Versican Processing by a Disintegrin-like and Metalloproteinase Domain with Thrombospondin-1 Repeats Proteinases-5 and -15 Facilitates Myoblast Fusion*

Background: Skeletal muscle fiber formation requires myoblast cell-cell membrane contact and fusion. Results: A versican-rich pericellular matrix surrounding myoblasts is proteolytically cleared by ADAMTS versicanases facilitating myoblast contact and fusion. Conclusion: Versican processing by ADAMTS versicanases contribute to muscle fiber formation. Significance: Targeting versican remodeling could enhance the regenerative capacity of muscle by improving muscle fiber fusion during regeneration. Skeletal muscle development and regeneration requires the fusion of myoblasts into multinucleated myotubes. Because the enzymatic proteolysis of a hyaluronan and versican-rich matrix by ADAMTS versicanases is required for developmental morphogenesis, we hypothesized that the clearance of versican may facilitate the fusion of myoblasts during myogenesis. Here, we used transgenic mice and an in vitro model of myoblast fusion, C2C12 cells, to determine a potential role for ADAMTS versicanases. Versican processing was observed during in vivo myogenesis at the time when myoblasts were fusing to form multinucleated myotubes. Relevant ADAMTS genes, chief among them Adamts5 and Adamts15, were expressed both in developing embryonic muscle and differentiating C2C12 cells. Reducing the levels of Adamts5 mRNA in vitro impaired myoblast fusion, which could be rescued with catalytically active but not the inactive forms of ADAMTS5 or ADAMTS15. The addition of inactive ADAMTS5, ADAMTS15, or full-length V1 versican effectively impaired myoblast fusion. Finally, the expansion of a hyaluronan and versican-rich matrix was observed upon reducing the levels of Adamts5 mRNA in myoblasts. These data indicate that these ADAMTS proteinases contribute to the formation of multinucleated myotubes such as is necessary for both skeletal muscle development and during regeneration, by remodeling a versican-rich pericellular matrix of myoblasts. Our study identifies a possible pathway to target for the improvement of myogenesis in a plethora of diseases including cancer cachexia, sarcopenia, and muscular dystrophy.


Skeletal muscle development and regeneration requires the fusion of myoblasts into multinucleated myotubes. Because the enzymatic proteolysis of a hyaluronan and versican-rich matrix by ADAMTS versicanases is required for developmental mor-
phogenesis, we hypothesized that the clearance of versican may facilitate the fusion of myoblasts during myogenesis. Here, we used transgenic mice and an in vitro model of myoblast fusion, C2C12 cells, to determine a potential role for ADAMTS versicanases. Versican processing was observed during in vivo myogenesis at the time when myoblasts were fusing to form multinucleated myotubes. Relevant ADAMTS genes, chief among them Adamts5 and Adamts15, were expressed both in developing embryonic muscle and differentiating C2C12 cells. Reducing the levels of Adamts5 mRNA in vitro impaired myoblast fusion, which could be rescued with catalytically active but not the inactive forms of ADAMTS5 or ADAMTS15. The addition of inactive ADAMTS5, ADAMTS15, or full-length V1 versican effectively impaired myoblast fusion. Finally, the expansion of a hyaluronan and versican-rich matrix was observed upon reducing the levels of Adamts5 mRNA in myoblasts. These data indicate that these ADAMTS proteinases contribute to the formation of multinucleated myotubes such as is necessary for both skeletal muscle development and during regeneration, by remodeling a versican-rich pericellular matrix of myoblasts.
Our study identifies a possible pathway to target for the improvement of myogenesis in a plethora of diseases including cancer cachexia, sarcopenia, and muscular dystrophy.
The extracellular matrix (ECM) 6 is a functionally critical component of the musculature, because it organizes and assimilates force transmission from contracting muscle fibers to the skeleton. It does so by stabilizing contracting muscle fibers via the membrane dystrophin-glycoprotein complex and by contributing the collagen-rich force transmitting matrix that fuses with tendons (1). Disorders of ECM associated with the dystrophin-glycoprotein complex are among the major causes of myopathies (2). What is not generally appreciated, and remains poorly understood is the potential role of ECM both as a barrier and as a contributor in myogenesis.
Myogenesis is a dynamic process in which proliferating myoblasts align and merge via membrane-membrane fusion to form multinucleated contractile muscle fibers (3,4). The efficient fusion of myoblasts into muscle fibers and their correct attachment to a highly specialized ECM surrounding those fibers are critical determinants of correct muscle formation, growth, regeneration, and contractile function (1). Thus, myoblast interactions with the ECM and their synthesis of ECM components are important for normal skeletal muscle development and regeneration (5) and as such the ECM needs to be coordinately remodeled during myoblast proliferation and fusion (6). Most embryonic cells and many adult cells have a surrounding matrix named the pericellular matrix, which is rich in hyaluronan (HA) and a HA-bound large proteoglycan, such as aggrecan in cartilage, and versican in most other connective tissue/mes-enchymal cells (7,8). This aggregating complex binds to cell surface receptors such as CD44, and contributes to formation of a hydrated pericellular matrix with the ability to exclude particulate matter. Functionally, such a matrix can be considered as a transitional matrix, a dynamic entity that is readily remodeled by new matrix synthesis and specific enzymes (9). Such enzymes include hyaluronidases and proteinases capable of clipping the versican core protein. Emerging evidence in the literature suggests that myoblasts may also express a pericellular matrix rich in versican (10,11). However, the role of versican and its remodeling is poorly defined, even though it may be an important driver in myogenesis and skeletal muscle regeneration.
Four splice variants of versican are known, and are derived from inclusion (or not) of alternative chondroitin sulfate-rich domains named GAG-␣ or GAG-␤. The isoforms V1 (containing GAG-␤) and V0 (GAG-␣ or GAG-␤) are the most abundant in non-neural tissues. Several among the 19 ADAMTS (a disintegrin-like and metalloproteinase domain with thrombospondin-1 repeats) proteinases were previously shown to attack the core protein of both versican isoforms at specific cleavage sites. In regard to V1 sequence enumeration, ADAMTS cleavage of the Glu 441 -Ala 442 peptide bond (12) within the GAG-␤ domain was shown to be crucial in a number of developmental contexts using ADAMTS knock-out mice (13,14). We hypothesized that during skeletal muscle development, one or more members of the subset of ADAMTS proteinases comprising ADAMTS1, -4, -5, -8, -9, -15, and -20 (13) could be involved in remodeling the versican-rich pericellular matrix to facilitate cell membrane contact and fusion. Therefore, the purpose of this study was to determine the significance of myoblast pericellular matrix and its clearance by ADAMTS proteinases during myotube formation. In particular, we focused on the potential role of ADAMTS5 in remodeling versican, because it is strongly expressed in developing skeletal muscle and cleaves both aggrecan and versican (15)(16)(17).
ADAMTS5 is a major target for arthritis therapy because cartilage aggrecan degradation is not seen in mice lacking ADAMTS5 following experimentally induced inflammatory arthritis and osteoarthritis (18,19). However, ADAMTS5 also has important functions in physiology, particularly during embryonic development. Thus a thorough understanding of its biological function(s) in skeletal muscle is paramount, given arthritis patients are also likely to present with muscle wasting due to the immobilizing nature of the disease. This current study exemplifies the importance of ADAMTS5 and the related versicanase, ADAMTS15, in the process of myoblast fusion leading to mature muscle fiber formation.

EXPERIMENTAL PROCEDURES
Mice and Tissues-Mouse experiments were carried out in accordance with National Health and Medical Research Council guidelines for the care and use of animals in research under the approval of the Animal Welfare Committee, Deakin University. Some experiments were done under an IACUC-approved protocol at the Cleveland Clinic. Six-week-old C57/Bl6J (Animal Resources Centre, WA, Australia) mice were mated after 1700 h and conception was confirmed before 0900 h the next morning by observation of vaginal mucous. The morning of conception was designated as gestational age 0.5 days postcoitus (E0.5). Proximal hind limbs of embryonic ages E12.5 through E14.5, or proximal hind limb muscles of older embryos (dissected away from osseous tissue) were collected and stored in TRIzol (Invitrogen, Mulgrave, Australia) at Ϫ80°C for mRNA analyses or fixed in 4% paraformaldehyde for immunostaining. Adamts5 Ϫ/Ϫ mice used in this study were from Jackson Laboratories and are as previously described (15,20).
Cell Culture-C2C12 myoblasts were maintained in growth medium (Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum (FBS) (Invitrogen)) in an atmosphere of 5% CO 2 at 37°C (HERAcell 150i, Thermo Scientific, Scoresby, Australia). For time course analysis of differentiation, cells were seeded at 20,000 cells/cm 2 and collected 24 (proliferation) and 48 h later (when confluent), after which differentiation medium (DMEM ϩ 2% horse serum) was added. For the first 48 h of differentiation, the medium contained 10 M Ara-C (Sigma) to stimulate cell cycle withdrawal. The time points were 6 h, 24 h, 48 h, 72 h, 4 days, and 6 days (differentiation). Media was changed every 24 h. Cells were harvested in triplicate wells, collected in TRIzol, and stored at Ϫ80°C. For siRNA transfection, C2C12 myoblasts were seeded into 12-well tissue culture plates at a density of 16,000 cells/cm 2 and grown to 60 -70% confluence. Cells were transfected with 25 nM Stealth RNAi targeting Adamts5 (Adamts5-1, 5Ј-gaucuucccgcauccugcaugucuauagacaugcaggaugcgggaagauc-3Ј; Adamts5-2, 5Ј-caggcuguacagucaugcaagcauuaaugcuugcaugacuguacagccug-3Ј) or control siRNA oligonucleotides (medium GC content, Invitrogen) diluted in Opti-MEM (Invitrogen) using Lipofectamine RNAi MAX (Invitrogen) according to the manufacturer's instructions. After 5 h, the transfection media were replaced with growth media. Differentiation was induced the following day when the cells reached ϳ90% confluence by changing to differentiation media. For the first 48 h of differentiation, 10 M Ara-C was included to stimulate cell cycle withdrawal. The differentiation medium was changed every 24 h. For the rescue experiments, conditioned medium from HEK293T cells transfected with cDNA expression constructs (see below) was used at a dilution of 1:4 in differentiation media from 24 h differentiation until the experimental end-point (72 h differentiation) when the fusion index was calculated.
Expression Constructs and Transfection of HEK293T Cells-HEK293T cells (ATCC, Manassas, VA) were grown in DMEM containing 10% FBS in an atmosphere of 5% CO 2 at 37°C. Cells were transfected using Lipofectamine 2000 (Invitrogen) with constructs encoding ADAMTS5 (wild type (WT) and E 411 A (16)), Adamts15 (WT and Glu 343 Ala), V1 versican construct (kindly provided by Professor Dieter Zimmermann), and empty vector control (pcDNA3.1MycHisAϩ (Invitrogen)). Serumfree conditioned medium was collected and cells were harvested in ice-cold PBS with a cell scraper from which cell lysate was prepared as previously described (16, 20 -22). Sterile conditioned medium was used for the rescue experiments.
RNA Extraction, Reverse Transcription, and Quantitative RT-PCR-RNA was extracted as per the manufacturer's protocol using TRIzol and 1 g of total RNA was reverse-transcribed with the iScript cDNA synthesis kit (Bio-Rad). Quantitative RT-PCR was performed on the cDNA using iQ SYBR Green Supermix (Bio-Rad) and oligonucleotide primers for the genes of interest (20) (additional primer sequences available upon request). Quant-iT OliGreen ssDNA Assay Kit (Invitrogen) was used to quantitate total cDNA input as per the manufacturer's instructions. Relative changes in mRNA levels to proliferating myoblasts were calculated using the ⌬C t method.
Immunocytochemistry and Immunohistochemistry-C2C12 cells were seeded into LabTek Permanox chamber slides (Electron Microscopy Sciences, Hatfield, PA) in growth medium and differentiated as described above. Cells were fixed in 4% paraformaldehyde/PBS for 5 min at room temperature before proceeding with immunostaining as previously described (20). Immunohistochemistry for V0/V1 versican (GAG-␤), cleaved versican (DPEAAE), and ADAMTS5 was performed as previously described (20), whereas co-localization of desmin included the addition of mouse monoclonal anti-desmin (Abcam, Cambridge, United Kingdom) with either rabbit anti-ADAMTS5 or anti-DPEAAE antibodies. The secondary antibodies used were FITC goat anti-rabbit IgG and Texas Red goat anti-mouse IgG (Invitrogen).
Quantitation of Fusion Index-For each experimental treatment, 15 digital images were captured with either an IX81 or IX71 Olympus inverted fluorescence microscope and a CC12 or an XM10 camera (Olympus). A well formed myotube was identified and an image of the founder myotube was captured, followed by two random images in the immediate vicinity (within 2 fields of view). This process was repeated for a total of five times, yielding 15 images per experimental group per experiment. To determine the fusion index, nuclei were counted and designated as either fused or unfused, using the Count tool in Adobe Photoshop CS5 Extended (Version 12.0) software. Fusion index was calculated as the percentage of fused nuclei as a proportion of total nuclei. Myoblasts were considered to be unfused if the cell contained only one or two nuclei; myotubes with three nuclei per cell were considered to be nascent myotubes for the purpose of calculating fusion index. Three independent experiments, performed in duplicate, were used for the determination of fusion index and ϳ2,000 -2500 nuclei were counted per experimental condition, per experiment.
To assess the effect of Adamt5 siRNA treatment on nascent myotube formation and maturation, the total number of myotubes per field of view was counted whereby myotubes containing 2-4 nuclei were designated as nascent and 5-9 nuclei or 10ϩ nuclei were designated as mature (24 -26). Approximately 520 -570 myotubes were counted per control or siRNA-treated groups across 6 independent experiments each performed in duplicate.
Creatine Kinase Activity Assay and Measurement of Total TGF-␤-Cells were lysed with 150 l of lysis buffer containing 40 mM MES buffer, 50 mM Trizma (Tris base), 1% Triton X-100, 1ϫ protease inhibitor mixture (Roche Applied Science). Insoluble material was collected by centrifugation and the supernatant was used for analysis of creatine kinase activity. Three technical replicates for each sample were performed in a 96-well plate using creatine kinase-NAC (Thermo Scientific). Creatine kinase activity was measured by the change in absorbance at 340 nm over 3 min (20-s intervals) at 37°C. Units per liter (units/liter) was calculated using the following formula: where TV is the total reaction volume (0.210 ml), A is the millimolar absorption coefficient of NADH at 340 nM (6.3), SV is the sample volume (0.010 ml), and P is the unique path length of light in the 96-well plate (determined to be 0.533 cm by spectrophotometric comparison of absorbance with known path lengths). Total levels of TGF-␤ were measured in 40-l aliquots of media collected from C2C12 cell cultures by an ELISA (R&D Systems, Minneapolis, MN) as per the manufacturer's instructions.
Cell Viability Counts-After 24 h differentiation, dead cells were collected and adherent cells were trypsinized and collected into the same FACS tube. The cells were resuspended in 100 l of 0.2% BSA/PBS containing 7-aminoactinomycin D. Cells were washed in PBS and fixed in 1% formaldehyde containing 2% FBS and 2 g/ml of actinomycin. The percentage of viable cells was determined by flow cytometric analysis using CELLQuest software. Cell number and viability was also determined using trypan blue (Sigma) exclusion staining and manual cell counts on a hemocytometer.
Particle Exclusion Assay and Quantitation of Pericellular Matrix Area-The assay was done essentially as previously described by Hattori et al. (27). C2C12 myoblasts transfected with Adamts5 siRNA were trypsinized and re-plated at a density of ϳ2,000 cells/well into 6-well plates 24 h after siRNA transfection. Cells were cultured for another 24 h and stained with 1 M calcein AM (Sigma). The medium was changed after 30 min and a red blood cell (RBC) suspension in serum-free DMEM at a density of 1 ϫ 10 8 erythrocytes/ml was added. After RBCs had settled to the bottom, images were taken with an Olympus IX71 inverted microscope (FITC channel for calcein-labeled cells and bright field for RBCs). As a control, hyaluronidase digestion was done using Streptomyces hyalurolyticus hyaluronidase (Sigma) prior to the particle exclusion assay. Resultant images of controls or Adamts5-1 siRNA treatments were analyzed using custom, semiautomated scripts generated in Image-Pro version 6.2 (Media Cybernetics, Silver Springs, MD) as previously described (27). In addition, scoring of the RBC exclusion area was performed by 3 investigators blinded to the experimental groups using a scale of 0 -3, where 0 ϭ no matrix accumulation and 3 ϭ maximal matrix accumulation. Approximately 20 images per experimental group were randomized and coded before the investigators scored them. Scores were whole number values only from 0 to 3, i.e. either 0, 1, 2, or 3.
Statistics-For quantitative PCR analyses, one-way analysis of variance followed by a Tukey's post hoc analysis were performed. All other analyses were performed using a Student paired, two-tailed t test between control and test groups. Gaussian distribution was assumed in all cases. Data were con-sidered statistically significant when a p value Ͻ 0.05 was obtained.

Versican Processing Is Absent in Postnatal Skeletal Muscle of Adamts5
Ϫ/Ϫ Mice-ADAMTS5 shows strong versicanase activity compared with other proteoglycan-degrading ADAMTS members (14); however, Adamts5 Ϫ/Ϫ mice do not display an overt myopathy (15,19). In both wild type and Adamts5 Ϫ/Ϫ embryonic hindlimb muscles versican processing was readily detectable by immunostaining with the neo-epitope antibody anti-DPEAAE that specifically detects the ADAMTS cleavage site in the V0 and V1 splice variants of versican (16,20,28,29) (Fig. 1A, top panels). These data suggest that versican processing is unimpaired in Adamts5 Ϫ/Ϫ mice during myogenesis, potentially as a result of participation by other proteinases. In contrast to wild type muscle, juvenile and mature skeletal muscle from 10 to 13 days and 3 weeks postnatal Adamts5 Ϫ/Ϫ mice had barely detectable versican cleavage when analyzed by FIGURE 1. Versican is proteolytically cleaved in embryonic and juvenile skeletal muscle. A, versican cleavage using anti-DPEAAE immunofluorescence (green signal, nuclei stained blue with DAPI) is readily detectable (arrows) in developing skeletal muscle of wild type and Adamts5 Ϫ/Ϫ mice at embryonic age (E) 13.5 days (top panels). It remains visible in 3-week-old wild type mice but not Adamts5 Ϫ/Ϫ mice (lower panels). Scale bars ϭ 100 m. The data are representative of embryonic hind limbs and 3-week-old skeletal muscle from 3 separate mice for each genotype. B, Western blot using anti-DPEAAE showing that versican cleavage at position E 441 A (V1) is detectable in postnatal skeletal muscle (black arrows) of wild type mice but absent in Adamts5 Ϫ/Ϫ mice. The Western blot shows two 10-day-old wild type and Adamts5 Ϫ/Ϫ mice skeletal hind limb muscle lysates, respectively (left-hand panel), and one 3-week-old wild type and Adamts5 Ϫ/Ϫ mice skeletal hind limb muscle lysates, respectively (right-hand panel). C, Adamts5 Ϫ/Ϫ mice have a greater number of centrally located nuclei than wild type mice in postnatal skeletal muscle (3 and 7 weeks old). D, ADAMTS5 co-localized with desmin in mature skeletal muscle fibers (arrowhead) and satellite cells (arrows). Scale bars ϭ 50 m. E, versican cleavage (DPEAAE neo-epitope) co-localized with desmin in mature skeletal muscle fibers (arrowhead) and satellite cells (arrows). Scale bars ϭ 50 m. The data for D and E represent muscle sections from 2 independent mice and these observations are representative of wild type mice (n ϭ 4). immunohistochemistry (Fig. 1A, bottom panels) or immunoblotting (Fig. 1B). This suggests that ADAMTS5 has a substantial role in versican processing in postnatal skeletal muscle growth and maturation. In agreement with this, higher levels of centrally positioned nuclei were seen in 3-and 7-week-old Adamts5 Ϫ/Ϫ mice compared with wild type mice (Fig. 1C). The presence of central nuclei suggests the absence of ADAMTS5 delayed maturation during postnatal muscle growth (30). Adult wild type skeletal muscle showed co-localization of ADAMTS5 (Fig. 1D) and cleaved versican (Fig. 1E) with desmin around mature muscle fibers and satellite cells, thus associating versican processing with functional skeletal muscle.
To elucidate the function for the ADAMTS proteolysis of versican during skeletal muscle development, we used a well characterized in vitro cell model of myoblast fusion, the C2C12 cell line (31). These cells remain undifferentiated in rich serum, but spontaneously undergo differentiation along a well characterized pathway when grown in a nutritionally poor medium containing horse serum. Quantitative RT-PCR showed that Adamts1, Adamts4, Adamts5, and Adamts15 mRNAs were dynamically up-regulated soon after induction of myoblast differentiation (Fig. 3, A-D), whereas Vcan mRNA was dramatically up-regulated upon myoblasts reaching confluence and contact inhibition, remaining high throughout the differentiation process (Fig. 3E). Pcsk6 mRNA was also dramatically upregulated at 72 h after onset of differentiation when myotube fusion is occurring (Fig. 3F). Maximal levels of myogenin upregulation were observed at 24 h with commencement of myotube formation (Fig. 4AЈ). RT-PCR showed that the versican V1 isoform was the major splice variant expressed both in E13.5 skeletal muscle and C2C12 myoblasts (Fig. 4B).
Next, we evaluated C2C12 cell expression of versican, cleaved versican, and ADAMTS5 during differentiation by confocal microscopy, using anti-versican GAG-␤, anti-DPEAAE, and anti-ADAMTS5 antibodies (20). Pericellular versican accumulation was detected as early as 6 h after onset of differ- entiation when the myoblasts appeared nestled within a versican-rich matrix alongside detectable versican cleavage (Fig. 4C, top panels). After 72 h differentiation, both ADAMTS5 and DPEAAE were localized at the ends of fusing myoblasts (Fig.  4C, bottom panels), suggesting a role for versican clearance, which could facilitate cell-cell contact and membrane fusion during myotube formation.
Catalytically Active ADAMTS5 and ADAMTS15 Proteinases Facilitate Myoblast Fusion-Given these parallels between the gene expression profiles and versican processing in embryonic muscle and in vitro myogenesis, we investigated the role of ADAMTS5 in the C2C12 cell line model. Adamts5 was targeted using two independent siRNA sequences each of which resulted in a 60 -75% reduction in Adamts5 mRNA levels (Fig.  5A) with a modest decrease in intracellular ADAMTS5 (Fig.  5B). Upon silencing Adamts5 mRNA, myogenin expression was reduced (Fig. 5C) and the myotubes formed were smaller and contained fewer nuclei (Fig. 5D) with a significant reduction in the fusion index (Fig. 5E), and creatine kinase activity (Fig. 5F). Of the two siRNA used, Adamts5-1 siRNA, showed the stronger reduction in Adamts5 mRNA levels. In separate experiments, manual cell counts and 7-aminoactinomycin D staining showed Adamts5 siRNA treatment did not reduce cell viability (data not shown). Further assessment of myotube formation and maturity showed significantly less mature myotubes (5ϩ nuclei (24 -26)) formed following Adamts5 siRNA treatment (Fig. 5G) alongside less total myotubes formed, suggestive of an initial impairment of myoblast fusion into nascent myotubes (Fig. 5G).
To validate the observed effects as being a consequence of specific siRNA targeting of Adamts5, and to determine whether ADAMTS proteolytic activity could indeed facilitate myotube fusion, we attempted to reverse the effects of siRNA using serum-free conditioned media containing either catalytically active or inactive (E 411 A active site mutant) human ADAMTS5 (16) or either catalytically active or inactive (Glu 343 Ala activesite mutant) mouse ADAMTS15 (data not shown). 7   both these ADAMTS versicanases effectively restored the fusion defect back to control levels (Fig. 6, A and B), indicating that ADAMTS versicanase catalytic activity is required for myotube fusion and suggesting that ADAMTS15, was as effective as ADAMTS5. Indeed, ADAMTS15 is now known to be a versican-degrading proteinase 7 and may compensate where ADAMTS5 activity is reduced. When we added ADAMTS5 and ADAMTS15 conditioned medium to differentiating myoblasts without prior Adamts5 siRNA treatment, we observed that neither catalytically active proteinase enhanced myotube fusion above the control (data not shown). In contrast, both catalytically inactive ADAMTS5 and ADAMTS15 impaired fusion by ϳ20 (p Ͻ 0.05) and ϳ30% (p Ͻ 0.01), respectively, suggestive of a dominant-negative effect, such as could occur by binding to versican or other substrates involved in myogenesis and preventing cleavage by ambient ADAMTS proteinases. In separate experiments, full-length V1 versican containing conditioned medium from HEK293T cells was added to the C2C12 media, leading to a ϳ70% reduction in fusion compared with empty vector control conditioned medium (p Ͻ 0.001, data not shown).
Reduced Adamts5 mRNA Results in the Accumulation of Pericellular Matrix Around C2C12 Myoblasts-Using the red blood cell exclusion assay (27), a significant increase in pericellular matrix was observed after treatment with Adamts5 targeted siRNA, using either computer imaging (Fig. 7, A and B,  Adamts5-1 siRNA versus control) or blinded visual analysis,  A, significant knockdown of Adamts5 mRNA levels upon siRNA treatments are seen at 72 h after differentiation. siRNA Adamts5-1 led to greater mRNA silencing. B, a modest decrease in intracellular ADAMTS5 protein levels is observed upon siRNA treatments. Pos. Cont. ϭ positive control (ADAMTS5 transfected HEK293T cells, which migrates at a higher apparent molecular weight due to the presence of a Myc-His 6 tag). C, reduced myogenin mRNA is observed in the Adamts5 siRNA-treated groups indicative of reduced differentiation. D, representative images of C2C12 myotubes at 72 h differentiation stained with desmin (green) and TO-PROா-3 (red). Control (scrambled G-C matched siRNA) treated myotubes shows efficient fusion (arrows). Two siRNAs targeted to separate regions of Adamts5 mRNA (Adamts5-1 or Adamts5-2) reduce myotube fusion and the myotubes formed in those siRNA groups are shorter and narrower (arrowheads). Scale bar ϭ 100 m. E, the calculated fusion index of n ϭ 3 biological replicates performed in duplicate, whose images are represented in A. Significantly less fusion is observed in the Adamts5 siRNA-treated groups. F, significantly less creatine kinase (CK) is present in the medium of the Adamts5 siRNA-treated groups. G, significantly fewer mature myotubes (5ϩ nuclei) and total myotubes are formed upon siRNA treatments. The data are representative of 6 biological replicates performed in duplicate. In all cases: *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001; error bars ϭ S.E. which included both Adamts5-1 and Adamts5-2 siRNA groups (data not shown). To ensure the pericellular matrix visualized was hyaluronan-based (and thus versican containing), Streptomyces hyaluronidase was added to siRNA-treated C2C12 myoblasts, which eliminated red blood cell exclusion (Fig. 7A, far  right panel).
Because it is likely that hyaluronidase activity is also implicated in the clearance of the hyaluronan and versican-rich pericellular matrix surrounding myoblasts, we analyzed the expression of Hyal2 across C2C12 differentiation but only modest changes in mRNA expression were observed (Fig. 7C). Thus, we conclude that the clearance of a versican-rich pericellular matrix by ADAMTS versicanases contribute to remodeling the pericellular matrix to enable myoblast fusion (Fig. 8).

DISCUSSION
The presence of a hyaluronate-rich pericellular matrix surrounding myoblasts derived from skeletal muscle was first described by Orkin et al. (32). In cardiac muscle, the disappearance of a hyaluronan-rich pericellular matrix coincided with myoblast fusion and altered levels of hyaluronan were also documented during myogenesis by Raganathan and Datta (33). Changes in pericellular matrix chondroitin sulfate proteoglycans during myogenesis were subsequently noted by Carrino et al. (34) in developing skeletal muscles of chick embryos; however, the mechanisms and functional consequence of pericellu-  lar matrix remodeling in myoblasts and during myoblast fusion has thus far been poorly characterized. We previously noted the expression of Adamts5 and its co-localization with myosin heavy chain in embryonic mouse hind limbs at a time when myoblasts initiate fusion into multinucleated tubes (15), which was supported by another study (35). ADAMTS versicanases including ADAMTS5 are localized to the cell surface or within the pericellular matrix of cell types such as chondrocytes (36,37) and it was therefore reasonable to hypothesize they might direct their activity toward versican in myoblasts.
The dynamic regulation of Adamts versicanases during in vivo and in vitro myotube formation suggested synergistic roles in processing of versican and other potential substrates during this process. The presence of ADAMTS5 in the pericellular matrix of C2C12 myoblasts, and its co-localization with cleaved versican (DPEAAE neoepitope) suggested that versican processing by this proteinase could facilitate myoblast fusion. Versican expression was recently shown to regulate proliferation and differentiation of avian myoblasts in vitro (38), whereas our data shows the specific relevance of versican proteolysis by ADAMTS versicanases during myoblast fusion to form nascent myotubes and a further requirement in facilitating the process of myotube maturation.
Recent studies identified a role for pericellular matrix clearance in isolated dermal fibroblasts from Adamts5 Ϫ/Ϫ mice (27,39) concomitant with increased or altered TGF-␤ signaling. In partial agreement with those data, we found the pericellular matrix was expanded upon silencing Adamts5 in the in vitro cell model of myogenesis; however, total TGF-␤ levels were unaffected by reducing Adamts5 levels in this process (data not shown). This observation allowed us to conclude that the accumulation of extracellular matrix was not a direct effect of changes in TGF-␤ levels, but rather is due to reduced processing by ADAMTS versicanases.
An endogenous inhibitor of ADAMTS5 and the related versicanase ADAMTS4 is TIMP-3 (40), which was previously reported as a regulator of myogenesis in C2C12 cells (41); although the overexpression of TIMP-3 inhibited myotube formation, a concordant reduction in TNF-␣ levels were observed suggesting an ADAM17-mediated effect. Furthermore, the overexpression of TIMP-3 alongside the addition of recombinant TNF-␣ effectively restored myotube formation (41). Thus, although TIMP-3 clearly has the capacity to inhibit several MMPs, ADAMs, and ADAMTS, some of which are important versicanases, the main action of TIMP-3 is to inhibit ADAM17 processing of TNF-␣ during myogenesis.
The Adamts5 Ϫ/Ϫ mouse presents with no overt myopathy or deficit in gait and mobility, likely due to compensation by other ADAMTS proteinases processing versican, although we did observe higher levels of centrally positioned nuclei in myofibers during postnatal growth. Initial postnatal muscle growth up to ϳ3 weeks of age is believed to be dependent upon satellite cell activation to form myoblasts, and their proliferation and fusion with existing myofibers, as there is little change in total myofiber number (42). Interestingly, we also showed that ADAMTS5 and cleaved versican localized to mature functional skeletal muscle fibers and satellite cells, indicating a possible role for ADAMTS versicanases during skeletal muscle homeostasis and, potentially, satellite cell activation and skeletal muscle growth.
The proteolysis of versican by ADAMTS proteinases is implicated in morphogenesis of the myocardium, heart valves, secondary palate, urogenital system (43)(44)(45), and interdigital web regression (20) during mouse embryogenesis, as well as ovulation in female mice (28,46). ADAMTS proteinases act alone or cooperatively in some of these contexts, such as web regression and palate closure. Despite combinatorial deletion of five alleles of three Adamts genes in mice (Adamst5, Adamts20, and Adamts9), skeletal muscle dysfunction was not elucidated (20). Adamts4 and Adamts5 combinatorial knockout mice have been previously reported as having no overt skeletal muscle myopathy (20,47) further suggesting a compensating role for Adamts15 during skeletal muscle development.
Skeletal muscle regeneration occurs after mechanical stress such as intense physical activity or trauma. In healthy skeletal muscle the regenerative process is usually relatively efficient. However, aging or muscle dystrophy can increase susceptibility to damage and impair regeneration. In a key feature of the regenerative process, myogenin-expressing myoblasts exit the cell cycle, align with other myoblasts, and fuse to form multinucleated myotubes, which ultimately fuse to existing muscle fibers (48). The current work demonstrated that suppressing ADAMTS5 or ADAMTS15 expression in myoblasts in vitro resulted in the impairment of myoblast fusion and subsequent myotube formation. Thus, inhibiting ADAMTS5 in the case of treatment for arthritis could be potentially detrimental in patients with compromised skeletal muscle. In addition, our data showed that catalytically inactive ADAMTS5, which nevertheless, retains substrate-binding ability impaired myoblast fusion. In agreement with this, addition of full-length V1 versican inhibited myoblast fusion, and pericellular matrix expanded upon silencing Adamts5.
Collectively, these data strongly suggest that ADAMTS versicanase activity mediates clearance of the pericellular matrix, which facilitates myotube formation, and that excess versican can impair this process. Our data support the notion that remodeling the transitional matrix is important for proper myotube formation. Thus, this work identifies a novel functional context for ADAMTS proteolytic activity during skeletal muscle development, and potentially, regeneration.