Phenotypic conversion of distinct muscle fiber ... - Semantic Scholar

THE JOURNAL OF COMPARATIVE NEUROLOGY 399:20–34 (1998)

Phenotypic Conversion of Distinct Muscle Fiber Populations to Electrocytes in a Weakly Electric Fish GRACIELA A. UNGUEZ* AND HAROLD H. ZAKON Department of Zoology, University of Texas, Austin, Texas 78712

ABSTRACT In most groups of electric fish, the electric organ (EO) derives from striated muscle cells that suppress many muscle phenotypic properties. This phenotypic conversion is recapitulated during regeneration of the tail in the weakly electric fish Sternopygus macrurus. Mature electrocytes, the cells of the electric organ, are considerably larger than the muscle fibers from which they derive, and it is not known whether this is a result of muscle fiber hypertrophy and/or fiber fusion. In this study, electron micrographs revealed fusion of differentiated muscle fibers during the formation of electrocytes. There was no evidence of hypertrophy of muscle fibers during their phenotypic conversion. Furthermore, although fish possess distinct muscle phenotypes, the extent to which each fiber population contributes to the formation of the EO has not been determined. By using myosin ATPase histochemistry and anti-myosin heavy chain (MHC) monoclonal antibodies (mAbs), different fiber types were identified in fascicles of muscle in the adult tail. Mature electrocytes were not stained by the ATPase reaction, nor were they labeled by any of the anti-MHC mAbs. In contrast, mature muscle fibers exhibited four staining patterns. The four fiber types were spatially arranged in distinct compartments with little intermixing; peripherally were two populations of type I fibers with small cross-sectional areas, whereas more centrally were two populations of type II fibers with larger cross-sectional areas. In 2- and 3-week regenerating blastema, three fiber types were clearly discerned. Most (. 95%) early-forming electrocytes had an MHC phenotype similar to that of type II fibers. In contrast, fusion among type I fibers was rare. Together, ultrastructural and immunohistochemical analyses revealed that the fusion of muscle fibers gives rise to electrocytes and that this fusion occurs primarily among the population of type II fibers in regenerating blastema. J. Comp. Neurol. 399:20–34, 1998. r 1998 Wiley-Liss, Inc. Indexing terms: skeletal muscle transdifferentiation; electric organ development; myosin heavy chain

Skeletal muscle cells of electric fish express a unique ability to undergo changes in their biochemical and morphological properties and give rise to electrocytes, the currentproducing cells of the electric organ (EO; reviewed in Bennett, 1971; Fox and Richardson, 1978, 1979). Previous light and electron microscopy studies have shown that electrocytes in all but one family (apternotidae) are myogenically derived (Bennett, 1971; Fox and Richardson, 1978, 1979). Mesodermal cells in the EO primordium differentiate into myoblast-like cells, which subsequently form multinucleated myotubes with a cross-striated contractile apparatus. In most groups of electric fish, further maturation of the EO results in the disassembly and/or degradation of the myofilamentous structures in parallel with striking alterations in the morphology of myogenic cells (Bennett, 1971; Fox and Richardson, 1978, 1979). Although EOs derive from a common precursor cell type, i.e., skeletal muscle, there is a striking diversity in their

r 1998 WILEY-LISS, INC.

size and shape among different fish (Bennett, 1971). In all electric fish, electrocytes are significantly larger than the muscle cells from which they derive. An increase in cell size might result from the recruitment of many muscle fibers to form a single electrocyte. Alternatively, a muscle fiber may convert and undergo hypertrophy to give rise to a single electrocyte. The extent to which cell fusion versus cell hypertrophy occurs during the formation of electrocytes has not been determined. Interestingly, EOs are formed from a large variety of skeletal muscles in fish representing at least six indepen-

Grant sponsor: National Institutes of Health; Grant number: R01 NS25513. *Correspondence to: Graciela A. Unguez, Department of Zoology, University of Texas, Austin, TX 78712. E-mail: [email protected] Received 9 January 1998; Revised 22 April 1998; Accepted 28 April 1998

TRANSDIFFERENTIATION OF MUSCLE TO ELECTRIC ORGAN dently evolved groups (Bennett, 1971; Bass, 1986). For example, electrocytes develop from extraocular muscles in Astroscopus, from brachial muscles in torpedinids, from pectoral muscles in Malapterurus, and from axial and tail muscles in the remainder groups (Bennett, 1971). The mechanisms by which only certain skeletal muscles undergo such phenotypic conversion are unknown. Furthermore, as in other vertebrates, skeletal muscles in fish are composed of different populations of fibers with distinct phenotypic properties. Previous studies on EO development, however, did not attempt to investigate the contribution of each fiber type population to the formation of electrocytes. Thus, whether all muscle fiber types are capable of converting their phenotype or whether transdifferentiation is a characteristic of only a subpopulation of fibers has not been determined. Sternopygus macrurus is a highly regenerative vertebrate species of Gymnotiform fish (Baillet-Derbin, 1978; Anderson and Waxman, 1981; Patterson and Zakon, 1993). After amputation of its tail, it can fully regenerate its spinal cord, skin, connective tissue, vertebrae, muscle, and EO (Patterson and Zakon, 1993). The recapitulation of muscle transdifferentiation to EO during regeneration has afforded the opportunity to study changes in some of the biochemical properties of developing electrocytes (Patterson and Zakon, 1993, 1997). The present study exploits the regeneration potential of S. macrurus to determine whether fusion or hypertrophy of all or some muscle fiber types gives rise to electrocytes. Our ultrastructural and biochemical analyses revealed that the fusion of muscle fibers gives rise to electrocytes and that this fusion occurs primarily among a select fiber type population. Preliminary results have been reported in abstract form (Unguez and Zakon, 1996).

MATERIALS AND METHODS Tissue preparation S. macrurus, a fresh-water species of knife fish native to South America, were obtained commercially from various fish importers. Adult fish of both sexes, 20–35 cm in length, were housed individually in 15 to 20-gallon aerated aquaria maintained at 25–28°C and fed three times weekly. Twentyfive fish were anesthetized using 2-phenoxy ethanol (1:1,500 in tank water), and the distal segment of the tail (1–2 cm) was transected at the caudal end of the anal fin. After tail amputation, the wound was treated with Woundex (a topical fungicide/antibiotic), and the fish were returned to their tanks and monitored until they fully recovered from anesthesia. At the level of the transection, EO predominates, but skeletal muscle, connective tissue, spinal cord, and skin are also present (Fig. 1). Tail segments cut from control fish (n 5 5) were mounted on cork, frozen in liquid nitrogen-cooled isopentane, and stored at 280°C. The other 20 fish were re-anesthetized 2 (n 5 10) or 3 (n 5 5) weeks after amputation, and their regenerating blastema were removed, mounted on cork, frozen in liquid nitrogen-cooled isopentane, and stored at 280°C. All procedures used in this study followed the American Physiological Society Animal Care Guidelines and were approved by the Animal Use Committee at the University of Texas at Austin, Austin, TX.

Electron microscopy Normal adult tails (n 5 2) and 2-week regenerating blastemas (n 5 3) were examined under a transmission

21

electron microscope (TEM). After surgical removal, blastemas and tails were fixed in 4% paraformaldehyde/2.5% gluteraldehyde overnight, transferred to sodium phosphate buffer (pH 5 7.2), post-fixed in osmium (2% osmium in sodium phosphate buffer) for 1 hour, dehydrated in an alcohol series and propylene oxide, and embedded in Spurrs’ plastic resin (Polysciences, Warrington, PA). Ultrathin (,90-nm) tissue cross sections were cut with a diamond knife, stained with uranyl acetate and lead citrate, and examined with the Hitachi HU 11-E TEM (Cell Research Institute of the University of Texas at Austin).

Biochemical analysis Transverse sections of control adult tails were cut at 12-µm thickness in a cryostat at 220°C, mounted on glass slides, and air dried at room temperature. Serial cross sections were stained qualitatively for myofibrillar ATPase (mATPase) with an alkaline (Alk-mATPase; pH 5 10.0) or acid (acid-mATPase; pH 5 4.35) preincubation using the method described by Mabuchi and Sreter (1980). Five monoclonal antibodies specific to different myosin heavy chain (MHC) isoforms were used. The reactivity patterns of these antibodies have been described for mammalian and avian skeletal muscle (Table 1). BF-F3 (IgM) was provided by Dr. S. Schiaffino (Institute of Pathology, Padova, Italy); N2.261 (IgG), A4.84 (IgM), A4.74 (IgG), and MF20 (IgG) were obtained from the Developmental Hybridoma Bank (Iowa City, IA); and anti-developmental (Dev; IgG) was obtained from Novocastra Laboratories (Vector Laboratories, Burlingame, CA). Supernatants of antibodies were used at the following dilutions: MF20 (1:5), N2.261 (1:5), A4.84 (1:1), A4.74 (1:2). Antibodies BF-F3 and Dev were used at dilutions of 1:1,000 and 1:100, respectively. The specific isotypes of MHC recognized by each of the monoclonal antibodies (Table 1) used in S. macrurus are not known. To test the specificity of these antibodies in S. macrurus, Western blots were performed on samples prepared from frozen tail muscle. All antibodies labeled a band with a molecular weight of approximately 200,000 Daltons, a molecular weight corresponding to that within the range of MHC isoforms (data not shown). Although distinct MHC isoforms were not distinguished on Western blots, these antibodies clearly labeled distinct populations of muscle fibers (see Results). In addition, monoclonal antibodies 5D2, specific to fast Ca21-ATPase sarcoplasmic reticulum (SR) in chick muscle (Kaprielian and Fambrough, 1987) and 12–101, specific to a membrane protein of unknown identity of the SR in newt muscle (Griffin et al., 1987), were also obtained from the Developmental Hybridoma Bank. Immunohistochemical analysis of MHC and SR content in individual fibers from control adult tails was performed on cross sections serial to those used for mATPase. Tissue sections were air dried, rehydrated in phosphate-buffered saline (PBS, pH 5 7.4) for 5 minutes, incubated in blocking solution (PBS, 2% BSA, 5% goat serum) for 30 minutes, and subsequently incubated overnight at 4°C with primary antibody diluted in blocking solution. Sections incubated without primary antibody were used as controls to visualize nonspecific staining. In some cases, primary antibody was visualized with fluorescein- or rhodamineconjugated secondary antibody (anti-mouse, 1:100; Cappel Laboratories, West Chester, PA). In other cases, the MHC antigen-antibody complex was visualized by using a biotinylated secondary antibody (Vectastain ABC kit, Vector

22

G.A. UNGUEZ AND H.H. ZAKON

Fig. 1. Cross section (12-µm thick) of a normal adult tail near the caudal end of the anal fin immunoreacted with desmin antibody D76 (Developmental Hybridoma Bank). Dorsal is up, ventral is down. The spinal cord (SC) is located dorsal to the central blood vessel (BV). Muscle fibers (arrows) are arranged in fascicles between the skin and

the electrocytes (EC). Both muscle fibers and electrocytes are immunolabeled with anti-desmin D76. The inset (enlargement of dashed box) shows an enlarged view of a fascicle from the ventral muscle group also present in this region of the tail. Scale bar 5 500 µm.

TABLE 1. Monoclonal Antibody Specificity for Skeletal Muscle MHCs

tissue sections were recorded and stored using a Nikon Diaphot epifluorescence microscope connected to a Cohu 4915 video camera and a Colorado Video frame store, interfaced to a Macintosh Quadra 800 running NIH Image 1.47 software. Final images were rendered using Adobe Photoshop (version 3.0.5; Adobe Systems, Mountain View, CA). The incidence of distinct fiber phenotypes according to mATPase and immunohistochemical staining in S. macrurus tail muscles was calculated based on a sample size of 100 fibers per fish from five control fish. Regenerating blastema from fish 2 and 3 weeks after tail amputation were sectioned in one of two ways. For immunohistochemical analysis, 12-µm-thick cross sections were cryostat sectioned at 220°C and processed as described for adult control tails. For confocal microscopy, 100- to 200-µmthick longitudinal sections were cut on a vibratome. These

Antibody MF202 BF-F33 A4.844 A4.744 N2.2614 Dev5

MHC1

type

All sarcomeric myosin IIb I IIa, IIx I, IIa Embryonic/neonatal

Species Chicken Rat Rat Mouse Mouse Rat

1MHC,

myosin heavy chain. et al. (1982). et al. (1989). 4Hughes and Blau (1992). 5Talmadge et al. (1995). 2Bader

3Schiaffino

Laboratories), and a horseradish peroxidase (HRP) reaction was run to amplify the signal by use of diaminobenzidine (DAB) and hydrogen peroxide (H2O2). Immunolabeled

TRANSDIFFERENTIATION OF MUSCLE TO ELECTRIC ORGAN longitudinal tissue sections were fixed in methanol (100%) at 220°C for 5–10 minutes before incubation in primary antibody. Images were captured on a Leica TCS 4D confocal microscope.

Fiber size measurements The cross-sectional areas (CSAs) of muscle fibers from adult control fish were measured on the sections stained for Alk-mATPase using the NIH Image 1.47 image processing system. At least 25 muscle fibers of each fiber phenotype (as determined above) were sampled from sections of five adult control fish. Data are presented as mean 6 standard error of the mean (SEM).

RESULTS In the weakly electric fish S. macrurus, the electric organ runs parallel to the vertebral column along the caudal twothirds of the fish’s body (Patterson and Zakon, 1993). At the level of transection, the spinal cord and blood sinus are found within the vertebral column at the center of the tail (Fig. 1). Surrounding the vertebral column are electrocytes that are cylindrical, cigar-shaped cells that range from 500 to 4,000 µm in length and 50 to 100 µm in width (Mills et al., 1992). Muscle fibers run longitudinally and are arranged in fascicles between electrocytes and the skin. At the same level of transection, fascicles of muscle fibers distinct from those adjacent to the electrocytes are also found, and these form the ventral muscle groups (Fig 1).

Tail regeneration occurs in a proximal to distal gradient After amputation of the tail, a blastema consisting of undifferentiated ependymal and mesenchymal cells formed at the site of transection. These cells differentiated into distinct organs like skin, muscle, EO, and spinal cord in a graded fashion from proximal to distal and from peripheral to central regions of the blastema (Patterson and Zakon, 1993). This regeneration gradient allows for the analysis of cells at early, intermediate, and late stages of differentiation within each blastema. For example, the proximal two-thirds of 2-week blastema contained muscle fibers adjacent to the epidermis and developing electrocytes medial to the muscle fibers, whereas most of the distal third contained only differentiating myogenic cells next to the epidermis (Fig. 2). Blastemas were on average 7.5 mm long (range 5 5–9 mm) after 2 weeks of regeneration and 16.7 mm long (range 5 13–20 mm) after 3 weeks

Fig. 2. Confocal microscopic image of a longitudinal section of a blastema labeled with anti-myosin heavy chain monoclonal antibody MF20 2 weeks after tail amputation. At this stage of regeneration, MF20 labels developing electrocytes and muscle fibers exclusively. The ventral epidermis is at the bottom edge of the photo and is not labeled. The proximal end (site where the tail was cut) is to the left, and the distal tip of the blastema is to the right. Cellular differentiation is just beginning at the distal tip of the blastema, whereas this process has been underway for 2 weeks near the amputation site. Developing electrocytes can be seen in the proximal region of the blastema (arrows). They are not fully mature, as mature electrocytes are negative for MF20 (Patterson and Zakon, 1996). Brightly labeled muscle cells (arrowheads) are found throughout at least two-thirds the length of the blastema between the developing electrocytes and the ventral epidermis. Note that only muscle cells are present at the distal tip of the blastema. Scale bar 5 500 µm.

23

24


of regeneration. The following sections present ultrastructural and immunohistochemical analyses of electrocyte and muscle fiber development in blastema 2 and 3 weeks after tail amputation.

Ultrastuctural analysis reveals fusion of muscle fibers To more conclusively determine whether or not cell fusion among muscle fibers occurs during EO formation, an ultrastructural analysis of developing muscle fibers and electrocytes in 2-week regenerating blastema was performed. Figure 3A shows an electron micrograph of a muscle fiber and a region of an electrocyte from an adult control tail. In the mature tail, skeletal muscle fibers revealed an ultrastructural arrangement characteristic of muscle fibers in other vertebrates. For example, each muscle fiber was surrounded by a basal lamina and had well organized myofibrils that were closely packed in a radial arrangement (Fig. 3A). The cytoplasm intervening between myofibrils contained SR and mitochondria. Mitochondria were also observed next to the plasma membrane along with myonuclei. As in muscle fibers, mature electrocytes showed peripheral mitochondria and nuclei in close proximity to the cell’s plasma membrane (Fig. 3A). Unlike muscle fibers, however, the cytoplasm of electrocytes was devoid of myofibrils. Instead, electrocytes contained loosely arranged filaments and vesicular structures throughout their cytoplasm. Electron micrographs of transverse sections taken near the middle of 2-week regenerating blastema are shown in Figure 3B–D. At this level of the blastema, muscle fibers showed an ultrastructure similar to that of fibers in adult control tails (Fig. 3B). Unlike control tails, however, there was a high incidence of differentiated muscle fibers adjacent to each other in regenerating blastema (Fig. 3B). Further inspection revealed gaps in the basal lamina of closely apposed fibers resulting in open connections between their cytoplasms (Fig. 3B). There was no evidence indicating the formation of single, larger muscle cells from fiber fusion. The fusion of muscle fibers with larger cells containing clusters of myofibrils dispersed throughout an amorphouslike cytoplasm was commonly observed (Fig. 3C,D). For example, in Figure 3C, gaps between the membranes of a muscle fiber and a cell with an elongated shape and clusters of myofilaments within an amorphous cytoplasm are revealed. Since the latter cells were found within the regenerating blastema and were at least three times the size of a normal muscle fiber, it is unlikely that these cells represent single degenerating or differentiating muscle fibers. Instead, the shape and cytoplasmic content resembled those of mature electrocytes. Often, a muscle fiber extended a process that fused with that of another muscle fiber displaying a lower density of myofibrils in its cytoplasm (Fig. 3D). Together, data from ultrastructural analyses are consistent with a process in which electrocytes are generated from the fusion of differentiated muscle fibers with a subsequent disassembly of their myofibril arrangement and a concomitant increase in cell size.

Histochemistry and immunohistochemistry distinguish four muscle fiber phenotypes in mature Sternopygus tail Phenotypic profiles of muscle fibers in the mature tail were studied by using mATPase histochemistry and MHC

immunohistochemistry. Four distinct fiber types that were spatially arranged in distinct compartments in the tail with little intermixing were distinguished in the tail of S. macrurus (Figs. 4 and 5). These are described below in order of their position along the superficial to deep axis. Electrocytes did not stain with either Alk-mATPase or acid-mATPase (Fig. 4) and were not labeled by any of the anti-MHC antibodies used in this study (Fig. 5). Myosin ATPase revealed the presence of type I (light with Alk-mATPase; dark with acid-mATPase) and two subtypes of type II (intermediate or dark with AlkmATPase; light with acid-mATPase) muscle fibers (Fig. 4). Type I fibers were located peripheral to the larger type II fibers (Fig. 4). Based on their staining patterns with the five anti-MHC antibodies used, type I fibers were further classified as fibers A and B, and type II fibers were classified as fibers C and D (Fig. 5). Type A fibers, located closest to the skin, had the smallest mean CSA (188 6 8 µm2) of all muscle fibers in the tail. These small fibers were immunoreactive for anti-slow (A4.84), anti-slow/IIa (N2.261), and anti-IIx (BF-F3) MHC antibodies (Fig. 5). However, these fibers did not react with anti-IIa/IIx (A4.74) or anti-embryonic/neonatal (Dev) MHC antibodies (Fig. 5). B fibers, located just medial to A fibers, had a mean CSA of 557 µm2 (6 38). Like A fibers, B fibers reacted positively with A4.84 and N2.261 (Fig. 6) but not with A4.74 or Dev MHC antibodies (Fig. 5). However, unlike A fibers, B fibers were not labeled with BF-F3 (Fig. 5). Medial to fibers A and B were C and D fibers, the two subpopulations of type II fibers (Fig. 5). C fibers had a larger mean CSA (1525 6 88 µm2) than either A or B fibers. Antibodies A4.84 and N2.261 did not label these fibers (Fig. 5). In contrast, C fibers reacted with antibodies BF-F3, A4.74, and Dev (Fig. 5). The most centrally located population of fibers were D fibers, which had the largest mean CSA (1934 6 34 µm2) of the tail muscle fibers. D fibers had immunolabeling patterns similar to those of C fibers. For example, whereas antibodies A4.84 and N2.261 did not label D fibers, antibodies BF-F3, A4.74, and Dev did (Fig. 5). However, unlike C fibers, D fibers stained intermediate after Alk-mATPase (Fig. 4) and often appeared to stain darker with antibody Dev (Fig. 5). The mean percentages (6 SEM) of fiber types A, B, C, and D were 38 (6 4.2), 8 (6 1.5), 27 (6 7.8), and 26 (6 3.3), respectively. Muscle fibers and electrocytes were also studied using antibodies 12–101, an antibody to an SR protein (Griffin et al., 1987), and 5D2, an antibody to fast Ca21-ATPase SR (Kaprielian and Fambrough, 1987). In normal adult tail, type I muscle fibers (A and B fibers) did not react with either 12–101 (Fig. 6) or 5D2 (not shown) whereas type II fibers (C and D fibers) were labeled with both antibodies (Fig. 6, 5D2 not shown). Mature electrocytes were not labeled with either anti-SR antibody (Fig. 6, 5D2 not shown). In summary, four muscle fiber phenotypes were identified in the myotome of S. macrurus. ATPase histochemistry showed that type I fibers were superficial and had smaller CSAs than the more medial type II fibers. MHC and SR immunohistochemistry revealed two subpopulations of type I fibers, fibers A and B, and two subpopulations of type II fibers, fibers C and D (Figs. 5, 6).

TRANSDIFFERENTIATION OF MUSCLE TO ELECTRIC ORGAN

Fig. 3. Electron micrographs of cross sections taken from an adult control tail (A) and the middle of 2-week regenerating blastema (B–D). A: A mature muscle fiber (M) and a region of an adjacent electrocyte (EC) are shown. The electrocyte reveals an amorphous cytoplasm with mitochondria (mit) and one nucleus (n) beneath the plasma membrane (pm). The inset is an enlargement of the region inside the dotted square showing the arrangement of myofibrils (mf) with intervening sarcoplasmic reticulum (SR) and mitochondria (mit) structures next to the plasma membrane (pm) of the fiber. B: Two closely apposed differentiated muscle fibers (M) are adjacent to a developing electrocyte containing at least two nuclei (n). Probable site of fusion of these two muscle fibers is shown by a gap in their basal lamina as indicated by the small arrow in the inset. C: A muscle fiber (M) shows an

25

ultrastructural arrangement of myofibrils with peripheral location of nuclei (inset, n) and mitochondria (inset, small arrows) that is similar to that of control fibers (compare with A). The inset is an enlarged view of a region where gaps in the membranes (large arrows) result in an open connection between the cytoplasm of the muscle fiber and developing electrocyte. A nucleus (n), mitochondria (small arrows), and myofibrils in clusters of different sizes (asterisks) are shown dispersed throughout the cytoplasm of the electrocyte. D: An extension between two muscle fibers indicates fusion of their cytoplasm (see inset). The fiber on the lower left has a cytoplasm packed with myofibrils. In contrast, the fiber on the upper right with three nuclei (n) reveals regions of cytoplasm devoid of myofibrils (asterisk). Scale bars 5 2 µm.

26


Fig. 4. Serial cross sections (12 µm) of a control adult tail showing the same region of muscle fibers stained for acid (pH 5 4.35) and alkaline (Alk; pH 5 10.0) ATPase histochemistry. A population of fibers located closest to the epidermis (e) stained dark with acid-ATPase and light with Alk-ATPase. These fibers are defined as type I fibers (I). A second population of fibers located more centrally (farther from the skin) stained light with acid-ATPase and dark or intermediate with Alk-ATPase. The latter two fiber populations are defined as type II fibers (II). Electrocytes (EC) were not labeled by either acid- or Alk-ATPase histochemistry. Scale bar 5 200 µm.

Muscle fiber types and developing electrocytes in regenerating blastema To identify the emergence of the four adult fiber types, mATPase histochemistry and MHC immunohistochemistry were performed on longitudinal and serial cross sections of blastema 2 and 3 weeks after tail amputation. All myogenic cells in regenerating blastema reacted with antibody MF20, and these cells were observed close to the epidermis in over 90% of the length of the blastema after 2 weeks of regeneration (Fig. 2). Immunohistochemically defined fibers A, B, and C/D were observed in regenerating blastema (Fig. 7). We could not distinguish between fiber types C and D because differences in mATPase activity and in staining intensity with antibody Dev were not always apparent. There was a higher incidence of A and B fibers than of C/D fibers among the population of MF20-positive fibers located most distally in the blastema (Fig. 8). The ratio of

A and B to C/D fibers increased along the proximal to distal axis (Fig. 8). Because the distal region of the blastema is the least differentiated region, these data demonstrate that fibers A and B emerge before fibers C/D. As in adult tails, C/D fibers had a noticeably larger cross-sectional area than A or B fibers (Fig. 8). Furthermore, A fibers were located closest to the epidermis, B fibers were just medial to A fibers, and C/D fibers were found farthest from the epidermis (Fig. 7). Fibers in the blastema with an immunohistochemical profile different from any of the four phenotypes found in the adult tail were not observed. Immunolabeling of MF20-positive cells with anti-SR antibodies 5D2 and 12–101 was generally weak. However, when label was present, it was confined to more centrally located fibers that reacted only with antibodies that discerned C or D fibers like antibody A4.74 (Fig. 9). Hence, the coordination of SR and MHC protein systems found in the adult were also found in regenerating blastema. In addition, MF20 labeling was found in the cytoplasm of developing electrocytes from 2- (Fig. 2) and 3-week blastema. Clusters of myosin immunoreactivity were often seen within electrocytes in the distal half of the blastema (Fig. 10). This MHC labeling pattern is consistent with the clusters of myofibrils revealed within developing electrocytes in electron micrographs (Fig. 10; compare with Fig. 3B–C). In 3-week blastema, the majority (. 90%) of electrocytes were MF20 negative. Few electrocytes (, 10%) were still labeled with MF20, and these were located at the most distal tip. Because our ultrastructural analysis showed that fusion of muscle fibers gives rise to electrocytes (Fig. 3B–D), we wanted to determine the phenotype of fibers that contribute to the formation of electrocytes. Our immunohistochemical analysis showed that developing electrocytes revealed a staining pattern similar to that of C and D muscle fibers of adult tail. For example, antibodies A4.84 (not shown) and N2.261 did not label immature electrocytes (Figs. 7, 8). Antibodies A4.84 and N2.261 very seldom showed any reactivity within developing electrocytes. In contrast, antibodies BF-F3, A4.74, and Dev labeled the cytoplasm of electrocytes and adjacent muscle fibers (Figs. 7, 8). Many developing electrocytes revealed a punctate labeling along the periphery of their cytoplasm with the latter three antibodies, a finding consistent with the close apposition and subsequent fusion of differentiated muscle fibers. Developing electrocytes also showed immunoreactivity with anti-SR antibodies 5D2 and 12–101 (Fig. 9). All developing electrocytes stained darker with 5D2 than with 12–101 (Fig. 9). Whether electrocytes downregulate their expression of these SR and MHC proteins synchronously or not with further maturation was not determined. Thus, as summarized in Figure 11, distinct fiber phenotypes were present in regenerating blastema, and their spatial distribution was established early during regeneration. However, not all fiber phenotypes participated in the formation of electrocytes. Based on both ultrastructural and immunohistochemical analyses, fusion of primarily type II (C and D) muscle fibers gave rise to electrocytes in the regenerating blastema.

DISCUSSION We found that skeletal muscle in the tail of S. macrurus is composed of at least four muscle fiber types and that electrocytes arise primarily from the fusion of a subpopula-


27

Fig. 5. Serial cross sections (12 µm) of a control adult tail showing the same region of muscle fibers reacted with anti-myosin heavy chain (MHC) antibodies A4.84, N2.261, BF-F3, A4.74, and Dev. Their specificity is listed in Table 1. Four fiber phenotypes were discerned by their immunohistochemical staining patterns. The smallest fibers located adjacent to the skin reacted with A4.84, N2.261, and BF-F3 and were labeled A fibers. Fibers located central to A fibers reacted with A4.84 and N2.261 but not with BF-F3, A4.74, or Dev and were labeled B fibers. C fibers were characterized by a negative reaction with A4.84 and N2.261 antibodies but a positive reaction with BF-F3, A4.74, and Dev antibodies. The most centrally located population of

fibers were named D fibers, and these fibers stained darker with Dev than did C fibers. Note that fibers labeled with A4.84 and N2.261 correspond to type I fibers, and fibers labeled exclusively with BF-F3, A4.74, and Dev correspond to type II fibers according to ATPase histochemistry (Fig. 4). Muscle fibers A, B, C, and D were arranged in distinct compartments in the tail with little intermixing. A schematic showing the position of the four fiber phenotypes and their classification based on histochemical (I or II) and immunohistochemical (A, B, C, or D) criteria is shown in the lower right corner. Electrocytes (EC) were not labeled by any anti-MHC antibody used this study. Scale bar 5 200 µm.

tion of muscle fibers. In mature tails, these fiber types are spatially arranged in discrete compartments within muscle bundles located between the electrocytes and the skin, with the smaller type I fibers being more superficial and the larger type II fibers being deeper. During regeneration of the blastema, differentiation of these muscle fiber types takes place with the emergence of type I fibers followed by that of type II fibers along a superficial to deep and proximal to distal gradient. Subsequent fusion of type II fibers precedes the complete disassembly of myofibril organization, a process that gives rise to mature electrocytes.

Muscle fiber types in adult Sternopygus tail We used anti-MHC antibodies raised against myosin from avian and mammalian muscle. However, the specific isotypes of MHC recognized by each of the monoclonal antibodies in S. macrurus are not known. Hence, the number and identification of MHC isoforms remains to be elucidated. Reliable identification of similarities between fish and mammalian MHC isoforms would require knowledge of amino acid and/or gene nucleotide sequences. Nevertheless, these antibodies clearly labeled distinct populations of muscle fibers.

28

G.A. UNGUEZ AND H.H. ZAKON that in other fish. Lastly, the exclusive labeling of type II fibers with anti-SR antibodies 5D2 and 12–101 in S. macrurus is characteristic of type II fibers in other teleost fish (12–101; Devoto et al., 1996; 5D2; Tullis and Block, 1996) and mammalian (5D2; Talmadge et al., 1996) muscles.

Differentiation of fiber types in regenerating blastema

Fig. 6. Serial cross sections (12 µm thick) of normal adult fish muscle stained with monoclonal antibody 12–101 and with anti-MHC antibodies A4.74 and N2.261. Antibody 12–101 is specific to a membrane component of the SR (see Materials and Methods). Type I muscle fibers (A4.74 negative and N2.261 positive) did not react with 12–101, whereas type II muscle fibers (A4.74 positive and N2.261 negative) were labeled with 12–101. Electrocytes (ECs) were not labeled by these antibodies. Scale bar 5 100 µm.

The mATPase activities and immunoreactivity of muscle fibers showed staining patterns similar to those of distinct fiber phenotypes found in muscles of other teleost fish species (Rowlerson et al., 1985; Scapolo et al., 1988; Chayen et al., 1993; Veggetti et al, 1993). For example, relatively small type I fibers and larger type II fibers that label with anti-slow and anti-fast antibodies, respectively, have been reported in various fish species from early development (Scapolo et al., 1988; Veggetti et al., 1993) to adulthood (Scapolo et al., 1988; Chayen et al., 1993; Veggetti et al., 1993). Furthermore, the spatial distribution consisting of type I fibers being closest to the skin and type II fibers being farther from the skin is characteristic of the muscle fiber organization among these fish. Thus, the myotomal organization in S. macrurus tail is similar to

Differentiation of type I muscle fibers appears to occur along the most superficial regions under the skin before deeper myogenic cells begin to differentiate into type II fibers. Co-labeling with anti-MHC antibodies that identified histochemically defined type I and type II fibers exclusively was not seen during regeneration. This suggests that either conversion from one phenotype to another did not occur or conversion occurred at low frequencies. However, the extent to which some fiber types arise during development from the phenotypic conversion of a different fiber type may vary among fish species (Scapolo et al., 1988; Veggetti et al., 1993). Antibodies raised against S. macrurus myosins would provide a more rigorous examination of the developmental MHC transitions during regeneration. Interestingly, mATPase activity appeared consistently only in the mature fish tail. In general, no mATPase activity could be demonstrated in the regenerating blastema even though most of the fibers were seen in electron micrographs to contain many myofibrils. On several occasions, sections from regenerating blastema and adult tail were processed together, and mATPase activity always was observed in adult tissue but only occasionally in blastema. Thus, the absence of histochemical mATPase activity in the blastema cannot be attributed to variability in methods used. Instead, this may be due to an intrinsic feature of the myosins at this early regenerating stage (Scapolo et al., 1988). Previous studies have suggested two contradictory temporal sequences for the generation of different types of muscle fibers in fish (Proctor et al., 1980; van Raamsdonk et al., 1978; Waterman, 1969). A more recent study in which myogenic precursor cells in zebrafish embryos were labeled and their fate examined at various stages of development demonstrated the presence of two distinct populations of precursor cells in the segmental plate (Devoto et al., 1996). The authors concluded that early embryonic myoblasts in zebrafish somites are specified to give rise to particular fiber phenotypes in the adult; one gives rise to slow (type I) and the other to fast (type II) muscle fibers. Thus, although slow fiber differentiation in zebrafish precedes that of fast muscle fibers, differentiation of fast fibers is not a result of fiber type conversion. This interpretation is consistent with studies of avian muscle that suggest that different fiber phenotypes originate during development from distinct myogenic lineages (Miller and Stockdale, 1986). It is tempting to postulate that, similar to zebrafish and avian muscle development, the four fiber phenotypes in tail muscle of S. macrurus derive not from a single myogenic lineage but from several. According to this hypothesis, early myogenic precursor cells would be specified to migrate to a superficial position and differentiate into type I fibers. In contrast, later-developing myogenic precursor cells would be specified to migrate to more central locations and differentiate into type II muscle


29

Fig. 7. Serial cross sections (12-µm thick) near the middle of a 2-week regenerating blastema immunoreacted with anti-MHC antibodies N2.261, BF-F3, A4.74, and Dev. N2.261 labeled small fibers (arrowhead) close to melanocytes (asterisk) and epidermis (e). Few N2.261-positive fibers were not labeled by BF-F3 (arrowhead), demonstrating the presence of A and B fibers in this blastema. Electrocytes

do not react with N2.261. Antibodies BF-F3, A4.74, and Dev labeled fibers located more centrally, demonstrating the presence of C/D fibers. These three antibodies also labeled developing electrocytes (arrows). A4.74 also labels nuclei of unidentified cells. A magnified view (inset) shows labeling of muscle fibers with Dev in the periphery of developing electrocytes. Scale bar 5 100 µm.

fibers. Bromodeoxyuridine (BrdU) labeling experiments in S. macrurus after tail amputation revealed a pool of myogenic reserve cells surrounding muscles and electrocytes at the wound site (Patterson and Zakon, 1993). Examination of the blastema at different times after amputation showed that BrdU-positive cells migrate into the regenerating blastema and subsequently co-express myosin and cytokeratin epitopes, markers of differentiating muscle and electrocytes. Additional experiments will be needed to determine whether these myogenic reserve cells consist of a heterogeneous population of precursor cells at the time they are activated and whether they give rise to muscle fibers of differing phenotypic characteristics.

electrocytes do not arise from myogenic cells at early stages of differentiation (Baillet-Derbin, 1978) but develop from myotubes when they are close to, or fully differentiated. Similar conclusions have been drawn previously from ultrastructural studies on the development of electrocytes in Torpedo (Fox and Richardson, 1978). There were no indications of hypertrophy of individual muscle fibers, in the absence of fusion with other fibers, giving rise to single electrocytes. The mechanisms that trigger the phenotypic conversion of myogenic cells appear to do so primarily by recruiting a select population of muscle fibers to form a syncytium, fuse, and contribute to the formation of electrocytes. To our knowledge, this is the only known instance of fusion among differentiated muscle fibers in vertebrates. Studies on membrane properties and biochemical status of the muscle fibers undergoing fusion should be of interest in relation to the current understanding of mechanisms that limit cell fusion during muscle fiber development. Transdifferentiation in which cells of one differentiated type and function switch to a second discrete identity is thought to be a relatively rare phenomenon. Prominent examples, such as occur during amphibian limb regenera-

Differentiation of electrocytes from skeletal muscle fibers Our ultrastructural analyses showed the fusion of myotubes with a morphology similar to that of adult muscle fibers and the subsequent presence of discrete, highly organized myofibrillar units within developing electrocytes. Eventually, the highly organized myofibrillar units disassemble and diffuse throughout the cytoplasm of the developing electrocyte. These results demonstrate that

30

Fig. 8. Sections (12 µm thick) from three levels (,200 µm apart) along the proximal-distal axis of a 2-week blastema were immunolabeled with anti-MHC antibodies N2.261 (left) and Dev (right). At the most distal regions of the blastema (top), myogenic cells adjacent to the epidermis (e) were labeled with N2.261 but not with Dev. It is noted that in S. macrurus, melanocytes (arrow) always display dark coloration. Furthermore, endogenous peroxidase activity in blood cells (arrowhead) was present. In the mid-region of the blastema, both


N2.261-positive and Dev-positive muscle fibers were observed. N2.261positive fibers were located closer to the epidermis, whereas Devpositive fibers were larger and were located more centrally, farther from the epidermis. In the proximal regions of the blastema, N2.261 label was restricted to muscle fibers close to the epidermis, whereas Dev label was found in more centrally located muscle fibers (arrows) and in developing electrocytes (arrowheads). Note the punctate labeling in the periphery of developing electrocytes. Scale bar 5 100 µm.


Fig. 9. Serial cross sections (12 µm thick) near the middle of a 2-week regenerating blastema immunoreacted with antibody A4.74, specific to IIa/IIx MHC; antibody 5D2, specific to fast Ca21-ATPase SR; and antibody 12–101, specific to a membrane protein of the SR. Only muscle fibers that reacted with A4.74 reacted with 12–101 and 5D2.

31

Unlike electrocytes in the adult, developing electrocytes are labeled by each of these antibodies (arrows). A4.74 also labeled nuclei from unidentified cells. The dark label adjacent to the epithelium (e) and denoted by an asterisk corresponds to melanocytes that always display dark coloration. Scale bar 5 200 µm.

32


Fig. 10. MF20 immunolabeling of a developing electrocyte (arrows) and muscle fibers (m) in a regenerating blastema. MF20 labeling is bright in muscle fibers. Regions of immunoreactivity within the

developing electrocyte (EC) correspond to clusters of myofilaments (see Fig. 4). An adjacent electrocyte is negative for MF20. Labeling with MF20 is not seen in mature ECs. Scale bar 5 100 µm.

tion (Brockes, 1994) and chick retina regeneration, occur in response to injury (Hitchcock and Raymond, 1992). In contrast, the skeletal muscle-to-EO conversion is part of the normal developmental program in Gymnotiform fish, and it is therefore most similar to the esophageal smooth muscle-skeletal muscle transition in mice (Patapoutian et al., 1995) or the modification of muscle cells for functions other than contraction as is found in the Purkinje fibers of vertebrate hearts (Gourdie et al., 1995). There are several cases of fish skeletal muscles being modified for non-muscular functions. The luminous organs (photophores) of the scopelarchid fish Benthalbella infans develop from replacement of the internal components of skeletal muscle fibers with luminescent granules (Johnston and Herring, 1985). Similarly, several groups of oceanic fish, commonly known as billfish, have extraocular muscles (EOMs) that function as a heat-generating organ located beneath the brain and close to the eyes (Block, 1986). The thermogenic organ is composed of modified EOM cells (heater cells) that are structurally distinct from all other types of muscle; myofibrils and contractile filaments are virtually absent, and the cell is packed with mitochondria and smooth membranes (Block, 1986). The derivation of luminous organs and thermogenic organs from skeletal muscle is evident by the presence of contractile proteins and isolated patches of disarrayed myofila-

ments in some of the cells of the adult organs (Johnston and Herring, 1985; Block, 1986, 1991; Bennett, 1971). Interestingly, biochemical studies on two key SR proteins associated with calcium transport, the SR Ca21ATPase and the SR Ca21 release channel, link heater cells to type II muscle fibers based on the expression of similar SR isoforms in both cell types (Block et al., 1994; Tullis and Block, 1996). These findings are consistent with our observations of the predominant restriction of a select fiber type population to transdifferentiate into electrocytes in S. macrurus. Together, these data demonstrate that muscle tissue has a high degree of functional plasticity. Furthermore, the incidence of a higher degree of plasticity among fast muscle fiber phenotypes may be an inherent characteristic maintained across species.

Possible mechanisms regulating phenotypic conversion of muscle fibers to electrocytes The origin and nature of the inductive interactions that regulate transdifferentiation of predominantly type II muscle fibers are unknown. The differential potential of muscle fibers of different phenotypic properties to convert to electrocytes may result from differences in (1) the commitment of myogenic precursor cells, (2) the plasticity of distinct muscle fiber types, and/or (3) the nature of


33

Fig. 11. Schematic showing the histochemical and immunohistochemical criteria used to define each fiber phenotype in tail myotome of S. macrurus (top) and the developmental fate of the four distinct fiber phenotypes in regenerating blastema (bottom). Histochemically identified type I fibers correspond to immunohistochemically classified type A and B fibers. These fiber phenotypes differentiate and do not change their phenotype during further maturation. In contrast, histochemically identified type II fibers, corresponding to immunohis-

tochemically typed fibers C and D, remain as fibers C or D in the adult, or fuse to give rise to electrocytes. During the formation of electrocytes, myofibrils disassemble after fusion of C and D fibers. Clusters of myofibrils from distinct fibers most likely diffuse throughout the cell’s cytoplasm and not segregate as depicted in the cartoon. MHC, myosin heavy chain; Abs, antibodies; SR, sarcoplasmic reticulum; 2, negative; 1, positive label, dark; 11, positive label, very dark.

cell-cell interactions acting upon muscle fibers in different regions. The lineage of distinct myoblast populations in skeletal muscle has been demonstrated for different vertebrate species (Miller and Stockdale 1986; Vivarelli et al., 1988; Devoto et al., 1996). Thus, the temporal emergence of distinct populations of muscle fibers with varying inherent phenotypic characteristics and response capabilities to extrinsic factors may underlie the fate of each cell phenotype and give rise to the spatial compartmentalization of fiber types and transdifferentiation of the most centrally located fiber population. However, the extent to which myogenic precursor cells may already be committed to either muscle or electrocyte is not known. The influence exerted on muscle phenotype by extrinsic factors such as innervation is well established in other vertebrates (Jolesz and Sreter, 1981). Hence, differences in neural input among fiber types in the blastema could induce phenotypic conversion and maintain the electrocytic properties of the cell in the adult. Studies have also provided evidence suggesting that local environmental cues may determine the phenotypic fate of muscle cells in mammals (Rubinstein and Kelly, 1981; McLennan, 1993). Indeed, components of the extracellular matrix have been implicated in the transdifferentiation of cell phenotypes in the jellyfish (Schmid et al., 1993). In addition, the inductive role of transcription factors in regulating cell phenotype and interconversion between different cell types has been well characterized in vitro for several systems: myogenic (Buckingham, 1992), adipogenic (Hu et al., 1995), and neurogenic (Lee et al., 1995). Similar components working alone or in concert may be responsible for controlling the transdifferentiation of type II muscle fibers into electrocytes in S. macrurus. Manipulations of the cellular environment of muscle and EO will

provide information on the influence that extracellular factors, such as innervation, hormones, and growth factors, have on the biochemical and morphological phenotype. The myogenic lineage of the EO during embryonic development and regeneration makes this an ideal system for studying the molecular and cellular mechanisms underlying the regulation of muscle phenotype, with particular emphasis on the unique plasticity of a select population of muscle fibers.

ACKNOWLEDGMENTS We are grateful to the Cell Research Institute of the University of Texas at Austin for the use of its electron microscopy facility and Ying Lu for her technical assistance, Dr. Jane Lubischer for assistance in confocal microscopy, and Drs. Lubischer and Melissa Coleman for critical reading of earlier versions of the manuscript.

LITERATURE CITED Anderson, M.J. and S.G. Waxman (1981) Morphology of regenerated spinal cord in Sternarchus albifrons. Cell Tissue Res. 219:1–8. Bader D., T. Masaki, and D.A. Fischman (1982) Immunochemical analysis of myosin heavy chain during avian myogenesis in vivo and in vitro. J. Cell Biol. 95:763–770 Baillet-Derbin, C. (1978) Cytodifferentiation of the regenerating electrocyte in an electric fish. Biol. Cell. 33:15–24. Bass, A.H. (1986) Electric organs revisited. In T.H. Bullock (ed): Electroreception. New York: John Wiley and Sons, pp. 19–100. Bennett, M.V.L. (1971) Electric Organs. In W.S. Hoar and D.J. Randall (eds): Fish Physiology, Vol. 5. New York: Academic Press, pp. 347–491. Block, B.A. (1986) Structure of brain and eye heater tissue in marlins, sailfish and spearfish. J. Morphol. 190:169–189.

34 Block, B.A. (1991) Evolutionary novelties: How fish have built a heater out of muscle. Am. Zool. 31:726–742. Block, B.A., J. O’Brien, and G. Meissner (1994) Characterization of the sarcoplasmic reticulum proteins in the thermogenic muscles of fish. J. Cell Biol. 127:1275–1287. Brockes, J.P. (1994) New approaches to amphibian limb regeneration. Trends Genet. 10:169–173. Buckingham, M. (1992) Making muscle in mammals. Trends Genet. 8:144–149. Chayen, N.E., A.M. Rowlerson, and J.M. Squire (1993) Fish muscle structure: fibre types in flatfish and mullet fin muscles using histochemistry and antimyosin antibody labelling. J. Muscle Res. Cell Motil. 14:533–542. Devoto, S.H., E. Melancon, J.S. Eisen, and M. Westerfield (1996) Identification of separate slow and fast muscle precursor cells in vivo, prior to somite formation. Development 122:3371–3380. Fox, G.Q. and G.P. Richardson (1978) The developmental morphology of Torpedo marmorata: Electric organ - myogenic phase. J. Comp. Neurol. 179:677–698. Fox, G.Q. and G.P. Richardson (1979) The developmental morphology of Torpedo marmorata: Electric organ - electrogenic phase. J. Comp. Neurol. 185:293–316. Gourdie, R.G., T. Mima, R.P. Thompson, and T. Mikawa (1995) Terminal diversification of the myocyte lineage generates Purkinje fibers of the cardiac conduction system. Development 121:1423–1431. Griffin, K.J., D.M. Fekete, and B.M. Carlson (1987) A monoclonal antibody stains myogenic cells in regenerating newt muscle. Development 101: 267–277. Hitchcock, P.F. and P.A. Raymond (1992) Retinal regeneration. Trends Neurosci. 15:103–108. Hu, E., P. Tontonoz, and B.M. Spiegelman (1995) Transdifferentiation of myoblasts by the adipogenic transcription factors PPAR and C/EBP. Proc. Natl. Acad. Sci. USA 92:9856–9860. Hughes, S.M. and H.M. Blau (1992) Muscle fiber pattern is independent of cell lineage in postnatal rodent development. Cell, 68:659–671. Johnston, I.A. and P.J. Herring (1985) The transformation of muscle into bioluminescent tissue in the fish Benthalbella infans Zugmayer. Proc. R. Soc. Lond. 225:213–218. Jolesz, F. and F.A. Sreter (1981) Development, innervation and activityinduced changes in skeletal muscle. Annu. Rev. Physiol. 43:531–552. Kaprielian, Z. and D.M. Fambrough (1987) Expression of fast and slow isoforms of the Ca21-ATPase in developing chick skeletal muscle. Dev. Biol. 124:490–503. Lee, J.E., S.M. Hollenberg, L. Snider, D.L. Turner, N. Lipnick, and H. Weintraub (1995) Conversion of Xenopus ectoderm into neurons by Neuro D, a basic helix-loop-helix protein. Science. 268:836–844. Mabuchi, K. and F.A. Sreter (1980) Actomyosin ATPase II: Fibre typing by histochemical ATPase reaction. Muscle Nerve 3:233–239. McLennan, I.S. (1993) Localization of transforming growth factor beta 1 in developing muscles: Implications for connective tissue and fiber type pattern formation. Dev. Dynam. 197:281–290. Miller, J.B. and F.E. Stockdale (1986) Developmental regulation of the multiple myogenic cell lineages of the avian embryo. J. Cell Biol. 103:2197–2208. Mills, A., H.H. Zakon, M.A. Marchaterre, and A.H. Bass (1992) Electric organ morphology of Sternopygus macrurus, a wave-type weakly electric fish with a sexually dimorphic EOD. J. Neurobiol. 23:920–932. Patapoutian, A., B.J. Wold, and R.A. Wagner (1995) Evidence for develop-

G.A. UNGUEZ AND H.H. ZAKON mentally programmed transdifferentiation in mouse esophageal muscle. Science 270:1818–1821. Patterson, J.M. and H.H. Zakon (1993) Bromodeoxyuridine labeling reveals a class of satellite-like cells within the electric organ. J. Neurobiol. 24:660–674. Patterson, J.M. and H.H. Zakon (1996) Differential expression of proteins in muscle and electric organ, a muscle derivative. J. Comp. Neurol. 370:367–376. Patterson, J.M. and H.H. Zakon (1997) Transdifferentiation of muscle to electric organ: Regulation of muscle-specific proteins is independent of patterned nerve activity. Dev. Biol. 186:115–126. Proctor, C., P.R.L. Mosse, and R.C.L. Hudson (1980) A histochemical and ultrastructural study of the development of the propulsive musculature of the brown trout, Salmo trutta, in relation to its swimming behaviour. J. Fish Biol. 16:309–329. Rowlerson, A., P.A. Scapolo, F. Mascarello, E. Carpene, and A. Veggetti (1985) Comparative study of myosins present in the lateral muscle of some fish: Species variations in myosin isoforms and their distribution in red, pink and white muscle. J. Muscle Res. Cell Motil. 6:601–640. Rubinstein, N.A. and A.M. Kelly (1981) Development of muscle fiber specialization in the rat hindlimb. J. Cell Biol. 90:128–144. Scapolo, P.A., A. Veggetti, F. Mascarello, and M.G. Romanello (1988) Developmental transitions of myosin isoforms and organization of the lateral muscle in the teleost Dicentrarchus labrax (L.). Anat. Embryol. 178:287–295. Schiaffino, S., L. Gorza, S. Sartore, L. Saggin, S. Ausoni, M. Vianello, K. Gundersen, and T. Lomo (1989) Three myosin heavy chain isoforms in type 2 skeletal muscle fibres. J. Muscle Cell Res. Motil. 10:197–205. Schmid, V., C. Bader, A. Bucciarelli, and S. Reber-Muller (1993) Mechanochemical interactions between striated muscle cells of jellyfish and grafted extracellular matrix can induce and inhibit DNA replication and transdifferentiation in vitro. Dev. Biol. 155:483–496. Talmadge, R.J., R.R. Roy, and V.R. Edgerton (1995) Prominence of myosin heavy chain hybrid fibers in soleus muscle of spinal cord-transected rats. J. Appl. Physiol. 78:1256–1265. Talmadge, R.J., R.R. Roy, G.R. Chalmers, and V.R. Edgerton (1996) MHC and sarcoplasmic reticulum protein isoforms in functionally overloaded cat plantaris muscle fibers. J. Appl. Physiol. 80:1296–1303. Tullis, A. and B. Block (1996) Expression of sarcoplasmic reticulum Ca21-ATPase isoforms in marlin and swordfish muscle and heater cells. Am. J. Physiol. 271:R262–R275. Unguez, G.A. and H.H. Zakon (1996) Transdifferentiation of striated muscle to electric organ in the weakly electric fish Sternopygus macrurus Distinct muscle fiber populations give rise to electrocytes. Am. Soc. Cell Biol. Abstr. 7:42a. van Raamsdonk, W., C.W. Pool, and G. te Kronnie (1978) Differentiation of muscle fiber types in the teleost Brachidanio rerio. Anat. Embryol. Berl. 153:137–155. Veggetti, A., F. Mascarello, P.A. Scapolo, A. Rowlerson, and M.D. Candia Carnevali (1993) Muscle growth and myosin isoform transitions during development of a small teleost fish, Poecilia reticulate (Peters) (Atheriniformes, Poeciliidae): A histochemical, immunohistochemical, ultrastructural and morphometric study. Anat. Embryol. 187:353–361. Vivarelli, E., W.E. Brown, R.G. Whalen, and G. Cossu (1988) The expression of slow myosin during mammalian somitogenesis and limb bud differentiation. J. Cell Biol. 107:2191–2197. Waterman, R.E. (1969) Development of the lateral musculature in the teleost, Brachydanio rerio: A fine structural study. Am. J. Anat. 125:457–493.