Islr regulates canonical Wnt signaling-mediated skeletal muscle regeneration by stabilizing Dishevelled-2 and preventing autophagy

Satellite cells are crucial for skeletal muscle regeneration, but the molecular mechanisms regulating satellite cells are not entirely understood. Here, we show that the immunoglobulin superfamily containing leucine-rich repeat (Islr), a newly identified marker for mesenchymal stem cells, stabilizes canonical Wnt signaling and promote skeletal muscle regeneration. Loss of Islr delays skeletal muscle regeneration in adult mice. In the absence of Islr, myoblasts fail to develop into mature myotubes due to defective differentiation. Islr interacts with Dishevelled-2 (Dvl2) to activate canonical Wnt signaling, consequently regulating the myogenic factor myogenin (MyoG). Furthermore, Islr stabilizes Dvl2 by reducing the level of LC3-labeled Dvl2 and preventing cells from undergoing autophagy. Together, our findings identify Islr as an important regulator for skeletal muscle regeneration.

R egeneration is critical to maintain the homeostasis of adult skeletal muscle in animals, and satellite cells (SCs) play an important role during this process. Adult skeletal muscle fails to regenerate by genetic ablation of SCs postnatally [1][2][3] . Upon adult skeletal muscle injury, quiescent Pax7 + SCs are activated and give rise to a population of Pax7 + /Myf5 + committed SCs 4,5 . These progenitor cells proliferate into Pax7 + /MyoD + myoblast cells 6,7 and differentiate into the myogenin + (MyoG + ) cells that fuse to form multinucleated myotubes 8,9 . SC differentiation is essential for skeletal muscle regeneration, and it is increasingly clear that SC differentiation is regulated by many signaling pathways. In particular, the canonical Wnt signaling pathway is important for promoting SC differentiation during skeletal muscle regeneration 10,11 . When Wnt ligands bind to Frizzled receptors and members of the low-density lipoprotein receptorrelayed protein (LRP) family, the canonical Wnt signaling is activated. The cytoplasmic part of Frizzled interacts with Disheveled-2 (Dvl2), facilitating interaction between Axin and the LRP tail, which destroys the β-catenin destruction complex and blocks the ubiquitination of β-catenin. Subsequently, β-catenin translocates into the nucleus and further forms a complex with the TCF transcription factors to activate transcription of Wnt target genes [12][13][14] . Dvl2 is the hub of the canonical Wnt signaling pathway; autophagy mediates the degradation of Dvl2 and further negatively regulates canonical Wnt signaling in response to cellular metabolic stress 15,16 . The deletion of some components of the canonical Wnt signaling pathway delays skeletal muscle regeneration 11 . Inactivation of β-catenin or BCL9 inhibits the differentiation of SCs 17,18 . It is well-known that autophagy is crucial for maintaining the energy balance and stability of the cellular environment 19,20 . Interestingly, autophagy regulates SC activation and skeletal muscle regeneration 21,22 . Given this, we wanted to know whether autophagy regulates muscle regeneration by affecting the stabilization of Dvl2 and how this process is precisely controlled during skeletal muscle regeneration.
Recently, the immunoglobulin superfamily containing leucinerich repeat (Islr) gene was identified as a new marker of mesenchymal stem cells and is expressed in SCs 23 . Islr mRNA is also detected in muscle tissues 24,25 . It is well-known that Duchenne muscular dystrophy (DMD) patients and dystrophin-null (mdx) mice have continuous regeneration of muscle and activation of SCs in humans and mice, respectively [26][27][28] . Islr mRNA is highly expressed in DMD patients and mdx mice in the Gene Expression Omnibus (GEO) database, suggesting a potential role of Islr in skeletal muscle regeneration. However, the in vivo functions of Islr in skeletal muscle regeneration are entirely unknown.
In this study, we found that Islr was highly expressed in differentiated myogenic cells. Utilizing an Islr loss-of-function mouse model, we demonstrated that Islr was required for skeletal muscle regeneration. Mechanistically, Islr activated the canonical Wnt signaling pathway by antagonizing autophagy to stabilize Dvl2.

Results
Islr is highly expressed in differentiated myogenic cells. The GEO database shows that Islr mRNA is highly expressed in mdx mice, thus we validated this information and found that Islr was indeed upregulated in mdx mice ( Supplementary Fig. 1a, b), indicating that it might be involved in skeletal muscle regeneration. To examine the expression of Islr during skeletal muscle regeneration, the tibial anterior (TA) muscles were injured with an injection of cardiotoxin (CTX) and allowed to regenerate. Islr protein levels were higher in the injured than in the contralateral TA muscles (CTL) at 3 d postinjury (Supplementary Fig. 1c). The expression level of Islr increased between 3 and 5 day postinjury, which is a critical stage during which SCs participate in skeletal muscle regeneration ( Supplementary Fig. 1d). To analyze the expression of Islr during myogenesis, we carried out immunohistochemical staining for Islr, Pax7, and MyoG on serial sections of TA muscles. No Islr protein expression was detected in Pax7 + quiescent SCs (Fig. 1a). However, Islr protein was expressed in Pax7 + activated SCs and MyoG + muscle progenitors (Fig. 1b, c). A combination of cell surface markers (CD45 − , CD31 − , Sca1 − , and α7-integrin + ) is widely used to purify SCs in skeletal muscle. Specifically, we isolated SCs from wild-type (WT) mice using fluorescence-activated cell sorting (FACS) 29 (Fig. 1d). Although Islr mRNA was not significantly different between freshly isolated SCs and activated SCs cultured for 24 h, the mRNA and protein levels of Islr were higher in differentiating cells as compared to proliferating cells ( Fig. 1e and Supplementary Fig. 1e).
To localize Islr expression in C2C12 cells in vitro, we generated an Islr expression vector in which green fluorescent protein (GFP) was inserted at the C-terminus of Islr (Islr-GFP). Islr colocalized with tubulin following plasmid transfection in C2C12 cells (Fig. 1f), indicating that Islr was expressed in the cytoplasm. Meanwhile, an expression vector fusing a Flag-tag after amino acid 18 (the leader peptide) of Islr was also constructed (L-Flag-Islr), yielding the same result as the Islr-GFP plasmid (Fig. 1g). Meanwhile, we found that endogenous Islr was mainly expressed in the cytoplasm of mature myotubes (Fig. 1h, i and Supplementary Fig. 1f). Furthermore, the expression of Islr gradually increased with the differentiation of C2C12 cells (Fig. 1j,  k), which was consistent with the results of primary myoblasts derived from FACS-purified SCs. Taken together, we concluded that Islr was highly expressed in differentiated myogenic cells.
Loss of Islr delays regeneration of adult skeletal muscle. In order to study the function of Islr in vivo, we crossed mice possessing the Islr Exons1b-2-loxP conditional allele with mice expressing a Cre protein driven in Myf5 + cells (conditional knockout [cKO]) 30 (Fig. 2a). To evaluate the recombination efficiency at the protein level, we purified SCs of control (Ctrl) and Islr cKO mice using FACS and found that Islr was not expressed at all in Islr cKO mice (Fig. 2b). Furthermore, we injected CTX into the TA muscles of control and Islr cKO mice and found that Islr was markedly deleted in most SCs of Islr cKO mice at 5 d postinjury (Fig. 2c, d).
The Islr cKO mice are fertile and healthy without any apparent change in gross phenotype under normal conditions. To understand the role of Islr in skeletal muscle regeneration, we histologically analyzed the TA muscles at 3.5, 5, and 14 d postinjury. At 3.5 d postinjury, control mice already displayed newly formed muscle fibers, whereas almost no muscle fiber regeneration was observed in Islr cKO mice ( Fig. 2e and Supplementary Fig. 2a). At 5 d postinjury, control mice had regenerated a large number of muscle fibers with centrally located myonuclei, whereas the diameters of the newly formed muscle fibers were very small in Islr cKO mice ( Fig. 2e and Supplementary Fig. 2a). At 14 d postinjury, control mice displayed tightly packed, well-formed muscle fibers, whereas Islr cKO mice had many small muscle fibers with single nuclei (Fig. 2e and Supplementary Fig. 2a). Five days after injury, the average cross-sectional area (CSA) of muscle fibers decreased by about 33% in Islr cKO mice compared to the control mice (Fig. 2f). In addition, the numbers of myofibers containing two or more centrally located nuclei were also significantly reduced in Islr cKO mice (Fig. 2g). We calculated the average muscle fiber areas and found that there were a large number of muscle fibers with an area less than 400 μm 2 in Islr cKO mice, and muscle fibers with an area greater than 1000 μm 2 were almost nonexistent in Islr cKO mice (Fig. 2h). These results demonstrated that the loss of Islr delayed regeneration of adult skeletal muscle.
Islr regulates the differentiation potential of SCs. The higher expression of Islr in differentiated myogenic cells prompted us to investigate the importance of Islr in the process of SCs differentiation during skeletal muscle regeneration. There was a significant reduction in the numbers of eMyHC + muscle fibers in Islr cKO mice compared to control mice at 3.5 d postinjury. At 5 d postinjury, the size of eMyHC + fibers was significantly reduced in the regenerating TA muscles of Islr cKO mice. Statistical analysis showed that the surface area of eMyHC + muscle fibers was significantly decreased in Islr cKO mice (Fig. 3a). This suggested that the differentiation potential of SCs decreased. To further test this idea, we quantified the numbers of MyoG + cells and found a significant reduction of MyoG + cells in Islr cKO mice (Fig. 3b). Consistently, Western blotting showed that the protein levels of MyoG and eMyHC were significantly reduced in Islr cKO mice at 3.5 and 5 d postinjury (Fig. 3c, d).
To confirm the results in vitro, we isolated single myofibers from the extensor digitorum longus (EDL) muscles of control and Islr cKO mice. The number of MyoG + cells in the clusters associated with single myofibers was greatly reduced in Islr cKO mice (Fig. 4a). Next, we analyzed FACS-purified SCs from control and Islr cKO mice. Although there was no significant difference during the proliferation period ( Supplementary Fig. 2b), we found that the numbers of MyoG + cells were decreased in Islr cKO mice after inducing differentiation (Fig. 4b). Consistently, Western blotting showed that the protein level of MyoG is reduced in Islr cKO mice (Fig. 4c). Further, we found that the ability of primary myoblasts derived from FACS-purified SCs to fuse myotubes was significantly decreased, and the fusion index was decreased by about 40% in Islr cKO mice compared to controls (Fig. 4d). Statistical analysis showed that Islr cKO mice had fewer myotubes containing three or more nuclei (Fig. 4e), and the levels of MyHC protein were much lower in Islr cKO mice (Fig. 4f). To confirm these results, we established primary myoblasts of control and Islr cKO mice. Similarly, we observed that the differentiation of primary myoblasts was significantly decreased in Islr cKO mice ( Supplementary Fig. 2c, d).
SC-specific deletion of Islr impairs skeletal muscle regeneration. The Myf5-Cre driver functions in the early embryonic development, and may cause compensatory mechanisms to influence the phenotype of Islr cKO mice. To address this concern, we crossed mice possessing the Islr fl/fl conditional allele with Pax7-CreER mice to generate the Pax7-CreER:Islr fl/fl mice. The mice were induced with tamoxifen (TAM) at the age of 10 weeks, followed by CTX damage (Fig. 5a). To evaluate the recombination efficiency at the protein level, we performed immunohistochemistry and Western blot experiments, and found that Islr was markedly deleted in the SCs of Pax7-CreER:Islr fl/fl mice at 5 d postinjury (Fig. 5b, c). To test whether Pax7-CreER:Islr fl/fl mice had a phenotype consistent with Islr cKO mice, we histologically analyzed the TA muscles at 5 and 14 d postinjury. At 5 d postinjury, Islr fl/fl mice had regenerated a large number of muscle fibers with centrally located myonuclei, whereas there were many blank areas in the TA muscles of the Pax7-CreER:Islr fl/fl mice ( Fig. 5d and Supplementary Fig. 2e). Five days after injury, the average CSA of muscle fibers was significantly decreased in Pax7- CreER:Islr fl/fl mice compared to the Islr fl/fl mice (Fig. 5e). In addition, the numbers of myofibers containing two or more centrally located nuclei were also significantly reduced in Pax7-CreER:Islr fl/fl mice (Fig. 5f). Meanwhile, the size of eMyHC + fibers was very small in the regenerating TA muscles of Pax7-CreER:Islr fl/fl mice (Fig. 5g). At 14 d postinjury, Islr fl/fl mice displayed tightly packed, well-formed muscle fibers, whereas Pax7-CreER:Islr fl/fl mice had many very small muscle fibers with single nuclei (Fig. 5d, h and Supplementary Fig. 2e). Overall, the phenotype of Pax7-CreER:Islr fl/fl mice was more severe, which demonstrated that Islr had a substantial effect on skeletal muscle regeneration. Islr functionating by Dvl2-mediated canonical Wnt signaling.
In order to study the molecular mechanisms of Islr during differentiation, we generated Islr shRNA stable (shIslr) C2C12 cells.
Islr mRNA was effectively knocked down in shIslr C2C12 cells compared to control shRNA (shCtrl) C2C12 cells ( Supplementary  Fig. 3a). The shCtrl and shIslr C2C12 cells were induced to differentiate, and the shIslr C2C12 cells failed to differentiate ( Fig. 6a, b). Subsequently, we performed RNA sequencing (RNAseq) analysis on shCtrl and shIslr C2C12 cells after 2 d of culture in growth medium (G2) or 3 d of culture in differentiation medium (D3). There are 19 signaling pathways whose expression was modulated in both the G2 and D3, in which the Wnt signaling pathway was closely related to C2C12 cells differentiation ( Supplementary Fig. 3b, c). Canonical Wnt pathway receptors and downstream target genes were significantly downregulated in shIslr C2C12 cells, and genes related to myogenic development were also significantly downregulated in shIslr C2C12 cells (Fig. 6c). The downregulation of Wnt target genes and myogenic genes were also confirmed by quantitative real-time PCR (qRT-PCR) (Fig. 6d). These results suggested that the differentiation phenotype of C2C12 cells might be associated with the perturbations of canonical Wnt signaling. Coincidentally, MyoG, a muscle differentiation marker, was reduced in the shIslr C2C12 cells (Fig. 6d). This is significant because MyoG was shown to be directly regulated by the canonical Wnt signaling pathway 31,32 . Consistently, we found that the level of active β-catenin in the nucleus was significantly reduced during the differentiation of shIslr C2C12 cells (Fig. 6e). The TOP/FOP reporter assay also showed that the activity of the canonical Wnt signaling pathway was significantly decreased in the shIslr C2C12 cells (Fig. 6f). Dvl2 is an important component involved in canonical Wnt signaling 16 , and Western blotting showed that the protein levels of Dvl2 were significantly decreased during the proliferation and differentiation of the shIslr C2C12 cells (Fig. 6g, h). Confirming the results obtained with C2C12 cells, we found that the protein levels of Dvl2 were significantly decreased during the differentiation of primary myoblasts from Islr cKO mice (Fig. 6i). Wnt3a is a classical ligand of the canonical Wnt signaling pathway. Exogenous addition of Wnt3a can activate canonical Wnt signaling. Axin1 was degraded following the addition of Wnt3a in shCtrl C2C12 cells. However, Axin1 still accumulated in shIslr C2C12 cells (Fig. 7a). Meanwhile, the ability of β-catenin to enter the nucleus was significantly reduced in response to Wnt3a addition in shIslr C2C12 cells (Fig. 7b). This suggested that Wnt3a required Islr for its function. To confirm the results obtained with C2C12 cells, we injected Wnt3a into the TA muscles of control and Islr cKO mice after CTX injury, and found that Wnt3a could not effectively activate canonical Wnt signaling during skeletal muscle regeneration (Fig. 7c, d), which is consistent with the results of C2C12 cells. Furthermore, 1azakenpaullone (1-AKP), which is known to activate the canonical Wnt signaling pathway independent of Dvl2 33 , reduced levels of Axin1 and GSK3β in shIslr C2C12 cells when added to the differentiation medium (Fig. 7e). As a consequence, MyHC was expressed in 1-AKP treated shIslr C2C12 cells (Fig. 7f). To confirm these results, we used another small-molecule Wnt Error bars represent the means ± s.d. * P < 0.05, ** P < 0.01, *** P < 0.001; Student's t test agonist, CHIR-99021 (CHIR) 34 , and obtained the same results ( Supplementary Fig. 4a). We also checked the expression of MyoG in shCtrl and shIslr C2C12 cells. Not surprisingly, the numbers of MyoG + cells were significantly decreased in shIslr C2C12 cells (Supplementary Fig. 4b); meanwhile, the protein levels of MyoG were also reduced ( Supplementary Fig. 4c). We treated shCtrl and shIslr C2C12 cells with 1-AKP and CHIR, and found that both drugs rescued the MyoG + cells (Fig. 7g, h). Meanwhile, we obtained the same results from primary myoblasts of control and Islr cKO mice that we established (Fig. 7i). Dvl2 complementation rescues the phenotype of Islr cKO mice.
The aforementioned data suggested that the loss of Islr blocked the canonical Wnt signaling pathway via Dvl2 (Fig. 6g-i). In order to validate the results obtained from the C2C12 cells, we checked the expression of Dvl2 in TA muscles at 3 d postinjury and found that its expression pattern was consistent with that of Islr ( Supplementary Fig. 5a). Meanwhile, Axin1 was also downregulated, suggesting activation of the canonical Wnt signaling pathway ( Supplementary Fig. 5b). Consistently, Dvl2 was highly expressed and β-catenin translocated to nuclei in the cells where Islr was highly expressed at 5 d postinjury (Fig. 8a, b). These results demonstrated that Islr expression correlated with Dvl2. Subsequent results showed that Dvl2 could not be efficiently expressed in Islr cKO mice at 3.5 d postinjury (Fig. 8c). Meanwhile, primary myoblasts derived from FACS-purified SCs of Islr cKO mice expressed lower levels of Dvl2 and higher levels of Axin1 than control mice after 2 d of culture in differentiation medium. These results suggested that the canonical Wnt signaling pathway was significantly inhibited in Islr cKO mice (Fig. 8d). Based on the above results, we injected CHIR into the TA muscles of control and Islr cKO mice after CTX injury, and found that CHIR rescued the phenotype of Islr cKO mice (Fig. 8e, f). Cumulatively, these data suggested that Dvl2 might be modulated by Islr. It has been reported that the inhibition of Dact1 upregulates the expression of Dvl2 and activates canonical Wnt signaling 35 . Therefore, we generated Islr/Dact1 cKO double-mutant mice. Intriguingly, the expression level of Dvl2 was rescued in Islr/ Dact1 cKO mice (Fig. 8g). Moreover, canonical Wnt signaling was reactivated in Islr/Dact1 cKO mice (Fig. 8h). As a consequence, the deletion of Dact1 rescued the phenotype of Islr cKO mice (Fig. 8i). The CSA of Islr/Dact1 cKO mice was significantly higher than in Islr cKO mice at 5 d postinjury (Fig. 8j). Meanwhile, the number of MyoG + cells was restored in Islr cKO mice (Fig. 8k). Taken together, the in vivo and in vitro data supported the notion that Islr regulated canonical Wnt signaling via modulation of Dvl2.
Islr stabilizes Dvl2 by antagonizing the autophagy system. The autophagy system regulates canonical Wnt signaling via modulation of Dvl2 levels 15 . We thus detected the occurrence of autophagy in C2C12 cells using a GFP-LC3 (a well-known marker of autophagosomes) plasmid 36,37 . Overexpression of Dvl2-Flag increased the formation of GFP-LC3 + puncta in C2C12 cells. However, co-overexpression of Islr-HA and Dvl2-Flag decreased GFP-LC3 + puncta formation compared to overexpression of Dvl2-Flag alone (Fig. 9a). Moreover, cooverexpression of Islr-GFP and Dvl2-Flag decreased the formation of Dvl2-Flag + puncta compared to overexpression of Dvl2-Flag alone in HEK293T cells ( Supplementary Fig. 6a). We found that the autophagy markers LC3II and p62 were significantly increased and decreased, respectively, in shIslr C2C12 cells (Fig. 9b). Furthermore, overexpression of Dvl2-Flag in shIslr C2C12 cells displayed more GFP-LC3 + puncta (Fig. 9c). Moreover, rapamycin (Rap), a well-known autophagy-inducing drug, was used to induce cell autophagy in shCtrl and shIslr C2C12 cells 38,39 . Endogenous Dvl2 formed more GFP-LC3 + puncta in shIslr C2C12 cells (Fig. 9d). To confirm this, we purified SCs from control, Islr cKO, Dact1 cKO, and Islr/Dact1 cKO mice. After rapamycin treatment, endogenous Dvl2 in primary myoblasts derived from FACS-purified SCs of Islr cKO mice formed more GFP-LC3 + puncta, but this was not observed in the Dact1 cKO mice. Meanwhile, GFP-LC3 + puncta were significantly reduced in Islr/Dact1 cKO mice (Fig. 9e), which confirmed the previous experiments performed in C2C12 cells. From above results, it was clear that the deletion of Islr increased the autophagy-mediated Dvl2 degradation. Therefore, we used bafilomycin A1 (BFA1) to treat primary myoblasts of control and Islr cKO mice that we established and found that Dvl2 and Wnt target genes Axin2 and Lef1 were rescued in primary myoblasts of Islr cKO mice (Fig. 9f). Meanwhile, the disruption of differentiation was rescued in the primary myoblasts of Islr cKO mice when BFA1 was added to the differentiation medium ( Supplementary Fig. 6b). To further determine whether Islr interacted with Dvl2 to antagonize the autophagy system, we performed immunoprecipitation analysis and found that Islr-GFP interacted with Dvl2-Flag ( Fig. 9g and Supplementary Fig. 6c). Meanwhile, we also found that Islr interacted Dvl2 in differentiated primary myoblasts (Fig. 9h). Dvl2 is ubiquitinated before entering the phagophore 15 . The ubiquitination level of Dvl2 was significantly increased in differentiated primary myoblasts of Islr cKO mice, indicating that Islr helped Dvl2 resist ubiquitination (Fig. 9i). Cumulatively, these results indicated that Islr stabilized Dvl2 protein by antagonizing the autophagy system.

Discussion
Our study found that Islr participated in regulating SC fate and skeletal muscle regeneration. The loss of Islr delayed skeletal muscle regeneration in Myf5-Cre:Islr fl/fl mice. Another SCspecific knockout (Pax7-CreER:Islr fl/fl ) mouse model confirmed these data with an even stronger phenotype. The SC differentiation capacity of Islr cKO mice was significantly decreased as a result of the inactivation of canonical Wnt signaling, in which Islr stabilized Dvl2 by antagonizing the autophagy system. Fig. 7 The failure of Islr-deficient myoblast differentiation is rescued by canonical Wnt signaling activation. a Western blot analysis of Axin1 protein levels in shCtrl and shIslr C2C12 cells incubated with Wnt3a or saline for 24 h after 2 d in differentiation medium. b Western blot analysis of active β-catenin protein levels in nuclear lysates extracted from shCtrl and shIslr C2C12 cells incubated with Wnt3a or saline for 24 h after 2 d in differentiation medium. Although we also found that Islr mRNA is expressed in quiescent SCs, similar to a recent report 23 , we found that Islr protein was not expressed in Pax7 + quiescent SCs. Islr was expressed in Pax7 + activated SCs and more highly expressed in differentiated myogenic cells with the progression of myogenesis, suggesting a potential role for Islr in myogenic differentiation. Supporting this idea, we found that C2C12 cells stopped differentiating into myotubes when disrupting the expression of Islr, and that the ability of myoblasts to differentiate in Islr cKO mice was also significantly decreased, as evidenced by the reduction of MyoG in both in vivo and in vitro experiments. It is well-known that MyoG is a key factor required for SC differentiation and myocyte fusion 40,41 . Skeletal muscle fails to develop normally in the absence of MyoG 8,9 . Taken together, our findings suggest that Islr regulates the differentiation of SCs and C2C12 cells. Although Islr was downregulated during mesenchymal stem cell differentiation into bone and adipose tissues 23 , there were many differences in the expressions of genes between these developing tissues and skeletal muscle. It is well-known that bone morphogenetic proteins are expressed in different patterns during bone formation and myogenic differentiation [42][43][44] .
Many reports have shown that the differentiation of C2C12 cells is mainly regulated by canonical Wnt signaling which can directly regulate the expression of MyoD and MyoG 31  directly binds the promoter of MyoG in conjunction with βcatenin and TCF4 to regulate C2C12 cell differentiation 32 . Consistent with previous reports, we found that the activity of the canonical Wnt signaling pathway was significantly perturbed in shIslr C2C12 cells, resulting in defective differentiation. Mechanistically, Islr interacted with and stabilized Dvl2 to activate the canonical Wnt signaling pathway. Dvl2 is an important factor required for canonical Wnt signaling 16,45 , and a reduction in Dvl2 expression abrogates canonical Wnt signaling 46 . Confirming this, activation of canonical Wnt signaling independent of Dvl2 was able to rescue the impaired skeletal muscle regeneration phenotype and increase the numbers of MyoG + cells both in the shIslr C2C12 cells and primary myoblasts of Islr cKO mice. Further, the deletion of Dact1 also rescued the impaired differentiation phenotype of Islr cKO mice by upregulating Dvl2 in vivo. In brief, Islr regulated Dvl2-mediated canonical Wnt signaling and further regulated SC differentiation and skeletal muscle regeneration. Additionally, because Dvl2 has some functions in noncanonical Wnt signaling during skeletal muscle regeneration 11 , the potential effect of Islr on other members of the Wnt family, including noncanonical Wnt signaling warrant future investigations.
Autophagy is associated with many important signaling pathways such as Wnt, Shh (Sonic Hedgehog), transforming growth factor β, and fibroblast growth factor (FGF) 47,48 . A key factor is that LC3 targets Dvl2, thus regulating the activity of the canonical Wnt signaling pathway through the autophagy system 15 . In our experiments, we found that Islr stabilized Dvl2 by inhibiting autophagy. GFP-LC3 + /Dvl2-Flag + puncta formed when the Dvl2-Flag plasmid was transfected into C2C12 cells. However, GFP-LC3 + /Dvl2-Flag + puncta could not form and Dvl2-Flag was uniformly dispersed in the cytoplasm when Dvl2-Flag and Islr-HA were co-transfected. These findings suggested that Islr antagonized the autophagy system, consequently protecting Dvl2 against degradation. Furthermore, we demonstrated that endogenous Dvl2 was more easily identified by LC3 both in shIslr C2C12 cells and Islr-deficient primary myoblasts derived from FACS-purified SCs when autophagy was induced.
Previous reports have shown that Dact1 can induce autophagy of Dvl2 and further inhibit canonical Wnt signaling 35,49 . In agreement with this, we found that the deletion of Dact1 inhibited the formation of Dvl2 + autophagic corpuscles in Islr-deficient primary myoblasts derived from FACS-purified SCs. BFA1, an inhibitor of autophagy, rescued Dvl2 protein level in primary myoblasts of Islr cKO mice. Reciprocal co-immunoprecipitation analysis demonstrated that Islr interacted with Dvl2 to antagonize the ubiquitination of Dvl2. It will be interesting to investigate how Islr affects the ubiquitination of Dvl2.
In conclusion, we found that Islr played an important role in the differentiation of SCs during skeletal muscle regeneration. At the molecular level, Islr protected Dvl2 against autophagymediated degradation, consequently activating the canonical Wnt signaling pathway (Fig. 10). Skeletal muscle injury and paraffin sections preparation. To induce injury, 100 μl of cardiotoxin (CTX, 10 μM in PBS, Sigma) was injected into hindlimb TA muscles. All CTX injured mice were 10 weeks old. The TA muscles were harvested at 3.5, 5, and 14 d postinjury. The TA muscles were fixed with 4% paraformaldehyde for more than 72 h and then subjected to dehydration embedding. Finally, paraffin sections of the muscles were obtained at a thickness of 2-4 μm.
Primary myoblast purification and culture. Primary myoblasts of control and Islr cKO mice were established by the method of differential adhesion. We cultured primary myoblasts in Ham's F10 nutrient mixture medium (Gibco) containing 20% FBS and 5 ng/ml bFGF. To induce differentiation, primary myoblasts were incubated with DMEM supplemented with 2% HS.
Single myofiber isolation and culture. The EDL muscles were isolated and digested with 0.2% type I collagenase (Sigma) in DMEM for 1 h. Single myofibers Plasmids and RNA interference. In order to construct a plasmid encoding Dvl2-Flag, the Flag-tag was added at the C-terminus of Dvl2. Dvl2-Flag was then cloned into the BamHI and EcoRI sites of the pcDNA3.1(+) vector (Invitrogen). In order to construct a plasmid encoding Islr-HA, the HA tag was added at the C-terminus of Islr. Islr-HA was then cloned into the BamHI and EcoRI sites of the pcDNA3.1 (+) vector. In order to construct a plasmid encoding leader peptide-Flag-Islr, the Flag-tag was added after amino acid 18 (the leader peptide of Islr). The leader peptide-Flag-Islr was then cloned into the BamHI and EcoRI sites of the pcDNA3.1 (+) vector. The GFP-LC3B plasmid was provided by Dr. Yeguang Chen (State Key Laboratory of Biomembrane and Membrane Biotechnology, Tsinghua University). The TOPflash/FOPflash and Renilla luciferase expression plasmids were kept in our lab. Islr-GFP (MG206818), scrambled shRNA, and Islr shRNAs (TG512833) were purchased from OriGene. The sequences of Islr shRNAs were as follows: Islr shRNA1, 5′-cctcagcggtgtctattccagaacaggac-3′; Islr shRNA2, 5′-gccaatgtgaccacactgagcctgtcagc-3′; Islr shRNA3, 5′-ggcttcgtgctggcactccactgtgacgt-3′; and Islr shRNA4, 5′-ggctgctatacggttgacaacgaggtaca-3′.
Cell lines and Islr shRNA stable (shIslr) C2C12 cells. C2C12 and HEK293T cell lines were purchased from the Chinese Academy of Medical Sciences and the School of Basic Medicine at Peking Union Medical College. To obtain stably expressing control shRNA (shCtrl) and shIslr C2C12 cells, we transfected C2C12 cells with scrambled shRNA plasmid or Islr shRNA3 plasmid via electric transfection (Lonza). At 24 h after transfection, the cells were diluted and cultured with puromycin (2 μg/ml, Sigma) for resistance screening for 1 week. The clones were then transferred to a 96-well plate to expand the culture. Finally, transfection was confirmed by PCR, and the primers used were as follows: forward: 5′-gaagagg gcctatttcccat-3′, reverse: 5′-gccttccatctgttgctgcg-3′. The efficacy of Islr shRNA interference was determined by quantitative real-time PCR (qRT-PCR) and Western blotting. Immunohistochemistry. Muscle sections were fixed in 4% paraformaldehyde and permeabilized for 10 min in 0.3% Triton X-100 in PBS. The samples were treated with 3% H 2 O 2 for 20 min and blocked in blocking buffer (ZSGB-BIO) for 2 h at room temperature. Primary antibodies were applied in blocking buffer at 4°C overnight. Subsequently, the samples were washed with PBS and stained with the appropriate biotin-labeled secondary antibody (goat anti-rabbit or anti-mouse) for 1 h at room temperature. After washing with PBS, the samples were stained with horseradish peroxidase (HRP)-labeled streptavidin for 1 h at room temperature. Finally, DAB (ZSGB-BIO) was used for color development. Samples were then washed with water, and hematoxylin was used to stain the nuclei for 5 min. The primary antibodies used were as follows: rabbit anti-Islr (cat. no. HPA050811, Sigma, 1:300), mouse anti-Pax7 (DSHB, 1:50), mouse anti-MyoG (cat. no. ab1835, Abcam, 1:200), rabbit anti-β-catenin (cat. no. 9562s, Cell Signaling Technology, 1:200), and rabbit anti-Dvl2 (cat. no. PA5-27965, Thermo Fisher, 1:200).
Plasmid transfection and luciferase reporter assay. Scrambled shRNA, Islr shRNA, Dvl2-Flag, Islr-GFP, Islr-HA, or GFP-LC3B plasmids were transfected into C2C12 cells, HEK293T cells, or SCs using Lipofectamine 2000 (Invitrogen). The samples were then harvested between 24 and 36 h after transfection. In order to identify the activity of canonical Wnt signaling, the TOPflash/FOPflash expression plasmids with the Renilla luciferase expression plasmid were transfected when shCtrl and shIslr C2C12 cells were cultured after 2 d in the differentiation medium. After 24 h, the reporter activity was measured using the Dual-luciferase Reporter Assay System (Promega).

Normal regeneration
Impaired regeneration  RNA-seq. The total RNA was extracted from shCtrl and shIslr C2C12 cells using an RNeasy Mini Kit (Qiagen) and on-column DNase digestion (RNase-Free DNase Set, Qiagen) to avoid contamination by genomic DNA. The cDNA library construction, sequencing, and transcriptome data analysis were performed by Gene Denovo Biotechnology Co., Ltd. (Guangzhou, China).
Autophagy analysis. C2C12 cells or SCs were transfected with GFP-LC3B plasmid. After 24 h, cells were treated with rapamycin (4 μM, Sigma) for 6 h to induce autophagy, followed by immunofluorescence analysis.
Treatment with reagents in cell culture. In order to verify the activity of canonical Wnt signaling, shCtrl and shIslr C2C12 cells were treated with either Wnt3a (100 ng/ml, Peprotech) or saline for 24 h after 2 d in differentiation medium. The cell samples were then collected for Western blot analysis. In order to activate canonical Wnt signaling in shIslr C2C12 cells, the shIslr C2C12 cells were treated with either 1-azakenpaullone (1-AKP; 3 μM, Sigma) or CHIR-99021 (CHIR; 3 μM, Stemgent) during differentiation. Western blotting and immunofluorescence analysis were performed after 3 or 7 d of differentiation. To verify the activity of canonical Wnt signaling in vivo, the TA muscles of control and Islr cKO mice were injected with Wnt3a (100 ng/ml, Peprotech) or PBS (0.1% BSA) at 1.5 d postinjury. The injured TA muscles were then collected for Western blot analysis at 4 d postinjury. To activate canonical Wnt signaling in vivo, the TA muscles of control and Islr cKO mice were injected with CHIR-99021 (50 ng/ml, Stemgent) or PBS (0.1% BSA) at 2.5 d postinjury. The injured TA muscles were then collected for H&E analysis at 5 d postinjury. To inhibit autophagy, the primary myoblasts of control and Islr cKO mice that we established were treated with bafilomycin A1 (BFA1; 200 nM, Sigma) for 6 h. Then cell samples were collected for Western blot analysis.
Statistical analysis. All experiments included at least three biological replicates. The areas of fluorescent staining for muscle fibers and nuclei were determined using ImageJ software. Data were analyzed using Microsoft Excel. A two-tailed Student's t test was used to assess statistical significance.

Data availability
RNA-seq data used in this study have been deposited in the National Center for Biotechnology Information Sequence Red Archive (SRA) under the accession code SRP160431. All other data are available within the Article and its Supplementary Files, or available from the authors upon request.