The Skeletal Muscle Satellite Cell: The Stem Cell That Came in From the Cold

Historical Perspective

The year 1961 was momentous for skeletal muscle, being marked by publications that established the dominant model of the cell biology of skeletal muscle for the remainder of the 20th century. The first was the demonstration that the multinucleated skeletal myofiber, the contractile unit of muscle, is formed by the fusion of large numbers of mononucleate myoblasts (Cooper and Konigsberg 1961; Stockdale and Holtzer 1961). This effectively resolved the problem first noted by Lewis and Lewis (1917) that myofibers seemed to increase in size and in content of nuclei in the absence of any observable nuclear division within the myofiber. The complementary discovery consisted of electron microscopic descriptions of an apparently quiescent cell lying on the surface of the myofiber, but beneath its basement membrane, where its peripheral position earned it the name ‘satellite’ cell (Katz 1961; Mauro 1961). Although first described in anuran amphibians, the satellite cell occupies an identical anatomical position in the majority of vertebrates and acquired immediate and almost unquestioned candidacy as the source of myogenic cells for growth and repair of postnatal skeletal muscle.

In general, stem/progenitor cells have been identified and characterized in terms of molecular markers, which have then been used to trace them to their anatomical niche within a tissue. In the case of the satellite cell, attribution of a stem cell-like status to an anatomically defined entity made it difficult to devise stringent tests because its role in regeneration usually moves it out of its position immediately beneath the basal lamina. Thus, the principal defining characteristics of a satellite cell are removed, destroying any formal connection between it and the myoblasts that eventually differentiate into newly formed muscle. Evidence that satellite cells function as myogenic precursors was initially based on studies of the distribution of labeled thymidine in growing or regenerating muscles (reviewed in Grounds and Yablonka-Reuveni 1993), which collectively led to the commonly accepted view that satellite cells divide to provide myonuclei to growing myofibers (Moss and Leblond 1971) before becoming mitotically quiescent in normal mature muscle (Schultz et al. 1978). Conclusive proof that satellite cells directly give rise to myoblasts was shown only with isolated myofibers, a technique pioneered by Bischoff (1975,1986) and Konigsberg et al. (1975), among others (Yablonka-Reuveni and Rivera 1994; Rosenblatt et al. 1995; Beauchamp et al. 2000). In these studies, viable myofibers were isolated from the muscle by enzymatic digestion, complete with their cohort of satellite cells still resident beneath the basal lamina. When cultured, the satellite cells proliferated, giving rise to satellite cell-derived myoblasts that differentiated to produce multi-nucleated myotubes. Recently, transplantation of such single myofibers into muscle has provided good evidence that the satellite cell indeed acts as a myogenic stem cell in vivo, able to give rise to both new myofibers and, importantly, also many new satellite cells (Collins et al. 2005). The satellite cell therefore fulfills the basic definition of a stem cell, in that it can give rise to a differentiated cell type and maintain itself by self-renewal.

What Is the Developmental Origin of Satellite Cells?

The prevailing hypothesis is that satellite cells in mammals and birds are derived from somites (Armand et al. 1983), transitory mesoderm-derived structures formed in pairs on either side of the neural tube. Somites then differentiate into dermomyotome and sclerotome, and mesodermal cells become specified as skeletal muscle precursors within the nascent myotome (Christ et al. 1975; Braun et al. 1992; Rudnicki et al. 1993). Myogenic precursors continue to be generated within the somite (Ben-Yair et al. 2003) and eventually give rise to epaxial muscles (deep back muscles), whereas others migrate from the myotomal lips to populate the muscles of the limb and diaphragm fields.

Myoblasts isolated from different developmental stages have distinct characteristics (e.g., Bonner and Hauschka 1974) and can be distinguished from those isolated from peri- and postnatal muscle (reviewed in Cossu and Molinaro 1987 and Yablonka-Reuveni 1995). Satellite cells can first be identified morphologically by their appearance underneath the forming basement membrane toward the end of the fetal period (Ontell and Kozeka 1984). Concurrence of the time when adulttype myoblasts can first be identified in later stages of embryogenesis and the time when satellite cells can be identified morphologically in situ has supported the notion that satellite cells represent a unique subset of myogenic progenitors (Hartley et al. 1991; Feldman et al. 1993). This may be further reflected by the varying roles of the myogenic regulatory factors (MRFs), a family of muscle-specific helix-loop-helix transcription factors comprising Myf5, MyoD, Mrf4, and myogenin. Mrf4 acts only as a myogenic determination factor in embryonic cells (Kassar-Duchossoy et al. 2004), whereas myogenin is required only for differentiation before birth (Knapp et al. 2006). Furthermore, the seminal observation that embryonic and fetal myogenesis is largely unperturbed in mice lacking the paired box transcription factor Pax7, although postnatal growth is severely affected, points to a unique requirement of satellite cells for this transcription factor (Seale et al. 2000). Proposed myogenic stem cells within the mouse somite express Pax3 and Pax7 but no myogenic-specific markers. It has been suggested that this same cell population later generates satellite cells, certainly during the postnatal period (Kassar-Duchossoy et al. 2005; Relaix et al. 2005) with similar observations made in chicken (Gros et al. 2005). These studies were restricted to examining late fetal and/or early postnatal stages for the presence of marked satellite cells. Therefore, cells identified with genetic markers at late fetal or early postnatal period may simply be required for postnatal growth and never actually become quiescent satellite cells in adult. The elegant work of Schienda et al. (2006) extends the examination period to 4-6 weeks and shows that many satellite cells in adult muscle are derived from the hypaxial somites, augmenting earlier studies.

Not all satellite cells are of somitic origin. Some head muscles are unique in that they do not derive from somites but rather from prechordal mesoderm and have a distinct genetic network controlling their formation (Tajbakhsh et al. 1997; Lu et al. 2002; Tzahor et al. 2003). Interestingly, mutant mice lacking Pax3 or c-Met (the hepatocyte growth factor receptor) do not have cells that migrate from somites to populate the limb and give rise to muscle. Rare myogenic cells, however, are still found in the limbs of embryos of these knockout mice, which share markers in common with hematopoetic and endothelial cells such as CD34 and flk (De Angelis et al. 1999), and CD34 is certainly expressed by adult satellite cells (Beauchamp et al. 2000). Furthermore, endothelial cells in skeletal muscle micro-vasculature contain Sca1 (Zammit and Beauchamp 2001), and rare satellite cells express eGFP in a Sca1 transgenic mouse (Mitchell et al. 2005). This raises the possibility of a common origin for endothelial cells and some satellite cells (Kardon et al. 2002), or that endothelial cells may be able to give rise to satellite cells. Intriguingly, mesangioblast cell lines derived from fetal vessels (Minasi et al. 2002; Sampaolesi et al. 2003) or cells from neonatal bone marrow and fetal liver (Fukada et al. 2002) are able to generate cells that can occupy the satellite cell niche following transplantation into adult muscle. Whether these tissues are normal sources of satellite cells during development, however, remains speculative at best. Indeed, rare myogenic cells can be found in a number of organs not sharing a common embryological origin including the brain, lungs, kidneys, and intestines during chicken fetal development (Gerhart et al. 2001), which may simply reflect aberrant migration or differentiation of pluripotent cells.

Molecular Markers of Satellite Cells

The anatomic definition of a satellite cell meant that their characterization initially relied on ultramicroscopic criteria and so, by definition, all cells located beneath the basal lamina of a myofiber are satellite cells, regardless of their function or gene expression profile. The relatively recent advent of molecular markers has allowed the reliable identification of satellite cells at the light microscope level (Figure 1 and Figure 2). The 3F-nlacZ-E transgene (Kelly et al. 1995) marks all myonuclei in fast myofibers with β-galactosidase; therefore, detecting nuclei lacking reporter gene activity with 4,6-diamidino-2-phenylindole provides an effective means of identifying the total population of satellite cells on isolated myofibers (Beauchamp et al. 2000). In normal mature muscle, satellite cells are mitotically quiescent (Schultz et al. 1978) and express Pax7 (Seale et al. 2000), the adhesion molecule M-cadherin (Irintchev et al. 1994), and saliomucin CD34 (Beauchamp et al. 2000), among others (reviewed in Charge and Rudnicki 2004) (Figure 1 and Figure 2). In addition, the Myf5 locus is clearly active in quiescent satellite cells as shown using the Myf5 ^nlacZ/+ mouse (Tajbakhsh et al. 1997; Beauchamp et al. 2000; Shefer et al. 2006) and several transgenic mouse lines (Zammit et al. 2004a), but whereas wild-type Myf5 mRNA can be detected in quiescent satellite cells (Day K and Yablonka-Reuveni Z, unpublished data) there is, as yet, no definitive evidence that the protein itself is present. The Myf5 ^nlacZ/+ mouse also provides a useful marker for the contribution of donor cells to the satellite cell pool following transplantation (Heslop et al. 2001; Collins et al. 2005). Recent markers that can be added to the list of those used to identify quiescent satellite cells include lysenin, which reports sphingomyelin levels in the cell membrane (Nagata et al. 2006) and caveolin 1 (Volonte et al. 2005).

Pax7 is probably the most useful current marker for identifying quiescent satellite cells due to the availability of a good antibody (Seale et al. 2000; Halevy et al. 2004; Zammit et al. 2004b; Shefer et al. 2006) (Figure 1 and Figure 2). It should be noted that markers such as CD34 are not specific to satellite cells but are useful markers on isolated myofibers because distinguishing satellite cells from other positive cells (e.g., endothelial cells) on muscle sections is more difficult. Such analysis often requires not only the satellite cell marker but also coimmunostaining to identify the basal lamina (see Figure 2).

To fulfill their role in muscle maintenance, hypertrophy, and repair, satellite cells must first be activated from this quiescent state to produce myoblast progeny (reviewed in Charge and Rudnicki 2004 and Wozniak et al. 2005). Satellite cell-derived myoblasts are generally characterized by the same set of myogenic markers as myoblasts derived from almost any developmental stage (Figure 3). When satellite cells are activated, they rapidly initiate MyoD expression (Fuchtbauer and Westphal 1992; Grounds et al. 1992; Yablonka-Reuveni and Rivera 1994) (Figure 2E) and undergo a CD34 isoform switch, continuing to coexpress Pax7, M-cadherin, and Myf5. They then begin to divide, expressing additional genes typical of cycling cells such as PCNA. Myogenin then marks the onset of myogenic differentiation (Fuchtbauer and Westphal 1992; Grounds et al. 1992; Yablonka-Reuveni and Rivera 1994; Yablonka-Reuveni et al. 1999b; Zammit et al. 2004b) together with a variety of regulatory and structural muscle genes typical of skeletal muscle myocytes. The pattern of Mrf4 expression during satellite cell myogenesis is less clear because it can be expressed either after, or prior to, initiation of myogenin expression (Smith et al. 1993,1994). A schematic of the dynamics of satellite cell markers as they transit from quiescence to differentiation is shown in Figure 3.

OPEN IN VIEWER

Figure 1

Distribution of muscle satellite cells on isolated myofibers. Myofibers were isolated from the extensor digitorum longus (EDL), soleus, and tibialis anterior (TA) of an adult mouse and immunostained for M-cadherin to locate satellite cells (A). Satellite cells appear distributed randomly along the entire length of TA and EDL myofibers, but soleus myofibers usually have an accumulation of satellite cells around the motor end plate where the motoneuron contacts the myofiber (arrow in A). An isolated soleus myofiber, coimmunostained with Pax7 (green, B) to identity satellite cells (arrows in B,C) and α-bungarotoxin (red, B) to locate the neuromuscular junction (arrowhead in B) clearly shows this accumulation of satellite cells around the motor end plate. All nuclei present were identified using 4,6-diamidino-2-phenylindole (DAPI) counterstaining (C). Bar: A = 200 μm; B,C = 14 μm. (A) Kindly provided by J. David Rosenblatt.

Are Satellite Cells a Heterogeneous Population?

The use of molecular markers has indicated that there may be heterogeneity within the satellite cell pool of young mice (Beauchamp et al. 2000). This could simply be due to dynamic expression of some of the antigens expressed by satellite cells in vivo and could be related, for example, to the length of time that the cells have been quiescent. That the satellite cell pool may be composed of a heterogeneous population is certainly suggested by various functional observations. During postnatal growth, cells in the satellite cell position in muscle can be separated into two typified categories according to their rate of cell division (Schultz 1996), although myogenic lineage markers were not used to confirm their identity. After muscle damage in adult, some myoblasts express myogenin within 8 hr and so presumably commit to differentiation with little or no proliferation, whereas most myoblasts do not even divide much before 24 hr (Rantanen et al. 1995). Differences between myogenic progenitors are also seen in culture where cells exhibit heterogeneity in proliferation rate and clonogenic capacity (Molnar et al. 1996). Finally, irradiation prevents muscle growth and maintenance due to the ablation of (most) satellite cells (Wakeford et al. 1991; LaBarge and Blau 2002), but a subpopulation of myogenic precursor cells survive and can still be recruited to regenerate muscle following a substantial injury (Heslop et al. 2000). At present there is no direct evidence linking behavioral and phenotypic heterogeneity, and it may simply reflect the state of a satellite cell in a dynamic system.

OPEN IN VIEWER

Figure 2

Identification of muscle satellite cells using molecular markers. There is an ever-increasing number of molecular markers for identifying quiescent satellite cells, but Pax7 is probably the most useful at present. An isolated EDL myofiber coimmunostained for Pax7 (green, A) and CD34 (red, B) shows a quiescent satellite cell on the surface of a myofiber (arrow in A-C). The 3F-nlacZ-E transgene marks all myonuclei in fast muscles and so provides an easy method to identify the whole satellite cell population on a myofiber, independently of antigen expression (Beauchamp et al. 2000). The Myf5 ^nlacZ/+ mouse expresses the Myf5/β-gal fusion protein in quiescent satellite cells and so makes their recognition straightforward (Beauchamp et al. 2000). Detection of myonuclei using X-gal from the 3F-nlacZ-E transgene (blue, D) and a satellite cell using salmon-gal from the Myf5 targeted locus (arrow in D) on an isolated EDL myofiber from a double Myf5 ^{nlacZ/3F-nlacZ-E} mouse (D). When satellite cells are activated (arrow in E), they rapidly coexpress MyoD (green, E) with CD34 (red, E). Pax7 can also be used to identify satellite cells on muscle sections, although this can be problematic on mouse tissue. Satellite cells (arrows in F,G) in a 49-day-old chicken pectoralis muscle section immunostained for Pax7 (green, F) with laminin (red, F) used to mark the basal laminae of muscle fibers. All nuclei present were identified by counterstaining with DAPI (blue, G). (F,G) Kindly provided by Benjamin Rosser and Mohammed Allouh; for further details see Halevy et al. 2004.

OPEN IN VIEWER

Figure 3

Schematic of satellite cell myogenesis and markers typical of each stage. Satellite cells are quiescent in normal adult muscle and can be activated by, for example, muscle damage. Once activated, satellite cells divide to produce satellite cell-derived myoblasts that further proliferate, before committing to differentiation and fusing to form myotubes, which then mature into myofibers (for clarity, satellite cell self-renewal is not included). CD34, Pax7, and Myf5/β-gal are expressed in quiescent satellite cells. Satellite cell activation is marked by the rapid onset of MyoD expression, whereas myogenin later marks the commitment to differentiation. The temporal expression pattern of MLC3F-tg is typical of many structural muscle genes such as skeletal muscle actin and MyHC, which mark sarcomeric assembly in the later stages of differentiation. Myf5/β-gal denotes the fusion protein product of the targeted allele of the Myf5 ^nlacZ/+ mouse (Tajbakhsh et al. 1997), whereas MLC3F-tg is the product of the 3F-nlacZ-E transgene (Kelly et al. 1995). (Adapted from Miller et al. 1999 with modifications by J. Beauchamp and the authors.)

The presence of different muscle fiber types classified by the myosin heavy chain (MyHC) isoform they contain provides a marker of muscle fiber heterogeneity that appears to be reflected in some populations of satellite cells. For example, the presence of a cat jaw muscle-specific superfast MyHC isoform in regenerated limb muscle is observed only following transplantation of jaw muscle-derived myogenic precursors (Hoh et al. 1988; Hoh and Hughes 1991). Similarly, the MyHC isoform that rodent satellite cell-derived myotubes express is correlated to the phenotype of the myofiber from which they originate (Dusterhoft and Pette 1993; Rosenblatt et al. 1996; Kalhovde et al. 2005). This does not appear to be the case in man (Bonavaud et al. 2001), but it is possible that the phenomenon is seen only when culturing myogenic cells on specialized matrixes that permit long-term cultures such as Matrigel (Hartley and Yablonka-Reuveni 1990). This predisposition of different satellite cell populations toward expression of specific MyHC isoforms is manifested only after differentiation, and no marker exists to distinguish them beforehand.

Heterogeneity of satellite cells between different muscles is clearer. With head muscle, for example, the masseter regenerates poorly compared with limb muscle (Pavlath et al. 1998), but the extent to which this is dictated by the environment of masseter tissue has not been fully evaluated. This is reflected, however, in poor proliferation and differentiation of masseter-derived myogenic cells in culture. Satellite cells in the extraocular muscles remain proliferative and add myonuclei to the uninjured myofiber when their limb counterparts are quiescent (McLoon et al. 2004), and these muscles are spared in Duchenne muscular dystrophy (Porter et al. 2003). These phenomena may be related to specific environmental cues, but the myogenic progenitors themselves may also be different, reflecting their different ontogeny from body muscles. Heterogeneity exists, however, even among somite-derived muscles. Mutant mouse lines in which the Pax3 locus has been targeted with eGFP provide the best current evidence of heterogeneity between muscles where, for example, only the gracilus in the hindlimb contains significant numbers of eGFP (Pax3)-expressing satellite cells. In the upper body, many muscles contain large numbers of satellite cells expressing eGFP (Pax3), the vast majority of which also express Pax7 (Relaix et al. 2006). There does not seem to be any obvious reason why in the Pax3 ^eGFP/+ mouse certain muscles contain eGFP+ve satellite cells, whereas neighboring ones do not. It does not appear related to the ontogeny of the muscle or to the muscle fiber type composition; therefore, the reason for this subpopulation remains a mystery. Although there is controversy as to whether this targeted allele faithfully reflects the expression of endogenous Pax3 in adult tissue (Horst et al. 2006), it certainly demonstrates heterogeneity among satellite cells. When transplanted, cells from Pax3 ^eGFP/+ mice maintain eGFP expression in a host muscle environment that does not normally contain such cells, indicating that this is probably lineage based (Montarras et al. 2005).

Whether this satellite cell heterogeneity is linked to multipotency is also unknown. Until recently it was considered that satellite cells were unipotent and that their function was restricted to supplying myoblasts for muscle maintenance and repair. Although cells isolated from muscle tissue are able to differentiate into both myogenic and neurogenic lineages (Alessandri et al. 2004) or adopt myogenic and endothelial fates (Tamaki et al. 2002), the relationship of these cells to satellite cells remains to be established. It has recently been shown that satellite cells themselves are also able to differentiate into both osteogenic and adipogenic (but not hematopoietic) lineages under standard cell culture conditions (Asakura et al. 2001; Shefer et al. 2004) or following exposure to osteogenic-(BMP2) or adipogenic-inducing factors (Asakura et al. 2001; Wada et al. 2002). It is still unclear if these alternative satellite cell fates represent trans-differentiation or amplification of distinct subpopulations within the satellite cell niche. The observations of Hashimoto et al. (2004) that single satellite cells can give rise to anatomically different progenitor types cells—myogenic “round cells” and “thick cells” that can give rise to myogenic and osteogenic cells—seems to support the notion of separate subpopulations within satellite cells. Moreover, in salamander, satellite cells (Pax7+ve) appear to contribute multipotential cells both in culture and during limb regeneration in vivo (Morrison et al. 2006). Whether mammalian satellite cells are also able to show alternative differentiation programs in vivo is unknown, but the ectopic muscle ossification that occurs in fibrodysplasia ossificans progressiva is thought to be the result of overexpression of BMP4 (Shafritz et al. 1996), which may act on satellite cells in a similar manner to BMP2 in culture. It is attractive to speculate that satellite-like cells unable to turn on the MyoD gene may retain a greater capacity to enter other mesenchymal lineages (discussed in Shefer et al. 2004). In support of this notion is the finding that myofibroblasts can express MyoD but are unable to fully execute the skeletal muscle differentiation program (Walker et al. 2001). Similarly, human rhabdomyosarcoma cells also express MyoD but do not differentiate into skeletal muscle because the factor is unable to activate transcription (Tapscott et al. 1993). Restoration of MyoD transcriptional activity, or the introduction of active Mrf4, promotes differentiation into skeletal muscle cells (Tapscott et al. 1993; Sirri et al. 2003).

Satellite Cells Remain Viable Throughout Life

The early experiments of Studitsky (1964) where muscles that were removed, minced, and replaced were still able to form a new functional muscle really demonstrate the remarkable ability of this tissue to regenerate. Effective regenerative ability is still retained after repeated cycles (up to 50) of extensive injury using myotoxins (Sadeh et al. 1985; Luz et al. 2002). Even in the absence of such overt damage, estimates of myonuclear turnover in rodents of 1-2% per week (Schmalbruch and Lewis 2000) combined with observations that most satellite cells only undergo one or two divisions in vivo before differentiation (Grounds and McGeachie 1987) mean that the pool of satellite cells would soon be exhausted in adult muscle without replenishment.

Aging is associated with sarcopenia, a significant decline in the mass, strength, and endurance of skeletal muscles in both human and animal models (Karakelides and Nair 2005). It has been proposed that compromised satellite cell function contributes to this age-linked muscle deterioration. Whether satellite cell numbers decline with age is controversial and probably varies among different muscles and species (e.g., Schafer et al. 2005; Shefer et al. 2006). Similarly, the amount of proliferation that they can undergo has also been proposed to vary with age (reviewed in Mouly et al. 2005). The age-linked decline in the rapidity at which satellite cells can be recruited into myogenesis clearly has an impact on satellite cell performance (Conboy et al. 2003) but may be due more to a delay of entry into proliferation and not necessarily an inherent declining capacity to differentiate (Shefer et al. 2006).

That said, exposure of myogenic progenitors from old muscle to a young environment rejuvenates the capacity of progenitors to contribute to repair (Carlson and Faulkner 1989; Carlson et al. 2001; Conboy et al. 2005) indicating that (some?) satellite cells do not change significantly with age. A reduction in the numbers of functional satellite cells along with a declining systemic environment and intramuscular changes in innervation and vascularization together may underlie failure of old muscles to repair properly following injury and during routine muscle utilization (Grounds 1998).

Is the Satellite Cell Compartment Maintained by Self-renewal?

The classic view that satellite cells self-renew, entailing either a stochastic event or an asymmetrical cell division at some point where one daughter cell is committed to differentiation whereas the second continues to proliferate or becomes quiescent, was initially proposed by Moss and Leblond (1971) over 30 years ago. When myoblast cultures are exposed to low serum medium, most respond by differentiating. However, some stop cycling and downregulate MyoD expression, characteristic of quiescence (Kitzmann et al. 1998; Yoshida et al. 1998). These “reserve cells” re-appear after cloning, indicating that it is a cell-intrinsic property and may reflect a stage in the cell cycle that leads to quiescence rather than differentiation when challenged with mitogen withdrawal, rather than having a genetic, lineage basis (Baroffio et al. 1996; Yoshida et al. 1998).

Culturing isolated myofibers in suspension provides a model that is a hybrid between in vivo regeneration and culture of dissociated single cells. Using this system, we have recently shown that satellite cell progeny adopt divergent fates. In culture, satellite cells synchronously activate to coexpress Pax7 and MyoD before dividing. The majority then suppress Pax7 expression, maintain MyoD, and differentiate, whereas others downregulate MyoD, maintain Pax7, and eventually stop cycling, entering a state resembling quiescence but can be restimulated and will again divide and differentiate (Zammit et al. 2004b). A similar diversification in satellite cell fate, with some cells ‘opting out’ of immediate differentiation, has also been described during posthatch chicken development (Halevy et al. 2004) and in the adult and aging mouse (Shefer et al. 2006). In both primary cultures and clonal cultures prepared from chicken and mouse, myogenic cells proliferate and coexpress Pax7 and MyoD but, with time, cells expressing Pax7 but not MyoD begin to accumulate. Rare satellite cells can be found in which the progeny of a recent division have MyoD distributed asymmetrically, with one satellite cell lacking expression, possibly indicating a divergence in the fate of the two progeny (Zammit et al. 2004b). Certainly the absence of MyoD also causes satellite cells to continue to proliferate and delay differentiation (Sabourin et al. 1999; Yablonka-Reuveni et al. 1999a). The role of MyoD in control of these cell fate decisions is central to muscle regeneration, yet the mechanism remains poorly understood.

In the absence of Pax7, satellite cells in mutant mice are rapidly depleted during the early postnatal period, showing that Pax7 has a crucial role in satellite cell function (Seale et al. 2000; Oustanina et al. 2004; Kuang et al. 2006; Relaix et al. 2006). Whether the absence of Pax7 causes a failure of self-renewal, however, is debatable. Constitutively expressed Pax7 is compatible with MyoD expression, proliferation, and myogenic differentiation (Relaix et al. 2006; Zammit et al. 2006 but see Olguin and Olwin 2004). Because infection of wild-type myoblasts with a dominant negative form of Pax7 leads to massive cell death, Pax7 may have anti-apoptotic functions and therefore play a role in maintenance and survival rather than self-renewal (Relaix et al. 2006). As Pax7 also has a role in neurogenesis and is expressed in adult brain, its deletion in vivo could also have indirect effects on muscle, meaning that conditional deletion of Pax7 in adult muscle would be useful to specifically dissect its role in satellite cells. Another consideration is that Pax3 has been shown to control cell surface properties during development (Mansouri et al. 2001). Loss of another member of the paired box transcription factor family Pax6 in the developing forebrain results in altered cellular adhesion and expression of R-cadherin (Stoykova et al. 1997). Therefore, effects of Pax7 deletion may well also perturb satellite cell function by affecting cell surface and consequently cell adhesion and/or cell-cell interactions.

Another possible candidate to control satellite cells is Notch signaling, which is involved in the decision of satellite cells to stop proliferating (Conboy and Rando 2002). The inhibitor of Notch, Numb, is asymmetrically distributed after cell division in some satellite cell progeny (Conboy and Rando 2002) and appears to segregate with Pax7 (Shinin et al. 2006), indicating that it may influence cell fate with a mechanism similar to that used in Drosophila development. Certainly with age, failure of the Notch pathway appears to contribute to the general reduction in efficiency of muscle regeneration (Conboy et al. 2003).

Transplantation of cells into muscle provides a useful assay for their fate in vivo. Much as stem cell function in the hematopoetic lineage has been explored using transplantation following whole-body irradiation to destroy endogenous stem cells, a similar assay of transplantation into locally irradiated muscle has been used to analyze the potential of myogenic precursor cells (e.g., Collins et al. 2005). Based on transplantation models, it has long been known that grafted myogenic cells not only contribute myonuclei but also produce myogenic precursors that can be re-activated by muscle damage, will proliferate and differentiate ex vivo, and form new muscle after serial transplantations (Watt et al. 1982; Partridge et al. 1989; Yao and Kurachi 1993; Morgan et al. 1994; Gross and Morgan 1999; Cousins et al. 2004). More recently, it has been shown that these donor-derived myogenic precursors can occupy the satellite cell niche and remain undifferentiated, retaining the ability to proliferate in response to future stimuli (Blaveri et al. 1999; Heslop et al. 2001). In these experiments, large numbers of muscle-derived cells were grafted (2-5 × 10⁵). Therefore, the exact nature of the cells is unknown, as to whether they proliferate before entering the satellite cell niche, i.e., self-renew. To address these points, we have now grafted a single myofiber, which has the advantage of having only a small number of associated satellite cells. For example, a single EDL myofiber with a mean of approximately seven satellite cells can give rise to considerable amounts of new muscle and, importantly, many new functional satellite cells. Because the number of new myonuclei and satellite cells far exceeds the number implanted with the single myofiber, satellite self-renewal must have occurred (Collins et al. 2005). Even taking into account any mitogenic effects of an irradiated muscle host environment, this is still a significant amount of proliferation (Morgan et al. 2002). Satellite self-renewal may be limited to a subpopulation of satellite cells and not necessarily be a property of all (Beauchamp et al. 1999). This is consistent with recent observations that a limited number of satellite cells retain incorporated BrdU in adult muscle from a pulse delivered perinatally. Some of these satellite cells do not then segregate the label upon cell division. Together this was interpreted as evidence that these particular satellite cells are stem cells with non-equivalent genomic DNA strands where one template strand is being protected from DNA replication errors (Shinin et al. 2006).

Finally, the phenomenon of the revertant fiber provides further evidence that satellite cell self-renewal is part of normal muscle homeostasis. The mdx mouse is a genetic and biochemical model of Duchenne muscular dystrophy, which lacks dystrophin protein due to a non-sense point mutation (Bulfield et al. 1984). Despite this translational block, rare revertant fibers can be observed, containing truncated dystrophin species produced by exon skipping forming an in frame transcript (Hoffman et al. 1990). These revertant fibers occur as singletons in newborn mdx mice but are then found as clusters that increase in constituent fiber number with age (Yokota et al. 2006). Because each cluster is composed of myofibers with an identical pattern of exon skipping often distinct from neighboring revertant clusters (Lu et al. 2000), this progressive accumulation of myofibers in a contiguous group containing this de facto heritable marker is consistent with a clonogenic event (Yokota et al. 2006). Although it is possible that the rare genetic event that results in the revertant fiber may have occurred during development in a presatellite cell, it is likely that ‘growth’ of the cluster is a satellite cell-mediated event requiring extensive proliferation and possibly self-renewal of these exceptional satellite cells.

Is There a Resident Satellite Cell-Stem Cell in Muscle Tissue?

Satellite cell self-renewal may not be the sole mechanism used to maintain a viable regenerative compartment. It has been proposed that, once activated, most satellite cells in healthy adult muscle are committed to myogenic differentiation (e.g., Yablonka-Reuveni and Rivera 1997) and so require replenishment from a non-satellite progenitor cell. In this scenario, there may be a progenitor of satellite cells residing outside of the satellite cell niche. Such a progenitor could be a multipotent stem cell able to give rise to several differentiated cell types including satellite cells. Candidates for this resident “stem cell of satellite cells” include endothelial-associated cells (De Angelis et al. 1999), interstitial cells (Tamaki et al. 2002; Polesskaya et al. 2003; Kuang et al. 2006), and side population (SP) cells (Gussoni et al. 1999; Asakura and Rudnicki 2002).

Mesonagioblasts, stem cells derived from the developing vasculature (Minasi et al. 2002), are able to rescue α-sarcoglycan^-/- adult muscle (Galvez et al. 2006) and also occupy the satellite cell niche after transplantation (Sampaolesi et al. 2003). Primary endothelium from developing/postnatal mice, and to a lesser extent adult mice, and cell lines derived from vasculature smooth muscle can also give rise to myogenic cells (Graves and Yablonka-Reuveni 2000; Cusella De Angelis et al. 2003), leading to the suggestion that these structures may be a source of satellite cells. Consistent with this is the observation that endothelium and satellite cells have certain molecular markers in common such as CD34 (De Angelis et al. 1999; Beauchamp et al. 2000), and that myogenic and endothelial cells probably share a common embryonic precursor (Kardon et al. 2002). Indeed, multipotent CD45-ve/CD34+ve/Sca1+ve cells able to generate both myogenic and endothelial cells are located in the muscle interstitium (Tamaki et al. 2003). More recently, it has also been shown that Pax3 can identify interstitial cells with myogenic potential (Kuang et al. 2006), but their relationship to other interstitial cells remains to be established.

SP cells were first identified as bone marrow-derived multipotent hematopoetic stem cells characterized by their ability to efflux the vital DNA dye Hoechst 33,342 (Goodell et al. 1996). Cells with similar physical properties can also be isolated directly from muscle (Gussoni et al. 1999; Jackson et al. 1999). Upon transplantation, muscle SP cells are able to contribute to both the hematopoietic (though poorly) and muscle lineages, producing both dystrophin-expressing myofibers in mdx mice and cells occupying the satellite cell niche (Gussoni et al. 1999; Asakura and Rudnicki 2002). Interestingly, muscle SP cells appear to be unable to adopt a myogenic lineage fate in culture in the absence of myogenic cells (but see Schienda et al. 2006). Muscle SP cells are a heterogeneous population, so far lacking a coherent set of molecular markers presumably because the conditions used for their fluorescence-activated cell sorting (FACS) isolation (e.g., enzymatic digestion conditions used prior to FACS and/or concentration of Hoechst dye) dictate the range of cells included in the SP fraction (Montanaro et al. 2004; Rivier et al. 2004). It does appear, though, that the majority of muscle SP cells express Sca1 but not CD45 (Montanaro et al. 2004). Most muscle SP cells comprise a resident population (Rivier et al. 2004), and recent evidence points to a large proportion of them being derived from cells in the hypaxial myotome that express Pax3, the same source as satellite cells (Schienda et al. 2006). Further studies are required to establish whether SP and satellite cells develop in parallel or whether one population is actually the progeny of the other.

Other populations of cells with myogenic potential can also be isolated from muscle tissue. Both CD45+/-ve and Sca1+/-ve muscle-derived cells are able to become incorporated into myofibers and express a muscle-specific transgene (McKinney-Freeman et al. 2002), although only the CD45+ve ones have hematopoetic potential. CD45+ve/Sca1+ve cells isolated from damaged, but not uninjured, muscle are able to adopt a myogenic fate directed by Wnt signaling and via a Pax7-dependent process (Polesskaya et al. 2003; Seale et al. 2004). Finally, multipotent cell lines have been isolated from muscle tissue using preplating techniques (e.g., Lee et al. 2000).

Notably though, the cell populations discussed in this section have only been characterized following purification from muscle tissue, such as by FACS. CD45 and Sca1+ve cells are widespread in muscle. For example, Sca1 is present on cells in blood vessels, but in quiescent satellite cells >99% do not express Sca1, whereas CD45 is not expressed at all (Zammit and Beauchamp 2001; Asakura and Rudnicki 2002; Mitchell et al. 2005). Interestingly, some satellite cell progeny in culture express Sca1 (Mitchell et al. 2005), a phenomenon whose relevance to in vivo myogenesis is unknown. The relationship, if any, of these various populations of cells to each other and, crucially, to satellite cells, remains to be determined.

Is the Satellite Cell Pool Replenished From Outside Muscle Tissue?

Grounds (1983) reported that skeletal muscle precursors do not arise from transplanted bone marrow. Fifteen years later, it was shown that muscle can be formed from this source (Ferrari et al. 1998). The difference in the two studies was the assay method used. Because sensitive molecular markers of donor cells were still several years away, Grounds (1983) had to rely on testing for the presence of a donor-specific isoenzyme in muscle extracts using gel electrophoresis, requiring a substantial amount of muscle to be made. Ferrari et al. (1998) were able to use a sensitive nlacZ reporter gene, enabling individual donor-derived nuclei to be identified.

Following bone marrow transplants, it has been shown that bone marrow-derived cells can be found in association with myofibers and express satellite cell markers (LaBarge and Blau 2002; Dreyfus et al. 2004; but see Wernig et al. 2005). Interestingly, mesenchymal cells from synovial membrane/fluid are also able to occupy the satellite cell niche (De Bari et al. 2003). Occupation of the satellite cell niche, albeit inefficiently, has lead to the assumption that any host myofiber with a donor contribution is a result of incorporation of these donor-derived cells. Recently, however, Sherwood et al. (2004) have shown that only satellite cells occupying the satellite cell niche are functional as myogenic precursors, and that the niche appears unable to instruct other cell types such as bone marrow-derived cells to function in a similar manner to satellite cells (Sherwood et al. 2004). Indeed, cells isolated from a variety of tissues including thymus (Pagel et al. 2000), nerve (Courbin et al. 1989), brain (Tajbakhsh et al. 1994; Galli et al. 2000; Rietze et al. 2001), and others (comprehensively reviewed in Grounds et al. 2002) have all been shown to generate skeletal muscle in vitro and, in some cases, also in vivo. How and why this occurs, though, is not clear. Certainly, cells are able to directly fuse with myofibers without first adopting a satellite cell-like cell intermediary step (Grounds et al. 2002). Myelomonocytic precursor cells are able to directly incorporate into regenerating myofibers (Camargo et al. 2003) where the syncytial nature of the myofiber environment may re-program their nuclei, as shown for heterokaryons (Blau et al. 1985). It should be noted, however, that a large proportion of bone marrow-derived cells incorporating into myofibers do not actually activate the myogenic program (Lapidos et al. 2004; Wernig et al. 2005).

The nature of cells with myogenic potential in non-muscle tissues has remained elusive, and their biological role, if any, is unclear. Bone marrow cells can home to muscle but not to any significant degree, with the best engraftment following bone marrow transplantation in the order of 5%, but the majority of muscles exhibit much lower levels (Brazelton et al. 2003; Wernig et al. 2005). Importantly, continual recruitment from the transplanted bone marrow is not significant, because the number of myofibers with a donor contribution quickly plateau in normal muscle and do not increase significantly with time (LaBarge and Blau 2002; Wernig et al. 2005), even in mdx mice whose muscles undergo chronic cycles of myofiber degeneration and regeneration (Ferrari et al. 2001). Similarly, in man, a Duchenne muscular dystrophy patient who underwent a bone marrow transplant still had only <1% of myofibers with donor-derived nuclei 13 years later (Gussoni et al. 2002). Presumably, therefore, no significant slowing in the progression of the disease was demonstrated. Furthermore, there does not appear to be much trafficking between bone marrow and muscle SP even after overt muscle injury (McKinney-Freeman et al. 2003; Rivier et al. 2004), although such damage does appear to recruit further bone marrow-derived cells to myofibers (Wernig et al. 2005).

Conclusions

The discovery of the satellite cell in 1961 provided the obvious candidate for the source of new muscle growth and repair. It remained the uncontested myogenic progenitor of skeletal muscle until a series of articles demonstrated that cells other than satellite cells could contribute to myogenesis. This led some to question this classic view that satellite cells are the sole supply of myogenic precursors during the life span of the animal. The contribution of these non-satellite cells to muscle in any case is very low and may well be unnecessary, considering the large amount of new muscle and viable satellite cells that can be generated from the few satellite cells resident on a single transplanted myofiber (Zammit et al. 2002; Collins et al. 2005). When the endogenous satellite cell compartment is rendered incapable of proliferation by localized irradiation, muscle regeneration is effectively prevented, indicating that circulating cells with myogenic potential are not able to contribute significantly and certainly not able to demonstrably affect muscle function (Wakeford et al. 1991; Heslop et al. 2000). Although lethally irradiated mice can be rescued by transplantation with hematopoetic stem cells, these cells do not appear to be able to become functional satellite cells (Sherwood et al. 2004). Whether non-satellite cells even constitute part of a physiologically relevant system for muscle repair or merely reflect some imprecision in the mechanisms that determine the fate of precursor cells remains to be determined.

As more genetic mouse models become available for tracing or abolishing satellite cells in vivo, it might be possible to address several long-standing questions regarding fundamental aspects of satellite cell biology. For example, how frequently are satellite cells required to supply new myonuclei during normal muscle utilization (Spalding et al. 2005)? Perhaps satellite cell self-renewal and asymmetrical cell division are compromised in aged adult muscle, resulting in the observed decline in satellite cells (Shefer et al. 2006). Likewise, despite the common notion that satellite cells participate in muscle hypertrophy as a result of exercise and weight overload, understanding of this field has progressed slowly since the initial observations by Schiaffino et al. (1976). Furthermore, the satellite cell niche is thus far a term without a base - is the region where the satellite cell resides on the myofiber surface different from that surrounding it? Does the basement membrane in that area have unique features that direct the regulation of satellite cells? Does the myofiber itself exert control over the quiescent vs active state of the satellite cell? Can donor satellite cells be modified so they participate in muscle repair along with self-renewal for the purpose of combating muscle deterioration in aging, muscular dystrophies, and as a consequence of diseases including cancer and AIDS? Finally, can we better understand the role of the non-satellite cell myogenic sources discussed here and whether they reflect a normal process. Although non-satellite cell sources might not be of significance to muscle repair under normal conditions, can they be recruited for enhancing myogenic stem cell function and myofiber maintenance in disease and aging? As much as 1961 defined the parameters in which the biology of adult muscle would be explored for the rest of the 20^th century, the opening years of the new millennium are providing us with new paradigms of adult myogenesis. Will these new models redefine the mechanisms of muscle growth and regeneration, or will they send us back to the future of 1961?