- Email: [email protected]
Desmin at myotendinous junctions
EXPERIMENTAL
CELL
RESEARCH
199,206-212
(1992)
Desmin at Myotendinous Junctions JAMES G. TIDBALL Department
of Kinesiology
Research Center, University and Jerry Lewis Neuromuscular 40.5 Hilgard Avenue, Los Angeles, California 90024-1527
Myofibrils are linked to the cell membrane at myotendinous junctions located at the ends of muscle fibers, and at costameres, sites positioned periodically along lateral surfaces of muscle cells. Both of these sites are enriched in proteins that link active components of myofibrils to the cell membrane. Costameres are also enriched in desmin intermediate filaments that link passive components of myofibrils to the lateral surfaces of muscle cells. In this study, the possibility that desmin is also found between the terminal Z-disk of myofibrils and the myotendinous junction membrane is examined by immunocytochemistry and by KI-extraction procedures. Data presented show that desmin is located in the filamentous core of cellular processes at myotendinous junctions at sites 30 nm or more from the membrane. This core lies deep to subsarcolemmal material previously shown to contain talin, vinculin, and dystrophin. The distance from desmin to the membrane suggests desmin does not interact directly with membrane proteins at the junction. Immunoblots and indirect immunofluorescence of junctional regions of muscle compared to nonjunctional regions show no apparent enrichment of desmin at junctional sites, although vinculin, another costameric and junctional component, is significantly enriched at junctional regions. These findings show that passive elements of myofibrils may be continuous from myotendinous junctions of muscle origin to insertion via desmin filaments located between terminal Z-disks and the junctional membrane. This can provide a system in parallel to that involving thin filaments, vinculin, and talin for linking myofibrils to the cell membrane at myotendinous junctions. o 1992 Academic
Press,
Inc.
INTRODUCTION Current evidence indicates that force generated by skeletal muscle cells is transmitted across the cell membrane at two distinct sites on the cell surface (e.g., [35], for review). One type of site, myotendinous junctions (MTJs), is located at each end of long, cylindrical muscle cells. The second type site, costameres, is distributed 0014-4827/92 $3.00 Copyright 0 1992 by Academic Press, Inc. All rights of reproduction in any form reserved.
206
of California,
periodically along the lateral surfaces of muscle cells with a periodicity identical to that of Z-disks. Force transmission at MTJs is inferred from: (1) their position in series and between force generating myofibrils and tendon collagen fibers, and (2) their enrichment in proteins implicated in mediating thin filamentmembrane association at sites of cell adhesion to extracellular structural molecules. Although costameres lie in parallel to myofibrils and are not, therefore, obvious components in force transmission, they are believed to serve this function in normal muscle because: (1) physiological observations show that damaged muscle fibers are capable of transmitting force from myofibrils laterally to the cell membrane and surrounding extracellular matrix (ECM) [31, 321, (2) electron microscopic observations show filamentous structures attaching myofibrils to the cell membrane at costameres [30], and (3) costameres are also enriched in proteins associated with force transmission from thin filaments to the cell membrane [l, 7, 23, 25, 33, 371. There are many similarities in protein populations at costameres and MTJs. Both sites are enriched in talin [ 371, vinculin [ 23,281, and integrin [ 11, all of which may be links in a chain of proteins that attach intracellular and extracellular structural elements. Costameres have been shown also to contain the intermediate filament protein, desmin [39]. Desmin apparently links Z-disks in adjacent myofibrils to one another and links the most peripheral Z-disks to the cell membrane [25,39]. These observations indicate that desmin can also be a loadbearing protein. In nonmuscle cells, other, nondesmin, intermediate filaments have been shown to interact with the cell membrane via a series of proteins distinct from those associated with actin (e.g., [lo, 17, 211). These intermediate filament-membrane associations occur commonly at some epithelial-epithelial cell junctions (desmosomes) and epithelial-ECM junctions (hemidesmosomes). Thus, hemidesmosomes and MTJs share some functional and structural similarities in that both are specialized sites of cytoskeletal-membrane interactions where cells adhere to extracellular matrix molecules. In this investigation, the possibility that desmin is a
DESMIN
AT MYOTENDINOUS
component of the dense, subplasmalemmal, filamentous mesh located at MTJs is investigated. The findings will permit further evaluation of the analogy between costameres and MTJs and indicate whether muscular contractile force can be transmitted across the MTJ interface via a molecular assembly including desmin in addition to the actin-vinculin-talin-integrin-fibronectin chain previously identified. MATERIALS
AND
METHODS
Electrophoresis and immunolabeling. Semitendinosus muscles from adult frogs (Rana pipiens; either sex) were dissected and placed in 3 n-&f sodium phosphate, pH 7.2, containing 115 mM sodium chloride, 2.5 mMpotassium chloride, and 1.8 mMcalcium chloride. Small groups of muscle cells (-8-12 fibers) were then dissected from the muscle and cleaned of excess connective tissue using fine needles and forceps. The terminal l-2 mm from each end of each fiber was then cut from the fibers and placed in sample buffer containing 80 mM Tris at pH 6.8 containing 0.1 M dithiothreitol, 70 mM sodium dodecyl sulfate (SDS), and 1.0 mM bromphenol blue at 4°C. These terminal regions contain MTJs. The remaining portions of the cylindrical muscle cells that typically measure -8 mm in length and 70 pm in diameter and contain no MTJs were processed separately in sample buffer. Junctional and nonjunctional samples were homogenized, boiled for 1 min, and then centrifuged for 3 min at 10,OOOgto remove insoluble material. The soluble fraction was then applied to SDS-polyacrylamide gels [22] and electrophoresed. Gels contained 10% acrylamide and 0.13% bisacrylamide. Gels were stained with Coomassie blue and the mass of myosin heavy chain (MHC), a prominent band at 200 kDa in each muscle lane, was compared to the mass of MHC in an adjacent lane loaded with molecular weight standard containing 1 pg of MHC. A corrected loading to obtain 4 pg of MHC in each lane was estimated based upon staining density. Gels were then run with corrected loading to provide equivalent quantities of MHC in lanes of junctional and nonjunctional regions of the muscle. Some gels were stained with Coomassie blue while other gels, identically loaded, were transferred to nitrocellulose by electrophoresis [2] at 1.0 A for 3 h at 4°C. The nitrocellulose sheets were washed overnight in 50 mM Tris, pH 7.6, containing 150 mM NaCl, 0.1% NaN, (buffer A) to which 3% bovine serum albumin (BSA), 0.05% Tween-20, and0.2% gelatin were added. The nitrocellulose sheets were then incubated for 90 min at room temperature with rabbit anti-chicken desmin (Sigma, St. Louis, MO) or mouse anti-human vinculin (Sigma). Each primary antibody was diluted 1:lOO in buffer A containing 0.2% gelatin, 0.5% Tween-20, and 5% inactivated horse serum. The nitrocellulose sheets were washed overnight in several changes of buffer A containing 0.2% gelatin and 0.05% Tween-20 and then incubated 90 min in ‘x51, affinity-purified, goat anti-rabbit IgG or iZ51, affinity-purified, goat anti-mouse IgG. These second antibodies were iodinated with “‘1 using the chloramine T method [15] and then diluted to 106cpm/ml in buffer A containing 2% hemoglobin, 0.2% gelatin, and 0.05% Tween-20. The nitrocellulose sheets were then washed overnight in several changes of buffer A, air-dried, and autoradiographed. This experiment was performed three times on three separate animals. Immunohistochemistry. Frog semitendinosus muscles were exposed by dissection, tied to an applicator stick, removed from the hindlimb, and fixed for 20 min in 2% formalin in 150 mM NaCl in 10 mM sodium phosphate buffer (PBS) at pH 7.2. The muscles were rinsed in PBS and frozen in OCT compound immersed in an isopentane bath cooled with liquid nitrogen. The muscles were then sectioned longitudinally in a cryotome at -20°C. Section thickness was 12 gm. The sections were washed in buffer A for 10 min, followed by buffer
207
JUNCTIONS
A containing 0.05% Tween, 0.2% gelatin for 30 min and then buffer A for 15 min. Sections were next incubated in anti-vinculin (1:50) for 2 h at room temperature and then washed three times for 10 min in buffer A. The sections were then incubated for 2 h in anti-desmin (1:50) followed by washes for three times 10 min in buffer A. Sections were covered with TRITC-conjugated, goat anti-mouse IgG (1:60; Sigma), again washed, and covered with FITC-conjugated, goat antirabbit IgG (1:50; Sigma). Following three, lo-min rinses in buffer A and one minute rinse in water, sections were mounted in water miscible mounting medium and viewed by epifluorescence. Control sections were prepared identically except primary antisera were replaced by nonimmunized rabbit antiserum. Electron microscopy and immunocytochemistry. Frog semitendinosus muscles were exposed by dissection and while in situ were tied to an applicator stick at physiological length. The muscles were removed from the hindlimb and immersed in 1.4% glutaraldehyde in 0.2 M sodium cacodylate at pH 7.0 containing 2 mM calcium chloride for 1 h. The muscles were then rinsed in cacodylate buffer and cut into small bundles of muscle cells attached to tendon. Samples used for conventional electron microscopy were then fixed in 1% osmium tetroxide for 30 min, rinsed in buffer, and dehydrated in a series of ethanols. They were embedded in epoxy resin and sectioned longitudinally at 60 nm. Samples used for immunocytochemistry were dehydrated at -20°C in dimethylformamide and then infiltrated with Lowicryl K4M plastic (Polysciences, Warrington, PA). Blocks were polymerized with ultraviolet light for 30 min at -20°C followed by 20 min at 20°C. Longitudinal sections 80 nm thick were cut from the blocks and placed on 200-mesh, nickel grids. Grids were immersed in buffer A containing 1% BSA, 0.05% Tween-20, and 0.2% gelatin for 30 min at 20°C. The grids were then incubated for 11 h at 4°C in (1) anti-desmin diluted in buffer A at 1:50, (2) buffer A only, or (3) serum drawn from a nonimmunized rabbit. Grids were then washed 30 min in buffer A with 1% BSA, 0.05% Tween-20, and 0.2% gelatin and then incubated for 5 h at 4°C in ferritin-conjugated, affinity-purified, goat anti-rabbit IgG (Cappel Laboratories, Cochranville, MD) diluted I:200 in 15 mM sodium phosphate, 150 mM sodium chloride at pH 7.0. The grids were rinsed for 7 h in several changes of buffer A, rinsed 10 min in distilled water, air-dried, and then stained for 15 min in 2% uranyl acetate. Potassium iodide extraction. Thin and thick filaments were extracted from skeletal muscle to facilitate visualization of intermediate filaments. Frog semitendinosus and sartorius muscles were exposed by dissection and tied to applicator sticks at physiological lengths. The muscles were then immersed in 68 mM NaCl, 1.5 mM KCl, 1 mM KH,PO,, 4 mM Na2HP0,, 0.5 mM EGTA, and 50% glycerol at pH 7.5 for 8 weeks at 20°C. The muscles were then transferred to 0.6 M KI, 20 mM Na,S,Os at 4°C for 2 weeks with occasional mixing. They were then fixed and embedded for thin sectioning and conventional electron microscopy. Other, identically extracted muscles were frozen in OCT compound and sectioned as described above for immunohistochemistry and then immunolabeled using antidesmin.
RESULTS
Anti-serum to chick skeletal muscle desmin that was used in this investigation binds specifically to a 54-kDa protein in reduced samples of frog skeletal muscle (Fig. 1) and binds to structures in sections of frog skeletal muscle that are distributed with a periodicity coinciding with Z-disks (Fig. 1). This molecular mass and cellular distribution are characteristic of desmin. Antisera to vinculin bind to a 116-kDa protein in frog skeletal mus-
208
-.__--
JAM&X3
- -_-._ I’IUBALL
ti.
FIG. 1. (A) SDS-PAGE separation of molecular mass standards with mass indicated in kilodaltons: myosin heavy chain (200 kDa); @-galactosidase (116 kDa); phosphorylase (95 kDa); bovine serum albumin (68 kDa); ovalbumin (43 kDa). (B) SDS-PAGE separation of nonjunctional frog semitendinosus muscle. (C) SDS-PAGE separation of junctional region of frog semitendinosus muscle. (D) Autoradiograph of immunoblot of nonjunctional region of frog skeletal muscle treated with anti-desmin and radioiodinated second antibody. (E) Autoradiograph of junctional region of frog skeletal muscle immunoblot treated with anti-desmin and radioiodinated second antibody. (F and G) Autoradiographs of immunoblots of nonjunctional (F) and junctional (G) frog skeletal muscle labeled with anti-vinculin and radioiodinated second antibody. (H) Indirect immunofluorescence of frozen, longitudinal section through frog skeletal muscle cell labeled with anti-desmin and FITC-conjugate second-antibody. Longitudinal axis of cell is oriented vertically. Horizontal bands displaying fluorescence are distributed with same periodicity as Z-disks, as determined by direct comparison of fluorescent image with image seen by Nomarski optics. Bar, 15 am.
cle (Fig. 1) and to the MTJ region of muscle cells (Fig. 2). This mass and distribution are characteristic of vinculin [26]. Double immunolabeling of MTJ regions using anti-desmin and anti-vinculin show both proteins appear present at MTJs (Fig. 2). Comparisons between junctional and nonjunctional regions of the masses of vinculin, desmin, actin, and cY-actinin relative to MHC were made in Coomassie blue-stained gels or in autoradiographic bands of immunoblots. In Fig. 1, all SDS-PAGE lanes are loaded with muscle homogenate containing 4 wg of MHC. A previous investigation [38] has shown that the band at - 110 kDa in frog semitendinosus is cz-actinin. Densitometric scans of Coomassie blue-stained gels show no difference in cy-actinin relative to MHC in junctional versus nonjunctional regions, which is consistent with previous findings showing that c-w-actinin is not enriched at MTJs [38]. Similarly, densitometric scans of stained gels (Fig. 1) show no difference in actin:MHC mass ratio between these regions of the muscle.
Autoradiographs of vinculin immunoblots show a marked increase in vinculin concentration in junctional regions (Fig. 1). This is consistent with previous observations showing that vinculin is enriched at MTJs [l, 281. However, desmin is shown here to display no increased concentration relative to MHC in junctional versus nonjunctional regions. Thus, desmin concentration in muscle scales with myofibrillar proteins (e.g., cr-actinin, actin, MHC) rather than membrane-associated proteins involved in force transmission between myofibrils and the cell membrane (e.g., vinculin). Electron microscopy of the digit-like processes at MTJs shows that these processes contain a dense meshwork of filamentous structures (Fig. 5). However, the dense packing of these components and the extensive crosslinking between filaments makes accurate, filament diameter measurements difficult. Electron microscopic observations of muscle fibers extracted with potassium iodide solutions to remove thin and thick filaments show that most myofibrillar material is removed by extraction (Fig. 3). Z-disks, M lines, and some lo-12 nm diameter filamentous material remain within the extracted sarcomeres. These residual filaments appear linked to one another at the presumed location of Nlines. The digit-like, junctional processes contain little extractable material (Fig. 4). However, even following extraction, measurement of individual filament diameters is not possible. Electron microscopic immunocytochemistry of MTJ regions labeled with anti-desmin reveals extensive labeling of the digit-like processes. Labeling in the processes is primarily 30 nm or more deep to the membrane (Fig. 6) rather than immediately subjacent. Thus, desmin appears to lie at the core of the digit-like processes. No similar labeling was seen on control sections (Fig. 7). Desmin labeling in the MTJ processes is greater than at sites between Z-disks of adjacent myofibrils lying in parallel (data not shown). Attempts using antibody labeling to test whether desmin constituted part of the filamentous material remaining following KI extraction were unsuccessful because KI extraction denatured desmin so that antibody recognition did not occur.
DISCUSSION Previous investigations of MTJ molecular structure (reviewed in [35]) have focused on the identity of proteins associated with the actin cytoskeleton. The possibility that the intermediate filament cytoskeleton may also play a role in force transmission from the terminal sarcomeres of myofibrils to the junctional membrane has not been previously examined, although it could provide an important function in force transmission from tendon to relaxed muscle myofibrils. Observations that show desmin located at costameres [39], where
DESMIN
AT MYOTENDINOUS
JUNCTIONS
209
FIG. 2. Longitudinal, frozen section of frog skeletal muscle and tendon. (A) Nomarski optics image showing interface between tendon (T) and muscle (M) and myotendinous junction (MTJ) of one muscle cell outlined in arrowheads. (B) Same MTJ as shown in A but viewed by epifluorescence optics and filtered to show distribution of anti-vinculin binding. (C) Same MTJ as shown in A but viewed by epifluorescence and filtered to show binding of anti-desmin. Bar, 40 pm.
force may also be transmitted across the cell membrane [31], and other studies that show MTJs and costameres have similar molecular structures [l, 7, 23, 25, 371 support the hypothesis that desmin is an MTJ component. Previous findings indicate that the actin cytoskeleton and intermediate filament cytoskeleton are associated in other cell types. Hubbard and Lazarides [ 141 demonstrated that extensive KI extraction of smooth muscle would not remove all cellular actin until intermediate filament extractions were used. They interpreted this as evidence of an intermediate filament-actin association. In that same study, the investigators showed that gel filtration of desmin in the presence of actin led to some actin and desmin coelution. This also indicates an interaction between the proteins. Finally, those investigators showed that gelation of mixtures of actin and desmin led to the formation of filaments that are labeled with both anti-desmin and anti-actin. These observations indicate that actin cytoskeletons and intermediate filament cytoskeletons may be functionally associated. Morphological data have also supported the possibility that actin and intermediate filaments interact. High voltage electron microscopic studies [26] have shown thin filaments form end-to-side contacts with intermediate filaments in fibroblasts, while double immunolabeling experiments [ll] show thin filament bundles converge on sites rich in intermediate filaments. Furthermore, pharmacologic or toxic disruption of the actin
cytoskeleton can lead to modifications in the intermediate filament cytoskeleton. For example, cytochalasin B treatments can lead to intermediate filament redistribution [6,43], suggesting intermediate filament association with actin. In the present study, it is shown that desmin is codistributed with actin thin filaments within the cellular processes of MTJs. Three morphological criteria for identifying desmin intermediate filaments in muscle are: (1) the structures are lo-nm diameter filaments, (2) the structures are not removed by the KI extraction procedures used in the present study, and (3) they bind antibodies to desmin. Data presented here show that two of these criteria are met by filaments present at MTJs. First, electron microscopic, immunocytochemical data obtained in the present study indicate that desmin is a prominent component of the digit-like processes at MTJs. Second, observations of KI-extracted muscle show that much of the dense material within the digit-like, junctional processes is unaffected by this extraction procedure. Although this is consistent with the presence of desmin in the processes, other nondesmin structural proteins may also remain. For example, Zdisks and M-lines are both discernible following KI-extraction, as are filamentous structures passing from Zdisk to Z-disk in series. However, not all of these filamentous structures linking Z-disks in series are intermediate filaments, if current views of intermediate
210
FIG. M-lines versely FIG. right in retained
JAMES
G. TIDBALL
3. Electron micrograph of longitudinal section of KI-extracted muscle. Myofibrils showing extraction resistant Z-disks (Z) and (M). Most other filamentous material is oriented along longitudinal axes of myofibrils except for some sites (arrows) where transoriented filaments appear to bind longitudinally oriented filaments together. Bar, 0.5 pm. 4. Electron micrograph of longitudinal section of KI-extracted muscle. MTJ with intracellular sites at left and extracellular sites at micrograph. Most myofibrillar material has been extracted although the dense, digit-like processes of the cell at the MTJ (arrows) have their density. Bar, 1.5 Frn.
filament distribution in muscle are accurate. Intermediate filaments have been located between Z-disks in series [39] and in parallel [24], but not within the myofibrils. In the present study, many filaments remaining following KI-extraction lie within the myofibrils. In addition, these fibrils display cross-linking at sites where N-lines would be located in unextracted muscle. This distribution suggests that some of these filaments may consist of titin and nebulin, filamentous proteins believed to have this distribution in myofibrils [42]. The third morphological criterion for identifying desmin filaments, measurement of 10 nm diameter, could not be determined even in KI-extracted muscle due to dense packing of the structures. Immunoblots of junctional and nonjunctional regions of muscle show that although desmin is present at MTJs, it is not apparently enriched at these sites, relative to myosin heavy chain molar concentration. This indicates that the number of desmin filaments involved in mediating attachments between the terminal Z-disk
and the junctional membrane is similar to the number involved in linking Z-disks in series in each nonterminal sarcomere. This is in contrast to other costameric proteins, such as talin [37] and vinculin [1, 281, that are substantially increased in junctional regions compared to nonjunctional. The electron microscopically discernible distribution of desmin at MTJs permits its location relative to other MTJ constituents to be compared. Previous electron microscopic investigations have shown that talin [37], vinculin [28], and dystrophin [5] are all located within the cell processes at MTJs. However, each of these cytoskeleta1 proteins has been demonstrated immediately subjacent to the junctional membrane and is believed to mediate thin filament attachments to the membrane. In contrast, desmin is shown here to be located deep in the cellular processes, within the filamentous core. This location for desmin is consistent with its hypothetical function proposed here in providing attachments between the terminal Z-disk and membrane-associated
DESMIN
FIG. 5. Electron micrograph within the process. Bar, 0.5 pm. FIG. 6. Electron micrograph second antibody. Ferritin grains, membrane. Bar, 0.5 pm. FIG. 7. Electron micrograph second antibody. Bar, 0.5 pm.
of single digit-like
AT MYOTENDINOUS
211
JUNCTIONS
process of muscle cell at MTJ
showing dense assemblage of filamentous
material
lying
of longitudinal section of digit-like, MTJ process of section labeled with anti-desmin and ferritin-conjugated indicating the location of desmin, lie within the process, but not immediately subjacent to the junctional of longitudinal
section of digit-like,
proteins to form a separate force transmitting system in parallel with thin filaments at MTJs. The identity of molecules involved in desmin-membrane attachment at MTJs is unknown. It is possible that hemidesmosomal proteins may serve this role at MTJs. However, much of what is currently known of hemidesmosome molecular composition has been obtained through identification of hemidesmosomal antigens recognized by autoantibodies in bullous pemphigoid (BP). These BP antigens are not known to be located in muscle and there is no evidence that muscle defects occur in BP patients. Other molecules that may link desmin directly or indirectly to the MTJ membrane include integrins. Integrin proteins have been shown capable of mediating transmembrane associations between the actin cytoskeleton and fibronectin [12, 131 and located at MTJs [l, 331. More recently [29, 301, an (r6/p4 integrin complex has been located at hemidesmosomes. It is feasible that dis-
MTJ process labeled with normal rabbit serum and ferritin-conjugated
tinct integrins are involved in forming separate associations of the actin cytoskeleton and intermediate filament cytoskeleton with the MTJ membrane. The functional significance of desmin at MTJs may relate to the mechanical behavior of the junction. Recently [16], the rheological properties of vimentin, another intermediate filament protein, and F-actin were compared using both steady and oscillatory shear loads applied by a torsion pendulum to solutions of the proteins. Although the values for stiffness and loads at failure during shearing are not directly applicable to those for in uiuo loading of desmin or actin filaments, the experimental comparisons suggest provocative differences in the behavior of these proteins during loading. F-actin is extremely stiff in shear loading, but fails at relatively low strain while vimentin filaments withstand large shear strains without failure but display low shear stiffness. It may be relevant to the presence of desmin at MTJs that these sites are expected to experience load-
212
JAMES
G. TIDBALL
ing almost exclusively in shear during muscle contraction [34, 361. Additional mechanical significance would be that the desmin cytoskeleton, if intact from origin to insertion of the cell, would contribute to muscle stiffness and elasticity, even while no actomyosin crossbridges are formed. The continuous intermediate filament cytoskeleton could thereby contribute to the elasticity of passive muscle, measured at - 10% that of active muscle in sinusoidal straining [ 191. A final, possible, functional implication of desmin located at MTJs relates to its putative role in myofibril formation. Results of some investigations indicate that desmin cytoskeleton assembly in straited muscle cells precedes the assembly of other, muscle-specific cytoskeletal structures [9]. This suggests that desmin intermediate filaments may provide a scaffold for assembly. If desmin does provide a scaffold for myofibril assembly, desmin located at MTJs would be functionally interpretable since the MTJs are the primary site of sarcomere addition. Although desmin may be involved in myofibril assembly and sarcomere addition, it is unlikely that it is essential. Recent findings [27] show that myocytes or myotubes, expressing a truncated form of desmin that does not form intermediate filaments, are still capable of myofibril formation. Thus, desmin may function primarily in a mechanical role at the MTJ. This investigation was supported by Grant AR-40343 from the National Institutes of Health and by a grant from the Muscular Dystrophy Association of America, Inc.
12. 13. 14. 15.
194,495-496. 16. 17. 18. 19.
21. 22. 23. 24. 25. 26. 27. 28. 29.
30.
Bozyczko, D., Decker, C., Muschler, J., and Horwitz, A. F. (1989) Exp. Cell Res. 183, 72-91. 2. Burnette, W. N. (1981) Anal. Biochem. 112, 195-203. 3. Burridge, K., and Connell, L. (1983) J. Cell Biol. 97, 359-367. 4. Burridge, K., and Mangeat, P. H. (1984) Nature (London) 308, 744-746. 5. Byers, T. J., Kunkel, L. M., and Watkins, S. C. (1991) J. Cell Biol. 115,411-422. 6. Celis, J. E., Small, J. V., Larsen, P. M., Fey, S. J., and DeMay, J. (1984) Proc. Natl. Acad. Sci USA 81, 1117-1121. 7. Craig, S. W., and Pardo, J. V. (1983) Cell Motil. 3,449-462. 8. Fine, J.-D., Horiguchi, Y., Jester, J., and Couchman, J. R. (1989) J. Invest. Dermatol. 92, 825. 9. Fiirst, D. O., Osborn, M., and Weber, K. (1989) J. Cell Biol. 109, 517-527. 10. Green, K. J., Parry, D. A. D., Steinert, P. M., Virata, M. L. A., Wagner, R. M., Angst, B. D., and Nilles, L. A. (1990) J. Biol. Chem. 265,2603-2612. 11. Green, K. J., Talian, J. C., and Goldman, R. D. (1986) Cell Motil. Cytoskel. 6, 406-418.
31. 32. 33.
Received July 31, 1991 Revised version received November
18, 1991
Janmey, P. A., Euteneuer, U., Traub, P., and Schliwa, M. (1991) J. Cell Biol. 113, 155-160. Jones, J. C. R., and Goldman, R. D. (1985) J. Cell Biol. 101, 506-517. Jones, J. C. R., and Green, K. J. (1991) Curr. Opin. Cell Biol. 3, 127-132. Kawai, M., and Brandt, P. W. (1980) J. Muscle Res. Cell Motil. 1,
279-303. 20.
REFERENCES 1.
Horwitz, A., Duggan, K., Buck, C., Beckerle, M. C., and Burridge, K. (1986) Nature (London) 320,521-533. Horwitz, A., Duggan, K., Greggs, R., Decker, C., and Buck, C. (1985) J. Cell Biol. 101, 2134-2144. Hubbard, B. D., and Lazarides, E. (1979) J. Cell Biol. 80, 166182. Hunter, W. M., and Greenwood, F. C. (1962) Nature (London)
34. 35. 36. 37. 38. 39. 40. 41. 42. 43.
Klatte, D. H., Kurpakus, M. A., Grelling, K. A., and Jones, J. G. R. (1989) J. Cell Biol. 109.337-3390. Kurpakus, M. A., and Jones, J. C. R. (1991) Exp. Cell Res. 194, 139-146. Laemmli, U. K. (1970) Nature (London) 227, 680-685. Pardo, J. V., Siliciano, J. D., and Craig, S. W. (1983) Proc. Natl. Acad. Sci. USA 80, 1008-1012. Price, M., and Sanger, J. (1979) J. Erp. 2001. 208,263-269. Repasky, E. A., Granger, B. L., and Lazarides, E. (1982) Cell 29, 821-833. Schliwa, M., and van Blerkom, J. (1981) J. Cell Biol. 90, 222235. Schultheiss, T., Lin, Z., Ishikawa, H., Zamir, I., Stoeckert, C. J., and Holtzer, H. (1991) J. Cell Biol. 114, 953-966. Shear, C. R., and Bloch, R. J. (1985) J. Cell Biol. 101, 240-256. Sonnenberg, A., Calafat, J., Janssen, H., Daams, H., van der Raaij-Helmer, L. M. H., Falcioni, R., Kennel, S. J., Aplin, J. D., Baker, J., Loizidou, M., and Garrod, D. (1991) J. Cell Biol. 113, 907-918. Stepp, M. A., Spurr-Michaud, S., Tisdale, A., Elwell, J., and Gipson, I. K. (1990) Proc. Natl. Acad. Sci. USA 87, 8970-8974. Street, S. F. (1983) J. Cell. Physiol. 114, 346-364. Street, S. F., and Ramsey, R. W. (1965) Science 149,1379-1380. Swasdison, S., and Mayne, R. (1989) Cell Tiss. Res. 267, 537543. Tidball, J. G. (1983) Cell Motil. 3, 439-447. Tidball, J. G. (1991) J. Biomech. 24(Suppl. l), 37-48. Tidball, J. G., and Daniel, T. L. (1986) Cell Tiss. Res. 245,315322. Tidball, J. G., O’Halloran, T., and Burridge, K. (1986) J. Cell Biol. 103,1465-1472. Tidball, J. G. (1987) Exp. Cell Res. 170,469-482. Tokuyasu, K. T., Dutton, A. H., and Singer, S. J. (1983) J. Cell Biol. 96,1727-1735. Tokuyasu, K. T., Maher, P. A., Dutton, A. H., and Singer, S. J. (1985) Ann. N.Y. Acad. Sci. 455, 200-212. Verando, P., Hsi, B.-L., Yeh, C.-J., Pisani, A., Serieys, N., and Ortonne, J.-P. (1987) Exp. Cell Res. 170, 116-124. Wang, K. (1985) Cell Muscle Motil. 6,315-369. Wolf, K. M., and Mullins, J. M. (1987) Cell Motil. Cytoskel. 7, 347-360.
CELL
RESEARCH
199,206-212
(1992)
Desmin at Myotendinous Junctions JAMES G. TIDBALL Department
of Kinesiology
Research Center, University and Jerry Lewis Neuromuscular 40.5 Hilgard Avenue, Los Angeles, California 90024-1527
Myofibrils are linked to the cell membrane at myotendinous junctions located at the ends of muscle fibers, and at costameres, sites positioned periodically along lateral surfaces of muscle cells. Both of these sites are enriched in proteins that link active components of myofibrils to the cell membrane. Costameres are also enriched in desmin intermediate filaments that link passive components of myofibrils to the lateral surfaces of muscle cells. In this study, the possibility that desmin is also found between the terminal Z-disk of myofibrils and the myotendinous junction membrane is examined by immunocytochemistry and by KI-extraction procedures. Data presented show that desmin is located in the filamentous core of cellular processes at myotendinous junctions at sites 30 nm or more from the membrane. This core lies deep to subsarcolemmal material previously shown to contain talin, vinculin, and dystrophin. The distance from desmin to the membrane suggests desmin does not interact directly with membrane proteins at the junction. Immunoblots and indirect immunofluorescence of junctional regions of muscle compared to nonjunctional regions show no apparent enrichment of desmin at junctional sites, although vinculin, another costameric and junctional component, is significantly enriched at junctional regions. These findings show that passive elements of myofibrils may be continuous from myotendinous junctions of muscle origin to insertion via desmin filaments located between terminal Z-disks and the junctional membrane. This can provide a system in parallel to that involving thin filaments, vinculin, and talin for linking myofibrils to the cell membrane at myotendinous junctions. o 1992 Academic
Press,
Inc.
INTRODUCTION Current evidence indicates that force generated by skeletal muscle cells is transmitted across the cell membrane at two distinct sites on the cell surface (e.g., [35], for review). One type of site, myotendinous junctions (MTJs), is located at each end of long, cylindrical muscle cells. The second type site, costameres, is distributed 0014-4827/92 $3.00 Copyright 0 1992 by Academic Press, Inc. All rights of reproduction in any form reserved.
206
of California,
periodically along the lateral surfaces of muscle cells with a periodicity identical to that of Z-disks. Force transmission at MTJs is inferred from: (1) their position in series and between force generating myofibrils and tendon collagen fibers, and (2) their enrichment in proteins implicated in mediating thin filamentmembrane association at sites of cell adhesion to extracellular structural molecules. Although costameres lie in parallel to myofibrils and are not, therefore, obvious components in force transmission, they are believed to serve this function in normal muscle because: (1) physiological observations show that damaged muscle fibers are capable of transmitting force from myofibrils laterally to the cell membrane and surrounding extracellular matrix (ECM) [31, 321, (2) electron microscopic observations show filamentous structures attaching myofibrils to the cell membrane at costameres [30], and (3) costameres are also enriched in proteins associated with force transmission from thin filaments to the cell membrane [l, 7, 23, 25, 33, 371. There are many similarities in protein populations at costameres and MTJs. Both sites are enriched in talin [ 371, vinculin [ 23,281, and integrin [ 11, all of which may be links in a chain of proteins that attach intracellular and extracellular structural elements. Costameres have been shown also to contain the intermediate filament protein, desmin [39]. Desmin apparently links Z-disks in adjacent myofibrils to one another and links the most peripheral Z-disks to the cell membrane [25,39]. These observations indicate that desmin can also be a loadbearing protein. In nonmuscle cells, other, nondesmin, intermediate filaments have been shown to interact with the cell membrane via a series of proteins distinct from those associated with actin (e.g., [lo, 17, 211). These intermediate filament-membrane associations occur commonly at some epithelial-epithelial cell junctions (desmosomes) and epithelial-ECM junctions (hemidesmosomes). Thus, hemidesmosomes and MTJs share some functional and structural similarities in that both are specialized sites of cytoskeletal-membrane interactions where cells adhere to extracellular matrix molecules. In this investigation, the possibility that desmin is a
DESMIN
AT MYOTENDINOUS
component of the dense, subplasmalemmal, filamentous mesh located at MTJs is investigated. The findings will permit further evaluation of the analogy between costameres and MTJs and indicate whether muscular contractile force can be transmitted across the MTJ interface via a molecular assembly including desmin in addition to the actin-vinculin-talin-integrin-fibronectin chain previously identified. MATERIALS
AND
METHODS
Electrophoresis and immunolabeling. Semitendinosus muscles from adult frogs (Rana pipiens; either sex) were dissected and placed in 3 n-&f sodium phosphate, pH 7.2, containing 115 mM sodium chloride, 2.5 mMpotassium chloride, and 1.8 mMcalcium chloride. Small groups of muscle cells (-8-12 fibers) were then dissected from the muscle and cleaned of excess connective tissue using fine needles and forceps. The terminal l-2 mm from each end of each fiber was then cut from the fibers and placed in sample buffer containing 80 mM Tris at pH 6.8 containing 0.1 M dithiothreitol, 70 mM sodium dodecyl sulfate (SDS), and 1.0 mM bromphenol blue at 4°C. These terminal regions contain MTJs. The remaining portions of the cylindrical muscle cells that typically measure -8 mm in length and 70 pm in diameter and contain no MTJs were processed separately in sample buffer. Junctional and nonjunctional samples were homogenized, boiled for 1 min, and then centrifuged for 3 min at 10,OOOgto remove insoluble material. The soluble fraction was then applied to SDS-polyacrylamide gels [22] and electrophoresed. Gels contained 10% acrylamide and 0.13% bisacrylamide. Gels were stained with Coomassie blue and the mass of myosin heavy chain (MHC), a prominent band at 200 kDa in each muscle lane, was compared to the mass of MHC in an adjacent lane loaded with molecular weight standard containing 1 pg of MHC. A corrected loading to obtain 4 pg of MHC in each lane was estimated based upon staining density. Gels were then run with corrected loading to provide equivalent quantities of MHC in lanes of junctional and nonjunctional regions of the muscle. Some gels were stained with Coomassie blue while other gels, identically loaded, were transferred to nitrocellulose by electrophoresis [2] at 1.0 A for 3 h at 4°C. The nitrocellulose sheets were washed overnight in 50 mM Tris, pH 7.6, containing 150 mM NaCl, 0.1% NaN, (buffer A) to which 3% bovine serum albumin (BSA), 0.05% Tween-20, and0.2% gelatin were added. The nitrocellulose sheets were then incubated for 90 min at room temperature with rabbit anti-chicken desmin (Sigma, St. Louis, MO) or mouse anti-human vinculin (Sigma). Each primary antibody was diluted 1:lOO in buffer A containing 0.2% gelatin, 0.5% Tween-20, and 5% inactivated horse serum. The nitrocellulose sheets were washed overnight in several changes of buffer A containing 0.2% gelatin and 0.05% Tween-20 and then incubated 90 min in ‘x51, affinity-purified, goat anti-rabbit IgG or iZ51, affinity-purified, goat anti-mouse IgG. These second antibodies were iodinated with “‘1 using the chloramine T method [15] and then diluted to 106cpm/ml in buffer A containing 2% hemoglobin, 0.2% gelatin, and 0.05% Tween-20. The nitrocellulose sheets were then washed overnight in several changes of buffer A, air-dried, and autoradiographed. This experiment was performed three times on three separate animals. Immunohistochemistry. Frog semitendinosus muscles were exposed by dissection, tied to an applicator stick, removed from the hindlimb, and fixed for 20 min in 2% formalin in 150 mM NaCl in 10 mM sodium phosphate buffer (PBS) at pH 7.2. The muscles were rinsed in PBS and frozen in OCT compound immersed in an isopentane bath cooled with liquid nitrogen. The muscles were then sectioned longitudinally in a cryotome at -20°C. Section thickness was 12 gm. The sections were washed in buffer A for 10 min, followed by buffer
207
JUNCTIONS
A containing 0.05% Tween, 0.2% gelatin for 30 min and then buffer A for 15 min. Sections were next incubated in anti-vinculin (1:50) for 2 h at room temperature and then washed three times for 10 min in buffer A. The sections were then incubated for 2 h in anti-desmin (1:50) followed by washes for three times 10 min in buffer A. Sections were covered with TRITC-conjugated, goat anti-mouse IgG (1:60; Sigma), again washed, and covered with FITC-conjugated, goat antirabbit IgG (1:50; Sigma). Following three, lo-min rinses in buffer A and one minute rinse in water, sections were mounted in water miscible mounting medium and viewed by epifluorescence. Control sections were prepared identically except primary antisera were replaced by nonimmunized rabbit antiserum. Electron microscopy and immunocytochemistry. Frog semitendinosus muscles were exposed by dissection and while in situ were tied to an applicator stick at physiological length. The muscles were removed from the hindlimb and immersed in 1.4% glutaraldehyde in 0.2 M sodium cacodylate at pH 7.0 containing 2 mM calcium chloride for 1 h. The muscles were then rinsed in cacodylate buffer and cut into small bundles of muscle cells attached to tendon. Samples used for conventional electron microscopy were then fixed in 1% osmium tetroxide for 30 min, rinsed in buffer, and dehydrated in a series of ethanols. They were embedded in epoxy resin and sectioned longitudinally at 60 nm. Samples used for immunocytochemistry were dehydrated at -20°C in dimethylformamide and then infiltrated with Lowicryl K4M plastic (Polysciences, Warrington, PA). Blocks were polymerized with ultraviolet light for 30 min at -20°C followed by 20 min at 20°C. Longitudinal sections 80 nm thick were cut from the blocks and placed on 200-mesh, nickel grids. Grids were immersed in buffer A containing 1% BSA, 0.05% Tween-20, and 0.2% gelatin for 30 min at 20°C. The grids were then incubated for 11 h at 4°C in (1) anti-desmin diluted in buffer A at 1:50, (2) buffer A only, or (3) serum drawn from a nonimmunized rabbit. Grids were then washed 30 min in buffer A with 1% BSA, 0.05% Tween-20, and 0.2% gelatin and then incubated for 5 h at 4°C in ferritin-conjugated, affinity-purified, goat anti-rabbit IgG (Cappel Laboratories, Cochranville, MD) diluted I:200 in 15 mM sodium phosphate, 150 mM sodium chloride at pH 7.0. The grids were rinsed for 7 h in several changes of buffer A, rinsed 10 min in distilled water, air-dried, and then stained for 15 min in 2% uranyl acetate. Potassium iodide extraction. Thin and thick filaments were extracted from skeletal muscle to facilitate visualization of intermediate filaments. Frog semitendinosus and sartorius muscles were exposed by dissection and tied to applicator sticks at physiological lengths. The muscles were then immersed in 68 mM NaCl, 1.5 mM KCl, 1 mM KH,PO,, 4 mM Na2HP0,, 0.5 mM EGTA, and 50% glycerol at pH 7.5 for 8 weeks at 20°C. The muscles were then transferred to 0.6 M KI, 20 mM Na,S,Os at 4°C for 2 weeks with occasional mixing. They were then fixed and embedded for thin sectioning and conventional electron microscopy. Other, identically extracted muscles were frozen in OCT compound and sectioned as described above for immunohistochemistry and then immunolabeled using antidesmin.
RESULTS
Anti-serum to chick skeletal muscle desmin that was used in this investigation binds specifically to a 54-kDa protein in reduced samples of frog skeletal muscle (Fig. 1) and binds to structures in sections of frog skeletal muscle that are distributed with a periodicity coinciding with Z-disks (Fig. 1). This molecular mass and cellular distribution are characteristic of desmin. Antisera to vinculin bind to a 116-kDa protein in frog skeletal mus-
208
-.__--
JAM&X3
- -_-._ I’IUBALL
ti.
FIG. 1. (A) SDS-PAGE separation of molecular mass standards with mass indicated in kilodaltons: myosin heavy chain (200 kDa); @-galactosidase (116 kDa); phosphorylase (95 kDa); bovine serum albumin (68 kDa); ovalbumin (43 kDa). (B) SDS-PAGE separation of nonjunctional frog semitendinosus muscle. (C) SDS-PAGE separation of junctional region of frog semitendinosus muscle. (D) Autoradiograph of immunoblot of nonjunctional region of frog skeletal muscle treated with anti-desmin and radioiodinated second antibody. (E) Autoradiograph of junctional region of frog skeletal muscle immunoblot treated with anti-desmin and radioiodinated second antibody. (F and G) Autoradiographs of immunoblots of nonjunctional (F) and junctional (G) frog skeletal muscle labeled with anti-vinculin and radioiodinated second antibody. (H) Indirect immunofluorescence of frozen, longitudinal section through frog skeletal muscle cell labeled with anti-desmin and FITC-conjugate second-antibody. Longitudinal axis of cell is oriented vertically. Horizontal bands displaying fluorescence are distributed with same periodicity as Z-disks, as determined by direct comparison of fluorescent image with image seen by Nomarski optics. Bar, 15 am.
cle (Fig. 1) and to the MTJ region of muscle cells (Fig. 2). This mass and distribution are characteristic of vinculin [26]. Double immunolabeling of MTJ regions using anti-desmin and anti-vinculin show both proteins appear present at MTJs (Fig. 2). Comparisons between junctional and nonjunctional regions of the masses of vinculin, desmin, actin, and cY-actinin relative to MHC were made in Coomassie blue-stained gels or in autoradiographic bands of immunoblots. In Fig. 1, all SDS-PAGE lanes are loaded with muscle homogenate containing 4 wg of MHC. A previous investigation [38] has shown that the band at - 110 kDa in frog semitendinosus is cz-actinin. Densitometric scans of Coomassie blue-stained gels show no difference in cy-actinin relative to MHC in junctional versus nonjunctional regions, which is consistent with previous findings showing that c-w-actinin is not enriched at MTJs [38]. Similarly, densitometric scans of stained gels (Fig. 1) show no difference in actin:MHC mass ratio between these regions of the muscle.
Autoradiographs of vinculin immunoblots show a marked increase in vinculin concentration in junctional regions (Fig. 1). This is consistent with previous observations showing that vinculin is enriched at MTJs [l, 281. However, desmin is shown here to display no increased concentration relative to MHC in junctional versus nonjunctional regions. Thus, desmin concentration in muscle scales with myofibrillar proteins (e.g., cr-actinin, actin, MHC) rather than membrane-associated proteins involved in force transmission between myofibrils and the cell membrane (e.g., vinculin). Electron microscopy of the digit-like processes at MTJs shows that these processes contain a dense meshwork of filamentous structures (Fig. 5). However, the dense packing of these components and the extensive crosslinking between filaments makes accurate, filament diameter measurements difficult. Electron microscopic observations of muscle fibers extracted with potassium iodide solutions to remove thin and thick filaments show that most myofibrillar material is removed by extraction (Fig. 3). Z-disks, M lines, and some lo-12 nm diameter filamentous material remain within the extracted sarcomeres. These residual filaments appear linked to one another at the presumed location of Nlines. The digit-like, junctional processes contain little extractable material (Fig. 4). However, even following extraction, measurement of individual filament diameters is not possible. Electron microscopic immunocytochemistry of MTJ regions labeled with anti-desmin reveals extensive labeling of the digit-like processes. Labeling in the processes is primarily 30 nm or more deep to the membrane (Fig. 6) rather than immediately subjacent. Thus, desmin appears to lie at the core of the digit-like processes. No similar labeling was seen on control sections (Fig. 7). Desmin labeling in the MTJ processes is greater than at sites between Z-disks of adjacent myofibrils lying in parallel (data not shown). Attempts using antibody labeling to test whether desmin constituted part of the filamentous material remaining following KI extraction were unsuccessful because KI extraction denatured desmin so that antibody recognition did not occur.
DISCUSSION Previous investigations of MTJ molecular structure (reviewed in [35]) have focused on the identity of proteins associated with the actin cytoskeleton. The possibility that the intermediate filament cytoskeleton may also play a role in force transmission from the terminal sarcomeres of myofibrils to the junctional membrane has not been previously examined, although it could provide an important function in force transmission from tendon to relaxed muscle myofibrils. Observations that show desmin located at costameres [39], where
DESMIN
AT MYOTENDINOUS
JUNCTIONS
209
FIG. 2. Longitudinal, frozen section of frog skeletal muscle and tendon. (A) Nomarski optics image showing interface between tendon (T) and muscle (M) and myotendinous junction (MTJ) of one muscle cell outlined in arrowheads. (B) Same MTJ as shown in A but viewed by epifluorescence optics and filtered to show distribution of anti-vinculin binding. (C) Same MTJ as shown in A but viewed by epifluorescence and filtered to show binding of anti-desmin. Bar, 40 pm.
force may also be transmitted across the cell membrane [31], and other studies that show MTJs and costameres have similar molecular structures [l, 7, 23, 25, 371 support the hypothesis that desmin is an MTJ component. Previous findings indicate that the actin cytoskeleton and intermediate filament cytoskeleton are associated in other cell types. Hubbard and Lazarides [ 141 demonstrated that extensive KI extraction of smooth muscle would not remove all cellular actin until intermediate filament extractions were used. They interpreted this as evidence of an intermediate filament-actin association. In that same study, the investigators showed that gel filtration of desmin in the presence of actin led to some actin and desmin coelution. This also indicates an interaction between the proteins. Finally, those investigators showed that gelation of mixtures of actin and desmin led to the formation of filaments that are labeled with both anti-desmin and anti-actin. These observations indicate that actin cytoskeletons and intermediate filament cytoskeletons may be functionally associated. Morphological data have also supported the possibility that actin and intermediate filaments interact. High voltage electron microscopic studies [26] have shown thin filaments form end-to-side contacts with intermediate filaments in fibroblasts, while double immunolabeling experiments [ll] show thin filament bundles converge on sites rich in intermediate filaments. Furthermore, pharmacologic or toxic disruption of the actin
cytoskeleton can lead to modifications in the intermediate filament cytoskeleton. For example, cytochalasin B treatments can lead to intermediate filament redistribution [6,43], suggesting intermediate filament association with actin. In the present study, it is shown that desmin is codistributed with actin thin filaments within the cellular processes of MTJs. Three morphological criteria for identifying desmin intermediate filaments in muscle are: (1) the structures are lo-nm diameter filaments, (2) the structures are not removed by the KI extraction procedures used in the present study, and (3) they bind antibodies to desmin. Data presented here show that two of these criteria are met by filaments present at MTJs. First, electron microscopic, immunocytochemical data obtained in the present study indicate that desmin is a prominent component of the digit-like processes at MTJs. Second, observations of KI-extracted muscle show that much of the dense material within the digit-like, junctional processes is unaffected by this extraction procedure. Although this is consistent with the presence of desmin in the processes, other nondesmin structural proteins may also remain. For example, Zdisks and M-lines are both discernible following KI-extraction, as are filamentous structures passing from Zdisk to Z-disk in series. However, not all of these filamentous structures linking Z-disks in series are intermediate filaments, if current views of intermediate
210
FIG. M-lines versely FIG. right in retained
JAMES
G. TIDBALL
3. Electron micrograph of longitudinal section of KI-extracted muscle. Myofibrils showing extraction resistant Z-disks (Z) and (M). Most other filamentous material is oriented along longitudinal axes of myofibrils except for some sites (arrows) where transoriented filaments appear to bind longitudinally oriented filaments together. Bar, 0.5 pm. 4. Electron micrograph of longitudinal section of KI-extracted muscle. MTJ with intracellular sites at left and extracellular sites at micrograph. Most myofibrillar material has been extracted although the dense, digit-like processes of the cell at the MTJ (arrows) have their density. Bar, 1.5 Frn.
filament distribution in muscle are accurate. Intermediate filaments have been located between Z-disks in series [39] and in parallel [24], but not within the myofibrils. In the present study, many filaments remaining following KI-extraction lie within the myofibrils. In addition, these fibrils display cross-linking at sites where N-lines would be located in unextracted muscle. This distribution suggests that some of these filaments may consist of titin and nebulin, filamentous proteins believed to have this distribution in myofibrils [42]. The third morphological criterion for identifying desmin filaments, measurement of 10 nm diameter, could not be determined even in KI-extracted muscle due to dense packing of the structures. Immunoblots of junctional and nonjunctional regions of muscle show that although desmin is present at MTJs, it is not apparently enriched at these sites, relative to myosin heavy chain molar concentration. This indicates that the number of desmin filaments involved in mediating attachments between the terminal Z-disk
and the junctional membrane is similar to the number involved in linking Z-disks in series in each nonterminal sarcomere. This is in contrast to other costameric proteins, such as talin [37] and vinculin [1, 281, that are substantially increased in junctional regions compared to nonjunctional. The electron microscopically discernible distribution of desmin at MTJs permits its location relative to other MTJ constituents to be compared. Previous electron microscopic investigations have shown that talin [37], vinculin [28], and dystrophin [5] are all located within the cell processes at MTJs. However, each of these cytoskeleta1 proteins has been demonstrated immediately subjacent to the junctional membrane and is believed to mediate thin filament attachments to the membrane. In contrast, desmin is shown here to be located deep in the cellular processes, within the filamentous core. This location for desmin is consistent with its hypothetical function proposed here in providing attachments between the terminal Z-disk and membrane-associated
DESMIN
FIG. 5. Electron micrograph within the process. Bar, 0.5 pm. FIG. 6. Electron micrograph second antibody. Ferritin grains, membrane. Bar, 0.5 pm. FIG. 7. Electron micrograph second antibody. Bar, 0.5 pm.
of single digit-like
AT MYOTENDINOUS
211
JUNCTIONS
process of muscle cell at MTJ
showing dense assemblage of filamentous
material
lying
of longitudinal section of digit-like, MTJ process of section labeled with anti-desmin and ferritin-conjugated indicating the location of desmin, lie within the process, but not immediately subjacent to the junctional of longitudinal
section of digit-like,
proteins to form a separate force transmitting system in parallel with thin filaments at MTJs. The identity of molecules involved in desmin-membrane attachment at MTJs is unknown. It is possible that hemidesmosomal proteins may serve this role at MTJs. However, much of what is currently known of hemidesmosome molecular composition has been obtained through identification of hemidesmosomal antigens recognized by autoantibodies in bullous pemphigoid (BP). These BP antigens are not known to be located in muscle and there is no evidence that muscle defects occur in BP patients. Other molecules that may link desmin directly or indirectly to the MTJ membrane include integrins. Integrin proteins have been shown capable of mediating transmembrane associations between the actin cytoskeleton and fibronectin [12, 131 and located at MTJs [l, 331. More recently [29, 301, an (r6/p4 integrin complex has been located at hemidesmosomes. It is feasible that dis-
MTJ process labeled with normal rabbit serum and ferritin-conjugated
tinct integrins are involved in forming separate associations of the actin cytoskeleton and intermediate filament cytoskeleton with the MTJ membrane. The functional significance of desmin at MTJs may relate to the mechanical behavior of the junction. Recently [16], the rheological properties of vimentin, another intermediate filament protein, and F-actin were compared using both steady and oscillatory shear loads applied by a torsion pendulum to solutions of the proteins. Although the values for stiffness and loads at failure during shearing are not directly applicable to those for in uiuo loading of desmin or actin filaments, the experimental comparisons suggest provocative differences in the behavior of these proteins during loading. F-actin is extremely stiff in shear loading, but fails at relatively low strain while vimentin filaments withstand large shear strains without failure but display low shear stiffness. It may be relevant to the presence of desmin at MTJs that these sites are expected to experience load-
212
JAMES
G. TIDBALL
ing almost exclusively in shear during muscle contraction [34, 361. Additional mechanical significance would be that the desmin cytoskeleton, if intact from origin to insertion of the cell, would contribute to muscle stiffness and elasticity, even while no actomyosin crossbridges are formed. The continuous intermediate filament cytoskeleton could thereby contribute to the elasticity of passive muscle, measured at - 10% that of active muscle in sinusoidal straining [ 191. A final, possible, functional implication of desmin located at MTJs relates to its putative role in myofibril formation. Results of some investigations indicate that desmin cytoskeleton assembly in straited muscle cells precedes the assembly of other, muscle-specific cytoskeletal structures [9]. This suggests that desmin intermediate filaments may provide a scaffold for assembly. If desmin does provide a scaffold for myofibril assembly, desmin located at MTJs would be functionally interpretable since the MTJs are the primary site of sarcomere addition. Although desmin may be involved in myofibril assembly and sarcomere addition, it is unlikely that it is essential. Recent findings [27] show that myocytes or myotubes, expressing a truncated form of desmin that does not form intermediate filaments, are still capable of myofibril formation. Thus, desmin may function primarily in a mechanical role at the MTJ. This investigation was supported by Grant AR-40343 from the National Institutes of Health and by a grant from the Muscular Dystrophy Association of America, Inc.
12. 13. 14. 15.
194,495-496. 16. 17. 18. 19.
21. 22. 23. 24. 25. 26. 27. 28. 29.
30.
Bozyczko, D., Decker, C., Muschler, J., and Horwitz, A. F. (1989) Exp. Cell Res. 183, 72-91. 2. Burnette, W. N. (1981) Anal. Biochem. 112, 195-203. 3. Burridge, K., and Connell, L. (1983) J. Cell Biol. 97, 359-367. 4. Burridge, K., and Mangeat, P. H. (1984) Nature (London) 308, 744-746. 5. Byers, T. J., Kunkel, L. M., and Watkins, S. C. (1991) J. Cell Biol. 115,411-422. 6. Celis, J. E., Small, J. V., Larsen, P. M., Fey, S. J., and DeMay, J. (1984) Proc. Natl. Acad. Sci USA 81, 1117-1121. 7. Craig, S. W., and Pardo, J. V. (1983) Cell Motil. 3,449-462. 8. Fine, J.-D., Horiguchi, Y., Jester, J., and Couchman, J. R. (1989) J. Invest. Dermatol. 92, 825. 9. Fiirst, D. O., Osborn, M., and Weber, K. (1989) J. Cell Biol. 109, 517-527. 10. Green, K. J., Parry, D. A. D., Steinert, P. M., Virata, M. L. A., Wagner, R. M., Angst, B. D., and Nilles, L. A. (1990) J. Biol. Chem. 265,2603-2612. 11. Green, K. J., Talian, J. C., and Goldman, R. D. (1986) Cell Motil. Cytoskel. 6, 406-418.
31. 32. 33.
Received July 31, 1991 Revised version received November
18, 1991
Janmey, P. A., Euteneuer, U., Traub, P., and Schliwa, M. (1991) J. Cell Biol. 113, 155-160. Jones, J. C. R., and Goldman, R. D. (1985) J. Cell Biol. 101, 506-517. Jones, J. C. R., and Green, K. J. (1991) Curr. Opin. Cell Biol. 3, 127-132. Kawai, M., and Brandt, P. W. (1980) J. Muscle Res. Cell Motil. 1,
279-303. 20.
REFERENCES 1.
Horwitz, A., Duggan, K., Buck, C., Beckerle, M. C., and Burridge, K. (1986) Nature (London) 320,521-533. Horwitz, A., Duggan, K., Greggs, R., Decker, C., and Buck, C. (1985) J. Cell Biol. 101, 2134-2144. Hubbard, B. D., and Lazarides, E. (1979) J. Cell Biol. 80, 166182. Hunter, W. M., and Greenwood, F. C. (1962) Nature (London)
34. 35. 36. 37. 38. 39. 40. 41. 42. 43.
Klatte, D. H., Kurpakus, M. A., Grelling, K. A., and Jones, J. G. R. (1989) J. Cell Biol. 109.337-3390. Kurpakus, M. A., and Jones, J. C. R. (1991) Exp. Cell Res. 194, 139-146. Laemmli, U. K. (1970) Nature (London) 227, 680-685. Pardo, J. V., Siliciano, J. D., and Craig, S. W. (1983) Proc. Natl. Acad. Sci. USA 80, 1008-1012. Price, M., and Sanger, J. (1979) J. Erp. 2001. 208,263-269. Repasky, E. A., Granger, B. L., and Lazarides, E. (1982) Cell 29, 821-833. Schliwa, M., and van Blerkom, J. (1981) J. Cell Biol. 90, 222235. Schultheiss, T., Lin, Z., Ishikawa, H., Zamir, I., Stoeckert, C. J., and Holtzer, H. (1991) J. Cell Biol. 114, 953-966. Shear, C. R., and Bloch, R. J. (1985) J. Cell Biol. 101, 240-256. Sonnenberg, A., Calafat, J., Janssen, H., Daams, H., van der Raaij-Helmer, L. M. H., Falcioni, R., Kennel, S. J., Aplin, J. D., Baker, J., Loizidou, M., and Garrod, D. (1991) J. Cell Biol. 113, 907-918. Stepp, M. A., Spurr-Michaud, S., Tisdale, A., Elwell, J., and Gipson, I. K. (1990) Proc. Natl. Acad. Sci. USA 87, 8970-8974. Street, S. F. (1983) J. Cell. Physiol. 114, 346-364. Street, S. F., and Ramsey, R. W. (1965) Science 149,1379-1380. Swasdison, S., and Mayne, R. (1989) Cell Tiss. Res. 267, 537543. Tidball, J. G. (1983) Cell Motil. 3, 439-447. Tidball, J. G. (1991) J. Biomech. 24(Suppl. l), 37-48. Tidball, J. G., and Daniel, T. L. (1986) Cell Tiss. Res. 245,315322. Tidball, J. G., O’Halloran, T., and Burridge, K. (1986) J. Cell Biol. 103,1465-1472. Tidball, J. G. (1987) Exp. Cell Res. 170,469-482. Tokuyasu, K. T., Dutton, A. H., and Singer, S. J. (1983) J. Cell Biol. 96,1727-1735. Tokuyasu, K. T., Maher, P. A., Dutton, A. H., and Singer, S. J. (1985) Ann. N.Y. Acad. Sci. 455, 200-212. Verando, P., Hsi, B.-L., Yeh, C.-J., Pisani, A., Serieys, N., and Ortonne, J.-P. (1987) Exp. Cell Res. 170, 116-124. Wang, K. (1985) Cell Muscle Motil. 6,315-369. Wolf, K. M., and Mullins, J. M. (1987) Cell Motil. Cytoskel. 7, 347-360.