Plectin 1 links intermediate filaments to costameric sarcolemma through β-synemin, α-dystrobrevin and actin

Skeletal muscle fibers possess heteropolymeric intermediate filaments (IFs) comprised of desmin, synemin, and paranemin (Bilak et al., 1998; Granger and Lazarides, 1980; Price and Lazarides, 1983; Schweitzer et al., 2001). These IFs are located at the periphery of myofibrillar Z-discs, interlink adjacent myofibrils at the Z-line level, and anchor onto costameres beneath the sarcolemma (Craig and Pardo, 1983; Fujimaki et al., 1986; Lazarides, 1980; Tokuyasu et al., 1983). The importance of IF-interlinkages between adjacent Z-discs and IF-anchorage on costameric sarcolemma in muscle function and integrity is supported by investigations of desmin-knockout mice, in which muscle fibers lacking desmin IFs develop reduced maximum force and their sarcolemmae are more susceptible to damage during contraction (Balogh et al., 2003; Li et al., 1996; Li et al., 1997; Sam et al., 2000). However, molecular organization of the linkages from IFs to Z-discs and costameres still remains unclear.

Within costameres, dystrophin-glycoprotein complex (DGC) provides structural support to the sarcolemma by linking the actin-based cytoskeleton with the extracellular matrix (ECM) (for a review, see Ozawa, 2006). It is comprised of dystrophin, α- and β-dystroglycan, sarcoglycans, sarcospan, α-dystrobrevin and syntrophin. Dystrophin is associated with β-dystroglycan and extracellular α-dystroglycan, which in turn binds laminin-2 in the basement membrane (BM) (Ervasti and Campbell, 1993), while it is indirectly associated with a group of five integral membranous proteins, the sarcoglycans and sarcospan (Ozawa et al., 1998; Yoshida et al., 1994). Other subsarcolemmal proteins in the complex include α-dystrobrevin and syntrophin, which directly interact with dystrophin (Ahn et al., 1996; Sadoulet-Puccio et al., 1997).

β-Synemin is a constituent of heteropolymeric IF in muscles, and has a very short N-terminal head domain, a central α-helical rod domain conserved in all IF proteins, and a very long C-terminal tail domain (Titeux et al., 2001). β-synemin molecules are assumed to link IFs to DGC, based on the in vitro findings that they can bind α-dystrobrevin, dystrophin and desmin (Bhosle et al., 2006; Mizuno et al., 2001). However, their binding sites for desmin, α-dystrobrevin and dystrophin are all confined to their rod domains, thereby raising a possibility that heteropolymerization of β-synemin with desmin may prevent β-synemin from binding α-dystrobrevin and/or dystrophin and therefore from linking IFs to DGC due to their competition for the rod domains.

Another candidate for the linker between IFs and costameres is the versatile crosslinker protein plectin. Plectin harbors a binding site for IF proteins at its C-terminus (Foisner et al., 1988; Nikolic et al., 1996; Reipert et al., 1999), whereas its N-terminal parts include binding domains for actin and integrin β4 (Andra et al., 1998; Geerts et al., 1999; Rezniczek et al., 1998). With these binding properties, plectin links cytokeratin IFs to the plasma membrane at hemidesmosomes of epidermal cells (Hieda et al., 1992; Litjens et al., 2006). By analogy, it is conceivable that plectin links desmin IFs to costameric sarcolemma. In fact, our previous immunoelectron microscopic study ultrastructurally revealed that plectin-labeled fine threads linked IFs to dystrophin- or vinculin-containing subsarcolemmal dense plaques, or costameres (Hijikata at al., 2003). In that study, however, we did not identify plectin-binding partners within costameres and, therefore, could not fully explore molecular organization from IFs to costameres through plectin.

To address these unexplored subjects, the present study was undertaken just to identify β-synemin as a protein binding plectin N-terminal fragments. The present immuno-EM analyses revealed that β-synemin was incorporated into costameric dense plaques associated with plectin threads. Furthermore, our in vitro analyses showed that β-synemin could bind actin as well as α-dystrobrevin, whereas plectin 1 – a plectin isoform expressed in skeletal muscles – could interact with α-dystrobrevin, β-synemin and actin. Based on the results obtained here, we propose a model of molecular organization from IF to costameres that includes DGC and integrin complexes.

Our previous immunoelectron microscopic analysis of rat skeletal muscle fibers revealed that plectin-labeled fine threads extended from desmin IFs to subsarcolemmal dense plaques containing dystrophin (Hijikata et al., 2003). Given this and the desmin-binding property of plectin at its C-terminus (Reipert et al., 1999), N-terminal parts of plectin would be expected to interact with as yet unidentified molecules within dystrophin-containing dense plaques. Thus, N-terminal fragments of plectin 1, an isoform of plectin expressed in skeletal muscles (Fuchs et al., 1999), were used as a probe for blot overlay assay to identify plectin-1-binding proteins within the subsarcolemmal dense plaques. The crude IF fraction prepared from muscle light microsome (LM) with detergents was separated by SDS-PAGE and blotted onto PVDF membranes. The membranes were overlaid with Myc-tagged N-terminal plectin-1-recombinant proteins (PleN1, 1-1273 aa), followed by detection of plectin-binding proteins with anti-Myc antibody. This blot overlay assay detected some Myc-positive bands representing possible plectin-1-binding proteins (data not shown).

Of these possible plectin-1-binding proteins, we focused our analysis on a protein of ∼160 kDa (supplementary material Fig. S1A). We determined its partial amino acid sequence and found that it contained PHEFH and VQLQRMVDQRS sequences. These amino acid sequences were used to search the GenBank database using BLASTP for similar sequences. The two sequences were identical to those of KIAA0353, an incomplete cDNA sequence isolated from a human brain cDNA library (GenBank™ accession number AB002351). To obtain the full-length cDNA, cloning of the 160 kDa plectin-1-binding protein was performed in a cDNA library prepared from skeletal muscle mRNA of newborn rats. By determining the complete DNA sequences of the obtained clones, we identified the plectin-1-binding protein as desmuslin (GenBank accession number AB091769). Desmuslin, a novel member of IF proteins, was found as a α-dystrobrevin-binding protein by the yeast-two-hybrid system (Mizuno et al., 2001). Comparison of the previously reported human and the present rat sequences revealed an overall identity of 72.4%. Desmuslin has subsequently been referred to as β-synemin (Mizuno et al., 2004; Titeux et al., 2001), and this name is also used in this article. The interactions of plectin with β-synemin and then α-dystrobrevin, a dystrophin-associated protein, agreed with our previous findings of the association of plectin threads with dystrophin-containing dense plaques.

In vitro interactions between plectin and β-synemin were further verified by pull-down assays using recombinant plectin 1 and β-synemin proteins. GST-fused full-length β-synemin was incubated with either Myc-tagged N-terminal plectin 1 (PleN1) fragments or Myc-tagged β-galactosidase (LacZ) and then immunoprecipitated by using anti-Myc antibody and protein L-agarose. The results indicated that PleN1 fragments coimmunoprecipitated with β-synemin, but neither LacZ nor PleN1 did this in combination with control IgG (supplementary material Fig. S1B). A reciprocal pull-down assay was performed using glutathione beads. GST-fused β-synemin pulled down PleN1, but not LacZ, whereas GST protein alone did not precipitate PleN1 (supplementary material Fig. S1C).

For in vivo analysis, plectin or β-synemin immune complex was immunoprecipitated from lysates of muscle LM using anti-plectin or anti-β-synemin antibody, respectively. Subsequent immunoblotting for β-synemin or plectin indicated that plectin and β-synemin formed protein complexes in vivo as well (supplementary material Fig. S1D). In control experiments, neither plectin nor β-synemin was immunoprecipitated by control IgG.

To assess the localization of β-synemin relative to plectin and α-dystrobrevin in skeletal muscle, rat diaphragm cryosections were doubly immunolabeled with anti-β-synemin and anti-plectin or anti-α-dystrobrevin antibodies, and observed by confocal laser scanning microscopy. As shown in Fig. 1A-C, β-synemin completely colocalized with plectin, which was found around Z-discs and beneath the sarcolemma (Hijikata et al., 1999; Hijikata et al., 2003; Schröder et al., 1997; Schröder et al., 1999). On longitudinal sections, both β-synemin and plectin displayed striated staining of Z-lines and subsarcolemmal intermittent staining confined to areas overlying Z-lines or costameres (Fig. 1D-F). By contrast, α-dystrobrevin was more diffusely distributed along the sarcolemma (Fig. 1H,K,N,Q). Doubly immunostained tangential sections including the sarcolemma clearly delineated costameric striations coinciding between β-synemin and α-dystrobrevin (Fig. 1J-L), whereas the longitudinal sections displayed intermittent superimposition of β-synemin-staining on diffuse α-dystrobrevin-staining along the sarcolemma (Fig. 1M-O). Consistent with the present observation of its complete colocalization with β-synemin, plectin also colocalized with α-dystrobrevin at costameres (Fig. 1P-R). These results indicated the colocalization of β-synemin and plectin with α-dystrobrevin at the costameric sarcolemma.

To further explore the precise localization of β-synemin beneath the sarcolemma, immunogold electron microscopy (EM) analysis was performed in rat skeletal muscle fibers chemically skinned with saponin. In this EM analysis, specific gold particles indicative of β-synemin were found along IFs, on the cytoplasmic side of subsarcolemmal dense plaques overlying Z-lines, namely costameres, and between IFs and costameric dense plaques (Fig. 2). The association of β-synemin-specific gold particles along IFs was consistent with the notion that β-synemin, together with desmin and paranemin, formed heteropolymeric IF (Hemken et al., 1997; Hirako et al., 2003). Extensive observations disclosed that IFs were interlinked with costameric dense plaques by fine threads, possibly plectin molecules, on both sides of which β-synemin labels were localized (Fig. 2C). Costameric dense plaques without association of IF were also immunolabeled with the gold particles (Fig. 2B). These results indicated discrete incorporation of β-synemin molecules into two distinct sites: IFs and costameric dense plaques.

Pull-down assay was carried out to define more precisely the β-synemin subdomains involved in the interaction with PleN1. Based on the β-synemin domain structure consisting of head, rod and tail domains, ten β-synemin mutant recombinant proteins were generated by subcloning rat β-synemin cDNA into a GST expression system (Fig. 3A). These mutant β-synemin proteins were incubated with Myc-tagged PleN1 and pulled down using anti-Myc antibody and protein L-agarose. The mutant proteins (Tail N1, N2 and Rod C), including the N-terminal part of β-synemin tail domains, were pulled down using PleN1. In addition, mutant proteins (Rod Ms and Ml) including the C-terminal part of the rod domain (rod domain 2B) were also precipitated, but in smaller amounts. Without anti-Myc antibody, however, none of the β-synemin mutant proteins were pulled down. These results indicated that the part of β-synemin around the boundary between its rod and tail domains preferentially interacted with PleN1.

Next, to refine β-synemin-interacting sites on PleN1, reciprocal pull-down experiments were performed using GST-fused N-terminal tail fragments of β-synemin (Tail N1) and a variety of mutant plectin recombinant fragments, as shown in Fig. 3B. β-Synemin Tail N1 fragments precipitated mutant plectin 1 fragments, including the first exon product, calponin-homology domains and the middle portion of plakin domain (Fig. 3B), indicating that PleN1 fragments included three β-synemin-binding sites.

The present mapping revealed that three mutant plectin 1 fragments, Ex1, CHDα and PlD-M could interact with β-synemin in vitro. To further assess these interactions within cells, we expressed the three Myc-tagged plectin fragments in C2C12 myoblasts and myotubes, and examined their subcellular localization relative to β-synemin by double immunostaining with anti-Myc and anti-β-synemin antibodies. Before the transfection experiments, we examined subcellular localization of β-synemin by immunohistochemistry using anti-β-synemin antibody and fluorescent phallotoxins in C2C12 myoblasts and myotubes. In most of C2C12 myoblasts, β-synemin was expressed in a diffuse dotted pattern throughout the cytoplasm, whereas in C2C12 myotubes and some myoblasts, it was localized along stress-fiber-like structures (SFLSs) or immature myofibrils in an intermittent pattern, in addition to its dotted sarcoplasmic distribution (Fig. 4A-F).

The three plectin fragments colocalized at least partly with β-synemin in C2C12 cells (Fig. 4J-X), suggesting their intracellular interactions with β-synemin. Plectin Ex1 fragments were primarily localized in the nucleus and did not colocalize with β-synemin in most of the transfectants (Fig. 4G-I). In some transfectants, however, they were weakly distributed throughout cytoplasm with faint staining of their colocalization with β-synemin (Fig. 4J-L). These results could be interpreted as follows: overexpression of Ex1 fragments resulted in their distribution at the cytoplasm, where they were preferentially distributed along β-synemin-associated SFLSs, although the fragments were mostly localized in the nucleus because of a putative nuclear localization signal included in them (Rezniczek et al., 2003). Plectin CHDα fragments appeared to colocalize with β-synemin along SFLS and in the sarcoplasm, where they were distributed in a pattern somewhat more patchy than a dotted pattern of β-synemin (Fig. 4M-R). Plectin PlD-M fragments were locally distributed in an irregularly intermittent pattern along a subset and some parts of β-synemin-associated SFLSs, and almost lack their sarcoplasmic distribution, indicating their partial colocalization with β-synemin along some portions of SFLSs (Fig. 4S-X).

The present pull-down assay showed that plectin CHDα, including actin-binding sites, interacted with β-synemin tail fragments (Tail N1). This prompted us to examine the effect of β-synemin Tail N1 on the actin-binding of plectin CHDα. First, we confirmed in vitro interaction of plectin CHDα with F-actin. Consistent with previous reports (Fontao et al., 2001; Geerts et al., 1999), plectin CHDα co-sedimented with F-actin, whereas it was found in the soluble fraction in the absence of F-actin (Fig. 5, lanes 2 and 3). Next, we tested in vitro interaction of β-synemin Tail N1 with F-actin. Unexpectedly, β-synemin Tail N1 fragments slightly co-sedimented with F-actin, but were hardly found in the pellet fraction in the absence of F-actin (Fig. 5, lanes 4-6). These results indicated a direct, but weak association of β-synemin tail fragments with F-actin. In a co-sedimentation assay to assess the effect of β-synemin Tail N1, F-actin was polymerized in the presence of plectin CHDα, incubated with a tenfold molar excess of β-synemin Tail N1, and then sedimented by centrifugation. Even in the presence of β-synemin Tail N1, plectin CHDα was found to be associated with F-actin and almost absent in the soluble fraction (Fig. 5, lanes 7 and 8). These results indicated that β-synemin Tail N1 did not inhibit interactions of plectin CHDα with F-actin under this experimental condition.

The present pull-down assays demonstrated that plectin 1 bound the rod domain 2B and N-terminal tail domain of β-synemin, whereas a previous study reported that α-dystrobrevin interacted with the β-synemin rod domain (Mizuno et al., 2001). These findings implied that plectin 1 and α-dystrobrevin compete with each other for β-synemin owing to the partial overlap of their interacting sites on the rod domain – although we initially presumed that plectin 1 interacted with β-synemin, which in turn bound α-dystrobrevin. This presumptive molecular array of plectin 1, β-synemin and α-dystrobrevin was examined in immunoprecipitation experiments using Myc-tagged PleN1 fragments, GST-fused full-length β-synemin and α-dystrobrevin fragments including products of exons 8 to 16, which were indispensable for interactions with β-synemin (Mizuno et al., 2001). Prior to this experiment, interactions of β-synemin with α-dystrobrevin were ascertained by pull-down assay. Using glutathione beads, GST-fused β-synemin pulled down α-dystrobrevin fragments, whereas GST protein alone did not (Fig. 6A). Next, interactions of plectin 1 with α-dystrobrevin were also tested by immunoprecipitation. In this experiment, Myc-tagged LacZ recombinant proteins and N-terminal fragments of plectin 1f (PleN1f) were used as control and comparison samples, respectively. Plectin 1f is an alternative variant differing from plectin 1 only in the first exon and is most likely to interact with α-dystrobrevin, because it is preferentially distributed along the sarcolemma of skeletal muscle fibers (Rezniczek et al., 2007). In contrast to this prediction, only PleN1 coimmunoprecipitated with α-dystrobrevin fragments, whereas neither PleN1f nor LacZ did (Fig. 6B). These results indicated that plectin 1 directly interacts with α-dystrobrevin through its unique exon 1 part.

To assess the formation of the presumptive molecular array, full-length β-synemin, α-dystrobrevin fragments and Myc-tagged PleN1, PleN1f or LacZ were incubated and then coimmunoprecipitated using anti-Myc antibody and protein L-agarose beads. As shown in Fig. 6C, PleN1 fragments coimmunoprecipitated with α-dystrobrevin fragments as well as β-synemin, whereas PleN1f fragments coimmunoprecipitated with β-synemin but not α-dystrobrevin fragments. Taking account of β-synemin- and α-dystrobrevin-binding sites on plectin 1 and plectin 1f, these results suggest that plectin 1 and plectin 1f can bind β-synemin at their CH and/or plakin domains, thereby probably preventing the association of β-synemin with α-dystrobrevin, whereas plectin 1 can interact further with α-dystrobrevin or both α-dystrobrevin and β-synemin through its exon 1 part. In control experiments, neither LacZ fragments nor PleN1 in combination with control IgG significantly precipitated α-dystrobrevin and β-synemin.

The results presented so far indicate that plectin PleN1 fragments can interact with α-dystrobrevin, F-actin and β-synemin. To assess whether plectin 1 forms a molecular complex with all of these three proteins, we performed a blot overlay assay. PleN1 fragments were immobilized on nitrocellulose membrane, overlaid and incubated with one of the three proteins or the mixture of all the proteins, followed by detection of bound proteins using antibody specific for each protein. As shown in Fig. 7, comparison of resulting signals indicated no significant differences between overlays of each protein alone and of the three proteins. These results suggest a possibility that plectin 1 can form molecular complexes with all of α-dystrobrevin, F-actin and β-synemin.

To explore in vivo association of β-synemin and plectin with costameric components including DGC and cytoskeletal proteins, β-synemin and plectin immune complexes were immunoprecipitated from lysates of muscle LM by anti-β-synemin and anti-plectin antibodies. Subsequently, the presence or absence of proteins in the immune complexes was determined by immunoblotting using antibodies against α-dystrobrevin, dystrophin, pan-actin, (meta)vinculin, desmin and α-actinin. All of the proteins examined, with the exception of α-actinin, were detected in both β-synemin and plectin immune complexes, but control IgG immune complex did not significantly contain any of these proteins (Fig. 8). Desmin coimmunoprecipitated more abundantly with β-synemin than plectin, whereas other proteins, such as dystrophin, metavinculin, α-dystrobrevin 1, α-dystrobrevin 2 and actin, coimmunoprecipitated in greater amounts with plectin. As reported previously (Hijikata et al., 2003), metavinculin, rather than vinculin, was preferentially immunoprecipitated by either of the antibodies.

Next, anti-nonmuscle actin (β- and γ-actin) antibody was used to determine whether costameric nonmuscle actin is associated with β-synemin and plectin in the immune complexes. Both the immune complexes included nonmuscle actin, but its levels were higher in plectin immune complexes. These results suggest that nonmuscle actin, possibly costameric γ-actin, is involved in the interactions of β-synemin and plectin with DGC and other costameric components beneath the sarcolemma. In fact, γ-actin was found to localize predominantly to the costameric sarcolemma, where it seemed to be associated with DGC through dystrophin (Rybakova et al., 2000).

Since plectin and β-synemin indirectly interacted with dystrophin through α-dystrobrevin, as described above, we postulated that their expression and/or distribution might be affected by the deficiency of dystrophin. Therefore, the expression and distribution of plectin and β-synemin in dystrophin-deficient skeletal muscles were explored by immunoblotting and immunohistochemistry, and compared with those in control muscles. In addition, those of other costameric proteins including α-dystrobrevin were also examined. The specimens were obtained from anterior tibial muscles of wild-type mice and dogs, and dystrophin-deficient X-chromosome-linked muscular dystrophy (mdx) mice and canine X-chromosome-linked muscular dytstrophy in Japan (CXMDJ) dogs.

The expression levels of plectin and β-synemin were significantly elevated in CXMDJ dogs relative to wild-type dogs, whereas they were almost equivalent between wild-type and mdx mice (Fig. 9). For utrophin, integrinβ1d and metavinculin, their expressions were increased in mdx mice and CXMDJ dogs as compared with wild-type controls. However, statistically significant differences were found only between CXMDJ and wild-type dogs, but not between mdx and control mice. The expression levels of vimentin showed a significant increase in mdx and CXMDJ as compared with wild-type controls, whereas those of α-dystrobrevin 1, α-dystrobrevin 2 and α-dystrobrevin 3 represented a significant decrease in dystrophin-deficient muscles relative to controls. The expression levels of vinculin, desmin and α-actinin were almost equivalent between wild-type and mdx or CXMDJ muscles.

The distributions of plectin and β-synemin were slightly altered in dystrophin-deficient muscles of mdx mice and CXMDJ dogs as compared with normal controls (supplementary material Fig. S2A-F). Dystrophin-deficient muscle fibers showed more intense staining for β-synemin and plectin along the sarcolemma as compared with normal muscles. More conspicuous phenotypic alterations were found in IF networks of CXMDJ dogs, but not in those of mdx mice. Aberrant IF networks and/or abnormal sarcoplasmic deposits or aggregates were observed in CXMDJ muscle fibers. Aberrant IF networks appeared as larger and more irregular honeycomb structures when compared with normal controls. The sarcoplasmic deposits were positive for β-synemin, plectin, vimentin and desmin, but negative for utrophin, α-dystrobrevin and integrin β1d. These aberrant IF networks and sarcoplasmic deposits were found mainly in regenerating fibers that were positive for vimentin or developmental myosin staining (supplementary material Fig. S2G-I).

It has been long assumed that desmin IFs anchor onto the costameric sarcolemma. Consistent with this assumption, our previous immuno-EM study ultrastructurally demonstrated that plectin linked desmin IFs to dystrophin- and vinculin-containing subsarcolemmal dense plaques, or costameres (Hijikata et al., 2003). Except for plectin, however, proteins involved in the linkages from IF to costameres and their molecular organization remained largely unknown. Recently, direct association of plectin 1f to dystrophin and β-dystroglycan was shown as a linkage between IFs and DGC (Rezniczek et al., 2007). Extending this knowledge further, this study provides new insights into the molecular organization from IF to costameres, by demonstrating that plectin 1 can bind actin, α-dystrobrevin and β-synemin, whereas β-synemin can also bind actin as well as α-dystrobrevin.

Unlike other IF proteins, β-synemin can bind actin and actin-associated proteins. These abilities are supported by several lines of evidence, provided by our results and those of others: (1) β-synemin tail fragments slightly co-sedimented with F-actin in the actin co-sedimentation assay; (2) β-synemin was found along SFLS, including actin, in C2C12 myotubes; (3) β-synemin coimmunoprecipitated with sarcomeric and nonmuscle actin from muscle LM lysates; (4) previous yeast-two-hybrid analyses using avian synemin cDNA as a bait demonstrated direct interactions of synemin with α-actinin and vinculin (Bellin et al., 1999; Bellin et al., 2001). However, mammalian β-synemin is unlikely to interact with α-actinin, although it seems to bind vinculin. In the present immunoprecipitation experiment, β-synemin immuno-complexes from LM lysates contained metavinculin but not α-actinin. This absence of β-synemin-interaction with α-actinin was possibly due to the lack of the approximately 300-residue-long C-terminal sequences of avian synemin that are required for binding α-actinin in rat β-synemin molecules.

In costameres, β-synemin seems to serve as an actin-associated protein rather than an IF protein to form heteropolymeric IFs, and is unlikely to link IFs to costameric components. As demonstrated by our immuno-EM analysis, β-synemin appeared to be incorporated into costameric dense plaques without associating with IF. Similar β-synemin-association with the sarcolemma, independent of IF, was found in muscle fibers of desmin knockout mice that lacked desmin IFs (Carlsson et al., 2000). Such β-synemin-associations with the sarcolemma possibly occur through its interactions with α-dystrobrevin, actin, and/or metavinculin within costameres, where β-synemin might contribute to the integration of actin-based molecular architectures. However, β-synemin molecules were incorporated into heteropolymeric IFs as well. These β-synemin molecules, which are heteropolymerized with desmin, would not interact with α-dystrobrevin or dystrophin anymore and, therefore would not link IFs to costameres, because desmin, α-dystrobrevin and dystrophin confine their binding sites onto β-synemin rod domain (Bhosle et al., 2006; Mizuno et al., 2001). This notion is supported by the present immunoprecipitation results, indicating that β-synemin associated with plectin 1f through a part of rod domain could not further co-precipitate with α-dystrobrevin.

Four plectin isoforms (plectin 1, plectin 1b, plectin 1d and plectin 1f) generated by alternative splicing are predominantly expressed in skeletal muscles (Fuchs et al., 1999). They differ from each other only in their small N-terminal sequences that are encoded by their own first exon (exon 1, exon 1b, exon 1d and exon 1f, respectively). A recent study using isoform-specific antibodies and isoform-expression constructs revealed the differential targeting of plectin isoforms in skeletal muscle fibers or myotubes (Rezniczek et al., 2007). Plectin 1 and 1f are preferentially associated with the sarcolemma, whereas plectin 1d is localized exclusively to Z-discs. The differential targeting of plectin 1 to the sarcolemma could be explained by the present finding that its exon 1 part bound α-dystrobrevin. This binding property would target plectin 1 to the sarcolemma associated with α-dystrobrevin. However, the differential localization of plectin 1f at the sarcolemma cannot be explained, because nothing is known of proteins specifically interacting with its exon 1f part. As presented here, plectin 1f fragments did not interact with α-dystrobrevin. Nevertheless, their unique exon 1f parts may interact with other costameric proteins or different domains of α-dystrobrevin, e.g. its exon parts 1 to 7, ultimately leading to its localization at the sarcolemma.

At its N-terminus, plectin possesses multifunctional CH domains to interact with multiple proteins, in addition to the plakin domains. The CH domains contain binding sites for actin (Andra et al., 1998), integrin β4 (Geerts et al., 1999; Rezniczek et al., 1998), vimentin (Sevcík et al., 2004), nesprin-3 (Wilhelmsen et al., 2005), the nonreceptor tyrosine kinase Fer (Lunter and Wiche, 2002), dystrophin and utrophin (Rezniczek et al., 2007), whereas the plakin domain was found to bind β-dystroglycan (Rezniczek et al., 2007). The present study has added β-synemin to the lists of CH-domain-binding or plakin-domain-binding proteins. Moreover, an isoform-specific binding site was disclosed in plectin 1, which has unique exon 1 part to bind β-synemin and α-dystrobrevin. These accumulating data underline a versatile binding property of N-terminal parts of plectin. However, this versatile binding property raises questions of how plectin selectively determines and regulates interactions with binding partners.

A model of molecular organization from IF to costameres including DGC and integrin complexes is proposed based on the results of the present and previous studies (Fig. 10A). Plectin 1 interacts with heteropolymeric IFs including desmin and β-synemin through its C-terminal IF-binding domain (Hijikata et al., 2003; Reipert et al., 1999), whereas its N-terminal part binds α-dystrobrevin, costameric actin and β-synemin. The costameric actin bound by plectin 1 might be associated with dystrophin and (meta)vinculin (Rybakova et al., 2000; Witt et al., 2004), whereas β-synemin on plectin 1 molecules might interact further with costameric actin. Moreover, α-dystrobrevin associated with plectin 1 might also bind β-synemin. In addition to this model, another one has been recently proposed (see Fig. 10B), in which plectin 1f links IFs directly to dystrophin and β-dystroglycan within DGC (Rezniczek et al., 2007). This direct binding of plectin to and its indirect interactions through actin with dystrophin well agreed with our previous findings that treatment of plectin immune complexes from LM lysates with gelsolin decreased dystrophin within the immune complexes, but did not completely remove it (Hijikata et al., 2003).

With respect to linking IFs to the BM, costameres in skeletal muscles are quite analogous to hemidesmosomes in epidermal cells, although costameric components are much more numerous and organized in more complicated manner than hemidesmosomal ones. Costameres including DGC and integrin complexes are extracellularly associated with the BM through α-dystroglycan and integrin α7β1d (Belkin et al., 1996; Eravsti and Campbell, 1993), whereas they are intracelluarly linked to IFs by plectin (Hijikata et al., 2003; Rezniczek et al., 2007; Schröder et al., 2002). Both the analogous structures are assumed mechanically to stabilize the plasma membrane and to protect cell against mechanical stress. Attesting this assumption, similar disruptions of plasma membranes and cell structures were reported to occur in skin or skeletal muscle by the deficiency or defects of constituents of hemidesmosomes or costameres, such as plectin (Andra et al., 1997; Gache et al., 1996; McMillan et al., 2007), IF proteins (desmin and keratin) (Chan et al., 1994; Li et al., 1997) and other membranous or membrane-associated proteins (integrin α7, dystrophin, α-dystrobrevin, sarcoglycan, integrin α6, integrin β4, BP180) (Dowling et al., 1996; Duclos et al., 1998; Grady et al., 1999; Hack et al., 1998; Hoffman et al., 1987; Huber et al., 2002; Mayer et al., 1997; Pulkinnen et al., 1997; van der Neut et al., 1996).

Quantitative alterations of costameric components were found in dystrophin-deficient muscles. In contrast to the reduction of DGC components (Ozawa, 2006; Straub and Campbell, 1997), other costameric components, such as plectin, β-synemin, (meta)vinculin, talin and integrins, increase their expressions in dystrophin-deficient muscles, as demonstrated in the present and previous studies (Hodges et al., 1997; Law et al., 1994; Rezniczek et al., 2007). Increased synemin- and plectin-staining along the dystrophin-deficient sarcolemma were also noted by the present and previous immunohistochemical studies (Schröder et al., 1997). Moreover, costameric γ-actin is also more abundantly expressed in mdx muscles lacking dystrophin, compared with control muscles (Hanft et al., 2006). These increased expressions of costameric components, including integrin complexes, might be a compensatory or an adaptive cellular response to unstable costameres and unstable anchorages of IF on costameres.

As a compensatory response in dystrophin-deficient muscles, increased β-synemin and plectin 1, together with costameric γ-actin, might preserve the subsarcolemmal localization of α-dystrobrevin, which loses a main binding partner dystrophin beneath the sarcolemma. As presented in this study, β-synemin and plectin 1 can bind α-dystrobrevin. Another possibility is that sarcoglycan-sarcospan complex also contributes to recruitment of α-dystrobrevin to the sarcolemma, because the complex interacts with the N-terminal portion of α-dystrobrevin and is still present in the dystrophin-deficient sarcolemma (Hack et al., 2000; Ozawa et al., 2000; Yoshida et al., 2000). In this context, it is worth noting that utrophin, a structurally related protein that can compensate for lack of dystrophin, might not retain α-dystrobrevin at the sarcolemma. α-Dystrobrevin 2 localized at the extrasynaptic sarcolemma was found not to interact with utrophin (Peters et al., 1998).

DGC and integrin complexes within costameres would have the similar function of linking the BM to IFs through plectin, costameric actin and β-synemin. This would result in stabilizing the sarcolemma and protecting it against contraction-imposed stress. Their functional similarity would be reasonably supposed, given that the reduction of DGC induced a compensatory response of increased expression of integrin complexes. However, the two complexes would also have their own functional roles, because dystrophin-deficient muscles still undergo degeneration despite increased expression of integrin complexes. This fact implies that increased integrin complexes are not sufficient, either quantitatively or qualitatively, to compensate for functional roles of DGC. The two complexes in combination with plectin and IFs might serve as platforms for distinct signal transduction, because DGC and integrin complexes are associated with distinct proteins involved in different signaling, i.e. nNOS and FAK, respectively (Brenman et al., 1996; Pham et al., 2000).

To clone cDNA of a plectin-binding protein, a rat skeletal muscle 5′-stretch plus cDNA library (Clontech) was screened with DNA probes, which were prepared by PCR using the cDNA library and a pair of primers predicted from partial amino acid sequences (5′-CCNCAYGARTTYCAY-3′ for PHEFH and 5′-RTCNACCATYCTYTG-3′ for QRMVD). This screening identified 47 positive clones that did not contain a 5′ sequence with an ATG start codon. Therefore, to clone the full-length cDNA, we constructed a cDNA library from single-stranded DNAs generated by reverse transcription using mRNA obtained from skeletal muscles of newborn rats, random hexamers and gene-specific primers based on the sequences of incomplete cDNAs (5′-CATGCGCTCAGGCAACGTTTCCTCCAGCTG-3′ and 5′-TTGACTCTGTCTGGTCAGGGCAGACAGCTG-3′). This cDNA library was screened with a new probe that had been generated from a positive cDNA clone obtained in the previous screening to identify ten overlapping cDNA clones including the full-length clone.

To obtain cDNA clones encoding the N-terminal parts of various plectin isoforms (exon 1∼exon 30), a partial cDNA library was constructed (as described above) with random hexamers, and rat plectin-specific primers (5′-TTCATCCAGAGCCTGCAGGCCCGCTGCTT-3′, 5′-TCTCTGCCTCGGCTTGGCACAGCCACCGTT-3′, and 5′-TTCTCCAGCTCCTGCTCAGCCAGCTCTCGC-3′). This library was screened with DNA probes generated using rat plectin cDNA fragments (plectin 1, 1-1311bp), which were obtained by RT-PCR using total RNA of new-born rat skeletal muscles and a pair of rat plectin 1-specific primers (5′-ATGGTGGCTGGCATGCTCATG-3′ and 5′-GTGGTGCCGGATCCACTHTAG-3′). Thirty-three positive clones were identified and found to include various plectin isoforms, such as plectin 1, 1a, 1b, 1d, 1f and 1e.

α-Dystrobrevin cDNA encoding exon 8 to exon 16 was amplified from the rat skeletal muscle 5′-stretch plus cDNA library by PCR using a pair of primers (5′-GCNAAYGTNGARAAYGTNTTYCAYCCNGTNG-3′ and 5′-RTAYTCYTCCATYTGNAGYTCRTTYTCNAG-3′). These primers were designed based on the published sequences of mouse and human α-dystrobrevin 1. A cDNA of approximately 1.5 kb was obtained and sequenced to confirm that it was an α-dystrobrevin cDNA fragment encoding amino acid residues 232-669 of α-dystrobrevin isoform1 isoform11 (XP_001054793).

β-Synemin constructs encoding the following segments of the protein were subcloned into pGEX-6p-1 glutathione S-transferase (GST) bacterial expression vector (GE Healthcare): full-length (amino acids 1-1255), Rod N (1-130), Rod Ml (56-286), Rod Ms (57-223), Rod C (169-318), Tail N1 (303-578), Tail N2 (303-432), Tail M (482-930), Tail C1 (822-1255), Tail C2 (931-1102), and Tail C3 (1071-1255). Similarly, cDNA fragments encoding full-length β-galactosidase (LacZ), the N-terminal portion of plectin 1 including exon 1 (PleN1, amino acids 1-1273), the N-terminal part of plectin 1f including exon 1f (PleN1f, 1-1121), exon 1 (Ex1, 1-181), calponin homology domain (CHDα, 183-507), various plakin domain fragments (PlD-N, 413-736; PlD-MC, 707-1273; PlD-M, 707-950; PlD-Cl, 854-1273; PlD-Cs, 965-1273) were subcloned into the pGEX-6p-1. All of these plectin fragments and β-galactosidase contained the Myc-tag epitope at their C-termini. α-Dystrobrevin cDNA fragment obtained from the rat skeletal muscle cDNA library was subcloned into the pET-19b vector carrying an N-terminal 6xHis tag sequence (Novagen). Escherichia coli BL21 (DE3) was transformed with the constructs described above. Expression of GST or 6xHis fusion proteins was induced by 0.2 mM IPTG for 3 h at 28°C. Purification of GST or 6xHis fusion proteins was performed as described in the respective manufacturer's protocols. GST was removed from plectin recombinant fragments and LacZ by digestion with PreScission protease (GE Healthcare) on glutathione bead columns.

Rabbit β-synemin antisera were raised against recombinant rat synemin tail fragments (amino acids 822-1255). The antisera were purified by passage over affinity columns conjugated with the synemin tail fragments. Other antibodies used for immunohistochemistry, immunoblotting, and immunoprecipitation were as follows: polyclonal and monoclonal anti-plectin antibody [(Hijikata et al., 2003) clone 7A8; Sigma-Aldrich], monoclonal and polyclonal anti-α-dystrobrevin antibody [Clone 23; BD Transduction Laboratories (Yoshida et al., 2000)], monoclonal anti-pan-actin antibody (C4; Novus Biological, Inc.), polyclonal anti-nonmuscle (β- and γ-) actin antibody (Cosmo Bio Co. Ltd.), monoclonal anti-integrin β1d antibody (clone 2B1; Chemicon International), monoclonal anti-α-actinin antibody (sarcomeric EA-53; Sigma-Aldrich), polyclonal and monoclonal anti-desmin antibody (Progen Biotechnik GmbH, DE-U-10; Sigma-Aldrich), monoclonal anti-Myc antibody (clone 9E10; Roche), monoclonal anti-vinculin antibody (clone hVin-1; Sigma-Aldrich), monoclonal anti-vimentin antibody (clone VIM13.2; Sigma-Aldrich), polyclonal anti-utrophin antibody (Imamura and Ozawa, 1998), monoclonal anti-dystrophin antibody (clone Dy8/6C5; Novocastra, clone mandy8; Sigma-Aldrich), monoclonal anti-developmental myosin heavy-chain antibody (clone RNMy2/9D2; Novocastra).

The secondary antibodies used in the present study were as follows: Alexa Fluor 488 goat anti-mouse IgG(H+L), Alexa Fluor 488 or Alexa Fluor 594 goat anti-rabbit IgG(H+L) (Invitrogen), peroxidase-conjugated goat anti-mouse IgG(H+L), goat anti-rabbit IgG(H+L) (Pierce), and rabbit anti-chicken IgY antibody (Promega), goat anti-rabbit IgG(H+L) conjugated to 5-nm gold particles (BioCell Research Laboratories).

C2C12 cells (2.0×10⁵ cells) were cultured on collagen-coated Aclar coverslips within 35-mm dishes in growth medium (DMEM containing 20% FCS, 100 U/ml penicillin G and 100 μg/ml streptomycin). Plasmid DNA was prepared by subcloning exon 1 (Ex1), calponin homology domains (CHDα), plakin domain (PlD-M) cDNA fragments including Myc sequence into pZac expression vector (kindly provided by James M. Wilson. Plasmid DNA (4 μg) was transfected into the cells within each dish by using Lipofectamine™ 2000 (Invitrogen). After washing out plasmid DNA, the transfected cells were cultured in growth medium overnight, and then their differentiation was initiated by switching the medium to the DMEM medium containing 5% horse serum, 10 μg/ml insulin, and the antibiotics. After 2 or 3 days, C2C12 myotubes were fixed with chilled (–20°C) methanol and processed for immunostaining (Hijikata et al., 1997).

Cryosections of rat skeletal muscles (diaphragm and tibialis anterior) were prepared and immunostained as described previously (Hijikata et al., 1999). These sections were observed under a confocal scanning laser microscope (Fluoview FV1000, Olympus). For F-actin staining in C2C12 cells, Alexa Fluor 594-conjugated phallotoxins (Invitrogen) were utilized.

For immunoelectron microscopy, small bundles of muscle fibers were carefully teased from glycerinated muscle strips, chemically skinned with 50 μg/ml saponin in EGTA rigor solution, immunolabeled, processed for thin-section EM using tannic acid enhancement, and observed, as described previously (Hijikata et al., 2003).

For pull-down assay, purified GST-synemin recombinant fragments were incubated with Myc-tagged PleN1 in the incubation buffer (50 mM HEPES pH 7.0, 10% glycerin, 1 mM DTT, 0.5% NP-40) for 3-4 hours at 4°C. The reaction mix was further incubated with anti-Myc mAb for 3-4 hours and then with protein L-agarose beads (Pierce) with rocking at 4°C overnight. The beads carrying the immune complexes were washed six times with the same buffer. The immune complexes were eluted by addition of SDS sample buffer. The Myc-tagged plectin recombinant fragments were also incubated with and pulled down by GST-synemin fragments Tail N1 and glutathione beads. Prior to the incubation, Myc-tagged plectin recombinant fragments were incubated with glutathione beads and precleared by centrifugation to remove contaminating plectin fragments still fused to GST. The glutathione beads carrying GST-synemin fragments associated with plectin fragments were washed six times with the incubation buffer and eluted with SDS sample buffer at 95°C.

For in vivo immunoprecipitation, light microsomes (LM), prepared from rat skeletal muscles according to the procedure described by Ohlendieck et al. (Ohlendieck et al., 1991), were lysed with solubilization buffer (150 mM NaCl, 1% Triton X-100, 0.1% SDS, 15% glycerol, 50 mM Tris-HCl pH 7.5) containing protease inhibitors (100 μg/ml PMSF, 2 μg/ml leupeptin). After centrifugation, the supernatant was precleared with protein G-Sepharose (Sigma-Aldrich). The precleared supernatant was incubated with either polyclonal anti-plectin or anti-synemin antibody and then with protein G-Sepharose overnight at 4°C. The beads carrying the immune complexes were washed three times with solubilization buffer, three times with solubilization buffer without SDS, and once with TBS (50 mM Tris-HCl, 150 mM NaCl, pH 7.5), followed by incubation with SDS sample buffer at 95°C.

Actin was purified from rabbit skeletal muscle as described previously (Matsumura et al., 1983). The purity of actin was more than 97%, determined by SDS-PAGE. The actin was allowed to polymerize in the presence of recombinant plectin and synemin fragments in actin polymerization buffer (20 mM Tris-HCl pH 7.5, 2 mM MgCl₂, 100 mM KCl, 0.5 mM ATP, 0.1 mM β-mercaptoethanol) for 1 h at room temperature. Actin filaments with bound proteins were sedimented by centrifugation for 1 hour at 100,000 g and 20°C, and corresponding amounts of pellet and supernatant were analyzed by SDS-PAGE.

Plectin PleN1 fragments (1 μg) were immobilized on nitrocellullose membranes, which were blocked in TBS containing 5% BSA and 0.2% Tween-20 for 5 hours at 4°C. Subsequently, membranes were overlaid and incubated with 3 μM actin or 1 μM α-dystrobrevin or 1 μM β-synemin or the mixture of the three proteins in 120 μl overlay buffer (20 mM HEPES pH 7.5, 150 mM NaCl, 2 mM MgCl₂, 1 mM DTT, and 3.5% BSA) overnight at 4°C. Bound proteins were detected by using the protein-specific antibody, HRP-conjugated secondary antibody and ECL system (GE Healthcare).

Blot membranes treated with ECL solutions were scanned and evaluated using luminescent image analyzer LAS-3000 and Multi Gauge software (Fuji film). The mean value of spot intensities measured in the overlay with a single protein was calculated, and then each spot intensity was represented relative to this mean value by calculating the ratio of measured value per the mean value. Similarly, intensity of each band obtained in immunoblottings of control and dystrophin-deficient muscles was represented relative to the mean value of intensities measured in control muscles.

We thank J. M. Wilson (University of Pennsylvania) for providing pZac vector. This work was supported by a Research Grant (17A-10) for Nervous and Mental Disorders from the Ministry of Health, Labor and Welfare, and by the High-Tech Research Center Project for Private Universities (MEXT. HAITEKU, 2004-2008).