Collagen XXV promotes myoblast fusion during myogenic differentiation and muscle formation

Fusion of myoblasts into multinucleated myofibers is crucial for skeletal muscle development and regeneration. However, the mechanisms controlling this process remain to be determined. Here we identified the involvement of a new extracellular matrix protein in myoblast fusion. Collagen XXV is a transmembrane-type collagen highly transcribed during early myogenesis when primary myofibers form. Limb muscles of E12.5 and E14.5 Col25a1−/− embryos show a clear defect in the formation of multinucleated myofibers. In cell culture, the cleaved soluble extracellular domain of the collagen XXV is sufficient to promote the formation of highly multinucleated myofibers. Col25a1 is transiently expressed during myogenic differentiation and Col25a1 transcripts are down-regulated in multinucleated myofibers by a muscle-specific microRNA, miR-499. Altogether, these findings indicate that collagen XXV is required in vivo and in vitro for the fusion of myoblasts into myofibers and give further evidence that microRNAs participate to the regulation of this process.


Results
Collagen XXV is required for muscle formation in vivo. Previous studies indicated that Col25a1 is highly transcribed in developing skeletal muscle cells in limbs from E11.5 mouse embryos and not detected in adult skeletal muscles 7,10 . To examine the early in vivo function of collagen XXV, we analyzed the muscle phenotype of limb muscles from Col25a1 knock out mice at E12.5 and E14.5 (Fig. 1). Histological analysis of transverse cross-sections of hindlimb muscles revealed the presence of myosin positive muscle cells in E14.5 Col25a1−/− mouse embryos (Fig. 1A). Their number is similar in Col25a1−/− and wild-type (WT) E14.5 embryos (Fig. 1B). However, muscle fibers of Col25a1−/− embryos are larger than controls and irregular in shape, whereas WT myofibers are linear (Fig. 1C). At this embryonic stage, the proportion of myofibers with a cross-sectional area above 80 µm 2 is much higher in the Col25a1−/− forelimb muscles than in the same muscle masses in the control embryos. While multinucleated myofibers were present in forelimbs from WT E14.5 embryos, many of the myosin positive longitudinal myofibers observed in the Col25a1 deficient forelimbs were much shorter, with few myonuclei (Fig. 1D). Some of them have bulges, which could explain their larger diameter compared to controls (Fig. 1C,D). Analysis of forelimbs from E12.5 WT and Col25a1−/− embryos strongly suggests that collagen XXV is required for the formation of multinucleated myofibers during primary myogenesis (Fig. 1E). Multinucleated myofibers with sarcomeres were detected in the control E12.5 embryos, whereas most of the Col25a1−/− myofibers seem to properly align but were much shorter. This suggests that collagen XXV expression is important for myoblasts to properly fuse, although we did detect multinucleated myofibers at later embryonic stages. As WT myofibers, Col25a1−/− fibers had sarcomeres and expressed embryonic MyHC. The diameter of these poorly nucleated myofibers seem to be larger, as already observed in forelimbs from E14.5 embryos (Fig. 1C-E). These in vivo data reveal that collagen XXV is required for muscle formation during primary myogenesis.
Col25a1 is transiently expressed during myogenic differentiation. To better characterize the role of collagen XXV during primary myogenesis, as well as in myogenic differentiation, we analyzed the precise timing of Col25a1 expression during mouse development and in cell culture. By RT-qPCR, we confirmed that Col25a1 transcripts were abundant in limbs from WT E12.5 embryos and decreased when skeletal muscles mature, as measured by the abundance of Myh7 transcripts, coding for the adult slow isoform of myosin heavy chain (type I MyHC) ( Fig. 2A) 7 . Next, we analyzed the precise timing of Col25a1 induction upon myogenic differentiation (Fig. 2B). Primary myoblasts were isolated from P10 neonatal WT mice and induced to differentiate. RT-qPCR analyses revealed a massive and transient induction of Col25a1 transcripts at two days of differentiation, a time point which corresponds with the peak of expression of the differentiation marker, myogenin, and the initiation of transcription of the embryonic isoform of MyHC, encoded by the Myh3 gene, specifically expressed in young myotubes. These results should be confirmed at the protein levels. Unfortunately, we were not able to detect the endogenous protein, since all antibodies tested were not specific enough. Altogether, our transcriptional expression results of col25a1 during myogenic differentiation strongly suggest that collagen XXV is involved in the formation of myotubes.
Collagen XXV promotes fusion of myoblasts. Once synthesized, collagen XXV is targeted at the plasma membrane where the extracellular C-terminal domain could be cleaved by furin 10 . To test the localization of collagen XXV in muscle cells, we transfected C2C12 myoblasts with expression vectors encoding either GFP alone (GFP) or collagen XXV fused to GFP at its C-terminal end (Collagen XXV-GFP) (Fig. 3A). Collagen XXV-GFP was detected on the surface of myoblasts, by staining live cells with an anti-GFP antibody, a method used to detect plasma membrane proteins 17 (Fig. 3Aa,Aa'). We observed that collagen XXV-GFP was concentrated on a restricted region of the cell surface in some myoblasts (Fig. 3Ab,Ab'). To further characterize the role of collagen XXV during muscle differentiation, 10T1/2 fibroblasts were transfected with expression vectors encoding either GFP alone or collagen XXV-GFP (Fig. 3B). 48 h after transfection, the collagen XXV-GFP fusion protein was barely detectable in the cellular extracts, whereas the control GFP was highly expressed in the control transfected cells (Fig. 3C). These results www.nature.com/scientificreports www.nature.com/scientificreports/ suggest that the ectodomain of collagen XXV fused to GFP might be cleaved by furin and liberated from the membrane as a secreted form in transfected 10T1/2 cells, as detected in the culture media of Col25a1-GFP transfected cells (Fig. 3C). To determine whether the ectodomain of collagen XXV is able to regulate myogenic differentiation, differentiating C2C12 myoblasts were cultured four days with the culture medium of GFP or Col25a1-GFP transfected 10T1/2 fibroblasts (Fig. 3B,D). Fusion index was significantly higher when C2C12 were differentiating in the presence of the ectodomain of collagen XXV in the medium (Fig. 3E). We observed a significant increase in large myotubes with more than 20 nuclei, as cells were cultured with the culture medium of Col25a1-GFP transfected 10T1/2 cells compared to the myotubes monitored in control conditions (Fig. 3D,F). To test whether the ectodomain www.nature.com/scientificreports www.nature.com/scientificreports/ of collagen XXV activates the differentiation program of C2C12 cells, we analyzed by RT-qPCR the expression of MyoD, Myf5 and myogenin in C2C12 cells cultured 1 or 2 days in the conditioned media from 10T1/2 cells previously transfected either with GFP or Col25a1-GFP (Fig. 3G). Although the amount of myod transcripts was similar in each condition tested, the expression of the pro-differentiation myogenic factor myogenin was down-regulated in C2C12 cells cultured 2 days in the conditioned medium of Col25a1-GFP transfected 10T1/2 cells, compared to C2C12 cells cultured 2 days in classic differentiation medium or in conditioned media from GFP transfected 10T1/2 cells. This result suggests that the presence of the ectodomain of collagen XXV in the culture medium likely inhibits the differentiation program of C2C12 cells. In contrast to myogenin, Myf5 is highly expressed in proliferating myoblasts and its expression decreases as myoblasts differentiate 18 . We observed a significant increase of myf5 transcript levels in C2C12 cells cultured 1 or 2 days in the presence of collagen XXV-GFP in the differentiation medium compared to control media or conditioned media from GFP transfected 10T1/2 cells (Fig. 3G). Altogether, these results indicate that conditioned media from col25a1-GFP transfected 10T1/2 cells promote fusion of C2C12 cells, without the activation of differentiation.
As conditioned media from 10T1/2 cells might contain many factors which could interfere with differentiation and fusion of C2C12 cells, we next tried to confirm the pro-fusion effect of the collagen XXV-GFP on differentiating C2C12 cells, by minimizing the action of these factors. To do so, we used HEK293T cells to get the highest rate as possible of collagen XXV-GFP proteins, after 48 h transfection. We then enriched collagen XXV-GFP from conditioned media of transfected HEK293T cells by ion exchange chromatography (Fig. 4A, Sup. Fig. 1). We identified and pooled the fractions the most enriched in collagen XXV-GFP. As negative control, we pooled the corresponding fractions eluted from the conditioned media of GFP transfected HEK293T cells (Fig. 4A, Sup. Fig. 1). We observed a significant increase in the number of myotubes containing more than 20 nuclei when C2C12 cells were treated with collagen XXV-GFP enriched fractions compared to myotubes monitored in cultures treated with the control fractions or with recombinant GFP (rGFP) (Fig. 4B-E). These results strengthen those obtained with differentiating C2C12 cultured in conditioned media from the different transfected 10T1/2 cells, indicating that the soluble collagen XXV ectodomain is sufficient to promote the fusion of myoblasts into myotubes. Combined with our in vivo data, these results demonstrate that collagen XXV is required for muscle formation during mouse development, in regulating fusion of myoblasts into multinucleated myofibers. www.nature.com/scientificreports www.nature.com/scientificreports/ Post-transcriptional regulation of Col25a1. Gene expression is regulated at the transcriptional level but also post-transcriptionally by classes of non-coding RNAs 19 . As Col25a1 expression is transient in vitro and in vivo (Fig. 2), we screened for microRNA binding sites within the 3′UTR of the Col25a1 gene using the miRDB program (http://mirdb.org/). We identified miR-208a (mmu-miR-208a-3p) and miR-208b (mmu-miR-208b-3p) www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ as putative candidates to repress Col25a1 in young myotubes, on a seed site highly conserved in mammals (Fig. 5A). miR-208a and miR-208b are both intronic and encoded by two distinct slow myosin genes, Myh6 and Myh7 respectively 20 . This family of miRNAs has a third member, miR-499 (mmu-miR-499-5p), with a sequence highly related to those of miR-208a and miR-208b (Fig. 5A). miR-499 is also intronic, since it is encoded by intron 19 of the slow Myh7b gene. As Myh6/miR-208a is exclusively expressed in the heart 21 , we focused our analyses on Myh7/miR-208b and Myh7b/miR-499, even if miR-499 was not predicted to target the 3′UTR of the Col25a1 transcripts (Fig. 5A). To test whether these miRNAs could repress Col25a1 expression, we analyzed by RT-qPCR the expression of miR-208b and miR-499 and their host genes, during mouse embryogenesis (Figs 2A, 5B) and myogenic differentiation (Fig. 5C). The sequences of miR-208a and miR-208b are so close to each other that it is impossible to discriminate them by RT-qPCR. However, we assumed that the miR-208 transcripts detected here were miR-208b transcripts, since miR-208a is specifically expressed in the heart. During embryonic myogenesis, miR-208b levels were relatively similar in limb muscles (Fig. 5B), despite a net increase of the transcripts of its host gene ( Fig. 2A). In contrast to miR-208b, miR-499 transcript levels exhibited a reverse expression profile to that of Col25a1, like its host gene, Myh7b (Figs 2A, 5B). miR-499 expression was indeed about eight fold higher in limb muscles of E18.5 WT embryos compared to limbs from E12.5 or E14.5 WT embryos, suggesting that miR-499 is a better candidate than miR-208b to repress Col25a1 expression during muscle development. Next, we analyzed the expression profile of miR-208b and miR-499 during differentiation of primary muscle cells (Fig. 5C). Both miRNA expression was low 2 days after differentiation, time point when Col25a1 transcripts were the most abundant. As Col25a1 transcripts decreased (from day 3 of differentiation), miR-208 and miR-499 levels increased, suggesting that these two miRNAs could target Col25a1 transcripts during myogenic differentiation in culture.
To more directly establish the regulation of Col25a1 by miR-208b and miR-499, we manipulated miR-208b and miR-499 in cultured differentiating myoblasts. Transient transfections of synthetic miR-208b, miR-499 or both decreased significantly endogenous Col25a1 transcripts of about 30%, in differentiating primary myoblasts after 48 h transfection and differentiation (Fig. 5D). The efficiency to repress Col25a1 transcripts was similar for each microRNA and co-transfection of both miR-208b and miR-499 did not induce further decrease in Col25a1 mRNA abundance, suggesting a competition between these microRNAs on the 3′UTR of Col25a1. In parallel, we noticed that increased levels of miR-499 specifically increased the amount of Myh7 transcripts, while Myh7b expression seemed to be repressed by miR-208b (Fig. 5D), suggesting a cross-regulation of these microRNAs family members with their host genes. Next, we tested the putative miR-208b/miR-499 binding site on the 3′UTR of the Col25a1 gene by luciferase assays. 10T1/2 fibroblasts were transiently co-transfected with miR-208b, miR-499 or both, in the presence of a miScript Target Protector specific for the miR-208b/miR-499 binding site on the 3′UTR of the Col25a1 gene or its related negative control and a luciferase reporter fused to the 3′UTR of the Col25a1 gene. Transient transfections of synthetic miR-499 alone or in combination with miR-208b significantly decreased the Col25a1 3′UTR reporter activity, and this repression was abolished in the presence of the miScript target protector specific to the miR-208b/miR-499 binding site of the 3′UTR of the col25a1 gene (Fig. 5E). In contrast to miR-499, transfection of synthetic miR-208b alone did not significantly modify the luciferase activity of the reporter, suggesting that miR-499 is probably the main regulator of col25a1 transcripts (Fig. 5E). These repressions were completely abrogated when miR-208b and/or miR-499 were co-transfected with a luciferase reporter construct harboring a mutated miR-208b/miR-499 binding site in the Col25a1 3′UTR. These results provide evidence for a consistent post-transcriptional regulation of Col25a1 transcripts mainly by miR-499, during muscle differentiation and development.

Discussion
Implication of the extracellular matrix in primary myogenesis. The ECM has long been known to be important for muscle integrity and function. In myogenesis, cell-ECM interactions contribute to the formation of myotubes in the direction and organization of muscles to be. In the adult, myofibers are surrounded by an ECM material, the basement membrane. The inner layer of the basement membrane of degenerated fibers, the basal lamina, has recently been shown to play a critical role in muscle regeneration, since it serves as a scaffold directing myogenic progenitors alignment for division, migration and therefore fusion into regenerating myotubes 22 . The basement membrane appears only 2-3 days before birth in mice. Building the muscle tissue in the embryo also need a scaffold. During fetal development, secondary myoblasts adhere to primary myotubes and use them as a scaffold for their alignment and fusion into fetal myofibers 23 . So, primary myofibers serve as a scaffold for the remaining stages of myogenesis and set the orientation of all muscles. In contrast to secondary myofibers, primary myoblasts fuse in an environment composed of an interstitial ECM. Cells interact with the ECM via a variety of receptors, the integrins being the most common. Several studies showed that β1 integrins are essential for primary myoblasts to interact with their surrounding ECM and properly fuse during embryonic myogenesis [24][25][26] . However, it is not clear which ligands in the ECM could be involved in these cell-ECM interactions. Here, we provide evidence that collagen XXV is part of the ECM involved in the proper set up of primary myofibers, since the genetic ablation of collagen XXV leads to the formation of short and misaligned myofibers in E12.5 and E14.5 limbs (Fig. 1). Indeed, collagen XXV expression is high during primary myogenesis and reduces as secondary myogenesis occurs (Fig. 2), suggesting an important role for collagen XXV in the early stages of the construction of the muscle tissue. Whether this collagen is able to bind β1 integrins during primary myogenesis remains to be determined. As the mouse model used in our study is a whole body knock out model, we cannot exclude that non muscle cells might express collagen XXV within the embryonic limb. However, our in vitro data clearly show that primary myoblasts express Col25a1 transcripts as they proliferate and differentiate (Fig. 2). Furthermore, Col25a1 transcripts were not detected in fibroblast-like cells isolated from limb skeletal muscles of newborn mice (data not shown). Therefore, we hypothesize that primary myoblasts might be the main source of collagen XXV within the limb bud. Overall, our data combined with studies from the literature point out the crucial role of ECM in the establishment of primary myotubes, which will be used as the scaffold for all subsequent stages of myogenesis.
www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ Role of collagen XXV in the fusion of myoblasts. Skeletal muscle development requires migration of muscle progenitor cells, their differentiation and fusion into myotubes before maturating. Fusion of muscle cells is probably the most poorly understood process in this mechanism, particularly in mammals. Some transmembrane molecules involved in the myoblast fusion process have recently been identified. Myomaker, a muscle specific transmembrane protein, is necessary for myoblast fusion during skeletal muscle development and regeneration 27 . To stimulate its fusogenic activity, it interacts with the transmembrane micropeptide Myomixer (also named Myomerger or Minion) 28 . This micropeptide also promotes myoblast fusion and skeletal muscle formation during development [28][29][30] . How this small peptide induces membrane fusion is not understood yet. Before fusion, the cells might recognize each other and adhere 31 . Moreover, during myotube formation, the negatively charged phospholipid, phosphatidylserine (PS), transiently translocates from the inner layer of the plasma membrane to the outer leaflet 32 . Recognition of cell surface-exposed PS by PS receptors, such as stabilin-2, also promotes fusion with neighboring myoblasts and skeletal muscle regeneration 33 . In addition, the externalization of PS controls the orientation of the fusion, thereby preventing random fusion and leading to the proper elongation of myotubes 34 . In this context, we identified a completely new regulator of fusion of myoblasts. In absence of collagen XXV, myoblast fusion is impaired in vivo at E12.5 and E14.5 (Fig. 1), whereas primary cells collected from E18.5 col25a1−/− mice could spontaneously fuse into myotubes as efficiently as WT ones (data not shown). Altogether, these results suggest that this collagen is rather involved in the fusion of myoblasts into myotubes during primary myogenesis, its implication during secondary myogenesis being not essential. In culture, increased levels of the soluble ectodomain of collagen XXV promotes fusion of myoblasts into multinucleated myotubes, in a manner independent from the differentiation process, since the differentiation marker myogenin was not increased by the soluble collagen XXV in the culture medium (Figs 3 and 4). Altogether, our data further point out the role of membrane-associated molecules in myoblast fusion, mainly during primary myogenesis. As collagen XXV is initially a transmembrane protein before being cleaved by furin (Fig. 3A), we tested whether overexpression of collagen XXV would promote fusion of 10T1/2 cells with C2C12 cells (data not shown). We did not observe any sign of heterologous fusion, indicating that collagen XXV is probably not a fusogenic protein, but rather involved in the recognition or adherence of fusion competent myoblasts. The precise mechanism by which collagen XXV is involved in the fusion process remains to be determined. miR-499, the first microRNA involved in myogenic fusion. In this study, we also described a new type of regulators of myogenic fusion. Indeed, by regulating the expression of Col25a1, miR-499 also participates to the fusion process. In contrast to miR-208b, miR499 expression is significantly increasing during embryonic myogenesis, as Col25a1 transcripts decrease (Fig. 5B). Although it was the only one from this family not to be identified in the miRDB database as targeting mouse Col25a1 expression, miR-499 is probably the microRNA most implicated in the Col25a1 repression during embryonic myogenesis (our results). Furthermore, miR-499 is identified in the database as a putative regulator of human Col25a1 transcripts. Interestingly, the expression of miR-499 and its host gene, Myh7b, is under the control of miR-208a in the heart 20 . This is not the case with miR-208b and Myh7b/miR499 in the skeletal muscle tissue, since overexpression of miR-208b in differentiating primary myoblasts decreased the levels of Myh7b transcripts, whereas miR-499 levels were not reduced (Fig. 5D). According to the miRDB database, the murine Myh7b gene would be targeted at its 3′ UTR by only three microRNAs on seed regions completely different from the miR-208b one. Therefore, our results rather suggest that one of the targets of miR-208b could be the Myh7b pre-mRNA itself. One can hypothesize that miR-208b targets the miR-499 sequence in intron 19 to regulate transcription and/or splicing of Myh7b, inducing downregulation of its expression. Indeed, it has been described that a specific alternative splicing of Myh7 transcripts blocks Myh7 translation and activates Myh7b mRNA degradation 35 . Moreover, our results indicate that miR-499 modulates Myh7 expression. Since overexpression of miR-499 induces Myh7 overexpression and not a downregulation, miR-499 could stabilize Myh7 transcripts (Fig. 5D). In total, these results point out an original cross-regulation of miR-499 and miR-208b on the expression of their host genes, Myh7 and Myh7b respectively. Similar to these specific functions on Myh7 and Myh7b expression, these microRNAs seem to have distinct mRNA targets, Col25a1 transcripts being a favorite target for miR-499, even if miR-208b overexpression could also repress Col25a1 expression in cultured differentiating primary myoblasts (Fig. 5D,E). Our results suggest that miR-499 transiently participates in vitro and in vivo to myogenic differentiation by controlling myoblasts fusion as myotubes mature and activates the expression of Myh7 and Myh7b. Thereby, this microRNA would control the amount of nuclei within each myotube by repressing expression of collagen XXV. Whether this microRNA could control expression of other transcripts involved in myoblast fusion remains to be determined.
Fusion of myoblasts during embryonic development, as well as in adulthood, appears to be a critical process, which must be highly and tightly controlled by several mechanisms. Here, we show for the first time the implication of a collagen in this process. This transmembrane collagen is transiently expressed and tightly regulated by a muscle specific microRNA, giving the first evidence of the involvement of microRNAs in this cellular mechanism.
Cloning. The Col25a1 expressing construct was generated from a mouse cDNA library. The Col25a1 sequence was cloned in frame into the pCI neo-EGFP reporter vector, using the following primer sequences: Forward 5′-ATAGGCTAGCCTCGAATGTTGGTGAAGAAGCTTGC-3′; Reverse 5′-TGGTGGCGACCGGTGGATCCCGGGACTTTTGCCAGCAGCC-3′.
The luc-Col25a1 vector contains 4 kb of the 3′UTR sequence of the Col25a1 transcript, cloned into the pGl 4.23 reporter vector at the Xho I site. The putative miR-499/miR208b target site was mutated in the 3′UTR of the Col25a1 gene by directed mutagenesis, using the following primers: The integrity of all constructs was confirmed by sequencing.
Conditioned media and in vitro differentiation assay. 10T1/2 cells were transfected 6 h with 1 µg GFP or Col25a1-GFP expression vectors using 10 mM PEI (Polyethylenimine) reagent. The next day, differentiation medium (DMEM high glucose, with 2% horse serum) was added to the cells (conditioned differentiation medium). In parallel, C2C12 cells were seeded at 2500 cells/cm 2 and cultured in proliferation medium (DMEM high glucose, with 20% FBS). The day later, the conditioned differentiation medium from transfected 10T1/2 cells was added to C2C12 cells (differentiation condition day 0). Two days later, the conditioned differentiation medium of C2C12 was replaced by new conditioned differentiation medium from freshly transfected 10T1/2 cells (differentiation condition day 2). C2C12 cells were fixed at differentiation condition day 4.

Enrichment of Collagen XXV-GFP soluble fractions and in vitro differentiation assay. HEK293T cells
were transiently transfected 8 h with plasmids encoding the Collagen XXV-GFP fusion protein or GFP alone using lipofectamine 2000 (Invitrogen). The next day, the culture medium was shifted to serum-free culture medium containing 0.02% BSA. For both conditions, conditioned media were collected 24 h later, adjusted to 50 mM Tris-HCl pH8 then loaded onto a 1 mL HiTrapQ anion-exchange column (GE Healthcare) equilibrated in the same buffer. Bound proteins were eluted in 0.5 mL fractions along a NaCl gradient (0-500 mM). All fractions were analyzed by immunoblot to identify those containing the Collagen XXV-GFP fusion protein. The 3 fractions containing the highest amounts of Collagen XXV-GFP fusion protein were pooled and frozen until used for in vitro differentiation assays. The corresponding fractions eluted from the GFP-alone conditioned media were processed identically and used as negative control. For the differentiation assay, C2C12 cells were seeded at 2500 cells/cm 2 and cultured in proliferation medium (DMEM high glucose, with 20% FBS). The next day, C2C12 cell medium was switched to differentiation medium, including 10 µL/mL of pooled fractions enriched in Collagen XXV-GFP or the same amount of the corresponding fractions from GFP-alone conditioned media, or 0.5 ng/mL of recombinant GFP (Novus biologicals). Half of the differentiation medium was refreshed 2 days later, prior fixation 4 days after the beginning of the treatments.
RNA extraction and quantitative RT-PCR. Total RNA from tissues or cultured cells was extracted using the TRIzol reagent (Invitrogen), followed by a RQ1 RNase-free DNase treatment (Promega). Reverse transcription was realized from 1 µg total RNA using the miScript II RT kit (Qiagen) according to the manufacturer's protocol. Quantitative real-time PCR were performed in triplicate using SYBR green Supermix (BioRad) with a CFX ™ 384 Touch qPCR system (BioRad). 26S was used as a normalizer. Primers used are listed in Table 1.

Western blot analysis.
Total proteins were extracted from cultured cells with protein lysis buffer (2% SDS, 50 mM Tris-HCl pH 6.8, 10% glycerol, 100 mM DTT). To analyze the presence of the fusion protein collagen XXV-GFP in the cell culture medium, transfected 10T1/2 cells were cultured overnight in a serum free medium. Cell culture media were then collected and concentrated using Amicon 10 K (Millipore). Bio-Rad protein assay with BSA standard was performed to measure protein concentration of samples. To detect collagen XXV-GFP proteins in conditioned media of transfected HEK293T cells, cells were cultured overnight in a serum free medium containing 0.02% BSA and an aliquot of the conditioned media was used for Western blotting. 20 µg proteins were separated by sodium dodecyl sulfate polyacrylamide gel 10% electrophoresis. The separated proteins were transferred to a PVDF membrane before immunoblotting with the following primary antibody: anti-GFP (1/2000, Santa Cruz sc-9996). Anti-mouse-HRP (Cell signalling 7076 S) was used as a secondary detection antibody. Finally, immunoreactive bands were detected using the ECL chemiluminescence detection. statistical analysis. The fusion index was calculated as the percentage of nuclei in myosin positive myotubes with more than 3 nuclei of total nuclei. The number of nuclei in myotubes was manually counted. The results are presented as means ± s.e.m. Statistical analyses were performed using t-test or two-way ANOVAs followed by a two-way repeated-measures analysis of variance with a Tukey's multiple comparisons test (Figs 3G, 4E). Statistical significance was accepted at a P value < 0.05. www.nature.com/scientificreports www.nature.com/scientificreports/