Expression of a developmentally regulated cross-linking intermediate filament-associated protein (ifapa-400) during the replacement of vimentin for desmin in muscle cell differentiation

During the course of development and differentiation, the transition of intermediate filament proteins (IFs) toward tissue-specific expression involves precise transcriptional regulation (Bloemendal and Pieper, 1989; Kelvin et al. 1989). In early vertebrate development, cytokeratins and vimentin are the first IF isoforms to be expressed. Whereas the former are specific to the epithelia, the latter has a widespread distribution during differentiation (Traub, 1985; Page, 1989). Vimentin is found in all mesenchymal cell derivatives as well as in progenitor cells of muscle and nerve tissues. Upon differentiation of these latter tissues it is progressively replaced by desmin, glial fibrillary acidic proteins (GFAP) or neurofilaments (NFs) according to the cell type (Traub, 1985). However, the structural organization of IFs in networks and, particularly, the fate of these complexes involving several associated proteins during cell differentiation are still poorly understood (Steinert and Roop, 1988; Bloemendal and Pieper, 1989; Robson, 1989).

Several proteins known to be involved in the filament superstructure have already been described and were classified as class II of intermediate filament-associated proteins (IFAPs) by Steinert and Roop (1988). Although a definition of IFAPs remains elusive, they can tentatively be defined as proteins appearing to be morphologically and functionally related to IFs but which do not form filaments (Yang et al. 1990). These proteins are characterized by their very high molecular weight and are thought to play a cross-linking role between IFs, shaping and anchoring the IF network in the cell. Plectin, which is probably related to LFAP-300 (Herrmann and Wiche, 1987), is by far the best characterized of these proteins. It has a widespread occurrence in tissues as well as in cultured cells and is connected to IFs, desmosomes, spectrins and microtubule-associated proteins in addition to Z-lines of cardiac and striated muscles (Wiche et al. 1983; Wiche, 1989). Filaggrin and neurofilaments H and M (NF-H and NF-M), although restricted to epidermal and nerve tissues, are also thought to connect IFs (Steinert and Roop, 1988). Synemin is similar to plectin in its cross-linking function, but different in its chicken-restricted occurrence and its 180 run periodicity along the filament core (Granger and Lazarides, 1980). Finally, paranemin, also expressed with synemin during chicken muscle development, is colocalized with IFs in cultured cells and muscle tissues (Breckler and Lazarides, 1982; Price and Lazarides, 1983).

The in vitro differentiation of fibroblast-like myoblasts fusing and forming multinucleated myotubes is a good model for the study of IF developmental shifts. During the differentiation process, desmin is progressively integrated into vimentin filaments present in myoblasts, and both are progressively associated with Z-lines during and after the assembly of the sarcomeres (Granger and Lazarides, 1979; Gard and Lazarides, 1980; Tokuyasu et al. 1984). This transformation of the IF network might involve heteropolymers of vimentin and desmin. Such continuous reorganization of IF heterofilaments has been proposed to involve paranemin CBreckler and Lazarides, 1982). Another high molecular weight IFAP (IFAPa-400) was also found to be transiently expressed during muscle in addition to neural cell differentiation (Vincent and Lahaie, 1988; Chabot and Vincent, 1990). This protein has been identified by a monoclonal antibody, F51H2, selected for its early recognition of myogenic and neurogenic cell lineages in developing chick embryos. IFAPa-400 is colocalized with and copurified with IFs. It has recently been shown to be strongly expressed during the early events of nervous tissue ontogenesis and to disappear thereafter, suggesting a possible involvement in structural remodeling during cytodifferentiation. Its expression in the chick embryo appeared to be confined to the cell lineages that transiently express vimentin during development (Vincent and Lahaie, 1988; Chabot and Vincent, 1990). Here we present results on IFAPa-400 expression during in vivo as well as in vitro myogenesis and suggest that it has a role in the reorganization of IFs during differentiation of muscle cells.

Anti-IFAPa-400 (clone F51H2) is a monoclonal IgM antibody produced by immunizing a mouse with embryonic day (ED) 3 chick embryo somites (Vincent and Lahaie, 1988). Hybridoma selection was carried out by screening of culture supernatants on polyethylene glycol chick embryo sections showing possible myogenic and neurogenic cell lineage specificity (Thibodeau et al. 1989). Clone F51H2 allowing the identification of IFAPa-400 was selected for its strong reactivity with the myotomes as well as the neural tube. The monoclonal IgG anti-vimentin (Clone 6G2) was produced by a mouse immunized with vimentin purified from chicken erythrocytes (Granger and Lazarides, 1982). This antibody was positively identified by Western blotting and immunofluorescence labellings of chicken embryo fibroblast (CEF) cells. Polyclonal antibody against desmin was produced in rabbits after immunization with chicken gizzard desmin purified as previously described (Geisler and Weber, 1980). The IFA antibody against IFs (Pruss et al. 1981) was obtained from the American Type Culture Collection (Clone TIB 131).

Myoblast primary cultures were obtained using the protocol of Konigsberg (1979) with minor modifications. Cell suspensions resulting from mechanical dissociation of ED 11 chicken pectoralis major muscles were pre-plated for 45 min on uncoated tissue culture dishes in order to remove most of the contaminating fibroblastic cells. Non-adherent myoblastic cells were then plated on gelatin-coated 35 mm dishes at a density of 2.5× 10⁴cells/dish in DME medium supplemented with 10 % horse serum, 3 % chick embryo extract, and 0.01 mM cytosine arabinoside. After 24 h and every 2 days thereafter, the medium was replaced with fresh medium containing 5% horse serum and 3% chick embryo extract.

Cultured cells were fixed for 20 min in methanol at —20 °C and washed with PBS. They were then exposed to a mixture of primary antibodies diluted in phosphate-buffered saline (PBS) containing 0.5 % BSA. This mixture was composed of F51H2 (IgM, ascitic fluid 1/100), biotin-labelled monoclonal anti-vimentin (IgG, culture supernatant), polyclonal rabbit anti-desmin (1/50 dilution). After a 45 min incubation at room temperature, cells were washed with PBS and the immunoreactivities were revealed, respectively, with FITC-conjugated IgM-specific goat anti-mouse (Sigma Chemical Co.), AMCA streptavidin conjugate (Jackson Immunological Labs) and TRITC-labelled goat antirabbit antibodies. Pre-immune rabbit serum, irrelevant hybridoma supernatants of the same classes and second antibodies alone were used as negative controls. Preparations were observed on a Zeiss Axiophot photomicroscope equipped with appropriate filter combinations. No significant fluorescence overlapping of fluorochromes was observed on control preparations.

Tissues obtained from White Leghorn chicks and embryos were snap frozen in liquid nitrogen and preserved from degradation using a cocktail of protease inhibitors (ImM phenylmethyl sulfonyl fluoride (PMSF), 5 μg ml⁻¹ leupeptin, 5 μg ml⁻¹ pepstatin, 5 μg ml⁻¹ antipain in dimethyl sulfoxide (DMSO)) during sample preparations. Cytoskeleton preparations of embryonic pectoral muscles were obtained after homogenization in 0.6 M KC1,1.0% Triton X-100,1 mM EDTAin PBS, pH 7.4, and a 20 min centrifugation at 15 000 g in a microfuge. Resulting preparations were separated on 8 % denaturating gels according to the ‘low bis’ procedure, from Thomas and Kornberg (1975). After electrophoresis, proteins were transfered to nitrocellulose (Towbin et al. 1979) for immunolabelling with F51H2 antibody (ascitic fluid, 1/1000), anti-vimentin 6G2 antibody (culture supernatant) and desmin antibody (1/1000) and subsequently revealed by ¹²⁵I-labelled secondary antibodies using 5% powdered skim milk in PBS as blocking agent.

Whole-mount preparations of IF networks were obtained according to a protocol adapted from others (Fey et al. 1984; Capeo and McGaughey, 1986; Katsuma et al. 1987). CEF cells were grown on gold grids previously covered with Fonnvar and coated with carbon and fibronectin. Cells were extracted in CSK buffer (100 mM NaCl, 300 mM sucrose, 10 mM Pipes, pH 6.8, 3 mM MgCl₂, 0.5% Triton X-100, 1.2 mM PMSF, 0.1 mM iodoacetamide) for 5 min at 0°C and eluted with CSK-AS buffer (250 mM (NH4ASO4, 300 mM sucrose, 10 mM Pipes, pH 6.8, 3mM MgCl₂, 0.5% Triton X-100,1.2 mM PMSF, 0.1 mM iodoacetamide) for 5 additional min. A fixation in 0.2% glutaraldehyde followed by a 5 min wash in 0.5 mg ml⁻¹ NaBH₄ preceded the incubation with the F51H2 antibody for l h. After two 15-min washes in PBS, cells were incubated with colloidal gold goat anti-mouse (10 nm) and washed again in PBS. After a post-fixation in 2% glutaraldehyde, cells were dehydrated through an ethanol series, critical point dried, and coated with carbon for unembedded whole-mount electron transmission observation on a JEOL ‘JEM 1200 EX’ electron microscope. Stereopair micrographs were taken with angles of +5° and −5° at magnifications from 40×10³ to 65×10³ and of + 10° and −10° at magnifications from 10³ to 15×10³.

Several IFAPs have been well characterized to date and two main characteristics emerge as a common denominator: colocalization with IF networks in cultured cells and cosedimentation with them in high-salt/detergent buffers (Traub, 1985; Steinert and Roop, 1988; Wiche, 1989). Triple-labelling experiments performed on myo-fibroblastic cell cultures could establish the clear colocalization of vimentin (Fig. 1A), desmin (B) and IFAPa-400 (C) in a putative myoblast. However, the largest cell shown in Fig. 1 contained a bright IFAPa-400 and desmin spot (arrow) that was vimentin negative, whereas the smallest cell (thick arrow) exibited IFAPa-400 and vimentin but no desmin. In addition to validating the triple-labelling efficiency, the different relative labelling intensities in these cells also suggest that the combination of both IF isoforms with IFAPa-400 may differ from one cell to another.

As shown in Fig. 2, stained gels (A,B) and corresponding immunoblots treated with the IF-specific IF A antibody (Pruss et al. 1981) (C,D) and anti-IFABa-4D0 (E,F) indicate that most of the IFAPa-400 labelling was associated with the high-salt and detergent-resistant IF fraction of ED 12 chick striated muscle. The soluble fraction (B, D, F) contained most of the protein mass (B) and a large amount of vimentin (D), whereas the fraction resistant to 0.6 M KC1 and 1 % Triton X-100 extraction (A, C, E) contained most of the IFAPa-400 (E). The relative proportion of insoluble IFAPa-400 was greater than that of vimentin, which appeared to be equally distributed in the soluble and insoluble fractions. Desmin was not detectable after extraction at this embryonic stage. A major band of 42×x10³M_r corresponding to actin was consistently retained in the resistant fraction along with vimentin. IFAPa-400 did not react with the anti-IF antibody that is epitope-specific to the conserved helical domain of all IFs (C, D). Further investigations, which are in progress, are required to study its relatedness to other IFAPs at the primary structure level.

Another characteristic described previously for plectin or IFAP-300 was its specific localization on the IF network where high molecular weight IFAPs are thought to act as cross-linkers (Foisner et al. 1988; Steinert and Roop, 1988). In order to examine the ultrastructural localization of IFAPa-400 with respect to IFs, we observed CEF cells as whole mounts using immunogold-IFAPa-400 labelling. Stereopair micrographs (Fig. 3A and B) show a threedimensional lattice of IFs where IFAPa-400 labelling was concentrated at the junction of intermediate-sized filaments (12–15 nm), either at the crossing point of two filaments (arrow) or at the junction of several (arrowhead). In a low magnification micrograph of the same area (Fig. 4A), the same network of IFs was also visible throughout the cell surface, where it appeared to maintain the cellular web. IFAPa-400 labelling was never observed on a single filament. Control experiments done on the same preparations using irrelevant IgM antibody, PBS-BSA or second antibodies alone also showed no labelling (Fig. 4B). The identification of the observed filaments as authentic IFs was strengthened by the observation of filaments interwoven at high magnification (Fig. 3B). Not only were gold particles clustered around the IF knot, but a periodicity on filaments similar to the 20–22 nm axial repeat was evident. Similar observations were described previously on reconstituted IFs using shadowing techniques (Henderson et al. 1982; Milam and Erikson, 1982).

The very early expression of IFAPa-400 at embryonic day 2 in myogenic tissues along with its absence in mature tissues (Vincent and Lahaie, 1988) led us to investigate its expression throughout muscle development in parallel with vimentin and desmin. Four tissues were studied; pectoralis major, heart, gizzard and aorta. They represent four types of muscle (skeletal, cardiac, visceral smooth and vascular smooth) that were dissected from ED 8 embryos through to adult chickens. Total extracts from these tissues were resolved by SDS-PAGE and immunoblotted with IFAPa-400, vimentin and desmin antibodies in order to compare their relative levels. As seen in Fig. 5, various patterns of relative expression of the three proteins were observed. No correlation appeared between IFAPa-400 expression and the vimentin/desmin ratio in the four tissues. In fact, we understood that the proportion of IFs in differentiated tissues in comparison to embryonic ones was difficult to evaluate precisely because of all the structural reorganizations occurring in these tissues. In the adult pectoralis major IFs were not detectable as in younger tissue, probably due to their small proportion compared to contractile proteins when measured as total protein content. Despite this situation, the restricted expression of IFAPa-400 to embryonic skeletal, visceral smooth and cardiac muscles was clearly demonstrated: whereas the aorta muscle still contained considerable amounts of IFAPa-400 in adults, pectoralis major and gizzard muscle completely lost their IFAPa-400 content after 18 days and the heart muscle appeared to lose it after hatching.

Embryonic muscle cells placed in culture are able to undergo several steps of myogenic differentiation and produce multinucleated myotubes showing sarcomeric striations. In Fig. 6, triple immunofluorescence labellings of myoblast cultures at different stages showed that vimentin, IFAPa-400 and desmin were simultaneously expressed in early as well as more differentiated myotubes. After one day in culture (Fig. 6A) vimentin was found in all cells, whereas desmin was expressed with IFAPa-400 only in presumably replicating myoblasts (Fig. 6A, thick arrows). Even if its expression showed some heterogeneity, desmin was exclusively found in cells expressing higher levels of IFAPa-400. After 7 days in culture, many vimentin-positive fibroblast-like cells that presumably escaped the cytosine arabinoside treatment also exhibited IFAPa-400 reactivity (Fig. 6C). When myotubes were maintained long enough in culture for us to observe a significant number of fibroblastic cells finally reaching confluence, we could observe a state of differentiation that could not otherwise be observed. After 12 days in culture, vimentin had completely disappeared from desmin-rich myotubes while IFAPa-400 was still abundant (Fig. 6D). Five days later, vimentin was confined to the fibroblast-like cells and desmin to myotubes, and IFAPa-400 had in turn disappeared from myotubes (Fig. 6E). When such cultures were re-exposed to cytosine arabinoside one week after their preparation in order to prevent the spreading of replicating fibroblasts, it was impossible to conserve them long enough to observe the vimentin disappearance from myotubes (data not shown). After 10 days, these cultures formed free myotubes floating in the dish and finally died a few days later.

IFs differ from the other cytoskeletal frameworks in their expression of cell type-specific isoforms in differentiated tissues. In muscle and nervous tissues, desmin and neurofilaments or GFALP are expressed after the transient expression of vimentin. This report describes the comparable transient expression of IFAPa-400 during in vivo and in vitro myogenesis. This protein has been identified by a monoclonal antibody, F51H2, selected for its early recognition of myogenic and neurogenic cell lineages in developing chick embryos (Vincent and Lahaie, 1988). As recently demonstrated in neurogenic tissues (Chabot and Vincent, 1990), its expression in myogenic tissues appeared clearly associated with that of vimentin before terminal differentiation. During in vitro myogenic differentiation, we demonstrate that IFAPa-400 expression dropped progressively after the disappearance of vimentin from myotubes in long-term cultures.

Several studies have permitted both the identification of IFAPa-400 as a member of the IFAP family, and its distinction from those already described. In cultured myoblasts, we have shown that IFAPa-400 was colocalized wnth vimentin and/or desmin filaments (Fig. 1). IFAPa-400 has also been previously demonstrated to be associated with the characteristic colchicine-induced juxtanuclear cables, showing its structural association with IFs (Vincent and Lahaie, 1988). This association between IFAPa-400 and IFs was further demonstrated by their copurification. In embryonic muscle, IFAPa-400 was shown to be mostly resistant to high-salt/detergent extraction. On the contrary, vimentin was distributed in equal proportions between the soluble and insoluble fractions (Fig. 2). Such soluble pools of IFs have been described previously (Ip and Fellows, 1990; Soellner et al. 1985). Although no evidence can directly relate the solubility of IFs to differentiation, it will be interesting to investigate further the stability of the IF network during development.

Despite many common features, IFAPa-400 demonstrated characteristics distinguishing it from other IFAPs. On the basis of one or more criteria such as restricted species or tissue occurrences, apparent molecular weight or ultrastructural localization, IFAPa-400 could be distinguished from plectin (Wiche, 1989), synemin (Granger and Lazarides, 1980), epinemin (Lawson, 1983), NAPA-73 (Ciment and Weston, 1986), p50 (Wang and Ramirez-Mitchell, 1983) or neurofilaments (Liem and Hutchison, 1982). However, IFAPa-400 has several properties similar to paranemin (Price and Lazarides, 1983). Both are chicken-specific, developmentally regulated, colocalized and copurified with desmin and vimentin. Their different apparent molecular weights cannot be considered as a distinguishing criterion, owing to the different SDS-PAGE systems used to extrapolate these values. However, IFAPa-400 was shown to be present in most neuroectodermal cells before their terminal differentiation (Chabot and Vincent, 1990), whereas paranemin expression was confined to both embryonic and adult neural crest-derived Schwann cells (Price and Lazarides, 1983). Ln addition, since heavy labelling of paranemin could be found on adult myocardium Z-discs where no IFAPa-400 could be observed, the two proteins are probably different, but any relationship will have to be directly established by using antibody cross-reactivity, peptide mapping or nucleic acid sequence analysis.

The ultrastructural localization of IFAPa-400 at IF crossing points is a characteristic previously described for plectin (Foisner et al. 1988). This localization, along with its association with vimentin led to the hypothesis that plectin and proteins of the same family cross-link IFs into networks (Steinert and Roop, 1988; Wiche, 1989). Such labelling, in addition to the absence of IFA reactivity, also argues against the possibility that IFAPa-400 could be an IF itself.

Our immunoblot studies have shown that IFAPa-400 disappeared completely from embryonic pectoralis major, heart and gizzard muscles while its expression increased continually in aorta muscle. No clear relationship could be established between the expression of IFAPa-400 and the vimentin/desmin ratio. The variable relative contents of desmin, vimentin and IFAPa-400 were probably due to the very different nature of these tissues. However, whereas the levels of IFAPa-400 expression could not be correlated with the IF ratio in the different muscles, its expression appeared to be associated with the persistence of vimentin in the vascular smooth muscle. These results are compatible with previous studies showing that only vascular smooth muscle may actually express vimentin constitutively in adults with various amounts of desmin (Frank and Warren, 1981; Gabbiani et al. 1981; Osborn et al. 1981; Schmid et al. 1982). This unusual expression in adult muscle tissue has been proposed to be related to the elastic nature of the proximal part of the aorta, given its gradual disappearance in the muscular distal part of the artery (Osborn et al. 1981). Therefore, the persistence of IFAPa-400 with vimentin in such tissues is very interesting with regard to its function during development.

As seen in the triple immunofluorescence experiment (Fig. 6), IFAPa-400 was expressed in myoblasts during the replacement of vimentin by desmin in mature myotubes, remaining colocalized with them until vimentin disappeared. In young cultures, desmin was always expressed in cells expressing IFAPa-400. There were similar variations in the relative labelling intensity between the three proteins. One vimentin-positive cell could express various levels of IFAPa-400 and desmin, IFAPa-400 always being the higher of the two. Later in culture, there was very high expression of IFAPa-400 with desmin in fusing cells still expressing vimentin. The ultimate absence of vimentin on Z-lines where desmin is still present with IFAPa-400 after 12 days could represent a further state of differentiation before the appearance of fully differentiated myotubes containing desmin only. Higher magnifications always showed the same colocalization of the three proteins, either in the loose filament network form or at the Z-lines. In fact, the accumulation of IFs at Z-lines appeared to be a continuous process of reorganization following the setting up of the contraction machinery (Tokuyasu et al. 1984). IFAPa-400 expression remained associated with this process until the vimentin disappeared. The possibility of epitope masking in our in vitro myogenesis experiments appeared improbable, since the same phenomenon was observed with denatured proteins on SDS-PAGE with both the F51H2 antibody (Fig. 5) and a purified polyclonal antibody (data not shown). Whether the disappearance of vimentin in myotubes can be attributed to the presence of the cell layer replacing the plastic substratum cannot be resolved at this point. However, such involvement of the extracellular matrix or cell-cell contacts in muscle culture have been clearly demonstrated previously. During myogenic differentiation in vitro, for example, fusion or further differentiation could not occur if the integrin was blocked with the CSAT antibody (Menko and Boettiger, 1987). Other experiments have also demonstrated that the extracellular matrix influenced the sarcomeric structure as well as the onset of myogenic differentiation (Kujawa et al. 1986; Foster et al. 1987; Hilenski et al. 1989). In our long-term cultures, the disappearance of vimentin from myotubes represents a terminal differentiation process that was observed only when the myotubes were lying on a vimentin-positive cell layer after 17 days in culture. Even if the nature of the apparent interaction has not been determined, the transition of IF isoforms by direct cell to cell contacts has been reported in other systems (Holtzer et al. 1982; Dulbecco et al. 1983; Ben-Ze’ev, 1984; Sanderson et al. 1986).

Altogether, the results presented in this report further establish IFAPa-400 as a new element associated with the process of IF shifts and reconstruction. Instead of forming a new network to replace the one specific to undifferentiated cells, newly expressed IF isoforms would instead be integrated into old networks, which could then be remodeled into new structures in the differentiated cells. There is considerable evidence to support the idea that IF networks are remodeled during differentiation. In addition to several reports of multiple IF colocalizations, transfection experiments involving exogenous IF genes placed under independent control express IF proteins that are subsequently incorporated into pre-existent filament networks without disturbing their integrity (Wang et al. 1984; Quax et al. 1985; Albers and Fuchs, 1987). The integration of desmin into an existing network would therefore promote new interactions with different cell constituents and play a role in terminal cytodifferentiation. In this sense, the IF isoform specificity might be related to functions in differentiated cells (Steinert and Roop, 1988). For example, IFs that have a loose network conformation in undifferentiated skeletal muscle cells are located progressively at Z lines during differentiation (Granger and Lazarides, 1979; Gard and Lazarides, 1980). Furthermore, in epidermal cells specific basal layer keratins are replaced by keratins associated with hyperproliferation in suprabasal epidermis (Stoler et al. 1988) whereas others are associated with the establishment of desmosomes (Erikson et al. 1987; Kopan et al. 1987). Finally, the relationship between isoform expression and function was also shown for vimentin during rapid growth (Connel and Rheinwald, 1983; Boyer et al. 1989), and in high-density fibroblast cultures where cell motility and cell-cell contacts govern IF expression pattern (Ben-Ze’ev, 1984). Taken together, these results showing the integration of newly expressed IF proteins into existing filaments and the relationship between isoform and function suggests that during differentiation, the replacement of one isoform by another may occur gradually during the reshaping of the IF network. The recent description of the occurrence of nestin in neuroepithelial stem cells, a protein sharing structural features with IFs and defining a new IF class, will lead to a more comprehensive understanding of IF transitions during differentiation (Lendahl et al. 1990). It is likely that there are several associated proteins still to be described in addition to transient isoforms of IFs.

The results presented in this paper support the view of a dynamic IF network in which two IF isoforms can be expressed together and found associated progressively with new structures. By co-culturing myogenic cells with fibroblasts, we were able to obtain myotubes showing advanced levels of maturation as suggested by their strong desmin reactivity and lack of vimentin. Owing to its transient expression, which is concomitant with the isoform shift, we suggest that IFAPa-400 represents an important element that is involved in this process. IFAPa-400 could be associated with the transient or hybrid coexpression of IF isoforms, and have a cross-linking function during the reshaping of the IF network.

We thank Drs R. W. Currie and P. A. Rogers for their helpful discussions and critical reading of the manuscript. We also thank Dr P. M. Charest for his help with the electron microscopy studies and Dr J.-P. Valet for the preparation of antibodies. This work was supported by the Medical Research Council of Canada (program PG-35). M.V. was a scholar of Le Fonds de la Recherche en Santé du Québec. L.C. received a studentship from Le ministère de l’éducation du Québec (Actions structurantes).