SEARCH

SEARCH BY CITATION

Keywords:

  • tadpole;
  • metamorphosis;
  • thyroid hormone;
  • muscle stem cells

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. ORIGIN OF ANURAN MESODERM
  5. SOMITOGENESIS
  6. EPAXIAL AND HYPAXIAL TRUNK MUSCLES
  7. FAST AND SLOW MUSCLES OF THE TRUNK AND TAIL
  8. LIMB MUSCLES
  9. METAMORPHOSIS AND FORMATION OF ADULT MUSCLE
  10. PERSPECTIVES
  11. Acknowledgements
  12. REFERENCES

Knowledge of muscle development in a vertebrate reflects strengths of the particular model system. For example, the origin of mesoderm is very well characterized in Xenopus laevis, where development of somites is less well understood. The major problem in muscle development, presented by frogs, is the complete replacement of larval muscles by adult muscles at thyroid hormone–dependent metamorphosis. All tail muscles die, all leg muscles form de novo, and muscles in the jaw and trunk show both processes. The nature of adult muscle progenitors remains unclear. Comparison of X. laevis development with divergent amphibian patterns, such as direct developers, which lack the larval tadpole, should highlight important steps in adult muscle formation. Developmental Dynamics 236:2444–2453, 2007. © 2007 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. ORIGIN OF ANURAN MESODERM
  5. SOMITOGENESIS
  6. EPAXIAL AND HYPAXIAL TRUNK MUSCLES
  7. FAST AND SLOW MUSCLES OF THE TRUNK AND TAIL
  8. LIMB MUSCLES
  9. METAMORPHOSIS AND FORMATION OF ADULT MUSCLE
  10. PERSPECTIVES
  11. Acknowledgements
  12. REFERENCES

Development of anuran amphibians, the frogs and toads, differs from other vertebrates by the radical reorganization of the body at metamorphosis. Adding complexity to this reorganization, the bodies of the tadpole and the frog are not only very different from each other but they are also very different from any other vertebrate (Handrigan and Wassersug,2007). The herbivorous tadpole lacks legs and swims by means of a muscular tail that lacks vertebrae. The carnivorous adult has greatly modified hind limbs for jumping and no tail. At metamorphosis, tadpole muscles die and adult muscles develop. This biphasic development requires formation of two sets of muscles, with temporal and functional differences.

There are three patterns of muscle change in the transition from tadpoles to frogs. In the tadpole tail, all of the muscles die (Nishikawa and Hayashi,1994; Berry et al.,1998; Elinson et al.,1999; Shi,2000). In the forelimb and hindlimb, new muscles arise (Muntz,1975; Brown et al.,2005; Satoh et al.,2005). Finally in the jaw and the trunk, both changes occur. Larval muscles die coincident with formation of new muscles for the frog's body (Alley,1989; Nishikawa and Hayashi,1994).

These patterns of muscle gain and loss have been analyzed recently by cellular and molecular approaches in the model anuran amphibian, Xenopus laevis, and this analysis has provided new insights into old problems. In addition to the importance of this frog model, there exist among anurans numerous independent evolutionary losses of the tadpole from the life history. One of these direct developers, Eleutherodactylus coqui, has emerged as a model for this derived development (Callery et al.,2001; Elinson and Beckham,2002). The coquí, as it is popularly called in its Puerto Rican homeland, provides opportunities to see how development in general, and muscle formation in particular is altered with the deletion of the tadpole stage.

ORIGIN OF ANURAN MESODERM

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. ORIGIN OF ANURAN MESODERM
  5. SOMITOGENESIS
  6. EPAXIAL AND HYPAXIAL TRUNK MUSCLES
  7. FAST AND SLOW MUSCLES OF THE TRUNK AND TAIL
  8. LIMB MUSCLES
  9. METAMORPHOSIS AND FORMATION OF ADULT MUSCLE
  10. PERSPECTIVES
  11. Acknowledgements
  12. REFERENCES

In vertebrates, the striated muscles of the voluntary muscle system are derived from the somites, which are in turn derived from the dorsal or paraxial mesoderm. In X. laevis, dorsal mesoderm arises in the late blastula at the intersection of two signaling pathways, initiated by molecules originally localized in the oocyte. These pathways have been the subjects of genetic regulatory network analyses (Loose and Patient,2004; Koide et al.,2005) as well as recent reviews (De Robertis and Kuroda,2004; Weaver and Kimelman,2004; Heasman,2006; Kimelman,2006), and a brief summary will follow.

In X. laevis, a key molecule is the transcription factor VegT, whose RNA is localized to the vegetal cortex of the full-grown oocyte. During oocyte maturation, VegT RNA diffuses into the vegetal cytoplasm and is translated. Following fertilization and cleavage, VegT enters nuclei and activates a number of genes, most importantly TGFβ ligands of the Nodal family. When these ligands interact with cell surface receptors, the cells are induced to form mesoderm and turn on transcription factors such as Brachyury and other mesoderm-specific genes.

VegT RNA represents a cytoplasmic determinant, necessary for formation of both mesoderm and endoderm. A second cytoplasmic determinant, also localized to the vegetal cortex of the oocyte, initiates a path that converts some of the mesoderm to dorsal mesoderm. The best current candidate for this dorsal determinant is Wnt11 RNA (Tao et al.,2005; Kofron et al.,2007). The dorsal determinant is shifted asymmetrically relative to VegT by a cortical rotation shortly after fertilization. In the region of the dorsal determinant, β-catenin moves into nuclei with Xtcf3 and activates genes such as Siamois. Siamois is a transcription factor that activates genes necessary for formation of the central signaling center, the organizer. The organizer secretes molecules, such as noggin and chordin, which bind to the ligand BMP4, preventing it from interacting with its receptor. The absence of BMP4 signaling in the mesoderm, along with other signaling conditions, causes specification of dorsal mesoderm.

SOMITOGENESIS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. ORIGIN OF ANURAN MESODERM
  5. SOMITOGENESIS
  6. EPAXIAL AND HYPAXIAL TRUNK MUSCLES
  7. FAST AND SLOW MUSCLES OF THE TRUNK AND TAIL
  8. LIMB MUSCLES
  9. METAMORPHOSIS AND FORMATION OF ADULT MUSCLE
  10. PERSPECTIVES
  11. Acknowledgements
  12. REFERENCES

Segmentation of the dorsal mesoderm into somites begins during neurulation, and this process has been reviewed by Radice and coworkers (1989) and by Keller (2000). About 45 somites are generated in X. laevis (Nieuwkoop and Faber,1956), but only nine are trunk with the rest being tail (Tucker and Slack,1995). Keller (2000) complained that little work was devoted to somite patterning in amphibians as compared to chick and mouse. A similar imbalance exists in understanding the molecular segmentation clock determining the periodicity of somite formation, despite the fact that some of the pioneering work on the somite clock was done in X. laevis by Cooke (1975,1978; Cooke and Zeeman1975). Recently, there have been recent inroads into the molecular control of X. laevis somite segmentation (Giacomello et al.,2002; Moreno and Kintner,2004; Kragtorp and Miller,2006; Nagano et al.,2006).

Although comparatively less is known about amphibian somitogenesis, there are aspects not seen in other taxa. Anurans exhibit several patterns of somite morphogenesis that do not fit neatly into phylogenetic categories (reviewed in Radice et al.,1989; Gatherer and del Pino,1992; Keller,2000). Depending on the species, cells may end up spanning the myotome of a somite by rotation of elongated cells, by interdigitation of cells followed by elongation, or by fusion of cells. The initial myotomal cells may be mononucleated or multinucleated. Somitogenesis in X. laevis is the most divergent as it involves mononucleated cells that undergo somite rotation (Hamilton,1969; Kiełbówna,1981; Youn and Malacinski,1981). The restriction of this rotation to X. laevis indicates that its stature as the model system should not imply generality. Hollway et al. (2007) recently documented somite rotation in zebrafish. Given the variation within anurans, this similarity may be due to evolutionary convergence rather than conservation. Perhaps some property that makes both X. laevis and zebrafish useful as model systems, such as their very rapid early development, led to the common mechanism of somite rotation.

Somite rotation involves movement of individual cells within a somite, rather than the rotation of the whole somite (Youn and Malacinski,1981; Afonin et al.,2006). The cell lies at first with its long axis in a mediolateral orientation, so it is perpendicular to the notochord. The cell then bends around so that it ends up with its long axis in a dorsoventral orientation and parallel to the notochord. Within a somite, the most anterior cell becomes the most lateral cell, and the most posterior cell comes to lie adjacent to the notochord (Afonin et al., 2006).

The somite partitions into sclerotome and myotome, but this event is not well characterized in anurans compared to other model vertebrates. The X. laevis orthologue of the sclerotome marker Pax1 has only recently been isolated, and information on its expression is preliminary (Handrigan and Wassersug,2007). Of particular relevance to anurans is the unusual utilization of the sclerotome from which the axial skeleton is derived (Duellman and Trueb,1986; Handrigan and Wassersug,2007). First, anurans have the smallest number of presacral vertebrae of any vertebrate, with only 5–8. Second, most families of anurans lack ribs, and when present, they do not attach to the sternum. Finally, anurans lack caudal vertebrae. This lack might be expected since frogs do not have tails, but with the exception of megophyrids (Haas et al.,2006; Handrigan and Wassersug,2007; Handrigan et al.,2007), there are no vertebrae in the tails of tadpoles either.

EPAXIAL AND HYPAXIAL TRUNK MUSCLES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. ORIGIN OF ANURAN MESODERM
  5. SOMITOGENESIS
  6. EPAXIAL AND HYPAXIAL TRUNK MUSCLES
  7. FAST AND SLOW MUSCLES OF THE TRUNK AND TAIL
  8. LIMB MUSCLES
  9. METAMORPHOSIS AND FORMATION OF ADULT MUSCLE
  10. PERSPECTIVES
  11. Acknowledgements
  12. REFERENCES

Development of trunk muscles from the somites was well described histologically (Ryke,1953; Lynch,1984,1990), but antibody staining of whole mounts has provided more dramatic and accessible data (Fig. 1) (Elinson and Fang,1998; Martin and Harland,2001). Epaxial muscles form in position within the somites. The first hypaxial muscles appear as small islands on the embryo's flanks (Fig. 2). The islands expand to produce bands of muscle that will become the rectus abdominis. In tadpoles of both X. laevis and Rana pipiens, there is a large gap between left and right abdominal muscles, leaving the region surrounding the ventral midline without muscle (Fig. 1). This gap closes to join at a thin linea alba at metamorphosis (Ryke,1953; Lynch,1984).

thumbnail image

Figure 1. Slow and fast muscles in Xenopus laevis early tadpoles. These pre-feeding tadpoles (NF 46) were stained with antibodies to slow (S58) and fast muscle (F59) (Crow and Stockdale,1986). A: Dorsal view of fast muscle staining highlights most if not all of the muscles and provides a comparison to B. B: Dorsal view shows slow muscle staining of the most dorsal fiber of each trunk and tail myotome (e.g., arrow). C: This ventral view of fast muscle staining provides a comparison to D. D: Ventral view shows slow muscle staining of the most ventral fiber of each tail myotome (e.g., arrow) as well as the most ventral fiber in the rectus abdominis (e.g., arrowhead). Snout-Vent Length (SVL): 3 mm.

Download figure to PowerPoint

thumbnail image

Figure 2. Origin of hypaxial muscles. The hypaxial muscles are first present as small islands on the embryo's flank. The top embryo is Xenopus laevis, and the bottom one is Rana pipiens. The muscles are visualized immunocytochemically with 12/101 primary antibody (Kintner and Brockes,1985).

Download figure to PowerPoint

These events in tadpoles have counterparts in the direct developer, E. coqui (Elinson and Fang,1998). The future rectus abdominis first appears as bands of muscle on the left and right side. The bands move laterally and ventrally over the large mass of yolky cells until they approach each other at the ventral midline (Fig. 3). In addition to the muscle bands of the rectus abdominis, there are muscle bands laterally, which will become the obliquus externus and the transversus, as well as a small pectoralis muscle anteriorly. These muscles cover the yolk mass only partially, but with continued development to a froglet, all of these muscles expand, filling the gaps between them including the ventral gap (Fig. 3). This growth of the muscles depends on thyroid hormone (Callery and Elinson,2000a). Methimazole, an inhibitor of thyroid hormone synthesis, blocks the muscle expansion, and exogenous exposure to thyroid hormone rescues it.

thumbnail image

Figure 3. Muscle development in the direct developer Eleutherodactylus coqui. A: This Townsend Stewart (TS) stage 11 embryo has two bands of the rectus abdominis muscle, approaching the ventral midline, and small pectoralis muscles (arrow at posterior tip), extending below the arms. There are large gaps between the rectus abdominis, the pectoralis, and the lateral muscles. SVL: 4.5 mm. B: By TS 15, thyroid hormone stimulation of muscle growth has led to complete coverage of the trunk. Compare posterior tip of pectoralis (arrow) to A. SVL: 5.5 mm. The time between TS 11 and TS 15 is about six days. The primary antibody is 12/101.

Download figure to PowerPoint

The prospective hypaxial muscle cells in X. laevis express in order the same set of genes expressed in amniotes (Martin and Harland,2001). These are Pax3, followed by XMyf5, and finally XMyoD as the muscles differentiate. Cells expressing Pax3 emerge from trunk somites I–VIII, but not from head somites I–IV, which degenerate towards the end of embryogenesis. Similarly, there is no cell migration from tail somites.

In addition to Pax3, XMyf5, and XMyoD, Lbx1 is also expressed in the prospective hypaxial musculature (Martin and Harland,2006). This result was unexpected, since in zebrafish, chicken, and mouse, Lbx1 is expressed in mesenchymal cells, involved in long-range migration from the somite and destined to form limb and other muscles (Jagla et al.,1995; Mennerich et al.,1998; Birchmeier and Brohmann,2000; Gross et al.,2000; Neyt et al.,2000; Haines and Currie,2001). Lbx1 is not expressed in the interlimb regions in mouse and chicken, where hypaxial muscles form by epithelial extensions, nor is it expressed in a chondrichthyan where both fin and hypaxial muscles form by epithelial extensions.

The correlation between Lbx1 expression and mesenchymal mode of appendicular muscle origin is alluring, and led Neyt et al. (2000) and Haines and Currie (2001) to propose that this link was present in the last common ancestor between teleosts and tetrapods. Results from X. laevis and reptiles, however, point to the need to test this hypothesis further. In X. laevis, Lbx1 may be correlated with mesenchymal migration, but this type of migration also occurs in the hypaxial muscles that form between limbs. In X. laevis, long-range migration of Lbx1-expressing cells into the head produces the hypoglossal muscle, as it does in amniotes (Martin and Harland,2006). The Lbx1-expressing cells that form the hypaxial muscles, however, do not migrate far before differentiating into muscle. Conversely, epithelial extensions appear to be involved in reptile limb development (Galis,2001; Haines and Currie,2001). It is premature to conclude that the last common ancestor of teleosts and tetrapods utilized mesenchymal Lbx1-expressing cells for limb muscles and epithelial non-Lbx1-expressing cells for body wall muscles. The anuran pattern, as suggested by X. laevis, may be derived, so it is important to investigate these patterns in phylogenetically relevant and accessible animals like urodele amphibians, lungfish, and sturgeon.

The ventral movement of the future rectus abdominis muscles from their dorsal origin is accompanied by movement of pigment (Fig. 4). The pigment movement is not just migration of individual melanophores but involves epidermis and likely the underlying tissues in both X. laevis and E. coqui (Elinson and Fang,1998). Collectively, these tissues have been called the body wall. Removal of neural crest, the source of the melanophores, does not interfere with hypaxial muscle migration and the development of the rectus abdominis in X. laevis (Martin and Harland,2001). Likewise, removal of the prospective hypaxial mesoderm leads to severe abnormalities and deficits of the rectus abdominis, obliquus externus, and the transversus muscles in E. coqui, but does not interfere with pigment movement (Elinson and Fang,1998). The movement of the body wall ventrally appears to be fueled mainly by growth of the constituent tissues.

thumbnail image

Figure 4. Body wall closure in Eleutherodactylus coqui. A: The pigmented body wall of this TS 9 embryo has moved from the dorsal midline to cover halfway over the mass of yolky cells. Coincident with the leading edge of pigment are the muscle bands of the rectus abdominis. B: By TS 11, about three days later, the two sides of the pigmented body wall are close to the ventral midline. They will fuse together shortly.

Download figure to PowerPoint

As mentioned earlier, the rectus abdominis muscles from either side fuse at the ventral midline in the direct developer E. coqui, and the pigmented surface accompanies the muscle (Fig. 4). Consequently, the body wall in E. coqui secondarily provides a new external surface, surrounding the large mass of yolky cells. This expansion and closure over the yolky cells bears a resemblance to the expansion of the prospective chorion and yolk sac over the uncleaved yolk in amniotes, such as birds. A similar capacity of the body wall may have preceded the large egg size and meroblastic cleavage, thought to have occurred in the evolution of the amniote egg (Elinson,1989; Collazo et al.,1994; Chea et al.,2005). There is a major difference between the expansions in E. coqui and in amniotes with large yolky eggs, however. In E. coqui, the rectus abdominis moves all the way around the yolk mass, while in amniotes, the somite-derived hypaxial muscle does not move around the yolk (Elinson and Beckham,2002). Its path around the yolk is blocked by the formation of the amnion. This difference in hypaxial muscle migration was likely critical in allowing the huge increase in the size of the yolk mass, characteristic of the amniote egg.

FAST AND SLOW MUSCLES OF THE TRUNK AND TAIL

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. ORIGIN OF ANURAN MESODERM
  5. SOMITOGENESIS
  6. EPAXIAL AND HYPAXIAL TRUNK MUSCLES
  7. FAST AND SLOW MUSCLES OF THE TRUNK AND TAIL
  8. LIMB MUSCLES
  9. METAMORPHOSIS AND FORMATION OF ADULT MUSCLE
  10. PERSPECTIVES
  11. Acknowledgements
  12. REFERENCES

The bulk of the trunk and tail muscles in the X. laevis embryo are fast muscles, with slow muscles forming in two striking patterns (Radice,1995; Elinson et al.,1999; Grimaldi et al.,2004). One set of slow muscle fibers, termed “first wave slow fibers” by Grimaldi et al. (2004), forms a superficial layer over the tail myotomes. A second set of slow muscle fibers are present as the most dorsal and the most ventral fiber of each trunk and tail myotome, including the most ventral fibers of the rectus abdominis (Fig. 1) (Elinson et al.,1999). These muscles in the tail were called “dorsal and ventral cords” by Elinson et al. (1999) and appear to be the same as the “second wave slow fibers” of Grimaldi et al. (2004). The dorsal cords in the tail are an attenuation of longissimus dorsi in the trunk that persists in the frog (Ryke,1953; Kordylewski and Gruszka,1986; Elinson et al.,1999). The dorsal and ventral cords are the last tail muscles to disappear during metamorphosis, and they are implicated in pulling the degenerating tail tissues into the body (Elinson et al.,1999).

Formation of the first wave slow fibers depends on Hedgehog signaling, leading Grimaldi and co-workers (2004) to propose that the formation of these superficial slow muscles is homologous between zebrafish, a teleost, and Xenopus, a tetrapod. The myotomal cells closest to the notochord would be induced to form slow muscle via notochordal Hedgehog, and they would then move from this deep position to a superficial one. The rotation of the X. laevis somite, mentioned previously, would represent the proposed movement.

There are several problems with this hypothesis. First, as with Lbx1, this homology requires a very large leap between two distantly related model organisms. Since somite rotation has only been found for X. laevis among anurans, other animals should be surveyed for embryonic formation of slow muscle and its relation to notochordal Hedgehog signaling. The obvious choices are lungfish and urodele amphibians. Among anurans, Rana catesbeiana, Bufo bufo japonicus, and Rana japonica appear to have peripheral slow muscle, based on histochemistry and electron microscopy (Sasaki,1974,1977), but its existence is questioned in Rana temporaria (Muntz et al.,1989). An immunohistological survey for slow muscles in various anuran embryos would help determine whether X. laevis represents a primitive or a derived state. Second, as discussed earlier, Afonin et al. (2006) present a different interpretation of somite rotation than Grimaldi et al. (2004). In Afonin's model, all somitic cells, rather than a subset, are initially in contact at their medial end with the notochord. Upon rotation, the most lateral cells are those that were originally at the anterior end of the new segment. This orientation could suggest that the signal for slow muscle specification occurs at segmental boundaries. Alternatively, a somitic cell at the anterior end of a segment may receive the Hedgehog signal at a different time in its development compared to a posterior cell, and as a result become slow muscle.

Formation of the dorsal and ventral cords of slow muscle (Fig. 1) are independent of Hedgehog signaling (Grimaldi et al.,2004), raising the question as to how these positions at the muscle boundaries are determined. Expression of XMyf5 at the dorsal and ventral extremes of the somite indicates that these regions are myogenic centers, equivalent to the dermomyotome. A hypothesis for the formation of the dorsal and ventral cords is that the termination of the function of the myogenic centers is linked to formation of slow muscle fibers.

LIMB MUSCLES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. ORIGIN OF ANURAN MESODERM
  5. SOMITOGENESIS
  6. EPAXIAL AND HYPAXIAL TRUNK MUSCLES
  7. FAST AND SLOW MUSCLES OF THE TRUNK AND TAIL
  8. LIMB MUSCLES
  9. METAMORPHOSIS AND FORMATION OF ADULT MUSCLE
  10. PERSPECTIVES
  11. Acknowledgements
  12. REFERENCES

One of the salient features of tadpoles is their delayed limb development, although it is a popular misconception that there are no limbs until metamorphosis. The forelimbs are covered by an opercular fold, which hides them until metamorphic climax in tadpoles of most species. The forelimbs are visible under the skin in X. laevis, however. The hindlimbs are tucked behind the body at the base of the tail. They grow slowly, but muscle and cartilage are present before metamorphic climax. To provide the necessary perspective, X. laevis tadpoles begin feeding at NF stage 46 (Nieuwkoop and Faber,1956), when they are 9–12 mm long. Hindlimb buds are first visible at NF 46. The first differentiated leg muscle fibers are present at NF 53, when the tadpole is 50–60 mm long and the hindlimb is about 10 mm long (Muntz,1975; Brown et al.,2005; Satoh et al.,2005). Metamorphic climax is signaled by the breaking of the forelimbs through the operculum at NF 58, when the tadpole is 80–110 mm long and the hindlimb is about 20 mm long. There are about 3 weeks between NF 46 and NF 53, and another 3 weeks until NF 58.

Limbs grow rapidly at metamorphic climax, when thyroid hormone levels rise dramatically. Low levels of thyroid hormone appear to be important in limb development before then, but it is difficult to implicate thyroid hormone in the initiation of limb muscle formation. The thyroid gland is histologically differentiated by NF 50 (Nieuwkoop and Faber,1956), but there is no expression of the thyroid stimulating hormone (TSH) gene at NF 52/53 (Buckbinder and Brown,1993). Plasma levels of thyroxine (T4) were reported to be very low at NF 54, and triiodothyronine (T3) was not detectable until NF 57 (Leloup and Buscaglia,1977). Morvan Dubois and co-workers (2006) recently reported, however, that both T4 and T3 are present in eggs and prefeeding tadpoles (NF 35–37), indicating a maternal source of thyroid hormones. In addition, D2, the gene for a type II deiodinase, is expressed as early as NF 49 in hindlimb buds (Cai and Brown,2004). Type II deiodinase converts T4 to the more active T3, so even a very low level of circulating T4 could act specifically on the limb bud. The question of levels of thyroid hormones at NF 53, when leg muscle first appears, should be reinvestigated with more sensitive methods.

There is some hindlimb development in tadpoles of X. laevis and other species lacking a thyroid or pituitary gland (Rot-Nikcevic and Wassersug,2004; see Elinson,1994 for older references), but the presence of hindlimb muscles is not explicitly mentioned in these cases. Chemical thyroidectomy can be achieved with methimazole, which blocks thyroid hormone synthesis. When X. laevis tadpoles are reared continuously in methimazole, the hindlimb buds develop only as far as NF 52 limb buds, before muscle is expected (Brown et al.,2005). When transgenic X. laevis carry a dominant negative thyroid receptor driven by a muscle-specific cardiac actin promoter, hindlimb muscles are present but greatly reduced (Das et al.,2002). The dominant negative receptor, however, would only be expressed after cells had started on a myogenic path. There remains ambiguity about the role of thyroid hormone in the initiation of hindlimb muscle differentiation.

The picture is clearer in the direct developer E. coqui. Limb buds appear at Townsend Stewart (TS) stage 4, shortly after neural tube formation (Townsend and Stewart,1985; Elinson,1994; Richardson et al.,1998; Hanken et al.,2001). This timing is similar to amniotes, rather than anurans with tadpoles. The limbs develop continuously, and hindlimb muscles are present at TS 9 (Elinson and Fang,1998), before the formation of the thyroid gland at TS 10 (Jennings and Hanken,1998). Limbs are short and stumpy in embryos treated with methimazole, but limb muscles are present (Callery and Elinson,2000a; Callery et al.,2001). RNAs for thyroid receptors are present in the oocyte and early embryo (Callery and Elinson,2000a), however, and thyroid hormone itself has been found in early embryos of Xenopus and the toad Bufo marinus (Weber et al.,1994; Morvan Dubois et al.,2006). There is the remote possibility that maternal thyroid hormone complexed to its receptor could persist and be inherited by cells in the limb bud. Barring this, it appears that thyroid hormone stimulates growth rather than initiation of limb muscles.

In X. laevis, differentiated muscle is present at NF 53, and several muscle-specific genes are expressed before then. Diffuse expressions of cardiac actin and XMyoD were detected at NF 51 by in situs, and myogenin and Pax3 RNA were detected at NF 52 (Satoh et al.,2005). Low levels of XMyf5 and MyoD were detected in hindlimb buds by RT-PCR, with myogenin RNA present at NF 52 (Nicolas et al.,1998). The late appearance of myogenic gene expressions in the X. laevis hindlimb raises the question of the origin of the muscle progenitors.

Satoh et al. (2005) recently made multiple attempts to identify limb muscle progenitors. Muscles of embryos, transgenic for GFP driven by the muscle-specific cardiac actin promoter, were labeled with GFP as expected. A diffuse GFP signal was present in limb buds as early as NF 48 and continuing to NF 52, but the authors suggested that this signal could result from early non-specific leaky expression. To check further for muscle progenitors in early limb buds, limb buds from transgenic NF 51 tadpoles were transplanted to unlabeled hosts. Limb muscle cells were GFP positive, indicating that muscle progenitor cells were present in the limb bud by NF 51. Satoh et al. (2005) report being unable to use standard fate mapping labels to determine whether these progenitors were derived from somites. Instead they used a plasmid carrying a Cre recombinase driven by the cardiac actin promoter to label somite-derived tissues with GFP in transgenic embryos. Although back and abdominal muscles were labeled in this way, muscles in forelimbs and hindlimbs were not.

There are several possible explanations for failure to demonstrate that limb muscle progenitor cells originate from the somite. First, the labeling may be too mosaic to detect reliably a small population of cells. Second, the prospective limb muscle progenitors may not express cardiac actin when the cardiac actin-Cre plasmid is available. Third, the limb muscle progenitors may come from another source.

A non-somitic origin for limb muscle progenitors is unlikely given the demonstrations of the somitic origin in amniotes and zebrafish. A non-somitic origin in amphibians, however, was the conclusion from the older literature (Byrnes,1898; reviewed in Nicholas1955). The convincing demonstration of the somitic origin in amniotes required the detailed cell fate mapping, provided by the chick/quail system (Christ et al.,1977; Chevallier et al.,1977). Such limb fate mapping has not been accomplished for anurans, and I have found no reports on urodeles either, even though there is a long tradition in the experimental analysis of their limb development.

In chick and mouse, there is a population of cells, which are derived from the somite and express Pax7 (Gros et al.,2005; Relaix et al., 2005; reviewed in Buckingham,2006). These cells function both as muscle progenitor cells for generating new myoblasts during embryonic and fetal development and as satellite cells for repair and regeneration of muscle in adults. Chen et al. (2006) recently identified satellite cells, expressing Pax7, in tail muscle of X. laevis tadpoles. These cells produce myoblasts and new muscle during tail regeneration. This linkage in X. laevis makes it likely that muscle stem cells can be identified elsewhere in the tadpole by their expression of Pax7. Neither Pax7 nor Pax3 expression was detected in X. laevis hindlimb buds at NF 51, although muscle progenitor cells are present (Satoh et al.,2005).

As discussed earlier, Lbx1 is expressed in somitic cells that migrate mesenchymally into limb buds to form muscle in chicken and mouse. Lbx1 is not expressed in the interlimb somites in these amniotes. In contrast, Lbx1-expressing cells migrate from all trunk somites in X. laevis (Martin and Harland,2006). Some of these produce the hypaxial muscles, but there was no demonstration that the Lbx1 cells are also limb muscle progenitors. Robust Lbx1 expression is present in limbs about three weeks later at NF 53, when limb muscles first appear. What is lacking in X. laevis and in any other anuran is information on whether Lbx1 expressing mesenchymal cells migrate from the somite into the limb bud, and if so, when this migration is.

At this moment, the nature of limb muscle progenitors in anurans and their origin are mysteries. Consideration of results from mouse and chicken may provide hints to solve these mysteries. For example, when MyoD is knocked out in mouse embryos, the formation of limb muscles is significantly delayed (Kablar et al.,1997). In X. laevis, Lbx1 downregulates MyoD and allows continued proliferation without differentiation of muscle precursor cells (Martin and Harland,2006). Lbx1 is expressed in limb buds at NF 53, so a continuous expression of Lbx1 in limb muscle progenitors from the time of somite migration until NF 53 could contribute to delays in anuran limb muscle development. An alternative possibility is the requirement for Wnt-6 expression in the ectoderm for muscle differentiation in the chicken limb bud (Geetha-Loganathan et al.,2006). A delay in anurans of the expression of an ectodermal stimulus like Wnt-6 could constitute the basis for delayed limb muscle development.

It is likely that examination of limb development in E. coqui would provide guidelines in searching for limb muscle precursors in the development of tadpoles. Both forelimb and hindlimb buds are large; they form early, and limb muscles appear about a week after limb buds. Development of E. coqui is derived. Its developmental trajectory can be represented as a deletion of the larval tadpole and the joining of embryonic events to metamorphic ones (Fig. 5) (Schlosser and Roth,1997; Schlosser,2001; Callery and Elinson,2000b; Callery et al.,2001). Embryonic events in X. laevis should be the same as those in E. coqui, so the origin of the limb myogenic cells is expected to be conserved between embryos of E. coqui and X. laevis.

thumbnail image

Figure 5. Relationship between a direct and an indirect developing frog. The entire tadpole stage in the indirect developer (bottom), which lasts months, is reduced to a few hours in the direct developer (top). In the indirect developer, the operculum covers the forelimb area just before the tadpole begins feeding. At metamorphic climax, the forelimbs burst through a hole in the operculum, and the tadpole stops feeding until it is a frog. In the direct developer E. coqui, maximal opercular growth and the initiation of thyroid hormone–dependent events both occur at TS 10 (Callery and Elinson,2000a,b). The nutrition, provided by tadpole feeding, is replaced by the larger yolk supply in the direct developer. Reproduced from Callery and Elinson (2000b) with kind permission of Springer Science and Business Media.

Download figure to PowerPoint

METAMORPHOSIS AND FORMATION OF ADULT MUSCLE

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. ORIGIN OF ANURAN MESODERM
  5. SOMITOGENESIS
  6. EPAXIAL AND HYPAXIAL TRUNK MUSCLES
  7. FAST AND SLOW MUSCLES OF THE TRUNK AND TAIL
  8. LIMB MUSCLES
  9. METAMORPHOSIS AND FORMATION OF ADULT MUSCLE
  10. PERSPECTIVES
  11. Acknowledgements
  12. REFERENCES

At metamorphosis, larval muscles degenerate and adult muscles develop (Chanoine and Hardy,2003). There is only muscle degeneration in the tail, and only new muscle formation in the limbs. Both processes occur simultaneously in the jaw and trunk. The jaw undergoes a tremendous remodeling at metamorphosis, and this is reflected in the musculature. Although observations in X. laevis, either morphological or molecular, are lacking, larval muscle degeneration and new muscle formation are evident histologically in jaws of R. pipiens (Alley,1989). In contrast, there is no jaw muscle degeneration in E. coqui (Hanken et al.,1997). Muscles originate in a mid-metamorphic configuration and develop directly to their adult forms. Albeit for different reasons, both E. coqui and R. pipiens jaw muscle development indicates that the adult jaw musculature is independent of the larval one. For E. coqui, there are no larval muscles. For R. pipiens, not only are the adult muscles new but also some of the skeletal anchorage points are altered.

The transition from larval to adult muscles in the trunk is better characterized. The gene for the myogenic regulatory factor, myogenin, is not expressed during primary myogenesis in the embryo, but it is expressed during the secondary myogenesis giving rise to the adult trunk muscles (Nicolas et al.,1998). There is an increase in the adult isoform of tropomyosin in the trunk during metamorphosis (Nishikawa and Hayashi,1994), and appearance of trunk muscle fibers expressing an adult form (MHC A7) of myosin heavy chain (Nicolas et al.,1998). Adult muscle fibers are also recognizable by their smaller diameter compared to larval ones.

Using differences in tropomyosin and myosin heavy chain detected immunologically, Nishikawa and Hayashi (1994) provide evidence that adult muscle fibers are produced as a wave over the trunk, beginning anteriorly and dorsally. One possibility is that there is a wave of conversion of larval muscle fibers to adult ones, with gene expression changes within cells. A second possibility is that adult muscle progenitors become active and produce new adult muscle fibers, which replace the dying larval muscle fibers. The latter possibility appears to be the case.

Myoblasts, isolated from tadpole tail, are expected to be larval type, and they differ from myoblasts isolated from adult leg muscle (Shibota et al.,2000). Both types of myoblasts form myotubes in culture, but they react differently to thyroid hormone, T3. T3 causes death of both larval myoblasts and myotubes formed from them, but not of adult myoblasts or myotubes. This differential sensitivity to T3 was then used to show that there are two populations of myoblasts in the NF 55 trunk. Isolated trunk myoblasts could be separated into two groups based on gravity sedimentation. One group did not die in response to T3, while the other had some death (Shimizu-Nishikawa et al.,2002). These results suggest that the co-existence in the metamorphosing trunk of larval muscle cell death and adult muscle cell development is due to different types of myoblasts, larval and adult.

PERSPECTIVES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. ORIGIN OF ANURAN MESODERM
  5. SOMITOGENESIS
  6. EPAXIAL AND HYPAXIAL TRUNK MUSCLES
  7. FAST AND SLOW MUSCLES OF THE TRUNK AND TAIL
  8. LIMB MUSCLES
  9. METAMORPHOSIS AND FORMATION OF ADULT MUSCLE
  10. PERSPECTIVES
  11. Acknowledgements
  12. REFERENCES

Alley (1989) aptly called the replacement of old larval muscles in the R. pipiens jaw by new adult muscles “retrofitting.” Retrofitting can be contrasted to “co-option” in which larval muscles would be co-opted for use in the adult. In surveying the metamorphic changes in jaw, trunk, and limb muscles, the balance of change is clearly on the retrofitting side. Indeed, there is presently no evidence for any larval muscle surviving into the adult. Given this retrofitting, a major open question for frog muscle development is the nature and location of adult muscle precursors in the embryo and larva.

We can consider two hypotheses. In the Muscle Stem Cell Hypothesis, there are muscle stem cells in the tadpole, which can become either larval myoblasts or adult myoblasts (Fig. 6). In the absence of thyroid hormone, the muscle stem cells produce larval myoblasts, which make larval muscle cells. A low level of thyroid hormone may promote adult myoblast development from muscle stem cells and subsequent adult myotube formation. In this scenario, there are no specifically adult type cells “hanging out” in early development, let's say before there is a thyroid gland at NF 52. These muscle stem cells could still migrate into the limb buds, prior to then, but they would need a further stimulus to produce adult myoblasts.

thumbnail image

Figure 6. Hypotheses on the origin of larval and adult muscle cells. In the Muscle Stem Cell Hypothesis, Somite Cells become Muscle Stem Cells. In the absence of thyroid hormone (TH), these stem cells produce larval myoblasts, which proliferate and differentiate to generate Larval Muscle Cells. With low levels of TH, some Muscle Stem Cells produce adult myoblasts, which generate Adult Muscle Cells. In the Somite Specification Hypothesis, a Somite Cell becomes a Larval Muscle Stem Cell upon receiving Signal A, and it becomes an Adult Muscle Stem Cell upon receiving Signal B. Larval Muscle Stem Cells generate Larval Muscle Cells in the absence of TH and die in its presence. Adult Muscle Stem Cells generate Adult Muscle Cells upon activation by TH.

Download figure to PowerPoint

In the Somite Specification Hypothesis, both larval and adult muscle stem cells are specified from the somite, when other specifications occur (Fig. 6). Specification would occur due to various signals, comparable to the specification of the slow peripheral muscles (Grimaldi et al.,2004). These adult muscle stem cells would hang out in the limb buds, at the dorsal periphery of the trunk musculature, and in the head muscles, until their activity rose above a basal level perhaps due to low levels of thyroid hormone.

Shimizu-Nishikawa et al. (2002) identified in X. laevis two populations of muscle precursor cells, which differed in their response to thyroid hormone. Larval type cells died, while adult type cells survived. They called these precursor cells “myoblasts,” but they likely were muscle stem cells or satellite cells, the precursors to myoblasts. This distinction can be made by determining whether they express Pax7, characteristic of stem and satellite cells. As mentioned previously, Pax7-expressing satellite cells are present in tadpole tail muscle (Chen et al.,2006). The existence of two populations of muscle stem cells, differentially sensitive to thyroid hormone, would support the Somite Specification hypothesis (Fig. 6). Although both would express Pax7, each would express other genes specifically marking them with differences important for larval vs. adult development.

Complementing this major question on the origin of adult muscle cells in frogs, there are amphibian models that do not show the typical frog biphasic pattern of development. The direct developing E. coqui was discussed here, but there are also urodeles, such as Ambystoma mexicanum, which are the amphibian sister group to the frogs. Examination of amphibians other than Xenopus is not only feasible, but can contribute to analysis of the origin of frog muscle in two ways. First, comparative analysis may reveal an interaction that would otherwise be obscure in Xenopus. For example, investigation of Lbx1-expressing cells would be easier in E. coqui, because of the larger limb buds and faster limb development. Second, there should be plausible evolutionary scenarios that link mechanisms of adult muscle formation in the different amphibians. Attempts to build those scenarios would highlight events requiring further explanation.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. ORIGIN OF ANURAN MESODERM
  5. SOMITOGENESIS
  6. EPAXIAL AND HYPAXIAL TRUNK MUSCLES
  7. FAST AND SLOW MUSCLES OF THE TRUNK AND TAIL
  8. LIMB MUSCLES
  9. METAMORPHOSIS AND FORMATION OF ADULT MUSCLE
  10. PERSPECTIVES
  11. Acknowledgements
  12. REFERENCES

I thank Tamara Holowacz, University of Toronto, and Boris Kablar, Dalhousie University, for their comments on this review. This work was supported by grant IOB-0343403 to R.P.E. from the National Science Foundation (NSF), USA.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. ORIGIN OF ANURAN MESODERM
  5. SOMITOGENESIS
  6. EPAXIAL AND HYPAXIAL TRUNK MUSCLES
  7. FAST AND SLOW MUSCLES OF THE TRUNK AND TAIL
  8. LIMB MUSCLES
  9. METAMORPHOSIS AND FORMATION OF ADULT MUSCLE
  10. PERSPECTIVES
  11. Acknowledgements
  12. REFERENCES