The composition and organization of myofibrils at extra-junctional membrane attachment sites in cultured neonatal rat cardiac muscle cells were analysed by immunofluorescence and electron microscopy. When myofibril terminals attached to the cell membrane via focal contacts at regions of the sarcolemma that lacked intercalated discs, they appeared to be non-striated and resembled thick actin cables. Although the non-striated terminals contained actin, myosin and alpha-actinin, the proteins were not organized into recognizable sarcomeres at the light microscopic level. Analysis of the structure of the terminals in the electron microscope confirmed that the usual sarcomeric organization and attachments to the sarcolemma were markedly modified. The non-striated myofibril terminals differed in structure from both stress fibres in non-muscle cells and stress fibre-like structures present in embryonic heart cells in culture. Non-striated myofibril terminals attached to the cell membrane by lateral contact with extra-junctional electron-dense membrane plaques rather than by insertion by their ends into the fascia adherens. It is proposed that the structure and composition of membrane-attachment points for myofibrils may have an influence on the structure, organization or stability of contractile elements in cardiac muscle.

The structure and function of the contractile machinery in muscle cells have logically lead to the proposal that similar structures and mechanisms would be found to underlie motile activities in non-muscle cells. In fact, a number of the contractile proteins found in muscle have also been found in non-muscle cells (for review, see Fischman, 1986). For example, stress fibres in non-muscle cells have been found to be somewhat analogous to myofibrils in muscle cells. Stress fibres are composed of bundles of filamentous actin that contain alpha-actinin and myosin arranged in periodic arrays along their length, and which resemble miniature sarcomeres (Gordon, 1978). At the ultrastructural level alpha-actinin is localized in regularly spaced dense bodies and myosin is present in the intervening spaces along the microfilament bundle (Langanger et al. 1986). Although non-muscle cells contain isotypes of alpha-actinin and myosin that are distinct from those in muscle cells (for review, see Fischman, 1986) myosin thick filaments do not appear to be present, the general organization of these proteins in the stress fibre is remarkably similar to their arrangement in the sarcomeres of myofibrils.

Structures resembling stress fibres have been noted in skeletal and cardiac muscle cells in culture (Dlugosz et al. 1984; Peng et al. 1981; Atherton et al. 1986). These so-called stress fibre-like structures (SFLS) were present at the tips of bona fide myofibrils and have been proposed to be precursors for the assembly of striated myofibrils (Dlugosz et al. 1984).

Recent evidence from rat cardiac muscle suggests that the intercalated disc may somehow stabilize the sarcomeric organization of the myofibrils. Previous work with cultured neonatal rat heart cells (Atherton et al. 1986) revealed that the sarcomeric structure of myofibrils was stable provided that the fibrils were attached to the sarcolemma via intercalated discs. However, when myofibril terminals were attached to the sarcolemma via focal contacts at regions of the membrane that lacked intercalated discs they appeared to lose their striations gradually as they approached the sarcolemma. These non-striated terminals resembled actin cables and were associated with vinculin in focal adhesion contacts (Atherton et al. 1986) that were reminiscent of focal contacts in fibroblasts where stress fibres attach to the membrane (Geiger, 1979).

Previously, the presence or absence of sarcomeres/striations was assessed indirectly by staining with rhodamine phalloidin (Atherton et al. 1986). Phal-loidin binds specifically to actin (Wulf et al. 1979) and silhouettes the A-and I-bands of mature sarcomeres (Dlugosz et al. 1984; Atherton et al. 1986). However, the non-striated myofibril terminals lacked any recognizable substructure. While it was clear from these results that marked modifications in the myofibrils were present, many questions regarding the details of their structure remained unanswered. Furthermore, the relation of these structures to stress fibres of non-muscle cells has not been established. In this study, the cells were stained for immunofluorescence with anti-bodies to contractile proteins and also examined in the electron microscope in order to assess in more detail the apparent atypical organization of contractile proteins in the myofibril terminals. The results suggest that marked alterations in the structure of myofibrils are associated with their attachment to adhesion plaques in cultured mammalian cardiac muscle. It is proposed that modifications in the structure of myofibril terminals may facilitate their attachment to the sarco-lemma at adhesion plaques. This idea is consistent with the hypothesis that critical differences exist between adhesion plaques and the fascia adherens junction that may have a marked influence on the structure of myofibrils (Atherton et al. 1986).

Materials

Pancreatin and Dulbecco’s modified Eagle’s medium (DMEM) were obtained from GIBCO Laboratories (Grand Island, NY). Rhodamine phalloidin was from Molecular Probes, Inc. (Junction City, OR). Normal goat and normal rabbit sera were obtained from Cooper Biomedical, Inc. (Malvern, PA). Goat anti-mouse and goat anti-rabbit secondary antibodies conjugated with fluorescein were from Kirke-gaard and Perry Laboratories, Inc. (Gaithersburg, MD). Rabbit antiserum to alpha-actinin from bovine heart was a generous gift from Dr Keith Burridge, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina (Bur-ridge & McCullough, 1980). Mouse monoclonal antibodies to vinculin (Geiger, 1981) were obtained from Miles Scientific (Napervilee, IL). Mouse monoclonal antibody MF20 against chicken skeletal muscle myosin (Bader et al. 1982) was obtained from the Developmental Studies Hybridoma Bank maintained by a contract from NICHD (NO1-HD-6–2915). Monoclonal antibody MF20 specifically recognizes sarco-meric myosin and cross-reacts with skeletal and cardiac muscle myosin in mammalian cells (Bader et al. 1982).

Cell culture

Primary cultures of cardiac muscle cells from normal 3-day-old neonatal rats were prepared as previously described (Atherton el al. 1986). Hearts were dissociated by multiple cycles of exposure to pancreatin. The resulting single cell suspensions were depleted of fibroblasts by differential ad-hesion and seeded into 35 mm tissue culture dishes containing glass coverslips at densities ranging from 0·7 × 106 to 1·4 × 106 cells per dish. Cells were fixed for immunofluorescence or electron microscopy 24, 48 or 72 h after plating.

Indirect immunofluorescence staining

Methods for immunofluorescence staining were similar to those reported previously (Atherton et al. 1986). The cells were fixed in 3·7% formaldehyde, permeabilized with acetone at − 20°C and incubated with antibodies. Polyclonal antibodies and rhodamine phalloidin were used at a final dilution of 1:20 in phosphate-buffered saline containing 10 mg ml-1 bovine serum albumin. Monoclonal antibody MF20 was applied to cells as undiluted culture supernatant according to the methods of Bader et al. (1982). Antibody incubations were for 30 min at 37°C in a moist chamber. Coverslips were mounted on glass slides in Aqua-Mount (Lerner Laboratories, New Haven, CT) containing 100mgml-1 of l,4-diazabicyclo-[2.2.2]-octane to retard the fading of fluorescein. Coverslips stained with normal rabbit serum diluted 1:20 and culture supernatant from cultures of SP2 myeloma cells gave negative staining reactions except for weak nuclear staining seen with all polyclonal sera used.

Slides were viewed with a Leitz Dialux 20 fluorescence microscope equipped with a 63×/1·40 piano apochromatic lens. Kodak Tri-X film was developed with Diafine (Acufine, Inc., Chicago, IL).

Electron microscopy

Cells grown on glass coverslips were fixed in 1% glutaraldehyde in 1 M-cacodylate buffer, pH 7·3, for 30 min at 24°C and then rinsed for 1 h. The cells were post-fixed in 1% osmium tetroxide in the same buffer for 30 min, rinsed in distilled water and stained for 30 min with 1% aqueous uranyl acetate. They were dehydrated by a series of 10-min exposures to 50, 70, 95 and 100% ethanol and finally incubated for 30 min in propylene oxide prior to embedding.

The cells were embedded in an EPON–Araldite mixture composed of 23% EPON 812, 18·5% Araldite 502, 55·5% DDSA, and 3% DMP-30 (% by vol.). Cells were first infiltrated with a 1:1 mixture of EPON–Araldite and propylene oxide for 2h and then incubated overnight in undiluted EPON–Araldite. After the coverslips were drained overnight they were placed on top of BEEM capsules (Ted Pella, Inc., Tustin, CA) previously filled with embedding medium and polymerized for 48 h at 60°C without vacuum.

Coverslips were removed from BEEM capsules by alternate exposure to heat and cold. Small 1 mm blocks containing the cells were cut from the flat end of the capsules and re-embedded at the conical tip of a second set of BEEM capsules. The blocks were oriented in the capsule so that the cell layer was either perpendicular or parallel to the anticipated plane of section.

Thin sections were mounted on copper grids and stained with 5% uranyl acetate for 30 min and then with lead citrate for 2–10min. Sections were viewed under a JEOL 100C× transmission electron microscope at an accelerating voltage of 60kV.

Immunofluorescence staining with antibodies to defined sarcomeric proteins

We have shown previously that myofibrils in cultured rat cardiac muscle cells exhibit striated and non-striated regions when stained with rhodamine phal-loidin, and that the non-striated myofibril terminals were associated with elongated membrane plaques that contained vinculin (Atherton et al. 1986). The vinculin plaques were shown to correspond to the location of focal contacts with the substratum.

Fig. 1 shows neonatal rat heart muscle cells after 72 h in culture stained with rhodamine phalloidin and antibodies to alpha-actinin and myosin. With all three labels the myofibrils exhibited striations centrally in the cell, but appeared to lack striations at their ends. Antibodies to myosin (Fig. 1A,B) and alpha-actinin (Fig. 1D) labelled A-bands and Z-bands, respectively, in the same regions of myofibrils that appeared striated after staining with rhodamine phalloidin (Fig. 1C). Similarly, terminal regions of myofibrils that lacked striations when stained with rhodamine phalloidin were also stained continuously throughout their length with both types of antibody to sarcomere components (Fig. 1A,B,D). These antibodies failed to define both A-and Z-bands in the terminal regions consistent with the results with phalloidin labelling. Therefore, although the terminals contained actin, myosin and alpha-actinin, the proteins were not organized into recognizable sarcomeres. This suggested that the terminal elements of these myofibrils were not organized in typical sarcomeric units. The details of their structure were investigated by electron microscopy.

Fig. 1.

Immunofluorescence staining of muscle cells after 72 h in culture with antibodies to defined sarcomeric proteins. A,B. Muscle cells stained with monoclonal antibodies to muscle-specific myosin. Note that the distal ends of myofibrils were labelled continuously throughout their length and lacked defined A-bands (arrows in A and B). At the same time, myosin was distributed in sarcomeres in central regions of the cell. C, D. Double-label staining of muscle cells with rhodamine phalloidin (C), and with antibodies to alpha-actinin (D). The pattern of staining with these probes was similar to that of myosin. The distal ends of myofibrils were labelled continuously along their length with both rhodamine phalloidin (C) and antibodies to alpha-actinin (D) (arrows), while centrally in the cell the myofibrils exhibited striations. Bar, 10 μm.

Fig. 1.

Immunofluorescence staining of muscle cells after 72 h in culture with antibodies to defined sarcomeric proteins. A,B. Muscle cells stained with monoclonal antibodies to muscle-specific myosin. Note that the distal ends of myofibrils were labelled continuously throughout their length and lacked defined A-bands (arrows in A and B). At the same time, myosin was distributed in sarcomeres in central regions of the cell. C, D. Double-label staining of muscle cells with rhodamine phalloidin (C), and with antibodies to alpha-actinin (D). The pattern of staining with these probes was similar to that of myosin. The distal ends of myofibrils were labelled continuously along their length with both rhodamine phalloidin (C) and antibodies to alpha-actinin (D) (arrows), while centrally in the cell the myofibrils exhibited striations. Bar, 10 μm.

Ultrastructure of non-striated myofibril terminals

Replicate cultures of cardiac muscle cells were fixed, stained and embedded for electron microscopy. Thin sections were taken from blocks in a plane parallel to the substratum in order to observe the ultrastructure of myofibrils in the same orientation as they were seen in the fluorescence microscope. The electron microscopic data corroborated previous immunofluorescence observations (Atherton et. al. 1986). The montage in Fig. 2 was constructed from electron micrographs of a myocyte after 72 h in culture. The structure of a typical myofibril terminal as it approaches the free edge of a cell is contrasted with the attachment of a myofibril to a fascia adherens at cell–cell contact sites. In the latter case, the typical sarcomeric organization of the myofibril was maintained all the way up to the fascia adherens (Fig. 2A). However, at the other end of the cell, where there were no contacts with neighbouring cells, the myofibril lacked typical sarcomeric structure (see inset C of Fig. 2 shown at high magnification in Fig. 3). The general transition pattern observed was as follows. First, there was a segmentation of the Z-line into linearly arranged fragments (Fig. 2B). This was followed one to two sarcomere lengths distally by a gradual shift to a longitudinally oriented array of dense bodies that were not in lateral register (Fig. 2, inset C). Similar dense bodies were also observed in cultured rat cells by Legato (1972). Some of the dense bodies were highly elongated in parallel with the myofilaments (Fig. 3). The dense bodies were very similar in distribution, size and shape to the vinculin plaques observed in the light microscope. The dense bodies varied from 0·2 to 10 μm in length and the vinculin plaques ranged from 0·7 to 10μm long.

Fig. 2.

The structure of remodelling terminals viewed in the electron microscope. This montage shows the transitional changes in the organization of myofibril terminals as they approach the free edge of a cell from a 72 h culture. Thin sections were taken parallel to the substratum. Note that the normal sarcomeric structure is retained where myofibrils attach to the sarcolemma via fasciae adhérentes (arrowheads, inset A). The opposite end of the cell is free of cell–cell contacts and the myofibril terminals are in the process of remodelling (montage B,C; and Fig. 4). The first sign of remodelling was a fragmentation of the Z-disc (arrows, inset B). More distally (inset C) Z-discs appeared to be replaced by elongated dense bodies. The area indicated by inset C is shown at high magnification in Fig. 4. Montage, × 5350; inset A, × 13 700; inset B, × 17000.

Fig. 2.

The structure of remodelling terminals viewed in the electron microscope. This montage shows the transitional changes in the organization of myofibril terminals as they approach the free edge of a cell from a 72 h culture. Thin sections were taken parallel to the substratum. Note that the normal sarcomeric structure is retained where myofibrils attach to the sarcolemma via fasciae adhérentes (arrowheads, inset A). The opposite end of the cell is free of cell–cell contacts and the myofibril terminals are in the process of remodelling (montage B,C; and Fig. 4). The first sign of remodelling was a fragmentation of the Z-disc (arrows, inset B). More distally (inset C) Z-discs appeared to be replaced by elongated dense bodies. The area indicated by inset C is shown at high magnification in Fig. 4. Montage, × 5350; inset A, × 13 700; inset B, × 17000.

Fig. 3.

Elongated dense bodies in a modified myofibril terminal. This is a high-magnification view of the area indicated by C in Fig. 2. Note the paucity of thick filaments (small arrowheads). The dense bodies (1arge arrowheads) appear to be contained within the filament bundles and they are oriented parallel to the long axis of the myofibril. l, A lipid droplet. × 31 530.

Fig. 3.

Elongated dense bodies in a modified myofibril terminal. This is a high-magnification view of the area indicated by C in Fig. 2. Note the paucity of thick filaments (small arrowheads). The dense bodies (1arge arrowheads) appear to be contained within the filament bundles and they are oriented parallel to the long axis of the myofibril. l, A lipid droplet. × 31 530.

Light microscopic observations showed that non-striated myofibril terminals were shorter in 48 h cultures than in 72 h cultures (Atherton et al. 1986). Fig. 4 shows a myofibril terminal in a myocyte after 48 h in culture. As expected, the transition changes in the myofibril near its terminus were considerably more abrupt. In this case, the large, elongated dense bodies often occurred at a site where you would expect the next Z-line to be. There was either an intermediate segmentation of the Z-line (not shown) or a streaming of the Z-lines of terminal sarcomeres (Fig. 4).

Fig. 4.

Modified myofibril terminals in a cell from a 48 h culture. Sections were taken parallel to the substratum. The free edge of the cell is to the right. Non-striated terminals of myofibrils are considerably shorter after 48 h in culture than they are after 72h. Notice the streaming Z-lines (arrows) and the elongated dense bodies distal to them (arrowheads). × 16000.

Fig. 4.

Modified myofibril terminals in a cell from a 48 h culture. Sections were taken parallel to the substratum. The free edge of the cell is to the right. Non-striated terminals of myofibrils are considerably shorter after 48 h in culture than they are after 72h. Notice the streaming Z-lines (arrows) and the elongated dense bodies distal to them (arrowheads). × 16000.

Another aspect of these transitional changes was the proportion of thick filaments relative to thin filaments. Regions of the myofibril containing segmented or ‘streaming’ Z-lines qualitatively appeared to have the normal ratio of thick and thin filaments (Figs 2, 4). However, areas containing the large longitudinally oriented dense bodies contained relatively few thick filaments that were not in register laterally (Fig. 3).

Clearly, the apparent lack of striations at the light microscopic level correctly indicated the absence of a typical sarcomeric organization in the myofibril terminals when intercalated discs were absent. Another objective of these electron microscopic studies was to determine whether any membrane specializations could be seen where the myofibrils contacted the cell membrane. Areas similar to those shown in Fig. 2 were sectioned perpendicular to the substratum and parallel to the long axis of myofibrils. This plane of section should reveal points of contact between the myofibril terminals and the sarcolemma. Fig. 5 shows a pseudopodium of a myocyte sectioned perpendicular to the substratum. A myofibril is seen to contain typical sarcomeres deep in the cell, but this organization gradually changed as it approached the edge of the cell.

Fig. 5.

The relation between remodelling myofibril terminals and electron-dense plaques on the cytoplasmic side of the cell membrane. Thin sections were taken perpendicular to the substratum. A. Medium-magnification view showing the lateral association between membrane plaques (small arrowheads) and remodelling terminals (inset C). In contrast, striated portions of myofibrils adjacent to the sarcolemma were not associated with membrane plaques. B. The inset in A at high magnification. Thick and thin filaments are given over to bundles of primarily thin filaments containing large deposits of electron-dense material within them (arrows). C. The inset in A shown at high magnification. Note that bundles of thin filaments with a few thick filaments (small arrowheads) appear to contact the plaques (1arge arrowheads) laterally along their length. A. × 21000; B,C, × 73400.

Fig. 5.

The relation between remodelling myofibril terminals and electron-dense plaques on the cytoplasmic side of the cell membrane. Thin sections were taken perpendicular to the substratum. A. Medium-magnification view showing the lateral association between membrane plaques (small arrowheads) and remodelling terminals (inset C). In contrast, striated portions of myofibrils adjacent to the sarcolemma were not associated with membrane plaques. B. The inset in A at high magnification. Thick and thin filaments are given over to bundles of primarily thin filaments containing large deposits of electron-dense material within them (arrows). C. The inset in A shown at high magnification. Note that bundles of thin filaments with a few thick filaments (small arrowheads) appear to contact the plaques (1arge arrowheads) laterally along their length. A. × 21000; B,C, × 73400.

The terminal had become attenuated in diameter and very few thick filaments could be seen. Moreover, terminals were consistently related laterally with deposits of electron-dense material on the cytoplasmic side of the sarcolemma (Fig. 5A,C). These electron-dense membrane plaques had an average length of 0·8 μm and ranged from 0·3 to 1·8 μm long. In addition, electron-dense material was also observed within the bundles of filaments in the terminals that did not appear to be part of a membrane plaque (Fig.5A,B).

Relation between modified myofibril terminals and stress fibres

Non-striated terminal regions of myofibrils that re-semble stress fibres have been described in embryonic chick cardiac and skeletal muscle cells in culture (Dlugosz et al. 1984; Antin et al. 1986), and in amphibian skeletal muscle cells developing in vitro (Peng et al. 1981). In both systems the myofibrils contained typical sarcomeric striations centrally in the cell, but the striations appeared to be lost at the distal ends of myofibrils as they approached the edge of the cell. At the electron microscopic level, myofibrils in cultured amphibian muscle cells exhibited a transition from a sarcomeric structure to bundles of thin filaments that lacked thick filaments and Z material (Peng et al. 1981). The filament bundles splayed out as they approached the edge of the cell.

In the chick, the non-striated terminals of myofibrils have been referred to as stress fibre-like structures (SFLS) and they were proposed to be precursors to the assembly of striated myofibrils (Dlugosz et al. 1984). Stress fibre-like structures in these cells were shown to resemble stress fibres of non-muscle cells in their pattern of immunofluorescence staining with anti-bodies to alpha-actinin and non-muscle myosin (Dlugosz et. al. 1984).

When SFLS of chick skeletal muscle cells in the process of recovery from exposure to ethylmethane sulphonate were observed in the electron microscope (Antin et al. 1986), they were seen to be composed primarily of 100se bundles of thin filaments with scattered dense bodies and few thick filaments. No data were presented that confirmed that there was continuity between SFLS and striated regions of myofibrils at the ultrastructural level.

In none of the studies in the chick or amphibian systems described above did investigators 100k for specializations of the sarcolemma such as attachment plaques or focal contacts where the myofibrils attached to the membrane. The sites where stress fibres in non-muscle cells terminate and attach to the cell membrane are associated with plaque-like deposits of vinculin, alpha-actinin and talin, which correspond to the location of focal adhesive contacts with the substratum (Geiger, 1979; Burridge & Feramisco, 1982; Burridge & Connell, 1983). In the electron microscope, electron-dense plaques seen on the cytoplasmic side of the membrane are associated with the attachment sites of stress fibre terminals (Heaysman & Pegrum, 1973).

Previous work with the rat system established that non-striated myofibril terminals attached to the cell membrane at points of focal contact with the sub-stratum that were associated with plaque-like deposits of vinculin (Atherton et al. 1986). It was not possible to determine whether the plaques also contained talin because antibodies to chicken gizzard talin did not cross-react well with rat cells (unpublished observations). Since the terminals were labelled throughout their length with alpha-actinin, no conclusions could be drawn concerning whether or not the plaques contained alpha-actinin as well as vinculin. Examination of the terminals in the electron microscope revealed that very prominent electron-dense membrane plaques were associated laterally with non-striated myofibril terminals near the edge of muscle cells (Fig. 5). That is, the membrane plaques seen in the electron micro-scope were found to be associated with myofibril terminals in the same areas of the cell as focal contacts. At present there is no direct evidence that these extra-junctional electron-dense membrane plaques cor-respond to focal contacts observed in the light micro-scope.

While SFLS in chick cardiac muscle cells closely resemble stress fibres in non-musclc cells, non-striated myofibril terminals in rat cardiac muscle cells differ from both SFLS and true stress fibres in a number of ways. First, rat myofibril terminals exhibited extensive lateral associations with electron-dense membrane plaques along their length while stress fibres appear to attach to them at their ends (Heaysman & Pegrum, 1973). In addition, scattered thick filaments were present in non-striated myofibril terminals, but they have not been observed in stress fibres of non-muscle cells (Langanger et al. 1986) and SFLS in chick cells were not labelled with antibodies to muscle-specific myosin (Dlugosz et al. 1984). Immunofluorescence studies have shown that stress fibres and SFLS in chick cells exhibit striations when labelled with antibodies to alpha-actinin or non-muscle myosin (Lazarides & Burridge, 1975; Gordon, 1978; Dlugoszet al. 1984), while non-striated myofibril terminals were labelled continuously throughout their length with antibodies to alpha-actinin and muscle-specific myosin. Therefore, although the non-striated terminals of myofibrils in rat heart cells may have some relation to stress fibres of non-muscle cells, recent observations suggest that they represent myofibrils in the process of assembly, disassembly or remodelling.

Functional significance of modified myofibril terminals

It remains to be determined whether or not non-striated regions of myofibril terminals are capable of contraction. Neonatal rat cardiac cells in 72 h cultures beat vigorously even though most of the cells contained long stretches of non-striated myofibril terminals. It is possible that only the central striated portions of myofibrils were responsible for the observed contractile activity. If this is so, then the non-striated terminals would function much like tendons by serving as an intermediate structure for the attachment of striated myofibrils to the sarcolemma.

The elongated Z-bodies and streaming Z-lines near the ends of myofibrils may indicate that the thin filaments have slid out of alignment relative to their arrangement in an I-band. It has been proposed that the width of the Z-disc reflects the amount of overlap of thin filaments with opposite polarities (Engel & Banker, 1986). A sliding out of alignment in the terminals was suggested by an apparent progressive change in the configuration and orientation of the Z material in rat cardiac cells. Adjacent to striated regions of the myofibrils, the Z-bands, although fragmented, were oriented transverse to the filament bundles. However, more distally, Z-bands were replaced by elongated Z-bodies that were oriented longitudinally with the fibrils. The thick filaments that remained were clearly out of alignment.

The structural characteristics of the non-striated myofibril terminals are subject to a number of alternative interpretations. For example, their structure could reflect transitional changes in myofibrils leading to the gradual breakdown or disassembly of sarcomeres. This possibility is consistent with the fact that the non-striated terminals of myofibrils elongated centripetally with time in culture (Atherton et al. 1986). On the other hand, their structure could represent an intermediate stage in myofibril assembly. This is supported by the fact that large numbers of polyribosomes were observed in the immediate vicinity of the terminals. It is also possible that disassembly and assembly of sarcomeres are occurring simultaneously in the same cells. However, if addition of sarcomeres were the dominant process, then one would expect the myo-fibrils to become gradually striated all the way to their ends. Instead the non-striated regions get progressively longer. Time-lapse observations of changes in the number of sarcomeres present in individual myofibrils of living muscle cells in culture would be required to decide whether sarcomeres are being lost or added.

A third possibility is that it may be necessary for the cell to remodel or modify the structure of myofibril terminals in order to facilitate their attachment to focal adhesive contacts. Thus, reorganization of the contractile proteins in the terminals may represent an adaptation for the attachment of the myofibrils to the sarcolemma in the absence of normal insertion points; namely, the fascia adherens.

There are insufficient data available at present to decide between the three alternative interpretations presented. However, it is intriguing to consider the possibility that the normal sarcomeric organization of myofibrils may be incompatible with membrane specializations associated with focal contacts. This notion would imply that critical differences exist in the composition and/or architecture of focal adhesion plaques versus the fascia adherens (Atherton et al. 1986). Placing rat cardiac cells in culture and depriving them of the opportunity to form fasciae adherentes may have revealed important regulatory or adaptive processes with regard to the structure and organization of myofibril terminals. Both fascia adherens junctions and focal adhesion plaques serve as points of attachment of actin filaments. However, the two types of membrane specializations may interact differently with cytoskeletal/contractile elements. Such differences may have lead to modifications in the organization myofibril terminals.

Moreover, it is interesting to note that the streaming Z-lines and highly elongated Z-bodies observed in the modified myofibril terminals (Figs 3, 4) resembled aberrant morphologies of Z material associated with certain muscle diseases (for review see Engel & Banker, 1986). It is possible that the adaptive changes in the structure of myofibrils seen in culture will prove to be quite relevant to understanding certain disease processes in muscle.

In conclusion, the non-striated myofibril terminals observed in cultured rat cardiac muscle cells differ considerably in their structure from stress fibres in non-muscle cells. Neonatal rat heart cells in culture may provide a useful in vitro model for a molecular analysis of the non-striated myofibril terminals. For example, a more complete analysis of the molecular composition of the terminals with particular attention to the isotypes of contractile proteins present might further elucidate the degree of relatedness of the terminals to non-muscle stress fibres. One would like to know whether the atypical structure observed in the terminals involves a transformation of the myofibril to a stress fibre with its characteristic non-muscle protein isotypes, or if instead it involves an alternative organization of myofibril-specific proteins.

We are most grateful for support of our research from the Chicago Heart Association and the National Science Foundation. The authors thank Dr Keith Burridge for generously providing us with antibodies to alpha-actinin. We also express our appreciation to Dr Robert Decker for reading the manuscript.

Antin
,
P. B.
,
Tokunaka
,
S.
,
Nachmias
,
V. T.
&
Holtzer
,
H.
(
1986
).
Role of stress fiber-like structures in assembling nascent myofibrils in myosheets recovering from exposure to ethyl methane sulfonate
.
J. Cell Biol
.
102
,
1464
1479
.
Atherton
,
B. T.
,
Meyer
,
D. M.
&
Simpson
,
D. G.
(
1986
).
Assembly and remodeling of myofibrils and intercalated discs in cultured neonatal rat heart cells
.
J. Cell Sci
.
86
,
233
248
.
Bader
,
D.
,
Masaki
,
T.
&
Fischman
,
D. A.
(
1982
).
Immunochemical analysis of myosin heavy chain during avian myogenesis in vivo and in vitro
.
J. Cell Biol
.
95
,
763
770
.
Burridge
,
K.
&
Connell
,
L.
(
1983
).
A new protein of adhesion plaques and ruffling membranes
.
J. Cell Biol
.
97
,
359
367
.
Burridge
,
K.
&
Feramisco
,
J. R.
(
1982
).
Alpha-actinin and vinculin from non-muscle cells: calcium-sensitive interactions with actin
.
Cold Spring Harb. Symp. quant. Biol
.
46
,
587
597
.
Burridge
,
K.
&
Mccullough
,
L.
(
1980
).
The association of alpha-actinin with the plasma membrane
.
J. supramolec. Struct
.
13
,
53
65
.
Dlugosz
,
A. A.
,
Antin
,
P. B.
,
Nachmias
,
V. T.
&
Holtzer
,
FI
. (
1984
).
The relationship between stress fiber-like structures and nascent myofibrils in cultured cardiac myocytes
.
J. Cell Biol
.
99
,
2268
2278
.
Engel
,
A. G.
&
Banker
,
B. Q.
(
1986
).
Ultrastructural changes in diseased muscle
.
In Myology
(ed.
A. G.
Engel
&
B. Q.
Banker
), pp.
909
1043
.
New York
:
McGraw Hill Book Company
.
Fischman
,
D. A.
(
1986
).
Myofibrillogenesis and the morphogenesis of skeletal muscle
.
In Myology
(ed.
A. G.
Engel
&
B. Q.
Banker
), pp.
5
37
.
New York
:
McGraw Hill Book Company
.
Geiger
,
B.
(
1979
).
A 130k protein from chicken gizzard: its localization at the termini of microfilament bundles in cultured chicken cells
.
Cell
18
,
193
205
.
Geiger
,
B.
(
1981
).
Involvement of vinculin in contact-induced cytoskeletal interactions
.
Cold Spring Harb. Symp. quant. Btol
.
46
,
671
682
.
Gordon
,
W. E.
(
1978
).
Immunofluorescent and ultrastructural studies of “sarcomeric” units in stress fibers of cultured non-muscle cells
.
Expl Cell Res
.
117
,
253
260
.
Heaysman
,
J. E. M.
&
Pegrum
,
S. M.
(
1973
).
Early contacts between fibroblasts-an ultrastructural study
.
Expl Cell Res
.
78
,
71
78
.
Langanger
,
G.
,
Moeremans
,
M.
,
Daneels
,
G.
,
Sobieszek
,
A.
,
De Brabander
,
M.
&
De Mey
,
J.
(
1986
).
The molecular organization of myosin in stress fibers of cultured cells
.
J. Cell Biol
.
102
,
200
209
.
Lazarides
,
E.
&
Burridge
,
K.
(
1975
).
Alpha-actinin: immunofluorescent localization of a muscle structural protein in non-muscle cells
.
Cell
6
,
289
298
.
Legato
,
M. J.
(
1972
).
Ultrastructural characteristics of the rat ventricular cell grown in tissue culture, with special reference to sarcomerogenesis
.
J. molec. cell. Cardiol
.
4
,
299
317
.
Peng
,
H. B.
,
Wolosewick
,
J. J.
&
Cheng
,
P.-C.
(
1981
).
The development of myofibrils in cultured muscle cells: a whole-mount study and thin section electron microscopic study
.
Devi Biol
.
88
,
121
136
.
Wulf
,
E.
,
Deboben
,
A.
,
Bautz
,
F. A.
,
Faulstich
,
H.
&
Wieland
,
T.
(
1979
).
Fluorescent phallotoxin, a tool for the visualization of cellular actin
.
Proc. natn. Acad. Sci. U.S.A
.
76
,
4498
4502
.a