Analysis of Pax7 expressing myogenic cells in zebrafish muscle development, injury, and models of disease


  • Claudia Seger,

    1. MRC Centre for Developmental and Biomedical Genetics, University of Sheffield, Firth Court, Western Bank, Sheffield, United Kingdom
    2. A*STAR Institute of Molecular and Cell Biology, Proteos, Singapore
    Search for more papers by this author
  • Murray Hargrave,

    1. MRC Centre for Developmental and Biomedical Genetics, University of Sheffield, Firth Court, Western Bank, Sheffield, United Kingdom
    Search for more papers by this author
  • Xingang Wang,

    1. A*STAR Institute of Molecular and Cell Biology, Proteos, Singapore
    Search for more papers by this author
  • Ruth Jinfen Chai,

    1. A*STAR Institute of Molecular and Cell Biology, Proteos, Singapore
    Current affiliation:
    1. School of Anatomy and Human Biology, The University of Western Australia, Perth 6009, WA, Australia
    Search for more papers by this author
  • Stone Elworthy,

    1. MRC Centre for Developmental and Biomedical Genetics, University of Sheffield, Firth Court, Western Bank, Sheffield, United Kingdom
    Search for more papers by this author
  • Philip W. Ingham

    Corresponding author
    1. MRC Centre for Developmental and Biomedical Genetics, University of Sheffield, Firth Court, Western Bank, Sheffield, United Kingdom
    2. A*STAR Institute of Molecular and Cell Biology, Proteos, Singapore
    3. Department of Biological Sciences, National University of Singapore, Singapore
    • IMCB, 61, Biopolis Drive, Proteos, Singapore 138673

    Search for more papers by this author


The transcription factor Pax7 is a marker and regulator of muscle progenitors and satellite cells that contribute to the embryonic development and postembryonic growth of skeletal muscle in vertebrates, as well as to its repair and regeneration. Here, we identify Pax7+ve myogenic cells in the zebrafish and characterize their behavior in postembryonic stages. Mononucleate Pax7+ve cells can first be found associated with myofibers at 72 hours post fertilization (hpf). To follow the behavior of muscle progenitor cells in vivo, we generated transgenic lines expressing fluorescent proteins under the control of the pax7a or pax3a promoters. We established an injury model using cardiotoxin injection and monitored cell proliferation and myogenic regulatory factor expression in myogenic precursors cells and muscle fibers after injury using proliferation markers and the transgenic lines. We also analyzed Pax7+ve cells in animals with dystrophic phenotypes and found an increased number compared with wild-type. Developmental Dynamics 240:2440–2451, 2011. © 2011 Wiley-Liss, Inc.


In vertebrates, the growth and repair of skeletal muscle depends critically upon a pool of tissue specific stem cells, termed satellite cells that are set aside from the proliferative pool of muscle progenitors at the end of embryogenesis. First identified by electron microscopic analysis of frog leg muscles in the early 1960s (Katz,1961; Mauro,1961), satellite cells have subsequently been described in numerous vertebrate species from fish to mammals (Muir et al.,1965; Reger and Craig,1968; Allbrook et al.,1971; Kryvi,1975). In undamaged adult skeletal muscle, satellite cells reside in a quiescent state between the basal lamina and plasmalemma of mature myofibers (Schultz et al.,1978). They are small, contain highly condensed single nuclei, have reduced cytoplasm and are distributed throughout the terminally differentiated adult muscle (Nag and Nursall,1972; Stoiber and Sanger,1996), their density depending on age, muscle fiber type, and species (Hawke and Garry,2001).

In mice, satellite cells account for approximately 30% of sublaminar muscle nuclei at birth, but this proportion decreases to less than 5% by the age of 2 months old (Bischoff,1994), reflecting their contribution to muscle fibers during postnatal growth (Gibson and Schultz,1983). Satellite cells are mitotically and physiologically quiescent but can be activated by a variety of local stimuli such as exercise, stretching and trauma due to injury and disease. Upon activation, the cells divide asymmetrically, giving rise to both uncommitted daughters that maintain the satellite cell population (Bischoff,1994; Angello and Hauscka,1996; Baroffio et al.,1996) and to committed myogenic precursor cells (MPCs) that proliferate and differentiate, either by fusing together to generate de novo fibers or by fusing with existing fibers (Moss and Leblond,1971; Grounds and Yablonka-Reuveni,1993). The ability to give rise to both undifferentiated and differentiating progeny by entering the cell cycle led to the designation of satellite cells as muscle stem cells.

In amniote embryos, satellite cells have been shown to arise from a distinct population of dermomyotomal progenitors that express and require the related transcription factors Pax3 and Pax7 for their specification (Gros et al.,2005; Kassar-Duchossoy et al.,2005; Relaix et al.,2005). Pax7 has been identified as a key determinant of satellite cells, as all skeletal muscle groups of Pax7 null mutant mice lack satellite cells and fail to grow or regenerate after birth (Seale et al.,2000). Pax3 is also expressed in satellite cells (Conboy and Rando,2002; Buckingham et al.,2003) and shares similar and overlapping functions with Pax7, as well as controlling precursor cell migration to sites of muscle formation in the limb and body wall (Relaix et al.,2006). Both Pax3 and Pax7 have been suggested to maintain cells in an undifferentiated state with the potential to proliferate, thereby maintaining the muscle stem cell pool (Oustanina et al.,2004; Relaix et al.,2006). Recently, however, it has been found that inactivation of Pax7 in mice 2–3 weeks after birth has no effect on the regenerative or self-renewal capabilities of satellite cells. This finding implies that, while Pax7 directs the survival and proliferation of the myogenic progenitor cells from which satellite cells derive, it is dispensable once the cells have entered their quiescent state (Lepper et al.,2009).

As in amniote embryos, Pax3 and Pax7 are initially expressed within the dermomyotome of the zebrafish embryo (Devoto et al.,2006; Hammond et al.,2007; Hollway et al.,2007), which in this species arises from the anterior portion of the somite and, following an anterior-to-lateral rotation, comes to cover the lateral surface of the primary myotome (Hollway et al.,2007; Stellabotte et al.,2007). Lineage tracing of individual cells in the Pax7+ve anterior region of the somite demonstrated that they give rise to cells of seven distinct fates, including muscle progenitors that generate fibers during a secondary period of larval muscle growth (Hollway et al.,2007). In addition, mitotically quiescent Pax7+ve cells were identified from early larval stages (3 dpf), located beneath the basal lamina of individual fibers and accordingly suggested to be satellite cells (Hollway et al.,2007).

Here, we characterize further the Pax7+ve myogenic cells in larval stages: we show that some Pax7+ve cells divide and contribute to myotomal growth during the larval stages, while some migrate deeper into the myotome. Following cardiotoxin-induced injury, we observe that Pax7+ve cells migrate to the site of injury, where they proliferate. Concomitantly, myogenic regulatory factors (MRFs) are up-regulated in myofibers adjacent to the site of injury. We also find that Pax7+ve cells numbers are increased in two zebrafish models of muscular dystrophy.


Ontogeny of Pax7+ve Cells in the Myotome

Using a mAb raised against the chicken Pax7 protein (DSHB) previously shown to recognize Pax7 in zebrafish (Kawakami et al.,1997; Feng et al.,2006; Hammond et al.,2007; Minchin and Hughes,2008), two different types of Pax7+ve cells could be clearly distinguished in embryos at 24 hpf: the majority (∼30 per somite) of labeled cells were rounded and located close to the surface of the somite, nestled between and superficial to the Prox1+ve slow-twitch fibers (Figs. 1A, 2M) (Feng et al.,2006). In addition, small numbers of much more intensely labeled cells with dorsoventrally elongated bean-shaped nuclei that correspond to pigment cells (xanthophores) were located in the dorsal superficial somite (Fig. 1B). Approximately a third of the Pax7+ve cells within the myotome were co-labeled with a polyclonal antibody specific for the zebrafish Myod protein (α-Myod), suggesting that they were fated to form myofibers (Fig. 1C,D; Supp. Fig. S1, which is available online).

Figure 1.

Ontogeny of Pax7+ve cells in the myotome. A–C,F–J: Lateral flatmounts of embryos stained with α-Pax7 monoclonal antibody (mAb): anterior is to the left and dorsal up. A: Around 30 weakly labeled rounded Pax7+ve cells nestle between slow muscle nuclei (labeled by Prox1 - red) in the lateral myotome of each somite at 24 hours post fertilization (hpf). B: At the same stage, weak Pax7 immunoreactivity is detected in round cell nuclei located more superficially (arrows) and at much higher levels in the xanthophores, bean-shaped cells that are elongated along the dorsoventral axis (arrowheads). C,D: At 24 hpf, approximately a third of Pax7+ve cells are differentiating as they co-express Myod (arrows). Numbers based on analysis of three somites in four to seven animals, error bars show standard error. Counts include only weakly stained rounded Pax7+ve cells and exclude intensely stained elongated Pax7+ve xanthophores. E,F: At 48 hpf, the xanthophores remain strongly labeled; weaker Pax7+ve cells (green) are dispersed throughout the somite but several have aligned along the vertical and horizontal myosepta, labeled with anti-Laminin (red). G,H: From 3 days postfertilization (dpf), some Pax7+ve cells associate closely with myofibers stained with Phalloidin (actin, red). I–M: Pax3/7 cells are surrounded by β-catenin staining in cryostat cross-sections at 6 dpf (arrows), while other myonuclei are not fully surrounded (arrowheads). J,K are magnifications of the boxed area in I. N–Q: Cross-section of muscle fibers labeled with αPax3/7 (red) and Laminin (green) show that Pax3/7+ve cells are surrounded by laminin staining (arrows) at 6 dpf. Scale bar = 50 μm in A–C,E,F, 25 μm in I–Q.

Figure 2.

Expression of TgBAC(pax7a:GFP)i131 and TgBAC(pax3a:GFP)i150 recapitulates the patterns of the endogenous genes in the myotome. A–C: At 24 hours post fertilization (hpf), Pax7 protein (A) and TgBAC(pax7a:GFP)i131 expression (B) mostly coincide (C), arrowheads highlight cells detected with α-Pax7, but barely GFP+ve. Green fluorescent protein (GFP) expression is also detected in processes extending from the xanthophores (B, arrows). D–F: At 48 hpf, labeling with α-DP312 (Pax3/Pax7) (D) and GFP expression driven by TgBAC(pax3a:GFP)i150 (E) show significant overlap (F); some nuclei labeled strongly by DP312 show little or no GFP expression, consistent with some cells expressing only Pax7 (arrows, inserts). G–I: Increasingly medial planes through 5 days postfertilization (dpf) TgBAC(pax7a:GFP)i131 larvae: at a superficial level, expression is observed in xanthophores (arrowheads) and along the horizontal and vertical myosepta (arrows, G). H: More medially, Pax7a cells align closely to myofibers, stained with Phalloidin (actin, red; arrows). I: Intense GFP expression in dorsal spinal cord (long arrow), dorsal neural crest cells (arrowheads), and alongside myofibers (small arrows). J: Individual fibers (α-actinin, red) are labeled in 48 hpf TgBAC(pax7a:GFP)i150 reflecting perdurance of the GFP protein. K,L: TgBAC(pax3a:GFP)i150 at 6 dpf: expression persists in fibers (K) and remains in rounded cells alongside the fibers at 10 dpf (L). M: Quantification of Pax7+ve nuclei at 24 hpf to 5 dpf detected in fixed specimens with αPax7 mAb and in vivo pax7a:GFP+ve cells. Counts include only weakly stained or weakly expressing Pax7+ve cells per somite and exclude intensely stained elongated Pax7+ve xanthophores. N–P: Cross-sections of Tg(Pax7a:GFP)i131 show a Pax7a:GFP+ve cell fully surrounded by β-catenin (arrows) and persistence of GFP in an individual fiber (arrowhead) at 6 dpf. N,O are magnifications of the boxed area in P. Scale bars = 50 μm in A–L, 25 μm in N–P.

At 48 hpf, toward the end of embryogenesis, the number of Pax7+ve cells found within the myotome was reduced by half and some were now located at the horizontal myoseptum, while others remained along the vertical myoseptum (Fig. 1E,F; 3M). By the early larval stage (72 hpf), the number of Pax7+ve cells on the superficial edges of the myotome decreased further to an average of 9 cells per somite; most such cells still accumulated at the horizontal and vertical myosepta, although individual cells could be found associated with myofibers (Figs. 1G,H, 3M). Counts at all stages included only weakly stained Pax7+ve cells and excluded intensely stained elongated Pax7+ve xanthophores. The cells nestled much closer within the fibers than at 48 hpf and had migrated deeper into the myotome. Analysis of sections of 6 dpf larvae stained with mAb DP312 that recognizes both Pax7 and Pax3 (Davis et al.,2001; Hammond et al.,2007), revealed labeled nuclei in between the fibers. At this stage, these Pax3/7+ve nuclei were fully surrounded by the membrane markers β-catenin and laminin, indicating that they belong to individual cells distinct from the multinucleated fibers (Fig. 1I–Q).

Generation of Pax3 and Pax7 Reporters That Label Muscle Progenitor Cells

To investigate further the behavior of the Pax7+ve cells in vivo, we generated transgenic lines that express fluorescent proteins under the control of the pax7a and pax3a promoters. BACs containing the genomic loci of each gene were identified and sequences encoding GFP (green fluorescent protein) or lyn-td-Tomato were inserted downstream of the translation initiation codons by recombineering (Lee et al.,2001; see the Experimental Procedures section and Supp. Fig. S2 for details). Approximately 350 embryos were injected and at least three founders were isolated for each transgene; only the founder lines showing the highest levels of fluorescence were used in subsequent analyses. For both genes, independently derived lines showed identical patterns of expression (Supp. Figs. S5, S6). Staining with an αPax7 mAb (Fig. 2A–C) and with mAb DP312 (Fig. 2D–F) confirmed that these lines largely recapitulated the endogenous expression patterns of these genes in the myotome (Fig. 2G–I). In addition, both the TgBAC(pax7a:GFP)i131 and TgBAC(pax3a:GFP)i150 lines showed expression in the neural crest cells, the brain and very strongly in the neural tube (Supp. Figs. S3, S4) consistent with the endogenous expression patterns of the two genes (Seo et al.,1998; Minchin and Hughes,2008).

Both the number of non-xanthophore αPax7 mAb stained and GFP-expressing Pax7a+ve cells decreased significantly between 24 hpf and 5 dpf of development (Fig. 2M). However, slightly fewer GFP+ve cells were present in the trunk myotome of live TgBAC(pax7a:GFP)i131 animals than were labeled in fixed specimens with the αPax7 mAb (Fig. 2M,B arrowheads). The zebrafish genome contains genes encoding two Pax7 orthologues, designated Pax7a and Pax7b, that share 94% amino acid identity, both of which have been suggested to be detected by the monoclonal antibody (mAb) raised against chicken Pax7 (Kawakami et al.,1997; Minchin and Hughes,2008). Thus the disparity between numbers labeled by the transgenic line vs. labeled by the antibody could be due to the presence of cells that exclusively express Pax7b, which is likely recognized by the Pax7 mAb. However, it is also possible that any heterogeneity in the expression of pax7a could result in the signal from some cells falling below the level of detection.

At later stages, from 48 hpf onward, we found that entire muscle fibers were labeled by the fluorescent proteins of both TgBAC(pax7a:GFP)i131 and TgBAC(pax3a:GFP)i150 lines (Fig. 2J,K,P). Such a pattern is not observed either by in situ hybridization or by αPax7 and αPax3 antibodies, suggesting that it reflects the perdurance of the GFP expressed by the Pax3a+ve/Pax7a+ve progenitor cells from which the fibers have derived. Counts include only weakly stained Pax7+ve cells and exclude intensely stained elongated Pax7+ve xanthophores.

Using time lapse video-confocal microscopy, one or more Pax7a+ve cells in each somite could be seen to divide between 30 hpf and 36 hpf; typically one of the two daughters migrated into the myotome whilst the other remained in the vicinity of the vertical myoseptum (Fig. 3A). At 58 hpf, individual Pax7a+ve cells could still be seen to divide (Fig. 3B; see also Supp. Movies S1 and S2 in Supp. data).

Figure 3.

Following Pax7a+ve cell division in live embryos. A: Sequences of images of a 34 hours post fertilization (hpf). TgBAC(pax7a:GFP)i131 embryo captured every 90 sec. Individual GFP+ve cells (red arrows) can be seen dividing in each somite. A–G, A′–D′, A″–G″: Three different cell divisions are shown. In each case, one of the two daughter cells moves into the myotome while the other remains close to the vertical myoseptum. B: Series of images of a TgBAC(pax7a:GFP)i131 embryo at 58 hpf showing a GFP+ve cell (red arrow) dividing in the dorsal somite (A*–L*). The more ventral daughter cell appears to elongate to form a myofiber. The individual images were taken approximately 160 sec apart. Scale bar = 50 μm.

At 4 and 5 dpf, significantly fewer Pax7a+ve cells were found within the myotome, the majority positioned along the vertical myosepta (Fig. 2G–I), and some also within the myofibers (Fig. 2H). They had a more flattened shape than at earlier stages. At 6 dpf, the GFP signal was fully surrounded by the membrane marker β-catenin, confirming the presence of individual Pax7a+ve cells between the myofibers (Fig. 2N–P).

Pax7+ve Cell Behavior in a Cardiotoxin Based Injury Model

The rattle snake (Naja naja atra) cardiotoxin is routinely used experimentally to injure muscle fibers in mammals: it acts by depolarizing the membrane, thereby disrupting sarcomeric structure (Chang et al.,1972). To investigate the behavior of the Pax7+ve cells in response to myofiber injury, we established a protocol for cardiotoxin-induced muscle injury in the zebrafish larva (see Experimental Procedures). Using a fine glass capillary, cardiotoxin was injected directly into the muscle of somite 10 or a neighboring somite around the end of the yolk sac, before the yolk extension at 3 dpf or later. Whereas very few rounded Pax7+ve cells (approximately 1–5 per somite, Fig. 4A) could be seen in control wild-type 5 dpf larvae, significant numbers of such cells accumulated at the site of injury within 24–48 hr of injection with 0.3 mg/ml Cardiotoxin (Fig. 4B,D; see also Supp. Movie S3 in Supp. data). By contrast, although injection with phosphate buffered saline (PBS) alone resulted in a separation of fibers creating a gap in the myotome by 5 dpf, very few if any Pax7+ve cells were present at the lesion (Fig. 4C).

Figure 4.

Behavior of Pax7+ve cells in response to cardiotoxin-induced injury. A: By 5 days postfertilization (dpf), Pax7+ve cells are relatively scarce in individual somites of uninjured larvae: in this image a single labeled cell (red: αPax7) is located at the dorsal extreme (arrow). B: Cardiotoxin injected larvae at 5 days postfertilization (dpf): several Pax7+ve cells accumulate at the site of injury. C: Control injection with phosphate buffered saline (PBS) creates a gap in the myotome (arrow) but no accumulation of Pax7+ve cells. D: numbers of Pax7+ve cells accumulated at the site of injury within 24–48 hr of injection into the somite with 0.3mg/ml Cardiotoxin (yellow bar) compared with noninjured (blue bar); injection with PBS alone resulted in fewer Pax7+ve cells present (red bar). Numbers based on analysis of three somites in four to seven animals, error bars show standard error. Asterisks: P < 0.001. Counts include only weakly stained Pax7+ve cells and exclude intensely stained elongated Pax7+ve xanthophores. E–G: Five dpf injured TgBAC(pax3a:GFP)i150 stained with DP312 (Pax3/Pax7), which labels both GFP+ve cells (arrows) and bright xanthophores (arrowheads). H: Numbers of Pax3+ve and Pax7+ve cells compared with their respective transgenic lines in noninjured and injured backgrounds. The numbers of Pax7+ve cells counted in fixed specimens was significantly higher in injured vs. noninjured somites in TgBAC(pax7a:GFP)i131 embryos and Pax7mAb stained embryos, while no significant increase was observed following injury of TgBAC(pax3a:GFP)i150 embryos compared with noninjured TgBAC(pax3a:GFP)i150 embryos. Numbers based on analysis of three somites in four to seven animals, error bars show standard error. Asterisks: P < 0.005. I–K: 5 dpf injured TgBAC(pax7a:GFP)i131 stained with αPax7 mAb (K, red). Individual Pax7+ve cells are positive both for Pax7and pax7a:GFP (arrows, J,K), but up-regulation of GFP is also seen in the injured fibers (arrowheads, J). L: Another example of an injured 5 dpf TgBAC(pax7a:GFP)i131 with small inserts showing the injury site and the GFP channel with individual cells alongside the injured fibers (arrows). M: Number of bromodeoxyuridine-positive (BrDU+ve) nuclei in injured vs. noninjured TgBAC(pax7a:GFP)i131 embryos at 4 dpf. While there is a significant increase of labeled BrDU+ve cells overall, the proportion of BrDU+ve GFP+ve labeled cells is significantly increased in injured vs. noninjured fish (marked in red shading in the graph). Numbers based on analysis of three somites in 4–7 animals, error bars show standard error. Asterisks: P < 0.005. Counts include only weakly expressing GFP+ve cells and exclude intensely expressing elongated Pax7a:GFP xanthophores. N–P: A GFP+ve cell (arrow) in an injured 4 dpf TgBAC(pax7a:GFP)i131(N) expresses pcna mRNA (O, red). Q–S: BrdU labeling (R, red) of GFP+ve cells in TgBAC(pax7a:GFP)i131 (Q) embryos injured at 4 dpf. Note also up-regulation of GFP expression in fibers around the injury site (S, arrows). Scale bar = 50 μm.

Live imaging of both TgBAC(pax7a:GFP)i131 and TgBAC(pax3a:GFP)i150 embryos following cardiotoxin injection revealed the presence of motile GFP+ve cells at the site of injury (Supp. Movie S4), consistent with the accumulation of Pax7+ve cells described above. Notably, however, whereas the numbers of such cells counted in fixed specimens was significantly higher in injured vs. non injured somites in TgBAC(pax7a:GFP)i131 embryos (Fig. 4E–I), no significant increase was observed following injury of TgBAC(pax3a:GFP)i150 embryos (Fig. 4J–L,H). Interestingly, muscle fibers at the site of injury also appeared to be GFP+ve, an effect seen predominantly in injected TgBAC(pax7a:GFP)i131 embryos (Fig. 4I–L).

Accumulation of pcna mRNA provides an assay for cell cycle entry (Mathews et al.,1984). In situ hybridization analysis of injured embryos revealed accumulation of pcna transcript in Pax7+ve cells as early as 20 hr postinjury, consistent with initiation of cell division (Fig. 4N–P). However, because PCNA (proliferating cell nuclear antigen) is also involved in DNA repair (Essers et al.,2005), some of this expression might be associated with damaged muscle tissue. We therefore used BrDU (5-bromo-2′deoxyuridine) incorporation to detect dividing cells over several rounds of cell division (Gratzner,1982). TgBAC (pax7a:GFP)i131 embryos were exposed to 10mM BrDU for 15 minutes at different times after injury, before being fixed and stained with α-BrDU antibody. The first proliferating cells could be seen at the site of injury within 24 hr (Fig. 4Q–S). Significantly more cells were labeled with BrDU in injured compared with noninjured larvae; the proportion of these BrDU+ve cells that were also GFP+ve (Fig. 4M) was also significantly higher than in noninjured controls (indicated by red shading in the histogram in Fig. 4M).

Up-regulation of MRFs in Injured Fibers

The expression of the MRFs myf5 and myod in muscle cells and fibers following injury was also monitored. Injection of cardiotoxin into the muscles of larvae from a myf5 reporter line Tg(myf5:YFP)CLGY237 (Ellingsen et al.,2005) led to the rapid up-regulation of YFP expression in individual fibers (Fig. 5A); staining with the Pax7 mAb revealed a close association between the myf5 reporter expression and Pax7+ve cells at the site of injury (Fig. 5B). We note that in wild-type larvae, or in larvae carrying nonmuscle GFP reporter genes, some faint fluorescence could be detected following injury, specifically in the green channel; however, this signal was well below the levels observed in the injured fibers of Tg(myf5:YFP)CLGY237 larvae (Fig. 5C,D). Cardiotoxin injection of larvae of a novel myod reporter line, TgBAC(myod:GFP)i124, revealed the presence of newly formed fibers. These fibers had very small diameters and in some cases traversed noninjured fibers (Fig. 5E,F). Analysis of endogenous myf5, myod, and myogenin expression by in situ hybridization of injured larvae confirmed that the MRFs were expressed in the fibers adjacent to the site of injury (Fig. 5G, and data not shown).

Figure 5.

myf5:YFP (YFP, yellow fluorescent protein) and myod:GFP (GFP, green fluorescent protein) in cardiotoxin-induced injury. A,B: Injured Tg(myf5:YFP)CLGY237 larva at 4 days postfertilization (dpf): note up-regulation of YFP expression in fibers around the injury site (A) and accumulation of Pax7+ve cells around the YFP+ve fibers (B). C,D: Injured wild-type larva at 5 dpf (C) and at 4 dpf (D) showing slight auto-fluorescence in the injured fiber: this is much weaker and more diffuse than the signal seen in the Tg(myf5:YFP)CLGY237 injured larvae. E,F: Control (E) and injured TgBAC(myod:GFP)i124 (F) larva at 4 dpf: note new small myod-expressing fibers that have formed in response to injury (arrows in E). G: Expression of myf5 (assayed by in situ hybridization) in fibers localized at the site of injury in a 4 dpf larva.

Pax7+ve Cells Accumulate at Lesion Sites in Zebrafish Models of Muscular Dystrophy

In human Duchenne muscular dystrophy (DMD) patients, muscle fibers undergo repeated cycles of degeneration and regeneration, leading to the activation and the rapid depletion of satellite cell reserves that correlates with the loss of regenerative capacity (Webster and Blau,1990). The sapje mutation disrupts the zebrafish orthologue of the human DMD gene, dystrophin, resulting in the failure of embryonic muscle end attachments and a progressive muscle degeneration phenotype (Bassett et al.,2003). To investigate whether this is similarly accompanied by the activation of myogenic progenitors, we analyzed Pax7+ve cells in sapje mutant larvae. The sapje mutants were readily identified by the loss of birefringency, a property imparted by the parallel thread-like myofibrils within muscle fibers (Fig. 6A,B); loss of birefringency is thus indicative of the absence of striated muscle tissue, lack of muscle fibers or loss of fiber organization. Myotomal lesions could be detected within the somites of sapje mutants by 48 hpf. Pax7+ve cells were found primarily within the damaged fiber areas of 2-4 dpf mutants, (Fig. 6C–E). At 3 dpf, fewer Pax7+ve cells were found in areas not strongly affected by the mutation relative to wild-type controls, whereas significantly more Pax7+ve nuclei were found in areas that showed major defects (Fig. 6F). When counting the Pax7+ve cells, Xanthophores were excluded, as described above.

Figure 6.

Behavior of Pax7+ve cells in dystrophic muscle. A,B: Birefringence of muscle fibers in polarized light reveals large gaps in sapjeta222a mutant larvae compared with wild-type at 5 days postfertilization (dpf). C: Pax7+ve cells can be seen within the gaps of degrading myofibers in sapjeta222a mutants at 2 dpf (arrows show a subset of Pax7+ve cells localized within the lesion sites). D,E: Pax7+ve cells continue to accumulate at 3 dpf and 4 dpf in sapjeta222a mutants as fewer fully intact fibers remain (arrows show Pax7+ve cells nestled in between detached myofibers, arrowheads xanthophores). F: Increase in numbers of Pax7+ve nuclei at 3 dpf in damaged somites (yellow bar) of sapjeta222a mutants compared with wild-type (blue bar). Counts in less severely damaged somites are slightly reduced (red bar). The Y-axis shows the number Pax7+ve cells per somite. Numbers based on analysis of three somites in four to seven animals, error bars show standard error. Asterisks: P < 0.001. Counts include only weakly stained rounded Pax7+ve cells and exclude intensely stained elongated Pax7+ve xanthophores. G,H: Accumulation of Pax7+ve cells in the myotome of 3 dpf larvae following morpholino mediated knock down of Integrin α7 expression. Scale bar = 50 μm.

The Integrin alpha7beta1 is a specific cellular receptor for the basement membrane protein Laminin-1, and like the Dystrophin-associated protein complex, is essential for the linkage between the extracellular matrix and cytoskeleton of muscle fibers (Mayer et al.,1997); Mice homozygous for a mutation of the integrin alpha7 gene are viable but develop symptoms typical of a progressive muscular dystrophy (Mayer et al.,1997), as do zebrafish embryos following injected of an antisense morpholino specific for integrin α7 (Postel et al.,2008). In such MO-injected embryos, Pax7+ve cells also accumulated in the lesion points, as in sapje (Fig. 6G,H).


Postembryonic muscle growth in amniotes depends upon a mitotically active population of progenitors cells located in the myotome. These muscle progenitors have been shown to express both Pax3 and Pax7, to originate from the central domain of the dermomyotome and to give rise to a persistent population of muscle stem cells, the satellite cells (Ben-Yair and Kalcheim,2005; Gros et al.,2005; Kassar-Duchossoy et al.,2005; Relaix et al.,2005). The Pax3/7+ve cells migrate directly beneath the central dermomyotome during its final dissociation, infiltrating the entire myotome (Ben-Yair and Kalcheim,2005; Gros et al.,2005). In teleosts, muscles continue to grow throughout life, both by hypertrophy and hyperplasia (Rowlerson and Veggetti,2001). In the pearl fish, mitotically active Pax7+ve cells have been shown to migrate from the posterior dermomyotomal lip (Steinbacher et al.,2008), forming a population of muscle progenitor cells that contribute to stratified fast muscle hyperplasia during the larval stages (Marschallinger et al.,2009).

Previous studies have identified putative satellite cells in several other fish species such as eel, seabass, rainbow trout, and carp, which in each case have been shown to enter the cell cycle and to be the source of newly forming fibers (Alfei et al.,1994; Akster et al.,1995). Adult zebrafish have previously been shown to be capable of regeneration in different skeletal muscle injury models. Rowlerson et al. (1997) injured both adult zebrafish and sea bream (Sparus aurata) to compare their regeneration potential; despite differences in their normal growth-related hyperplasia, dynamic regeneration processes took place in both species with the formation of new fibers, showing that the very limited degree of postlarval hyperplasia of the zebrafish cannot be attributed to a paucity of myogenic progenitor cells. Consistent with this, Hollway et al. (2007) reported the association of quiescent Pax7+ve cells with myofibers, which they described as lying underneath the basal membrane (revealed by anti-laminin staining), consistent with them being satellite cells. In another teleost, Coregonus lavaretus, Pax3 and Pax7 have also been described as being expressed in muscle progenitor cells (Kacperczyk et al.,2009). In the carp Cyprinus carpio, and the teleost Atlantic herring, Clupea harengus, a population termed “myosatellite cells” have been found positioned outside as well as underneath the basal lamina (Koumans et al.,1993; Johnston and Horne,1994; Johnston and Cole,1998).

Here, we present several lines of evidence that Pax7 expression marks muscle progenitor cells in the zebrafish. The number of Pax7+ve cells decreases from 24 hpf to 5 dpf of development. Those Pax7+ve cells that take up positions between the myofibers beneath the superficial external cell layer appear fully encapsulated by laminin, suggesting that they are interstitial rather than satellite cells. When larvae are injured with cardiotoxin, Pax7+ve cells migrate around the site of injury and enter the cell cycle while adjacent fibers up-regulate MRF expression. In both sapje mutants and itga7 morphants, two independent models of muscular dystrophy, we observed an increase of Pax7+ve cells in areas of muscle fiber damage. Taken together, these findings indicate that Pax7+ve cells have typical muscle progenitor cell properties.

Given the perdurance of GFP implied by the detection of labeled fibers in the myotome of both pax3a;GFP and pax7a:GFP transgenic embryos, one might expect that similarly labeled fibers would appear following cardiotoxin induced injury or in dystrophic fish. While we did observe labeled fibers close to the site of injury following cardiotoxin injection, their morphology and the rapidity with which they became GFP+ve argues against them being newly derived from muscle progenitor cells. On the other hand, only very rarely did we observe newly formed fibers that were GFP+ve in injured fish. We suspect that this implies that under regeneration conditions, cells recruited to the site of injury down-regulate the pax7a:GFP as they enter the myogenic program. A similar phenomenon has been reported in mouse transgenic lines, in which following activation, Pax7 became progressively transcriptionally inactivated until undetectable in differentiated cells (Zammit et al.,2006). Even in the myotome of transgenic embryos, the frequency of GFP+ve cells is much lower than would be expected if all the derivative of Pax3 or Pax7 expressing progenitor were labeled. Tracing the fate of such progenitors will require the generation of pax3a:CreER and pax7a:CreER lines which is currently ongoing in our laboratory.

It has been suggested that in the zebrafish, myosepta act as “migration highways” from the lesion area to surrounding areas, and new fibers form away from the myosepta (Rowlerson et al.,1997), which has already been implied by Stoiber and Sanger (1996). While such behavior was also seen in our experiments, with some Pax7+ve cells migrating along the myosepta, others seemed to remain within the centre of the somite, with no contact to the myosepta. Future investigations into the later phases of the regeneration processes will allow more insights into myogenic progenitor cell behavior in the zebrafish.

In conclusion, zebrafish, like other teleost species and amphibians, have muscle progenitor cells with similar characteristics to those found in amniotes: expression of the key marker Pax7 which labels them throughout various larval stages and their participation in injury and muscle repair in dystrophic conditions.


Zebrafish Lines and Maintenance

Wild-type embryos were obtained from AB or LWT strains. Transgenic embryos were created: TgBAC(pax7a:GFP)i131, TgBAC(pax3a:GFP)i150, TgBAC(pax7a:lyndttom)i167, TgBAC(myod:GFP) i124. Tg(Fli:GFP)y1 (Lawson and Weinstein,2002) was obtained from Tim Chico (University of Sheffield). Tg(myf5:GFP)CLGY237 was obtained from Thomas Becker (University of Bergen, Norway). Sapjeeta222a was obtained from the Tuebingen Stock Centre at the Max Planck Institute for Developmental Biology. Adult fish were maintained on a 14-hr light 10-hr dark cycle at 28° C in UK Home Office approved facilities in the MRC CDBG aquaria at the University of Sheffield and were staged according to (Kimmel et al.,1995).

In Situ mRNA Hybridization and Immunohistochemistry

In situ hybridization was performed essentially as previously described (Oxtoby and Jowett,1993). pax7b, the duplicate gene identified by (Minchin and Hughes,2008) with 510 amino acid with 94% identity to the pax7a gene, is expressed very similar and also labels cells within the dermomyotome. myf5 probe was from (Coutelle et al.,2001) and the pcna probe from EST clone fc43g05 (MPMGp609L0932, RZPD).

Antibody Staining

Immunohistochemistry was performed essentially as previously described (Roy et al.,2001). Embryos were fixed in 4% paraformaldehyde in PBS. Mouse monoclonal anti-Pax7 (Hybridoma Bank) was used at 1:20 dilution, anti-Laminin (Sigma) at 1:100, DP312 at 1:50 (Davis et al.,2001), α-actinin (Sigma) at 1:500, rabbit polyclonals α-GFP (1:1,000, Torrey Pines Biolabs), mouse α-GFP (1:500, Clontech), β-catenin (1:500, Abcam), and α-Prox1 (1:5,000). α-Prox1 was raised against recombinant zebrafish Prox1 purified from E. coli (A.M. Taylor, personal communication). α-mouse or rabbit secondary antibodies Alexa 488 or Alexa 568 (Molecular Probe) and antibody Cy3 (Jackson lab) were used at 1:500. Nuclei were stained using TOTO-3 iodide (Molecular Probes) at 1:1,000 or DAPI (1:500) and Actin stained using Phalloidin (Sigma) at 1:100. If embryos older than 3 dpf were used, the embryos were incubated with Proteinase K (10 μg/ml) for at least 30 min at room temperature, washed with PBTX, incubated in acetone for 7 min at −20°C, washed with PBTX and incubated roughly 2 hr per dpf (maximal overnight) in 0.1% collagenase in PBS solution, before the addition of the primary antibody. If the signal of a particular antibody proved too weak, the Tyramide Signal Amplification kit by Molecular Probes was used to intensify the signal.

BrDU Staining

Embryos were kept for at least 10 min in 10 mM BrDU (Sigma) and 15% dimethyl sulfoxide [DMSO] in Ringers Saline (116 mM NaCl, 2.9 mM KCl, 1.8 mM CaCl2, 5 mM HEPES in H2O) on ice before fixation. The embryos were washed in PBT and incubated in 2N HCl for 30 min at room temperature before a 90-min 10 μg ml−1 proteinase K digestion and standard immunofluorescence, Bu20a anti-bromodeoxyuridine (1:100, DakoCytomation).

Myod Antibody

Anti-Myod sera was raised in rabbits against a fragment of the zebrafish Myod protein corresponding to amino acid 170–275 fused to glutathione S-transferase (GST), which was expressed in and purified from E. coli. The same Myod-GST fusion protein was used to purify polyclonal anti-Myod antibody from the sera as previously described (Robinson et al.,1988). To verify the specificity of the purified antibody, Western blots of Myod-MO injected fish were performed: these showed a specific reduction in the band detected by the α-Myod antibody (Supp. Fig. S1).

Homologous BAC Recombination

A GFP-SV40pA-FRT-Kn-FRT recombineering targeting cassette and red recombineering system in EL250 cells were used to insert EGFP and lyn-dt-tomato with an SV40 polyadenylation site at the pax7a and pax3a ATG start site in BACs CHORI173_62K19 (for pax7a, which has approximately 45 kb upstream and approximately 100 kb downstream of the pax7a transcription unit), CH211-20F20 (for pax3, which contains approximately 100 kb upstream and 55 kb downstream of the pax3 transcription unit) and zC173-A2 (for myod, which contains approximately 52 kb upstream and 81 kb downstream of the myod transcription unit). Nothing was excised from the BAC when the GFP or lyn-td-tomato construct was inserted, but the Kanamycin cassettes between LoxP sites that were inserted for antibiotic resistance selection were removed again using L-Arabinose treatment. An Iscei- site was inserted into each BAC using further homologous recombination. These modified BACs, linearized with I-SceI, were used to generate stable transgenic lines TgBAC(pax7a:gfp)i131, TgBAC(pax7a:lyn-tom)i167 and TgBAC(pax3a:gfp)i150 as previously described (Higashijima et al.,1997).

Cardiotoxin Injury

A total of 0.3 mg/ml Cardiotoxin I from Naja naja atra (Sigma) was injected using a fine glass capillary needle, that is commonly used to inject fish embryos, into one of the somites number 6–12 (counting from anterior) of 3 dpf (unless otherwise stated) anesthetized fish that had been embedded in 1.5% low melting agarose. Various volumes from 0.5 nl to 4 nl were injected.

Morpholino Injections

Approximately 3 nl of itga7 morpholino(AAGTCAGTCATCTTCTGCATGG TTG) was injected at 0.3 mM concentration into 1–2 cell stage zebrafish embryos (Rosemary Kim, unpublished observations). The morpholino was initially designed against the ATG site according to the Ensembl database 2005, but according to the 2010 Ensembl database, the sequence is approximately 1 kb upstream of the start site of the gene, but its phenotype is identical with the one observed in (Postel et al.,2008). The ATG site Myod Morpholino used (ATATCCGA CAACTCCATCTTTTTTG) was injected at 0.5 mM concentration into 1–2 cell stage zebrafish embryos (Lin et al.,2006).

Quantitation and Statistical Analysis

All cell counts in this study included only weakly stained Pax7+ve cells and excluded intensely stained elongated Pax7+ve xanthophores. Cell counts were performed on confocal stacks spanning an entire segment on one side of the animal. Three somites in 4–7 animals were scored; error bars in the graphs represent the standard error. All statistical tests were Student t-tests and performed in Excel (Microsoft Office). A P value below 0.05 was considered statistically significant.

Confocal and Spinning Disc Microscopy

Whole-mount and flat-mount fluorescent embryos were photographed using a Leica DMRE and an Olympus confocal microscope. Still images were processed using Photoshop software (Adobe). For the creation of movies, the Leica Spinning Disc microscope was used and videos in Quicktime format exported. Embryos were mounted in 1.5% low melting agarose with 6% Tricane.


We thank Caroline Parkin for instruction in Leica spinning disc microscopy; Nipam Patel for antibodies; Thomas Becker and Tim Chico for transgenic lines; Phil Elks for the lyn-td-tomato construct; Steven Moore for help with screening transgenics; Raymond Teck Ho Lee and Lee May Yin for help with the DAB staining protocol and reagents; Dr. Studipto Roy's lab for antibody reagents; Tanya Whitfield, Rosemary Kim, Ashish Maurya, Anne-Cecile Burguiere, and Sarah Baxendale for helpful discussions; and Tan Swee Chuan, Claire Allen, Matt Green, Susie Surfleet, and Lisa Van Hateren for their excellent zebrafish husbandry. This work was funded by a MRC Programme Grant, by the European Union Framework 6 Network of Excellence “Cells into Organs” and by A*STAR (Agency for Science Technology and Research) Singapore. The MRC CDBG zebrafish facility is supported by an MRC Centre Grant. The authors declare no competing financial interests. C.S., M.H., and P.W.I conceived the study and designed the experiments. C.S. performed the majority of the experiments, S.E. generated the myodGFPBAC, R.J.C., and W.XG. raised the anti-myod antibody. C.S., M.H., and P.W.I. analyzed the data. C.S., and P.W.I. wrote the study.