A Novel Isoform of Met Receptor Tyrosine Kinase Blocks Hepatocyte Growth Factor/Met Signaling and Stimulates Skeletal Muscle Cell Differentiation*

Background: Alternative splicing gives a variation from the same gene. Results: A novel alternative splicing form of Met exists in human skeletal muscle, and its depletion decreases muscle differentiation, whereas its overexpression acts oppositely. Conclusion: A novel alternative splicing form of Met regulates skeletal muscle differentiation. Significance: This is an example of regulation by alternative splicing, which affects skeletal muscle differentiation. Hepatocyte growth factor (HGF) and its receptor, Met, regulate skeletal muscle differentiation. In the present study, we identified a novel alternatively spliced isoform of Met lacking exon 13 (designated Δ13Met), which is expressed mainly in human skeletal muscle. Alternative splicing yielded a truncated Met having extracellular domain only, suggesting an inhibitory role. Indeed, Δ13Met expression led to a decrease in HGF-induced tyrosine phosphorylation of Met and ERK phosphorylation, as well as cell proliferation and migration via sequestration of HGF. Interestingly, in human primary myoblasts undergoing differentiation, Δ13Met mRNA and protein levels were rapidly increased, concomitantly with a decrease in wild type Met mRNA and protein. Inhibition of Δ13Met with siRNA led to a decreased differentiation, whereas its overexpression potentiated differentiation of human primary myoblasts. Furthermore, in notexin-induced mouse injury model, exogenous Δ13Met expression enhanced regeneration of skeletal muscle, further confirming a stimulatory role of the isoform in muscle cell differentiation. In summary, we identified a novel alternatively spliced inhibitory isoform of Met that stimulates muscle cell differentiation, which confers a new means to control muscle differentiation and/or regeneration.

Myogenic differentiation is a cascade of intracellular events involving the coordination of muscle-specific gene expression and exit from the cell cycle, resulting in terminally differenti-ated myotubes. The ability of adult muscle tissue to grow or regenerate in response to injury depends on the activation and proliferation of quiescent satellite cells in response to growth stimuli (1,2). Their progeny myoblasts proliferate and eventually fuse with each other to replace degenerated muscle fibers in the injured area (3). Several growth factors and cytokines are implicated in stimulating or inhibiting satellite cell proliferation and differentiation (4 -6).
Activation of Met receptor tyrosine kinase upon binding with hepatocyte growth factor (HGF) 5 induces pleiotropic responses, such as proliferation, motility, morphogenesis, and angiogenesis in several cell types including tumor cells (7) and functions in vertebrate development (8 -11). In muscle, HGF is responsible for the proliferation of quiescent satellite cells and is involved in muscle regeneration (12). Met is expressed on quiescent satellite cells in normal muscle tissue (13), and the addition of HGF to cultured satellite cells promotes the entry into the cell cycle (14 -16). In addition, HGF/Met signaling is required for the migration of satellite cells to injured site (17). Although HGF can increase the number of myoblasts in regenerating muscle, it conversely inhibits muscle differentiation in vivo and in vitro (18,19), which is one example of the widely accepted dogma that differentiation and proliferation are opposite processes. Thus, HGF/Met signaling needs to be tightly regulated during the muscle differentiation process; it should be turned on at the early phase to activate the satellite cells and increase the progeny myoblasts but turned off at the late phase to induce terminal differentiation. The regulation of HGF/Met signaling during the muscle cell differentiation/regeneration is known to be achieved mainly by regulating expression levels of HGF and/or Met during muscle regeneration (18, 20 -22).
Alternative splicing is one of the regulated processes during gene expression that generates structural or functional diversities necessary to regulate various physiological processes including development and differentiation. Several spliced Met variants have been reported, such as an 8-kb major Met transcript, a 7-kb Met transcript, and Sm-Met, a small isoform of Met (23). However, except for the 8-kb major Met transcript, which yields the wild type Met protein, no known splice isoforms of Met are shown to be involved in normal human physiology.
In the present study, we found a novel alternatively spliced form of Met lacking exon 13, which yielded a C-terminal truncated Met protein having dominant negative activity. The inhibitory Met variant was induced in primary human skeletal muscle myoblasts at the onset of differentiation, stimulating differentiation process both in vitro and in vivo. Although the role of Met variants in the skeletal muscle differentiation is poorly understood, this study provides evidence that HGF/Met signaling during the skeletal muscle differentiation is inhibited by increasing the expression level of inhibitory Met variant, as well as decreasing the level of wild type Met via alternative splicing.

EXPERIMENTAL PROCEDURES
Reagents and Antibodies-Phosphotyrosine 4G10, human Met DO-24, and human Met DL-21 antibodies were from Upstate Biotechnology (Lake Placid, NY). Anti-phospho-p44/ p42 MAP kinase (pERK1/2) and anti-p44/p42 MAP kinase (ERK1/2) were from Cell Signaling Technology (Mississauga, Canada). C-28 antibody for human Met and B-2 antibody for murine Met were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA), and HGF antibody was from R&D Systems (Minneapolis, MN). Cy3-conjugated goat anti-mouse IgG was obtained from Jackson ImmunoResearch Laboratories, Inc. (West Grove, PA). Recombinant human HGF was generously provided by Dr. G. F. Vande Woude (Van Andel Research Institute, Grand Rapids, MI). HGF concentrations are presented as scatter units/ml, and 5 units are equivalent to ϳ1 ng of protein. RPMI 1640 medium, DMEM, minimum essential medium, F-10 medium, FBS, horse serum, trypsin-EDTA, antibiotic-antimycotic containing penicillin G sodium, streptomycin sulfate, and amphotericin B were obtained from Invitrogen. Protease inhibitor mixture tablet was purchased from Roche Applied Science. Phosphatase inhibitors and other reagents not specified were from Sigma.
Cell Culture and Transfections-NIH3T3 cells were obtained from American Type Culture Collection (Manassas, VA) and cultured in DMEM supplemented with 10% FBS. Transfections were performed using the Lipofectamine Plus TM (Invitrogen) according to the manufacturer's directions. The plasmid encoding human Met was a gift from Dr. G. F. Vande Woude. To generate ⌬13Met-overexpressed NIH3T3 cells, cells were co-transfected with a plasmid pSV2neo conferring resistance to G418 and a plasmid encoding the ⌬13met gene. Control cells were transfected with the resistance plasmid and empty vector. Cells were selected for 2 weeks with 800 g/ml G418 and then grown as pools.
Lentiviral Vector Production-The control lentiviral vector plasmid (pLenti-hrGFP) expresses humanized Renilla green fluorescent protein (hrGFP; Stratagene) driven by human EF1 ␣ gene promoter (24). To increase the expression of reporter gene, hrGFP gene was linked to the Woodchuck hepatitis virus post-transcriptional regulatory element (WPRE) as previously described (25). Human EF1 ␣ gene promoter and WPRE were kind gifts from Dr. Dong Wan Kim (Changwon National University, Korea) and Dr. Thomas J. Hope (Salk Institute, CA), respectively. To construct ⌬13Met lentiviral vector plasmid (pLenti-⌬13Met), hrGFP plus WPRE fragment in the pLenti-hrGFP vector was replaced with human ⌬13met gene. High titer lentiviral vector stock was produced in 293T cells by calcium phosphate-mediated transient transfection of the pseudotyped lentiviral vectors (hrGFP or ⌬13Met) along with packaging vectors as described (26).
Immunoprecipitation and Western Blotting-Cells were lysed in radioimmune precipitation assay buffer (50 mM Tris-HCl, pH 7.4, 1% Nonidet P-40, 0.1% SDS, 150 mM NaCl, 5 mM EDTA) containing freshly added protease inhibitor mixture solution and/or phosphatase inhibitors (1 mM sodium fluoride and 1 mM sodium orthovanadate). Cell lysates (1 mg) were incubated with 1 g of anti-Met antibody (C-28 or DO-24) for 16 h at 4°C with gentle rotation. Then 20 l of 50% slurry of rProtein G-agarose (Invitrogen) was added into each lysate, incubated for an additional hour, and centrifuged at 5,000 rpm for 3 min to collect immune complexes. The immunoprecipitates were washed three times with the lysis buffer and analyzed by immunoblotting. The protein expression was detected with ECL detection kit after incubation with HRP-conjugated secondary antibodies (Amersham Biosciences).
Immunofluorescence-Cells grown on glass coverslips were fixed by immersion in 4% paraformaldehyde for 15 min and permeabilized for 5 min in Tris-buffered saline containing 0.075% Triton X-100 (TBS-T). Nonspecific binding was blocked with 5% bovine serum albumin (BSA). Coverslips were then incubated with the primary antibody at room temperature in a humidified atmosphere. After three washes with PBS, the coverslips were incubated with Cy3-conjugated goat antimouse IgG antibody diluted in PBS containing 5% BSA, washed repeatedly with PBS, and mounted with Vectashield mounting medium containing DAPI to visualize the nuclei (Vector Laboratories). To detect Met or myosin heavy chain (MHC), we used DL-21 or MF-20 (Developmental Studies Hybridoma Bank, University of Iowa), respectively, and anti-desmin antibody was used to detect the myoblast-positive cells. Expression and localization of Met were observed under a confocal microscope (Olympus Fluoview FV300), and MHC was observed under a fluorescence microscope (Axio Imager; Carl Zeiss) using constant contrast and brightness conditions. Subcellular Fractionation-To obtain the cytoplasmic fraction, cell pellets were resuspended in cytoplasmic buffer (20 mM Tris, 250 mM sucrose, 10 mM EGTA, and 2 mM EDTA, pH 7.5), incubated for 3 h on ice after sonication, and then centrifuged at 100,000 ϫ g for 1 h at 4°C. The insoluble pellet was resuspended in Nonidet P-40 buffer (20 mM Tris, 1% Nonidet P-40, 1 mM EGTA, and 1 mM EDTA, pH 7.5), incubated for 15 min on ice, and then centrifuged at 50,000 ϫ g for 1 h at 4°C to obtain the membrane fraction. Each fraction was resolved on 8% SDS-PAGE and immunoblotted with DL-21 antibody to detect Met isoforms. PKC-␣ (Santa Cruz, CA) was used as a marker for cytosolic fraction.
HGF Binding Assay-⌬13Met-or control vector-transfected NIH3T3 cell pools were plated at 5 ϫ 10 6 cells in a 100-mm culture dish and cultured overnight. After washing three times with PBS, the cells were grown in 5 ml of DMEM, 0.02% FBS for another 24 h. Each conditioned medium (CM) was harvested, clarified by centrifugation, and concentrated 10-fold by using Centriplus centrifugal filter devices (Millipore Co., Bedford, MA). The concentrated conditioned media were incubated with 100 ng of rhHGF for 16 h at 4°C with rotating and then immunoprecipitated with anti-Met antibody (DO-24) or anti-HGF antibody. Western blot analysis was followed.
Cell Proliferation-Cells were seeded in 96-well plates and cultured overnight. After 24 h of serum starvation, the cells were incubated for the indicated time periods in the presence of HGF (20 units/ml). Cell numbers were counted after trypsinization using hemocytometer.
Scratch Wound Motility Assay-HaCaT cells were seeded (7 ϫ 10 5 cells/well) in 24-well plates in RPMI 1640 supplemented with 10% FBS and allowed to adhere overnight. The cells were then starved with 0.1% FBS for 24 h. The monolayered cells were then carefully scratched with a sterile 200-l pipette tip and incubated with the prepared conditioned media from ⌬13Metor control vector-transfected NIH3T3 cells for 48 h. After washing three times with cold PBS, the cells were fixed in acetone:methanol (1:1, v/v) for 15 min on ice and stained with 2% crystal violet. Cell migration was assessed by measuring the gap at three different sites and calculating the mean value.
Source of Human Skeletal Muscle-Skeletal muscle tissues were obtained from patients during surgical treatment for nonmuscle problems of larynx. This study was approved by the Institutional Review Board of Ajou University Hospital (AJIRB-CRO-05-087), and all patients gave written informed consent.
Primary Skeletal Muscle Cell Culture-Primary skeletal muscle cells were isolated from the tissues according to a modified method of Foulstone et al. (27). Briefly, biopsy samples were dissected into 1-mm 3 pieces and digested in TE buffer (0.05% trypsin and 0.02% EDTA in PBS) for 1 h with gentle mixing at 37°C. Then the reaction was stopped by adding FBS up to 10%, followed by filtration through 100-m cell strainer (Falcon; BD Biosciences) before centrifugation at 1000 rpm for 5 min. The cell suspensions were plated onto a 100-mm culture plate precoated with 0.2% gelatin, incubated, and passaged when 80 -90% confluent. Cells in 80% confluence were induced differentiation by washing twice in PBS and adding the differentiation media (minimum essential medium supplemented with 2% horse serum, 50 IU/ml penicillin, and 50 g/ml streptomycin). Isolated myoblasts were identified by immunostaining with anti-desmin antibody (Dako, Ely, UK). On average, primary cultures contained over 60% myoblasts, which did not change with passaging. All experiments were performed with cells between passages 2 and 5.
In Vivo Skeletal Muscle Regeneration-All experiments involving animals were guided by the Ajou University School of Medicine Institutional Animal Care and Use Committee. In C57BL/6 mice, tibialis anterior (TA) muscle was injured by the myotoxic agent notexin, as described previously (28). Briefly, mice were anesthetized with avertin (2,2,2-tribromoethanol; Sigma-Aldrich), and an adequate depth of anesthesia was maintained during the treatment. The right hind limb was shaved, and a small portion of the anterior aspect of the TA muscle was surgically exposed. Twenty l of notexin (10 g/ml in isotonic saline; Latoxan, Valence, France) was injected to the muscle with a 29-guage fixed needle, whereas the left leg was used as uninjected control. After intramuscular injection, the skin incision was closed with surgical sutures, and the mouse was allowed to recover in a warm place. Three days after notexin injection, the mouse was anesthetized, and the notexin-injected TA muscle was surgically exposed, and 300 ng of p24 of lentiviral stock was injected into the muscle. A group of mice was sacrificed on days 3, 6, 8, or 10 postinfection.
Immunohistochemical Analyses-At the time points indicated, TA muscles were harvested and fixed in 4% paraformaldehyde overnight, dehydrated, and embedded in paraffin. Sections for histological analysis (4 -6 m thick) were rehydrated and performed hematoxylin/eosin staining. The damaged area of transverse sections and cross-sectional areas of myofibers were measured using ImageJ version3.0 software and Carl Zeiss AxioVision version 4, respectively. The data are presented as the means Ϯ S.D. The p values were determined using two-way analysis of variance or two-tailed unpaired Student's t test with statistical significance as indicated.
Statistical Analysis-The data are presented as the means Ϯ S.D. The p values were determined using two-tailed unpaired Student's t test with statistical significance defined as p Ͻ 0.05.

A Novel Alternatively Spliced Form of Met Exists in Human
Tissues-The Met receptor tyrosine kinase in mouse tissues is present as two major isoforms, and the only difference between these two forms is the presence of a 47-amino acid segment in the juxtamembrane region that can be deleted by alternative splicing of exon 14 (30). Because this segment is responsible for down-regulating activated Met, the smaller sized Met isoform lacking this segment is highly active (31).
To evaluate the presence of this isoform, which is called Sm-Met in human tissues, we amplified the cDNA region from exon 11 through exon 16 from the BD TM human multiple tissue cDNA panel (Clontech) by using a specific primer set (Fig. 1A, top panel). In all tissues tested, an 806-bp band amplified from wild type c-met was observed, even though the expression levels were different among tissues under the same PCR conditions. Interestingly, an additional band was detected in the skeletal muscle sample below the 806-bp major band (Fig. 1A, middle panel). A similar band was observed in kidney, liver, pancreas, and placental tissues; however, the expression levels were much lower than that of skeletal muscle. To confirm the presence of the smaller band in tissues other than skeletal muscle, we amplified the same region from the liver cDNA by increasing the number of PCR cycles. As shown in Fig. 1A (bottom panel), two discrete bands were amplified after the PCR (first), and the reamplified bands from each band (second) clearly showed the presence of two different-sized cDNA fragments in liver tissue. To our surprise, sequencing analysis revealed that the smaller fragments from liver and skeletal muscle were not from exon 14 deletion. Instead, both liver and skeletal muscle lacked the 157-bp exon 13. The presence of this isoform was also confirmed in two different human primary skeletal muscle tissues by RT-PCR with other pairs of primers followed by sequencing (Fig. 1B).
Because the entire single transmembrane domain of Met is encoded by the second half of exon 13 (32), deletion of exon 13 in skeletal muscle or liver tissues is expected to eliminate the transmembrane domain of Met (Fig. 1C, top panel). Notably, exclusion of 157 bp should induce a frameshift, resulting in encoding five new amino acids followed by early termination (Fig. 1C, middle panel). Thus, the deduced protein of this variant is expected to contain only the extracellular domain and five additional amino acids (Ile-Trp-Ala-Val-Asp; Fig. 1C, bottom panel).
The Novel Isoform, ⌬13Met, Yields a Detectable Protein in Vivo-Because not all alternatively spliced variants of Met yield detectable protein products in vivo (32,33), we examined the expression of the encoded protein (hereafter designated as ⌬13Met) in human primary skeletal muscle tissue, where the mRNA was abundantly found (Fig. 1A, middle panel). Lysates of primary skeletal muscle tissue and human ⌬13Met-overexpressing NIH3T3 (⌬13Met-NIH3T3) cells were immunoprecipitated with the DO-24 anti-human Met antibody, and the precipitates were subjected to Western blotting with the DL-21 anti-human Met antibody (neither antibodies detect mouse Met). Because both antibodies target the extracellular portion of human Met, both wild type and ⌬13Met could be detected.
As shown in Fig. 1D (top panel), ␤-chains of ⌬13Met (ϳ100 kDa) and wild type Met (ϳ145 kDa) were recognized in primary human skeletal muscle tissue. To determine the specificity of the detection, the blot was deprobed and then incubated with DL-21 anti-Met antibody, which had previously been adsorbed with Met antigenic peptide. As shown in Fig. 1D (middle panel), the Met and ⌬13Met bands could no longer be detected, confirming the specificity. Next, the blot was reincubated with C-28 antibody raised against the C terminus of human Met (Fig.  1D, bottom panel). Although the band representing wild type Met was recovered, the presumed ⌬13Met bands did not reappear because of a lack of the Met C terminus. Thus, these results suggest that a considerable amount of ⌬13Met protein is expressed in human primary skeletal muscle tissue. The presence of ⌬13Met protein was also evident from subsequent analysis using primary human myoblasts (see Fig. 4D).
Membrane-localized ⌬13Met Is Shed into Culture Medium and Binds to HGF-To evaluate the localization of ⌬13Met in cells, NIH3T3 cells were transfected with human ⌬13Met or Met cDNA followed by immunostaining for human Met with DL-21 anti-Met antibody. Immunoreactivity of Met in ⌬13Met-transfected NIH3T3 was detected in both cytoplasm and plasma membrane as in Met-transfected cells, which suggests that ⌬13Met is localized in plasma membrane even though it lacks transmembrane and cytoplasmic domains ( Fig.  2A). To confirm the presence of ⌬13Met in plasma membrane, NIH3T3 cells were co-transfected with ⌬13Met and Met cDNA followed by fractionation cytoplasm and membrane parts. Although a relatively larger amount of ⌬13Met was observed in the cytoplasmic fraction, a small proportion was localized in the membrane fraction (Fig. 2B). It should be noted that the sizes of ⌬13Met in the two fractions were different. It is highly likely that ⌬13Met in the membrane represents a ␤-chain of the mature form, which is linked with ␣-chain in disulfide bond, whereas the cytoplasm contains only the premature form in which ␣and ␤-chains are not yet cleaved by proteases. Our data therefore suggest that exogenously expressed ⌬13Met is mainly present as an intracellularly localized precursor form, whereas a small amount is targeted to the plasma membrane as mature protein.
In view of the absence of the transmembrane domain, we examined whether ⌬13Met is released into the culture medium after targeting to the membrane. Thus, the CM of Met-or ⌬13Met-NIH3T3 cells was harvested, concentrated, and then subjected to Western blotting (Fig. 2C). Unexpectedly, a band of ϳ80 kDa that was smaller than ⌬13Met was detected in the CM from both Met-and ⌬13Met-NIH3T3 cells. This fragment represents a soluble, truncated form of Met that is released from the membrane-localized Met by proteolytic cleavage, i.e. ectodomain shedding, which is mediated by certain metalloproteases and occurs at or near the cell surface (34). Collectively, it is highly probable that ⌬13Met is not secreted as itself but somehow localizes at the cell surface like peripheral protein and is shed into the medium.
Because the shedding Met released from the Met has binding ability with HGF because of the presence of the SEMA domain necessary for HGF binding (35), we examined whether the shedding Met cleaved from ⌬13Met can also interact with HGF. The results of IP-Western blotting of the conditioned media preincubated with rhHGF revealed that HGF co-immunoprecipitated with the shedding Met, which originated from ⌬13Met and vice versa (Fig. 2D).
⌬13Met Attenuates HGF/Met Signaling-Upon binding with HGF, Met is dimerized and autophosphorylated on specific tyrosine residues, followed by transferring downstream signals including Ras-Raf-ERK (36). However, the presence of ⌬13Met, which has no cytosolic domains, probably interferes with this HGF/Met signaling pathway by sequestering HGF.
To address this possibility, ⌬13Met was transfected in NIH3T3 or human Met-overexpressed NIH3T3 cells followed by analyzing Met phosphorylation and ERK1/2 activation upon HGF stimulation by Western blotting. ⌬13Met overexpression clearly decreased tyrosine phosphorylation of Met in the presence of HGF, as well as in the absence of HGF in both cell lines (Fig. 3A). ⌬13Met also inhibited ERK1/2 phosphorylation by HGF in both cell lines, demonstrating that ⌬13Met effectively inhibits HGF-induced downstream signaling. To further evaluate whether the inhibitory effect of ⌬13Met on HGF/Met signaling pathway is dose-dependent, NIH3T3 cells were transfected with different concentrations of ⌬13Met cDNA followed by stimulation with HGF. As shown in Fig. 3B, different amounts of ⌬13Met led to a decrease in HGF-induced ERK1/2 phosphorylation in a concentration-dependent manner.
Because cell proliferation and migration are dependent on ERK1/2 signaling (37, 38), we also determined the inhibitory effect of ⌬13Met on HGF-induced NIH3T3 cell proliferation.
Overexpression of ⌬13Met decreased cell proliferation compared with mock transfected NIH3T3 cells (Fig. 3C). In addition, the CM from ⌬13Met-NIH3T3 cells prevented HGF-induced migration of HaCaT keratinocytes, whereas the CM from mock NIH3T3 cells did not affect cell migration (Fig. 3D), confirming the possible function of ⌬13Met in sequestering HGF as a soluble ⌬13Met. These findings collectively suggest that ⌬13Met is capable of blocking HGF-induced cell proliferation and migration via inhibition of HGF/Met signaling in a concentration-dependent manner.
⌬13Met Is Induced during Skeletal Muscle Differentiation-Because a comparable level of ⌬13Met was detected in skeletal muscle, it is possible that ⌬13Met plays a physiological role in this tissue. Initially, we analyzed de novo HGF-dependent Met activation in primary human myoblasts cultured in growth medium (F-10 containing 20% FBS). Different batches of primary myoblasts were grown with or without neutralizing anti-HGF antibody for 48 h followed by immunoprecipitation with anti-Met antibody. As shown in Fig. 4A, although the extents were different between preparations, Met autophosphorylation was abolished by incubating with neutralizing antibody, suggesting that Met is activated by endogenous HGF.
Next, we evaluated the expression levels of Met and HGF during the differentiation of cultured myoblasts. Differentiated myoblasts displayed a characteristic multinucleated morphology and MHC induction in a time-dependent manner during the differentiation processes (Fig. 4B). As expected, the expression levels of c-met and hgf mRNA were maximal in actively proliferating myoblasts at day 0, followed by an evident decrease in the c-met transcript thereafter, but not hgf mRNA (Fig. 4C). On the other hand, the ⌬13met transcript was barely detectable at day 0, but rapidly increased after the induction of differentiation as evidenced by myogenin expression, another marker for muscle differentiation. Consistent with RT-PCR data, ⌬13Met protein expression was clearly increased 2 days after the induction of differentiation and gradually decreased thereafter, whereas Met expression gradually decreased during the whole period (Fig. 4D). These data indicate that although ⌬13Met is an alternatively spliced form of Met, its expression is differently regulated from the Met expression during myoblast differentiation.
Suppression of ⌬13Met Decreases Skeletal Muscle Differentiation-Because ⌬13Met inhibits HGF/Met signaling and the expression of Met and ⌬13Met are differently regulated during myogenic differentiation, it is highly probable that the concomitant increase of ⌬13Met and decrease of Met expression are important for attenuating HGF/Met signaling, which is required for the transition to differentiation from proliferation of myoblasts.
To address this hypothesis, primary myoblasts were transfected with ⌬13Met-specific siRNA followed by inducing differentiation. ⌬13Met-specific siRNA was designed to target junctional sequences between exon 12 and exon 14 because the junction is the only region in ⌬13met different from the wild type c-met. Therefore, use of another siRNA(s) of independent sequences was not an option. Specificity of the siRNA was demonstrated by Western blotting that only the protein level of ⌬13Met but not Met was depleted by the siRNA in Met/ ⌬13Met-overexpressed NIH3T3 cells (Fig. 5A). When the primary myoblasts were induced for differentiation, ⌬13Met siRNA-transfected primary myoblasts expressed myogenin mRNA 1 day later than control myoblasts, and the mRNA level slightly decreased on days 3 and 4 compared with the levels in control cells (Fig. 5B). Indeed, multinucleated myotube formation was clearly reduced in ⌬13Met siRNA-transfected cells on differentiation day 3 compared with that of control siRNA-transfected cells (Fig. 5C). The analysis of the extent of differentiation by measuring the total myotubular area disclosed a decreased differentiation in ⌬13Met siRNA-transfected cells on days 3 and 4 (Fig. 5D). In addition, ϳ40% of MHC-positive cells from control siRNAtransfected cells contained more than two nuclei, whereas less than 30% of positive cells from ⌬13Met siRNA-transfected cells contained more than two nuclei, confirming the decreased differentiation capacity of ⌬13Met-depleted cells (Fig. 5E).
To evaluate whether Met depletion rescues the decreased differentiation of ⌬13Met-depleted primary myoblast, cells were co-transfected with ⌬13Metand Met-specific siRNA followed by inducing differentiation. The reduced level of MHCpositive cells in ⌬13Met-depleted cells was significantly overcome by depleting Met (Fig. 5F), suggesting that induction of ⌬13Met expression plays a role in the skeletal muscle differentiation process by inhibiting HGF/MET signaling.
Overexpression of ⌬13Met Enhances Differentiation in Primary Human Skeletal Muscle Cells-To confirm the role of ⌬13Met during muscle differentiation, we overexpressed ⌬13Met in primary myoblasts followed by inducing differentiation. Primary myoblasts were transduced with ⌬13Met-producing lentivirus or hrGFP-producing lentivirus, which was used as a control. Fig. 6A shows that the protein expression of ⌬13Met was increased in a dose-dependent manner in the infected cells.
When the differentiation was induced in both cultures, the number of MHC-positive cells began to be seen at differentiation day 1, and the myotube formation was evident from day 2 in both cultures (Fig. 6B). However, the myotubes in ⌬13Metoverexpressed myoblasts were longer and thinner than those in control cells, and the number of MHC-positive cells was signif-icantly higher than that of control cells on differentiation days 2 and 3 (Fig. 6C). When the myotubes of day 3 were classified by the number of nuclei within each myotube, the myotubes in ⌬13Met-overexpressed cells contained more nuclei than those in control (Fig. 6D). Specifically, over 50% of myotubes in ⌬13Met-overexpressed cells had six or more nuclei in a single myotube, whereas only ϳ10% of myotubes in control cells had six or more nuclei. Furthermore, the mRNA of myogenin was induced 1 day earlier in ⌬13Met-overexpressed cells than control cells (Fig. 6E). Because we used different batches of primary cells for the experiments shown in Figs. 4 -6, variations in the day of myogenin induction or ⌬13Met expression between different experiments were unavoidable. However, comparison between experimental groups in the same experimental setting was still possible because the same batch of primary cells was used for the specific experiment. were transfected with control siRNA or two different doses (50 or 100 nM) of ⌬13Met-specific siRNA using Lipofectamine 2000. Control siRNA was used as for control transfection. Two days after transfection, cells were harvested and subjected to Western blotting. B, reduced level of myogenin mRNA induction by ⌬13Met-specific siRNA transfection. Primary myoblasts were transfected with the indicated siRNA 1 day prior to differentiation, and the cells were harvested at the indicated times. RT-PCR analysis for myogenin was performed. Actin was used as the loading control. All samples were analyzed in triplicate, and the data are presented as means Ϯ S.D. C, representative MHC staining of differentiated myoblasts. One day before differentiation, primary myoblasts cultured on coverslips were transfected with the indicated siRNA (100 nM). After induction of differentiation, coverslips were subjected to immunostaining for MHC expression. Representative images on differentiation day 3 are shown. DAPI was used to stain nuclei. D, the extent of differentiation is represented by arbitrary units of MHC-stained area, measured by Carl Zeiss AxioVision 4. *, p Ͻ 0.05; **, p Ͻ 0.01 compared with control siRNA at the corresponding day. E, decrease in myotube formation by ⌬13Met-siRNA transfection. Myotubes showing MHC immunoreactivity with multinuclei were counted at day 3 in samples processed as for C. Myotubes with binuclei and multinuclei (n Ն 3) are separately counted. *, p Ͻ 0.05; **, p Ͻ 0.01 compared with con siRNA. F, primary myoblasts were co-transfected with ⌬13Met and Met siRNAs followed by inducing differentiation for 3 days. The cells were immunostained for MHC as described in C. MHC-positive cells were counted in samples, and the extent of differentiation is quantified and represented as percentages of MHC-positive cells among total cells. con, control; d, day.
Collectively, these results strongly suggest that the induction of ⌬13Met by alternative splicing significantly affects human skeletal muscle differentiation.
⌬13Met Accelerates Muscle Regeneration in Vivo-To address whether ⌬13Met is a positive regulator of muscle regeneration in vivo, we injected ⌬13Met-producing lentivirus into notexin-injured mouse TA muscle. We used a lentiviral vector encoding the murine version of ⌬13Met instead of the human form in this experiment because human Met poorly binds murine HGF, whereas murine Met binds both human and murine HGF (39). Additionally, ⌬13Met is hardly detected in mouse cells or tissues (our unpublished observation), so a knockdown approach was not an option to prove our hypothesis.
The inhibitory effects of the murine ⌬13Met on HGF/Met signaling were confirmed because the increase of ERK1/2 and AKT phosphorylation by HGF exposure was inhibited by the presence of murine ⌬13Met (Fig. 7A). Fluorescence from the hrGFP-expressing lentiviral vector control was observed as early as 1 day after the viral injection into TA muscle (Fig. 7B). Next, we evaluated the expression of endogenous Met in TA muscle by immunostaining transverse sections with anti-murine Met antibody (Fig. 7C). Positive immunoreactivities were detected near the nuclei localized outside muscle bundles, which are regarded as muscle satellite cells (arrows). In notexininjured TA muscle, the cross-sectional area of muscle bundle was reduced, and bundles were dissociated from other bundles. However, the positive immunoreactivity of Met was increased, further suggesting that muscle precursor cells are activated as early as 1 day after notexin exposure by increasing the myogenic cell numbers. The surface of TA muscle turned to be rather white and edematous on the third day after the notexin injection, and the damaged area was expanded up to 80% until day 8 in both groups (Fig. 7D, left panel). The white-colored damaged surface areas remained as 80% on day 10 after injury in control mice, whereas the color of the surface was significantly recovered in the ⌬13Met-injected mice, and the damaged surface area, measured by image analysis, was significantly diminished at day FIGURE 7. Enhancement of muscle regeneration by ⌬13Met overexpression. A, NIH3T3 cells expressing murine Met were subsequently transfected with empty vector or murine version of ⌬13Met. After serum depletion for 24 h, cells were treated with 20 units/ml of HGF for 10 min. Lysates from the samples were prepared and analyzed for ERK1/2 and AKT phosphorylation directly by Western blotting. B, representative fluorescence photomicrographs of hrGFP in cross-sections of TA muscle 1 day postintramuscular injection of hrGFP-expressing lentivirus. The muscles were surgically excised at the indicated day after injection, frozen, and sectioned transversely (8 m) through the mid-belly region on a cryostat microtome at Ϫ20°C. C, the cross-sections of TA muscle were immunostained with anti-murine Met antibody (green). DAPI was used to stain nuclei (blue). D, comparison of the surface of TA muscle between hrGFP and ⌬13Met-injected mice. Representative photomicrographs were taken at the indicated day postinjury (left panel). The estimated damaged area (white-colored area) was measured by AxioVision version 4 (right panel). The p values were determined using two-way analysis of variance. **, p Ͻ 0.01 compared with hrGFP-injected TA muscle. E, effect of ⌬13Met expression in TA muscle regeneration. Cross-sections of hrGFP or ⌬13Met-injected TA muscle, which was harvested at day 10 postinjury, were stained with hematoxylin/eosin (left panel). Scale bar, 200 m. The damaged myofibers stained dark purple were depicted as yellow dots using Image-Pro Plus 4.5 after photographed by digital camera (right panel, inset). The obtained results were quantified and represented in a form of a bar graph (right panel). *, p Ͻ 0.05 compared with hrGFP-injected muscle. F, representative photomicrographs of hematoxylin/eosin-stained transverse TA muscle section. Note the greater number of infiltrated lymphocytes between and on the muscle fibers (arrows) and the small, irregular shape of damaged muscle fibers in hrGFP-injected muscle (arrowheads). Scale bar, 20 m. G, comparison of the cross-sectional area of myofibers between hrGFP and ⌬13Metinjected muscle (upper panel). The distribution of the cross-sectional area of myofibers measured by AxioVision version 4 was plotted (lower panel). The data are means Ϯ S.D. *, p Ͻ 0.05; **, p Ͻ 0.01 compared with hrGFP-injected muscle. d, day; DIC, differential interference contrast. 10 (Fig. 7D, right panel; p Ͻ 0.01 compared with control). In support of the above observations, hematoxylin/eosin staining of transverse sections of TA muscles clearly revealed reduction of the damaged fibers stained dark purple in ⌬13Met-introduced TA muscle compared with controls (Fig. 7E, left panels). The damaged fibers were depicted as yellow dots that were quantified using Image-Pro Plus 4.5 (Fig. 7E, right panel, inset). The mean damaged area of ⌬13Met-injected mice was reduced by ϳ60% compared with the controls (Fig. 7E, right panel; p Ͻ 0.05). Moreover, the clearance of inflammation and the replacement of damaged myofibers by newly formed fibers, evidenced by the central localization of nuclei, were more clearly observed in the ⌬13Met-introduced TA muscle compared with the control mice (Fig. 7F). In addition, the median cross-sectional area of myofibers was significantly increased in the ⌬13Met-introduced group compared with the control group (Fig. 7G, upper panel; n ϭ 300, p Ͻ 0.05). A plot of the area of each myofiber as a function of frequency distributions showed a rightward shift in the ⌬13Met-introduced muscle, indicating an evident increase in the percentage of larger fibers (Fig. 7G,  lower panel).
Collectively, these data clearly show that ectopic expression of ⌬13Met enhances muscle regeneration in vivo. Although ⌬13Met is not expressed in murine skeletal muscle tissue, the above in vivo effect of expression of artificial murine ⌬13Met on muscle regeneration shows the regulatory effects of ⌬13Met in skeletal muscle differentiation and/or regeneration, as well as a possibility that the human form of ⌬13Met also enhances muscle regeneration in human skeletal muscle tissues.

DISCUSSION
Regarding muscle differentiation and/or regeneration, HGF is responsible for stimulation of satellite cell activation after muscle injury (15) and migration of muscle precursors into the limb buds of developing embryos (11,40). Although HGF was first shown to induce DNA synthesis in quiescent satellite cells, thereby driving them into the cell cycle, it also inhibits myogenesis in cultured myoblasts (18). When HGF was ectopically expressed in primary satellite cells, it suppressed the activation of muscle regulatory gene reporter constructs (MCK, MRF4, MEF2, and 4Rtk-CAT), as well as the gene expression of MyoD, myogenin, and MHC (22). These data imply that the effect of HGF on muscle regeneration probably depends on whether HGF is applied before or after the activation of satellite cells. Because HGF is normally present in the extracellular matrix surrounding muscle fibers and the activated satellite cells, a regulatory system to inhibit HGF/Met signaling at the right time is required to achieve successful myotube formation during muscle regeneration. It has been reported that the transcription of hgf and c-met genes is down-regulated when myoblasts stop proliferating and begin to differentiate, providing a down-regulation mechanism of HGF/Met signaling (15).
Important results of this study indicate that a novel Met variant is expressed in human skeletal muscle that is produced by alternative splicing. Exon 13 skipping in the mRNA induces a reading frameshift followed by an early termination, which produces C-terminally truncated Met (⌬13Met). The structure is very similar to previously reported Met recombinants, which had the extracellular domain of Met and showed HGF/Met signaling inhibition by interfering with HGF binding to Met and Met homodimerization (34,40). Met-934, one of the splicing variants of Met, is composed of the first 12 exons of c-met followed by an extension of the twelfth exon and a stop codon, implying a Met variant resembling ⌬13Met (41). The authors obtained this variant by clustering and assembling the information from dbEST (GenBank TM ) and the human genome. Similar to ⌬13Met, this form reduces Met activation-induced cell proliferation and survival. However, the authors used an artificial fusion protein with the Fc portion of human IgG1 instead of the endogenous form for the functional study. In addition, the endogenous expression of this truncated isoform was not documented. Interestingly, our data prove that both mRNA and protein of ⌬13Met are expressed in human skeletal muscle, and the protein is localized in the plasma membrane, which could be released by proteolytic cleavage. Moreover, the inhibitory effect of ⌬13Met in HGF/Met signaling pathways corresponds with results from other inhibitory Met recombinants in which ⌬13Met overexpression inhibits HGF-induced Met phosphorylation, ERK1/2 signaling, cell proliferation, and cell migration. Our data suggest that ⌬13Met expression caused by alternative splicing is a novel mechanism of down-regulating HGF/Met signaling.
Several elements are involved in the regulation of alternative splicing. One of these elements is direct or indirect interactions of cis-acting regulatory elements and trans-acting activators or repressors of splicing (41). Ron, another member of the scatter factor receptor family like Met, has an alternatively spliced isoform (⌬Ron) generated by the deletion of exon 11 (42). It has been reported that the deletion of exon 11 of Ron is controlled by an exonic splicing silencer and exonic splicing enhancer located in exon 12 and that SF2/ASF, a trans-acting serine-and arginine-rich splicing factor (SR proteins), binds directly to exonic splicing enhancer and enhances its splicing activity (43). We found several cis-acting elements around exons 12-14 of c-met using ESE Finder program. In addition, SR proteins displayed different expression patterns between proliferating and differentiating primary human skeletal muscle cells. 6 The possibility of the involvement of specific cis-acting elements and SR proteins in the regulation of ⌬13Met expression is presently under investigation.
The cellular localization of ⌬13Met is predicted to be extracellular including cell wall by PSORT, a computer program for the prediction of protein localization sites in cells. However, the presence of ⌬13Met was not identified in the conditioned medium, but rather a shedding Met released from ⌬13Met was identified, suggesting that ⌬13Met is localized on the plasma membrane where ectodomain shedding is occurred by metalloproteases (34). Our results of immunostaining and cellular fractionation also support the presence of ⌬13Met on the plasma membrane, suggesting possible dimerization between Met and ⌬13Met. Although we have addressed whether ⌬13Met forms a heterodimer with wild type Met by using cross-linking agents but failed to confirm its presence (data not shown), we suspect that the structural conformation of ⌬13Met on the plasma membrane might not be the same as that of Met that spans through the plasma membrane and that might make unfavorable conditions for heterodimerization.
The most important finding in the study is a novel mechanism for down-regulating HGF/Met signaling during muscle cell differentiation. We prove that ⌬13Met expression is upregulated during muscle differentiation (Fig. 4), and the abrogation of ⌬13Met expression with a specific siRNA results in a decrease of muscle differentiation (Fig. 5). Furthermore, ectopic expression of ⌬13Met significantly enhances muscle differentiation both in vitro and in vivo (Figs. 6 and 7). These data led us to our view that after the activation and proliferation of myoblasts in early phase of skeletal muscle regeneration and/or differentiation, increase of ⌬13Met expression relative to Met contributes to the down-regulation of HGF/Met signaling, which subsequently leads to the stimulation of terminal muscle cell differentiation.
To the best of our knowledge, this is the first report on the role of an alternatively spliced form of Met in human physiology. Furthermore, the activating role of ⌬13Met in skeletal muscle differentiation through its antagonistic activity against HGF may be valuable in the development of new strategies for enhancing muscle differentiation/regeneration in clinical situations.