Journal of Cell Science 107, 377-386 (1994) 377 Printed in Great Britain © The Company of Biologists Limited 1994

Rearrangement of mRNAs for costamere proteins during costamere development in cultured from chicken

E. J. Morris and A. B. Fulton* Department of Biochemistry, University of Iowa, Iowa City, IA 52242, USA *Author for correspondence

SUMMARY

Mature skeletal myofibrils are surrounded by costameres, earlier in development. Controls for bleed through of fluo- ribs of metavinculin, , intermediate filaments, and rescence, RNase H sensitivity, hybridization without probe, other proteins that connect the myofibril to the extracellu- and binding to the myofibril all gave appropriate results. lar matrix. Costameres have recently been shown to be the Probes to glyceraldehyde-3-phosphate dehydrogenase, a sites at which the forces generated by the myofibril are glycolytic enzyme, stained diffusely and did not associate transduced laterally into the extracellular matrix. We with the myofibril. These results show that components of observed costameres developing in cultured skeletal the costamere arrive at the structure in a defined sequence, muscles, grown in micromass culture from cells taken from and that mRNA organization is a conspicuous, precise and embryonic chicken leg. We detected proteins by immuno- temporally controlled aspect of costamere development. fluorescence and mRNA by in situ hybridization. Antibody These results may have wider implications. In these cells, and probe signals were imaged by laser scanning confocal some mRNAs are positioned with submicrometer precision microscopy. Antibody to vimentin protein is first detected in space and differentially over time. Particular mRNAs in stripes in register with the Z line of the myofibril, at differ in the time and place of such positioning. This implies ~day 12 after fusion; soon thereafter probe to vimentin both that cellular structures provide physical cues for such mRNA is also detected in the same stripes. Optical sections positioning and that mRNA contains information that indicate that vimentin mRNA and protein are very close, no interacts with such cues in a message-specific manner. If more than 0.1 mm apart and possibly in immediate contact. such precision in mRNA location is found in other somatic Antibody to vimentin is detected in stripes only in cells that cells, it could have significant implications for the ways in twitch spontaneously. Antibodies and probes to and which cells generate and maintain cellular structures. vinculin protein and mRNA are next detected in stripes of the same periodicity, at ~day 17 after fusion. Vinculin Key words: in situ hybridization, costamere, myofibrillogenesis, protein (but not mRNA) is detected at focal contacts much intermediate filaments, mRNA

INTRODUCTION of proteins and occurring in highly restricted locations, costameres are one of the cellular structures whose assembly Costameres are ribs of protein that contain vinculin and lie must be explained in order to understand the mature phenotype between the cell membrane and the Z line in some skeletal of a . muscles and in (Craig and Pardo, 1983; Pardo We reported the organization of vimentin protein into bands et al., 1983). Costameres encircle the myofibril and offer a with the spacing of the Z line in muscles grown in micromass contact point to couple contractions of the myofibril to the cultures (Cripe et al., 1993). In these cells, vimentin mRNA extracellular matrix along the length of muscle cells (Shear and was also organized into bands with the same spacing. We were Bloch, 1985). A recent report established that costameres are interested to discover whether these bands of vimentin were sites of force transduction in cardiomyocytes (Danowski et al., costameres developing in cultured muscle cells and whether 1992). Costameres include various proteins known elsewhere message organization might play a role in the development of to be associated with cell matrix adhesion sites. These proteins these structures. For these reasons, we have examined the include talin (Beckerle and Yeh, 1990), spectrin and ankyrin organization of vimentin, vinculin and desmin protein and (Nelson and Lazarides, 1984), intermediate filament proteins, mRNAs during the later stages of muscle development. and (Porter et al., 1992). Danowski et al. (1992) presented a detailed review of recent work on costameres. MATERIALS AND METHODS Costameres are one structure among several that permit muscle cells to function as contractile machinery and transduce Muscle culture contraction to the rest of the organism. Composed of a variety Cultures were prepared weekly from thigh muscle of 12-day-old 378 E. J. Morris and A. B. Fulton chicken embryos (Denning et al., 1988). Synchronized muscle cells showed a striated pattern of protein, mRNA, or both. The highest level were grown on -coated glass coverslips employing a newly of organization in a given cell was the level at which the cell was developed micromass method (Cripe et al., 1993). scored. A minimum of 15 fields of view were scored for each coverslip. The percentage of cells displaying a given trait was calcu- Fixation lated and recorded. Cells were fixed in 1:9 (v/v), formaldehyde/methanol for 20 minutes These cultures were fairly synchronous throughout the first week. at −20¡C. Solutions of 37% formaldehyde sealed under N2 were With longer times in culture, the muscles showed some variation in obtained from Electron Microscopy Sciences, and a new ampule was the times at which given patterns were seen, but did not vary in the used each time. This procedure is modified from that of Bresser and sequence in which they were seen. To compare samples cultured in Evinger-Hodges (1987). If fixed cells were stored in 70% ethanol at different weeks, samples were grouped into 3- or 4-day periods (7- −20¡C, there was some loss of resolution. All the images shown here 10, 14-17, 20-24). The data from these periods were then averaged to were photographed on the same day on which the cells were fixed. account for variability within any one culture and between different Some samples were also fixed for protein staining only, in absolute cultures depending on the overall maturity. In addition, since these ethanol at −20¡C. To optimize double staining for desmin and were micromass cultures, the cells in the centers of the micromasses vinculin, other samples were fixed in 0.1% glutaraldehyde in were at higher densities and showed more advanced development than methanol for 10 minutes at −20°C. All three fixation methods were cells at the edge. For these two reasons, reporting development as the compared for all three proteins. average percentage gives an impression of precision that could be misleading. For example, within a single culture, the cells in the center In situ hybridization of the micromass might have 70% of the cells with periodic vimentin The procedure of Lawrence and Singer (1985, 1986) was used with a mRNA and 35% periodic desmin mRNA; at the edge of such a mass, few modifications, described previously (Cripe et al., 1993). This the comparable numbers might be 40% and 20%. Since it was imposs- procedure was optimized by Lawrence and Singer for efficient hybrid- ible to collect data as a function of the distance from the micromass, ization and reduced background. It was shown to detect single copy the averaged data for development was expressed on a scale from 1 genes of DNA and cytoplasmic mRNAs of low abundance. We chose to 4, based on the percentage of cells that showed a striated pattern the present fixation because it improved cellular preservation and (0, did not stain for the protein (or mRNA); 1, stained diffusely; 2, retention of mRNA in these cells. between 0 and 30% were periodic; 3, between 30 and 70% were periodic; and 4, between 70 and 100% were periodic). Between 14 Probes and 21 samples were scored for each data point. A 425 base pair DNA probe to a vimentin-specific sequence, a 450 The variations in development after the first week make tests such base pair DNA probe to a vinculin-specific sequence, and a 389 base as Student’s t-test inappropriate for detecting differences within a pair DNA probe to a desmin-specific sequence were generated by the single culture. Therefore, the probability (P) that the fraction of cells PCR technique. The probes were constructed using digoxigenin-11- showing a pattern of vimentin protein in stripes was different from dUTP in the reaction mixture. A 640 base pair DNA probe to glyc- the fraction of cells showing vimentin mRNA in stripes was calcu- eraldehyde-3-phosphate dehydrogenase (GAPDH) was generated in a lated using the Wilcoxon two-sample test. Probabilities (P) for the similar fashion. Details of the PCR conditions and the probes to testing of the vimentin, desmin and vinculin protein or message were vimentin and GAPDH have been described (Cripe et al., 1993; Zehner calculated using the Mann-Whitney U-test. and Paterson, 1985). The 389 bp desmin mRNA probe was constructed using the same PCR technique. The plasmid pD8 was supplied by S. Capetanaki. The RESULTS 5′ 24-nucleotide primer was derived from the nucleotide sequence beginning at nucleotide 56 (5′-ATGTCAAGATGGCCTTG- GACGTGG-3′) and the 3′ 24-nucleotide primer ended with nucleotide We have previously shown that vimentin mRNA is localized 445 (5′-CTGGTCTTGCGCTAGCGCGAAGCC-3′). The 450 bp in cultured muscle cells (Cripe et al., 1993). The patterns seen vinculin mRNA was constructed using the same PCR technique. The for vimentin mRNA change during development. The most plasmid p1020 was supplied by S. Craig. The 5′ 30-nucleotide primer highly organized pattern detected is a series of stripes, coinci- was derived from the nucleotide sequence beginning at nucleotide dent with the vimentin protein, that appear after muscles begin 2704 (5′-ATCAGCCCATGATGATGGCTGCTAGGCAGT-3′) and to contract. These bands of vimentin protein and mRNA the 3′ 37-nucleotide primer ended at nucleotide 3144 (5′-GATGCT- overlie and are somewhat wider than the Z line. ′) GCTTCAGCTTCTCTCACAGTT-3 . Costameres are circumferential bands of protein that include Detection of probe and protein vinculin, desmin and vimentin (Craig and Pardo, 1983; Pardo The hybridized probe was detected with anti-digoxigenin antibody et al., 1983). We were therefore interested to determine labeled with either fluorescein or rhodamine as previously described whether these bands of vimentin were costameres. To examine (Cripe et al., 1993). Vimentin, vinculin and desmin protein were this possibility, we used immunocytochemistry to detect detected by monoclonal antibodies and fluorescein-labeled anti-IgG vinculin and desmin protein and mRNA in parallel with studies also as before. The monoclonal antibody used to detect vimentin was for vimentin protein and mRNA at various stages of muscle monospecific for vimentin, based on immunoprecipitation of labeled development. proteins in cultured muscle and on two-dimensional western blots During the first week of development, after fusion was (Isaacs et al., 1989a). The monoclonal antibodies used to detect initiated with divalent cations (see Materials and Methods), desmin and vinculin were the monoclonal antibodies D-3 (against muscle cells fused to form myotubes that assembled and desmin) and VN 3-24 (against vinculin); both were obtained from the extended myofibril. At the end of this week, antibodies to Hybridoma Bank, University of Iowa. Samples were viewed in the Bio-Rad 600 laser scanning confocal microscope, which permits vimentin showed diffuse or fibrillar staining and probes to optical sectioning and computerized quantitation. vimentin mRNA showed diffuse staining in most cells at this time (Fig. 1). Antibodies to vimentin began to display a Statistical analysis periodic pattern of staining with a spacing of the Samples were observed and scored for the percentage of cells that lying above the Z line between days 7 and 10 of muscle devel- Proteins and mRNAs at costameres 379

Fig. 1. Diffuse organization of costamere proteins and mRNAs after the first week of development. Left columns show antibody staining, right columns show staining with probe to mRNA. Top half shows muscles stained with antibodies or probes to vimentin; bottom half shows muscles stained with antibodies or probes to vinculin. ×2,000.

Fig. 2. Increasing organization of vimentin protein and mRNA during the second week of development. Left columns show antibody staining; right columns show staining with probe to mRNA. Top panel shows muscles that stain periodically with antibodies to vimentin, but stain diffusely with probes to vimentin mRNA; bottom panel shows muscles, stained, that stain periodically both with antibodies to vimentin and with probes to vimentin messenger. Top, ×2,000; bottom, ×3,000. opment; this organization continued to increase during the and those that stained periodically for vimentin mRNA; the second week of development. The probes to vimentin mRNA protein staining was slightly more organized (Fig. 3). By days showed a high degree of organization in some cells; other cells 20 to 24 after fusion there was no statistically significant dif- still displayed a diffusely fibrogranular staining pattern, ference in the periodicity of staining for either protein or although all myotubes at this stage contained extensive message for vimentin (Fig. 3, Fig. 4). In addition, by this time myofibril (Fig. 2). By days 14 to 17 after fusion, greater than in culture the intensity of staining for vimentin was decreas- 75% of the cells displayed periodic staining for vimentin ing. protein (Fig. 2, Fig. 3). The probes to vimentin mRNA also Next we compared the organization of vimentin protein to showed periodic staining in most cells at this time; however, that of vinculin and desmin protein. As shown in Fig. 5, sub- there was a statistically significant difference between the stantial fractions of the cells displayed periodic staining for number of cells that stained periodically for vimentin protein vimentin protein at days 7 to 10. However, no cells displayed 380 E. J. Morris and A. B. Fulton

pattern had become largely periodic, the vinculin and desmin Fig. 3. Organization of staining patterns were only periodic in about half as many cells vimentin protein and (Fig. 5). There was no significant difference in the fraction of mRNA at costameres cells that stained periodically for vinculin and desmin. By the during development. last period examined (days 20 to 24 of culture), there were no Cells were grown in significant differences in the fraction of cells that stained peri- micromass cultures, odically with any of the three antibodies (Fig. 5, Fig. 4A). In fixed at various times during development, and Fig. 5, it appears that a larger fraction of the cells displayed stained to reveal protein periodic staining for vimentin; however, the asynchrony of the and mRNA. At one cultures had increased at that point and made it difficult to week of development, detect small differences. The middle panel in Fig. 4A shows a fewer than half the cells culture stained for desmin protein and mRNA. This sample contained periodic was fixed using the glutaraldehyde fixation described in vimentin protein; by Materials and Methods, and was chosen as an example for this week two, more than fixative. The glutaraldehyde fixation showed good preservation three-quarters did. of details but suffered from higher background fluorescence Vimentin mRNA also became periodically organized, but did so than the simultaneous methanol-formaldehyde fixation. For slightly later than did the protein. desmin only, methanol-formaldehyde fixation produced samples that were less intensely stained than did glutaralde- periodic staining for either vinculin or desmin at this time hyde fixation (Fig. 4B). With both fixatives, the presence of (vinculin shown in Fig. 1). All muscle cells displayed staining periodic staining was unambiguous for all three proteins and for vinculin and desmin, but the pattern was either diffuse or mRNAs. fibrogranular during this first week and up to a week and a half We also examined the organization of mRNA at the of development. By days 14 to 17, when the vimentin protein costamere. From Fig. 6 it is clear that a substantial fraction of

A

B

Fig. 4. (A) Organization of costamere proteins and mRNAs after the third week of development. Staining for vimentin, desmin and vinculin proteins and mRNAs all show periodic patterns. Left columns show antibody staining; right columns show staining with probe to mRNA. Top panel shows muscles stained with antibodies or probes to vimentin. Middle panel shows muscles stained with antibodies or probes to desmin; bottom panel shows muscles stained with antibodies or probes to vinculin. The desmin sample was fixed using glutaraldehyde fixation and shows some background fluorescence; it is included for comparison with simultaneous methanol-formaldehyde fixation. ×2,000. (B) Sample fixed by simultaneous methanol/formaldehyde fixation, and stained for desmin protein and mRNA; overall intensity is lower but periodicity of staining is unambiguous. Proteins and mRNAs at costameres 381

Some additional observations are illustrated by Fig. 7. There are brightly staining granules that stain only for message Fig. 5. Organization of (green), especially in the vicinity of the nuclei. These granules costamere proteins are most prominent in the less mature portions of the cell and during development. do not stain for vimentin protein. These brightly staining Cells were grown in granules have been observed at earlier time points as well. In micromass cultures, the more mature portion of the cell, small streaks of green stain fixed at various times during development, can be detected. These are most conspicuous in more mature and stained to reveal cells. These images were collected simultaneously by separate vimentin, vinculin or detectors. The images can be fused to determine the degree to desmin protein. At one which the protein and mRNA are colocalized. The fused image and two weeks of is shown at the bottom. development, more Finally, it is quite likely that this does represent a single cell cells contained and that the cytoplasm is common space. However, domains periodic vimentin in which message is organized periodically lie immediately protein than contained adjacent to domains in which it is still diffuse or fibrogranular. periodic vinculin or desmin protein; by week three, the three proteins were equally well organized. Similar contrasts in organization are also seen along the long axis of a single myotube. Various controls were used to reduce the possibility of artefact (Fig. 8). The pattern depends on RNA-DNA hybrids, Fig. 6. Organization of because treating samples with RNase H (specific for DNA- costamere mRNAs RNA hybrids) after the hybridization reaction abolishes the during development. Cells were grown in signal (Fig. 8, top). The pattern does not reflect non-specific micromass cultures, binding to costameres, because not all mRNAs occur in these fixed at various times stripes. For example, the mRNA for the glycolytic enzyme during development, GAPDH is not localized in these cells (Fig. 8, bottom). When and stained to reveal stripes of vimentin mRNA are seen, they are not binding non- vimentin, vinculin or specifically to the myofibril, because cells at earlier stages have desmin mRNA. At one myofibrils (stained here with antibodies to ) but no and two weeks of stripes of vimentin mRNA (not shown; compare with Cripe et development, more cells al., 1993). Controls for bleed-through between channels and contained periodic samples prepared without a probe also gave appropriate images vimentin mRNA than contained periodic as before (Cripe et al., 1993), but are not shown here. vinculin or desmin mRNA; by week three, the three mRNAs were equally well organized. DISCUSSION cells began displaying periodic staining for vimentin mRNA as early as the first 10 days of culture. By the second week of We will begin by presenting a summary of these results in culture, 14 to 17 days after fusion, most cells displayed which the simplest interpretation is that the images seen by periodic staining for vimentin mRNA. At this same stage of antibody or probe staining are reasonable representations of development, fewer than half of the cells displayed periodic protein and message distribution. We will then examine several staining for vinculin and desmin mRNA. There was no statis- possible sources of artifact and discuss why we do not think tically significant difference in the fraction of cells that stained that they affect the interpretation of our results. Finally, we will periodically for either these two mRNAs. For the last period compare these results with those of previous studies of inter- examined, no statistically significant differences could be mediate filaments and costameres in muscle cells. found between any of the three mRNAs observed. If the staining for protein and mRNA represents the distrib- In some cases a single cell would display several patterns of ution of protein and message in these cells, the following development at one time. Such a cell is shown in Fig. 7. In the sequence can be determined. Muscle cells in these cultures go lower half of the cell, both the staining for vimentin protein through fusion and develop myofibrils. These myofibrils begin (red) and for vimentin messenger (green) show substantial to come into register with each other laterally during the latter periodicity. In the upper portion of the cell, the staining for part of the first week of development. After muscles in the vimentin protein (red) is periodic; however, the staining for cultures begin spontaneous contractions, the vimentin protein, vimentin mRNA still displays the fibrogranular or diffuse which until this time had been longitudinally distributed in staining seen earlier in development. It is likely that this cell loose fibers, comes to be organized and concentrated above the represents a more mature myotube into which a less mature Z line and around the Z disk. This process occurs more rapidly myotube has recently fused and that the two cytoplasmic com- in muscles cultured on plastic (cf. Fig. 8g, Isaacs et al., 1992). partments are beginning to come into register with each other. Within one to two days of the organization of vimentin protein These differences in organization can also be seen by taking in this pattern, vimentin mRNA becomes concentrated samples at different times during development, as shown in (although not exclusively) in the same positions. Within one Figs 1, 2 and 4. to two days after vimentin mRNA redistribution, both the 382 E. J. Morris and A. B. Fulton

Fig. 7. Colocalization of protein and mRNA. Micrographs show staining for vimentin protein (red) and mRNA (green). Bottom panel shows a merged image of the top two micrographs. mRNA and protein for vinculin and desmin also come to lie in for rejecting them. First, it is possible that the staining with these narrow bands. anti-vimentin antibodies contains a component of cross- The presence of vinculin in these strips confirms our initial reaction to desmin protein. This explanation was one proposed belief that we are observing the development of costameres in for periodic staining of vimentin (Holtzer et al., 1982), pre- cultured muscle. It also appears that these cells contain suffi- viously observed by some investigators (Gard and Lazarides, cient information to localize message and protein with submi- 1980). However, the antibody used here is a monoclonal, not crometer precision. This organization occurs at different times a polyclonal, antibody. In addition, it has been shown to be for different messages and proteins. monospecific for vimentin by immunoprecipitation and by Are all of the observations obtained in this way sufficiently western blotting after two-dimensional gel electrophoresis. In robust to permit this interpretation? We will consider four both of these assays of specificity, the samples used included possible alternative explanations for the results and our reasons desmin and ought to have detected any cross-reactions with Proteins and mRNAs at costameres 383

Table 1 System Event in development of skeletal muscle observed Reference Myoblast fusion, myotube formation Various 1 Early myofibril formation Various 2,3 Lateral registry of slender myofibrils Various 4-8 Spontaneous contraction In culture 5,7,8 Organization of vimentin protein above I band In culture 8.9 Organization of desmin, vinculin, other costamere In culture 4,5,7,8 proteins above I band In embryo 6 Disappearance (absence) of vimentin from costamere In culture 5,7,8 (requires coculture with fibroblasts) In embryo 6 ‘Condensation’ of costamere proteins 5,6,7,8

These references are given as examples and are not intended to be comprehensive, particularly for the stages preceding costamere development. 1, Colley et al. (1990); 2, Hill et al. (1986); 3, reviewed by Fulton and Isaacs (1991); 4, Gard and Lazarides (1980; leg); 5, Holtzer et al. (1982; breast); 6, Tokuyasu et al. (1983, 1984; leg); 7, Cossette and Vincent (1991; breast); 8, this report (leg); 9, Cripe et al. (1993; leg).

others not, thus generating an apparently periodic pattern when in fact the proteins are homogeneous? The possibility of dif- Fig. 8. Controls (left panel in each pair shows staining for protein; ferential fixation is impossible to eliminate completely. right panel shows staining for mRNA). Top: vimentin protein or However, we have used three different fixations, some mRNA in muscle cells. The sample was treated with RNase H (specific for DNA-RNA hybrids) after in situ hybridization. The involving chemical modifications and others not; the protein fluorescent signal for mRNA is nearly abolished. Bottom: muscles staining patterns appear to be independent of fixation. It is were stained for vimetin protein and GAPDH mRNA. GAPDH difficult to see how differential fixation could generate the peri- mRNA is not localized with costameres, but appears diffuse; odicity of the patterns, since the younger, and presumably more therefore neither the probes nor the cellular mRNAs bind to fragile, cells are the ones that display the most even staining costameres nonspecifically. for protein. It is only in later development that periodic patterns of staining are seen for the protein. Do the differences in time at which the staining for vimentin, desmin that this monoclonal antibody displays. For our vinculin and desmin become organized result from cryptic immunofluorescent staining to represent cross-reaction with epitopes? There are several examples of a protein that is found desmin, the antibody would have to detect an epitope that is to be present by biochemical criteria but is undetectable by displayed by vimentin protein under all circumstances and by immunofluorescence until some spatial reorganization has desmin protein only under circumstances of immunofluores- happened. However, this seems unlikely in this case. We do cence. not observe an absence of staining first, followed by a pattern. This seems unlikely. Other investigators have detected Rather, we observe one pattern of staining and then a different striped patterns of vimentin in muscle cells by using polyclonal pattern of staining, with no difference in intensity between the antibodies that have been cross-absorbed with the opposite two patterns. For cryptic epitopes to be misleading here, there antigen (Tokuyasu et al., 1983, 1984). Thus, we are not the first would have to be some configuration that vinculin and desmin to detect with high reliability the organization of vimentin enter into transiently as they move between the diffuse, fibro- protein with the periodicity and position of the costamere. granular pattern of earlier development and the periodic pattern Does the staining that we see represent non-specific binding seen later. While this is not impossible, it seems unlikely. to the myofibril? This seems unlikely for several reasons. First, We are reasonably confident, therefore, that the patterns of these cells contain abundant myofibril for several days before staining that we have observed are reasonable reflections of the any of the probes or antibodies begin to display periodic actual kinetics of development for the messages and proteins staining. Second, the various antibodies and probes become examined here. This raises the question of how our results periodic at different times. It is difficult to see how non-specific compare with earlier observations of intermediate filament dis- staining could be temporally regulated for different probes. tribution during muscle development in vitro and in vivo. Third, the periodic patterns of probes to all three mRNAs Some discrepancies in vimentin and desmin distribution had disappear when the samples are treated with RNase H, which already been reported when these two intermediate filaments is specific for RNA in RNA-DNA hybrids. RNase H should began to be the subject of extended investigation. We believe, not reduce non-specific staining. Finally, probes to GAPDH however, that most of the disagreements in the literature can never stain in stripes, but display a diffuse pattern at all times be reconciled if due weight is given to the variations in devel- (Cripe et al., 1993). Chemically the process used to detect opment that occur with differences in muscle source, GAPDH mRNA would be expected to produce the same kinds embryonic age, culture conditions and other variables that are of artifacts. biologically often irrelevant, but experimentally significant. Are the periodic staining patterns that we see the result of We would like to propose a synopsis of muscle development differential fixation? That is, are the proteins differentially (without appending days to it; Table 1) that may account for preserved during development or are some parts preserved and most or all of the observations in the literature and reconcile 384 E. J. Morris and A. B. Fulton apparent disagreements. Representative references for this vimentin in filaments may also reflect the post-replicative state synopsis are given in Table 1. of these cells (Vikstrom et al., 1992). Mononucleated myoblasts fuse to form multinucleated Similarly, desmin staining would then be likely to represent myotubes. One of the earliest events in these myotubes is the desmin-rich filaments, not pure desmin homopolymer. We formation of periodically organized myofibril, in which have reported that vimentin is assembled in part cotranslation- myosin, , α-actinin and all display a characteristic ally; about half of the vimentin polypeptides associate with periodicity. These myofibrils begin as several slender strands cytoskeletal structures during translation (Isaacs et al., 1989a). that come into register laterally with each other during the If desmin filaments, which are similar chemically (reviewed by next period of development. As the myofibrils come into Stewart, 1993), also undergo this process one might expect to register, cells begin spontaneous contractions. Soon after cells see homogeneous domains, the length of which would reflect spontaneously contract, vimentin protein in these cells comes the processivity of cotranslational assembly. This remains to to overlie and surround the Z disk. Shortly thereafter, be tested. vimentin mRNA is also concentrated there. Shortly after that, If there is a stage in cultured muscle development that has vinculin mRNA and protein, and desmin mRNA and protein, vimentin-rich and desmin-rich filaments that are not seen in the also come to lie in the region of the costamere. With embryo, can these observations have relevance to studies of the continued development, the vimentin protein disappears from embryo in which the segregation of intermediate filament the muscle cell, leaving vinculin concentrated at the proteins was not as marked (Tokuyasu et al., 1985)? We costameres, and desmin underlying the costameres, and sur- propose that the slower kinetics of development observed in rounding the Z disk. culture offer an opportunity to observe transients that would be Some features of this synopsis that warrant attention include difficult to observe in the embryo. Tokuyasu (1989) has the mechanical event of contraction and its apparently invariant reported on events during cardiac myofibrillogenesis in the occurrence before periodic intermediate filaments are observed embryo that occur in hours; in culture these events require days (Cossette and Vincent, 1991, and this report). Furthermore, the (Lu et al., 1992). We propose that the constraints on growing disappearance of vimentin with further culture may depend cells under conditions that permit in situ hybridization have upon the presence of fibroblasts in the same culture (Cossette also slowed down development enough to permit detection of and Vincent, 1991; Holtzer et al., 1982). If these two features a transient intermediate that would be difficult to detect during are given due weight, it appears that almost all of the reported the rapid development observed in embryo. An additional results are mutually consistent. The one possible exception is observation that is consistent with this organization having the detection by immunofluorescence of vimentin around the functional significance is that only vimentin mRNA has been Z disk of mature muscle. This may be a consequence of cross- observed becoming organized into strands and cables at the reaction with the vimentin-like protein observed by Tokuyasu periphery of regions in which protein is becoming organized et al. (1983). into costameres. If this synopsis represents muscle development accurately, Alternatively, what we see as a two-step process, in which it presents two new questions about the results presented here. vimentin-rich filaments are first organized at the costamere, First, how can we observe a concentration of vimentin protein followed by desmin-rich filaments, may occur simultaneously occurring at the costameres before desmin protein if desmin in the embryo. Crosslinking proteins that interact with vimentin and vimentin protein are copolymerized throughout intermedi- may tether intermediate filaments at one site, while interactions ate filaments? The evidence for some measure of copolymer- between desmin and other proteins tether those proteins ization is strong (Tokuyasu et al., 1985, among others). elsewhere on the same intermediate filaments. More detailed However, these authors reported clustering of each isoform immunoelectron microscopy will be needed to resolve this. The within one filament. Some stretches of filament appeared to be organization of intermediate filaments at this site also suggests almost pure homopolymer, while other stretches were more that intermediate filaments may help anchor mRNA here. mixed. The authors comment on the marked non-uniformity of Perhaps the use of the truncated desmin gene that disrupts inter- staining in these filaments. mediate filaments (Schultheiss et al., 1991) would resolve this. We suggest that, in the cells we observed, there is still a sub- This is the first report of highly localized mRNAs for desmin stantial number of vimentin filaments, synthesized several days and vinculin, and only the second to correlate protein and earlier when the ratio of vimentin to desmin was fairly high. mRNA position in in such finely spaced patterns (Cripe et al., Such filaments might not be pure homopolymer but would be 1993). As we discussed in that report of vimentin mRNA vimentin-rich filaments. For them to retain the vimentin-rich location, a striped pattern with spacing of less than a micro- composition that they had when first assembled, we further meter apart cannot serve a function in the soluble phase; any propose that, as in the case of myosin isoforms, there is less gradient formed would disperse too rapidly. Vimentin has been exchange observed in vivo than might be predicted from results reported as assembling during translation (about 50% in these in vitro. Reduced exchange of myosin isoforms in vivo has cells; Isaacs et al., 1989a). For a protein that assembles during been documented (Bandman, 1985; Gauthier, 1990). It may be translation, the position of the mRNA determines the site of that here we are observing a similarly stable filament for the initial assembly; later redistribution or disassembly and re- intermediate filament protein vimentin. Such increased assembly would affect the relation between accumulated stability of the filaments in vivo might result from the presence protein and its mRNA. The relation between localized mRNA of ancillary factors or from excluded-volume effects on disas- and cytoskeletal assembly has been discussed at length recently sembly and reassembly. Excluded-volume effects would be (Fulton, 1993). expected to decrease the rate of disassembly and favor the per- Another striking feature of these observations is that desmin sistence of the homopolymer, once formed. The stability of and vinculin mRNAs become concentrated at the stripes later Proteins and mRNAs at costameres 385 than does vimentin mRNA. This implies either that mRNA is of myofibrillar proteins in round postmitotic myoblasts of embryonic skeletal located with more spatial precision than we can detect (i.e. muscle. J. Cell Sci. 95, 11-22. these patterns do differ, but at the ultrastructural level) or that Cossette, L. J. and Vincent, M. (1991). Expression of a developmentally regulated cross-linking intermediate filament-associated protein (IFAPa- the cell can modulate the delivery of different mRNAs differ- 400) during the replacement of vimentin for desmin in muscle cell entially. Either explanation implies more cellular control over differentiation. J. Cell Sci. 98, 251-260. mRNA location than previously appreciated. Craig, S. W. and Pardo, J. V. (1983). Gamma actin, spectrin, and intermediate How is the mRNA localized to these stripes? Since vimentin filament proteins colocalize with vinculin at costameres, myofibril-to- undergoes cotranslational assembly to a significant extent, one attachment sites. Cell Motil. 3, 449-462. Cripe, L., Morris, E. and Fulton, A. B. (1993). Vimentin mRNA location possibility is that nascent peptides are ‘captured’ at the appro- changes during muscle development. Proc. Nat. Acad. Sci. USA 90, 2724- priate site. Actin mRNA accumulates at the appropriate 2728. location when protein synthesis is inhibited (Singer, 1992). Danowski, B. A., Imanaka-Yoshida, K., Sanger, J. M. and Sanger, J. W. This approach cannot be used here, because the organization (1992). Costameres are sites of force transmission to the substratum in adult of mRNA occurs over several days. Such extended treatments rat cardiomyocytes. J. Cell Biol. 118, 1411-1420. Denning, G., Kim, I. S. and Fulton, A. B. (1988). Shedding of cytoplasmic with inhibitors are toxic. A diffferent mechanism for mRNA by developing muscle cells. J. Cell Sci. 89, 273-289. localization is suggested by the high sequence conservation Fulton, A. B. and Isaacs, W. B. (1991). Titin, a huge, elastic myofibril protein seen in the 3′ untranslated regions of these mRNAs; the with a probable role in morphogenesis. BioEssays 13, 157-161. vimentin 3′ untranslated region is 80% identical between Fulton, A. B. (1993). Spatial organization of the synthesis of cytoskeletal chicken and hamster. The actin 3′ untranslated region can proteins. J. Cell. Biochem. 53, 1-5. Gard, D. L and Lazarides, E. (1980). The synthesis and distribution of desmin localize heterologous mRNA (Singer, 1992). Experiments are and vimentin during in vitro. Cell 19, 263-275. underway to examine this region of the vimentin mRNA for Gauthier, G. F. (1990). Differential distribution of myosin isoforms among the the ability to target mRNA appropriately. myofibrils of individual developing muscle fibers. J. Cell Biol. 110, 693-701. We conclude that vimentin protein is the first component Hill, C. S., Duran, S., Lin, Z., Weber, K. and Holtzer, H. (1986). Titin and detected in register with the myofibril above the I band, at ~day myosin, but not desmin, are linked during myofibrillogenesis in postmitotic mononucleated myoblasts. J. Cell Biol. 103, 2185-2196. 12 of culture; soon after, vimentin mRNA is also present in the Holtzer, H., Bennett, G. S., Tapscott, S. J., Croop, J. M. and Toyama, Y. same stripes. Desmin and vinculin mRNAs and proteins are (1982). Intermediate-size filaments: changes in synthesis and distribution in next detected in stripes of the same periodicity, starting at week cells of the myogenic and neurogenic lineages. Cold Spring Harb. Symp. 2. Controls for bleed through, RNase H sensitivity, hybridiz- Quant. Biol. 46, 317-329. ation without probe, and binding to the myofibril all gave Isaacs, W. B., Cook, R. K., Van Atta, J. C., Redmond, C. M., Lin, H. L. and Fulton, A. B. (1989a). Assembly of vimentin in cultured cells varies with appropriate results. Probes to GAPDH, a glycolytic enzyme, cell type. J. Biol. Chem. 264, 17953-17960. did not associate with the myofibril. These results indicate that Isaacs, W. B., Kim, I. S., Struve, A. and Fulton, A. B. (1989b). Biosynthesis components of the costamere arrive in the structure in a defined of Titin in cultured skeletal muscle cells. J. Cell Biol. 109, 2189-2195. sequence; mRNA organization is a conspicuous aspect of Isaacs, W. B., Kim, I. S., Struve, A. and Fulton, A. B. (1992). Association of titin and myosin heavy chain in developing muscle. Proc. Nat. Acad. Sci. costamere development. USA 89, 7496-7500. These results may have wider implications. These cells Lawrence, J. B and Singer, R. H. (1985). Quantitative analysis of in situ position several mRNAs with submicrometer precision in hybridization methods for the detection of actin gene expression. Nucl. Acids space and differentially over time. Different mRNAs differ in Res. 13, 1777-1799. the time and place of such positioning. Therefore, cellular Lawrence, J. B. and Singer, R. H. (1986). Intracellular localization of mRNA for cytoskeletal proteins. Cell 45, 407-415. structures offer physical cues for such positioning, and mRNA Lu, M.-H., DiLullo, C., Schultheiss, T., Holtzer, S., Murray, J. M. and interacts with these cellular cues in a message-specific, devel- Choi, J., Fischman, D. A. and Holtzer, H. (1992). The vinculin/sarcomeric- opmentally regulated manner. Many cells show localized a-actinin/a-actin nexus in cultured cardiac myocytes. J. Cell Biol. 117, 1007- mRNA for cytoskeletal proteins (reviewed by Singer, 1992; 1022. Fulton, 1993). If similar precision in mRNA location is present Nelson, W. J. and Lazarides, E. (1984). Goblin (ankyrin) in striated muscle: identification of the potential membrane receptor for erythroid spectrin in in other somatic cells, it could have significant implications for muscle cells. Proc. Nat. Acad. Sci. USA 81, 3292-3296. the ways in which cells generate and maintain cellular struc- Pardo, J. V., Siliciano, J. D., Craig, S. W. (1983). A vinculin-containing tures. cortical lattice in skeletal muscle: transverse lattice elements (‘costameres’) mark sites of attachment between myofibrils and sarcolemma. Proc. Nat. This work was supported by grants from the Muscular Dystrophy Acad. Sci. USA 80, 1008-1012. Association and the American Heart Association. We thank Janelle Porter, G. A., Dmytrenko, G. M., Winkelmann, J. C. and Bloch, R. J. Beck for excellent technical assistance with cell culture and in situ (1992). Dystrophin colocalizes with b-spectrin in distinct subsarcolemmal hybridizations. This article is dedicated to the memory of Benjamin domains in mammalian skeletal muscle. J. Cell Biol. 117, 997-1005. Schultheiss, T., Lin, Z., Ishikawa, H., Zamir, I., Stoeckert, C. J. and Pearce Morris, 1931-1991. Holtzer, H. (1991). Desmin/vimentin intermediate filaments are dispensable for many aspects of myogenesis. J. Cell Biol. 114, 953-966. Shear, C. R. and Bloch, R. J. (1985). Vinculin in subsarcolemmal densities in REFERENCES chicken skeletal muscle: localization and relationship in intracellular and extracellular structures. J. Cell Biol. 101, 240-256. Bandman, E. (1985). Distribution of slow myosin in dystrophic chicken Singer, R. H. (1992) The and mRNA localization. Curr. Opin. muscle. Advan. Exp. Med. Biol. 182, 63-72. Cell Biol. 4, 15-19. Beckerle, M. C. and Yeh, R. K. (1990). Talin: Role at sites of cell-substratum Stewart, M. (1993). Intermediate filament structure and assembly. Curr. Opin. adhesion. Cell Motil. Cytoskel. 16, 7-13. Cell Biol. 5, 3-11. Bresser, J. and Evinger-Hodges, M. J. (1987) Comparison and optimization Tokuyasu, K. T. (1989). Immunocytochemical studies of cardiac of in situ hybridization procedures yielding rapid, sensitive mRNA myofibrillogenesis in early chick embryos. III. Generation of fasciae detections. Gene Anal. Tech. 4, 89-104 adherentes and costameres. J. Cell Biol. 108, 43-53. Colley, N. J., Tokuyasu, K. T. and Singer, S. J. (1990). The early expression Tokuyasu, K. T., Dutton, A. H. and Singer, S. J. (1983). Immunoelectron 386 E. J. Morris and A. B. Fulton

microscopic studies of desmin (skeletin) localization and intermediate Steady state dynamics of intermediate filament networks J. Cell Biol. 118, filament organization in chicken skeletal muscle. J. Cell Biol. 96, 1727-1735. 121-129. Tokuyasu, K. T., Maher, P. A. and Singer, S. J. (1984). Distributions of Zehner, Z. E. and Paterson, B. M. (1985). The chicken vimentin gene: aspects vimentin and desmin in developing chick myotubes in vivo. I. of organization and transcription during myogenesis. Ann. NY Acad. Sci. 455, Immunofluorescence study. J. Cell Biol. 98, 1961-1972. 79-94. Tokuyasu, K. T., Maher, P. A. and Singer, S. J. (1985). Distributions of vimentin and desmin in developing chick myotubes in vivo. II. Immunoelectron microscopic study. J. Cell Biol. 100, 1157-1166. (Received 23 August 1993 - Accepted, in revised form, Vikstrom, K. L., Lim, S. S., Goldman, R. D. and Borisy, G. G. (1992). 3 December 1993)