Journal of Cell Science 102, 643-652 (1992)Printed in Great Britain (E) The Company of Biologists Limited 1992

643

Formation of highly organized skeletal muscle fibers in vitro

Comparison with muscle development in vivo

SOMPORN SWASDISON and RICHARD MAYNE*

Department of Cell Biology, University of Alabama at Birmingham, UAB Station, Box 302, Birmingham, Alabama 35294, USA

*Author for correspondence

Summary

Two methods were developed in which long-termcultures of quail skeletal muscle were established so thatall of the muscle fibers develop in a highly orientedmanner. The muscle fibers became spontaneously andvigorously contractile and established strong connec-tions with the extracellular matrix at their ends thatclosely duplicate the structure of the myotendinousjunction. A continuous basal lamina was formed aroundeach muscle fiber that contained type IV collagen,laminin and heparan sulfate proteoglycan. With one ofthe methods, an extensive extracellular matrix devel-oped around each muscle fiber that was highly organizedwith the formation of a distinctive epimysium, perimy-

sium and endomysium. Analysis of the cultures by bothmethods for different isoforms of myosin snowedexpression of an adult form of myosin by some of themuscle cells. The results therefore demonstrate thatmuscle development in the present culture systemsproceeds extensively for several weeks. It will now bepossible to investigate directly the structure of theconnections between muscle fibers and the extracellularmatrix.

Key words: muscle fiber formation, collagen types I andIII, basement membrane formation, myosin isoforms.

Introduction

The structure and development of the contractileapparatus of skeletal muscle have been extensivelyinvestigated (Kedes and Stockdale, 1989). However,less is known of the mechanism of attachment of themuscle fibers to the extracellular matrix, which occursat the myotendinous junction (Garrett and Tidball,1988) and also through lateral associations calledcostameres (Shear and Bloch, 1985). There is consider-able evidence that at the myotendinous junctionstructures occur that are similar to those that arepresent in the focal adhesion sites of cells in culture. Atfocal adhesion sites several proteins are known to beinvolved in the attachment of actin filaments to the faintegrin family of transmembrane proteins (reviewed byTurner and Burridge, 1991). In adult muscle, bothvinculin (Shear and Bloch, 1985) and talin (Tidball etal., 1986) are concentrated at the myotendinousjunction as also is the fa chain of the integrin family(Garrett and Tidball, 1988; Bozyczko et al., 1989;Swasdison and Mayne, 1989). The location of a-actininat the myotendinous junction is less clear and has beenreported to be both present (Trotter et al., 1983) andabsent (Tidball, 1987). However, o--actinin is known tointeract with the cytoplasmic domain of the fa integrinchain (Otey et al., 1990) and is present at focal adhesion

sites (Burridge et al., 1988; Pavalko et al., 1991). InDrosophila, a-actinin is required for the attachment offlight muscle to the epithelial tendon cells of the cuticle(Fyrberg et al., 1990). Interestingly, dystrophin alsoappears to be concentrated at the myotendinousjunction (Shimizu et al., 1989; Samitt and Bonilla, 1990;Byers et al., 1991) where it may be functionallyimportant. In support of this, the myotendinousjunction is improperly formed in the mdx mouse, whichlacks dystrophin (Tidball and Law, 1991).

In a previous paper, a cell culture system wasdescribed in which quail skeletal muscle cells differen-tiated into myofibers while entrapped within a collagengel (Swasdison and Mayne, 1991). Such cultures weremaintained for several weeks and became highlycontractile with extensive infolding of the ends of themyofibers when examined by electron microscopy.Attachment of the myofibrils to the end of the musclefiber was found to duplicate closely the structure of themyotendinous junction in vivo. These results thereforesuggest that the formation and structure of themyotendinous junction and its attachment to theextracellular matrix may be studied in cell culture.However, in the previous culture system the musclefibers developed randomly such that an interlockingmeshwork of fibers eventually formed. It thereforeproved difficult to locate the end of an individual fiber,

644 S. Swasdison and R. Mayne

especially by electron microscopy. In this paper, wedescribe two new culture methods in which all of themuscle fibers develop in a highly oriented manner.These culture systems make it possible to analyze theattachment sites at the ends of the muscle fibers andalso to investigate the long-term development ofskeletal muscle in vitro.

Materials and methods

Cell cultureSkeletal muscle cells were obtained from the pectoralismuscle of 10-day-old quail embryos as described previously(Swasdison and Mayne, 1991). The muscle was cleaned ofconnective tissue, minced and mechanically dissociated bypipetting up and down in culture medium (Caplan, 1976). Themedium consisted of minimal essential medium (MEM)

supplemented with 1% antibiotic-antimycotic solution, 1% L-glutamine, 10% horse serum (all from GIBCO, Grand Island,NY) and 5% embryo extract. The muscle cell suspension wasfiltered to give a single cell suspension, which was brieflycentrifuged and resuspended in fresh medium. The cells werediluted to a concentration of 5 x 10s cells/ml before use.Tendon fibroblast cultures were prepared from the legtendons of 17-day-old chick embryos. The tendons werecleaned of skin and muscle, minced and incubated with 0.25%trypsin and 0.1% collagenase (Worthington BiochemicalCorporation, Freehold, NJ) in simple MEM at 37°C for 40min. Enzymatic dissociation was stopped by addition of horseserum (10%) and the cell suspension was filtered to removeundigested pieces of tendon fibers. After centrifugation andwashing once with fresh medium, tendon fibroblasts werediluted to a concentration of 105 cells/ml before use.

Cells were grown initially on the surface of a Teflonmembrane (Bionique Laboratories Inc., Saranac Lake, NY).The membrane was scratched in a parallel manner with fine

1. Scratch a Bionique membranewith fine sand paper and pin itonto 2 small pieces of wax in aculture dish

TENDONFIBROBLASTS.

MYOBLASTS*

WAX

4. Change the medium dailyand maintain the culturefor 4 weeks

5. Muscle fibers contractfreely, detach from themembrane and form amuscle bundle-likestructure

2. Plate skeletal myoblasts in themiddle of the membrane andtendon fibroblasts at both endsof the muscle cells

3. Cover the culture with thefresh medium

4. Maintain the culture for2 days, then transfer to acollagen gel

Remove the pins from bothends of the membrane andtransfer the membrane to aculture dish containing aneutralized collagen solutionso that the cells are on thesurface of polymerizingcollagen gel

Remove the membrane afterseven days and seta secondcollagen gel above the cells.Grow for 3 weeks and changethe medium daily

Fig. 1. Diagram showing the two methods used for long-term growth of oriented skeletal muscle fibers in culture(A) Method without a collagen gel in which the myotubes come off the surface of the Bionique membrane but remainattached to fibroblasts at the ends of the culture. (B) Method in which the culture at two days is transferred to a formingnative collagen gel and, after a further seven days, a second gel is set above the muscle fibers

Skeletal muscle development in vitro 645

Fig. 2. Phase-contrast micrograph of muscle fibers grownwithin a collagen gel as described in method B of Fig. 1.The micrographs was prepared six days after removal ofthe Bionique membrane. Note that all of the muscle fibersare aligned. x72.

sandpaper, cut into 0.4 cm x 1.2 cm strips and cleaned withdetergent. The strips were soaked in ethanol for 1 h beforebeing pinned at both ends onto two small pieces of wax in a 35mm culture dish (see Fig. 1). The culture chamber wasexposed to ultraviolet light for 3 h before use. One drop ofmuscle cell suspension was plated in the middle of thescratched membrane and two drops of tendon fibroblastsuspension were plated at each end of the muscle cells. Theculture was incubated at 37°C in a humidified atmosphere of5% CO2, 95% air for 3 h or until the cells attached to themembrane. During this time the drops did not coalesce. Freshmedium was then added to cover the cultured cells on themembrane, and the medium was changed daily.

In one series of experiments (Fig. 1, method A) themedium was changed daily for four weeks and then the musclecells were prepared for immunofluorescent staining andelectron microscopy. In another series of experiments (Fig. 1,method B) the myoblasts and tendon fibroblasts were grownon the scratched membrane for two days before beingtransferred to a collagen gel. The membrane strip plusattached cells was turned over and transferred to anotherculture dish containing a freshly neutralized collagen solution,so that the cells were located on the surface of a polymerizingcollagen gel. The gel was covered with fresh medium and,after seven days of culture on the gel layer, the membrane wasgently removed leaving muscle and tendon cells attached tothe gel. A second collagen gel was set above the myotubes andthe first gel. Medium was changed daily for three weeks and

Fig. 3. Electron micrograph ofa longitudinal section of fourmuscle fibers from a culturegrown on a scratchedmembrane and transferred to acollagen gel for three weeks.The muscle fibers showextensive development ofsarcomeres and are surroundedby a continuous basal lamina.Note a cell (star) that does notcontain myofilaments and yetappears to be surrounded bythe same basal lamina as themuscle fibers (arrowheads) andis therefore potentially asatellite cell. Also, note thepresence of small cellprocesses (labeled F) that arelacking a basal lamina and arepresumably derived from oneor more fibroblasts. x 10,650.

646 5. Swasdison and R. Mayne

the culture then was prepared for immunofluorescent stainingand electron microscopy.

Indirect immunofluorescence microscopySmall pieces of sample from cultured muscle cells (in thepresence or absence of collagen gel) were rapidly frozen byimmersion in isopentane cooled by liquid nitrogen. Cryosec-tions were cut in longitudinal or cross-section (6-8 fim thick),transferred to gelatin-coated glass slides and air dried. Eachsection was incubated with primary monoclonal antibody for30 min at room temperature in a moist chamber. Sectionswere subsequently washed with phosphate-buffered saline(PBS) and incubated for 30 min with fluorescein-conjugatedgoat anti-mouse IgG (Cappel Laboratories, Cochranville,PA) diluted 1:100 in PBS. After washing with PBS, thesections were mounted with Immu-mount (Shandon, Pitts-burgh, PA) and examined with a Leitz Ortholux II microscopeequipped with an epiilluminator at an excitation wavelengthof 450-490 nm and emission wavelength greater than 515 nm.For control, this primary antibody was substituted with PBS.Under these conditions no fluorescent staining was observed.

The antibodies used in this study were monoclonalantibodies specific for type I and type III collagens (DD4 and3B2, respectively; Swasdison et al., 1992), type IV collagen(1D2; Mayne et al., 1983), a basement membrane form ofheparan sulfate proteoglycan (antibody 33; Bayne et al.,1984), laminin (antibody 31; Bayne et al., 1984), vinculin(VN3-24; Saga et al., 1985), adult fast-twitch myosin heavychain (MHC) (MF1; Obinata et al., 1984), adult MHC(MF14; Bader et al., 1982), "all stages" myosin (MF20; Baderet al., 1982) and cardiac myosin (CCM-52; Clark et al., 1982).

Electron microscopyThe whole culture in the chamber (in either the presence orabsence of a collagen gel) was fixed with 3% glutaraldehyde,1% tannic acid in 0.1 M phosphate buffer (PB), pH 7.3, at 4°Cfor 1 h. Small pieces of fixed muscle were selectively cut fromthe culture and postfixed with 2% osmium tetroxide in PB for45 min. After postfixation, the samples were rinsed exten-sively with 0.1 M PB and dehydrated in a graded ethanolseries and propylene oxide before being embedded in Spurr'sresin (Electron Microscopy Sciences, Ft. Washington, PA).Ultrathin sections were prepared by using an LKB NovaUltratome and stained with 1% uranyl acetate and 0.2% leadcitrate. The sections were examined at 80 kV on a JEOL-100Celectron microscope.

Results

The initial objective of the experiments was to developa cell culture method in which developing muscle fibersare all aligned in parallel with, the ends of the musclefibers forming attachment sites to the extracellularmatrix. The procedures that were consistently success-ful are shown diagrammatically in Fig. 1. Quailmyoblasts (with a small contamination of fibroblasts)were plated onto a scratched Bionique membrane and,after attachment, became aligned along the scratchlines with extensive fusion occurring from 48-72 h. Themyotubes that formed were generally long and thin andeventually began to contract spontaneously. For someexperiments (Fig. 1, method A), muscle and tendon

Fig. 4. Electron micrograph of a muscle fiber grown in a collagen gel for three weeks. Note that the muscle fiber containswell-organized and aligned sarcomeres with a well-developed basal lamina on the surface of the fiber (arrows). Also notethe membranous organelles that arise in culture during development of the T-tubule system (arrowheads). Note theextensive array of collagen fibrils on the surface of the muscle fiber present in both cross and longitudinal arrays, x 14,500.

Skeletal muscle development in vitro 647

BFig. 5. Electron micrographs of longitudinal sections of muscle fibers grown in a collagen gel for 3 weeks. (A) Note theextensive subsarcolemmal electron density (star) and the extensive infolding of the sarcolemma. (B) Note the terminal Zband and the filaments that extend to a subsarcolemmal electron density (arrow). Also, note the collagen fibrils of theextracellular matrix and the banded fibrils that consist of aggregated microfibrils of type VI collagen (arrowheads).A, x 18,167; B, x 13,900.

648 S. Swasdison and R. Mayne

cells were grown on the membrane in the absence of acollagen gel for four weeks. During this time the musclefibers became freely contractile, detached from themembrane and formed a bundle in which both ends ofthe fibers remained attached to the tendon fibroblastsand stabilized by the fixed pins at each end of theculture. With this method, although the muscle fibersunderwent extensive development, it was difficult tolocate the ends of the muscle fibers and to analyze theirstructure. A second method was therefore developed(Fig. 1, method B) in which each muscle-fibroblastculture after two days was transferred to, and eventu-ally embedded within, a gel of native type I collagen.Fig. 2 shows a phase-contrast micrograph of numerousaligned muscle fibers embedded within a collagen gel,which was prepared six days after removal of theBionique membrane. With this method muscle fiberscould be grown within the collagen gel for at least 3weeks during which time strong attachments wereformed to the gel with the ends of each muscle fiberbeing clearly observed by phase-contrast microscopy tocontract and pull on the gel. Electron micrographs wereprepared of a three-week muscle-tendon fibroblastculture in the presence of a collagen gel. Fig. 3 showsfour myofibers that are aligned with each other withlittle connective tissue between the cells and with each

fiber surrounded by a continuous basal lamina. In thiselectron micrograph a cell is also present (labeled with astar) that lacks myofilaments and yet is. apparentlysurrounded by the same basal lamina as the muscle fiber(arrowheads). It is therefore potentially a satellite cell.Each muscle fiber was well differentiated with highlyorganized sarcomeres that filled most of the cell and anextensive array of large mitochondria that werearranged laterally between the Z bands. Fine collagenfibrils were observed in the extracellular matrix andsmall projections of cells without a basal lamina(presumably from fibroblasts and labeled F) were alsopresent. Fig. 4 shows a similar culture at three weeksand at higher magnification. A continuous basal laminahas developed on the surface of the muscle fibers(arrows) with prominent development of sarcomeres.Also note that intracellular membranous organelles arepresent in Fig. 4 (arrowheads), which are considered toarise during the development of the T-system tubules inculture (Ishikawa, 1968; Shiozaki and Shimada, 1990).

Examination of the ends of muscle fibers by electronmicroscopy (Fig. 5A and B) showed extensive infold-ings of the sarcolemma with the presence of acontinuous basal lamina as observed previously(Swasdison and Mayne, 1991). Extensive electron-dense areas were commonly observed in the subsarco-

Fig. 6. Indirectimmunofluorescent staining ofa cross-section of skeletalmuscle grown for four weeksin the absence of a collagengel (method A). (A) Control,no primary antibody;(B) laminin (antibody 31);(C) type IV collagen (antibody1D2); (D) heparan sulfateproteoglycan (antibody 33);(E) type 1 collagen (antibodyDD4); (F) type III collagen(antibody 3B2). Note theprominent basal lamina aroundeach muscle fiber that ispositive for laminin, type IVcollagen and heparan sulfateproteoglycan. Also note theprominent staining forinterstitial collagens withstrong staining for type IIIcollagen in an apparentepimysium (arrow),perimysium (double smallarrow) and endomysium(arrowheads). X165.

Skeletal muscle development in vitro 649

lemma (Fig. 5A) and in Fig. 5B filaments are observedto extend from the last Z band to an extensivesubsarcolemmal structure (arrow). Interestingly, thissite lacked an extensive accumulation of collagen fibrilsin the extracellular matrix. In addition to the collagenfibrils of the extracellular matrix, a second fibrillarsystem (arrowheads) was commonly observed thatrepresents aggregates of type VI collagen microfibrils(Timpl and Engel, 1987).

Analyses were also performed on the cultures grownwithout a collagen gel as shown in Fig. 1A. Cross-sections were prepared from the center of the muscle inwhich all of the fibers were aligned in parallel. Fig. 6shows indirect immunofluorescent staining for laminin(B), type IV collagen (C) and a basement membraneform of heparan sulfate (D). In each instance acontinuous basal lamina was observed delineating eachmuscle fiber. Immunofluorescent staining for type Icollagen (E) and type III collagen (F) showed colocal-ization throughout the extracellular matrix of themuscle bundle. Type III collagen was prominentthroughout the muscle and in the septa of connectivetissue between the muscle fibers, giving rise to anapparent epimysium (arrow), perimysium (doublesmall arrow) and endomysium (arrowheads), indicatingthat the organization of the muscle bundle closelycorrelates to muscle organization in vivo. Using methodB, described in Fig. 1, the muscle cells grow as a sheetonly a few cells thick (see Fig. 7A) and do not form anorganized connective tissue matrix around the fibers.

Experiments were also performed to localize vinculinin four-week muscle cultures grown in a collagen gel(Fig. 7); A shows that in cross-section vinculin is locatedat the surface of each muscle fiber. However, in B theconcentration of vinculin at the end of the muscle fiberwas clearly demonstrated (arrow). Such results are verysimilar to the previously described distribution ofvinculin in muscle fibers in vivo (Shear and Bloch,1985).

Comparisons were also made for the different typesof myosin present in cultures grown either without (Fig.8A, C, E and G) or with a collagen gel (Fig. 8B, D, F,H). The experiments showed that no difference couldbe detected between the cultures in the two differentconditions. For adult fast-twitch myosin heavy chain(antibody MF1, A and B) and for adult myosin heavychain (antibody MF14, C and D) selective staining ofonly some of the muscle fibers was observed. However,for "all stages" myosin heavy chain (antibody MF20, Eand F) and ^-cardiac myosin (antibody CCM-52, G andH) all muscle fibers were stained brightly. This resultshows that at least some of the muscle fibers aresynthesizing adult forms of myosin, as would beexpected to occur in vivo.

Discussion

In the present study methods are described for the long-term culture of aligned quail skeletal muscle fibers suchthat the cells are spontaneously and vigorously contrac-

tile, do not become overgrown by fibroblastic cells anddisplay many of the functional properties of adultskeletal muscle. Of particular interest are the develop-ment of a continuous basal lamina on the surface of themuscle fibers and the development of structures at thetermini of the fibers that are very similar to themyotendinous junction. The development of such aculture system now makes it possible to investigatedirectly in cell culture the mechanism by which a musclefiber interacts with its surrounding extracellular matrix.The methods that we have described align all of themyofibers in one direction and with method B theirtermini are readily accessible to electron microscopy. Inprevious work, alignment of muscle fibers was achievedby mechanically stretching an elastic substratum coatedwith collagen, onto which the cells are plated (Vanden-burgh and Karlisch, 1989; Vandenburgh et al., 1991).Examination of cultures established in the manner byelectron microscopy showed the formation of manyhighly organized muscle fibers with a continuous basal

Fig. 7. Indirect immunofluorescent staining for vinculin ofa four-week-old muscle culture in a collagen gel. (A) Across-section of the gel; and (B) a longitudinal section.Note the presence of vinculin in both the subsarcolemmaof the muscle fibers (A) and also the concentration ofvinculin at the end of the muscle fiber (B, arrow). X200.

650 S. Swasdison and R. Mayne

Fig. 8. Indirectimmunofluorescent staining ofmuscle cultures withmonoclonal antibodiesrecognizing different myosins.(A,C,E and G) Cross-sectionsof muscle cultures (4 weeks) inthe absence of a collagen gel.(B,D,F and H) Cross-sectionsof muscle cultures grown forfour weeks in a collagen gel.(A and B) Fast-twitch myosinheavy chain (antibody MF1);(C and D) adult myosin heavychain (antibody MF14); (E andF) "all stages" myosin heavychain (antibody MF20);(G and H) /3-cardiac myosin(antibody CCM-52). Note thatonly some of the myofibers arestained in A-D but brightstaining for all myofibers isobserved in E-H. x!65.

lamina together with the formation of an extensiveextracellular matrix between the muscle fibers. Anouter capsule of connective tissue and fibroblastssurrounded the muscle fibers (Vandenburgh et al.,1991). A number of unidentified mononuclear cells thatwere extensively vacuolated were also observed. How-

ever, examination of these cultures established bymembrane stretching by electron microscopy failed tolocate the ends of the muscle fibers (H. Vandenburgh,S. Swasdison and R. Mayne, unpublished obser-vations). In another recently described method toobtain oriented muscle fibers, cells were grown on

Skeletal muscle development in vitro 651

Saran wrap that was pinned to Sylgard. After a periodof growth the cells came off the Saran wrap butremained attached at the pins (Strohman et al., 1990).The muscle fibers that developed also appeared to beorganized into an epimysium, a perimysium and anendomysium, and an extensive basal lamina developedaround each muscle fiber. In this study, the appearanceof adult myosin heavy chain (positive staining withmonoclonal antibody MF14) was also observed,although all muscle fibers were also positive for MF20("all stages" myosin). However, with this culturesystem, it is not possible to analyze the ends of themuscle fibers or to investigate the interaction of themuscle fiber with surrounding connective tissue. Long-term growth of skeletal muscle was also recentlyobtained by growth on Matrigel (a basement membraneextract of the EHS mouse tumor) and formation ofadult forms of myosin was described (Hartley andYablonka-Reuveni, 1990). However, no attempt wasmade to orient the fibers in this system and an analysisof the attachment of the cells to the matrix was notpresented.

In all of the culture systems described above,fibroblasts were not removed from the dissociatedmuscle cells, and the results suggest that it is aninteraction between the fibroblasts and the developingmuscle that is responsible for the formation of thehighly organized muscle tissue, especially for culturesgrown without a collagen gel. In previous work, aninteraction between fibroblasts and the developingmyotubes was found to be required for the formation ofa continuous basal lamina (Kiihl et al., 1984; Sandersonet al., 1986), and such interactions are undoubtedlyimportant for the establishment of an intact muscle andits associated connective tissue.

This work was supported by N1H grant AR 37984 and bythe Muscular Dystrophy Association. We thank Dr. KeithBurridge, Dr. Thomas F. Linsenmayer and Dr. Radovan Zakfor providing antibodies, respectively, to vinculin, type Icollagen and cardiac myosin. The MF1, MF14 and MF20monoclonal antibodies were obtained from the DevelopmentStudies Hybridoma Bank maintained by the Department ofPharmacology and Molecular Sciences, Johns HopkinsUniversity School of Medicine, Baltimore, MD, and theDepartment of Biology, University of Iowa, Iowa City, 1A,under contract NO1-HD-6-2915 from the NICHD.

References

Bader, D., Masaki, T. and Fischman, D. A. (1982). Immunochemicalanalysis of myosin heavy chain during avian myogenesis in vivo andin vitro. J. Cell Biol. 95, 763-770.

Bayne, E. K., Anderson, M. J. and Fambrough, D. M. (1984).Extracellular matrix organization in developing muscle:Correlation with acetylcholine receptor aggregates. J. Cell Biol. 99,1486-1501.

Bozyczko, D., Decker, C , Muschler, J. and Horwitz, A. F. (1989).Integrin on developing and adult skeletal muscle. Exp. Cell Res.183, 72-91.

Burridge, K., Fath, K., Kelly, T., Nuckolls, G. and Turner, C. (1988).Focal adhesions: transmembrane junctions between theextracellular matrix and the cytoskeleton. Anna. Rev. Cell Biol. 4,487-525.

Byers, T. J., Kunkel, L. M. and Watkins, S. C. (1991). Thesubcellular distribution of dystrophin in mouse skeletal, cardiacand smooth muscle. J. Cell Biol. 115, 411-421.

Caplan, A. I. (1976). A simplified procedure for preparing myogeniccells for culture. J. Embryol. Exp. Morphol. 36, 175-181.

Clark, W. A., Chizzonite, R. A., Everett, A. W., Rabinowitz, M. andZak, R. (1982). Species correlations between cardiac isomyosins: Acomparison of electrophoretic and immunological properties. J.Biol. Chem. 10, 5449-5454.

Fyrberg, E., Kelly, M., Ball, E., Fyrberg, C. and Reedy, M. C.(1990). Molecular genetics of Drosophila alpha-actinin: mutantalleles disrupt Z disc integrity and muscle insertions. J. Cell Biol.110, 1999-2011.

Garrett, W. and Tidball, J. (1988). Myotendinous junction: structure,function and failure. In Injury and Repair of the MusculoskeletalSoft Tissues (ed. S. L.-Y. Woo and J.A. Buckwalter), pp. 171-207.Park Ridge, Illinois: American Academy of Orthopaedic Surgeons.

Hartley, R. S. and Yablonka-Reuveni, Z. (1990). Long-termmaintenance of primary myogenic cultures on a reconstitutedbasement membrane. In Vitro Cell Develop. Biol. 26, 955-961.

Ishikawa, H. (1968). Formation of elaborate networks of T-systemtubules in cultured skeletal muscle with special reference to the T-system formation. /. Cell Biol. 38, 51-66.

Kedes, L. H. and Stockdale, F. E. (1989). Cellular and MolecularBiology of Muscle Development. New York: Alan R. Liss, Inc.

Kuhl, U., Ocalan, M., Timpl, R., Mayne, R., Hay, E. and Von DerMark, K. (1984). Role of muscle fibroblasts in the deposition oftype IV collagen in the basal lamina of myotubes. Differentiation28, 164-172.

Mayne, R., Sanderson, R. D., Wiedemann, H., Fitch, J. M. andLinsenmayer, T. F. (1983). The use of monoclonal antibodies tofragments of chicken type IV collagen in structural and localizationstudies. J. Biol. Chem. 258, 5794-5797.

Obinata, T., Reinach, F. C , Bader, D. M., Masaki, T., Kitani, S. andFischman, D. A. (1984). Immunochemical analysis of C-proteinisoform transitions during the development of chicken skeletalmuscle. Develop. Biol. 101, 116-126.

Otey, C. A., Pavalko, F. M. and Burridge, K. (1990). An interactionbetween a-actinin and the /31 integrin subunit in vitro. J. Cell Biol.111, 721-729.

Pavalko, F. M., Otey, C. A., Simon, K. O. and Burridge, K. (1991).<x-Actinin: a direct link between actin and integrins. Biochem. Soc.Trans. 19, 1065-1069.

Saga, S., Hamaguchi, M., Hoshino, M. and Kojima, K. (1985).Expression of meta-vinculin associated with differentiation ofchicken embryonal muscle cells. Exp. Cell Res. 156, 45-56.

Samitt, C. E. and Bonilla, E. (1990). Immunocytochemical study ofdystrophin at the myotendinous junction. Muscle and Nerve 13,493-500.

Sanderson, R. D., Fitch, J. M., Linsenmayer, T. F. and Mayne, R.(1986). Fibroblasts promote the formation of a continuous basallamina during myogenesis in vitro. J. Cell Biol. 102, 740-747.

Shear, C. R. and Bloch, R. J. (1985). Vinculin in subsarcolemmaldensities in chicken skeletal muscle: localization and relationship tointracellular and extracellular structures. J. Cell Biol. 101, 240-256.

Shimizu, T., Matsumura, K., Sunada, Y. and Mannen, T. (1989).Dense immunostainings on both neuromuscular and myotendonjunctions with an anti-dystrophin monoclonal antibody. Biomed.Res. 5, 405-409.

Shiozaki, M. and Shimada, Y. (1990). Three-dimensional structure ofintracellular membranous organelles in embryonic chick skeletalmuscle cells in vitro revealed with scanning electron microscope. J.Electron Microsc. 39, 18-25.

Strohman, R. C , Bayne, E., Spector, D., Obinata, T., Micou-Eastwood, J. and Maniotes, A. (1990). Myogenesis and histogenesisof skeletal muscle on flexible membranes in vitro. In Vitro CellDevelop. Biol. 26, 201-208.

Swasdison, S., Mayne, P. M., Wright, D. W., Accavitti, M. A., Fitch,J, M., Linsenmayer, T. F. and Mayne, R. (1992). Monoclonalantibodies that distinguish avian type I and type III collagens:isolation, characterization and immunolocalization in varioustissues. Matrix 12, 56-65.

652 S. Swasdison and R. Mayne

Swasdison, S. and Mayne, R. (1989). Location of the integrin complexand extracellular matrix molecules at the chicken myotcndinousjunction. Cell Tiss. Res. 257, 537-543.

Swasdison, S. and Mayne, R. (1991). In vitro attachment of skeletalmuscle fibers to a collagen gel duplicates the structure of themyotendinous junction. Exp. Cell Res. 193, 227-231.

Tidball, J. G. (1987). Alpha-actinin is absent from the terminalsegments of myofibrils and from subsarcolemmal densities in frogskeletal muscle. Exp. Cell Res. 170, 469-482.

Tidball, J. G. and Law, D. J. (1991). Dystrophin is required fornormal thin filament-membrane associations at myotendinousjunctions. Amer. J. Patliol. 138, 17-21.

Tidball, J. G., O'Halloran, T. and Burridge, K. (1986). Talin atmyotendinous junctions. J. Cell Biol. 103, 1465-1472.

Timpl, R. and Engel, J. (1987). Type VI collagen. In Structure and

Function of Collagen Types (ed. R. Mayne and R.E. Burgcson),pp. 105-143. Orlando: Academic Press.

Trotter, J. A., Eberhard, S. and Samora, A. (1983). Structuralconnections of the muscle-tendon junction. Cell Motil. 3, 431-438.

Turner, C. E. and Burridge, K. (1991). Transmembrane molecularassemblies in cell-extracellular matrix interactions. Curr. Opin.Cell Biol. 3, 849-853.

Vandenburgh, H. H. and Karlisch, P. (1989). Longitudinal growth ofskeletal myotubes in vitro in a new horizontal mechanical cellstimulator. In Vitro 25, 607-616.

Vandenburgh, H. H., Swasdison, S. and Karlisch, P. (1991).Computer-aided mcchanogcncsis of skeletal muscle organs fromsingle cells in vitro. FASEB J. 5, 2860-2867.

(Received 27 February 1992 - Accepted 7 April 1992)