Binding of ADAM12, a marker of skeletal muscle regeneration, to the muscle-specific actin-binding protein, alpha -actinin-2, is required for myoblast fusion.

ADAM12 belongs to the transmembrane metalloprotease ADAM ("a disintegrin and metalloprotease") family. ADAM12 has been implicated in muscle cell differentiation and fusion, but its precise function remains unknown. Here, we show that ADAM12 is dramatically up-regulated in regenerated, newly formed fibers in vivo. In C2C12 cells, ADAM12 is expressed at low levels in undifferentiated myoblasts and is transiently up-regulated at the onset of differentiation when myoblasts fuse into multinucleated myotubes, whereas other ADAMs, such as ADAMs 9, 10, 15, 17, and 19, are expressed at all stages of differentiation. Using the yeast two-hybrid screen, we found that the muscle-specific alpha-actinin-2 strongly binds to the cytoplasmic tail of ADAM12. In vitro binding assays with GST fusion proteins confirmed the specific interaction. The major binding site for alpha-actinin-2 was mapped to a short sequence in the membrane-proximal region of ADAM12 cytoplasmic tail; a second binding site was identified in the membrane-distal region. Co-immunoprecipitation experiments confirm the in vivo association of ADAM12 cytoplasmic domain with alpha-actinin-2. Overexpression of the entire cytosolic ADAM12 tail acted in a dominant negative fashion by inhibiting fusion of C2C12 cells, whereas expression of a cytosolic ADAM12 lacking the major alpha-actinin-2 binding site had no effect on cell fusion. Our results suggest that interaction of ADAM12 with alpha-actinin-2 is important for ADAM12 function.

of ADAMs share a high sequence homology and domain organization with the class III snake venom metalloprotease-disintegrins. Both ADAMs and snake venom metalloprotease-disintegrins contain a metalloprotease-like domain with an associated regulatory prodomain, a disintegrin-like domain, a cysteine-rich domain, and an epidermal growth factor-like domain. In addition, ADAMs contain a transmembrane domain and a cytoplasmic tail (1)(2)(3)(4). ADAMS with active metalloprotease are involved in diverse and important cellular processes (5)(6)(7). ADAM17 (tumor necrosis factor converting enzyme; TACE), ADAM10 (Kuzbanian), and ADAM9 function in the shedding of the ectodomain of membrane-anchored proteins, such as the cytokine tumor necrosis factor-␣ (8,9), the tumor necrosis factor receptor, transforming growth factor-␣ (10,11), the heparin-binding epidermal growth factor-like growth factor (12), and the cleavage of the amyloid precusor protein (13)(14)(15). ADAM12 is also an active metalloprotease, which binds to ␣2-macroglobulin in vitro (16), but its physiological substrates have not been identified.
The other extracellular domains of ADAMs, the disintegrin and the cysteine-rich domains, also play important roles in cell-cell and cell-matrix adhesion, cell differentiation, and fusion. The disintegrin-like domains of ADAMs are the most highly conserved. They seem to have retained the ability to bind to integrins and play a role in cell adhesion processes such as fertilization, as suggested by the interaction of sperm ADAM 2 (fertilin ␤) with integrin ␣ 6 ␤ 1 on the egg (17). Two recent reports show that the disintegrin-like domain and the cysteinerich domain of ADAM12 mediates adhesion of C2C12 cells and tumor cell lines (18,19). ADAMs 1, 3, 12, and 14 possess in their cysteine-rich domain a motif similar to viral fusion peptides (1), and these ADAMs are expressed in cells that participate in cell fusion events, such as fertilization, myogenesis, and osteogenesis. However, the direct involvement of such a motif in any cell fusion process has not been demonstrated yet.
In contrast to the well conserved organization of the extracellular regions of ADAMs, their cytoplasmic tails are much less conserved. Approximately half of the known ADAM proteins have cytoplasmic tails of up to 200 amino acids containing consensus SH3-binding motifs, whereas others have shorter cytoplasmic tails that lack any identifiable motif. ADAM9 contains two SH3-binding motifs and binds to Src SH3 domains, but not to Abl SH3 domains in vitro (20). A recent report from Howard et al. (21) identified two SH3-domain-containing proteins that bind to ADAM9 and ADAM15. ADAM9 also binds to protein kinase C through its cytoplasmic tail, and this interaction regulates the metalloprotease activity of ADAM 9 toward its substrate HB-epidermal growth factor (12). Thus, these data suggest that the cytoplasmic tails of ADAMs play a role in signal transduction to regulate the activity of the extracellular region. The cytoplasmic tail of ADAM12 is among the longest of all ADAM proteins and contains potential SH3-binding motifs, suggesting of interactions with cytoplasmic proteins.
Skeletal muscle differentiation and regeneration require several events, including activation and inhibition of growth factors, matrix remodeling, cell-cell interaction, and cell fusion. Some of these processes could be controlled by ADAMs. Several ADAMs are expressed in skeletal muscle. Among them, ADAM12 may play a role in muscle differentiation, as it is required for fusion of C2C12 mouse myoblasts into multinucleated myotubes (22). Moreover, a short form of a soluble splice variant of ADAM12, which is expressed in human placenta, has myogenic activity in vivo (23).
To better understand the role of ADAMs in skeletal muscle development and regeneration, we used the C2C12 myoblast cell line, derived from mouse satellite cells, as a model. We analyzed the expression pattern during C2C12 differentiation of six known ADAMs predicted to be active metalloproteases, ADAMs 9,10,12,15,17,and 19. We show that ADAM12 expression is induced at the onset of fusion, whereas the other ADAMs are constitutively expressed. We also show that ADAM12 is not expressed in normal adult muscle but is reexpressed in regenerating muscle.
We also used the yeast two-hybrid system to identify cytoplasmic proteins interacting with the cytoplasmic tail of ADAM12. We show that ADAM12 interacts with the musclespecific ␣-actinin-2 in vitro and in vivo. The major site of interaction is located in the membrane-proximal region of ADAM12 tail. Disruption of interaction between ADAM12 and ␣-actinin-2 in C2C12 cells by overexpression of an ADAM12 cytoplasmic tail prevents cell fusion. Our results identify for the first time a potential linkage between an ADAM and the cytoskeleton mediated by an actin-binding protein and suggest that skeletal muscle regeneration requires this interaction.

EXPERIMENTAL PROCEDURES
Culture of C2C12 Cells-C2C12 cells (ATCC) were grown in supplemented Dulbecco's modified Eagle's medium (Dulbecco's modified Eagle's medium with 2 mM glutamine, 1 mM sodium pyruvate, 100 g/ml streptomycin, and 100 g/ml penicillin) containing 10% fetal calf serum. When cells reached confluence, differentiation was induced by culturing in supplemented Dulbecco's modified Eagle's medium containing 2% horse serum.
Detection of Various ADAMs mRNAs in C2C12 Cells by Reverse Transcription-Polymerase Chain Reaction (RT-PCR)-Total RNA was extracted from subconfluent proliferating C2C12 cells (D-1), confluent C2C12 cells (D0), or C2C12 cells at various days after differentiation (D1 to D7) using the Trizol Reagent (Roche Molecular Biochemicals) according to the manufacturer's instructions. mRNAs for various AD-AMs, myogenin, and ␤-tubulin were detected by one-step RT-PCR Ti-Tan (Roche Molecular Biochemicals) using 1 g of total RNA per reaction. The sequences of primers and size of PCR products are summarized in Table I.
Detection of ADAM12 Protein in C2C12 Cells by Immunoblotting-Undifferentiated and differentiating C2C12 cells were detached from the culture dishes with 5 mM EDTA and pelleted by centrifugation. Cells were lysed in lysis buffer (20 mM Tris, pH 8, 150 mM NaCl, 1 mM MgCl 2 , 1 mM CaCl 2 50 mM octylglucosid, 10 g/ml aprotinin, 10 g/ml leupeptin, and 10 g/ml pepstatin). Insoluble material was removed by a quick centrifugation at 1000 rpm, and the supernatant was then centrifuged at 12,000 rpm for 45 min to recover a membrane-enriched pellet, which was extracted in lysis buffer. The protein concentration of each extract was determined using the Bio-Rad assay. Ten g of protein were electrophoresed on 4 -20% polyacrylamide gels (Novex), and proteins were transferred onto a nitrocellulose filter. ADAM12 was detected by an antiserum generated by immunizing a rabbit with a peptide corresponding to the C-terminal end of mouse ADAM12 (CHQVPRPSHNAYIK). The mouse ␤ 1 D integrin antiserum has been described (24). Antibodies were used at a 1/200 dilution.
Immunodetection of ADAM12 in Normal and Regenerating Mouse Muscle-Indirect immunofluorescence was performed as described (25). Immunostaining was performed on muscle from mdx dystrophin-deficient mice, used as a model for spontaneous muscle regeneration. ADAM12 was detected with an antiserum generated against a peptide corresponding to the C-terminal end of human ADAM12 (CHQVPRST-HTAYIK), almost identical to the mouse sequence previously mentioned. The antiserum was diluted 1/200.
Plasmid Constructions-The entire cytoplasmic tail of mouse ADAM12 was amplified using PCR from a corresponding full-length ADAM12 cDNA-plasmid and directionally inserted in-frame with the DNA binding domain of the LexA transcription factor in plasmid pBTM116 (kindly provided by Dr. S. Tartare-Deckert). DNA sequencing confirmed in-frame insertion and absence of mutations. All deletions and mutations of the cytoplasmic tail of ADAM12 were made by PCR of the specific regions and cloned into pBTM116 using the same procedure.
Yeast Two-hybrid Screening-A human skeletal muscle cDNA library, constructed in plasmid pGAD10 containing sequences encoding the GAL4 activation domain, was obtained from CLONTECH. One and a half million transformants of the yeast strain L40 were screened according to the protocol described by Hollenberg et al. (26).
Preparations of GST Fusion Proteins-The entire cytoplasmic tail of ADAM12 (ADAM12cyt) was sub-cloned into pGEX-5 ϫ 1 (Amersham Pharmacia Biotech) and expressed in XL1-blue bacteria cells (Stratagene). The procedure for extraction of GST fusion proteins was standard except that 1.5% N-lauroylsarcosine was added to the lysate before extraction with 2% Triton X-100. GST alone was prepared in parallel, and GST fused to the SH2 domain of Shc (herein referred to as GST-SH2) was kindly provided by Dr. V. Dodelet.
In Vitro Transcription and Translation-A plasmid for in vitro transcription of the full-length ␣-actinin-2 was kindly provided by Dr. A. Beggs. In vitro transcription and translation was performed with the TNT kit from Promega according to the manufacturer's instructions.
In Vitro Binding Assay-GST, GST-ADAM12cyt, or GST-SH2 fusion proteins coupled to glutathione beads (Amersham Pharmacia Biotech) were incubated overnight at 4°C with 5 l of [ 35 S]Met-labeled, in vitro translated ␣-actinin-2 in 200 l of binding buffer (2 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.5 mM dithiothreitol, 0.1% Triton X-100, 10 g/ml aprotinin, 10 g/ml leupeptin, and 10 g/ml pepstatin). The beads were then washed four times in washing buffer (25 mM Tris-HCl, pH 7.5, 1 mM EDTA, 0.5 mM dithiothreitol) and boiled in SDS sample buffer. Samples were electrophoresed on polyacrylamide gels, and the gels were treated with Amplify (Amersham Pharmacia Biotech), dried, and exposed to x-ray film overnight. Equivalent amounts of GST fusion proteins were electrophoresed on a separate gel and stained with Gelcode blue stain (Pierce).
Transfection of C2C12 Cells with ADAM12 Cytoplasmic Tail Constructs-The entire ADAM12 cytoplasmic tail (ADAM12cyt) was expressed in pCDNA3.1 vector (Invitrogen). For the constructs myr-ADAM12cyt and myr-⌬45ADAM12cyt, a myristoylation motif coding sequence (GGGAGTAGCAAGAGCAAG) was inserted in-frame into the 5Ј end of ADAM12cyt cDNA or -⌬45ADAM12cyt, ADAM12cyt cDNA ADAM12 Binding to ␣-Actinin-2 encoding a truncated ADAM12cyt lacking the N-terminal 45 amino acids. C2C12 cells were plated in 12-well dishes and grown to 70% confluence. Two g of vector containing ADAM12cyt, myr-ADAM12cyt, myr-⌬45ADAM12cyt, or vector without insert and 0.2 g of CMV-␤-gal vector were cotransfected into cells using the Fugene 6 reagent (Roche Molecular Biochemicals) in Dulbecco's modified Eagle's medium with 10% fetal calf serum. The next day, cells were switched to differentiation medium. Staining for ␤-gal was performed on the cells after 4 days of differentiation. Each transfection was done in duplicate, and the number of blue myotubes was determined in 8 fields for each duplicate. The transfection efficiency was similar for each vector construct, as determined by the number of ␤-galactosidase-positive cells. In addition, analysis of the efficiency of expression of each construct transfected was performed by immunoblotting of cell lysates.
In Vivo Binding of ADAM12 Cytoplasmic Tail with ␣-Actinin-2-ADAM12cyt, myr-ADAM12cyt, and myr-⌬45ADAM12cyt constructs were cotransfected with ␣-actinin-2 cDNA-plasmid in CHO cells. Cells were lysed in lysis buffer in 50 mM Tris-Cl, pH 7.5, 150 mM NaCl, 1% Triton X-100, 0.5% deoxycholate, and 10 g/ml each of pepstatin A, leupeptin, and aprotinin. After incubation on ice for 15 min, the lysate was centrifuged at 10,000 rpm for 20 min, and the supernatant was recovered. Immunoprecipitation was performed by incubating the supernatant with 4 l of the polyclonal mouse ADAM12cyt antibody or 4 l of the corresponding preimmune serum for 4 h at 4°C under gentle agitation. Thirty l of protein A-Sepharose was then added for an additional 1 h of incubation. After three washes in the lysis buffer, 25 l of sample buffer was added to the pelleted beads, and the immunoprecipitated proteins were resolved by SDS-polyacrylamide gel electrophoresis. Immunoblotting was then performed as described above with the monoclonal sarcomeric ␣-actinin antibody (clone EA-53, Sigma) at dilution of 1:5000 and with the polyclonal mouse ADAM12cyt antibody at dilution of 1:1000.

Expression of ADAMs in C2C12
Cells-C2C12 cells have been extensively used as an in vitro model of skeletal muscle differentiation. As these cells are satellite cells derived from an adult mouse, they can also be considered as a model for skeletal muscle regeneration (27). Therefore to define the roles of AD-AMs in muscle regeneration, we analyzed by RT-PCR the expression of six ADAM proteins, all of which are potentially active metalloproteases. As shown in Fig. 1a, undifferentiated C2C12 myoblasts express mRNAs for all ADAMs tested, namely, ADAMs 9, 10, 12, 15, 17, and 19. During differentiation of myoblasts to myotubes, monitored here by an increase in myogenin mRNA expression, ADAM12 expression increased in the early stages and then decreased between day 2 and 3, whereas the expression of the other ADAMs tested remained constant throughout differentiation.
The changes in ADAM12 expression during differentiation of C2C12 cells were confirmed at the protein level by immunoblotting of extracts of a membrane-enriched fraction from C2C12 cells. An antibody directed against the C-terminal end of ADAM12 (Fig. 1b) detected a 100-kDa protein, presumably corresponding to the full-length ADAM12, and a 54-kDa protein, likely representing a processed form of ADAM12 lacking the metalloprotease domain, as described previously (22). During C2C12 differentiation, here monitored by the increase in ␤ 1 D integrin, full-length ADAM12 and particularly the processed form increased until day 2 and then decreased to low levels in more differentiated cells (Fig. 1b). The changes in expression of ADAM12 during C2C12 differentiation, observed at both mRNA and protein levels, suggest a role for ADAM12 in skeletal muscle regeneration, specifically at the onset of cell fusion.
Expression of ADAM12 in Fetal, Adult, and Regenerating Skeletal Muscle-Immunolocalization of ADAM12 in adult skeletal muscle revealed no staining. In contrast, in mdx muscle, which undergoes spontaneous regeneration (28), the small, newly formed muscle fibers were strongly positive for ADAM12 (Fig. 2). These results suggest that ADAM12 expression in skeletal muscle is restricted to developing myofibers during embryonic development, as described previously (29), and during regeneration in the adult, as shown here.
Interaction of ADAM12 with the Cytoskeleton-We screened a human skeletal muscle cDNA library in a yeast two-hybrid system using the ADAM12 cytoplasmic tail as a bait (ADAM12cyt). Out of 1.5 ϫ 10 6 transformants, we isolated FIG. 1. Expression of ADAMs during myogenic differentiation of C2C12 cells. a, mRNA for ADAMs 9, 10, 12, 15, 17, and 19 was detected by RT-PCR as described under "Experimental Procedures." Expression of ADAMs 9, 10, 15, 17, and 19 is constant during differentiation. ADAM12 is expressed in myoblasts until day 1 after differentiation and is then down-regulated. A positive marker of myogenic differentiation, myogenin is up-regulated during differentiation. ␤-Tubulin mRNA was monitored to assess equivalent amounts of RNA from each reaction. b, immunoblotting analysis of ADAM12 during C2C12 differentiation. Ten g of protein from a membrane fraction of C2C12 cells harvested at different days of differentiation was separated by SDS-polyacrylamide gel electrophoresis under reducing conditions, transferred to nitrocellulose membranes, and reacted with a polyclonal antibody against mouse ADAM12 (upper panel) or against ␤ 1 D integrin (lower panel). ADAM12, particularly its processed form, increased up to day 2 of differentiation. Arrowheads indicate the two major processed forms of ADAM12: the form that lacks the prodomain, with a molecular mass of approximately 97 kDa (open arrowhead), and the fully processed form, which lacks the protease domain, with a molecular mass of approximately 54 kDa (filled arrowhead). The muscle-specific ␤ 1 D integrin, a marker of differentiation, was induced after 2 days in differentiation medium. The arrow indicates the doublet forms of ␤ 1 D integrin (140 and 130 kDa).

FIG. 2. Expression of ADAM12 in regenerating muscle.
Immunostaining of ADAM12 in regenerating dystrophic muscle. mdx dystrophic muscle undergoes active regeneration between 1 and 2 months of age. ADAM12 is up-regulated in the small, newly formed fibers in the muscle of mdx mice. Bar, 10 m. several clones interacting with ADAM12cyt but not with an irrelevant bait, lamin C. Among them, eight distinct clones encoding the muscle-specific isoform of ␣-actinin, ␣-actinin-2, were isolated. As linkage of ADAM12 to the cytoskeleton may be important for its function, we further characterized the ␣-actinin binding. Based on the overlaps between the ␣-actinin-2 clones, it is likely that the binding site for ADAM12 is contained within the COOH-terminal half of ␣-actinin, between half of the spectrin-type repeat 3 and the domain with EF-hand-like repeats (Fig. 3).
Interaction of ADAM12 with ␣-Actinin-2 in Vitro-To confirm the interaction of ADAM12cyt with ␣-actinin-2, we performed in vitro binding assays. ␣-Actinin-2 synthesized by in vitro translation in the presence of [ 35 S]methionine was incubated with GST-ADAM12cyt and control fusion proteins coupled to glutathione-Sepharose beads. Gel electrophoresis of proteins bound to the beads and autoradiography of the gel showed that ␣-actinin-2 bound to GST-ADAM12cyt but not to GST, GST-SH2, or the glutathione beads alone (Fig. 4a). The quantities of GST proteins in the assay were estimated on a separate gel (Fig. 4b).
Identification of the ␣-Actinin-2 Binding Sites in ADAM12-To further characterize the interaction of ADAM12cyt with ␣-actinin-2, we generated deletion constructs of ADAM12cyt and tested them for binding to ␣-actinin-2 in yeast. As shown in Fig. 5, M-C and M-G, composed of the membrane-proximal 44 and 28 amino acids, respectively, interacted as strongly as did ADAM12cyt with ␣-actinin-2, whereas other constructs, M-D, -E, -F, and -H, did not bind at all. The C-terminal truncated ⌬2, ⌬3, and ⌬4, which contain all the proximal part of ADAM12cyt, bound to ␣-actinin-2 as expected. Surprisingly, constructs ⌬5ЈA and ⌬5ЈB, which lack the proximal 45 and 32 amino acids, respectively, but retain the rest of the cytoplasmic tail, bound to ␣-actinin-2, although less strongly (Fig. 5a). The binding affinity was estimated from ␤-galactosidase activity: 2.5 h of color development was required to obtain the same intensity as that obtained within 20 min with the M-C construct. This suggests that the first 30 amino acids of ADAM12cyt contain a major ␣-actinin-2 binding site, whereas another region, which is probably conformational, as it can not be mapped by using different deletions of ADAM12cyt, contains a minor site of interaction.
The sequence of the membrane-proximal region of ADAM12cyt is unique among the cytoplasmic domains of AD-AMs and might therefore confer ␣-actinin-2 binding to ADAM12 only among the proteins of the ADAMs family. Two arginine residues, at positions 21 and 26 from the membrane-proximal portion, are conserved between mouse and human ADAM12. To test whether these positively charged residues could be involved in the interaction with ␣-actinin-2, we mutated either residue arginine 21 or arginine 26 to leucines in the M-G deletion construct and tested binding to ␣-actinin-2. Neither of these mutations affected the interaction of M-G with ␣-actinin-2 (Fig. 5b), suggesting that the binding site for ␣-actinin-2 is probably contained within the first residues of the membrane-proximal portion of ADAM12cyt.
The predicted structure of the 30 amino acids of the membrane-proximal region of ADAM12cyt is a helix, which is also a feature of some other longer cytoplasmic domains of ADAMs, such as ADAM15. To test the hypothesis that conformation of the membrane-proximal helix accounts for binding to ␣-actinin-2, we analyzed binding of the mouse ADAM15 (metargidin) cytoplasmic tail with ␣-actinin-2. We found that the ADAM15 cytoplasmic tail bound ␣-actinin-2, although less strongly than did ADAM12cyt (Fig. 5b). This result indicates that other ADAM cytoplasmic domains that possess structural homologies with ADAM12cyt can bind ␣-actinin.
Overexpression of ADAM12 Cytoplasmic Tail Has a Dominant-negative Effect on Fusion of C2C12 Cells-To investigate the physiological significance of ADAM12cyt binding to ␣-actinin-2, we disrupted this interaction by overexpressing a fulllength cytoplasmic tail of ADAM12 in C2C12 cells and analyzed the effects of such an overexpression on C2C12 cell fusion. To optimize the potential competition between soluble ADAM12cyt and the endogenous ADAM12 cytoplasmic tail, we FIG. 3. Schematic representation of ␣-actinin-2 and the ␣-actinin-2 clones binding to ADAM12 in the yeast two-hybrid screen. ␣-Actinin-2 contains a globular N-terminal actin-binding domain, a rod-like domain with four spectrin-like repeats, and a C-terminal domain with two calcium-binding domain (EF-hands). Mapping of the ␣-actinin-2 clones defines a putative binding region of ␣-actinin-2 with ADAM12 between the C-terminal half of repeat 3 and the C terminus. targeted ADAM12cyt to the plasma membrane by adding a myristoylation site to the N terminus of the peptide. We cotransfected a ␤-gal reporter plasmid (ratio, 10:1) to identify transfected cells and to score, after 5-bromo-4-chloro-3-indolyl ␤-D-galactopyranoside (X-gal) staining, the number of myotubes resulting from fusion between ADAM12cyt-transfected myoblasts. Overexpression of myristoylated ADAM12-cyt reduced fusion by 70%, overexpression of ADAM12cyt without myristoylation reduced fusion by 30%, and a truncated, myristoylated ADAM12cyt, missing the proximal region containing the major binding site for ␣-actinin-2, had no effect on fusion (Fig. 6a). This experiment was repeated at least three times. FIG. 5. Identification of the ␣-actinin-2 binding sites of ADAM12cyt. Deletion constructs of ADAM12 were tested for interaction with ␣-actinin-2 by the yeast two-hybrid system using a ␤-galactosidase filter assay (a), and the results are summarized in b. The strength of twohybrid interactions of the various deletion constructs with ␣-actinin-2 was estimated on the basis of ␤-galactosidase color development. Numbers in b refer to amino acid residues at the boundaries of each construct. The major ␣-actinin-2-binding site was located in the first 28 amino acid residues. A second, minor site was located in the membrane-distal region of the ADAM12 cytoplasmic domain.
ADAM12 Binding to ␣-Actinin-2 13937 Immunoblotting with an antibody specific for the C-terminal of the ADAM12 cytoplasmic tail showed that all ADAM12 constructs were equally expressed in transiently transfected CHO cells (Fig. 6b). These results suggest that a membrane-targeted ADAM12cyt competes with wild type ADAM12 for binding to ␣-actinin-2 and that the ADAM12-␣-actinin interaction is essential in cell fusion.
In Vivo Binding of ADAM12 Cytoplasmic Tail with ␣-Actinin-2-To confirm that the soluble ADAM12 cytoplasmic tails bind to ␣-actinin-2, we did immunoprecipitation experiments by cotransfecting ADAM12 cytoplasmic tail constructs and ␣-actinin-2-encoding plasmid in CHO cells. Immunoprecipitations were performed with polyclonal ADAM12 antibody or preimmune serum, and a monoclonal ␣-actinin antibody was used for immunoblotting. ␣-Actinin-2 was specifically coimmunoprecipitated with myristoylated ADAM12cyt (Fig. 7a). Fig.  7b shows that ␣-actinin-2 was better coimmunoprecipitated with the myristoylated ADAM12cyt than with the ADAM12cyt without myristoylation, suggesting that membrane-targeted ADAM12 cytoplasmic tail binds ␣-actinin-2 more efficiently. Moreover, much less ␣-actinin-2 was coimmunoprecipitated with the truncated, myristoylated ADAM12cyt, which is consistent with the fact that the truncated ADAM12cyt lacks the major binding site for ␣-actinin-2. These results demonstrate that ADAM12 cytoplasmic tail binds ␣-actinin-2 in vivo and that capacity of inhibition of fusion by the ADAM12 cytoplasmic tails is correlated with their capacity to bind ␣-actinin-2.

DISCUSSION
The present study shows that ADAM12 is a marker of skeletal muscle regeneration and is specifically up-regulated at the onset of myoblast fusion. We show that the cytoplasmic domain of ADAM12 interacts with the muscle-specific ␣-actinin-2, and we provide evidence that this interaction is necessary for ADAM12 to promote muscle cell fusion.
The major ␣-actinin-2 binding site in the ADAM12 cytoplasmic domain was mapped to the first membrane-proximal 30 amino acids. By comparing all known bindings of transmembrane proteins to ␣-actinin, Heiska et al. (39) noted that positively charged and hydrophobic amino acids are the only common elements in the ␣-actinin binding site of these proteins. The sequence KRKTLMRLLFTHKK in the membrane-proximal region of mouse ADAM12 fits this concept. A second and minor binding site for ␣-actinin-2 was found in the distal region of the ADAM12 cytoplasmic domain. As we could not identify a precise binding site in this region, it is possible that the site requires a particular conformational structure in this area of the ADAM12 cytoplasmic tail that is not maintained in the fragments of the tail. This second binding site could reinforce or stabilize the interaction of ␣-actinin-2 with the major binding site.
The amino acid sequence of the major binding site for ␣-actinin-2 is unique to ADAM12 in the superfamily of ADAM proteins, suggesting that the property of binding ␣-actinin-2 could be specific to ADAM12. However, the predicted structure of this domain is primarily helical and is, in this respect, similar to the corresponding domain of several other ADAMs with long cytoplasmic domains. 2 It is therefore possible that other ADAMs might bind ␣-actinin. Indeed, we observed binding between the ADAM15 (metargidin) cytoplasmic domain and ␣-actinin-2 in yeast. Although the ADAM15-␣-actinin interaction was not examined in detail, this observation suggests that other ADAMs may bind ␣-actinin and therefore be linked to the cytoskeleton. It further suggests that the binding of ADAM12 to ␣-actinin is not specific to ␣-actinin-2, which is muscle-specific; ADAM12 may also bind to other ␣-actinins. Co-immunoprecipitation of ␣-actinin-2 and ADAM12cyt. a, CHO cells were cotransfected with ␣-actinin-2 and myr-ADAM12cyt (myrA12). Immunoprecipitation was done with ADAM12cyt antibody or preimmune serum (Ϫ), and Western blot was performed with ␣-actinin antibody. The same blot was reprobed with ADAM12cyt antibody to confirm the presence of myr-ADAM12cyt. b, CHO cells were cotransfected with ␣-actinin-2 and myr-ADAM12cyt (myrA12), ADAM12cyt (A12), or myr-⌬45ADAM12cyt (myr⌬45A12). Immunoprecipitations were done with ADAM12cyt antibody, and Western blots were performed first with ␣-actinin antibody and then with ADAM12cyt antibody.
Early evidence for a role for ADAM12 in muscle differentiation was presented by Yagami-Hiromasa et al. (22), when they showed that transfection of C2C12 cells with antisense mRNA encoding ADAM12 inhibited cell fusion. The same study showed that overexpression of full-length ADAM12 also inhibited myoblast fusion. In contrast, overexpression of a truncated soluble ADAM12 lacking the metalloprotease domain enhanced fusion, suggesting that the metalloprotease domain negatively regulates cell fusion (22). In addition, a truncated and soluble ADAM12 lacking the metalloprotease domain promotes ectopic muscle formation in tumor cells grown in nude mice (23). The mechanism by which ADAM12 controls myoblast fusion is unknown.
Our study shows for the first time that an ADAM is linked to the cytoskeleton and provides evidence that this linkage might be necessary for ADAM12 function during myoblast fusion. This study brings new insight to how ADAM12 functions in skeletal muscle cells and suggests that the cytoskeleton may regulate distribution of ADAM12 on the cell surface, thereby exposing its metalloprotease and disintegrin domains in specialized areas of the cell surface, where localized proteolysis and/or cell-cell contacts need to occur.