Removing Cdon from muscle stem cells reduces their ability to activate key signals and regenerate muscle

To generate Cdon conditional knockout mice, three embryonic stem (ES) cells lines with conditional knockout potential for Cdon were obtained from the European Conditional Mouse Mutagenesis Program. Southern blotting analysis was performed by standard protocols to confirm the conditional Cdon allele (Cdon^tm3a in Supporting Information, FigureS1) in two of the three ES cell lines (C03 and H02). Mouse lines were generated from these ES lines by the Mouse Genetics and Gene Targeting Core of Icahn School of Medicine at Mount Sinai. Both lines were similar, and line H02 was used for further analysis. The probes for the Southern blots were derived by PCR with the following probes: NsiI forward, AATGGACTGGCGTTGTGTACTGG; NsiI reverse, GTTTGTGTGTGCCTGCCCTTTC; NheI forward, ATGGAGATGCTTGTGGATGGG; NheI reverse, AAAGCCTCATTTCAGCCCAGC. See the legend to FigureS1 for further details.

Cdon^fl/fl mice were crossed with Pax7^CreERT2 mice (obtained from the Jackson laboratory). The primers used for genotyping are listed in TableS1. Tamoxifen (tmx) (Sigma‐Aldrich, 1.5 mM) was administered intraperitoneally every 2 days for five times prior to muscle injury. To induce the skeletal muscle regeneration, mice were anaesthetized with a 1–2% isoflurane followed by CTX (Sigma‐Aldrich, 10 μM) injection into tibialis anterior (TA) muscle in a volume of 25 μl. Zmpste24^−/− mice (kindly gifted by Dr. Zhongjun Zhou, University of Hong Kong) and wild type (WT) control mice were sacrificed to analyze satellite cells. All animal experiments were approved by the Institutional Animal Care and Research Advisory Committee at Sungkyunkwan University School of Medicine Laboratory Animal Research Center and complied with the regulations by the institutional Ethics Committee.

C2C12 myoblasts and 293 T cells were cultured as previously described.27 To determine the response to basic FGF (bFGF; Sigma‐Aldrich), C2C12 myoblasts were changed from Dulbecco's Modified Eagle Medium (DMEM) containing 15% foetal bovine serum (FBS) to DMEM containing 0.2% bovine serum albumin for overnight, followed by 5 ng/mL bFGF treatment as indicated. For adhesive culture, C2C12 cells were cultured either on matrigel (Corning) or poly‐L‐lysine (PLL) (Sigma‐Aldrich) coated petri dishes for 2 days. Primary myoblasts were isolated from hindlimbs of Cdon^+/+, Cdon^−/−, and Cdon^fl/fl;Pax7^CreERT2 mice as previously described.28 For the Cre recombinase‐mediated in vitro deletion of Cdon, 4‐hydroxytamoxifen (4‐OHT; Sigma‐Aldrich, 1 µM) were treated for 4 days after amplification of isolated Cdon^fl/fl;Pax7^CreERT2 primary myoblasts in F10 medium (Gibco) containing bFGF (2.5 ng/mL) and 20% FBS. Single myofiber isolation was performed as previously described.27 Briefly, extensor digitorum longus (EDL) muscles were digested with collagenase 2 (Worthington, 400 unit/mL), followed by seeding on matrigel‐coated plates in DMEM medium containing 20% FBS. Lentivirus‐expressing shRNA Cdon (shCdon) was generated with a modified lentiviral vector derived from pLKO.1 (Sigma‐Aldrich) in HEK 293 T cells using helper plasmids pCMV‐VSV‐G and pCMV dR 8.9. The shRNA sequences for Cdon used in this study are as previously reported.28

For histology and immunostaining of muscles, snap freezing and cryosections were carried out as previously described.,To assess the structure and fibrosis, fixed cryosections were stained with Mayer's haematoxylin and eosin (BBC Biomedical) and Picro‐Sirius red (Abcam) according to the manufacturer's protocol. Immunostaining was performed as previously described.,Cultured cells, single myofibers, and frozen muscle sections were fixed, permeabilized and processed for incubation with primary antibodies against Cdon, Pax7, Laminin, eMyHC, activated‐integrin β1 (*Itgb1), cleaved Caspase3, pγH2AX, and secondary antibodies (Thermo Fisher). SA‐β‐gal staining was conducted by using SA‐β‐gal kit (Cell Signaling Technology) as previously described.SA‐β‐gal staining for muscle tissue was performed as previously described.Briefly, freshly frozen sections were fixed, followed by incubating X‐gal mixture (pH 5.5) for 2 days at 37°C. 5‐Ethynyl‐2'‐deoxyuridine, a nucleoside analog of thymidine (EdU) incorporation assay was performed by adding EdU with a final concentration of 10 μM for 2 h before fixation and then analyzed using the Click‐iT EdU kit (Invitrogen) according to the manufacturer's protocol. Images were acquired under LSM‐710 Meta confocal microscope or Nikon ECLIPSE TE2000‐U microscope. 29 30 29 30 31 32

Quantitative real‐time PCR analysis was examined as previously described.33 cDNA was obtained from total RNA using a PrimeScript RT reagent kit (TAKARA) and analyzed by quantitative real‐time PCR using SYBR Premix Ex Tag^TM (TAKARA). The primers sequences are listed in TableS1, and each transcript levels were normalized with ribosomal protein L32. RNA sequencing analysis was performed with the Illumina NextSeq500 (ebiogen, Korea). All gene set enrichment analysis (GSEA) plots including GSEA enrichment plot and map were generated with the GSEA software (Broad Institute; software.broadinstitute.org/gsea↗ /).

Immunoblot and immunoprecipitation analyses were carried out as previously described.30 Briefly, tissues and cells were lysed in radioimmunoprecipitation assay (Thermo) buffer with complete protease inhibitor cocktail (Roche) and separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis. For immunoprecipitation, precleared cell extracts were incubated with primary antibodies overnight at 4°C, followed by incubation with protein A‐agarose or G‐agarose beads (Roche Diagnostics) for 1 h and washing three times with cell extraction buffer. Precipitates were analyzed by western blotting. Surface biotinylation was performed as previously explained.34 Briefly, biotinylated surface proteins were precipitated with streptavidine‐agarose beads (Pierce) and analyzed with sodium dodecyl sulfate–polyacrylamide gel electrophoresis. Primary antibodies are listed in TableS2.

All values were analyzed using GraphPad Prism 7 or Excel spreadsheet. To assess the statistical difference, Student's t‐test with two‐tailed paired comparison was performed. In case of comparison groups as indicated in the figure legend, we determined the statistical significance by one‐way analysis of variance with Tukey post hoc test. Statistical significance was indicated as P < 0.05 (*), P < 0.01 (**), and P < 0.001(***). Error bar were represented by mean ± standard deviation (mean ± SD).

To investigate the role of Cdon in muscle regeneration, we have analyzed the expression of Cdon in satellite cells by immunostaining of myofibers isolated from EDL muscles. Cdon is expressed in Pax7‐positive satellite cells (Figure1A), preferentially at the basal membrane (FigureS2). To define the in vivo function of Cdon in muscle regeneration, we generated mice carrying a floxed allele of Cdon (FigureS1). These mice were crossed with Pax7^CreERT2 mice, which express tmx‐inducible Cre recombinase in satellite cells but not other cells in skeletal muscle tissue35; 3‐month‐old Cdon^fl/fl;Pax7^CreERT2 mice were injected five times with saline (mock) or tmx starting 8 days prior to the first CTX injury of TA muscles; 21 days after the first injury, a second injury was initiated (Figure1B). Immunoblotting of TA muscles revealed that Cdon protein was depleted in tmx‐treated muscles 7 days after the second injury, compared with the mock saline‐treated muscles (FigureS3A and S3B). In addition, quantitative RT‐PCR analysis of isolated myoblasts and fibroblasts treated with mock or 4‐hydroxytamoxifen (4‐OHT) showed that Cdon transcription was specifically decreased in myoblasts (FigureS3C). Cdon ablation in satellite cells mildly impaired muscle regeneration as assessed by myofiber cross‐sectional area 21 days after the first injury (FigureS4A and S4B). After the second injury, a more severe impairment in muscle regeneration was observed (Figure1C). Control muscles displayed myofibers with centrally localized nuclei, characteristic of regenerating muscles 7 days after the second CTX injury (PID7), while Cdon‐depleted muscles accumulated single cells in interstitial areas between regenerating myofibers. In addition, immunostaining for laminin and embryonic myosin heavy chain (eMyHC, a marker for immature myofibers) demonstrated that more myofibers in Cdon‐depleted muscles at PID7 were positive for eMyHC, compared with control muscles (Figure1C and 1D). Consistent with this, eMyHC transcript levels were significantly increased in Cdon‐depleted muscles, relative to control muscles (Figure1E). The myofiber size in Cdon‐depleted muscles 21 days after the second injury (PID21) was greatly reduced, compared with the control muscles (Figure1C and 1F). Impaired muscle regeneration processes are frequently linked with aberrant extracellular matrix (ECM) remodelling and fibrosis.36, 37 In agreement with this notion, Sirius red staining showed that Cdon depletion in satellite cells led to an exacerbated fibrotic response at PID7 and PID21 (Figure1G and 1H). Similarly, transcript levels of Collagen 1a1 (Col1a1) and Col1a3 were mildly but significantly increased in Cdon‐depleted muscles (Figure1I). The weights of Cdon‐depleted muscles at PID7 or PID21 were also significantly decreased, compared with control muscles (FigureS5). Taken together, these data demonstrate that Cdon ablation impairs satellite cell function.

To investigate the detailed mechanism underlying the impaired regenerative capacity of Cdon‐ablated satellite cells, we first analyzed their proliferative capacity. To do so, TA muscles of mock or tmx‐treated mice were injured and injected with bromodeoxyuridine (BrdU) 2 days after a single injury to label proliferating cells, followed by immunostaining analysis 24 h later (Figure2A). Forty per cent of cells in control regenerating muscles were BrdU‐positive, while ~20% of cells in the tmx‐treated muscles were BrdU‐positive (Figure2B and 2C). In addition, TA muscles of mock or tmx‐treated mice at 4 days after CTX injury were immunostained for Pax7 and Ki67 (FigureS6). In the control muscles, about 27% of cells were double positive for Pax7 and Ki67, while ~13% of cells in tmx‐treated muscles were double positive. To further examine this, single myofibers from EDL muscles were isolated from mock or tmx‐treated mice and cultured for 48 h in the presence of bFGF followed by 5‐ethynyl‐2'‐deoxyuridine (EdU) incorporation assay (Figure2D and 2E). In agreement with the results from in vivo BrdU incorporation, EdU incorporation was significantly lower in satellite cells on myofibers from tmx‐treated mice, compared with control myofibers. Furthermore, Cdon^−/− primary myoblasts exhibited significantly reduced BrdU incorporation relative to that of Cdon^+/+ myoblasts (Figure2F and 2G). However, Cdon‐deficient myoblasts showed no overt signs of cell death, as measured by cleaved Caspase 3 immunostaining (FigureS7). Self‐renewal of satellite cells is critical for muscle regeneration and maintenance.7, 9 Mock or tmx‐treated TA muscle sections from PID21 were therefore immunostained for Pax7 and laminin, and the number of Pax7‐positive cells was quantified (Figure2H and 2I). In control muscle sections, there were about 12 Pax7‐positive cells per 100 regenerating myofibers 21 days after the first injury, which decreased to ~10 such cells after the second injury. Cdon‐depleted muscles had a significantly decreased number, 5–6 Pax7‐positive cells per 100 regenerating myofibers 21 days after each injury. Taken together, these data reveal that Cdon depletion causes reduced proliferative and self‐renewal capacity of satellite cells.

We next analyzed muscles at PID21 for production of Pax7 and pγH2AX (Figure3A and 3B). Cells double positive for Pax7 and pγH2AX or positive for pγH2AX only were quantified. Cdon depletion resulted in an increased proportion of pγH2AX‐positive cells in both Pax7‐positive and Pax7‐negative populations, compared with controls. To address whether impaired self‐renewal and proliferative capacity in Cdon‐depleted satellite cells are associated with cellular senescence, mock and tmx‐treated muscle sections from PID21 were subjected to SA‐β‐gal staining (Figure3C and 3D). SA‐β‐gal‐positive cells were rarely found in the control muscles, while Cdon‐depleted muscles averaged 11 SA‐β‐gal‐positive cells per field. To confirm the effect of Cdon depletion on cellular senescence, primary myoblasts isolated from Cdon^fl/fl;Pax7^CreERT2 mice were treated in vitro with dimethyl sulfoxide or 4‐OHT for 4 days in the presence of bFGF and subjected to SA‐β‐gal staining (Figure3E and 3F). More than 30% of Cdon‐depleted myoblasts were positive for SA‐β‐gal staining, while <5% of dimethyl sulfoxide‐treated control myoblasts were SA‐β‐gal positive. Additionally, Cdon^−/− primary myoblasts exhibited significantly increased pγH2AX‐positive cells relative to that of Cdon^+/+ myoblasts (FigureS8). These data indicate that Cdon depletion resulted in cellular senescence and suggested it produces defective responsiveness to bFGF.

Similar to Cdon‐deficient satellite cells, C2C12 cells infected with a Cdon shRNA lentivirus expressed higher levels of p21 and p16 proteins, compared with control lentivirus‐infected cells (Figure3G). Cdon‐depleted C2C12 cells also proliferated less efficiently than control cells (Figure3H). SA‐β‐gal staining and pγH2AX immunostaining revealed that Cdon depletion significantly elevated SA‐β‐gal‐positive cells and pγH2AX‐foci, compared with control (Figure3I–3K). To further examine this, qRT‐PCR analysis for SASP genes (Igfbp5 and interleukin‐6) and E2F target genes was performed in control or Cdon‐depleted C2C12 cells (Figure3L). Similar to the SA‐β‐gal staining, the expression of Igfbp5 and interleukin‐6 was significantly increased in Cdon‐depleted cells. In contrast, the E2F target genes Cyclin E1, Lamin B1, Mcm3, and Cdc6 were mildly but significantly decreased in Cdon‐depleted cells, compared with the control cells. Taken together, these data suggest that Cdon depletion in satellite cells and myoblasts results in cellular senescence and impaired stem cell function, likely contributing to decreased muscle regeneration.

To determine the regulatory pathways underlying regeneration defects caused by Cdon deficiency, we performed RNA sequencing analysis with muscles of either mock‐injected or tmx‐injected mice 4 days after a single injury (Figure4A). To identify groups of genes associated with observed phenotypes in Cdon‐deficient satellite cells, GSEA, as an unbiased approach, was performed (Figure4B–4E). GSEA analysis revealed that gene sets implicated in endogenous stimulus, external stimulus, and negative regulation of response to stimulus were significantly altered (dashed lines indicate nominal P = 0.05 and false discovery rate q = 0.05) (Figure 4B and Figure S9). Moreover, KEGG pathway mapping showed that genes associated with arginine and proline metabolism, ECM‐receptor interaction, pathways in cancer, and MAPK signalling pathway were altered in Cdon‐depleted muscles. Enrichment Map (Figure 4C), a network‐based method for gene set enrichment visualization, revealed that the key seven gene sets are tightly associated with each other, implying that genes of those seven pathways underpin the phenotypes of Cdon‐deficient satellite cells. Additionally, the gene expression patterns of those gene sets (e.g. KEGG MAPK signalling pathway and GO Response to external stimulus) seemed to be divided in half (Figure 4D and 4E). Furthermore, genes involved in inflammatory response, ECM, negative regulation of cell population proliferation were perturbed in Cdon‐deleted muscles (Figure S10). Interestingly, the expression of myogenic regulatory factors, including Myf5, Myogenin, Pax7, Pax3, Myod1, and Mef2c were not significantly altered between control and Cdon‐deleted muscles (Figure S11A), suggesting satellite cell activation initiated normally.

To further examine whether decreased proliferation of Cdon‐depleted satellite cells is because of perturbed MAPK signalling, qRT‐PCR analysis of genes regulated by the ERK was performed. Expression of Fosb, Fos, and Egr1, immediate early response genes induced by ERK, was significantly decreased, while expression of Oas1a and Oas2, which is inhibited by ERK,38 was elevated (Figure5A). Furthermore, Cdon‐deficient primary myoblasts cultured in the presence of bFGF and Cdon‐depleted C2C12 cells in growth medium exhibited significantly decreased Egr1 expression (Figure5B). The RNA sequencing data showed no significant alteration in expression of growth factors (various FGFs and HGF) or FGF receptors that are related to ERK activation (FigureS11B). Similarly, Cdon^+/+ and Cdon^−/− myoblasts had similar levels of FGFR1 and FGFR4 proteins (Figure5C). To explore further, control and Cdon‐depleted C2C12 cells were serum starved for 16 h and then treated with bFGF (5 ng/mL) for 5, 10, or 30 min and analyzed for expression of Fos and Egr1 by qRT‐PCR (Figure5D). In line with the results shown in Figure5B, Cdon‐depleted C2C12 cells showed significantly delayed and reduced Fos and Egr1 expression, compared with the control cells. Treatment with bFGF for 2 min greatly elevated active, phosphorylated ERK1/2 (p‐ERK1/2) levels in control cells, and this was further increased at 5 min of treatment (Figure5E). In contrast, Cdon‐depleted cells exhibited starkly attenuated ERK activation in response to bFGF. To assess whether this decrease in ERK activation also occurred in regenerating Cdon‐depleted muscles, mock and tmx‐treated muscles at PID4 were subjected to immunoblotting. The levels of p‐ERK1/2 were substantially reduced in Cdon‐depleted muscles, compared with control muscles (Figure5F and 5G). Collectively, these data demonstrate that the ablation of Cdon in satellite cells impairs ERK activation in response to FGF signalling, likely contributing to defective proliferation and regeneration.

To investigate the potential mechanism leading to the impaired FGF responsiveness in Cdon‐deficient satellite cells, we analyzed a potential interaction between FGFR and Cdon in C2C12 cells by coimmunoprecipitation with control IgG or Cdon antibodies (Figure5H). FGFR1 and FGFR4 proteins were coprecipitated with Cdon in C2C12 cells. Next, we assessed whether Cdon regulates the surface localization of FGFRs by surface biotinylation (Figure5I). Interestingly, the surface levels of FGFR1 and FGFR4 decreased in Cdon‐depleted cells while the surface cadherin level was largely unchanged, relative to the control cells. In a previous study, we demonstrated that Cdon levels were decreased in C2C12 myoblasts cultured on a non‐specific cell adhesion promoting substratum, PLL. Cdon depletion by siRNA or by culture on PLL decreased integrin/FAK/ERK signalling.26 To further explore this observation, C2C12 cells were cultured on PLL‐coated or Matrigel‐coated petri dishes to promote integrin signalling in presence of serum. The levels of Cdon, p‐ERK, and pFAK were greatly decreased in cells cultured on PLL while p21 levels were strongly enhanced (FigureS12A and S12B). Cells cultured on PLL exhibited starkly blunted proliferation, evident by significantly decreased BrdU incorporation and a concurrent increase in SA‐β‐gal expression, compared with cells cultured on Matrigel (FigureS12C and S12D). Similar to Cdon‐depleted cells, C2C12 cells cultured on PLL had diminished surface levels of FGFR1, FGFR4, and Cdon while cadherin levels were largely unchanged (Figure5J). Collectively, these data suggest that Cdon regulates surface localization of FGFR1 and FGFR4, thereby regulating FGF responsiveness of myoblasts.

A previous study suggested that defective integrin signalling is linked with impaired FGF responsiveness, thereby contributing to reduced proliferation of aged muscle stem cells.20 We therefore examined integrin activation in satellite cells on freshly isolated myofibers from vehicle and tmx‐treated mice by immunostaining with antibodies to active integrin β1 (Itgb1*) and Cdon (Figure6A and 6B). About 76% of control Pax7‐positive satellite cells had polarized localization of Itgb1* at the basal lamina contact site of satellite cells. In contrast, only ~40% of Cdon‐depleted satellite cells had nonpolarized Itgb1* proteins, and they frequently had weaker overall Itgb1* signals. This result resembled the phenotype observed in aged satellite cells.20 The level of Itgb1* protein was assessed in control and Cdon‐deficient primary myoblasts (Figure6C and 6D). Itgb1* levels decreased in Cdon‐deficient myoblasts, relative to the control myoblasts while total Itgb1 proteins were unchanged. Additionally, we did not observe any coimmunoprecipitation between Cdon and Itgb1 when overexpressed in 293T cells nor did Cdon depletion reduce cell surface levels of Itgb1 in C2C12 cells (FigureS13). A reduction in expression of the integrin ligand fibronectin has been implicated in aged satellite cell function.22 However, the RNA sequencing data revealed that Fibronectin expression was not altered in Cdon‐depleted muscles whereas distinct collagen types (Col1a1, Col5a1, Col3a1, and Col14a1) were significantly altered, likely reflecting the enhanced fibrosis in Cdon‐depleted muscles (FigureS10).

To further understand the role of Cdon in muscle aging, we examined expression of Cdon and Itgb1* in satellite cells from mice with the Zmpste24 protease deficiency, of a model for progeria.39 These mice exhibit, among other aging‐related phenotypes, premature muscle loss associated with elevated cellular senescence and reduced proliferation of muscle stem cells.40, 41, 42 Myofibers isolated from EDL muscles of 4‐month‐old wild type or Zmpste24^−/− mice were immunostained for Cdon, Itgb1*, and Pax7 (Figure6E–6G). About 90% of Pax7‐positive satellite cells from control myofibers were positive for Cdon expression, while about 63% of Pax7‐positive cells from Zmpste24‐deficient myofibers were positive for Cdon. Similarly to Cdon‐deficient satellite cells, more Zmpste24‐deficient satellite cells exhibited either weaker or nonpolarized Itgb1* signal, compared with the wild type cells. These data collectively suggest that Cdon and integrin signalling are perturbed in aging satellite cells. Furthermore, SA‐β‐gal‐positive cells were easily observed in Zmpste24‐deficient muscles while wild type muscles had only very few cells positive for SA‐β‐gal‐positive cells (Figure6H and 6I). Additionally, Zmpste24‐deficient muscles also exhibited greatly elevated immunostaining for pγH2AX in both Pax7‐positive and negative cells compared with wild type muscles (Figure6J and 6K). These results suggested that Cdon depletion is correlated with satellite cell aging. In summary, our study demonstrates a key role of Cdon in the maintenance of satellite cell self‐renewal and proliferation through FGFR trafficking and signalling.