Control of muscle regeneration in the Xenopus tadpole tail by Pax7

Following amputation, the tail of the Xenopus tadpole will regenerate over 2-3 weeks to restore a functional appendage of similar size to the original (Slack et al.,2004). Much of the tissue mass of the regenerated tail consists of striated muscle, but the origin of this muscle has been a matter of some uncertainty and controversy. In this paper, we show that an adult stem cell:the muscle satellite cell, already known to be able to replace fibres in damaged muscle, is also capable of building a new mass of tissue in an extended regenerated appendage.

In principle, three possibilities have been proposed for the origin of regenerated muscle: de-differentiation of myofibres, muscle satellite cells and side population (SP) cells. During regeneration of the limbs of urodele amphibians (newts and salamanders), it has been well documented that the striated muscle fibres can de-differentiate(Namenwirth, 1974; Lo et al., 1993; Kumar et al., 2000; Echeverri et al., 2001). In this process, the nuclei of the myofibres re-enter S-phase and the fibres break apart to become mononuclear cells. These cells then participate in the regeneration blastema, proliferate and eventually become re-differentiated,mostly as new myofibres, but also as some other tissue types. In postnatal mammals, there is no regeneration of appendages such as limbs or tails. However, striated muscle does have some ability to repair itself following tissue damage. In this process, the new fibres are generally considered to be derived from muscle satellite cells: a population of small mononuclear cells located beneath the basement membrane of the myofibres(Seale and Rudnicki, 2000). Muscle satellite cells can be considered as a type of adult stem cell as they are able both to reproduce themselves and to produce myoblasts, which can differentiate to form new myofibres. However, this idea has been challenged because in skeletal muscle, there also exists another kind of stem cell called side population (SP) cells that can be recognized by a DNA-binding Hoechst(33342) dye (Gussoni et al.,1999; Jackson et al.,1999; Seale et al.,2001). Transplanted SP cells have also been shown to be able to adopt the myogenic lineage and to participate in muscle repair.

Previous studies of Xenopus tail regeneration showed that there was no de-differentiation of the myofibres(Ryffel et al., 2003; Gargioli and Slack, 2004). The fibres near the cut surface simply degenerate and die. This suggested a different mode of muscle regeneration from that found in urodeles. It was conjectured, on the basis of the labelling patterns from different types of graft, that satellite cells might be the precursors of the striated muscle of the regenerate (Gargioli and Slack,2004). In the present paper, we show that this is the case. We show that, as in mammals, Pax7 is expressed in muscle satellite cells and that many satellite cells, but not myonuclei, are active in DNA replication. We show that a domain swap inhibitor, Pax7EnR, can antagonise the biological activity of Pax7. We then use an inducible pax7EnR transgene to deplete the tail of satellite cells and show that this reduces or inhibits the regeneration of new muscle.

Muscle satellite cells were actually first discovered in frogs, visualised by electron microscopy as mononucleated cells wedged between the basement membrane and the plasma membrane (Mauro,1961). Later, muscle satellite cells were also found in skeletal muscles of mammals and birds, and their role in muscle repair has been studied recently (Chargé and Rudnicki,2004). Upon muscle injury, quiescent satellite cells residing in the damaged myofibres are rapidly activated to re-enter the cell cycle. The upregulation of Myf5 and MyoD, two myogenic regulatory factors, confers on them a myoblast identity. After proliferation, most of the cells differentiate and fuse to generate new myofibres(Cornelison and Wold, 1997; Cooper et al., 1999). A minority of satellite cells are retained in the sub-laminar location to replenish the pool of cells for subsequent muscle repair(Chargé and Rudnicki,2004).

An important gene expressed in satellite cells is pax7(Seale et al., 2000; Halevy et al., 2004). It is a member of the Pax (paired box) family of transcription factors, which play important roles in cell fate, early embryonic patterning and organogenesis(Mansouri et al., 1996; Mansouri et al., 1999; Ziman et al., 2001a; Lamey et al., 2004). Pax7 has three conserved protein domains, a DNA-binding domain called the paired domain, a paired-type homeodomain and an octapeptide(Jostes et al., 1990). Early studies of the pax7 gene mainly focused on its biological function in the central nervous system because of its abundant expression there(Mansouri et al., 1996; Kawakami et al., 1997; Ziman et al., 2001a). Its activity in muscle satellite cells has been established more recently(Seale et al., 2000). It has been found that Pax7 is only expressed in quiescent and newly activated muscle satellite cells. Upon myogenic differentiation, it is rapidly downregulated. The Pax7^-/- mice contain reduced number of muscle satellite cells and they are progressively lost during postnatal growth(Oustanina et al., 2004; Kuang et al., 2006; Relaix et al., 2006). These findings revealed an important feature of satellite cells in their capacity for self renewal and the role of Pax7 in this context(Zammit et al., 2004; Collins et al., 2005; Montarras et al., 2005).

Our results presented here show that the satellite cells are dividing during normal tail growth and, most importantly, that they are responsible for forming the muscle masses of the regenerated tail. This means that the cellular processes of regeneration in the anuran tadpole is much more akin to the tissue repair events of mammals than to the de-differentiation/re-differentiation process found in urodeles.

Xenopus laevis embryos were obtained by in vitro fertilization and staged according to the Nieuwkoop and Faber (NF) tables(Nieuwkoop and Faber, 1967). Embryos were dejelled with 2% cysteine (Sigma) (pH 7.8) and then cultured in 0.1×NAM. From stage 46, they were transferred to recirculating aquarium and fed on tadpole diet (Blades Biological, Redbridge, UK).

Xenopus pax7 (Xpax7) gene was isolated from cDNA of stage 25 embryos by RT-PCR. The primers were designed according to the sequence submitted by the laboratory of Dr Richard Harland (NCBI:AF725267): sense primer, 5′-CAA CTT GTG AGG ACT CTT CTA GGC T-3′; antisense primer,5′-TTT TCA CCA AGT GGC AGA CAT-3′. The pax7 DNA fragment obtained by RT-PCR was ligated into pGEM-T easy vector (Promega) to generate pax7-pGEMT, which was then sequenced. The full-length Xenopus pax7 gene was excised from pax7-pGEMT plasmid by SpeI and ApaI (Promega), and cloned into SpeI and ApaI sites of pSL1180 vector (Amershan). This plasmid is called pax7-pSL1180 and used for subsequent cloning.

To generate the construct for RNA injection, the full pax7sequence was isolated from pax7-pSL1180 with HindIII and EcoRI (Promega) and cloned into the same sites of pcDNA3 vector(Invitrogen). Capped Pax7 RNA was then transcribed in vitro with T7 RNA polymerase (Promega) after linearization with Tth111I(Promega).

The dominant-negative form of pax7 was made by the following steps. The N-terminal region of the Xenopus pax7 gene (1-245 amino acids) was excised from pax7-pSL1180 with HindIII and SmaI (Promega), and was then cloned into HindIII and ClaI (blunt filled) sites of ENR-N-pCS2+ vector (kind gift of Dan Kessler). This pax7EnR plasmid is designed to produce a fusion protein, which includes the pax7 DNA-binding domain and Drosophila engrailed repressor domain. For RNA injection, the sequence of pax7EnR was cloned into pcDNA3 with HindIII and XbaI (Promega) to generate pax7EnR-pcDNA3. This plasmid was then linearized with SmaI and transcribed in vitro with T7 RNA polymerase.

A pax7EnR plasmid suitable for transgenesis was made by excising the sequence of pax7EnR from pax7EnR-pcDNA3 with HindIII and XbaI (blunt filled), and then cloning into HindIII and SmaI sites of HGEM. The pax7EnR-HGEM was linearized with XmnI before use in transgenesis.

Transgenic Xenopus laevis tadpoles were made as previously described (Amaya and Kroll,1999), except for the omission of restriction enzyme from the reaction. The transgenics were sorted out by GFP expression in the lens at stage 42.

For Pax7 overexpression, 500 pg pax7 and 80 pg gfp mRNA were injected into one side of dorsal animal hemisphere at the four-cell stage. For pax7EnR injection, 200 pg pax7EnR,together with 80 pg gfp mRNA, was injected into one side of dorsal animal hemisphere at the four-cell stage. For rescue experiments, 500 pg pax7, 200 pg pax7EnR and 80 pg gfp mRNA were co-injected.

The tadpoles at stage 49 were anaesthetized in 0.02% MS222, and injected with 2 μl of the thymidine analog 5-bromo-2′-deoxyuridine (BrdU)labelling reagent from the Cell Proliferation Kit (Amersham). The injection was performed 24 hours before fixation.

Whole-mount in situ hybridization was performed according to the standard protocol (Harland, 1991). The antisense pax7 probe was designed to hybridize specifically with the C-terminal region of Pax7. The pax7-pGEMT was linearized with SmaI and transcribed with T7 RNA polymerase. For morphology or immunohistochemistry, embryos or tadpoles were fixed in Zamboni's fixative (40 mM NaH₂PO₄, 120 mM Na₂HPO₄, 2% PFA, 0.1% saturated picric acid) overnight at 4°C. Myofibres were stained with 12/101 monoclonal antibody(Kintner and Brockes, 1984) at 1:100 dilution of medium. The secondary antibody was horse anti-mouse IgG whole molecular alkaline phosphatase-conjugated (Vector Labs) at 1:1000 dilution. The colour was developed with the BM purple reagent (Roche).

Activated satellite cells were stained with MyoD monoclonal antibody (kind gift of John Gurdon) at 1:4 dilution. The basement membrane of myofibres was stained with laminin (Sigma) at 1:100 dilution. Their secondary antibodies were Texas Red-conjugated anti-mouse IgG and fluorescein-conjugated anti-rabbit IgG (Vector Labs), respectively. The monoclonal PCNA antibody(Dako Cytomation) was used at 1:500 dilution. Muscle satellite cells were stained with anti-Pax7 monoclonal antibody (Developmental Studies Hybridoma Bank, University of Iowa) at 1:300 dilution. Antigen retrieval was performed for Pax7 antibody staining by boiling the slides in citrate buffer (Vector Labs) for 5 minutes in the microwave oven. For vibratome sections, the secondary antibody was horse biotinylated anti-mouse IgG (Vector Labs) at 1:500 dilution. Then the sections were incubated in ABC reagent (Dako Cytomation), followed by colour development using a DAB kit (Vector Labs). For paraffin sections, Pax7 antibody staining was either used with the ABC method as described above, or with a fluorescence method. In the latter case, the secondary antibody is Texas Red-conjugated anti-mouse IgG (Vector Labs). The slides were counterstained in 0.5% Methyl Green solution (Fluka) and mounted in Depex (BDH) or Gel mounting medium (Biomeda) before observation under the microscope.

The tadpoles at stage 49 were fixed in Zamboni's fixative containing 0.5%glutaraldehyde overnight at 4°C, washed in PBS and embedded in 5% low melting agarose (Sigma). Transverse sections (100 μm) were cut with a Leica vibratome and then immunostained as described above. Following this, the vibratome sections were washed in PBS, and post-fixed in osmium tetroxide (1%w/v) in 0.1 M sodium cacodylate buffer (pH 7.6) for 1 hour. The post-fixed sections were dehydrated in a series of ethanols and embedded in Epon resin(TAAB). The polymerized resin blocks were trimmed and transversely sectioned with a Reichert Ultracut-E ultramicrotome (Leica, Wein, Austria). Sections(100 nm) were collected onto copper grids, some being stained with uranyl acetate and lead citrate, and examined under a JEOL JEM1200EX transmission electron microscope (JEOL, Tokyo, Japan). Owing to the narrow spacing between each section, for the morphometric studies a nucleus within myofibers was counted only once in each resin block.

The Xenopus laevis tadpoles were anaesthetized in 1/3000 MS222 and kept in the anaesthetic solution during the operation. For the heat-shock experiments, tadpoles were placed into warmed water at 34°C for 30 minutes 3 hours before tail amputation and again each day during tail regeneration. Initially, the distal 50% length of the tail was removed. Twelve or 14 days later, when the tail regenerated to its full length, the distal 75% of the regenerated tail was amputated again. The tadpoles were allowed to recover from anaesthesia in tap water before returning to aquarium tanks.

Muscle satellite cell counting was based on the Pax7 antibody labelling on tissue cross-sections of regenerated tails. Cells were counted on a series of sections comprising the 50 μm length of the tail that is nearest to the first amputation site or near to the second tail amputation site, as the regenerated muscle near the tail tip is too small for any quantification. Ten tails of similar size in each group were examined and statistical analysis was performed using Student's t-test.

Ten regenerating tails of each experimental group were collected for RNA isolation. Total RNA was prepared using Trizol and reverse transcribed into cDNA with Superscript III system (Invitrogen). Primers used for pax7EnR are: sense, 5′-GCTCTGTCCCCTCAGGTTTAGT-3′;antisense, 5′-GGTGGTGTGCGTCTGATTGTG-3′. Primers for pax7are: sense, 5′-TCAATAATGGTCTCTCCCCGC-3′; antisense,5′-TTGCCAGGTAATCAACAGCGG-3′. Primers for pax6 are: sense,5′-GCA ACC TGG CGA GCG ATA AGC-3′; antisense, 5′-CCT GCC GTC TCT GGT TCC GTA GTT-3′.

For apoptosis detection, paraffin tissue sections were prepared as described above. The TUNEL assay was applied on 7 μm section with the in situ cell death detection kit (Roche) as instructed, followed by colour development with Fast Red (tablets from Sigma). For detection of apoptosis in Pax7-expressing cells, Pax7 immunohistochemistry was carried out first,developed with DAB, and then the apoptosis labelling reaction performed and visualized with a GFP filter set. For detection of apoptosis in MyoD-expressing cells, MyoD immunohistochemistry was performed first, and then followed by the apoptosis labelling reaction. Apoptotic cells were counted in three series of sagittal sections of 3 days regenerated tails.

GFP was observed in live tadpoles under anaesthesia as described above,using a Leica Fluo III fluorescent dissecting microscope with a GFP2 filter set. Stained sections were visualized with a Leica DMRB microscope. Images were captured using a SPOT RT camera (Diagnostic instruments) and processed with Photoshop software (Adobe).

We isolated the Xenopus pax7 (Xpax7) gene from cDNA of stage 25 Xenopus laevis embryos by RT-PCR according to the sequence submitted by Harland's laboratory (NCBI: AY 725267). The expression pattern was determined by in situ hybridization with a pax7-specific antisense probe. Pax7 transcripts are first detectable at stage13 as two bilateral stripes in the neural plate(Fig. 1A). At stage 16, pax7 is expressed with the highest intensity in the anterior neural fold (Fig. 1B). Transverse section of the anterior region shows that the expression domain is in the sensorial layer of the neural ectoderm and the dorsal lateral somitic mesoderm(inset in Fig. 1B). As the neural folds fuse, Pax7 transcripts become restricted to the future brain (Fig. 1C). At tailbud stage, pax7 expression in the head extends from the forebrain to the hindbrain and, by stage 35, to the whole spinal cord. The transcripts in the spinal cord are all concentrated on the dorsal side(Fig. 1G,J,L). From the early tailbud stage, two semi-circles of pax7 expression in head mesenchyme cells are evident surrounding the eye (Fig. 1D-F). Cross and horizontal sections show that it is expressed in the mesenchyme cells in the dorsal and anterior region of the developing eye(Fig. 1H,M,N). Moreover, pax7 transcripts are also detected in the pituitary anlage(Fig. 1F, inset; Fig. 1I) and in the pronephric anlage (Fig. 1E,J). A horizontal section through the pronephric anlage showed two strips of pax7 transcripts are present in the posterior region(Fig. 1J,K).

The segmented expression pattern of pax7 is obvious in anterior dorsal lateral somites starting from stage 19 (inset in Fig. 1C). A series of faint chevrons of pax7 expression appears in the trunk region of early tailbud stage embryo (Fig. 1D),and intensify as embryo develops (Fig. 1D-F). Parasagittal section shows that pax7 is expressed in scattered cells at the anterior and posterior edges of individual somites(Fig. 1N) and a slight random scatter of cells in the undifferentiated presomitic mesoderm of the tail bud(Fig. 1O).

At later tadpole stages, in situ probes are unable to penetrate the tadpole skin. Therefore, we used anti-Pax7 monoclonal antibody to detect expression of Pax7 protein on paraffin sections. Consistent with our in situ results, the immunohistochemistry study shows that high levels of Pax7 expression are present in forebrain (data not shown), midbrain(Fig. 2A), hindbrain(Fig. 2B) and dorsal spinal cord (Fig. 2C). Moreover, Pax7 is expressed in the eye muscle and pituitary gland(Fig. 2A,D). The cells with Pax7 positive signals in the tadpole trunk and tail muscle are flat,peripheral and squeezed beneath the basement membrane, as revealed by laminin antibody staining (Fig. 2E). On the basis of their position, these cells are mostly likely to be muscle satellite cells.

In summary, the in situ hybridization and immunohistochemistry studies in Xenopus laevis demonstrate that Xenopus pax7, like its homologues in other species, is strongly expressed in the central nervous system and muscle. This expression pattern continues to adult stage, as detected by RT-PCR (data not shown). Apart from that, Xenopus Pax7 is also expressed in pituitary gland (Fig. 1I and Fig. 2A) and adult testis (data not shown).

The expression study of Pax7 suggested that the cells squeezed between myofibres with positive Pax7 immunostaining signals may be Xenopusmuscle satellite cells. To investigate this, we performed immunoelectron microscopy (IEM). Muscle satellite cells are located in dentations between the basement membrane and plasma membrane of myofibres, while the myonucleus lies within the plasma membrane of myofibres(Mauro, 1961). These morphological criteria enable us to distinguish between the nuclei of satellite cells and the myonuclei by transmission electron microscopy.

Fig. 3A,B show muscle satellite cells that are squeezed between the basal lamina (indicated by arrows in Fig. 3C,D) and the plasma membrane. The positive Pax7 antibody signals visualized by black electron dense dots concentrate in the nuclei of satellite cells. By contrast,no Pax7 antibody activity is detected in the myonuclei(Fig. 3E). Statistically, among 25 muscle satellite cells from five individual tails, 22 cells were found to be positive for Pax7 (88%, n=25); by contrast, all the myonuclei identified are negative for Pax7 (0%, n=41; Table 1). As negative control,sections stained without Pax7 antibody do not show any electron dense granules in the nucleus (data not shown). Thus, we are able to take Pax7 as a reliable muscle satellite cell marker in the tail of Xenopus tadpole.

Occasionally the Pax7 antibody-positive signals are also found in some other cell types in tadpole tail, outside of the muscle fibres. These cells are identified based on their characteristic structures as described in the Boston University ultrastructure website(http://www.bu.edu/histology/m/t_electr.htm). The Pax7 antibody label is present in the nucleus of Schwann cells whose cell membrane forms myelin coils around the axon(Fig. 3F). A lymphocyte that is characterized by its nucleus filling virtually the entire volume of the cell is also found to be positive (Fig. 3G). Moreover, a faint Pax7-positive signal is located in the nucleus of a fibroblast surrounded with collagen fibrils(Fig. 3H). This shows that not every Pax7-positive cell in the tail is necessarily a satellite cell.

Previously, we observed many BrdU-labelled cells in the muscle of the growing tadpoles (Gargioli and Slack,2004). To find whether these are, in fact, satellite cells, we injected BrdU into stage 49 tadpoles and fixed the tadpole tails 1 day after injection. They were processed for immuno-electron microscopy, using antibody to BrdU. We found that 60% of satellite cell nuclei are labelled with anti-BrdU (Fig. 3I, Table 1). By contrast, almost all the myonuclei examined from four tails are negative for BrdU(Fig. 3J, Table 1). Only one myonucleus was detected with a faint BrdU antibody signal. This is perhaps due to DNA repair synthesis, or to recent incorporation of a satellite cell in the fibre. As satellite cells are multiplying and myonuclei are not, it seems highly likely that the satellite cells are the source of the new fibres, or contribute to fibre expansion during growth, as they do in mammals(Moss and Leblond, 1971).

Tails of Xenopus tadpole will regenerate fully after amputation(Slack et al., 2004). To test whether Pax7 plays a role in this process, we performed RT-PCR assays with regenerating tails 3 days post amputation. As shown in Fig. 4A, the pax7 mRNA level is increased in the regenerating tails, compared with that in tails without amputation. We consider that this is due to an increase in number of Pax7-positive cells. It is not possible to make a precise quantitative comparison of cell numbers because the structure of mature tail muscle and regeneration bud are so different. However, at this stage, there is a considerable amount of cell proliferation in the regeneration bud, indicated by PCNA antibody staining (Fig. 4B), and many Pax7-positive cells are found free of the muscle lying close to the amputation level in the blastema(Fig. 4C), which is the undifferentiated region of the regeneration bud(Gargioli and Slack, 2004).

To further investigate the function of Pax7 in muscle satellite cells during Xenopus tail regeneration, we generated a domain-swapped construct pax7EnR, in which the C-terminal region of XenopusPax7 was replaced with the transcriptional repression domain of Drosophila Engrailed (Han and Manley, 1993).

This method has been used many times to generate transcription factor domain swaps, but we felt it important to confirm that it really had the predicted biological activity: namely that it can inhibit normal Pax7 function. To do this, we overexpressed Pax7EnR in the brain region, which is rich in endogenous expression of Pax7. We injected 200 pg pax7EnR,together with 80 pg gfp mRNA, into the left side of the dorsal animal hemisphere of four-cell stage embryos. The injected neurulae show defects in the anterior neural fold on the left (Fig. 5A) and later on in the developing left eye(Fig. 5C and Table 2). These defects become obvious at advanced stages. The left eye is absent or smaller, while the right eye has fully developed (Fig. 5C,D,F). The GFP fluorescence on the injected side indicates that it still has a small lens underneath the epidermis (inset in Fig. 5C). To test the specificity of Pax7EnR, we co-injected the pax7EnR RNA together with 500 pg wild-type pax7 RNA. As shown in Fig. 5B,E and Table 2, the pax7 RNA is able to rescue these eyes back to approximately normal size.

The above results suggested the involvement of Pax7 in eye development in Xenopus laevis. As pax7EnR inhibits eye development, it is possible that pax7 itself can promote eye development on overexpression. Indeed this is the case. When 500 pg pax7 and 80 pg gfp mRNA were co-injected into one side of the dorsal animal hemisphere at the four-cell stage, some tadpoles developed an ectopic eye at the injected site (Fig. 5G-I, Table 2). Although only a small proportion of the injected tadpoles developed an ectopic eye, when found it is very well formed, containing a retinal pigmented epithelial layer, outer nuclear layer, inner nuclear layer, ganglion cell layer and lens in a spatial arrangement identical to the endogenous eyes. Detailed examination of these tadpoles shows that the ectopic eye is often accompanied by a tube protruding from ventral midbrain (Fig. 5J). No ectopic eyes were seen in tadpoles resulting from embryos injected into the same region with gfp mRNA alone.

We are not claiming that Pax7 has a normal function in Xenopus eye development, but the phenomena described here provide us with a useful bioassay. The experiments show that Pax7 promotes eye development, Pax7EnR represses it and sufficient Pax7 can restore eye development in the presence of Pax7EnR. This proves that Pax7EnR is able to inhibit the function of Pax7 and we can therefore use Pax7EnR as a dominant-negative form of Pax7 to investigate the function of Pax7 during muscle regeneration.

As the RNA experiments showed that overexpression of Pax7EnR disturbs the early embryonic development of Xenopus laevis, we generated transgenic tadpoles in which the pax7EnR gene is driven by a heat-shock promoter in response to a 34°C warm pulse(Beck et al., 2003). This means that the gene is inactive through development and activated only when the temperature is raised. We performed RT-PCR analysis on four groups of tadpole tails. They are: (1) wild-type tails regenerated for 7 days without heat shock; (2) pax7EnR transgenic tails regenerated for 7 days without heat shock; (3) wild-type tails regenerated for 7 days with one heat shock each day after amputation; and (4) pax7EnR transgenic tails regenerated for 7 days with one heat shock each day after amputation. As shown in Fig. 6A, pax7EnR is induced only in the case of pax7EnR transgenic tails with heat shock,and no leaking expression of pax7EnR is detected in tails without heat shock. These results also show that the endogenous pax7 mRNA is reduced in the regenerating tails with Pax7EnR expression. By contrast, there is no effect on pax6 mRNA, which is also present in the regenerating tail (Fig. 6A).

The reduction of pax7 level might arise from repression of transcription or from a reduction in the number of cells expressing pax7. To investigate these possibilities we compared the number of muscle satellite cells in wild-type and pax7EnR regenerates. Both wild-type and pax7EnR transgenic tadpoles were given the first heat shock three hours before tail amputation. Then the distal 50% of the tail was cut off and sectioned to generate the samples: wild type d0 and Pax7EnR d0. During tail regeneration, a heat shock was administered once every day. Seven days later, the regenerating tails were fixed as the samples wild type d7 and Pax7EnR d7. We counted the number of muscle satellite cells in the 50 μm tail region that is nearest to the amputation surface. Based on Pax7 antibody staining of ten tails of each group, we found the number of satellite cells in wild type d0 and Pax7EnR d0 is similar(Fig. 6B,D,F). After 7 days regeneration, the number of satellite cells falls a little in the wild-type regenerates but falls substantially in the pax7EnR regenerates(Fig. 6C,E,F). After 14 days regeneration, the number of satellite cells in the pax7EnR tadpoles was still about half the number in wild type (Fig. 6F). The 14-day sections are from a more distal level of the regenerate, corresponding to the level of second amputation (see below), and so the absolute cell numbers per section are much reduced owing to the smaller area. For all the specimens, the level of Pax7 in the satellite cells appears approximately similar, indicating that the reduction of pax7 mRNA level by Pax7EnR arises from a reduction in the number of satellite cells.

The reduction of the number of muscle satellite cells in pax7EnRtransgenic tails might arise from a reduction in the normal proliferation or from promotion of cell death of satellite cells during muscle regeneration. We examined cell proliferation by PCNA antibody staining in 3 day regenerating tails of wild-type and Pax7EnR transgenic tadpoles and this showed no obvious difference (data not shown).

Then, we performed TUNEL staining to test for apoptosis in the regenerating tails. In the case of pax7EnR transgenic tails, most apoptotic cells are present in the blastema region, the same location that muscle satellite cells are found freed from the myofibres(Fig. 7A-C). In the wild-type tails, the number of apoptotic cells is smaller than that in the pax7EnR transgenic tails (Fig. 7E). Moreover, in many clusters of cells, possibly indicating small clones, those with low level of Pax7 expression are often entering apoptosis (red arrows in Fig. 7D), although those with high level of Pax7 expression are not. Given that the heat shock promoter is expected to have a similar activity in all cells, it may be that the cells with lower content of pax7 are more vulnerable to the effect of the pax7EnR. Quantification of cell death shows that cell death in mesenchyme and blastema of pax7EnR transgenic tails is more severe than that in wild-type tails, whereas cell death in the regenerating spinal cord and notochord is unaffected(Fig. 7E).

To further investigate apoptosis of muscle satellite cells during regeneration, we used the MyoD antibody(Hopwood et al., 1992) to label the activated satellite cells(Zammit et al., 2004) and then performed a TUNEL assay. The cell quantification result demonstrates that over 60% of MyoD-labelled cells in the mesenchyme and blastema region of pax7EnR transgenic tails are undergoing apoptosis, whereas only 25%of MyoD-positive cells undergo apoptosis in wild-type tails. This result confirms a protective role of Pax7 in cell survival, something that has also been noticed in mammalian studies (Kuang et al., 2006; Relaix et al.,2006). In other words, Pax7EnR reduces the number of satellite cells not by reducing their proliferation rate but by increasing their rate of cell death.

The above results show that the number of satellite cells in regenerating tails is sharply decreased when pax7EnR is activated by heat shock. So we predict that less muscle would form in the regenerating pax7EnRtransgenic tails compared with wild type, if muscle satellite cells are the sole source of muscle regeneration. However, after 14 days heat-shock treatment, when muscle in wild-type tail has terminally differentiated, we did not find a significant reduction of muscle in the transgenic pax7EnRtails. As shown in Fig. 8B,C,the staining strength of 12/101, a marker of muscle differentiation, is similar in pax7EnR and in wild-type tails.

However, we know that the first regenerate is depleted in satellite cells and this raises the question of whether it can regenerate yet more muscle following a second amputation. To investigate this, we followed the procedure as depicted in Fig. 8A. The Pax7EnR expression is induced with a daily heat shock throughout the regeneration period. The first regenerate is allowed to grow for 14 days and then the distal 75% of the primary regenerated tail is re-amputated and allowed to regenerate for another 14 days. As shown in Fig. 6F, the number of satellite cells in heated pax7EnR transgenics is about half that in wild types at this amputation level. After the second regeneration, we performed 12/101 immunostaining on both regenerates. As shown in Fig. 8B,D,G, the muscle in both first and second regenerates of wild-type tails was fully formed. By contrast,for the case of pax7EnR heat shocked tadpoles, 61% of the second tail regenerates showed substantially reduced differentiated muscle, and another 11% had no muscle at all (Fig. 8E,F, Table 3). In these individuals the spinal cord and the notochord do regenerate fully in the second regenerate, so even though Pax7 is expressed in some spinal cord cells it is presumably not necessary for regeneration(Fig. 8H). This experiment clearly shows that the regeneration of muscle depends on the presence of a population of satellite cells in the amputated tail.

Our in situ hybridization and immunohistochemistry studies suggest that pax7 expression is highly conserved in vertebrates. As in mouse and chick, pax7 transcripts in Xenopus are mainly found in the central nervous system, the embryonic muscle and in muscle satellite cells. pax7 expression in embryonic brain and spinal cord shows high regional specificity, as indicated by its dorsal expression in midbrain and the whole spinal cord, with ventral expression in a subpopulation of neural cells in the hindbrain (Figs 1, 2). This pax7expression pattern is similar to that in mouse and chick(Jostes et al., 1990; Kawakami et al., 1997). During embryonic myogenesis, the chevron pattern observed in Xenopus somites is very similar to the Pax7 expression pattern in the E12.5 mouse myotome (Horst et al., 2006). Expression of pax7 is also observed in a subset of muscle cells, with the position and appearance of satellite cells(Fig. 2). This similar expression pattern suggests a conserved involvement of pax7 in embryonic myogenesis and specification of muscle progenitor cells(Halevy et al., 2004; Seale et al., 2004; Gros et al., 2005; Kassar-Duchossoy et al., 2005; Relaix et al., 2005).

However, there exist minor differences of pax7 expression profile between species. pax7 expression is found in the melanocyte lineage in chick and quail, but not in the melanocyte lineage in Xenopustadpoles, mouse and rat (Lacosta et al.,2005). Expression of pax7 in head mesenchyme is also different in Xenopus and mouse. As shown in Fig. 1H,M,N, pax7transcripts appear earlier and stronger than that in the segmented somites,while in mouse, pax7 expression in the head mesenchyme is not detectable until E11.5 (Horst et al.,2006).

The presence of pax7 transcripts in the pronephros of the Xenopus embryo is interesting and its function during organogenesis is currently under investigation. Our preliminary result suggests that pax7 also plays an important role in the growing pronephros, probably synergizing with Pax2 and Pax8 by acting as an anti-apoptotic factor (Y.C., G.L. and J.M.W.S., unpublished) (Relaix et al.,2006).

Our testing of the function of pax7 and pax7EnRconstructs by injecting them into the dorsal side of four-cell stage embryos resulted in a very surprising phenotype: the induction of a well-formed extra eye. This phenotype is reproducible, despite the low frequency of occurrence(10% of cases). In an elegant manner, this overexpression phenotype is the`opposite' of the Pax7EnR phenotype, which involves reduction of the normal eye. The eye-reduction phenotypes of pax7EnR can be rescued by co-injection of pax7. An overexpression phenotype does not, of course, prove that the gene in question is involved in the normal development of the affected structure. In this case, it is probably not directly involved as Pax7 is not normally expressed in the eye itself. However, it is expressed in the region immediately anterior to the eye and so the overexpression effects may indicate an indirect function for Pax7 in eye development involving signals from the surrounding tissues.

Consistent with this, pax7 is expressed in chick optic tectum and mouse superior colliculus, one of the major visual targets in brain(Nomura et al., 1998; Ziman et al., 2001b; Thomas et al., 2004; Thompson et al., 2004). When misexpressed, Pax7 can induce ectopic formation of superior collicular tissue,with characteristic laminae innervated by retinal ganglion cell axons(Matsunaga et al., 2001; Thompson et al., 2004). Interestingly, among the tadpoles with an extra eye, we very often observed formation of an extra tube-like structure protruding from the midbrain(asterisks in Fig. 5J). Detection of a series of transcription factors related to eye field specification show that some of the transcription factors, such as Rx1 and Tll, are downregulated by pax7EnR injection (data not shown). Thus,these findings do support a role of Pax7 in the development and maturation of the vertebrate visual system, even if it is indirect. It is also possible that the overexpression experiments are interfering in some way with pax6,well known as a master controlling gene in eye development(Gehring and Ikeo, 1999). However, Fig. 6A does not indicate any effect on pax6, at least at the level of mRNA content.

In this report, we show that Pax7 is a reliable satellite cell marker by combining nuclear Pax7 expression with morphological observations of satellite cells by electron microscopy. We need a good molecular marker, because it is impossible reliably to identify a satellite cell under light microscope by morphology alone. As shown in Fig. 3 and our statistical analysis, the immuno-electron microscope result proves that Pax7 is a very good satellite cell marker. The observation that 88% of morphologically identified satellite cells express Pax7 would allow us to pick up most of satellite cells in a pool of cell population. The minority of negative cells are presumably those that have commenced differentiation to myoblasts. Despite the occasional expression of Pax7 in other cell types, expression of Pax7 in myonuclei is never found, providing a very clear discrimination between the satellite cells and myofibres themselves. The presence of Pax7 in some Schwann cells may possibly be associated with their ability to re-enter the cell cycle for peripheral nerve regeneration (Hall, 2005).

Several other molecules, including neural cell adhesion molecule (NCAM),Met and M-cadherin, have been used to identify satellite cells in various experimental approaches (for a review, see Holterman and Rudnicki, 2005),so we also checked the expression of these candidates in Xenopus. In our hands, the efficiency of these markers to recognize satellite cells in Xenopus tadpoles is not as good as for Pax7. We observed significantly more Pax7-positive cells in tadpole muscle tissues than cells positive for other markers (data not shown) and so consider Pax7 as the most reliable satellite cell marker in Xenopus.

Our results support our previous conjecture that satellite cells residing in the tail muscle fibres are the origin of muscle regeneration in Xenopus tadpoles (Gargioli and Slack, 2004). First, additional expression of pax7 in the regenerating tail suggests that satellite cells are activated to divide after tail amputation. In regenerating tails 3 days post amputation, a large number of Pax7-positive cells appears in the region where muscle fibres are undergoing degeneration, and many Pax7-positive cells are visible in the growing blastema (Fig. 4). Second, our experiments show that inhibition of Pax7 action by Pax7EnR causes apoptosis of muscle satellite cells, and reduces the number of satellite cells in regenerating tails. The tails that are depleted of satellite cells do still regenerate, but most regenerates contain little or no muscle(Fig. 8; Table 3). Thus, our result demonstrates that muscle regeneration in Xenopus tails is similar to muscle repair in mammals, where it has been generally accepted that satellite cells are adult muscle precursor cells and contribute to repair of damaged muscle (Wagers and Conboy,2005). The actual requirement for Pax7 function must be in the survival of the proliferating satellite cells rather than in their differentiation into muscle, as the ability to form a first regenerate is not significantly impaired. This is consistent with the observed loss of Pax7 expression on differentiation (Seale and Rudnicki, 2000).

Recent studies have demonstrated that Pax7 has an important role in cell survival of adult muscle satellite cells(Kuang et al., 2006; Relaix et al., 2006). Using activated caspase 3 as a marker, Relaix et al. showed that apoptosis occurs in activated satellite cells in the skeletal muscle of the pax7 mutant(Relaix et al., 2006). Consistent with this, our detection of cell death with TUNEL assay suggest that cells with low level of Pax7 tend to undergo apoptosis, while those with high level of Pax7 expression can be protected from dying. In the regenerating tails of pax7EnR tadpoles, the number of cells undergoing apoptosis is increased (Fig. 7) and this ultimately results in the failure of regeneration of muscles in pax7EnR transgenic tails.

It was found that Pax3, which is expressed both in quiescent and activated muscle satellite cells in mammals, has partially redundant function with Pax7 in satellite cells (Relaix et al.,2005; Relaix et al.,2006). As we do not obtain 100% suppression of muscle regeneration with pax7EnR, it is possible that there is a redundant component present such as Pax3, but it is also possible that we do not obtain 100%inhibition of pax7 function.

Our experiments do not support the view that myofibres regenerate from resident or circulating stem cells of the side population (SP) type(Chargé and Rudnicki,2004). As discussed above, these stem cells should be available to reconstitute the muscle fibres in pax7EnR tadpoles but do not do so. Our results also differ from the previous findings on limb regeneration in newts and axolotl, in which it has been well documented that de-differentiation and re-differentiation of multinucleate muscle fibres occurs (Lo et al., 1993; Kumar et al., 2000; Echeverri et al., 2001; Echeverri and Tanaka, 2002). We see no particular reason to doubt the results from the urodele species but consider that the mode of Xenopus tail muscle regeneration to be much closer to the regeneration of damaged muscle in mammals. Interestingly, a recent paper on newt regeneration has also highlighted a role for satellite cells (Morrison et al.,2006).

Together, our experiments support the idea that Pax7 is required for the survival of the proliferating satellite cells in muscle and hence the maintenance of the satellite cell pool during regeneration(Kuang et al., 2006; Relaix et al., 2006). This satellite cell pool is the origin of the regenerated muscle in the Xenopus tadpole tail.

We thank Dr Ian Jones and Mrs Ursula Potter for technical advice on immunoelectron microscopy. We also thank Drs John Gurdon and Henrietta Standley for giving us the monoclonal antibody specific for Xenopus laevis MyoD. This work was supported by a Wellcome Trust Programme grant to J.M.W.S.