A myogenic precursor cell that could contribute to regeneration in zebrafish and its similarity to the satellite cell



A. L. Siegel, Australian Regenerative Medicine Institute, Level 1, Boulevard 75, Monash University, Clayton Campus, Wellington Road, Clayton, Victoria 3800, Australia

Fax: +61 (0)3 9905 9862

Tel: +61 (0)3 9902 9638

E-mail: ashley.siegel@monash.edu


The cellular basis for mammalian muscle regeneration has been an area of intense investigation over recent decades. The consensus is that a specialized self-renewing stem cell, termed the satellite cell, plays a major role during the process of regeneration in amniotes. How broadly this mechanism is deployed within the vertebrate phylogeny remains an open question. A lack of information on the role of cells analogous to the satellite cell in other vertebrate systems is even more unexpected given the fact that satellite cells were first designated in frogs. An intriguing aspect of this debate is that a number of amphibia and many fish species exhibit epimorphic regenerative processes in specific tissues, whereby regeneration occurs by the dedifferentiation of the damaged tissue, without deploying specialized stem cell populations analogous to satellite cells. Hence, it is feasible that a cellular process completely distinct from that deployed during mammalian muscle regeneration could operate in species capable of epimorphic regeneration. In this minireview, we examine the evidence for the broad phylogenetic distribution of satellite cells. We conclude that, in the vertebrates examined so far, epimorphosis does not appear to be deployed during muscle regeneration, and that analogous cells expressing similar marker genes to satellite cells appear to be deployed during the regenerative process. However, the functional definition of these cells as self-renewing muscle stem cells remains a final hurdle to the definition of the satellite cell as a generic vertebrate cell type.


alveolar rhabdomyosarcomas


Met proto-oncogene encoding HGF receptor


external cell layer


embryonal rhabdomyosarcomas


hepatocyte growth factor


myogenic regulatory factor


basic helix-loop-helix transcription factor, myogenic regulatory factor 5


basic helix-loop-helix transcription factor, myogenic differentiation 1


paired box transcription factor 3


paired box transcription factor 7


The regeneration of skeletal muscle is one of the most astonishing regenerative capacities exhibited by adult mammals. Recently, there has been much interest in the aspects of regeneration that are conserved across vertebrates and how the variable regenerative capacity seen across Metazoa compares specifically with that evident in vertebrates. Most commonly, organisms such as the zebrafish or amphibians have been investigated for regenerative processes that are distinct from the limited regenerative capacities evident in mammals. The majority of these studies have focused on discovering pathways or cells types that could be manipulated to stimulate regeneration in a context where it does not occur. By contrast, the present minireview aims to examine skeletal muscle repair, a regenerative process, which has been well described in amniotes but far less investigated in the more basal vertebrates. Two possibilities can be envisaged for the relationship between muscle regeneration in basal vertebrates and that described in amniotes. First, stem cell-based muscle regeneration may be an evolutionarily conserved process basal in the vertebrate phylogeny or, second, the competence for muscle regeneration via specific resident muscle stem cells arises as a derivation of a new developmental process distinct from the general regenerative capacity evident in specific lower vertebrates. This review discusses recent and past data with respect to the identification and role of the satellite cell, an amniote postnatal muscle stem cell, in earlier branching vertebrates during muscle formation and repair, focusing on the zebrafish, with the aim of distinguishing between these two possible scenarios.

What is a satellite cell?

Vertebrate skeletal muscle is composed of multinucleated myofibres that form a syncytium of fused, postmitotic myocytes. During amniote postnatal growth and development, it has been shown that cells located peripheral to the growing myotube become covered by the basal lamina and either continue dividing and contributing to the peripheral pool, fuse into the differentiating myofibre, or become part of the reserve stem cell pool. These reserve cells become the skeletal muscle adult stem cells and are termed satellite cells [1]. The growth of the myofibre in size during the juvenile phase of amniotes relies on the fusion of satellite cells with the existing myofibres. Once the myofibres mature, the myonuclear number stabilizes and the satellite cells appear quiescent, with respect to both the cell cycle and activity [2]. The number and status of the satellite cells remains constant in the healthy adult mammal. In response to injury, the satellite cells become active and interact with their environment to reform innervated, vascularized contractile muscle, mirroring muscle that has been lost [3].

Despite being discovered in the frog, the role of the satellite cell in frogs and other basal vertebrates remains poorly defined. In 1961, Mauro [4] named satellite cells in reference to their position relative to the myofibre. He was able to clearly identify cells in the tibialis anticus muscle of frog, Rana temporaria, as independent units with their own unique cell membrane and characteristic relationship to the muscle fibre. Mauro [4] quickly compared what he saw with a similar cell described in the rat sartorius muscle by Palade, already making the inference that this cell was conserved across Metazoa. The first description of these cells providing insight into their function was not made in rats or frogs but, instead, in studies of the fruit bat.

Satellite cells in the fruit bat were described as small, bipolar, cells with a heterochromatic nucleus, scanty cytoplasm, and cytoplasmic processes that trailed on for many micrometres leading to a total cell length of up to 25 μm [5]. Microscopy is still the most accurate and accepted way of identifying a quiescent satellite cell, although they are regularly described by the factors that they express (Table 1) enabling the performance of functional studies and manipulations. The factors identified so far include paired box transcription factors (Pax)3/7 [1, 6, 7], basic helix-loop-helix transcription factor myogenic factor 5 (Myf5) [8], surface proteoglycans syndecan-3 and syndecan-4 [9], tyrosine receptor kinase Met proto-oncogene encoding hepatocyte growth factor (HGF) receptor (c-Met) [10], M-cadherin [11, 12], intergin-α7 [13, 14], chemokine receptor CXCR4 [15], calcitonin receptor [16], neural cell adhesion molecule [17], homeobox transcription factor Barx2 [18], and nuclear envelope proteins lamin A/C and emerin when used in conjunction with caveolin-1 [19, 20]. Most of these markers have only been identified in satellite cells of the mouse; a few have also been identified in satellite cells in humans and in satellite-like cells of the zebrafish (Table 1).

Table 1. Frequent used markers of satellite cells
Satellite cell markerQuiescent mouse satellite cellsActivated mouse satellite cellsDetection methodHuman satellite cellsZebrafish satellite-like cellsExpressed in other cells
  1. Ab, antibody; ISH, in situ hybridization; TM, transgenic mouse; TZ, transgenic zebrafish

Pax3+TM?TZNeural crest
Myf5−/++TM?TZMyogenic cells
Syndecan-3++Ab??Macrophages, leukocytes, chondrocytes
Syndecan-4++Ab??Macrophages, leukocytes
c-Met−/++Ab?ISHNeural crest, pericytes, motorneurones and others
M-cadherin++Ab+?Granular cells
Integrin α7++Ab??Muscle, neurones, vasculatuor
CXCR4++Ab??Migratory cells
Calcitonin receptor++Ab??Kidney, nervous system, osteoclasts
Neural cell adhesion molecule+Ab+?Lymphocytes, natural killer cells, dendrites, neurones, glia
Barx2++Ab??Interstitial muscle cells, myogenic cells, chondrogenic cells, and many epithelial tissues
Lamin A/C++Ab+?Developmentally regulated initially in myocyctes then later in all other nuclei except germ cells and haematopoietic cells
Emerin++Ab??Ubiquitously expressed but increased in skeletal and cardiac nuclei
Caveolin-1++Ab+?Adipocytes, neurones, fibroblasts, smooth muscle

Pax7 has become the accepted marker of satellite cells across many vertebrate species, including humans [12, 20], pig, horse, cow, sheep [21], chick [22], salamander [23, 24], frog [25], zebrafish [26, 27] and, recently, an invertebrate chordate amphioxus [28] (Table 2). Pax7 has become the canonical biomarker because Pax7 positive cells are required in the regeneration [29] and expression in all quiescent and proliferating satellite cells of the mouse [6]. Over the last 50 years, the mouse satellite cell has been established as a self-renewing, multipotent stem cell population that is necessary for the maintenance and repair of skeletal muscle, and is so defined as the adult stem cell of skeletal muscle [30]. However, in non-murine systems, the establishment of Pax7 positive cells as a homologous cell type to that of the Pax7 positive mouse satellite cell is not clear, and how widely deployed the satellite cell system is in animals remains unresolved (Table 2). Pinpointing when the satellite cell can be identified in amniotes and what features of a satellite cell are important with respect to making them a muscle stem cell are critical for determining whether the presence of Pax7 positive cells in basal vertebrates indicates that stem cell-based muscle regeneration evolved basally in the vertebrate phylogeny or even earlier in the chordate lineage.

Table 2. Comparison across species of satellite-like cells and their capacity for regeneration relative to amniote standards
SpeciesSatellite cell identification and methodCapacity for regeneration (epimorphic/tissue)Confirmation of satellite cells as the source for muscle regenerationReference
Drosophila ‘Adult muscle precursor’, asymmetric division Numb segregation, association with myofibreTissue regeneration in adultNot assessed [122]
AmphioxusMicroscopy, Pax3/7EpimorphicPH3 increase in regeneration only in Pax3/7 cells [28]
Lamprey‘Lateral cells’ Pax3/7Tissue regeneration in adultNot assessed [123]
SharkMicroscopy   [124]
FrogMicroscopy, Pax7Tissue as adult both as a larvaeIn larvae, epimorphic regeneration confirmed by reduction in Pax7 cells before amputation and reduction of muscle in the regenerate [125]
NewtPax7, morphology different with two basal laminaBothInjected Pax7 cultured progeny participate in limb regeneration Inhibition of Shh reduced the number of Pax7 cells, resulting in less epimorphic regeneration [24, 126]
AxolotlPax7, microscopyBothSeen in the regenerating tail and limb [90, 127]
LizardPax7, microscopyEpimorphic regeneration of the tailNot assessed but myoblasts noted in the regenerating tail of the gecko [128, 129]
ZebrafishPax7, microscopyBothPax7 positive cells in larval injury [26, 95]
Atlantic salmonMicroscopy, c-metAdult tissue regeneration from heart and skeletal muscle inflammation and mechanical injuryNot assessed [130-132]
CarpMicroscopy, cell culture   [133]
Rainbow troutMicroscopy, cell cultureAdult tissue regeneration from mechanical injuryNot assessed [69, 134, 135]
Electric fishPax7, microscopyBothBrdu positive Pax7 positive cells give rise to the regrown muscle, spinal cord and electrocytes [93]

Satellite cell origins: dermomyotome versus external cell layer

The first identification of mouse satellite cells occurs towards the end of fetal development, around embryonic day 16.5 [31, 32]. This identification is made using the classic definition of an adult stem cell to define the satellite cell, in that only a few satellite cells (and possibly only one) are sufficient to replenish the entire satellite cell pool and provide the host muscle with robust regenerative capacity [33, 34]. Satellite cells self-renew and divide asymmetrically [35], are multipotent [3], and exist in multiple populations with different properties [36]. Experiments in mice and chick have shown that adult muscle satellite cells are originally derived from a transient structure known as the dermomyotome, found in the developing embryo [37, 38]. The dermomyotome is formed from the dorsal portion of the amniote somite, is epithelial in nature, and functions as the source of dermal structures and skeletal musculature of the trunk and limbs [39, 40] (Fig. 1).

Figure 1.

Schematic comparing the processes of myogenesis and the formation of adult myogenic precursors in the amniote (A–D) and zebrafish (E–H). In the amniote, the dermomyotome (A, red and yellow layer) forms on the lateral surface of the somite and is responsible for early myotome development and growth. An initial differentiation event occurs when the pioneer fibres (A–C, blue fibres) are formed, which act as a scaffold for the developing myotome. Subsequently, waves of cells delaminate from the lips of the dermomyotome (A, B, red block to red fibres), migrate inwards and form the slow and fast muscle fibres of the myotome. The central region of the dermomyotome contributes to the second wave of myogenic growth, as well as serving as a source of satellite cells in the adult (B–D, yellow block to yellow fibres/cells). In the zebrafish, no classical dermomyotome is present initially. The somites form the adaxial cells, which are the first cells to differentiate in zebrafish myogenesis (e, blue fibres). The adaxial cells migrate through the developing myotome to become the slow muscle fibres at the lateral surface of the myotome (E, blue fibres). The somite then undergoes a rotational event, allowing the posterior section of the somite to migrate laterally and form the fast muscle fibres comprising the bulk of the myotome (E, F, red block to red fibres). The anterior portion of the somite migrates even further laterally, forming the external cell layer (ECL) (E, F, yellow block to yellow cells). The ECL appears to persist for much longer than the amniote dermomyotome. The ECL is assumed to contribute to larval and adult myogenesis, as well as generate satellite-like cells (G, H, yellow cells/yellow fibres). The neural tube is shown in purple; the notochord is shown in green, and presumptive satellite-like cells are shown in orange. Adapted with permission [61].

The muscle precursors that arise from the dermomyotome express the paired box transcription factors Pax3 and Pax7, as well as low levels of the basic helix-loop-helix myogenic transcription factor Myf5 [41-43]. The first events in the establishment of skeletal muscle involve the migration of postmitotic myocytes, which express Myf5 and another myogenic transcription factor basic helix-loop-helix transcription factor, myogenic differentiation 1 (MyoD), underneath the developing dermomyotome. These myocytes align along the axis of the embryo and consequently differentiate into myofibres [44, 45], with the resultant ‘pioneer’ fibres acting as a scaffold for the developing myotome [46, 47]. Subsequently, postmitotic myocytes delaminate from the dermomyotome, migrate inwards and form the muscle of the primary myotome. This process continues until the dermomyotome is completely de-epithelialized. Cells present in the epaxial part of the dermomyotome give rise to the dorsal skeletal muscle of the trunk, whereas cells from the hypaxial domain form lateral trunk limb musculature [48]. During these events, muscle progenitors intercalate into the primary myotome and consequently contribute the satellite cells found in postnatal muscle (Fig. 1). Specifically, the central region of the dermomyotome that undergoes an epithelial to mesenchymal transition last has been shown to contain two populations of myogenic progenitor cells, both of which express Pax3 and Pax7 and translocate directly into the primary myotome once the scaffold is formed [37, 39, 49] (Fig. 1).

Pax7 is expressed in mononucleated cells that either differentiate into myotubes or give rise to the satellite cell compartment [50]. Pax7 expression continues in the adult, as indicated by high levels of expression in all satellite cells. The current model therefore suggests that satellite cells are somitically derived progenitors trapped or halted in their differentiation programme, and that these cells can then later reactivate the myogenic programme. Evidence of lineage plasticity of satellite cells, as demonstrated by their ability to differentiate into multiple mesogenic tissues (myogenic, osteogenic and adipogenic), supports the theory of reactivation of the somitic differentiation programme [51]. In addition, the identification of satellite cells in a specific muscle corresponds to the timing of the generation of the myoblasts for that particular muscle, which will subsequently fuse to form the muscle fibres of the back, limb and head [37, 38, 49, 52].

When attempting to identify the likelihood of a satellite cell in the basal vertebrates, in particular the zebrafish, a comparison of the origin of satellite cells is complicated by the difference in their embryonic formation. Specifically, the zebrafish lacks a classic dermomyotome, utilizing a different mechanism to initially establish muscle tissue and muscle progenitors. The developmental origin of muscle progenitor and satellite-like cells in the zebrafish myotome was identified as the anterior somitic compartment of the early zebrafish somite. The origin of these cells was established through lineage tracing, whole somite imaging, and genetic loss of function experiments [26]. Cells of the anterior somitic compartment express Pax3 and Pax7, similar to the myotomal and satellite cell progenitors of the amniote dermomyotome [26]. After whole somite rotation where the anterior portion of the somite migrates to the lateral most aspect of the myotome, the external cell layer (ECL) becomes identifiable and can be found covering the myotome (Fig. 1). The cells in this layer are mitotically active and contribute to muscle growth, functioning as an equivalent to the amniote dermomyotome. The ECL is visible by Pax7 expression and continues to cover the developed myotome. A few Pax7 cells also exist interspersed within the zebrafish myotome. This continued Pax7 expression in cells that will maintain mitotic ability in the skeletal muscle is reminiscent of amniote Pax7 satellite cells [26, 53, 54]. Pax7 expression in the ECL persists well into larval development of trunk musculature [54-56]. Exactly how long the ECL lasts for, whether it gives rise to all satellite-like cells or just a subset, and when and where these satellite-like cells gain their characteristic morphology and expression pattern, all remain unknown.

By the end of primary myogenesis, the zebrafish somite contains a myotome, sclerotome and the dermomyotome-like ECL, which gives rise to mature muscle, skeleton and the myogenic progenitors responsible for subsequent growth, just as for the amniotes [57, 58]. However, some differences in myogenesis have been observed at this early developmental time. For example, in some muscle tissue, such as muscle pioneer cells, myogenic commitment occurs relatively earlier in the zebrafish compared to the amniote (Fig. 1). Furthermore, genetic analyses suggest that two methods of myogenesis occur: one where Myf5 and MyoD act together to drive slow myogenesis and the initial fast myogenesis in the medial somite, and another where only MyoD is required such as that of the pectoral fin and cranial muscle development [59]. The zebrafish myotome segregates into lateral slow muscle and medial fast muscle populations, which, again, represents a very different arrangement to the amniote where multiple muscle fibre types are located in the same muscle in a ‘salt-and-pepper’ pattern. How these distinct populations grow and regenerate remains to be determined.

These observed differences in myogenesis might largely be a result of the fact that, unlike the amniote, new skeletal muscle fibres in the zebrafish are derived from and produced in different areas of the myotome throughout the life of the fish. This ability to continue producing new fibres is termed eternal hyperplasia, and is a direct counterpoint to the exclusive reliance of mammalian muscle growth via hypertrophy postnatally [60, 61]. The most likely candidate for the continued growth in the zebrafish is the Pax7-expressing lineage [54]. These cells have been shown to form muscle fibres in the embryonic and early larval stages [53, 62]. In support of this suggestion, isolation of adult zebrafish skeletal myoblasts reveals a population of cells that express low levels of Pax7. As in mammals, zebrafish mononuclear cell cultures derived from adult muscle can differentiate into myofibres in culture [63, 64]. During myogenic differentiation in vitro, these cells are responsible for downregulating markers such as pax7a and Myf5, as well as subsequently upregulating differentiation markers such as myogenin when forming myotubes in culture [64]. Similar to mammals, zebrafish muscle fibres are multinucleated [65, 66]. Furthermore, zebrafish fibre-associated Pax7 cells express markers specific to quiescent mammalian satellite cells (e.g. the HGF receptor c-Met) that are not universally expressed by the external cell layer, suggesting that they form a discrete population of cells analogous to mammalian satellite cells [26].

Satellite-like cells have been identified and cultured in other fish species, including Atlantic salmon [67], carp [68] and rainbow trout [69] (Table 2). Although salmon are clearly different to zebrafish with respect to growth and lifespan, the identification of c-Met positive mononuclear cells within the muscle of both fish species suggests that muscle growth and possible regeneration in the adult fish are likely to be conserved [70]. Further evidence for such conservation is found in the Atlantic herring, where the satellite-like cells have characteristics similar to mammalian satellite cells; in histological section, they were observed in the form of unique elongated spindle-shaped cells distributed normally amongst myofibres [71].

Zebrafish muscle fibres are attached from one myosepta to the next, where a myosepta comprises a connective tissue structure initially established along the somite boundaries functioning in a similar fashion to a tendon in an amniote. Once attached between two myosepta, the myofibre then grows in diameter, adding new myonuclei, meaning that satellite-like cells in zebrafish could position themselves anywhere in the myotome and still enable the access of fusion-competent cells to existing myofibres. This makes the identification of possible myogenic stem cell populations more challenging. In fish myogenesis, it is also unknown whether there is a separate muscle stem cell population for hyperplastic fibre recruitment and another for hypertrophic fibre growth, similar to that described in the early post-embryonic amniote. It is hypothesized that, in the zebrafish, a pool of embryonic cells are formed at the end of development and that these cells have a high capacity for migration during muscle development [72] and regeneration [73]. We suggest that some of these cells may migrate within the muscle or stay at the position of the ECL to form pools of cells analogous to satellite cells [74]. Although gene expression and histological analyses suggest that a cell that resembles the amniote satellite cell exists in fish and amphibia, we prefer using the term ‘satellite-like cell’ to define this population until a functional validation of the stem-like properties of these cells has been achieved.

Cellular basis of regeneration in basal vertebrates

In non-disease states, it is assumed that very little turnover of mammalian muscle tissue occurs in vivo, except when muscle repair and/or regeneration is required to respond to localized trauma or damage [75]. In such cases, satellite cells are required to provide a source of new muscle nuclei because differentiated myonuclei are terminally postmitotic. Skeletal muscle is capable of a robust regeneration response when injured. Numerous mammalian injury models have been used to investigate this response, including physical manipulations such as crush and stab injuries, temperature-based injuries such as heat and freeze injuries, and chemical-based injuries such as those induced by cardiotoxin and barium chloride [3, 76-78]. As in postnatal muscle growth, satellite cells are the proliferative cell that is responsible for the ability of skeletal tissue to completely and repeatedly regenerate, forming both fusion competent myoblasts, as well as replenishing their own numbers, thus restoring muscle architecture within 2 weeks [79-81]. The similarities between the activation of satellite cells and somitic myogenesis suggest that post-embryonic myogenesis recapitulates elements of embryonic development, including the expression of myogenic regulatory factors (MRFs) [82], morphogen signalling through bone morphogenic protein [83, 84], HGF [85] and fibroblast growth factor [86]. Recently, the conditional ablation of Pax7-expressing cells from adult mouse muscle has been shown to result in a complete lack of muscle repair upon subsequent injury, thus demonstrating the requirement for Pax7-expressing cells in fibre regeneration and satellite cell replenishment in adult mammalian muscle [29, 87-89].

Are the satellite-like cells identified in lower vertebrate species such as the zebrafish capable of the same function? This definition has relied on microscopy and Pax7 expression as a starting point (Fig. 1). Early studies in the frog showed that mincing the gastrocnemius and replacing it in the location from which it was taken would result in tissue regeneration. As noted at the time, the histology of regeneration of the frog muscle was very similar to rat regeneration from minced muscle. The regeneration of the frog muscle occurred from the periphery inward and was never completely regenerated; at best, regeneration produced a ratio of 75% muscle to 25% connective tissue. Interestingly, it appeared that the frog was able to regenerate muscle with a similar efficiency when no minced muscle was introduced in place of the excised gastrocnemius, suggesting that satellite-like cells from neighbouring muscle could contribute to regeneration [90].

Although the adult anuran shares an ability to regenerate skeletal muscle with the amniote, it also shares a unique ability with some urodeles to regenerate whole structures (i.e. the limb as an early anuran tadpole). The amputated limb undergoes epimorphic regeneration, which is defined as dedifferentiation of an adult structure to form an undifferentiated mass of cells, known as the blastema. The blastema is critical for epimorphic regeneration and is similar to the early limb bud with respect to following a distinct set of morphological stages of growth, patterning and re-differentiation. This is unlike the muscle tissue regeneration that occurs through satellite-like cell proliferation and subsequent differentiation. The finding that cultured newt myotubes would break up into mononucleated cells when they were placed into a blastema [91] was evidence for epimorphic regeneration of the limb. One question arising from this experiment was that, if the urodele has satellite-like cells, do they contribute to the epimorphic regeneration, or is the limb repair solely through the dedifferentiation process regenerating the entire limb? Using Pax7 as a marker for these cells, the muscle of the regenerated limb appears to originate from both sources in the urodele: dedifferentiated mononucleate cells from the partially injured myofibres and resident satellite-like cells [24].

With regard to the similarity of the satellite-like cells found in newts and those found in mammals, there appears to be some difference with respect to the histological position. It appears that, in the newt, two basal lamina separate the satellite cell and the myofibre that it is associated with. The adult axolotl (another urodele with the ability for tissue and epimorphic regeneration) does have cells located under the muscle basal lamina in the same way that they were described originally. However, these cells were not investigated for Pax7 expression [92]. The axolotl does afford the opportunity to activate both tissue regeneration and epimorphic regeneration at the same time. This experiment was carried out by mincing an axolotl limb muscle and subsequently amputating the leg through the level of the minced muscle. It appeared that the epimorphic ability prevailed and the minced muscle participated in the formation of a new limb [90]. This leads to the intriguing question of what factor is mechanistically lost between the early anuran that is able to epimorphically regenerate and the adult that cannot? In the future, we may be able to remove all satellite-like cells from an adult urodele and examine whether epimorphic regeneration still occurs, including skeletal muscle regeneration.

Very few experiments have been carried out to test the tissue regeneration potential of many of the animals in which satellite-like cells have been identified (Table 2), although a few recent studies are able to shed some light on this issue. The cephalochordate amphioxus is capable of epimorphic regeneration of its tail, with the progression and timing being very close to that observed in frog limb regeneration. A recent study by Somorjai et al. [28] reported very few peripheral, flattened Pax3/7 positive cells residing under the basal lamina before tail amputation and a marked increase of Pax7 positive cells in a regenerative stage-dependent manner after amputation. These cells were also positive for pH3 (histone H3 phosphorylation), indicating that they were proliferative as a result of the injury. These data suggest that a population of quiescent myoblast progenitors (satellite-like cells) exists in amphioxus and were likely present in the chordate ancestor. In yellow stripe knife fish, a very unique vertebrate belonging to the gymnotiformes, which have a robust ability to regenerate their tails after amputation, the contribution of Pax7 cells to tail regrowth has been investigated. Weber et al. [93] found that a few Pax7 cells were associated with the adult skeletal muscle, as well as the muscle-derived electric organ. After amputation, the Pax7 cells proliferated and gave rise to the new muscle fibres and electrocytes that regrew from the blastema. No dedifferentiation of the muscle or electrocytes was detected at the wound. If only muscle was removed from the ventral wall, the knife fish would regenerate the lost muscle tissue, although at a much slower rate than the tail regenerated and with less proliferation of the Pax7 cells, suggesting a responsive pool of cells that was activated by blastema formation.

The zebrafish adult and larva do have the ability to epimorphically regenerate a number of organ systems, although whether skeletal muscle also exhibits an epimorphic regenerative capacity has been a matter of contention until recently. The larval tail resection injury model was used in conjunction with CRE/Lox mediated lineage tracing of zebrafish to test whether the mature skeletal muscle cells, once injured, would be dedifferentiated to repair the injury or whether new muscle was formed from muscle progenitor cells. Rodrigues et al. [94] determined that the injured myofibres did not epimorphically regenerate the injury. The zebrafish has been shown to regenerate a number of tissues and three types of injury have been performed on zebrafish skeletal muscle: needle stab, cardiotoxin injection and laser-induced micro-injury. Ablation of fibres of young larval fish via cardiotoxin injection results in a regeneration response involving a population of Pax7-expressing cells that migrate into the injury site within 48 h after injury [62, 95]. Repair of myofibres that have been laser damaged (but not ablated) at the same developmental stage appears to be a process independent of this Pax7-expressing population [96]. In the cardiotoxin model of regeneration in larval fish, the number of myogenic precursor cells can be increased by progranulin expression, suggesting that there is a population of cells responsible for muscle growth that exists in a stem cell state similar to that for progenitor cells in the zebrafish regeneration models of the fin, heart, retina and liver [97]. In adult fish, proliferative myogenic cells actively entered the injury site after needle stab injury, with new fibres being visible 11 days after injury [73]. Most of these fibres expressed all the hallmarks that are associated with muscle repair in the amniote, and subsequently expressed markers associated with fast-white muscle tissue [73]. However, the confounding issue of continual growth in the fish gives rise to the question of whether the fish would need a quiescent population in place for repair. These cells must have the ability to perpetuate themselves through self-renewal and generate mature skeletal muscle through differentiation. The ability to self-renew is also shared with cancer models induced in fish muscle, strengthening the idea that the same pathways of self-renewal may be operating and that an adult stem cell within the skeletal muscle of zebrafish could exist.

Rhabdomyosarcoma and the satellite cell

One of the most recent models of tumourigenesis suggests that the origin of cancer lies in excessively proliferating and abnormally altered stem cells [98]. Supporting the theory that the lineage and status of the cell mediates its potential in muscle-specific tumours, or rhabdomyosarcomas, when different groups of single cells isolated from muscle were sorted and induced to form tumours, the satellite cells were the only ones that induced true rhabdomyosarcomas in the injected host [99]. Conservation of cancer-related pathways and cells between fish and human has been demonstrated both histopathologically and genetically. Relative to the skeletal muscle, there are two forms of human cancer, embryonal rhabdomyosarcomas (ERMS) and alveolar rhabdomyosarcomas (ARMS). ERMS arise in the head, neck and extremities, characterized by mutations or dysregulation of the RAS pathway [100, 101]. ARMS tend to occur in limbs and the axial musculature, and the molecular basis of these tumours is a chromosomal translocation between either Pax3 or Pax7 [102] (Fig. 2). Early studies of induced rhabdomyosarcomas in rat muscle suggested that the source of the tumour was satellite cells and also that human rhabdomyosarcomas occur through a failure of some satellite cells to undergo appropriate terminal differentiation [103] (Fig. 2). Satellite cells are implicated because they possess characteristics in common with cancer cells: self-renewal, high rates of proliferation, resistance to senescence, and reversal of quiescence. Overall, ERMS-specific gene expression shows great similarity to activated satellite cell gene expression, including Pax7, c-Met and MRFs; the finding that murine satellite cells are capable of inducing ERMS experimentally also supports the idea [104-106]. Although the exact cell of origin of rhabdomyosarcomas remains unknown, the correlation of incidence and periods of high muscle growth and satellite cell activity in humans supports the theory of ERMS being induced by cells in a myogenic precursor state instead of an early expression of myogenic markers in a non-muscle lineage.

Figure 2.

Possible cellular origins of rhabdomyosarcoma in the amniote and zebrafish and the MRFs that they express. Muscle specification from the mesoderm depends on the expression levels of MRFs depicted in the nuclei. During the experimental induction of rhabdomyosarcoma in the amniote, different types of rhabdomyosarcomas can be generated depending on the oncogenes used and the developmental age of the cells. When uncommitted mesodermal cells were forced to express the fusion gene of Pax3-FOXO1 to model the human rhabdomyosarcoma where Pax3 translocates so that it is juxtaposed to FOXO1, the cells generate ARMS. When committed myoblasts are induced to express similar oncogenes, they form ERMS. It is assumed that ARMS can also arise from fully formed skeletal muscle. In the zebrafish, the induction of oncogenes to form rhabdomyosarcomas both presomitically and 24 h after fertilization resulted in the formation of ERMS from myogenic and progenitor cells. Adapted with permission [136].

Frequently, aberrations in the animal can lead to a better understanding of what is naturally occurring in the tissue, and this may be the case with human cancers in the context of zebrafish stem cells. The first studies of carcinoma in zebrafish induced rhabdomyosarcomas by embryo immersion of carcinogens [107, 108]. Two genetic models of rhabdomyosarcomas in the zebrafish have been developed subsequent to this discovery. The first model was based on driving the expression of rag2-hKRASG12D that unexpectedly expressed in the mononuclear cells within the myotome, differentiating myoblasts (and the rare fusing myoblasts) but not multinucleated terminally differentiated muscle fibres. This result suggested that, as in the amniote, there was heterogeneity within the proliferating zebrafish myocytes [100]. The mosaic larvae developed tumours with a histology and a gene expression signature similar to those of clinical ERMS in human paediatric patients. The ERMS propagating cells did not express the terminal myogenic differentiation gene α-actin but did express early muscle progenitor makers Myf5, c-Met and M-cadherin [100]. The second model was an inducible model based on using the ubiquitously expressed β-actin promoter driving the same oncogene hKRASG12D at 24 h after fertilization, with the fish being heat shocked to induce oncogenic expression in a temporally controlled fashion [109]. Both models generated tumours that expressed myogenin, MyoD and desmin, suggesting a possible predetermined stem cell residing within the skeletal muscle. It is interesting that the oncogene hKRAS under a β-actin promoter, which should express in the whole organism, induced zebrafish rhabdomyosarcoma in 75% of the tumours formed after 24 h of induction, suggesting that, at this time point, there may be more cells susceptible to Raf-induced proliferation in the myotome than in the rest of the fish [110, 111]. Supporting this idea, experiments were performed with the transgenic Myf5 zebrafish, in which the early muscle progenitor cells were marked with green fluorescent protein. These fish were injected with rag2-KRASG12D, and it was noted that all the tumours formed were green fluorescent protein positive, suggesting that the muscle progenitor cells were induced to form ERMS. The ERMS generating cells could be isolated and transplanted into a new host and were still able to induce ERMS without re-injection of the oncogene [112]. They found that these serially transplantable ERMS propagating cells could give rise to metastatic myogenin positive cells in much the same way as in humans, suggesting that the pathways governing the cell type and disease in humans are similar in the zebrafish. The similar induction of rhabdomyosarcomas in the zebrafish compared to that in mammals suggests the existence of a stem cell within the zebrafish myotome that can be transformed, thus strengthening the argument for the existence of a cell similar to the satellite cell in the zebrafish.


Regenerative ability is already established in numerous zebrafish structures, including the heart, brain, retina, spinal cord, sensory organs, scales, liver and appendages. However, the zebrafish appears to employ different strategies to repair certain tissues. Some rely on sequestered stem cell populations (brain, spinal cord, retina and neuromasts), whereas others are able to regenerate through dedifferentiation (fin and heart). Although the regenerative basis of zebrafish skeletal muscle clearly requires further investigation, it appears not to occur through dedifferentiation. If stem cell-based muscle regeneration occurs in basal vertebrates, it would suggest that the mechanism of satellite cell regeneration might be an evolutionarily conserved process. Although the functional definition of a satellite cell as a self-renewing muscle stem cell remains an experimental imperative, there is compelling circumstantial data available for the zebrafish and other basal vertebrates to suggest the broad phylogenetic existence of a satellite-like cell in vertebrates. Although these cells possess many of the defining characteristics of satellite cells, the ability to self-renew must be defined so that these cells can be referred to as a homologous cell type.

Satellite cell self-renewal and stem cell potential is one of the most intensely researched topics in muscle regeneration research. In the amniote, satellite cells comprise only 3–5.5% of the myonuclei of their associated fibre but can produce sufficient myoblasts to replace all the myonuclei of the myofibre within 4–5 days [113] and data such as this are still lacking in the zebrafish. More importantly, the amniote satellite cell pool continues to replace myonuclei when the muscle is subjected to repeated severe damage [114, 115]. Transplantation of myoblasts has shown that grafted myoblasts not only generate myonuclei [116], but also give rise to viable myogenic precursors [117-121]. Engraftment of a single mouse myofibre, with seven satellite cells, resulted in ten times as many new satellite cells in addition to many more myonuclei in the host that it was injected into, highlighting their ability to act as myogenic stem cells [33]. As we establish the similarities and differences in zebrafish and other vertebrate models versus amniotes in the context of skeletal muscle regeneration, many important questions remain to be answered. Do satellite-like cells express all the same markers? Are they also a heterogeneous population? Most importantly, can satellite-like cells proceed with efficient and repeatable muscle regeneration and self-renewal? Although many of these studies are ongoing, the results obtained will impact our understanding of how stem cells across Metazoa are identified, as well as what molecules govern their formation and function.


This work was funded by a grant from The Australian National Health and Medical Research Council Project Grant 1045468. The Australian Regenerative Medicine Institute at the Monash University is supported by grants from the Australian Government and the State Government of Victoria, Australia.