Muscle formation in regenerating Xenopus froglet limb


  • Akira Satoh,

    1. Department of Developmental Biology and Neurosciences, Graduate School of Life Sciences, Tohoku University, Aobayama Aoba-ku, Sendai, Japan
    Search for more papers by this author
  • Hiroyuki Ide,

    1. Department of Developmental Biology and Neurosciences, Graduate School of Life Sciences, Tohoku University, Aobayama Aoba-ku, Sendai, Japan
    Search for more papers by this author
  • Koji Tamura

    Corresponding author
    1. Department of Developmental Biology and Neurosciences, Graduate School of Life Sciences, Tohoku University, Aobayama Aoba-ku, Sendai, Japan
    • Department of Developmental Biology and Neurosciences, Graduate School of Life Sciences, Tohoku University, Aobayama Aoba-ku, Sendai, 980-8578, Japan
    Search for more papers by this author


A spike, a resultant regenerate made after amputation of a Xenopus froglet limb, has no muscle tissue. This muscle-less phenotype was analyzed by molecular approaches, and the results of analysis revealed that the spike expresses no myosin heavy chain or Pax7, suggesting that neither mature muscle tissue nor satellite cells exist in the spike. The regenerating blastema in the froglet limb lacked some myogenesis-related marker genes, myoD and myf5, but allowed implanted muscle precursor cells to survive and differentiate into myofiber. Implantation of hepatocyte growth factor (HGF) -releasing cell aggregates rescued this muscle-less phenotype and induced muscle regeneration in Xenopus froglet limb regenerates. These results suggest that failure of regeneration of muscle is due to a disturbance of the early steps of myogenesis under a molecular cascade mediated by HGF/c-met. Improvement of muscle regeneration in the Xenopus adult limb that we report here for the first time will give us important insights into epimorphic tissue regeneration in amphibians and other vertebrates. Developmental Dynamics 233:337–346, 2005. © 2005 Wiley-Liss, Inc.


During limb regeneration in urodele amphibians, muscle regenerates in an epimorphic way as a component of an entire regenerate that has complete structure and morphology. In the early stage of this process, dedifferentiation of myofibers is evident (Kintner and Brockes, 1984; Lo et al., 1993; Kumar et al., 2000) and dedifferentiated cells at the amputation site are thought to mainly contribute to the muscle formation in the regenerate (Carlson, 2003). Moreover, some experiments have suggested that dedifferentiated cells derived from myofibers are transdifferentiated into other tissue types, including cartilage and connective tissue (Lo et al., 1993; Kumar et al., 2000). On the other hand, muscle in some urodele species has been described to possess potential mononucleate cells called satellite cells (Cameron et al., 1986), although the contribution of satellite cells to limb regeneration is uncertain. Epimorphic muscle regeneration in the regenerating blastema formed at an amputated urodele limb is a fascinating and remarkable phenomenon compared with the poor muscle regeneration in other vertebrates such as mammals, which barely regenerate muscle tissue and not by an epimorphic mode but by a tissue mode of regeneration, i.e., by a process in which muscle fibers are directly formed from cellular elements (mainly satellite cells) remaining after tissue injury. However, because of the long and complicated process of muscle regeneration during limb regeneration, many points in the above issues remain obscure, such as the molecular nature of de- and redifferentiation in limb muscle regeneration, the degree to which transdifferentiation of muscle-derived mononucleate cells contributes to limb regeneration, the cellular origin of regenerated muscle tissue, and the contribution of satellite cells.

The African clawed frog (Xenopus laevis), an anuran amphibian, has some interesting features of limb regeneration. Xenopus can perfectly regenerate limbs during larval stages but loses the ability as development and metamorphosis progress (Dent, 1962; Overton, 1963; Muneoka et al., 1986). In the late tadpole, the skeletal pattern of a regenerated structure becomes more hypomorphic, and, eventually, young adults (froglets) after metamorphosis merely regenerate a cartilaginous spike, a simple cone-shaped structure without any segmentation. Adult Xenopus limb regeneration is thought to be epimorphic with regrowth of undifferentiated blastema cells like urodele limb regeneration, because the Xenopus limb blastema contains proliferating cells, these cells express msx1, a key molecule for an undifferentiated state, and the blastema formation in Xenopus shows nerve dependency as in urodeles (Endo et al., 2000). Nevertheless, limb regeneration in Xenopus froglets is incomplete in two aspects: one is its morphological feature as a nonpatterned skeletal structure, and the other is limited tissue redifferentiation. A spike made on an amputated Xenopus froglet limb is known to be muscle-deficient (Dent, 1962; Kurabuchi et al., 1983; Korneluk and Liversage, 1984; Muneoka and Sassoon, 1992), whereas other components, such as cartilage, connective tissue, nerves, blood vessels, and skin, are differentiated. This muscle-less situation could serve as a good model to investigate molecular and cellular aspects of epimorphic muscle regeneration during limb regeneration.

There are some possible explanations for this muscle-deficient phenotype. One possible explanation is that dedifferentiation of myofibers does not occur or the source of muscle precursor cells is absent in Xenopus adult limbs. Of interest, a recent work by (Gargioli and Slack, 2004) has suggested that myofibers of regenerates in an amputated tail of the Xenopus tadpole arise from satellite cells, not from pre-existing myofibers. Axolotls (Carlson and Rogers, 1976) and other urodeles (Cherkasova, 1983; Schrag and Cameron, 1983; Cameron et al., 1986) possess satellite cells or postsatellite cells that are localized under or between basal laminae, associated with myofibers, although it is not clear whether these cells contribute to epimorphic muscle regeneration. Little is known about the role of satellite cells in Xenopus adult limb muscle.

In developing embryos, myoblasts are a source of limb muscle precursor cells that are derived from the migrating myotome with expression of some transcription factors, such as Lbx1, Pax3, and Pax7. Particularly for limb muscles, a c-met–positive hypaxial myotome migrating into the limb bud is attracted by a motogen, HGF (hepatocyte growth factor), produced by limb mesenchymal cells (Brand-Saberi et al., 1996; Dietrich et al., 1999; Scaal et al., 1999). Muscle differentiation of myoblasts into myotubes begins with induction of myogenic basic helix–loop–helix transcription factors, so-called myogenic determination factors, myoD and myf5, both of which are required for commitment of myogenic lineage. Regarding muscle redifferentiation during Xenopus spike formation, it is possible that the series of molecular cascade(s) during myogenesis is disrupted. It is also possible that a certain blastema environment may repress gene expressions for the differentiation of myoblasts into more mature myotubes.

In the present work, we investigated the above aspects in terms of the cause of muscle-deficient phenotype in the Xenopus spike. First, we found that neither mature muscle tissue nor Pax7-positive satellite cells exist in the spike, although muscle tissue in the froglet leg contains Pax7-positive cells. Molecular marker analyses for muscle differentiation demonstrated that muscle precursor cells are absent in the regenerating blastema. However, introduction of muscle satellite cells into the froglet blastema showed that the blastema environment does not suppress muscle differentiation. Finally, we found that implantation of HGF-releasing cell aggregates induces successful muscle regeneration in Xenopus froglet limb regenerates.


Xenopus Froglet Limb Regenerates a Muscle-Less Structure After Amputation

A spike, a hypomorphic regenerate found after amputation of a Xenopus froglet limb (Fig. 1A,B), has been reported to have little muscle tissue (Dent, 1962; Korneluk and Liversage, 1984; Muneoka and Sassoon, 1992). Histological observation (Fig. 1C) and immunostaining with MF20 (anti–myosin heavy chain [MyHC] antibody as a muscle cell marker, Fig. 1D) and anti-Pax7 (a satellite cell marker, Fig. 1E) antibodies confirmed that spike structures regenerated at the wrist level had little muscle tissue and satellite cells (26 of 30), although a small contribution of stump muscle tissue in the most proximal region of the spike structure was observed (arrowheads in Fig. 1C,D). Of 30 specimens, 4 had frail muscle tissue in the spike (Table 1). In contrast, amputation of a hindlimb bud in a stage 52 tadpole resulted in regenerates that included segmented cartilage (Fig. 1F), MF20-positive muscle cells, and Pax7-positive cells (Fig. 1G,1H; Table 1).

Figure 1.

A Xenopus froglet limb regenerates a muscle-less spike after amputation. A,B: The same individual of Xenopus froglet before and after limb regeneration. Amputation of a forelimb at wrist level (A) gives rise to a spike (B). C–H: Histological and immunohistochemical analyses of muscle formation in Xenopus limb regeneration. C: A section of the spike was stained with eosin, Mayer's hematoxylin solution, and Alcian blue. Note that no muscle tissue can be observed in the spike, except in the most proximal region (arrowheads). D: Anti–myosin heavy chain immunoreactivity (fluorescein isothiocyanate [FITC] in green) was detectable in mature muscle in the stump and the most proximal region of the spike (indicated by white arrowheads). E: Pax7-positive cells (in green) were dotted in muscle tissue of the stump, but no signal can be seen in the spike (see also Figs. 2E, 4A). F–H: Sections of a regenerate after amputation at the ankle level of a stage 52 hindlimb bud. F: A histological section that shows muscle fibers (arrowheads) and cartilage. G,H: Immunostaining with MF20 (G) and anti-Pax7 (H) antibodies (FITC in blue–green). MF20-positive mature muscle cells (G) and Pax7-positive muscle precursor cells (H) were detectable in the regenerate as well as in the proximal stump. Lines indicate the estimated amputation sites. Scale bars = 400 μm in C–H.

Table 1. Muscle Structure in Regenerates After Limb Amputationa
 Muscle + n (%)Muscle − n (%)
  • a

    All samples were amputated at the level between zeugopod and autopod.

Stage 52 hindlimb bud10 (100)0 (0)
Froglet forelimb4 (13.3)26 (86.7)
 +Satellite cell implanted4 (66.7)2 (33.3)
 +P14 cell implanted10 (71.4)4 (28.6)
 +C127 cell implanted5 (31.3)11 (68.7)
Figure 2.

Myogenic marker gene expressions in the regenerating blastema. A–F: Sections of the froglet blastema that developed for 14 days after amputation. G–L: Sections of the tadpole blastema at 5 days after amputation. A,G:Sox9 transcripts were found in the middle of the blastema. B,H: Slight myoD expression was observed only in the stump region (H) and not within the froglet blastema (B). MyoD is expressed both in the tadpole blastema and stump region (H). C,D,I,J: Faint expressions of hgf (C)and c-met (D) were detectable in the froglet blastema, and in the tadpole blastema, significant signals for both hgf (I) and c-met (J) were detected. E,F,K,L: Immunostainings for myosin heavy chain (MyHC) and Pax7 proteins. E,K: Pax7-positive cells were scattered in the proximal stump region (E,K), and the froglet blastema was Pax7-negative (E), although Pax7-positive cells were detected also in the tadpole blastema (K). F,L: Anti-MyHC immunoreactivity was located only in the stump region. Scale bars = 400 μm n A (applies to A–F), in G (applies to G–L).

Figure 4.

Froglet blastema has a permissive environment for survival and differentiation of muscle precursor cells. A,B:Xenopus limb muscle tissue in the adult frog possesses Pax7-positive cells (A, fluorescein isothiocyanate [FITC] in blue–green); a transverse section of the muscle tissue was stained with anti-laminin (FITC in green) and anti-Pax7 (rhodamine in red) antibodies, and a high magnification of a section shows that a Pax7-positive in red is located inside basal laminae visualized by anti-laminin immunoreactivity in green (B). C–K: Cells isolated from muscle tissue of froglet femora were cultured for 1 day and stained with anti-Pax7 antibody. C: In low cell density, many cultured cells were Pax7-positive (C, insert, diaminobenzidine staining in brown). D: After 10 days incubation, a lot of elongated multinucleate cells were MF20-positive. E: An overlap image of bright and fluorescent field embosses implanted cell aggregates that contain 1 day-cultured Pax7-positive cells from cmv–green fluorescent protein (GFP) transgenic frogs. F–J: After 14–21 days, samples were stained with anti-GFP (F,I) and MF20 (G,J) antibody; a brightfield view (H). F,I: GFP-positive cells were recognizable in the blastema (arrows and arrowheads). F,G,I,J: Some GFP-positive cells were MF20-positive (arrowheads), and other ones were MF20-negative (arrows). K: The 4′,6-diamidine-2-phenylidole-dihydrochloride (DAPI) staining (blue) shows that some of these double-positive cells are multinucleate (arrowheads). Scale bars = 400 μm.

To elucidate the molecular nature of the muscle-less phenotype, we examined the expression of some myogenesis-related molecules in the regenerating blastema. At 14 days after amputation at the wrist level of a froglet forelimb, sox9, an early marker for cartilage differentiation, was expressed in the middle of the blastema (Fig. 2A), suggesting that tissue differentiation had begun by this stage. On the other hand, myoD, a marker for differentiating myoblasts, was not detectable (Fig. 2B). Hgf (Fig. 2C) and its receptor c-met (Fig. 2D), which are essential factors for migration of myoblasts, were detectable in the froglet blastema, but the staining was very faint. Pax7 protein, which is expressed in migrating myoblasts during limb development, was not detected in the blastema at 14 days, although Pax7-positive cells were found in the stump muscle tissue (Fig. 2E). In addition, there were no MF20-positive cells in the blastema (Fig. 2F), suggesting that muscle differentiation had not been completed at this point. All transcripts and proteins except for MyHC were positive in the growing blastema made in the tadpole hindlimb bud at stage 52 (Fig. 2G–L).

To confirm deficiency of myogenic marker gene expressions in the froglet blastema, we performed reverse transcription-polymerase chain reaction (RT-PCR) analysis using nucleotide sequence information on myoD, myf5, hgf, c-met, msx1, sox9, and fgf10 of Xenopus laevis (Fig. 3). As expected, neither the expression of myoD nor that of myf5 was detected in the froglet blastema even by RT-PCR, although expressions of both genes were detectable in the stage 52 limb bud. Expression of hgf and c-met in the froglet blastema was detected by RT-PCR, suggesting that the slight signals for both genes detected by in situ hybridization (Fig. 2C,D) were positive ones and that hgf and c-met are not absent from the froglet blastema but present there at a low level. Other markers examined, such as msx1 (for undifferentiated distal mesenchyme), sox9 (for differentiating chondrocytes), and fgf10 (for growth activity), were all positive both in the froglet blastema and tadpole limb bud.

Figure 3.

Reverse transcription-polymerase chain reaction analysis of stage 52 limb buds and froglet blastemas. Names of the transcripts are indicated to the left of the figure.

Implanted Muscle Satellite Cells Can Differentiate Into Muscle Cells in the Froglet Blastema

The above-described findings suggesting that there is no myoD/myf5-positive muscle precursor cells in the regenerating blastema raise some reasonable explanations for the muscle-less phenotype of the spike: absence of muscle precursor cells in the blastema, absence of signals inducing myogenesis, or presence of an environment that represses myogenesis. To determine whether the froglet blastema has a repressive environment for muscle differentiation, muscle satellite cells were implanted into the regenerating froglet blastema. First, we examined whether muscle tissue in the Xenopus adult limb includes muscle satellite cells. Many Pax7-positive cells were observed in limb muscle tissue (Fig. 4A), and they resided outside the sarcolemma but inside the basal laminae (detected by an antibody against laminin) of individual myofibers (Fig. 4B). Cells isolated from muscle tissue in the froglet femur were cultured for 1 day and stained with anti-Pax7 antibody. To identify cultured cells when grafted into the froglet blastema, we used cmv–green fluorescent protein (GFP) transgenic Xenopus froglets for the cell culture. The cultured cells included many Pax7-positive cells (Fig. 4C). Few MF20-positive cells were detectable at this point (data not shown, Table 2), but many MF20-positive multinucleate cells were seen after 10 days (Fig. 4D; Table 2). These results indicate that 1-day cultured cells for aggregates contained many muscle precursor cells that have the ability of muscle differentiation. Cell aggregates prepared from 1-day cultured cells were implanted into the froglet blastema 10–14 days after amputation (Fig. 4E; grafted cells were visualized by self-fluorescence of GFP). After 14–21 days, the blastemas were fixed, sectioned (Fig. 4H), and triple-stained simultaneously with anti-GFP antibody (Fig. 4F,I), MF20 antibody (Fig. 4G,J), and 4′,6-diamidine-2-phenylidole-dihydrochloride (DAPI; in blue at Fig. 4K). GFP-positive implanted cells were recognizable in the blastema (arrowheads in Fig. 4F,I), suggesting that some of implanted cells were able to survive in the environment in the froglet blastema. Some GFP-positive cells (arrows in Fig. 4F,I) were MF20-negative (arrows in Fig. 4G,J), and other GFP-positive cells were MF20-positive (arrowheads in Fig. 4F,G,I,J). Moreover, some of these double-positive cells were multinucleated; arrowheads in Fig. 4K show DAPI-stained nuclei (in blue) overlapping continuous staining by MF20 (in red), suggesting that the implanted cells were able to differentiate into myofibers that express MyHC. It is likely that the environment in the froglet blastema serves as a myogenic niche that allows survival and differentiation of myogenic cells.

Table 2. Character of Cultured Cells From Muscle Tissue
Antibody1 Day10 Days
  • a

    The number of immunoreactivity-positive cells was counted in a certain area of 10 independent samples.

Pax778.3 ± 5.423.00 ± 2.55
MF203.10 ± 2.2437.7 ± 10.2

HGF Produces Differentiated Muscle Cells in a Regenerated Spike of the Xenopus Froglet Limb

We then tried to make the froglet itself create muscle in a regenerated spike. The results, suggesting that the froglet blastema allows survival and differentiation of myogenic precursor cells, raise the possibility that the muscle-less situation of the spike may be due to absence of myogenic precursor cells in the blastema, and we, therefore, focused on a mitogen/motogen for myogenic cells, HGF, that is known to be secreted from the developing limb mesenchyme and to control migration of myogenic precursor cells into limb bud (Bladt et al., 1995; Thery et al., 1995; Brand-Saberi et al., 1996; Heymann et al., 1996). We detected hgf transcripts in the froglet blastema, but the expression level seemed to be very low (Figs. 2C, 3); therefore, we decided to supply exogenous HGF to the froglet blastema.

First, we tried to apply purified recombinant HGF in beads, but unsuccessful. Because it was possible that blastema cells should be treated by HGF for a long period, we tried HGF-producing cells. P14 cells, mouse C127-based and hgf-expressing cells, were cultured for a few weeks, and supernatant of the cultured cells was analyzed by Madin–Darby canine kidney (MDCK) cells assay (Konishi et al., 1991) to confirm that the cells under our culture condition produced HGF protein (not shown). Then, some P14 cell-aggregates labeled with PKH26 (red fluorescent dye) in advance were implanted into the froglet blastema (Fig. 5A), and the operated froglets were allowed to grow for 14–21 days. Of interest, after 2 weeks, a fluorescent signal for PKH was observed in the aggregates-implanted blastema (Fig. 5B), suggesting that the mammal cells might survive in the Xenopus environment. An anterior–posterior series of sections from the regenerated spike revealed that significant myogenesis occurred after implantation of P14 cells. MF20-positive cells were clustered at several positions through the blastema (Fig. 5C–K; Table 1), and the positive cells sometimes appeared fiber-like (Fig. 5D,F). Some clusters of the positive cells were also located in the middle and distal regions of the blastema (Fig. 5G–J). MF20-positive cells must be derived from host Xenopus cells, because MF20-positive cells were GFP-positive but PKH26-negative when PKH26-labeled P14 cells were implanted into the blastema of a cmv-GFP transgenic frog (data not shown). As shown in Figure 4A,B, muscle tissue in the Xenopus froglet limb contains many Pax7-positive satellite cells, and Pax7-expressing cells were also detectable in the MF20-positive region in this experiment (Fig. 5K), suggesting that the MF20-positive region is well-organized as muscle tissue. However, the muscle-like tissue had random locations and disorganized directions, and its connection with bone was not seen.

Figure 5.

P14 cells stimulate muscle formation in the froglet blastema. A,B: Hepatocyte growth factor (HGF) -releasing P14 cells labeled with red fluorescent dye PKH26 were implanted into the froglet blastema (A), and after 2 weeks, implanted P14 cells were detected in the froglet blastema (B). C–J: An anterior–posterior series of sections made from a specimen shown in B was stained with MF20 antibody. Arrows indicate that clusters of differentiated muscle cells existed in the manipulated spike, and higher magnifications of the regions indicated by the arrows are shown in D,F,H,J. These well-organized muscle fibers seem to consist of the bundled multinucleate fibers. K: Immunostaining with anti-Pax7 antibodies shows Pax7-expressing cells detectable at the MF20-positive region. Scale bars = 400 μm.


Spike in the Xenopus Limb Lacks Muscle Tissue and Myogenic Cells

A spike, a regenerate made after amputation of an adult Xenopus limb, is a hypomorphic structure consisting of unsegmented shaft-like cartilage and other tissues, including connective tissue, dermis, vessels, and nerves. Histological analyses have shown that the spike has no muscle tissue (Dent, 1962; Kurabuchi et al., 1983), but few reports focused on this issue. Our immunohistochemical observations confirmed the muscle-less phenotype of the spike (Fig. 1). However, our results showed that 13% of specimens (4 of 30) yielded muscle, although the muscle tissue was very poor, as reported by Fujikura et al. (1986). Even in these muscle-differentiated spikes, we did not observe any pattern or segmentation of cartilage, suggesting that there is no relationship between the pattern deficiency of cartilage and the differentiation deficiency of muscle in the spike.

Results of molecular analyses by in situ hybridization and RT-PCR speculated the muscle-less situation in the spike and also gave some insights into the reason for muscle deficiency. Lack of myoD and myf5 in the froglet blastema suggests that the muscle-less phenotype of the spike may be due not to imperfection or delay of terminal differentiation from muscle precursor cells into mature muscle fiber but rather to the absence of muscle precursor cells themselves or a defect of early gene regulation for muscle differentiation in muscle precursor cells. The lack of Pax7-positive cells in the froglet blastema suggests the absence of muscle precursor cells in the blastema. Pax7, which is expressed in myoblasts migrating into limb buds, is a key molecule for delamination from somite and migration (reviewed by Christ and Brand-Saberi, 2002, and Duprez, 2002). Indeed, the regenerating blastema in a tadpole limb bud that can differentiate muscle had Pax7-positive cells (Fig. 2K). Signaling mediated by HGF and c-met is a key molecular system for muscle development as well, and especially HGF has an attractive activity to migrating muscle precursor cells during limb muscle development (reviewed by Christ and Brand-Saberi, 2002, and Duprez, 2002). Although hgf and c-met transcripts were detected by RT-PCR (Fig. 3), the expression levels may be very low because only faint signal for either of the genes was observed by in situ hybridization (Fig. 3C,D). This incomplete condition of these signaling molecules in the froglet blastema suggests failure of migration of muscle precursor cells into the blastema. Alternatively, muscle precursor cells present in the blastema need to congregate for muscle differentiation, and HGF may be essential for this process. It is also possible that HGF is necessary for the survival and proliferation of muscle precursor cells because HGF is known as a mitogen for muscle precursor cells (Gal-Levi et al., 1998). Taken together, our results presented here suggest that early events for muscle differentiation, including cell migration, survival, proliferation, and myogenic gene expression, are key points for the muscle-less phenotype of the spike.

Environment in the Froglet Blastema Allows Muscle Differentiation

The possibility that the muscle-less phenotype is due to a disturbance of early steps for muscle differentiation in the froglet blastema suggests the explanation that a certain circumstance in the blastema prevents myogenesis-related gene expressions and resultant muscle formation. During muscle development in the limb bud, it is evident that the limb distal mesenchyme and the overlying ectoderm prevent muscle differentiation. Mouse myoblasts implanted into not proximal but distal mesenchyme of the developing chick limb bud hardly differentiate (Robson and Hughes, 1996). Overexpression of fibroblast growth factor (FGF) 4, which has its expression domain in the apical ectodermal ridge (AER) in the limb bud, results in a reduction in the number of terminally differentiated myogenic cells both in vivo (Edom-Vovard et al., 2001) and in vitro (Robson and Hughes, 1996). These studies suggest that the distal mesenchyme and the AER in the developing limb bud have inhibitory effects on muscle differentiation, and several molecules expressed in these regions, such as FGFs and bone morphogenetic proteins, are thought to play roles in the repression of muscle differentiation (reviewed by Francis-West et al., 2003). Considering the evidence of a negative environment in the developing limb bud for muscle differentiation, the blastema environment in the amputated Xenopus froglet limb might be suppressive for muscle redifferentiation.

However, when Pax7-positive muscle precursor cells, which can differentiate into myotube in vitro, were implanted into the froglet blastema, they survived and differentiated into MyHC-positive muscle fibers. These results suggest that the blastema environment does not prevent muscle differentiation of muscle precursor cells; rather, the environment seems to provide a niche for muscle formation. Of interest, it took approximately 2–3 weeks to observe the differentiated muscle cells after implantation of muscle precursor cells, although these cells differentiated into MyHC-positive muscle fiber within 10 days in vitro (Fig. 4D). Therefore, it is possible that the permissive environment of the froglet blastema may be insufficient for normal muscle differentiation.

HGF Reconstructs Muscle Tissue in the Spike

Implantation of HGF-releasing cells into the regenerating froglet blastema resulted in successful muscle formation in the spike. We found Pax7-positive cells around MF20-positive cells, suggesting that muscle satellite cells, which we have never seen in the normal spike (Fig. 1), were also reconstituted. This significant improvement in muscle formation suggests that HGF is a key molecule in muscle regeneration in the Xenopus limb. However, the C127 cell line, the base cell line of P14 that would express negligible levels of hgf (Jeffers et al., 1996), also activates muscle formation in the spike, although these cells have much less ability than do P14 cells (Table 1). This finding suggests that not only HGF but also other factors emitted from C127 cells are responsible for the reconstitution of muscle tissue. There are some possible mechanisms by which HGF (+X) reconstitutes muscle in the spike, including stimulation of the migration of muscle precursor cells from the stump into the blastema, stimulation of the assemblage/proliferation/differentiation of muscle precursor cells that reside in the blastema, and change in cell fate of multipotent blastema cells into myogenic lineage. As described above, HGF has both abilities as a mitogen and a motogen. Of interest, it was demonstrated that HGF activates muscle satellite cells from quiescence in vivo and in vitro (Bischoff, 1986, 1990; Allen et al., 1995; Tatsumi et al., 1998; Hawke and Garry, 2001) and that c-Met protein is expressed in both quiescent and activated satellite cells (Cornelison and Wold, 1997; Tatsumi et al., 1998; Hawke and Garry, 2001).

The HGF-releasing cell line P14 gave rise to clusters of muscle tissue that included muscle fibers and Pax7-positive satellite-like cells. Nevertheless, the regenerate in which muscle tissue was induced by P14 cells had an unsegmented and pattern-less shaft of cartilage. In addition, these myofibers appeared not to be in the correct location, and the direction of myofibers was random. Muscle patterning might be dependent on cartilage patterning that P14 cells could not rescue. Although we demonstrated that application of exogenous HGF produced by P14 cells into the froglet blastema resulted in the formation of well-organized but nonpatterned muscle tissue in the spike, we cannot exclude the possibility that endogenous HGF might not contribute to muscle formation during limb regeneration. Results of further investigations to compare differences between myogenesis in the Xenopus froglet blastema and that in the urodele blastema should provide insights into the ability of organ regeneration consisting of epimorphic tissue regeneration and pattern reconstruction.


Experimental Animals and Limb Amputation

Xenopus laevis tadpoles were raised in our laboratory from induced breeding of adult frogs and allowed to develop until they reached stages 51–55 (Nieuwkoop and Faber, 1956) or froglets (young adult frogs) after metamorphosis. For limb amputation, Xenopus tadpoles at stages 52–53 and froglets were anesthetized in 1:3,000 ethyl-3-aminobenzoate (Sigma) dissolved in Holtfreter's solution. Hindlimb buds in tadpoles and forelimbs in froglets were amputated at the prospective ankle (Tschumi, 1957) and at wrist level, respectively, with an ophthalmological scalpel. The manipulated tadpoles and froglets were allowed to develop for a few days until the blastemas reached appropriate sizes (a cone shape or a spike).

Histology and Immunohistochemistry

For histology, paraffin sections were triple-stained with Alcian blue, hematoxylin, and eosin. Immunohistochemical staining was carried out using MF20 (a monoclonal antibody against the myosin heavy chain [MyHC; Bader et al., 1982], purchased from Development Studies Hybridoma Bank, Iowa), anti-Pax7 antibody (kindly provided by Dr. Nomura, Tohoku University, Kawakami et al., 1997), anti-laminin antibody (Sigma), and anti-GFP antibody (Molecular probe). Samples were fixed in 4% paraformaldehyde/phosphate buffered saline (PBS) at 4°C overnight, washed with PBS several times, and immersed in 10% and 20% sucrose/PBS overnight. Fixed limbs were embedded in OCT compound (Miles), frozen in liquid nitrogen, and sectioned at 5–10 μm using a cryostat (Leica). Slides were treated with 0.25% Triton X-100 and subsequently with 0.5% skim milk/TBST for blocking. They were incubated first with antibodies diluted to 1/500 at 4°C overnight. After three washes with PBS, sections were incubated with an appropriate second antibody: fluorescein isothiocyanate–labeled anti-mouse IgG (Chemicon), rhodamine-labeled anti-mouse IgG (Chemicon), fluorescein-labeled anti-rabbit IgG (Chemicon), or horseradish peroxidase (HRP) -labeled anti-mouse and rabbit antibody mixture (DAKO). Cell nuclei were stained with DAPI. For staining of cultured cells, diaminobenzidine (DAKO) was used as the substrate for HRP reaction.

In Situ Hybridization

Xenopus myoD cDNA for a probe was a generous gift from Drs. Takahashi and Asashima (Tokyo University, Japan). Partial sequences for probes of Xenopus sox9, hgf, and c-met were isolated by RT-PCR using oligo (dT)-primed stage 51–54 limb bud cDNA generated with Superscript II (Invitrogen). The region cloned for each gene was as follows: 243-921 for Xenopus sox9 (GenBank accession no. AY035397), 31-860 for Xenopus hgf (GenBank accession no. S77422), and 3252-3824 for Xenopus c-met (GenBank accession no. AB027411). PCR products were cloned into the TOPO vector (Invitrogen) and sequenced.

Riboprobes were prepared by in vitro transcription with T7 or T3 RNA polymerase and with digoxigenin–11-UTP (Roche), according to the manufacturer's instructions. In situ hybridization on the Xenopus limb bud was performed as described previously (Endo et al., 2000; Yokoyama et al., 2000). For making serial cryosections, regenerating limbs were fixed in MEMFA (0.1 M MOPS, pH 7.4, 2 mM ethyleneglycoltetraacetic acid, 1 mM MgSO4, 3.7% formaldehyde), embedded in OCT compound (Miles), and serially sectioned at 10 μm.


Total RNA was extracted from regenerating blastemas in froglet forelimbs and tadpole hindlimb buds using a RNeasy Kit (Qiagen). DNaseI-treated total RNA was used as a template for first-strand cDNA synthesis using oligo (dT) primers. Superscript II reverse transcriptase (Invitrogen) was used for the extension, according to the manufacturer's instructions. Each PCR cycle was 94°C for 30 sec, 55 °C for 30 sec, 72°C for 1 min, and a final extension of 10 min at 72°C. We used 25 cycles (22 cycles for EF1a) for each assay. The forward and reverse primers used for each gene and the size of the PCR product were as follows: xmyoD, 5′-acagctgtggatagtggatccgag-3′ and 5′-gaagcatccggagcattcacacag-3′; xmyf5, 5′-ctcaatggtctggaagaaacag-3′ and 5′-cagcatgtttattaacacaatagc-3′; xmsx1, 5′-gggattcgttgtatggatcgc-3′ and 5′-aatggcccctacagggtaacg-3′; xsox9, 5′-tgaacttcttggactccttc-3′ and 5′-tgttctccaggagagtggac-3′; xfgf10, 5′-tacactaagtactttctccag-3′ and 5′-cataggtgttgtagccattc-3′; xhgf, 5′-gtaactaagaggggtcttgc-3′ and 5′-aattgcacagaactcccagg-3′; xc-met, 5′-ctgtgtctctcatggtactc-3′ and 5′-gctcctcttgtcataagttc-3′; EF-1a, 5′-cagattggtgctggatatg-3′ and 5′-actgccttgatgactcctag-3′.

Cell Culture and Cell Aggregate Implantation

Adult muscle tissue was isolated from the femur region of froglet hindlimbs (cmv-GFP transgenics generated in our laboratory) and cultured as described (Shibota et al., 2000). Myofibers not touching the culture dish (Becton Dickinson, 4459) were washed out, and the remaining monolayer of cells on the dish were cultured for 1 day or 10 days in culture medium (70% DMEM supplemented with 10% fetal calf serum [JBC], 20 μg/ml insulin, and 200 μg/ml gentamicin). P14 cells were kindly gifted by Dr. Noji (Tokushima University, Japan) and cultured for a couple of weeks with gentamicin (Sigma) to eliminate non–HGF-expressing cells. The MDCK assay (Konishi et al., 1991) for estimating HGF activity with supernatant of the cultured P14 cells was significantly positive (data not shown). Cell aggregates made of cultured muscle precursor cells and P14 cells labeled with PKH-26 (Zynaxis) in advance were implanted with a glass capillary, knife, and tweezers, as described previously (Yokoyama et al., 1998).


We thank Drs. Yuji Takahashi and Makoto Asashima for Xenopus myoD plasmid, Drs. Makoto Nomura and Atsushi Kawakami for anti-Pax7 antibody, Dr. Sumihare Noji for P14 cell line, and Dr. Sayuri Yonei-Tamura for critical reading of the manuscript.