Association of Syncoilin and Desmin LINKING INTERMEDIATE FILAMENT PROTEINS TO THE DYSTROPHIN-ASSOCIATED PROTEIN COMPLEX*

, We recently identified a novel protein called syncoilin, a putative intermediate filament protein that interacts with (cid:1) -dystrobrevin, a member of the dystro-phin-associated protein complex. Syncoilin is found at the neuromuscular junction, sarcolemma, and Z-lines and is thought to be important for muscle fiber integrity. Based on the similar protein structure and cellular localization of syncoilin and desmin, we proposed that these proteins interact in vivo . The data presented confirm an interaction between syncoilin and desmin and demonstrate their co-localization in skeletal muscle. Intriguingly, whereas these proteins interact, COS-7 cell expression studies show that desmin and syncoilin do not assemble into heterofilaments. Fur-thermore, fractionation assay and immunofluorescence study of H2K myoblasts and myotubes suggest that, unlike typical intermediate filament proteins, syncoilin does not participate in filament formation with any protein. However, it is possible that syncoilin is involved in the anchoring of the desmin intermediate filament network at the sarcolemma and the neuromuscular junction. This interaction is likely to be important for maintaining muscle fiber integrity and may also link the dystrophin-associated protein complex to the cytoskeleton. The dysfunction or absence of syncoilin may result in

The muscular dystrophies are a group of clinically heterogeneous diseases characterized by muscle wasting. Mutations in various muscle genes have been shown to be responsible for such disorders. Duchenne muscular dystrophy is caused by deletions and mutations in dystrophin, which encodes a protein of 427 kDa expressed in skeletal and cardiac muscle and in brain (1,2). In skeletal muscle, dystrophin is located at the sarcolemma, where it interacts with a number of proteins, including the ␣-dystrobrevins, to form the dystrophin-associ-ated protein complex (DAPC) 1 (3,4). Many members of this complex have already been implicated in muscle disorders, demonstrating the importance of the DAPC in muscle function. It is widely held that one function of the DAPC is to provide a molecular link between the actin cytoskeleton and the extracellular matrix, thereby sustaining sarcolemmal integrity during muscle contraction (3,5,6).
We recently identified a new member of the DAPC, syncoilin, through its interaction with ␣-dystrobrevin-1 and -2 in muscle (7). Syncoilin is a 64-kDa protein found in skeletal and cardiac muscle and was proposed to be a member of the intermediate filament (IF) protein superfamily based on sequence analysis (7). IF proteins are characterized by their ability to form 10nm-diameter filaments and form the basis of the higher eukaryotic cytoskeleton together with thin filaments and microtubules (for review, see Ref. 8). All IF proteins share a common structural organization, consisting of an N-terminal head domain, a C-terminal domain, and a highly conserved central rod domain. On the basis of sequence homology within this central region, IFs are classified into six groups, and syncoilin is most similar to type III and IV IFs such as ␣-internexin (7). Although syncoilin has the central coiled-coil domains typical of IF proteins, it failed to form filamentous structures in transfected COS-7 cells, suggesting that other factors may be required to initiate heteropolymeric filament formation (7). Whereas the patterns of expression of IFs vary, these proteins are often considered to play a role in structural support and the mechanical preservation of cellular space (for review, see Refs. 8 and 9). Desmin, for example, is the main IF protein in mature striated muscle cells and was postulated to maintain the integrity of muscle fibers by linking adjacent myofibrils together and linking peripheral myofibrils to the sarcolemma (10,11). Its localization at the sarcolemma (10), the Z-lines (11), and the neuromuscular junction (NMJ) (12), where syncoilin is expressed, makes it an ideal candidate binding partner of syncoilin. In view of the structural roles generally performed by IF proteins, the association of syncoilin with ␣-dystrobrevin is important because it may provide an alternate route by which the DAPC links to components of the cytoskeleton, in this case, the IF network.
The aim of this study was to investigate our hypothesis that syncoilin is associated with IF proteins at the sarcolemma and the NMJ. We therefore performed a yeast two-hybrid assay with syncoilin as bait. Desmin was identified as a putative binding partner of syncoilin. This interaction was confirmed in vitro and in vivo, through co-localization and co-immunoprecipitation assays. The ability of desmin and syncoilin to inter-act provides evidence for an important association between IF proteins at the Z-lines in muscle and the extracellular matrix via the DAPC. Mutations in the desmin gene cause myopathy together with cardiomyopathy (13). Syncoilin, like desmin, is up-regulated in some muscular dystrophies, making it a candidate gene for muscular dystrophy as well as cardiomyopathy.
Library Screening-The construct Syn5.1:pDB and 10 -20 g of a mouse skeletal muscle cDNA library (Invitrogen) were co-transformed into the yeast strain MaV203 (Invitrogen) and plated onto SC-Leu-Trp-His media (Sigma), containing 83 mM 3-aminotriazole (3-AT) (Invitrogen). Sequential transformation of these constructs was also performed. Co-transformed and sequentially transformed colonies were incubated for 2 days at 30°C, replica-cleaned, and further incubated for 3 days. Growth was confirmed by re-streaking resultant colonies onto fresh SC-Leu-Trp-Hisϩ100 mM 3-AT plates. Three master plates were created by patching colonies onto SC-Leu-Trp with yeast control strains a-e (Invitrogen) and replica-plated onto SC-Leu-Trp-Ura, SC-Leu-Trpϩ0.2% 5-fluoroorotic acid (5-FOA), and SC-Leu-Trp-Hisϩ100 mM 3-AT media. Yeast colonies streaked onto nitrocellulose membranes were assayed for ␤-galactosidase activity. DNA was isolated from positive clones and electroporated into Escherichia coli DH10B electrocompetent cells (Invitrogen). The 5Ј ends of the plasmids were sequenced using vector primer pDBFor (5Ј-GAATAAGTGCGACATCATCATC-3Ј). To confirm the interaction in yeast, the prey plasmids were co-transformed with Syn5.1:pDB and assayed for growth on selective plates and ␤-galactosidase activity. Negative control experiments were performed using empty vectors, pDBLeu and pEXP-AD502, to verify that the prey and the bait plasmids did not nonspecifically activate the yeast reporter genes.
Co-immunoprecipitation Experiments-Skeletal muscle was dissected from a normal adult C57 BL/10 mouse, flash-frozen in liquid nitrogen, and homogenized in 8 ml of solubilization buffer (150 mM NaCl, 1% (v/v) Nonidet P-40, 0.05% (w/v) SDS, and 50 mM Tris, pH 7.4) containing protease inhibitors (Sigma). Samples were incubated on ice for 30 min and centrifuged at 30,000 rpm at 4°C for 45 min. The following procedures were performed at 4°C unless stated otherwise. 1.5 mg of soluble protein, in a volume of 1 ml, was pre-cleared with 50 l of protein G-Sepharose (Amersham Biosciences, Inc.) on a blood mill for 3 h. Anti-syncoilin antibody SYNC-FP (10 l) and anti-desmin antibodies (10 l) DE-U-10 (Sigma) and Y-20 (Santa Cruz Biotechnology) were applied and incubated for 18 -24 h. 50 l of protein G was then added to capture the immune complexes for 4 h. The beads were washed four times with solubilization buffer by centrifugation at 1000 rpm for 2 min, eluted in 60 l of treatment buffer (75 mM Tris, pH 6.8, 3.8% (w/v) SDS, 4 M urea, 5% (v/v) ␤-mercaptoethanol, and 20% glycerol), and analyzed by Western blotting. A sample lacking primary antibody acted as a negative control.
Western Blotting-10 l of samples from immunoprecipitation experiments was separated on 10% SDS-polyacrylamide gels and electrophoretically transferred onto nitrocellulose membranes (Schleicher and Schuell). After blocking the membranes overnight at 4°C in 5% milk in TBST, SYNC-FP (1:200) and anti-mouse desmin monoclonal antibody DE-U-10 (Sigma) (1:2000) were applied and incubated for 1 h. The membranes were washed three times with TBST, and horseradish peroxidase-conjugated donkey anti-rabbit (Jackson Immuno Research Laboratories) and horseradish peroxidase-conjugated donkey antimouse (Sigma) antibodies were applied for 1 h. After washing with TBST, the membranes were developed using BM chemiluminescence substrate system (Roche Molecular Biochemicals) following the manufacturer's instructions.
Immunofluorescence Microscopy-Cryosections (8 m) of mouse quadricep muscle from normal C57 BL/10 mice were prepared as described previously (14). H2K myoblasts were cultured as described previously and allowed to differentiate for 4 days into myotubes (15). Sections were fixed in 0.5% paraformaldehyde for 15 min, whereas cultured cells (H2K myoblasts, myotubes, and transfected COS-7 cells) were fixed in 4% paraformaldehyde. The sections and cells were then permeabilized with 0.05% Triton X-100 in PBS for 15 min and blocked in 10% (v/v) fetal calf serum in PBS for 30 min at room temperature. The anti-syncoilin antibody SYNC-FP (7), anti-desmin goat polyclonal antibody Y-20 (Santa Cruz Biotechnology), and the anti-dystrobrevin rabbit polyclonal antibody ␤CTFP (16) were diluted at 1:50, 1:100, and 1:400 in PBS, respectively, and applied to the sections for 1 h at room temperature. Samples were washed twice with PBS and incubated for 1 h with donkey anti-goat fluorescein isothiocyanate-conjugated antibody (Jackson Immuno Research Laboratories) and donkey anti-rabbit rhodamine red antibody (Jackson Immuno Research Laboratories). Samples were washed twice with PBS and mounted with Vectashield (Vector Laboratories) and imaged using a Leica DMLD Light Microscope (Leica Microsystems Imaging Solutions, Cambridge, United Kingdom) and Leica DMRE TCS SP Cofocal Microscope (Leica Microsystems Imaging Solutions).
Fractionation of Muscle and Cellular Extracts-Protein extract from normal C57 BL/10 mouse skeletal muscle and transfected COS-7 cells were fractionated as described previously (17). Muscle or cellular extract was resuspended in 8 and 3 ml of nuclear extraction buffer, respectively (10 mM HEPES-KOH, pH 7.9, 0.5% Triton X-100, 0.5 M sucrose, 0.1 mM EDTA, 10 mM KCl, 1.5 mM MgCl 2 , 1 mM dithiothreitol, and protease inhibitors; Sigma). The resuspension volume remained constant throughout. The samples were homogenized for 20 s and centrifuged briefly at 10,000 ϫ g, and the supernatant was collected as the cytosol fraction. The resulting pellets were further lysed in nuclear extraction buffer II (nuclear extraction buffer supplemented with 0.5 M NaCl and 5% glycerol) at 4°C for 30 min. The samples were then centrifuged at 14,000 ϫ g for 20 min, and the supernatant fractions were collected as the nuclear fractions, whereas the pellets were homogenized in nuclear extraction buffer II and taken as the cytoskeletal fractions.

Identification of Desmin as a Potential Binding Partner of Syncoilin by Yeast Two-hybrid Analysis
To elucidate the role of syncoilin in muscle, a yeast twohybrid assay was performed to identify putative binding partners. The bait construct Syn5.1:pDB, containing the syncoilin coiled-coil and C terminus regions, was utilized to screen a skeletal muscle cDNA library. The full-length syncoilin cDNA was unsuitable because it nonspecifically activated transcription of the yeast reporter genes. A total of ϳ5.5 ϫ 10 6 clones were screened in two independent experiments, resulting in Ͼ400 positive colonies. Interactions between bait and prey plasmids were confirmed by replica plating onto selection media and assaying for ␤-galactosidase activity. Fifty-six colonies demonstrated the expected phenotype; they grew vigorously on SC-Leu-Trp-Ura and SC-Leu-Trp-Hisϩ100 mM 3-AT media, 5-FOA inhibited their growth, and they demonstrated ␤-galactosidase activity. Sequence analysis identified 30 clones that contained the entire coding sequence of desmin and 3 clones that contained various regions of desmin (nucleotide 238 -, 904 -, and 1022-end). A re-transformation assay was performed, in which the plasmid containing full-length desmin cDNA was co-transformed with the Syn5.1:pDB construct. The growth pattern of resultant yeast colonies was compared with control strains representing positive and negative interactors (Invitrogen). The phenotypic behavior of yeast colonies that contain syncoilin and full-length desmin is shown in Fig. 1. In brief, colonies demonstrated a phenotype similar to the positive strain, reaffirming the interaction between syncoilin and desmin. In negative control experiments, the desmin and syncoilin plasmids were co-transformed with empty vectors, resulting in no interaction. Thus, the interaction between syncoilin and desmin was not due to nonspecific activation of the yeast reporter genes.

Co-immunoprecipitation of Syncoilin and Desmin
To confirm the interaction of syncoilin and desmin, a coimmunoprecipitation experiment was performed on mouse skeletal muscle protein extracts. Prior to the assay, the solubilities of the two proteins were assessed. Desmin was found mainly in the insoluble fraction, although a small amount was found to be soluble. Syncoilin, in contrast, was found primarily in the soluble fraction (data not shown). When co-immunoprecipitation experiments were performed on the soluble extracts, syncoilin was detected in the immune complexes generated using two different desmin antibodies ( Fig. 2A). In the reciprocal experiment, the use of SYNC-FP resulted in the precipitation of desmin (Fig. 2B). A negative control lacking primary antibody resulted in no detection of desmin or syncoilin, indicating that protein G did not nonspecifically interact with either protein.

Co-localization of Syncoilin and Desmin in Skeletal Muscle
To demonstrate that syncoilin and desmin co-localize in normal muscle, the expression patterns of these proteins were assessed by immunofluorescence microscopy. Fig. 3 shows normal C57 BL/10 quadricep cryosections immunolabeled with anti-syncoilin antibody SYNC-FP (Fig. 3, A and D) and antidesmin antibody Y-20 (Fig. 3, B and E). The overlay of the two antibodies is shown in Fig. 3, C and F. Syncoilin, as reported previously (7), is concentrated at the NMJ of normal mouse skeletal muscle (as confirmed by double staining experiments with ␣-bungarotoxin; data not shown) and weakly stains the sarcolemma (Fig. 3A). On longitudinal sections, syncoilin was found at the Z-lines and along the sarcolemma (Fig. 3D). Desmin also displayed a similar staining pattern at the NMJ, the Z-lines, and the sarcolemma (Fig. 3, B and E). The overlay of both SYNC-FP and Y-20 (Fig. 3, C and F) confirms the co-localization of these proteins and hence supports an interaction between syncoilin and desmin at the NMJ, Z-lines, and sarcolemma.

In Vitro Expression Studies of Syncoilin and Desmin in COS-7 Cells
Immunofluorescence Microscopy-To further investigate the interaction of syncoilin and desmin, mammalian expression constructs containing the entire open reading frames of syncoilin and desmin were transfected into COS-7 cells independently and in parallel. The resultant protein distribution was detected by indirect immunofluorescence. Syncoilin expression was detected using the SYNC-FP antibody, whereas desmin expression was detected using the anti-desmin antibody Y-20. FIG. 1. Yeast two-hybrid analysis of desmin and syncoilin interaction. The interaction between syncoilin and desmin was examined in yeast using four selective systems: SC-Leu-Trp-Hisϩ100 mM 3-AT, SC-Leu-Trp-Ura, ␤-galactosidase assay, and SC-Leu-Trpϩ0.2% 5-FOA. Results were compared with control strains a-e (Invitrogen), which contain interacting proteins of varying strengths. The Positive column indicates the phenotype obtained with a control strain containing strong interactors. Colonies grew on SC-Leu-Trp-Hisϩ100 mM 3-AT and SC-Leu-Trp-Ura media and exhibited ␤-galactosidase activity, and 5-FOA inhibited their growth. When noninteractors were used (shown in the Negative column), colonies failed to grow on SC-Leu-Trp-Hisϩ100 mM 3-AT or SC-Leu-Trp-Ura media, did not exhibit ␤-galactosidase activity, and grew on media containing 5-FOA. Full-length desmin clone co-transformed with Syn5.1:pDB resulted in yeast colonies that grew on SC-Leu-Trp-Hisϩ100 mM 3-AT and SC-Leu-Trp-Ura media and exhibited ␤-galactosidase activity, and 5-FOA inhibited their growth. All four phenotypes are consistent with the behavior of an interactor. Negative control experiments involved co-transformation of the desmin clone with empty vector pDBLeu and Syn5.1:pDB with empty vector pEXP-AD502 and resulted in a phenotype indicative of noninteractors.  Fig. 4 shows a mixed population of transfected cells. In cells transfected with syncoilin only, SYNC-FP staining was concentrated in the cytoplasm, but syncoilin failed to form filaments, as reported previously (Fig. 4A) (7). In cells positive for desmin but negative for syncoilin, cage-like filamentous structures were observed (Fig. 4C). The filaments were far apart and were bundled adjacent to the nucleus. In some cell clusters, desmin also formed an extended structure spanning multiple cells. Single transfection with a desmin expression construct also resulted in filament formation in all transfected cells (data not shown). When both syncoilin and desmin were present, instead of self-assembling into an IF network, desmin assumed a punctate pattern in the majority of cells, and its immunoreactivity coincided with that of syncoilin (Fig. 4B). In some cells, a few filaments were visible, but they were unorganized and tended to be more densely packed. The cage-like desmin structure prevalent in singly transfected cells was not observed in any co-transfected cell. Irrespective of the staining pattern of desmin, syncoilin assumed a punctate cytoplasmic pattern in all transfected cells. Thus, not only did syncoilin and desmin fail to form filaments, the overexpression of syncoilin modified or inhibited the ability of desmin to self-assemble into 10-nm filaments. This ability of syncoilin to relocalize desmin demonstrated that the two proteins could interact in co-transfected cells.
To demonstrate the specificity of this interaction, mammalian expression constructs containing the entire open reading frames of ␣-dystrobrevin-1 and desmin were transfected into COS-7 cells in parallel. In co-transfected cells, desmin was able to form filaments that were indistinguishable from those found in singly transfected cells, showing that the modification or disruption of the desmin network was specific to syncoilin (Fig.  4D).
Fractionation of Transfected COS-7 Cells-Fractionation experiments were performed to investigate the subcellular localization of syncoilin and desmin in transfected COS-7 cells and in skeletal muscle. In COS-7 cells singly transfected with desmin, desmin was concentrated in the cytoskeletal fraction (Fig. 5A). In contrast, desmin was found in both the cytosolic and cytoskeletal fractions of co-transfected cells (Fig. 5B). These results support the observation that desmin assembled into insoluble filaments when singly transfected, whereas cotransfection of syncoilin caused the disassembly of desmin filaments into soluble units. Syncoilin, in contrast, was detected mostly in the cytosolic fraction in singly and doubly transfected cells, as is consistent with the cytoplasmic staining observed (Fig. 5, C and D). This lends support to the observation that syncoilin failed to form filaments with desmin in transfected COS-7 cells.
In skeletal muscle, desmin was detected in the cytoskeletal fraction, as is consistent with the reported insolubility of IF proteins (Fig. 5E). On the other hand, syncoilin was found mainly in the cytosolic fraction, although it was sometimes also detected in the cytoskeletal fraction (Fig. 5F).

Immunofluorescence Studies of H2K Myoblasts and Myotubes
To assess the endogenous expression pattern of syncoilin and desmin, H2K myoblasts and myotubes were immunolabeled with SYNC-FP and anti-desmin Y20 (Fig. 6A). Syncoilin could be found in all cells and assumed a punctate cytoplasmic pattern. In contrast, desmin was detected in only 40 -60% of cells and assembled into a filamentous structure. Unlike the staining pattern seen in transfected COS-7 cells, desmin filaments are maintained in the presence of syncoilin.

FIG. 4. Transfection of syncoilin and desmin in COS-7 cells. Expression constructs containing the open reading frames of syncoilin and
desmin were co-transfected in COS-7 cells. Syncoilin and desmin expression was detected using SYNC-FP and Y-20. An overlay of desmin and syncoilin immunoreactivity confirmed co-localization. In cells singly transfected with syncoilin (A), syncoilin was found in a punctate pattern in the cytoplasm. In cells positive for desmin but negative for syncoilin, cage-like IF networks could be observed (C). In cells containing both desmin and syncoilin (B), syncoilin and desmin both assumed a punctate pattern, and the two proteins co-localized in the cytoplasm. As a control, expression constructs containing the open reading frames of ␣-dystrobrevin and desmin were co-transfected in COS-7 cells (D). ␣-Dystrobrevin assumed a punctate cytoplasmic staining, whereas desmin assembled into a cage-like structure similar to that observed in cells singly transfected with desmin. ϫ63 magnification.
Myoblasts were allowed to undergo fusion for 4 days in differentiation medium, and resultant myotubes were again immunolabeled with SYNC-FP and anti-desmin Y-20 (Fig. 6, B and C). In H2K myotubes, syncoilin was again found throughout all tubes. On the other hand, desmin was only detected in a fraction of myotubes with variable intensity, and staining within each myotube was nonuniform. Desmin immunoreactivity was detected as spot-like structures, bundles of filaments, and sheets of longitudinally aligned filaments on the surface of myotubes. This may represent the integration of desmin filaments in individual myoblasts into a layer of desmin filaments at various stages of differentiation.

DISCUSSION
Here we describe an interaction between syncoilin and desmin, an association that may be important for linking the desmin intermediate filament network to the DAPC. We first identified syncoilin as a binding partner of ␣-dystrobrevin-1 and -2 using a yeast two-hybrid assay, and we classified the protein as a tentative member of the IF protein superfamily by sequence analysis (7). Based on the cellular localization and the protein structure of desmin, we speculated that syncoilin and desmin might interact. We have confirmed this hypothesis by identifying desmin as a binding partner of syncoilin. Intriguingly, syncoilin failed to form filaments when co-transfected with desmin. This suggests that syncoilin is not a component of desmin filaments but may be involved in the dynamic organization of these structures. Based on the structural functions postulated for desmin and the DAPC, we propose that syncoilin is responsible for attaching/organizing desmin filaments at the sarcolemma via the DAPC to provide structural support to muscle fibers.
Desmin is the main IF protein in muscle and is located predominantly at the neuromuscular junction, the sarcolemma, and the Z-lines. In knockout studies, muscle lacking desmin generated less stress (18) and showed increased susceptibility to damage (19), implicating desmin in the maintenance of muscle fiber integrity. Syncoilin is a member of the DAPC, which is thought to act as a mechanical bridge between the basal lamina and the actin cytoskeleton to sustain sarcolemmal integrity (3,5,6). By binding desmin, syncoilin may organize desmin filaments at the sarcolemma and link the extracellular matrix via the DAPC to the intermediate filament network. This role could also be extended to a pathological state, where syncoilin is up-regulated as a possible attempt to protect against muscle damage (7).
Syncoilin, desmin, and ␣-dystrobrevin are highly expressed at the NMJ (4,7,12). The loss of ␣-dystrobrevin leads to NMJ abnormalities, including the irregular distribution of acetylcholine receptors and a 50% reduction in the number of junctional folds, suggesting a disturbance in synaptic structure (20). Together, desmin and actin have been proposed to form submembranous support for acetylcholine receptors and to mediate the excitation of acetylcholine receptors to the sarcomeric contraction system (21). Syncoilin, by linking desmin filaments to the DAPC via ␣-dystrobrevin, may be involved in the maintenance of synaptic structure.
To understand the mechanism by which syncoilin functions in muscle, we performed cell transfection experiments to assess its ability to form filaments with desmin. As a type III or IV IF protein, desmin is capable of forming filaments with itself and with other IFs, such as synemin and paranemin (22,23). Desmin successfully assembled into an IF network when transfected alone, indicating that COS-7 cells support filament formation. In contrast, syncoilin failed to form filaments when transfected on its own or when co-transfected with desmin. However, because COS-7 cells do not naturally express desmin or syncoilin, some muscle-specific cofactors necessary for heterofilament formation may be absent. Therefore, H2K myo- In co-transfected cells, a cytosolic as well as a cytoskeletal pool of desmin could be detected (B). In singly transfected and co-transfected cells, syncoilin was found predominantly in the cytosolic fraction, although it was weakly present in the other fractions (C and D). Protein extract from normal mouse skeletal muscle was similarly fractionated. Desmin was concentrated in the cytoskeletal fraction (E), whereas syncoilin was found predominantly in the cytosolic fraction, although it was weakly present in the other fractions (F).
FIG. 6. Immunofluorescence study of H2K myoblasts and myotubes. Syncoilin and desmin expression in H2K myoblasts and myotubes was detected using SYNC-FP and Y-20. In myoblasts, syncoilin assumed a punctate cytoplasmic pattern, whereas desmin showed filamentous staining (A). In myotubes, desmin was nonuniformly distributed and was present as various desmin structures (B and C). Longitudinally aligned desmin filaments could be detected on the surface of myotubes (closed arrow). Spots of desmin immunoreactivity were also detected along the myotubes (open arrow), as were small bundles of filaments. Syncoilin was present throughout the tube and could be detected in all cells. ϫ100 magnification. blasts and myotubes were utilized to study the endogenous expression of syncoilin and desmin. As seen in transfected COS-7 cells, desmin, but not syncoilin, assembled into filaments in both myoblasts and myotubes; but unlike the COS-7 cells, desmin filaments were maintained in the presence of syncoilin. This suggests that the disruption of desmin filaments may be a result of the perturbation of the natural ratio between desmin and syncoilin. The disruption of the IF network has already been described in a study in plectin, a protein thought to cross-link the desmin IF network to other subcellular structures (24). High levels of full-length plectin and the IF-binding domain of plectin caused the assembly inhibition and disassembly of vimentin and cytokeratin 5/14 filaments in vitro and in expression assays (25,26). Therefore, the collapse of desmin filaments may be caused by the competition of syncoilin for sites important for IF formation and may not represent the physiological role of syncoilin.
The immunofluorescence study of H2K cells confirmed that syncoilin did not participate in filament formation with desmin or, more importantly, with any IF protein in myoblasts and myotubes. Fractionation experiments, which showed syncoilin to be highly soluble, also suggested that syncoilin was not a component of the insoluble desmin heterofilaments in normal skeletal muscle. This result is very intriguing because most IF proteins are thought to function by polymerizing into a filamentous network to perform structural roles. Therefore, rather than associating with desmin to form filaments, syncoilin may be involved in anchoring of the desmin IF network, which may be important for linking desmin filaments to the sarcolemma and to the Z-lines. Syncoilin expression was highly independent of that of desmin in both COS-7 and H2K cells, and syncoilin was expressed before desmin in myoblasts. These observations are both consistent with the role of syncoilin as a structural support for desmin filaments.
The association between cytoplasmic IFs and the plasma membrane has been documented (27). Directly, desmin has already been shown to bind ankyrin in avian erythrocytes by solution binding assay (28), thus syncoilin may be another point at which desmin filaments are attached to the sarcolemma. Indirectly, synemin, a possible component of desmin heterofilaments, has also been shown to interact with ␣-actinin and vinculin at the sarcolemma and at the costameres (22,29). In contrast to syncoilin, whose expression is independent of desmin expression in both COS-7 and H2K cells, synemin expression is highly dependent on that of other IF proteins. In desmin knockout mice, synemin is lost at the Z-lines, where it is normally expressed (30). On the cellular level, synemin is co-localized with vimentin in SW13 vimϩ cells but is not detected in SW13 vimϪ cells, where vimentin is absent (22). This shows that synemin expression is only maintained in the presence of other IF proteins. Unlike syncoilin, there is no report of synemin up-regulation in pathological states. Therefore, it is likely that syncoilin and synemin serve different roles in muscle.
Intermediate filaments have long been implicated in a range of diseases; desmin, for example, is involved in a group of muscle disorders collectively termed "desmin-related myopathies." Desmin-related myopathies are a group of often hereditary myopathies marked by an accumulation of desmin (for review, see Ref. 31). In the majority of cases, a mutation in the desmin gene was not detected (32), hence mutations or abnormalities of desmin interactors are thought to contribute to these diseases. An example is ␣B-crystallin, a chaperone-like protein whose gene was mutated in some patients with desminrelated myopathies (33). We propose that syncoilin is responsible for anchoring desmin filaments; hence, its dysfunction or absence is likely to result in a disturbance in the desmin IF network. Syncoilin is therefore a candidate gene for this disease. Furthermore, within these desmin inclusion bodies, dystrophin was detected (34,35), giving further support to a relationship between the DAPC and desmin.
During the preparation of this manuscript, Mizuno et al. (36) reported a protein that links dystrobrevin and desmin at the sarcolemma. This protein, desmuslin, is located at the sarcolemma, whereas syncoilin is found at both the NMJ and the sarcolemma. This protein is not up-regulated under pathological conditions, unlike syncoilin. It will be interesting to see how desmuslin relates to syncoilin in both normal and disease states.
In summary, we present evidence that desmin interacts with syncoilin, a peripheral member of the DAPC. Based on the structural roles of desmin IFs and the DAPC, it is likely that syncoilin contributes to the maintenance of muscle integrity by linking the two components together. Exactly how this is achieved is unclear at present because syncoilin does not appear to behave like other IF proteins. It is likely that syncoilin is responsible for organizing the desmin IF network, which may be important for the maintenance and integrity of muscle membrane and the localization of the DAPC. Its localization at the sarcolemma, the Z-lines, and the NMJ, as well as its up-regulation in pathological muscle, suggests that syncoilin serves diverse and important roles in muscle.