THE ANATOMICAL RECORD PART A 278A:571–578 (2004) Development of a Mammalian SeriesFibered Muscle ANGELIKA C. PAUL,1 PHILIP W. SHEARD,2 AND MARILYN J. DUXSON1* 1 Department of Anatomy and Structural Biology, University of Otago, Dunedin, New Zealand 2 Department of Physiology School of Medical Sciences, University of Otago, Dunedin, New Zealand ABSTRACT This study examines the processes by which multiply innervated, serially ﬁbered mammalian muscles are constructed during development. We previously reported that primary myotubes of such a muscle, the guinea pig sternomastoid muscle, span from tendon to tendon and are innervated at each of the muscle’s four innervation zones. Secondary myotubes form later, in association with each point of innervation (Duxson and Sheard, Dev. Dyn., 1995; 204:391– 405). We now describe the further growth and development of the muscle. Secondary myotubes initially insert onto and grow along the primary myotube. However, as they reach a critical length, they encounter other secondary myotubes growing from serially adjacent innervation zones and may transfer their attachment(s) to these serially positioned secondary myotubes. Other secondary myotubes maintain attachment at one or both ends to their primary myotube. Thus, an interconnected network of primary and secondary myotubes is formed. Patterns of reactivity for cell adhesion molecules suggest that early attachment points between myotubes are the embryonic precursors of adult myomyonal junctions, characterized by the expression of ␣7B␤1 integrin. Finally, the results show that secondary myotubes positioned near a tendon are generally longer than those lying in the mid belly of the muscle, and we suggest that the environment surrounding the tendinous zone may somehow stimulate myotube growth. Anat Rec Part A 278A:571–578, 2004. © 2004 Wiley-Liss, Inc. Key words: skeletal muscle; development; multiple innervation; secondary myotubes; intrafascicularly terminating ﬁbers; alpha 7 integrins; myomyonal junction The architecture of many larger mammalian skeletal muscles differs from the standard textbook view in that they have multiple endplate bands along the length of their fascicles, the majority of individual ﬁbers are shorter than the fascicles, and ﬁbers terminate intrafascicularly (Loeb et al., 1987; Paul, 2001). Intrafascicular ﬁber terminations (IFTs) are characterized by elaborate folding of the sarcolemma and high expression levels of cell-matrix adhesion molecules, in particular the ␣7␤1 integrins and dystrophin (Young et al., 2000; Paul et al., 2002). From the terminal sarcolemma of the IFT, ﬁbers of the extracellular matrix extend to contact other ﬁbers terminating in a complementary manner to form myomyonal junctions. These junctions may involve entire terminating ﬁbers, or multiple complementary branches of neighboring muscle ﬁbers (Torigoe and Nakamura, 1987; Young et al., 2000), and link muscle ﬁbers within each fascicle into a complex network, quite unlike anything seen in smaller muscles. We have previously characterized aspects of the development, functional anatomy, and physiology of such muscles © 2004 WILEY-LISS, INC. (Duxson and Sheard, 1995; Sheard et al., 1999; Young et al., 2000; Paul, 2001; Paul et al., 2002). Here we ask how the interlinking network of short ﬁbers and myomyonal junctions that characterizes these muscles is formed during development. Angelika C. Paul’s present address is Department of Cellular, Molecular, and Developmental Biology, University of Colorado, Boulder, CO 80309. *Correspondence to: Marilyn J. Duxson, Department of Anatomy and Structural Biology, School of Medical Sciences, University of Otago, P.O. Box 913, Dunedin 9001, New Zealand. Fax: 64-3-479-7254. E-mail: email@example.com Received 23 April 2003; Accepted July 11 2003 DOI 10.1002/ar.a.20020 572 PAUL ET AL. Fig. 1. Comparison of development of a singly innervated muscle (left) with that of a multiply innervated, series-ﬁbered muscle (right). The positions of innervation zones are indicated by arrows in each case. Primary myotube formation is similar in both muscle types (top), but primary myotubes of multiply innervated muscles are each multiply innervated. Secondary myotubes are initiated in association with all innervation zones, so that multiply innervated muscles have multiple foci of secondary myotube formation (middle). As secondary myotubes grow, they will all extend to the tendons in the singly innervated muscle. In contrast, most secondary myotubes will form intrafascicularly terminating ﬁbers in the multiply innervated muscle (bottom). The cellular pattern of embryogenesis of multiply innervated, serially ﬁbered muscles shows many features in common with development of simple muscles with a single zone of innervation, but also some crucial differences (Fig. 1). In both muscle types, the muscle is initially formed by a generation of primary myotubes that extend the entire muscle length, deﬁne the fascicular architecture, and become innervated by the motor nerve (Fig. 1, top). However, in the multiply innervated muscles, each primary myotube receives innervation at multiple sites spread evenly along the muscle length, rather than at a unique centrally located site as in the classical small muscle (Duxson et al., 1989; Duxson and Sheard, 1995). At the next stage of development, secondary myotubes form exclusively in association with sites of innervation on the primary myotubes in both muscle types (Fig. 1, middle). In the singly innervated muscle, this means that new secondaries are associated only with the single centrally located innervation zone of the primary myotube. However, in the multiply innervated muscle, sites of secondary myotube initiation are staggered along most of the length of the fetal muscle, corresponding to the multiple innervation sites on the primary myotubes (Duxson et al., 1989; Duxson and Sheard, 1995). Finally, as secondary myotubes elongate, differences again arise. In the singly innervated muscle, all secondary myotubes will extend fully to the tendon and insert directly onto it, whereas secondary myotubes in the multiply innervated muscle will terminate within the muscle belly at one or both ends, forming the characteristic series-ﬁbered arrangement of these muscles (Fig. 1, bottom). The question then is, how does this staggered array of short ﬁbers form into a coherent network that can effectively transmit force through the muscle? We hypothesized that in the developing multiply innervated muscle, secondary myotubes might make overlapping contacts with each other, with some or all of these sites of myotube-to-myotube contact being maintained to form the myomyonal junctions seen in adult serially ﬁbered muscles. It is well known that developing singly innervated muscles form transient junctional complexes between primary and secondary myotubes characterized by interfolding of the sarcolemmae and expression of a range of cell-cell adhesion molecules (Duxson and Usson, 1989; Rosen et al., 1992; Fredette et al., 1993; Kaufmann et al., 1999), but these are lost during maturation. Here, we investigate whether similar sites of myotube interaction occur in developing multiply innervated muscles and whether these involve secondary-to-secondary myotube contacts as well as primary-to-secondary contacts. We also compare the adhesion molecules expressed at sites of developmental myotube contact with those expressed at mature myomyonal junctions. Continuity of these expression patterns would be consistent with a developmental relationship between the two sites. MATERIALS AND METHODS Tissues for Immunohistochemistry Guinea pig sternomastoid muscles were collected from 2– 4 animals at each of embryonic day 37 (E37), E47, postnatal day 1 (PN1), and from mature adults. This age range covers the period from formation to maturity of secondary myotubes in this muscle (Duxson and Sheard, 1995). Dated pregnancies were the result of immediate postpartum matings between stably coupled pairs. Euthanasia of adults and neonates was by overdose of sodium pentobarbitone; of embryos, by cold anesthesia followed by decapitation, in accord with protocols approved by the University of Otago Committee on the Ethical Use of Animals in Research. Transmission Electron Microscopy (TEM) A semiserial section electron microscopic (EM) analysis was made of a single E37 muscle. The embryo was deeply anesthetized on ice, then the head ﬁxed in place to maintain the sternomastoid muscles at resting length. Fixation was by cardiac perfusion with 1% glutaraldehyde, 1% paraformaldehyde, and 0.05 M glucose in a 0.1 M phosphate buffer (pH 7.3). After perfusion, muscles were excised and immerse-ﬁxed in the same solution for a further 2 hr, postﬁxed in 2% osmium tetroxide, stained en bloc in 2% aqueous uranyl acetate for 45 min, then dehydrated in alcohols and propylene oxide before embedding in TAAB epoxy resin. Semiserial sectioning extended over about 3/4 of the muscle length (about 5 mm) with a group of 10 ultrathin (90 –100 nm) sections being collected onto Formvar-coated slot grids every 9 m and subsequently stained with lead citrate and uranyl acetate. Semiserial photomontages of ﬁve adjacent and easily identiﬁed myotube clusters, taken over the entire extent of the section series, were used to construct a longitudinal map identifying the extent, origin, and insertion of every myotube within the selected clusters [some results reported in Duxson and Sheard (1995)]. Outlines of selected regions of each cluster were subsequently traced into a 3D reconstruction program, using a digitizing pad, to produce 3D-rendered illustrations showing relationships between developing myotubes. DEVELOPMENT OF A SERIES-FIBERED MUSCLE Immunohistochemistry Immunohistochemical techniques were used to examine the presence or absence of selected integrins or their ligands and of cadherins and N-CAM at myomyonal junctions at each developmental stage. Techniques were essentially identical to those described previously (Paul et al., 2002). The integrin subunits examined were those we have previously localized to adult myomyonal junctions, namely, the ␣7A and ␣7B subunits, plus the ␣6 subunit, which we have not previously investigated. Antibodies used were rabbit polyclonal anti-␣7A (␣7CDA2) and anti␣7B (␣7CDB1) (Song et al., 1993; Martin et al., 1996) and a rat monoclonal anti-␣6 integrin (GoH3) (Immunotech, Marseille, France). A monoclonal antitenascin antibody (M1-B4) (Chiquet and Fambrough, 1984) was obtained from the Developmental Studies Hybridoma Bank maintained by the Department of Biological Sciences, University of Iowa (Iowa City, IA 52242), under contract NO1HD-7-3263 from the National Institute for Child Health and Development (NICHD). To look for the presence of cadherins, we used a mouse monoclonal antibody raised against ␤-catenin (15B8) kindly gifted by K. Knudson (Johnson et al., 1993). The anti-N-CAM antibody was a rabbit polyclonal against rat N-CAM (AO49) from the laboratory of Urs Rutishauser. The monoclonal mouse antislow-myosin heavy-chain antibody (NOQ7.1.1A) was from Dr. Robyn Fitzsimons (Harris et al., 1989). Sections were observed with an Olympus microscope and digital ﬂuorescent or bright ﬁeld images taken with a Pixera PVC 100C digital camera (Pixera). Two-color images were produced by merging within Adobe Photoshop 5.5. RESULTS Our initial aim was to understand the sequence of relationships between new secondary myotubes and their primary and secondary myotube neighbors as they form and then extend in length within the belly of a multiply innervated muscle. Transmission Electron Microscopy We used semiserial EM analysis to visualize the relationships between developing myotubes in the multiply innervated guinea pig sternomastoid muscle at embryonic day 37 (E37). This time point falls within the period of very active formation and growth of new secondary myotubes, so that the muscle contains a complete continuum of size of secondary myotube, ranging from absolutely new to almost fully grown. New myotubes appeared in association with each of the muscle’s four innervation zones, as previously reported (Duxson and Sheard, 1995). As secondary myotubes associated with a particular primary myotube elongated, those initiated at one innervation zone soon overlapped longitudinally with those initiated at adjacent innervation zones. Three-dimensional reconstruction revealed complex associations between growing secondary myotubes and their parent primary myotube and, later, between overlapping or adjacent secondary myotubes. Some examples of these associations are illustrated in the reconstruction and accompanying electron micrographs of Figure 2. Figure 3 shows a more quantitative analysis in which all the secondary myotubes sectioned and photographed from end to 573 end (37 secondary myotubes in total) are logged in terms of their length and the nature of their insertion. In their initial stages of growth (length of 1– 400 m; 2– 8 nuclei), young secondary myotubes formed complex interfolding adhesive junctions with the primary myotube along most of their length, just as previously seen in small muscles such as the rat lumbrical (Duxson et al., 1989). An example of such a very young myotube is cell 8 which is illustrated in Figure 2 at section level E only. This was a myotube of 350 m and 7– 8 nuclei and showed complex interfolding with the primary myotube, similar to that seen in Figure 2E, along most of its length. In older/longer secondary myotubes, these interfolded junctional zones became conﬁned to the cell extremities, and there was a less intimate relationship with the primary in the central zone. Cell 7 (Fig. 2D–F) is an example of a more developed secondary myotube, which extended over 1,100 m along the axis and formed a complex insertion onto the primary myotube near both its ends (Fig. 2D and F), while in its central zone (Fig. 2E) it was in simple apposition to the primary myotube. A crucial change occurred once secondary myotubes reached a length of 1,100 –1,400 m. At this length, a proportion of them released their insertion onto the parent primary, at one or both ends, and instead formed a close interdigitating insertion on a neighboring secondary myotube. One such pair of secondary myotubes is illustrated in Figure 2. Cell 9 (in blue) is a secondary myotube centered on an innervation point at 1,760 m along the longitudinal axis (top arrow) and extends down along the primary to 310 m, where it terminates. Cell 3 (in green) is centered on the innervation zone at 430 m on the axis and extends up the primary to 1,970 m before terminating. In Figure 2B–F, the relationship between these two cells can be seen in section at selected levels. In Figure 2B, cell 9 initially appears as an independent myotube, close to, but not inserted onto, the primary mytoube. At the next level (Fig. 2C, 1,600 m), the beginning of cell 3 is seen. It inserts deeply into cell 9 over about 100 m of axial length. The two cells then continue side by side, but without any intimate junctional contact (Fig. 2D, 860 m) until the end of cell 9 is approached (Fig. 2E, 360 m). At this level, an insertion of secondary to secondary again occurs, but in a reciprocal direction, so that the terminal portion of cell 9 inserts cell processes into the substance of cell 3. Finally, cell 3 continues alone after the termination of cell 9 (Fig. 2F). The quantitative analysis of secondary myotube insertion (Fig. 3) demonstrates the initial formation and growth of secondary myotubes on the primary myotube (squares, lower line), followed by transfer of the insertion of a proportion of secondaries onto other developing secondaries (diamonds, upper line of plot). The data show ﬁrst that all new secondary myotubes arise exclusively on the surface of the primary myotubes, never purely in association with an existing secondary myotube. Secondly, the length at which some secondaries transfer their insertions is signiﬁcant; 1,100 –1,400 m is the range of separation between adjacent innervation zones in the muscle at this developmental stage. Thus, it is the length at which secondary myotubes initiated at neighboring zones on the same primary myotube will ﬁrst encounter each other along the longitudinal axis. It seems to be a matter of chance whether a particular myotube retains insertion on the parent primary myotube or encounters a suitable 574 PAUL ET AL. Fig. 2. A cluster of developing myotubes from E37 guinea pig sternomastoid muscle. A: Reconstruction of part of a primary myotube (red/orange) and some of its associated secondary myotubes (all other colors) from semiserial (10 m interval) electron micrographs. The longitudinal axis is shown in m at left. Solid arrows at left indicate the levels of two innervation zones that lie within the region of reconstruction. Of particular interest are cell 9 (blue, centered on the innervation zone at 1,760 m and terminating at 310 m) and cell 3 (green, centered on the innervation zone at 430 m and terminating at 1,700 m). These two well-grown secondaries form complementary insertions onto each other, close to their respective points of termination. Cell 4 (gray) is a large secondary myotube that extends throughout the reconstructed region. Note that some myotubes (in particular myotubes 7 and 8, visible in D–F) were removed from the reconstruction for clarity. A–F: Electron micrographs from selected levels of the reconstruction showing details of some myotube-myotube relationships. Note particularly the interdigitating insertions between myotubes 3 and 9 close to their respective terminations at section levels C and E. Myotube 7 is a small myotube (1,100 m long) seen ﬁrst at level D and terminating shortly after level F; it inserts on the surface of the primary myotube. Myotube 8 (350 m in length) is seen at section level E only. Myotubes referred to in the text have been artiﬁcially outlined to allow resolution of the sarcolemmae at this low magniﬁcation. Cell numbers correspond to those in A. DEVELOPMENT OF A SERIES-FIBERED MUSCLE other secondary myotube and transfers its insertion. In the present study, about half of the secondary myotubes over 2,400 m in length continued to use the primary myotube as their main substrate for growth and as their sole insertion. Figure 3 also shows that very few secondary myotubes were found within the length range of 1,400 –2,100 m. Possible explanations for this gap in the length distribution of secondary myotubes are considered in the discussion. In summary, this TEM study shows that developing secondary myotubes in this series-ﬁbered muscle ﬁrst grow on and maintain insertion on the primary myotube. As they reach a length where they encounter other extending secondaries, a proportion transfer their insertion at one or both ends to these contemporaries. Others retain all insertion on the primary myotube. A proportion of both types will ﬁnally extend into the tendon zone, and insert at one end on the tendon, and at the other end on a primary or secondary myotube. This process results in formation of an interconnected network of primary and secondary myotubes, with complex interdigitations between myotubes taking the place of myomyonal junctions in the adult. Immunohistochemistry We next examined the cell adhesion molecules expressed at the interfaces between developing myotubes to see how these related to those expressed at adult myomyonal junctions. In the adult guinea pig sternomastoid muscle, we have previously reported strong immunoreactivity for ␣7B integrin and weak immunoreactivity for ␣7A integrin at myomyonal junctions of adult guinea pig sternomastoid muscle, but found no evidence for localization of N-CAM or cadherins (Paul et al., 2002). The developmental stages examined immunohistochemically were embryonic day 37 (E37), i.e., early secondary myotube generation (corresponding to the EM study above); E47, i.e., late secondary myotube development; postnatal day 1 (PN1), i.e., maturing muscle ﬁbers with few myotubes remaining; and adult. In the fetal material, a slow-myosin heavy-chain antibody (antislowMyHC, NOQ7.1.1A) was in some instances used to identify myotube type, as only primary myotubes react positively to the antibody in fetal sternomastoid muscle (Duxson and Sheard, 1995). Integrin Expression At E37, the anti-␣7B integrin antibody localized to the cytoplasm of myotubes, and at a modest level to selected interfaces between E37 myotubes. The arrow in Figure 4B indicates one reactive secondary-secondary myotube interface. At E47, irregular but intense staining was more frequently present at myotube interfaces, particularly those between adjacent secondary myotubes (arrow, Fig. 4E), while cytoplasmic expression had faded. Generalized labeling of the sarcolemma of both slow and fast ﬁbers was present by P1 (Fig. 4G and H), and there were localized regions of more extensive sarcolemmal staining at myotube interfaces (arrow in Fig. 4G), reminiscent of the pattern seen at adult myomyonal junctions viewed in crosssection. Blood vessels were also labeled by the anti-␣7B integrin antibody: the bright dots surrounding the muscle ﬁbers in Figure 4G–I are capillaries. In adult sternomas- 575 toid, the anti-␣7B integrin antibody localized to the entire sarcolemma of all muscle ﬁbers, with localized enhancement at sites of myomyonal junctions (e.g., arrow in Fig. 4I), as we have previously reported. Thus, the pattern of immunolocalization of the ␣7B integrin subunit suggests a continuity in identity between early myotube interfaces and adult myomyonal junctions. In developing sternomastoid muscle, we saw no sign of reactivity with the anti-␣7A integrin antibody at myotube interfaces. Embryonic day 37 myotubes showed some cytoplasmic localization but this faded by E47, with localization to myomyonal junctions visible only in the adult [embryonic data not shown; see Paul et al. (2002) for adult]. Finally, we examined expression of the ␣6 integrin subunit, as its presence has been reported during formation of myotubes (Bronner-Fraser et al., 1992), and of the ␣9 integrin subunit, which is characteristically expressed at sites of tension transmission such as mature tendons and tendon primordia. We screened for ␣9 integrin using an antibody to its only known extracellular ligand, tenascin (Wang et al., 1995; Yokosaki et al., 1998). No evidence was found for localization of either ␣6 integrin or tenascin at intrafascicular ﬁber terminations at any time, although tenascin always localized to ﬁber terminations at the tendon (results not shown). Cell-Cell Adhesion No cell-cell adhesion molecules have been found at adult myomyonal junctions (Paul et al., 2002), but their presence is widely reported at myotube surfaces and interfaces during development of singly innervated muscles. We wondered if they would be present at sites of primarysecondary and secondary-secondary myotube interaction in the embryonic series-ﬁbered muscle. We screened for cadherins using an antibody against their most widespread intracellular ligand, ␤-catenin. Myotube type was here identiﬁed by size and position rather than with the antislow-MyHC antibody due to problems with antibody cross-reactivity. During early secondary myogenesis (E37, Fig. 5A and B), ␤-catenin was localized over most myotube surfaces but enhanced at both primary-secondary (arrowhead in Fig. 6B) and secondary-secondary (arrow) myotube interfaces, supporting the previously reported role of cadherins in early myotube adhesion. At E47 (Fig. 5C and D), it was hard to identify primary myotubes based on their size, but ␤-catenin expression was much reduced at all myotube interfaces (the arrow in Fig. 5C and D indicates one myotube interface with slight immunoreactivity) and was absent by P1 and at adult myomyonal junctions (not shown). The strong expression visible in Figure 5D was associated with the developing capillary bed, rather than myotubes. Immunoreactivity to the N-CAM antibody showed an initial pattern of expression similar to ␤-catenin, being prominently expressed at virtually all myotube interfaces as well as on the free surfaces of myotubes at E37 (Fig. 6A and B). However, by E47, staining at many myotube interfaces was speciﬁcally reduced (e.g., arrow in Fig. 6D). With further maturation, N-CAM expression on myotubes was generally downregulated and largely restricted to satellite cells at P1 and later (Fig. 6E and F), consistent with previous reports of N-CAM expression patterns in developing muscle (Fredette et al., 1993). N-CAM reactiv- 576 PAUL ET AL. ity was never seen at interfaces between muscle ﬁbers at adult myomyonal junctions. DISCUSSION Fig. 3. A plot to show the nature of insertion of secondary myotubes of various lengths. The plot includes all secondary myotubes traced from end to end in the serial section study. The bottom line (squares) shows secondary myotubes that insert entirely on the primary myotube. The top line (diamonds) shows secondary myotubes that insert on other secondaries, at one or both ends. Secondary myotubes that extend into the tendon region are marked with an asterisk. Fig. 4. Localization of ␣7B integrin during development of guinea pig sternomastoid muscle. Primary myotubes are identiﬁed by antislow-MyHC antibody in red, with ␣7B integrin shown in green. A–C: Embryonic day 37. A: The primary myotube of a cluster, identiﬁed by its reaction for slowMyHC. B: Diffuse ␣7B cytoplasmic is present in all myotubes with occasional stronger sarcolemmal staining at interfaces between secondary myotubes (arrow). C: Overlay of A and B. D–F: Embryonic day 47. D: AntislowMyHC-positive primary myotubes. E: Cytoplasmic staining for ␣7B is Multiply innervated, series-ﬁbered muscles have a physical architecture very different to that of simple muscles. Many previous studies have led to a good descriptive understanding of how the form of simple muscles develops. An initial, rather sparse, framework of primary myotubes lays out the skeleton of the muscle and provides a substrate on which a much larger number of secondary myotubes will grow. The latter are initiated as binucleate cells near the central innervation zone of the primary myotubes, then elongate through to the tendon, where they form attachments to the ECM and separate entirely from the primary myotube to form independent contractile units (Duxson and Usson, 1989; Duxson et al., 1989). This current study shows that more complex muscles use the reduced, but sarcolemmal staining at myotube interfaces is more widespread (arrows). F: Overlay of D and E. G and H: Postnatal day 1. G: ␣7B integrin is present on the sarcolemma of all young muscle ﬁbers, with areas of more extensive staining (arrow) perhaps corresponding to forming myomyonal junctions. H: Overlay of slow-MyHC immunoreaction. I: Adult. ␣7B is localized on capillaries (bright dots) and all regions of the sarcolemma, with intense staining (arrow) indicating regions of myomyonal junctions (L). Scale bars ⫽ 50 m. DEVELOPMENT OF A SERIES-FIBERED MUSCLE 577 Fig. 5. Localization of ␤-catenin during development (B and D) compared with differential interference contrast (DIC) images (A and C). A and B: Embryonic day 37. ␤-catenin reactivity is present at primarysecondary (arrowhead) and secondary-secondary (arrow) myotube interfaces. C and D: Embryonic day 47. Strong ␤-catenin reactivity is now associated mainly with the developing capillary bed, although occasional secondary-secondary interfaces still react weakly (arrows in C and D). Scale bars ⫽ 50 m. same basic principles of development, but work several variations on this basic scheme. In multiply innervated muscles, primary myotubes again form a simple scaffold, predicting the muscle form and attachments. However, secondary myotubes are initiated at many rather than at a single innervation site on the primary myotubes (Duxson and Sheard, 1995). This present study shows that secondary myotubes at ﬁrst elongate on the surface of the primary myotube, but soon encounter myotubes growing in the opposite direction from adjacent innervation zones. At this point, many of them form connections onto other secondary myotubes and transfer one or both of their terminal points of insertion away from the primary myotube. Not all secondary myotubes transfer away from the primary myotube, however. At the stage observed in this serial EM study, about half of all the longer (⬎ 1,400 m) secondary myotubes maintained connection with the primary myotube at both ends, while others inserted at one end on the primary and at the other end on a secondary myotube. Thus, an interconnecting network of secondary and primary myotubes is slowly built, in contrast to the progressive production of independent parallel myotubes in simple muscles. What is the relation between the sites of insertion of developing myotubes onto each other and the myomyonal junctions that connect adult ﬁbers in serially organized muscles? Our studies of the cell adhesion molecules characterizing these two structures suggest developmental continuity, and therefore possible identity, between the two. Adult myomyonal junctions are most easily characterized by their high speciﬁc levels of expression of the ␣7B integrin subunit so this is a key marker. Early myotubes (E37) showed signiﬁcant widespread expression of cell-cell adhesion molecules (␤-catenin/cadherins and NCAM) over the whole sarcolemma and at myotube inter- Fig. 6. Localization of N-CAM (B, D, F) with corresponding differential interference contrast (DIC) images (A, C, E). A and B: Embryonic day 37. N-CAM immunoreactivity is ubiquitously present on myotube surfaces. C and D: Embryonic day 47. N-CAM is still widely present on the sarcolemma, but some myotube interfaces seem to show speciﬁcally reduced immunostaining (arrow). E and F: Postnatal day 1. N-CAM immunoreactivity restricted to satellite cells (arrow). Scale bars ⫽ 50 m. faces, while ␣7B integrin was detected only at select interfaces, generally between more mature secondary myotubes (e.g., Fig. 4C). As development progressed, ␤-catenin expression was generally lost from the sarcolemma and that of N-CAM was speciﬁcally downregulated at myotube interfaces, whereas strong expression of ␣7B integrin continued to appear at a selection of myotube interfaces (Fig. 4F). At postnatal day 1 (by which most myotubes are mature in the guinea pig), select areas of close apposition between adjacent myotubes/muscle ﬁbers showed intense reactivity for ␣7B integrin, but none at all for the cell-cell adhesion molecules tested. This expression pattern is typical of that seen at adult myomyonal junctions (Paul et al., 2002). Thus, physical attachment between young myotubes in guinea pig sternomastoid muscle is generally facilitated by a dynamically modulated pattern of expression of cell-cell adhesion molecules, as previously shown by other authors in simple muscle systems (Knudsen et al., 1990a, 1990b; Fredette et al., 1993; Irintchev et al., 1994). However, as myotubes in the multiply innervated muscle begin to mature, some of these sites begin to express the ␣7B␤1D integrin, which binds myotube to myotube across the developing basal lamina. Given their relative scarcity, we suggest that the sites expressing ␣7B␤1D integrin likely correspond to the sites of more specialized myotube-to-myotube attachment 578 PAUL ET AL. found near the ends of the developing myotubes, and that these sites mark the beginnings of the formation of the permanent myomyonal junctions of adult series-ﬁbered muscles. A remaining puzzle is the presence of a discontinuity in the length-distribution of the growing secondary myotubes. Figure 3 shows that only a single myotube was found within the length range of 1,400 –2,300 m. We do not believe this is a sampling error, as the serial section study would tend to select against the inclusion of complete, very long myotubes, rather than excluding those of middle length. One possible explanation is that end-to-end fusion of extending myotubes might sometimes occur. For example, two myotubes initiated from adjacent innervation zones might each grow to 1,200 m, at which point they encounter each other and fuse, producing one myotube of 2,400 m. Unfortunately, close analysis does not support this idea. Of the 16 myotubes above the critical length of 2,400 m, 10 of them overlap only a single innervation zone, and so could not possibly have formed by fusion of two separately initiated myotubes. An alternative explanation arises from the observation that nearly 80% (10/13) of the myotubes longer than 2,300 m extend into the zone where the tendon is forming, distal to the last innervation zone (Fig. 3, data points marked with a star). We suggest that the length of secondary myotubes may be inﬂuenced by the region of the muscle in which they grow. Myotubes that lie partly within the tendon may be stimulated to grow faster, or for a more extended period, whereas myotubes with two intrafascicular terminations might slow their growth prematurely. Our own observations of ﬁber length in adult guinea pig sternomastoid muscles would support the idea of a growth-stimulating effect of the tendon zone; in whole-mount immunohistochemical studies, ﬁbers with one end in a tendon and the other terminating intrafascicularly were always signiﬁcantly longer than ﬁbers terminating intrafascicularly at both ends (data not shown). In summary, this study traces the relationships between developing myotubes in a series-ﬁbered muscle and shows how they evolve into the physically interconnected network of adult muscle ﬁbers. A highly dynamic system is apparent, where early myotube interactions are dominated by cell-cell adhesion, whereas adult ﬁbers are connected by junctions bridging through the extracellular matrix. Finally, there is evidence suggesting differential rates of myotube growth in the region of the tendon versus the mid belly of the muscle. ACKNOWLEDGMENTS The authors thank Ms. Judy Rodda and the staff of EMTech for their capable technical assistance with these studies. The work was supported in part by grants from the Health Research Council of New Zealand, Lottery Health, and the Deans’ Fund of the University of Otago Medical School. LITERATURE CITED Bronner-Fraser M, Artinger M, Muschler J, Horwitz AF. 1992. Developmentally regulated expression of alpha 6 integrin in avian embryos. Development 115:197–211. Chiquet M, Fambrough DM. 1984. Chick myotendinous antigen: I, a monoclonal antibody as a marker for tendon and muscle morphogenesis. J Cell Biol 98:1926 –1936. Duxson MJ, Usson Y. 1989. Cellular insertion of primary and secondary myotubes in embryonic rat muscles. Development 107:243–251. Duxson MJ, Usson Y, Harris AJ. 1989. The origin of secondary myotubes in mammalian skeletal muscles: ultrastructural studies. Development 107:743–750. Duxson MJ, Sheard PW. 1995. Formation of new myotubes occurs exclusively at the multiple innervation zones of an embryonic large muscle. Dev Dyn 204:391– 405. Fredette B, Rutishauser U, Landmesser L. 1993. Regulation and activity-dependence of N-cadherin, NCAM isoforms, and polysialic acid on chick myotubes during development. J Cell Biol 123:1867– 1888. Harris AJ, Fitzsimons RB, McEwan JC. 1989. Neural control of the sequence of expression of myosin heavy chain isoforms in foetal mammalian muscles. Development 107:751–769. Irintchev A, Zeschnigk M, Starzinskipowitz A, Wernig A. 1994. Expression pattern of M-cadherin in normal, denervated, and regenerating mouse muscles. Dev Dyn 199:326 –337. Johnson KR, Lewis JE, Li D, Wahl J, Peralta Soler A, Knudsen KA, Wheelock MJ. 1993. P- and E-cadherin are in separate complexes in cells expressing both cadherins. Exp Cell Res 207:252–260. Kaufmann U, Martin B, Link D, Witt K, Zeitler R, Reinhard S, Starzinski-Powitz A. 1999. M-cadherin and its sisters in development of striated muscle. Cell Tissue Res 296:191–198. Knudsen KA, McElwee SA, Myers L. 1990a. A role for the neural cell adhesion molecule, NCAM, in myoblast interaction during myogenesis. Dev Biol 138:159 –168. Knudsen KA, Myers L, McElwee SA. 1990b. A role for the Ca2(⫹)dependent adhesion molecule, N-cadherin, in myoblast interaction during myogenesis. Exp Cell Res 188:175–184. Loeb GE, Pratt CA, Chanaud CM, Richmond FJR. 1987. Distribution and innervation of short, interdigitated muscle ﬁbers in parallelﬁbered muscles of the cat hindlimb. J Morphol 191:1–15. Martin PT, Kaufman SJ, Kramer RH, Sanes JR. 1996. Synaptic integrins in developing, adult, and mutant muscle: selective association of alpha1, alpha7A, and alpha7B integrins with the neuromuscular junction. Dev Biol 174:125–139. Paul AC. 2001. Muscle length affects the architecture and pattern of innervation differently in leg muscles of mouse, guinea pig, and rabbit compared to those of human and monkey muscles. Anat Rec 262:301–309. Paul AC, Sheard PW, Kaufman SJ, Duxson MJ. 2002. Localization of alpha7 integrins and dystrophin suggests potential for both lateral and longitudinal transmission of tension in large mammalian muscles. Cell Tissue Res 308:255–265. Rosen GD, Sanes JR, LaChance R, Cunningham JM, Roman J, Dean DC. 1992. Roles for the integrin VLA-4 and its counter receptor VCAM-1 in myogenesis. Cell 69:1107–1119. Sheard PW, McHannigan P, Duxson MJ. 1999. Single and paired motor unit performance in skeletal muscles: comparison between simple and series-ﬁbered muscles from the rat and the guinea pig. Basic Appl Myol 9:79 – 87. Song WK, Wang W, Sato H, Bielser DA, Kaufman SJ. 1993. Expression of alpha 7 integrin cytoplasmic domains during skeletal muscle development: alternate forms, conformational change, and homologies with serine/threonine kinases and tyrosine phosphatases. J Cell Sci 106:1139 –1152. Torigoe K, Nakamura T. 1987. Fine structure of myomyomous junctions in the mouse skeletal muscles. Tissue Cell 19:243–250. Wang A, Patrone L, Mcdonald JA, Sheppard D. 1995. Expression of the integrin subunit alpha 9 in the murine embryo. Dev Dyn 204: 421– 431. Yokosaki Y, Matsuura N, Higashiyama S, Murakami I, Obara M, Yamakido M, Shigeto N, Chen J, Sheppard D. 1998. Identiﬁcation of the ligand binding site for the integrin alpha9 beta1 in the third ﬁbronectin repeat of tenascin-C. J Biol Chem 273:11423–11428. Young M, Paul A, Rodda J, Duxson M, Sheard P. 2000. Examination of intrafascicular muscle ﬁber terminations: implications for tension delivery in series-ﬁbered muscles. J Morphol 245:130 –145.