Forkhead box M1 (FoxM1) transcriptional factor has a principal role in regulating cell proliferation, self-renewal, and tumorigenesis. However, whether FoxM1 regulates endogenous muscle development and regeneration remains unclear. Here we found that loss of FoxM1 in muscle satellite cells (SCs) resulted in muscle atrophy and defective muscle regeneration. FoxM1 functioned as a direct transcription activator of adenomatous polyposis coli (Apc), preventing hyperactivation of wnt/β-catenin signaling during muscle regeneration. FoxM1 overexpression in SCs promoted myogenesis but impaired muscle regeneration as a result of spontaneous activation and exhaustion of SCs by transcriptional regulation of Cyclin B1 (Ccnb1). The E3 ubiquitin ligase Cdh1 (also termed Fzr1) was required for FoxM1 ubiquitylation and subsequent degradation. Loss of Cdh1 promoted quiescent SCs to enter into the cell cycle and the SC pool was depleted by serial muscle injuries. Haploinsufficiency of FoxM1 ameliorated muscle regeneration of Cdh1 knock-out mice. These data demonstrate that the Cdh1–FoxM1–Apc axis functions as a key regulator of muscle development and regeneration.
Skeletal muscle is a homeostatic tissue that requires myogenesis upon postnatal growth and muscle regeneration following an injury1. Myogenesis is a tightly controlled process including muscle stem cells (also called muscle satellite cells (SCs)), which proliferate, differentiate to myoblasts, and subsequently form multinucleated myotubes2,3. A portion of activated SCs return to quiescence and replenish the SC pool4. The balanced transition of SCs between quiescence, proliferation, and differentiation is critical for SC maintenance and muscle homeostasis.
Muscle-specific transcriptional factors, including Pax3, Pax7, Mfy5, MyoD, and Myogenin, play critical roles in the determination of the fate of SCs, as well as in muscle development and regeneration5. Several key components of SC niche, such as laminin-α1 and collagen VI, are necessary for self-renewal of SCs and muscle regeneration6,7. Notch signaling, as well as p38/mitogen-activated protein kinase, wnt, phosphatase and tensin homolog, and several cell cycle regulators have been shown to regulate SC expansion and muscle development8–11. Recently, the ubiquitin–proteasome system has been demonstrated to be essential for the maintenance of SCs. Knockout of the Rp13 proteasomal component impairs the ability of SCs to proliferate, survive, and differentiate, resulting in defective muscle regeneration12. The anaphase-promoting complex/cyclosome (APC/C) is a multisubunit E3 ubiquitin ligase complex that regulates cyclic degradation of cell cycle regulator via the Cdh1 or Cdc20 adaptor proteins13. APC/CCdh1 has a broad spectrum of substrates in and beyond the cell cycle14,15. However, the role of APC/CCdh1 in SC maintenance is unclear.
Forkhead box M1 (FoxM1) is one of the substrates of APC/CCdh1 in several cancer cells16,17 and is a well-established regulatory transcriptional factor for cell cycle progression18. FoxM1 has been reported to play critical roles in stem cells19,20. FoxM1 is essential for proliferation of human embryonic stem cells and protects them against oxidative stress by transcriptionally regulating Cyclin B1 (Ccnb1) and cyclin-dependent kinase 1 (Cdk1)19. The higher expression of FoxM1 is essential for the maintenance of hematopoietic stem cells by transcriptionally regulating Nurr120. We previously described that FoxM1 promotes the proliferation and survival of C2C12 cells (an immortalized myoblast cell line) in vitro partially by the involvement of long noncoding RNAs21. FoxM1 expression displays dynamic changes during the transition of C2C12 between quiescence and activation21, implying the important roles of FoxM1 in SCs. However, the precise functions of FoxM1 in SCs and whether FoxM1 protein is a substrate of APC/CCdh1 are largely unknown.
Besides its transcriptional regulation of target genes, FoxM1 has been reported to directly interact with proteins, such as Npm122, Rnf16823, Melk24, and beta (β)-catenin25,26. β-Catenin is the core factor of the wnt pathway. Degradation of β-catenin is promoted by a cytoplasmic destruction complex consisting of adenomatous polyposis coli (Apc), Axin, Cki, and Gsk3β, which phosphorylates β-catenin27. Conversely, FoxM1 binds directly to β-catenin and enhances β-catenin nuclear localization in glioma cell lines25 and pancreatic cancer cells26. The constitutive activation of β-catenin in mice has no obvious effect on the amount of SCs under homeostatic conditions, but these mice show muscle regeneration defects with aberrant myofiber morphology and atrophy, which is mainly attributed to hyperactive transforming growth factor-beta (TGF-β) signaling28. However, little is about the role of FoxM1 in muscle regeneration and the correlation between FoxM1 and wnt signaling in SCs.
In this study, we demonstrate that the loss of FoxM1 results in muscle atrophy and damaged muscle regeneration. Both FoxM1 deficiency and overexpression impaired SC maintenance. Unexpectedly, FoxM1 promoted β-catenin degradation by transcriptionally regulating Apc. Moreover, the accumulation of FoxM1 protein was controlled by Cdh1 via ubiquitylation. The identified Cdh1–FoxM1–Apc regulatory axis provides new insight on the regulators of SC maintenance and muscle homeostasis.
To determine whether FoxM1 regulates muscle homeostasis, we crossed mice with loxP-flanked FoxM1 alleles (FoxM1fl/fl)29 with Pax7-Cre mice30 to generate control FoxM1fl/fl mice and FoxM1fl/flPax7-Cre mice (designated here as “FoxM1-cKO mice”) (Supplementary Fig. S1a, b). The expression of FoxM1 in SCs was efficiently knocked out (Supplementary Fig. 1c, d). The weight of tibialis anterior (TA) muscle in FoxM1-cKO mice showed no obvious differences with FoxM1fl/fl mice at 2 months of age but showed decreased weight at 8 months of age (Fig. 1a, b and Supplementary Fig. 1e). Moreover, the myofibers of extensor digitorum longus (EDL) in FoxM1fl/fl-cKO mice showed reduced numbers of myonuclei/myofiber than FoxM1fl/fl mice at 8 months of age (Fig. 1c, d and Supplementary Fig. 1f). Histological analysis of TA revealed muscle atrophy with age in FoxM1-cKO mice compared with control mice (Fig. 1e, f and Supplementary Fig. 1g). FoxM1-cKO mice were inferior in the maximum running distance compared with control mice at 8 months of age (Fig. 1g and Supplementary Fig. 1h). These data suggested that loss of FoxM1 in SCs resulted in muscle loss with age.
To explore the effect of FoxM1 deficiency on muscle regeneration, we induced muscle injury by injecting BaCl2 into muscles of mice at 2 months of age (Fig. 1h) Histological analysis of the TA muscles at 7 days and 14 days after injury revealed a more severe regeneration defect in FoxM1-cKO mice, as evidenced by the larger unrepaired areas (Fig. 1i). Deletion of FoxM1 in SCs resulted in smaller-sized regenerative myofibers, compared with control mice at 14 days post-injury (DPI) (Fig. 1j). Together, these data suggested that loss of FoxM1 in SCs resulted in muscle atrophy and impaired muscle regeneration.
The finding of decreased muscle mass in FoxM1-cKO mice prompted us to examine the abundance of SCs in the skeletal muscle. Since SCs have definite cell surface markers (defined as CD45−Sca1−CD11b−CD31−CD34+α7-integrin+)31, we utilized flow cytometry to analyze the SC pool. FoxM1 deletion had no obvious effect on the abundance of SCs in 2-month-old mice (Supplementary Fig. 2a) but considerably reduced SCs abundance in 8-month-old mice (Fig. 2a, b). Immunostaining revealed considerably fewer numbers of Pax7+ SCs per fiber in 8-month-old FoxM1-cKO mice than in their control littermates (Fig. 2c, d). These data suggested that loss of FoxM1 impaired SC maintenance.
We analyzed the cell cycle of SCs and observed a significantly increased frequency of quiescent SCs in FoxM1-cKO mice (Fig. 2e, f). We also found a decreased percentage of S phase SCs in FoxM1-cKO mice compared with control mice (Fig. 2g, h). In vitro analysis showed that deletion of FoxM1 inhibited the incorporation of bromodeoxyuridine (BrdU) and expansion of SCs (Supplementary Fig. 2b, c and Fig. 2i). Loss of FoxM1 may result in cell apoptosis. Deletion of FoxM1 resulted in a slight increase in apoptosis (Fig. 2j). The differentiation potential of SCs is vital for muscle regeneration. However, we found that the loss of FoxM1 had no obvious effect on SC differentiation (Supplementary Fig. 2d–f). Together, these data suggested that FoxM1 was essential for SC maintenance by regulating the proliferation and survival of SCs.
We performed global gene expression profiling of SCs isolated from 2-month-old FoxM1-cKO mice that were uninjured or 48 h post-injury. We found that 141 genes had over twofold differential expression in FoxM1-cKO SCs relative to their expression in FoxM1fl/fl SCs in uninjured mice, In injured mice, 515 genes were differentially expressed (Fig. 3a, b). Notably, cell cycle regulation and the ribosome pathway were the most commonly modified cellular function pathways in FoxM1-cKO SCs under homeostatic status, and p53, Foxo, and wnt pathways were enriched in injured mice (Fig. 3c, d). The differential expression of 28 genes, including the cell cycle regulator, Ccnb1, occurred in both the homeostatic and injured conditions (Fig. 3e). These results revealed that FoxM1 promoted cell cycling of SCs at the molecular level.
We also found genes of wnt signaling pathway that displayed downregulated expression in SCs in injured mice. These genes included Apc, Gsk3β, Axin1, and Skp1a (Fig. 3e). Wnt signaling plays an essential role during embryonic muscle development and in the maintenance of skeletal muscle homeostasis32,33. Quantitative real-time polymerase chain reaction (qPCR) results confirmed the significant downregulation of the aforementioned genes (Fig. 3f). The expression of Apc showed similar pattern of change in SCs during muscle regeneration (Supplementary Fig. 2g, h). Moreover, the protein level of Apc in SCs in vivo was reduced in FoxM1-cKO mice compared with control mice (Fig. 3h, i). The cultured FoxM1-deficient SCs displayed increased total and nuclear β-catenin proteins compared with control SCs in vitro (Fig. 3g). Consistently, the protein level of β-catenin in SCs in vivo was upregulated in FoxM1-deficient mice compared with control mice (Fig. 3j, k). The expression of Tgf-β2 and Tgf-β3 was also upregulated in FoxM1-deficient SCs in the injured condition (Supplementary Fig. 2i, j). Together, these data suggested that loss of FoxM1 resulted in activation of wnt/β-catenin signaling in SCs.
To determine whether FoxM1 regulates Apc through direct transcriptional activation, we searched for the consensus FoxM1-binding site (C/T)AAA(C/T)AA in the proximal promoter regions of Apc. A chromatin immunoprecipitation (ChIP) assay of SCs revealed the direct binding of FoxM1 to site 1 in the Apc promoter regions (Fig. 4a, b). The luciferase activity of a construct containing the wild-type (WT)-binding site showed apparent increases with FoxM1 expression. However, the construct containing the mutants did not (Fig. 4c, d). We generated mice with FoxM1 overexpression in SCs (designated as “Tg-FoxM1” mice here). The strategy efficiently promoted the overexpression of FoxM1 in SCs (Supplementary Fig. 3a–c). Inspiringly, FoxM1 overexpression significantly upregulated the expression of Apc in SCs (Fig. 4e), as well as the protein level of Apc (Fig. 4f, g). Subsequently, the total protein level of β-catenin was strikingly reduced (Fig. 4h). Loss- and gain-of-function analyses demonstrated a novel mechanism in which FoxM1 repressed wnt/β-catenin signaling by transcriptionally regulating Apc in SCs.
To determine whether the downregulated expression of Apc hindered muscle regeneration of FoxM1-cKO mice by the Apc-β-catenin axis, the mice were administered ICG-001, a small molecule inhibitor of wnt/β-catenin34. Treatment with ICG-001 partially rescued the regeneration capacity of FoxM1-cKO SCs, as shown by larger cross-sectional area (CSA), normal regenerative myofiber, and enhanced exercise performance (Fig. 4i–l). These data suggested that the regeneration defect of FoxM1-deficient SCs was partially due to the hyperactivation of wnt/β-catenin signaling resulting from reduced Apc expression.
We observed obvious muscle hypertrophy in Tg-FoxM1 mice compared with 2-month-old WT mice (Supplementary Fig. 3d and Fig. 5a, b). However, FoxM1 overexpression in SCs eventually resulted in striking muscle atrophy at 8 months of age (Supplementary Fig. 3e–i). Moreover, the regenerative myofibers of 2-month-old Tg-FoxM1 mice were significantly smaller than those in WT mice (Fig. 5c, d). These data indicated that FoxM1 overexpression in SCs promoted myogenesis at an early age but impaired muscle regeneration under stress and muscle development of mice with increasing age.
The transition of muscle hypertrophy to atrophy implied the exhaustion of SC pool in Tg-FoxM1 mice. We observed a decrease in the SC abundance in Tg-FoxM1 mice compared with WT mice at 2 months of age (Fig. 5e, f). The difference was more striking at 8 months of age and at 30 DPI (Supplementary Fig. 3j–l). The numbers of Pax7+ SCs in single fibers in Tg-FoxM1 mice were also significantly reduced compared to WT mice at 8 months of age (Fig. 5g, h). These data suggested that FoxM1 overexpression resulted in SC exhaustion, which helped explain the eventual muscle atrophy and the impaired muscle regeneration in Tg-FoxM1 mice.
FoxM1 overexpression promoted the entry of quiescent SCs into the cell cycle (Fig. 5i, j and Supplementary Fig. 3m, n). Flow cytometric analysis revealed that FoxM1 overexpression had no obvious effect on apoptosis of SCs (Fig. 5k). FoxM1 reportedly regulates a subset of G1–G2 and G2–M transition genes. Ccnb1 showed downregulated expression in FoxM1-deficient SCs compared with control SCs (Supplementary Fig. 4a). FoxM1 overexpression significantly upregulated the expression of Ccnb1 in SCs (Supplementary Fig. 4a). The expression of Ccnb1 was strikingly induced by muscle injury (Supplementary Fig. 4b). Moreover, the protein level of Ccnb1 in SCs showed a converse change in FoxM1-cKO and Tg-FoxM1 compared to control mice in vivo (Supplementary Fig. 4c, d) and in vitro (Supplementary Fig. 4e). ChIP and luciferase reporter assays demonstrated that FoxM1 transcriptionally regulated Ccnb1 expression (Supplementary Fig. 4f–h).
Ccnb1 complexed with Cdk1 is one of the key mitotic kinases, and Cdk1/Ccnb1 governs the entry into mitosis from the G2 phase of the cell cycle35,36. We cultured SCs of FoxM1-cKO mice in vitro and enforced Ccnb1 expression in these SCs. Enforced expression of Ccnb1 partially rescued BrdU incorporation in FoxM1-deficient SCs (Supplementary Fig. 4i, j). SCs cultured in vitro were treated with RO-3306, an inhibitor of Cdk1/Ccnb137. RO-3306 treatment in vitro abolished the increased BrdU incorporation of FoxM1-overexpressing SCs (Supplementary Fig. 4i, k). Although the injection of RO-3306 in vivo affected all muscle cells, RO-3306 treatment efficiently restored the frequency of quiescent SCs in Tg-FoxM1 mice and rescued the decreased SC pool (Supplementary Fig. 5a–d). Treatment with RO-3306 had no obvious effect on the numbers of Pax7+ SCs but considerably recovered the decreased SC pool in Tg-FoxM1 mice (Supplementary Fig. 5e). The treatment of RO-3306 also abolished muscle hypertrophy in 2-month-old Tg-FoxM1 mice (Supplementary Fig. 5f, g). Since Cdk1 mediates the phosphorylation of FoxM1, it is possible that the inhibition of Cdk1/Ccnb1 eventually affected the phosphorylation of FoxM1 and its transcriptional activity. Together, these data demonstrated that FoxM1 promoted cell cycling of SCs partially by transcriptionally regulating Ccnb1.
The APC/C is a multisubunit E3 ubiquitin ligase complex that regulates cyclic degradation of cell cycle regulator via Cdh1 or Cdc20 adaptor proteins38. We knocked down the expression of Cdh1 or Cdc20 by two independent short hairpin RNAs (shRNAs) in C2C12 cells. Reduction of Cdh1, but not Cdc20, resulted in increased FoxM1 protein in proliferative C2C12 cells (Fig. 6a). Knockdown or knockout of FoxM1 had no obvious effect on the expression of Cdh1 and Cdc20 in C2C12 or SCs (Supplementary Fig. 6a, b). Furthermore, co-immunoprecipitation assay revealed the direct interaction between FoxM1 and Cdh1, but not Cdc20, in SCs (Fig. 6b and Supplementary Fig. 6c). APC/CCdh1 induces the degradation of its substrates through ubiquitin-dependent proteolysis39. As expected, knockdown of Cdh1 reduced poly-ubiquitylation of FoxM1 in C2C12 cells (Fig. 6c, d). These data suggested that APC/CCdh1 promoted FoxM1 degradation through the ubiquitin–proteasome pathway in C2C12 cells in vitro.
We generated the conditional knockout mouse Cdh1fl/flPax7-Cre (designated as Cdh1-cKO) and its control (Cdh1fl/fl) (Supplementary Fig. 6e, f). The FoxM1 protein level in Cdh1-cKO SCs was increased compared to Cdh1fl/fl SCs in vitro (Fig. 6e). The mRNA and protein levels of Cdh1 showed peaks at 7 DPI (Supplementary Fig. 6d). The protein level of FoxM1 in SCs was strikingly elevated in Cdh1-cKO mice (Fig. 6f, g). Together, these data suggested that Cdh1 prevented accumulation of FoxM1 protein by ubiquitylation.
Knockout of Cdh1 evidently decreased the percentage of quiescent SCs (Fig. 7a, b) and depleted the SC pool at 2 months of age (Fig. 7d, e). Moreover, we detected muscle hypertrophy in Cdh1-cKO mice compared to Cdh1fl/fl mice at the same age (Fig. 7c). We generated Cdh1fl/fl, FoxM1fl/+, and Pax7-Cre mice to test whether haploinsufficiency of FoxM1 could rescue the exhaustion of SCs in Cdh1-cKO mice. Deletion of Cdh1 resulted in a remarkable reduction of SCs at 30 DPI, while haploinsufficiency of FoxM1 partially restored the quantity of SCs (Fig. 7f). Consistently, serial injuries resulted in a severe regeneration defect in Cdh1-cKO mice, but haploinsufficiency of FoxM1 improved the muscle regeneration of Cdh1-deficient mice (Fig. 7g, h). Together, these data suggested that Cdh1 prevented FoxM1 accumulation to maintain muscle homeostasis.
FoxM1 is a well-established regulatory transcriptional factor for cell cycle progression40,41. Recently, increasing evidence has suggested the pleiotropic role of FoxM1 in stem cells19,20. In this study, we identified previously unrecognized roles of FoxM1 in muscle homeostasis and SC maintenance. We found that FoxM1 promoted cell cycling of SCs partially by transcriptionally regulating Ccnb1. FoxM1 also preserved regenerative potential of SCs by transcriptionally regulating Apc to inhibit wnt/β-catenin signaling. The degradation of FoxM1protein is regulated by APC/CCdh1 via ubiquitylation.
FoxM1 acts as a cell cycle regulator and induces cell proliferation42–44. In this study, we found that loss of FoxM1 in SCs resulted in decreased SC pool with age, and the percentage of quiescent SCs considerably reduced. However, overexpression of FoxM1 evidently promoted cell cycle entry of SCs and enhanced proliferative capacity of SCs. It is possible that the quiescence exit and the increased proliferation of SCs in Tg-FoxM1 mice in combination results in SC exhaustion with age or under stress. Muscle development relies on proliferation and differentiation potential of SCs. We found that loss of FoxM1 had no obvious effect on SC differentiation. Thus the eventual muscle atrophy observed in FoxM1-cKO and Tg-FoxM1 mice could be attributed to the altered cell cycle of SCs and the reduced SC pool.
FoxM1 has been reported to regulate a subset of G1–G2 and G2–M transition genes in stage- and context-dependent manners42–44. Ccnb1 is a transcriptional target of FoxM1 in human hepatocellular carcinoma cells45 and human hTERT RPE-1 cells46. Ccnb1 binds to its partner Cdk1, and Cdk1/Ccnb1 is crucial for G2–M transition of the cell cycle47. Cdk1 can directly phosphorylate FoxM1 to increase its transcriptional activity46,48. Presently, FoxM1 directly regulated Ccnb1 expression by binding to its proximal promoter region. Inhibition of Cdk1/Ccnb1 partially rescued the loss of quiescent SCs in FoxM1-overexpressing mice. Thus we consolidated the reported mechanisms of FoxM1 transcriptionally regulating Ccnb1. We also found that ribosome-related genes were downregulated in FoxM1-deficient SCs, which helped explain the phenomenon that the enforced expression of Ccnb1 could not completely recover proliferation capacity of FoxM1-deficient SCs. Our data suggest that FoxM1 maintains the SC pool partially by transcriptionally regulating Ccnb1. Whether FoxM1 promotes the proliferation of SCs by ribosome biogenesis requires further studies.
Besides the effect of FoxM1 on SC cell cycling, we found that loss of FoxM1 severely impaired the regeneration potential of SCs. FoxM1-cKO mice showed largely unrepaired muscle regions compared to FoxM1fl/fl mice, which could not be simply attributed to the altered cell cycle of SCs. Through RNA-Sequencing, we found that knockdown of FoxM1 resulted in downregulation of genes related to the wnt pathway, including Apc, Axin1, and Gsk3β. Importantly, muscle regeneration required inhibited wnt/β-catenin signaling and Apc interaction with Axin, Cki, and Gsk3β to form a cytoplasmic destruction complex to phosphorylate β-catenin, leading to destruction of β-catenin via the ubiquitylation pathway27. The loss of Apc in SCs does not affect skeletal muscle tissue integrity but results in severe muscle regeneration defects33. β-Catenin constitutively active mice are not affected with respect to the number of SCs in the homeostatic condition. However, these mice display muscle regeneration defects with aberrant myofiber morphology and atrophy, which is mainly attributed to hyperactive Tgf-β signaling28. The phenotype of muscle regeneration in FoxM1-cKO mice was similar to that in Apc-deficient and β-catenin constitutively active mice. Furthermore, the protein level of β-catenin in FoxM1-deficient SCs was considerably increased following injury, raising a possibility that FoxM1 can inhibit wnt/β-catenin signaling to preserve regenerative potential of SCs.
The crosstalk of FoxM1 and wnt/Apc/β-catenin signaling is complicated. It has been reported that Gsk3 phosphorylates FoxM1 on serine 474, which induces FoxM1 ubiquitylation and subsequent degradation in LN229 and U87 cells25. Wnt3a increases the level and nuclear translocation of FoxM1, which binds directly to β-catenin and enhances β-catenin nuclear localization in 293T cells26. Presently, FoxM1 transcriptionally regulated the expression of Apc to restrict the activation of β-catenin signaling in SCs under stress. We speculate that the differing observations of our study may be attributed to the differences in cell types. Inhibition of wnt/β-catenin signaling by ICG-001 efficiently rescued the aberrant myofiber morphology and atrophy in FoxM1-cKO mice after injury. These findings expand our insights into the interwoven intercommunication between FoxM1 and wnt/Apc/β-catenin signaling and reveal the protective function of FoxM1 in muscle regeneration. This may be an adaptive strategy for SCs upon injury. Whether the FoxM1–Apc–β-catenin axis exists in other stem cells is unclear and warrants further investigation.
FoxM1 was required for SC proliferation and muscle regeneration, but excessive FoxM1 resulted in SC exhaustion. The expression of FoxM1 in SCs during muscle regeneration also showed a process of upregulation to downregulation. The protein level of FoxM1 was evidently increased in activated C2C12 compared to quiescent cells21 and notably decreased when returning to quiescent state (for C2C12 and SCs). Thus a regulatory network may precisely control the expression of FoxM1 at the transcriptional and epigenetic level. Proteolysis by the ubiquitin–proteasome system is a major mechanism involved in myofibrillar protein degradation. FoxM1 interacts with the complex of E3 ligase APC/C and its adaptor, Cdh1. Cdh1 stimulated degradation of FoxM1 protein in HeLa49, U2OS17, and MCF-7 cells50. Moreover, Fbxl2 (a component of Skp-Cullin-F box ubiquitin E3 ligase)51 and VprBP (a component of cullin 4-based E3 ubiquitin ligase)52 have been reported to regulate the degradation of FoxM1 in different cells. Here we discovered that FoxM1 directly interacted with Cdh1, rather than the adaptor Cdc20. Knockdown or knockout of Cdh1 resulted in the accumulation of FoxM1 protein in SCs. Cdh1 was required for FoxM1 ubiquitylation and subsequent degradation. The findings expand our knowledge of the relationship between Cdh1 and FoxM1 in stem cells and demonstrate that APC/CCdh1 mediates ubiquitylation and degradation of FoxM1 protein in SCs.
Cdh1 has been implicated in regulating cell cycle entry. Presently, deletion of Cdh1 resulted in decreased quiescent SCs and SC exhaustion. This phenotype is similar to that of FoxM1-deficient SCs. The haploinsufficiency of FoxM1 partially recovered the maintenance of SCs after injury and rescued the muscle regeneration of Cdh1-cKO mice, implying that FoxM1 was one of the key downstream targets of Cdh1 in vivo. The accumulation of ubiquitinated protein increased at 3 DPI, indicating that muscle regeneration is associated with the activation of machinery involved in protein degradation12. We observed that knockout of Cdh1 in SCs resulted in impaired muscle regeneration, indicating the important function of ubiquitin–proteasome system in muscle regeneration. Whether Cdh1 regulates SCs maintenance and muscle development with age requires further investigation.
In summary, the data presented here demonstrate that FoxM1 was critical for muscle development and regeneration. FoxM1 regulated cell cycle entry of SCs to meet the needs of myogenesis and muscle development, partially via Ccnb1. Upon injury, FoxM1 transcriptionally regulated Apc expression to repress wnt/β-catenin signaling. The ubiquitylation and degradation of FoxM1 protein in SCs was promoted by APC/CCdh1. These findings uncover an Cdh1–FoxM1–Apc axis that regulates muscle development and regeneration. Whether this axis is associated with muscle diseases and whether it could be potential intervention targets warrants further investigations.
The mice used had a C57BL/6J background. The FoxM1fl/+ mice carry a FoxM1 locus in which exons 4–7 were flanked by 2 loxp sites, while Cdh1fl/+ mice carry a Cdh1 locus in which exons 2–14 were flanked by 2 loxp sites (Cyagen, China). The FoxM1fl/fl mice and Cdh1fl/fl was crossed with Pax7-Cre mice to generate FoxM1fl/fl-Pax7-Cre SC-specific knockout mice (FoxM1-cKO) and Cdh1fl/fl-Pax7-Cre SC-specific knockout mice (Cdh1-cKO). For SC-specific overexpression of FoxM1, we generated a bacterial artificial chromosome (BAC) transgenic mouse lines, in which a “FoxM1 cds—rBGpA” cassette was inserted into the BAC RP23-267J16 following the start codon (ATG) in exon 1 of Pax7 gene (Tg-FoxM1). All compound genetically engineered mice were a result of breeding of the above mice followed by appropriate PCR-based genotyping with specific primers (Table S1). For drug treatments, ICG-001 (20 mg/kg, 3 injections/week, Selleck) and RO-3306 (4 mg/kg, 4 injections/week, Selleck) were intraperitoneally injected into mice for 2 and 5 weeks, respectively. All the mice experiments were approved by the Animal Committee of the Institute of Zoology, Third Military Medical University.
TA muscles harvested and were fixed for 12 h using 4% paraformaldehyde and subsequently transferred to 20% sucrose overnight for dehydration. For the assessment of muscle morphology, 10-μm-thick transverse sections of TA muscle were subjected to hematoxylin and eosin staining. For quantitative analysis, the CSA was performed using the AxioVision software (Zeiss).
To induce injury, 50 μl 1.2% BaCl2 in phosphate-buffered saline (PBS) or PBS alone was injected into and along the length of TA and gastrocnemius muscles. The hindlimb muscles or TA muscles were harvested 2–30 days after injury depending on the experiments.
Mice were acclimated to moderate treadmill running (10 m/min for 5 min daily) on a treadmill (ZH-PT, Zheng Hua) for 3 consecutive days before the test. Then mice ran on the treadmill at 10 m/min for 5 min, and the speed was increased from 10 to 16 m/min and then maintained constant until exhaustion. Running time and running distance were measured.
Single myofibers were isolated from EDL muscles and digested in Dulbecco’s modified Eagle’s medium (DMEM; catalog number: SH30022.01, Hyclone, GE, USA) with 0.2% collagenase (Sigma-Aldrich, USA) at 37 °C for 90 min. Fibers were liberated by trituration in DMEM medium with Pasteur pipettes.
To culture SCs in vitro, we obtained SCs using single myofiber cultures. Briefly, single fibers were placed in Matrigel-coated dishes (BD Bioscience, USA) in fiber medium consisting of DMEM (Hyclone) with 20% fetal bovine serum (Hyclone), 1% penicillin/streptomycin (Hyclone), and 1% Chick embryo extract (US biological, USA) at 37 °C with 5% CO2. SCs migrated off the myofibers in 3–4 days. The correct enrichment of SCs were validated by fluorescence staining of Pax7. To analyze the growth of SCs, isolated SCs were cultured in fiber medium. For inducing the reversed state of SCs, cells were cultured in serum-free DMEM for 72 h. For differentiation, SCs were cultured in DMEM containing 2% horse serum (Gibco, Thermo Fisher, USA) on Matrigel. For the inhibition of Cdk1/Ccnb1, SCs were treated with 0.5 μM RO-3306 (Selleck) for 24 h before analysis.
SC isolation and purification was performed according to the established methods. Briefly, TA muscle of mice was subjected to 0.2% collagenase (Sigma) for 90 min and then 0.2% dispase (Sigma) for 30 min. The cell suspension was filtered through a 70-μm nylon filter (Falcon), and mononuclear cells were collected and subjected to fluorescence-activated cell sorter (FACS; BD FACSAriaII) using immunostaining with the following: biotin anti-mouse CD45 (Biolegend, 103104), biotin anti-mouse/human CD11b (Biolegend, 101204), biotin anti-mouse CD31 (Biolegend, 102404), biotin anti-mouse Sca1 (Biolegend, 108103), streptavidin-APC/Cy7 (Biolegend, 405208), CD34-Alexa Fluor 647 (Biolegend, 152205), and Integrin α7-FITC (eBioscience, 11-5867-42). For the intracellular staining, the SCs were first stained with the surface marker and then subjected to Fixation/Permeabilization Kit (BD) according to the manufacturer’s instruction. Then the cells were stained with anti-Ki67-PE (Biolegend, 652404) or anti-Cyclin B1-PE (Biolegend, 647903) or stained with anti-Apc (Beyotime, AF2113) or anti-Cdh1 (Proteintech, 16368-1-AP) and incubated with anti-rabbit-FITC (Biolegend, A120-101F).
For BrdU incorporation analysis, SCs cultured in vitro were incubated with 10 μM BrdU (Sigma) for 30 min, washed with PBS, fixed with 70% absolute ethyl alcohol for 6 h, permeabilized with 0.1% Triton X-100 for 10 min, incubated with 2 M HCl for 30 min, and blocked with 2% horse serum for 1 h. Then cells were stained with anti-BrdU-FITC (Biolegend, 364104) for 30 min and analyzed by flow cytometry.
Myofibers or SCs were fixed with 4% paraformaldehyde for 30 min at room temperature, washed, and then incubated in 0.1% Trixon X-100 for 10 min. Then samples were blocked with 2% horse serum for 1 h and incubated with primary antibodies at 4 °C overnight. The following antibodies were used: Pax7 (Invitrogen, PA1-117, 1:100), MyHC (R&D, MAB4470, 1:50), and β-catenin (Proteintech, 51067-2-AP, 1:200). After the primary antibody incubation, myofibers or SCs were incubated with secondary antibodies conjugated with Alexa Fluor 488 or 647 (Beyotime, 1:500). Finally, the nuclei were staining by 4,6-diamidino-2-phenylindole (5 μg/ml, Sigma) for 10 min, and the fluorescent pictures were captured by confocal microscopy (Olympus).
To assess the altered cell phenotypes, we seeded the SCs in 96-well plates with 2000 cells per well in triplicate. At 24, 48, and 72 h post culture, cell proliferation was assessed using the Cell Counting Kit-8 Kit (Beyotime) according to the manufacturer’s instruction.
About 800 SCs from mice (homeostatic condition or 48 h-post injury) were isolated by FACS, and the mRNA library was constructed using the QIAseq FX Single Cell RNA Library Kit (QIAGEN) according to the manufacturer’s instruction. Libraries were sequenced by the Illumina HiSeq 2000 platform as 150-bp pair-ended reads. Reads were aligned using bowtie v0.12.9. FPKM estimation was performed with Cufflinks v2.1.1, aligned reads were counted with HTSeq, and differential expression analysis was performed with DESeq2. Differentially expressed genes were selected using a cut-off at a p value of <0.05 (false discovery rate adjusted for multiple testing).
Total cellular RNA was isolated from 3000 SCs immediately after FACS sorting using the Total RNA Isolation Kit (Thermo Fisher) according to the manufacturer’s instructions. cDNA was reverse transcribed using the PrimeScript RT Reagent Kit (Takara) and subjected to real-time PCR with SYBR Green Supermix (Bio-Rad) in an iCycler iQ Real Time PCR Detection System (Bio-rad). All primers are listed in Table S1. All samples were run in triplicate. β-Actin (Actinb) was used as an internal control for mRNA.
To generate the vectors for the expression of FoxM1, CDC20, and CDH1-specific shRNA, we designed the sequence of shRNAs and cloned shRNAs into the vector pLKO.1-puro (primer sequences, Table S1). For generation of the vectors of expression of FoxM1, we amplified transcriptional regions by reverse transcription–PCR and cloned the regions into pCDH-puro (primer sequences, Table S1). The HA-Ubiquitin plasmid was a gift from Dr. Zhigang Li. For plasmid transfection, C2C12 or 293T cells were transfected with the indicated plasmids using polyetherimide (Sigma-Aldrich, USA). For lentivirus production, 293T cells were transfected with the helper plasmid pSPAX2 and pMD2.G. The medium was replaced with fresh medium at 10 h after transfection. The culture supernatants were collected at 48 h after transfection and filtered by 0.22-μm membrane. Virus was stored at −80 °C until use. For lentiviral infection of cells, lentiviral stock was added to C2C12 or SCs with polybrene (8 μg/ml, Sigma-Aldrich, USA). After incubation overnight, supernatants was removed, and cells were washed with PBS and then resuspended in fresh medium with or without puromycin (3 μg/ml, Sigma-Aldrich) to screen stable cell lines.
Cells were prepared using RIPA buffer (Beyotime) containing protease inhibitor mixture (Beyotime). Immnoprecipitates or total cell lysates were analyzed by western blot according to standard procedures. The antibodies used in this study were against: Cdc20 (Proteintech, 10252-1-AP, 1:1000), Cdh1 (Santa, sc-56312 1:1000), FoxM1 (Proteintech, 13147-1-AP, 1:1000), Cyclin B1 (CST, 4138T, 1:1000), Cyclin B2 (Beyotime, AF2509, 1:1000), Apc (Beyotime, AF2113, 1:1000), β-catenin (Proteintech, 51067-2-AP, 1:1000), HA (Proteintech, 66006-2-Ig, 1:10,000), and Tubulin (Beyotime, AF0001, 1:1000).
SCs were harvested and used for ChIP assays according to the manufacturer’s instruction of the EZ-ChIP Chromatin Immunoprecipitation Kit (Millipore). In brief, the cells were fixed with 1% formaldehyde for 10 min, and the fixation reaction was quenched with glycine to a final concentration of 125 mM. The cells were lysed and sonicated until the desired lengths were achieved (100–500 bp). Then 5 μg of anti-FoxM1 (Proteintech) or control IgG were used for immunoprecipitation. After elution of DNA from the precipitated immunocomplexes, PCR or qPCR were performed with the specific primers (Table S1).
The corresponding genomic regions that included the putative binding sites was amplified by PCR from WT C57BL/6 mouse genomic DNA (primer sequences, Table S1), then were subsequently cloned into a pGL3-basic luciferase reporter vector. The vectors with mutant-binding sites were generated by site-directed mutagenesis with 3 bases’ mutation (primer sequences, Table S1). The vectors were transfected into 293T cells through the use of the polyetherimide, together with either pCDH or pCDH-FoxM1, as well as phRL-SV40 vector as an internal control. Cells were collected at 36 h after transfection, and both luciferase activities were assessed with the Dual Luciferase Reporter Gene Assay Kit according to the manufacturer’s instructions (Beyotime Biotechnology, China).
All experiments included at least three biological replicates. Comparisons between two groups were analyzed using two-tailed Student’s t test. Differences among more than two groups were analyzed using one-way analysis of variance followed by Tukey–Kramer post hoc tests. Values of p < 0.05 were considered statistically significant. All data are means ± SD.
Supplementary Information accompanies this paper at (10.1038/s41419-020-2375-6).
We thank Professor You-Yang Zhao (University of Illinois College of Medicine) for providing the FoxM1fl/fl mice. This work was supported by grants from National Key R&D Program of China (2017YFA0106700) and National Science Foundation of China (81670096, 31401253, and 81700135) and sponsored by Natural Science Foundation of Chongqing, China (CSTC2016JCYJA0354), Major Technological Innovation Plan of Southwest Hospital (SWH2016ZDCX1010), Army Major Scientific Research Projects (AWS17J007), and Military Medical Innovation Program (SWH2018LJ-07 and SWH2018QNKJ-28).