) and did not show Ki67 nuclear localisation.

The expression of Act-β-Cat therefore shows a comparable expression profile, which is dependent on the degree of satellite-cell activation, in two different models of muscle regeneration. In both models, Act-β-Cat is located in the nucleus during regeneration and after regeneration its expression not only decreases, but also is no longer located in the nucleus.

Satellite cells have a remarkable capacity to repair damaged skeletal muscle. Quantitative analysis carried out by Collins et al. (Collins et al., 2005) of the regenerative capacity of satellite cells has elegantly demonstrated this property. They were able to show that approximately 30,000 myonuclei developed in a matter of 3 weeks from a single transplanted muscle fibre that contained ten satellite cells. However, the mechanisms controlling satellite cell activation and proliferation are only just beginning to be understood.

This study relied on the timely development of a powerful tissue culture protocol, first devised by Bischoff et al. (Bischoff et al., 1986) but more recently refined by Zammit and colleagues (Zammit et al., 2002), which permits the isolation of muscle fibres decorated with their satellite cells in a nearly pure state. Importantly, contamination, especially from cells of the connective tissue, can be easily avoided. Isolated myofibres retain their satellite cells beneath the basal lamina that become activated to initiate the expression of myogenic markers, culminating first in cell proliferation, and eventually in differentiation. Zammit et al. (Zammit et al., 2004) have performed a detailed investigation into the temporal nature of these events. They showed that quiescent satellite cells could be identified by their expression of Pax7, a protein that is not expressed in the myonuclei. The process of isolating single fibres leads to the activation of satellite cells, an event that is marked by the expression of MyoD and the maintenance of Pax7 expression. Activated satellite cells undergo cell division for the first time 24 hours after fibre isolation, and all daughter cells continue proliferating and coexpressing MyoD and Pax7. After approximately 72 hours in culture, the progeny appear to make a decision regarding their future development. They either continue to coexpress MyoD and Pax7 and divide, or downregulate Pax7, upregulate myogenin, and subsequently differentiate. Alternatively, the cells can downregulate MyoD while maintaining Pax7 expression and return once more to their quiescent state. This temporal sequence has been used as a platform for our study in the assessment of the influence of Wnt proteins on the regulation of satellite-cell activity.

We provide four lines of evidence to demonstrate that Wnt signalling is active during satellite-cell proliferation. First, we established that Wnt signalling has an active role during satellite-cell activation and proliferation by surveying single muscle fibres for the production of Wnt ligands and ascertaining the presence of an active signalling cascade. Second, we found that a number of Wnt ligands are expressed during the first 24 hours after fibre isolation. Our data not only confirmed previous reports of Wnt5a expression in muscle (Polesskaya et al., 2003), but also now extends the list to include Wnt1, Wnt3a and Wnt11. Interestingly, we were unable to detect the expression of Wnt4 or Wnt6 in fibres that had been cultured for 24 hours. Our experiment did not resolve whether the Wnt proteins were being expressed by the satellite cells or the myotubes. However, other authors have shown that both cell types can express each of the Wnt proteins we have studied (Polesskaya et al., 2003). Third, we made the interesting and novel discovery that isolated myofibres containing satellite cells following 48 hours in culture were able to activate a TCF/LEF luciferase reporter construct, suggesting that the fibres and/or satellite cells produce canonical Wnt proteins during a muscle-repair response. Levels of TCF/LEF reporter activated by single fibres were similar to those induced by reporter cells stimulated with 10 mM lithium chloride, a known canonical Wnt-signalling activator (Klein and Melton, 1996), indicating that relatively high levels of Wnt are being produced by the satellite cells and/or myofibres. The lack of specific antibodies to the relevant Wnt proteins prevented us from making a quantitative analysis of exactly how much Wnt protein is actually being produced. In addition, we showed that ECGC, a potent inhibitor of Wnt signalling, significantly inhibited the proliferation of satellite cells. This is noteworthy as it establishes that autocrine or paracrine Wnt signalling is promoting the proliferation of satellite cells. Last, we showed that β-catenin, a key component of the canonical Wnt-signalling cascade, is present in quiescent satellite cells in the inactive form, but subsequently becomes activated following satellite-cell activation. This observation suggests that the proliferation initiated by the Wnt-signalling cascade does not have to rely on transcription of β-catenin, but rather on activation of this protein, which is already present within the quiescent satellite cells. This seems quite a common mechanism, because it is found in many other tissues, including human colorectal tissue and B cells (Iwao et al., 1998; Khan et al., 2007).

We demonstrate that the proliferation of satellite cells is controlled by the identity of the Wnt ligand. In our screen of five different Wnt proteins, we found that the action of these signalling molecules could be grouped into two readily distinguishable categories. Wnt1, Wnt3a and Wnt5a all induced a statistically greater degree of proliferation than control cells. By contrast, Wnt4 and Wnt6 inhibited proliferation. Comparisons are invited between the identity of ligands in these two categories and the ex vivo expression of Wnt proteins following fibre isolation. We were able to detect the expression of all the activating Wnt proteins, but were unable to detect the expression of inhibitory Wnt proteins. These data correspond with published work that demonstrated that inhibitory Wnt proteins are rapidly downregulated following fibre isolation (Polesskaya et al., 2003). By assimilating these findings, we can offer an explanation as to why a degree of satellite-cell proliferation takes place in the presence of the inhibitory Wnt proteins. We propose that the action of preparing fibres leads to the production of the activating Wnt proteins and that the external coculture-derived inhibitory Wnt proteins must overcome this effect completely to shut down proliferation. The results suggest that they do so only partially.

The intracellular response of β-catenin, a key component of the canonical Wnt-signalling pathway to the activation of satellite-cell proliferation, reveals a mechanistic insight into how the signalling molecules determine whether a cell should divide or not. We show that in untreated fibres, the satellite cells become activated, marked by the expression of MyoD, which is followed by the appearance of Act-β-Cat. In untreated samples, Act-β-Cat is found initially solely in the nucleus until the cells divide, at which point it is translocated out of the nucleus and becomes primarily located at the membrane interface between the two daughter cells. During later stages of development, the nuclear expression of Act-β-Cat is totally absent in cells that have withdrawn from the cell cycle and are undergoing differentiation, marked by the expression of myogenin (Zammit et al., 2004). However, we readily detected the nuclear localisation of Act-β-Cat in cells directly adjacent to Act-β-Cat^– myogenin⁺ nuclei (Fig. 1M). We suggest that in this situation, one cell was progressing along the differentiation pathway whereas the other was once more undergoing cell division. Our data showing nuclear location of Act-β-Cat during cell division and non-nuclear localisation after division or differentiation is highly reminiscent of its distribution during growth and differentiation in the C2C12 and L8 skeletal muscle cell lines (Goichberg et al., 2001). Furthermore, it is tempting to speculate, based on work performed in very many tissues, including skeletal muscle cells lines, that nuclear translocation of Act-β-Cat could result in the expression of cyclin D1 (CCND1), which is a potent activator of the cell cycle (Shtutman et al., 1999; Tetsu and McCormick, 1999, Goichberg et al., 2001). Furthermore, our observation of the mutually exclusive expression profile of Act-β-Cat and myogenin can be explained by the finding that Act-β-Cat represses the expression of myogenin but not MyoD, thus offering an attractive model to explain the regulation of proliferation versus the control of differentiation (Petropoulos and Skerjanc, 2002; Goichberg et al., 2001).

A novel finding of our study concerned the discovery that the two classes of Wnt proteins, causing radically differing outcomes in terms of cell division, both resulted in an upregulation in the levels of Act-β-Cat. We suggest that the usual distinction, in terms of the role of β-catenin during canonical versus noncanonical Wnt signalling, could be applied in satellite cells based not on whether it is present or not, but rather on its cellular function. We suggest that the defining feature of canonical Wnt signalling is the nuclear translocation of Act-β-Cat, permitting it to act as a transcriptional activator. There are now many studies documenting the presence of Act-β-Cat that nevertheless fails to initiate Wnt-mediated gene transcription (e.g. Na et al., 2007). A recent study has shown that Act-β-Cat can exist in a number of functionally related conformations. Act-β-Cat in specific conformations can act as a transcriptional activator, whereas other conformations are localised outside of the nucleus (Gottardi and Gumbiner, 2004). In light of the model of Gottardi and Gumbiner (Gottardi and Gumbiner, 2004), we suggest that Wnt1, Wnt3a and Wnt5a result in the accumulation of Act-β-Cat in a conformation that allows it to act as a transcriptional activator. On the other hand, Wnt4 and Wnt6 lead to the accumulation of a type of Act-β-Cat that forms a dimer, which is able to interact with cadherins and α-catenin, localised to the cell membrane and not the nucleus.

We provide evidence that when satellite cells that have been exposed to an inhibitory Wnt are moved to an environment either lacking the inhibitory signal or to an environment containing proliferation-supporting signalling, they respond by initiating robust cell division. Therefore, a cessation in proliferation is not marked by the onset of differentiation. However, the opposite may be the case: that is, satellite cells in a proliferation-enhancing environment are able to differentiate. The reversible nature of inhibition/proliferation is compatible with the distinct forms of Act-β-Cat that arise as a result of specific Wnt action, as discussed above. We propose that the inhibitory Wnt proteins would give rise to the membrane-associated form of Act-β-Cat and continue to do so as long as that signal is being received by the satellite cells. Once satellite cells are no longer influenced by the inhibitory signal they would express the proliferation supporting Wnt protein that gives rise to Act-β-Cat in a conformation able to act as a transcriptional activator and promote cell division.

There is a considerable volume of work that supports the notion that Wnt signalling regulates the expansion of specified myogenic cells (Munsterberg et al., 1995), which we now extend to satellite cells. Furthermore, a number of studies have shown that activation of the Wnt-signalling pathway leads to the transformation of nonmyogenic cells to the myogenic lineage (Petropoulous and Skerjanc, 2002; Polesskaya et al., 2003). However, recently, a study proposed that Wnt signalling, specifically through the action of Act-β-Cat, results in a fibrogenic conversion of muscle cells (Brack et al., 2007). In our experiments, the progeny of satellite cells in all the different conditions maintained their myogenic identity. Cells in clusters were identified and counted by staining with DAPI. The myogenic nature of all the constituents of each cluster was determined with a panel of antibodies that recognise skeletal-muscle-associated proteins (e.g. Pax7, MyoD, myogenin). Every cell within a cluster was positive for at least one of the markers. Therefore, we can state with some certainty that exposure of satellite cells to different Wnt proteins did not cause a detectable level of fibrogenic conversion.

A strong case has been made that all regeneration is due to these resident stem cells, based on the observation that genetically marked muscle fibres containing a few satellite cells gave rise to all the myonuclei in a regenerated muscle (Collins et al., 2005). We provide evidence that the programme of satellite-cell proliferation and differentiation, in the context of Act-β-Cat expression on isolated fibres, is recapitulated during chronic and acute in vivo muscle regeneration. In cases of muscle regeneration following muscle injury through the injection of cardiotoxin, and during human muscle regeneration in muscular dystrophy, we observed the same relationship between Act-β-Cat expression and tissue repair. During phases of repair, we found high levels of nuclear-associated Act-β-Cat expression as well as high levels of Pax7 and Ki67 expression, indicating that the cells are capable of undergoing division. After regeneration, the levels and distribution of Act-β-Cat changed considerably. There was a significant decrease in the level of Act-β-Cat but more importantly, when detected, it was located outside the nucleus, usually associated with the cell membrane. Furthermore, there was a decrease in Ki67 expression. Taken together, these results suggest that nuclear localisation of Act-β-Cat results in the proliferation of satellite cells and that its distribution outside of the nucleus is indicative of cells that are mitotically inactive. This work is in agreement with the finding that the forced overexpression of β-catenin during ischaemia-induced muscle damage leads to increased myoblast proliferation and enhanced muscle repair (Kim et al., 2006).

Finally, we propose an explanation of how Wnt signalling controls the proliferation of satellite cells. The rationale of the explanation is built around a model proposed by Olguin et al. (Olguin et al., 2007), who suggest that the ratios of Pax7 and MyoD activity are important in controlling the myogenic fate of satellite-cell progeny. They suggest that a high ratio of Pax7 to MyoD (as found in quiescent satellite cells) render satellite cells mitotically inactive. A cell at an intermediate ratio of Pax7 to MyoD is able to proliferate, but cannot differentiate. However, a cell with a low Pax7 to MyoD ratio can differentiate, because it is able to induce myogenin, which further decreases the ratio since myogenin represses Pax7 (Olguin et al., 2007). We suggest that Wnt proteins that promote the formation of Act-β-Cat and its translocation to the nucleus result in an increase in the expression of Pax7, which has been documented to stimulate satellite-cell proliferation (Petropoulus and Skerjanc, 2002). This probably occurs in concert with MyoD, because Pax7 induces the expression of MyoD (Relaix et al., 2005). However, the ratio of Pax7 to MyoD is maintained at the intermediate level, but in the case of the Wnt-stimulated cells, there are increased levels of the proteins that enhance the rate of proliferation. Furthermore, we suggest that nuclear translocation of Act-β-Cat is required to promote proliferation, because Wnt proteins that cause it to be excluded from the nucleus inhibit proliferation. As long as satellite cells are in a proliferation-enhancing environment, for example in the presence of Wnt1, Wnt3a or Wnt5a, they continue to proliferate. However, the signal needs to be dampened or antagonised in order for differentiation to take place. This could be achieved through the expression of sFRPs (small Frizzled-related proteins), which are a family of potent Wnt inhibitors that are expressed in adult skeletal muscle (Polesskaya et al., 2003).

However, this model makes a clear distinction between proliferation and activation. We have shown that following single fibre isolation, MyoD is expressed prior to the appearance of Act-β-Cat. This suggests that the process of satellite-cell activation, that is, the initiation of new gene expression in a satellite cell, is independent of Act-β-Cat activity.

Eight-week-old C57/Bl6 mice were sacrificed through Schedule 1 killing and EDL muscle carefully dissected without damaging the muscle fibres or proximal and distal tendons. Individual fibres from the EDL were dissociated with 0.1% (w/v) type 1 collagenase (Sigma) in Dulbecco's modified Eagle medium (DMEM) (Gibco) containing Glutamax and 100 μg/ml penicillin/streptomycin (Sigma) for 2 hours at 37°C. Using graded glass pipettes, single fibres were liberated from the muscle into fresh DMEM and either fixed in 4% paraformaldehyde in PBS for 10 minutes, or cultured.

Fibres were cultured in suspension in 24-well plates (NUNC) with 2-4 fibres in 0.5 ml single fibre culture medium (SFCM DMEM containing Glutamax, 100 μg/ml penicillin/streptomycin, 10% (v/v) horse serum and 0.5% (v/v) chick embryo extract) per well at 37°C in 5% CO₂ for either 2, 6, 24, 36, 48 or 72 hours. Myofibres were fixed in 4% PFA for 10 minutes. BrdU was administered at a concentration of 10 μM to myofibre SFCM at the start of 36 hour cultures or at 48 hours during 72 hour cultures.

Rat osteogenic sarcoma cells (ROS17/2.8) were plated out in 48-well plates and transfected with Super8XTOPFlash (courtesy of Randall Moon, University of Washington, Seattle, WA) and pRL-CMV Renilla luciferase vectors. To prepare vectors for transfection, 1 mg/ml lipofectin reagent (Gibco) was mixed in serum-free DMEM and incubated at room temperature (RT) for 30 minutes. 0.5 μg Super8XTOPFlash and 0.05 μg pRL-CMV Renilla luciferase DNA (per well) was diluted in serum-free DMEM, combined with the diluted lipofectin reagent and the mixture incubated at RT for 20 minutes. The growth medium was removed from ROS cells and replaced with 100 μl serum-free medium containing DNA per well. ROS cells were incubated at 37°C in 5% CO₂ for 6 hours to allow transfection to take place. Following transfection, cells were reincubated for 18 hours in normal growth medium (DMEM containing Glutamax, 10% foetal calf serum, 100 μg/ml penicillin/streptomycin and 1 mM sodium pyruvate) at 37°C in 5% CO₂. Transfected ROS cells were cultured for a further 30 hours under the same conditions with and without the addition of approximately 100 freshly isolated single muscle fibres (extracted as above) per well. 10 mM LiCl was used as a positive control. Cells were lysed in 1× passive lysis buffer (PLB; Promega) and lysates were analysed for both constitutively active Renilla and inducible Super8XTOPFlash luciferase activities using a BIO-TEK FL600 plate reader and KC4 software. Relative Super8XTOPFlash luciferase activity was calculated.

Murine NIH-3T3 Wnt1, Wnt3a, Wnt4, Wnt5a and Lac-Z (control) expressing cells lines (gifts from A. Kispert) as described (Kispert et al., 1998) and Murine NIH-3T3 Wnt6-expressing cells (gift from Seppo Vanio, University of Oulu, Oulu, Finland), were selected in 0.025% G418 and grown in DMEM containing 10% foetal calf serum, 100 μg/ml penicillin/streptomycin and 1 mM sodium pyruvate at 37°C in 5% CO₂.

Murine NIH-3T3 Wnt1, Wnt3a, Wnt4, Wnt5a, Wnt6 and Lac-Z-transfected control cells were grown in 24-well plates to 80% confluency. Cells were transferred to SFCM and 5-10 myofibres were added per well and cocultured for 2, 36 or 72 hours. Myofibres cultured for 72 hours were transferred to different culture conditions after 36 hours. Myofibres cultured for 2 hours with cells expressing Wnt1, Wnt6 and LacZ were administered with 10 μM BrdU. Single fibres were also cultured with 25 μM epigallocatechin-3-gallate (EGCG, Sigma) for 72 hours with control fibres cultured in normal SFCM. Cultures were fixed in 4% PFA/PBS for 10 minutes and fibres were processed for immunostaining.

Fixed myofibres or tissue cryosections were permeabilised in a solution of 20 mM HEPES, 300 mM sucrose, 50 mM NaCl, 3 mM MgCl₂ and 0.5% Triton X-100 (pH 7) at 4°C for 15 minutes. Fibres were washed in PBS and nonspecific binding was blocked using wash buffer (5% v/v FCS in PBS with 0.05% v/v Triton X-100) for 30 minutes. Primary antibodies used were monoclonal mouse anti-Pax7 IgG (Developmental Studies Hybridoma Bank, 1:4), monoclonal mouse anti-Pan-β-catenin IgG (Santa Cruz Biotechnology SC7963, 1:200), monoclonal mouse anti-activated-β-catenin IgG (Upstate 05665, 1:200), polyclonal rabbit anti-MyoD IgG (Santa Cruz Biotechnology M-318, 1:200), polyclonal rabbit anti-myogenin IgG (Santa Cruz Biotechnology M-225, 1:200), polyclonal rabbit anti-Ki67 (Novo castra-NCL Ki67p, 1:200), monoclonal rat anti-BrdU (Abcam ab6326-250, 1:100), polyclonal rabbit anti-Laminin IgG (Sigma L-9393, 1:200) and mouse anti-Caveolin3 IgG (BD Biosciences 610420, 1:200). Mouse primary antibodies were visualised using Alexa Fluor 488 fluorochrome-conjugated goat anti-mouse IgG (Molecular Probes A11029, 1:200) and rabbit primary antibodies were visualised using monoclonal swine anti-rabbit biotinylated IgG (DAKO E0353, 1:200) followed by streptavidin-conjugated Alexa Fluor 594 (Molecular Probes S11227, 1:200). All antibodies were diluted in wash buffer; samples were incubated in primary antibodies for 18 hours at 4°C, the anti-rabbit secondary antibody was incubated for 1 hour at room temperature and all fluorescent antibody labelling was carried out at room temperature for 45 minutes. Myofibres and tissue sections were mounted using fluorescent mounting medium (DakoCytomation) containing 2.5 μg/ml DAPI for nuclear visualisation. Subsequent image analysis and manual counting of labelled cells was done using either a Leica DM4000B fluorescent microscope and DC500 camera system or a Leica TC5 SP confocal microscope system.

Total RNA was extracted from myofibres harbouring activated satellite cells after 48 hours in culture. RT-PCR was performed for analysis of Wnt1, Wnt3a, Wnt5a and Wnt11 gene expression. The following primers were used: Wnt1, F:5′-GCTGCTGCCCAGCTGGGTTTCTACTAC, R:3′-TAGCTTTCCGTGCCCTTTCAACTCG. Wnt3a, F:5′-GCGGAGATCCTACCTGTGAG, R:3′-TCTGCCAAGGAGACTAGGAAAGCC. Wnt5a, F:5′-GAAGAAGCCCATTGGAATATTAAGC, R:3′-TTAGCGTGGATTCGTTCCCTTTCTC. Wnt11, F:5′-CCTCCCCGGCCCGACCCCTCCTTTGTAATTTGAATAAAAC, R:3′-TTTATTGGCTTGGGATCCTG.

The skin overlying the tibialis anterior (TA) muscle of the right lower hindlimb of 8-week-old C57Bl/ScSn mice was incised and 30 μl of 10 μM cardiotoxin (Latoxan, Rosans, France) injected through the investing fascia into the right TA muscle using a 50 unit insulin syringe. One single injection was undertaken per muscle to prevent leakage of the cardiotoxin from the muscle. The skin overlying the TA was closed by suture. As a control, the contralateral left TA muscle of each mouse was injected with 30 μl sterile phosphate buffered saline (PBS). At a variable number of days following cardiotoxin injection, TA muscles were dissected from the hind limbs of the mice and snap-frozen for cryosectioning.

The patients analysed in this paper had been referred to the Dubowitz Neuromuscular Unit (Hammersmith Hospital London, UK) for diagnostic purposes. Needle muscle biopsies were obtained with informed consent from the quadriceps and rapidly frozen in isopentane cooled in liquid nitrogen according to standard techniques. After sectioning, the muscle biopsies were processed for routine histological and histochemical stains. Patient 1 was 3 years old and was affected with a limb girdle muscular dystrophy with reduced expression of glycosylated dystroglycan (`dystroglycanopathy'). Mutations in the genes known to cause a dystroglycanopathy were ruled out in this patient (Godfrey et al., 2007), The muscle biopsy showed features compatible with muscular dystrophy, with increased variability in fibre size, increase in fat and connective tissue and internal nuclei, and features of regeneration and degeneration. Regenerated fibres were basophilic upon haemotoxylin and eosin staining and also had increased expression of neonatal and foetal myosin (data not shown). Muscle of patient 2 came from a 1-week-old child who was eventually deemed free of any neuromuscular condition and was therefore classified as normal.

Data were processed for statistical significance using independent sample t-tests at the 95% confidence interval. If data were significant at a higher interval then the P value is stated. All data are presented as mean ± s.e.m.

We thank the anonymous reviewers for insightful comments that have much improved the manuscript. Also, we would like to thank Helen Smith and Stephen Poutney for excellent assistance with confocal imaging. This work was funded by BBSRC for A.O. and G.L., AFM and MDC for D.L.-W. and Wellcome Trust (077750/205/Z) for P.V.