HMGB2 regulates satellite-cell-mediated skeletal muscle regeneration through IGF2BP2

Myogenesis is the term describing the process of myofiber formation, which occurs during embryonic development, postnatal growth and muscle regeneration (Chargé and Rudnicki, 2004; Tidball and Villalta, 2010). This process is a highly coordinated event and implicated in a series of transcriptional networks, among which Pax7, Myf5, MyoD (also known as MYOD1), myogenin and Mrf4 (also known as MYF6) are crucial and indispensable (Buckingham and Rigby, 2014; Mok and Sweetman, 2011). Skeletal muscle repair from injury involves in a series of complicated cascade events, which can be divided into the following steps. First, quiescent satellite cells, a pool of adult muscle stem cells, are activated and become committed myoblasts in response to the damage (Kang and Krauss, 2010; Kuang and Rudnicki, 2008). Then, these myoblasts begin to proliferate to expand the population of progenitor for repair (Chargé and Rudnicki, 2004; Kuang and Rudnicki, 2008). Finally, proliferating myoblasts withdraw from the cell cycle, undergo terminal differentiation and fuse with each other to form renascent myofibers that replace the damaged and dead ones (Apponi et al., 2011; Kang and Krauss, 2010). Sets of transcription factors have been identified that regulate the proliferation of satellite cells and terminal differentiation. However, more precise molecular mechanism and alternative key regulators are required to be further elucidated.

The high-mobility group (HMG) superfamily consists of three families, HMGA, HMGB and HMGN, all of which are chromatin-binding proteins that share similarities in structural and functional properties (Agresti and Bianchi, 2003; Hock et al., 2007). It has been well documented that HMG proteins not only act as dynamic modulators of chromatin architecture but also especially influence DNA replication, recombination, repair and transcriptional regulation (Agresti and Bianchi, 2003; Bianchi and Agresti, 2005; Ueda and Yoshida, 2010). Multiple transcription factors have been verified to interact with HMG proteins at specific gene locus during the process of transcription (Ueda and Yoshida, 2010). Moreover, it has been reported that both HMGA and HMGB families are mainly expressed in stem cells or progenitors to maintain proliferation and undifferentiated status (Hock et al., 2007). In particularly, HMGA2 can induce mouse embryonic stem cells to commit to the myogenic lineage and regulate myoblast proliferation and skeletal muscle development (Caron et al., 2005; Li et al., 2012).

There are three members of HMGB proteins in mammals, HMGB1, HMGB2 and HMGB3. All of them contain two DNA-binding domains (basic HMG boxes) followed by a long acidic C-terminal tail and have more than 80% identity in protein sequences (Bustin, 1999; McCauley et al., 2005; Catena et al., 2009). HMGB1 has been demonstrated to play a key role in multiple biochemical and molecular activities, such as the innate immunity response, necrosis, arthritis and tumorigenesis (Hock et al., 2007; Yanai et al., 2012). Although it has been demonstrated that as a transcriptional co-regulator, HMGB2 regulates various differentiation programs, including erythropoiesis, chondrogenesis and spermatogenesis, its role in myogenesis and muscle regeneration remains largely unclear (Laurent et al., 2010; Ronfani et al., 2001; Taniguchi et al., 2011).

Post-transcriptional regulation of myogenesis is an important process for muscle development and regeneration (Apponi et al., 2011). Insulin-like growth factor-II mRNA-binding proteins (IGF2BP2), also known as IMP2, is a member of IMP family that contains two RNA-binding domains and four K homology motifs and functions in mRNA localization, turn over and translation modulation (Nielsen et al., 2002, 2003). The involvement of RNA-binding proteins, such as IGF2BP2, HuR (also known as ELAVL1), CUGBP1 and Lin-28, in muscle biology has been extensively highlighted in recent investigations (Apponi et al., 2011; Li et al., 2012). A previous study found that IGF2BP2 functions by binding to the 5′ UTR of insulin-like growth factor 2 (IGF2) mRNA and controlling IGF2 translation (Dai et al., 2011). More recently, IGF2BP2 was also suggested to regulate myoblasts proliferation through binding to various target mRNAs and modulating their translation, such as Myc, Sp1, IGF1R, Ccng1 and Nras (Gong et al., 2015; Li et al., 2012). Interestingly, IGF2BP2 has been identified as a direct target of HMGA2 in proliferating myoblasts, NIH/3T3 fibroblasts and mouse embryos (Brants et al., 2004; Cleynen et al., 2007; Li et al., 2012). However, whether or not HMGB2 is able to regulate the expression of IGF2BP2 during myogenesis is unknown.

In the present study, we first found that HMGB2 was upregulated significantly in the early stage of muscle regeneration and highly expressed in proliferating myoblasts. Then, we used gain- and loss-of-function approaches to demonstrate that HMGB2 maintains myoblast proliferation and impedes myogenic differentiation by controlling IGF2BP2, which in return enhanced the protein production of Myf5 and cyclin A2. It was further demonstrated that IGF2BP2 could bind either to Myf5 or to cyclin A2 mRNA to improve the mRNA stability or enhance the translation, respectively. In vivo data also implied that HMGB2 was required for satellite cell proliferation and muscle repair after injury. In summary, we have identified and characterized the role of the HMGB2–IGF2BP2 axis in myogenesis and muscle regeneration.

To determine the expression pattern of potential genes in the early stage of muscle repair after injury. RNA sequencing (RNA-seq) was employed to analyze the genome-wide gene expression of damaged muscle tissues at various time points after injury. Among sets of differentially expressed genes, HMGB2 was focused on because of its upregulated expression (Fig. 1A). Real-time quantitative PCR (qPCR) and western blotting also confirmed that the expression of HMGB2 significantly increased in the early period of muscle repair, peaking at day 3 after cardiotoxin (CTX) injection, indicating its potential role in muscle regeneration (Fig. 1B). Furthermore, large numbers of HMGB2-positive cells were found in the damaged adult tibialis anterior muscle but not in the uninjured one (Fig. 1C,D). It has been well documented that activation and proliferation of quiescent satellite cells are required for adult muscle regeneration (Kang and Krauss, 2010). Importantly, immunofluorescence revealed that most of HMGB2-positive cells were simultaneously Pax7-positive, a marker of muscle stem cell or progenitor (Fig. 1D). Further evidence revealed that both mRNA and protein of HMGB2 were highly expressed in neonatal muscle but decreased during the progress of postnatal development and were not detectable in mature muscle (Fig. 1E). Collectively, these data support that HMGB2 is expressed mainly in myoblasts but not in mature myofibers.

Intriguingly, the myogenic genes Myf5, myogenin and Mrf4 showed a similar expression pattern during muscle development, implying that HMGB2 might be expressed during myogenesis (Fig. S1A). However, MyoD and Pax7 have differing expression patterns to HMGB2 in muscle development, which indicates that they potentially operate through a different mechanism. In addition, the expression of HMGB2 in other tissues of adult mice was also detected. We found that HMGB2 was widely expressed in various adult tissues (Fig. S1B), raising the possibility that HMGB2 might participate in maintaining the physiological function and homeostasis of these organs. Taken together, HMGB2 might play a vital role in muscle development and regeneration.

In order to further investigate the impact of HMGB2 on myogenesis, the C2C12 myoblast differentiation model was studied. The expression trend of HMGB2 was evaluated during myogenic differentiation. Similar to the results in vivo, both HMGB2 and Pax7 were highly expressed in proliferating C2C12 myoblasts but promptly downregulated after differentiation (Fig. 2A,B). Few HMGB2 proteins were detected in mature myotubes (Fig. 2C). Both in vivo and in vitro analyses showed that HMGB2 proteins were mainly located in Pax7-positive muscle stem cells or progenitors but rarely in mature myofibers (Figs 1C,D, 2C), indicating that HMGB2 is mostly expressed in undifferentiated myoblasts.

As HMGB2 is a chromatin-binding protein and also interacts with transcription regulators in the nucleus, its location in C2C12 myoblasts and primary satellite cells isolated from adult mice was detected in this work. Primary satellite cells were successfully isolated and became activated myoblasts. As expected, immunofluorescence staining showed that HMGB2 proteins were only located in the nuclei of C2C12 and primary myoblasts (Fig. 2D,E), which prompted us to speculate that HMGB2 operates with other myogenic regulators in the nucleus.

Next, three small interfering RNAs (siRNAs) were designed to knock down HMGB2 in C2C12 myoblasts (denoted si-HM-1, si-HM-2 and si-HM-3), two of which were efficient (Fig. S2A). Si-HMGB2, a mixture of si-HM-2 and si-HM-3, caused a substantial knockdown of HMGB2 proteins and was used in all of the following analysis (Fig. S2B). As shown, depletion of HMGB2 accelerated the initiation of myogenic program, because of a significant increase in the mRNA expression of early myogenic marker myogenin at 24 h after differentiation induction (Fig. 3A). Myogenin-positive cells also increased prominently at 24 h after differentiation when HMGB2 was knocked down (Fig. 3B). A similar increase in the expression of fast myosin skeletal heavy chain (MyHC, also known as Myh1) was also observed in differentiating C2C12 cells after HMGB2 knockdown (Fig. 3C,E). Consistently, addition of si-HMGB2 in C2C12 cells caused a significant enhancement of myotube formation (Fig. 3D). However, ectopic expression of HMGB2 led to a remarkable concomitant repression of myotube formation (Fig. 3F). Thus, HMGB2 was identified as a crucial myogenic regulator.

It has been well established that HMG proteins are mainly specific to stem cells, and they are implicated in the regulation of stem cell proliferation and differentiation (Agresti and Bianchi, 2003; Hock et al., 2007). More recently, Zhizhong Li and his colleagues have reported that the HMGA2–IGF2BP2 axis affects muscle development through controlling myoblast proliferation (Li et al., 2012). Simultaneously, the expression of HMGA2 decreased significantly when HMGB2 was downregulated (Fig. S2C–E). These results drove us to hypothesize that HMGB2 probably suppresses myogenesis by means of regulating myoblast proliferation and self-renewal. To test this, the proliferation rate of C2C12 cells and muscle satellite cells transfected with si-HMGB2 and si-NT was examined, respectively. Data from a real-time monitoring system revealed that the proliferation of C2C12 myoblasts and primary satellite cells was reduced after HMGB2 depletion with siRNAs (Fig. 4A; Fig. S4B). Flow cytometry analysis followed by propidium iodide staining further revealed that an arrest of cell cycle in S phase occurred in response to a HMGB2 decrease in C2C12 myoblasts (Fig. 4B).

To address the molecular mechanism underlying the control of cell cycle progression through HMGB2, HMGB2 was knocked down in myoblasts. The expression of several myogenic-associated factors (Myf5, MyoD, Pax7, id2 and YY1) and cell cycle regulators [cyclin A2, cyclin E, CDK2, CDK4, Rb, P27 (CDKN1B) and Akt (Akt1 isoform)] was detected by qPCR. A significant difference in mRNA level between the si-HMGB2 group and the control group was not observed for any of the detected genes (Fig. 4C,D). Intriguingly, a pronounced reduction of Myf5 and cyclin A2 proteins occurred in response to si-HMGB2 treatment both in C2C12 myoblasts and primary muscle satellite cells. However, there was no significant difference for YY1, Pax7 and MyoD proteins (Fig. 4E; Fig. S4C). In addition, overexpression of HMGB2 did not affect the mRNA levels of Myf5 and cyclin A2, but promoted their protein expression (Fig. S3A,B). Consistent with the siRNA experiment, overexpression of HMGB2 did not affect either the RNA or protein levels of Pax7 and MyoD (Fig. S3A,B).

Next, a model of CTX-mediated muscle injury concomitant with HMGB2 depletion was used to examine the expression of the above regulators (Fig. 4F). Both mRNA and protein levels of HMGB2 were successfully downregulated in tibialis anterior at day 3 after injury by intramuscular injection of si-HMGB2 (Fig. 4G,H), which resulted in significant reduction in Myf5, cyclin A2 and Pax7 proteins (Fig. 4H). In addition, the number of EdU (5-ethynyl-2′-deoxyuridine)-positive cells decreased substantially in tibialis anterior when HMGB2 was downregulated (Fig. 4I), implying that the proliferation of satellite cells was restrained after HMGB2 was depleted. Based on in vitro and in vivo results, we conclude that HMGB2 maintains myoblast proliferation and self-renewal by controlling protein expression of Myf5 and cyclin A2.

We found that HMGB2 regulates Myf5 and cyclin A2 at the protein but not mRNA level, suggesting that a post-transcriptional regulation mechanism needed to be further elucidated. IGF2BP2, a RNA-binding protein, can bind to the 5′-UTR of target mRNAs and enhance their stability and/or translation (Dai et al., 2011; Nielsen et al., 2002). A recent report has also suggested that HMGA2 directly regulates IGF2BP2 to affect myoblast proliferation and skeletal muscle development (Li et al., 2012). Intriguingly, the expression of IGF2BP2 was found to decrease in tibialis anterior muscle as well during postnatal development and was at a low level in adult muscle (Fig. 5A). This expression pattern was similar to that of HMGB2. Whether or not IGF2BP2 is a downstream effector of HMGB2 and regulates the protein levels of Myf5 and cyclin A2 in myoblasts needs to be further investigated.

To verify this possibility, both HMGB2 knockdown and overexpression experiments were first conducted. At 2 days after transfection, qPCR and western blotting analyses were performed in C2C12 myoblasts cultured in growth medium. Results showed that depletion of HMGB2 reduced IGF2BP2 expression, whereas ectopic expression of HMGB2 caused an increase in IGF2BP2 (Fig. 5B,C). Furthermore, consistent with in vitro results, in vivo analysis showed that the expression of IGF2BP2 was upregulated significantly in the early stages of muscle injury and downregulated when HMGB2 was depleted (Fig. 5D,E). Next, three siRNAs targeting IGF2BP2 (denoted si-IG-1, si-IG-2 and si-IG3) were used to repress IGF2BP2 expression. As shown, all of them were efficient (Fig. 5F, left). Si-IGF2BP2, a mixture of these three siRNAs, could also significantly inhibit protein expression of IGF2BP2 (Fig. 5F, right), and was applied in the following experiments. It was shown that there was no significant difference in HMGB2 expression when IGF2BP2 was knocked down in myoblasts (Fig. 5G).

To test the role of IGF2BP2 in myogenesis, both C2C12 cells and muscle satellite cells were induced to differentiate after IGF2BP2 was knocked down. More and longer fused myofibers were observed in IGF2BP2 knockdown group than that in the control one, as it is indicated by the higher fusion index (Fig. 5H; Fig. S4D). Thus, similar to HMGB2 depletion, IGF2BP2 downregulation also significantly promotes myogenesis in vitro. Taken together, these results imply that IGF2BP2 is a downstream target of HMGB2 both in vivo and in vitro.

To gain insights into the mechanism by which IGF2BP2 regulates myoblasts proliferation, IGF2BP2 was knocked down in myoblasts. Flow cytometry analysis after propidium iodide staining demonstrated that a decrease in IGF2BP2 caused the cell cycle arrest in S phase, as it is evidenced by a higher proportion of myoblasts in S phase after si-IGF2BP2 treatment (Fig. 6A). These observations recapitulate the results of HMGB2 knockdown (Fig. 4B). Then, RNA-binding protein immunoprecipitation (RIP) was carried out to identify target mRNAs of IGF2BP2. It was observed that, compared to the control group, both Myf5 and cyclin A2 mRNAs, but not Pax7, MyoD,  CDK2, smurf1 and CDK4 mRNAs, were well enriched by the IGF2BP2 antibody, suggesting that IGF2BP2 binds to Myf5 and cyclin A2 mRNAs in myoblasts (Fig. 6B,C).

Further analysis indicated that depletion of IGF2BP2 could not affect mRNA levels of Myf5 and cyclin A2 but downregulated both protein levels in C2C12 myoblasts and muscle satellite cells (Fig. 6D,E; Fig. S4E). However, MyoD was not regulated by IGF2BP2 (Fig. 6D,E). In contrast, an increase in Myf5 and cyclin A2 proteins accompanied the overexpression of IGF2BP2 (Fig. 6F). Remarkably, overexpressing IGF2BP2 could partially rescue the si-HMGB2-mediated reduction of Myf5 and cyclin A2 proteins (Fig. 6G). Moreover, immunofluorescence analysis demonstrated that overexpression of IGF2BP2 inhibited the enhanced myoblast differentiation caused by HMGB2 knockdown in C2C12 cells (Fig. 6H). Taken together, these data indicate that HMGB2 maintains myoblast proliferation and fate determination through regulating IGF2BP2, which binds to Myf5 and cyclin A2 mRNAs and enhances their protein production.

To further elucidate the mechanism that IGF2BP2 regulates the protein generation of cyclin A2 and Myf5 after binding to their mRNAs, mRNA decay analyses of IGF2BP2, cyclin A2 and Myf5 were performed. As expected, IGF2BP2 mRNA degraded with a half-life of nearly 4.5 h in control cells, but its half-life was reduced to 2.5 h after IGF2BP2 downregulation (Fig. 6I, left). Interestingly, there was no significant difference in Myf5 mRNA half-life between the siRNA treatment and the control group (Fig. 6I, middle), whereas si-IGF2BP2 treatment led to a nearly two-fold acceleration in the decay of cyclin A2 mRNA (Fig. 6I, right). In order to elucidate the mechanism by which IGF2BP2 regulates Myf5 protein expression, we performed a further analysis that indicated that when mRNA transcription and protein degradation were simultaneously inhibited, Myf5 protein decreased significantly after IGF2BP2 was knocked down (Fig. 6J). These findings indicate distinctive mechanisms by which IGF2BP2 promotes protein production of Myf5 and cyclin A2 after binding to their mRNAs. It is likely that IGF2BP2 enhances the translation of Myf5 without altering its mRNA stability. Nevertheless, the binding of IGF2BP2 improves cyclin A2 mRNA stability, subsequently increasing its protein production.

To define whether the effect of HMGB2 on myogenesis can be recapitulated in an in vivo context, a CTX-mediated muscle regeneration model was employed. In order to ensure that HMGB2 was always inhibited during muscle regeneration, si-HMGB2 or non-targeting siRNA (si-NT) was injected into tibialis anterior every 4 days and HMGB2 was maintained at a low level (Fig. 7A,B). It was shown that, compared to the control muscle, both the centrally located myonuclei number and the myofiber cross-sectional area (CSA) in tibialis anterior administrated with si-HMGB2 were significantly reduced at day 5 and day 10 after injury (Fig. 7C,D). Thus, intramuscular addition of si-HMGB2 impaired muscle regeneration. This result was contrary to the observation in vitro that knocking down HMGB2 promoted myogenesis (Fig. 3), which might be ascribed to the decrease in activated satellite cells caused by perturbation of satellite cell proliferation (Fig. 4). Collectively, HMGB2 decrease severely compromises muscle regeneration, suggesting a general requirement for HMGB2 in muscle remodeling and maintenance of muscle integrity.

Myogenesis is a well-orchestrated gene regulation program, and elucidating the underlying molecular mechanism is quite a challenging work. In this study, we identified the involvement of HMGB2 in myogenesis and muscle regeneration. Previous reports have revealed that HMGB2 is required for cell cycle progression and is mainly expressed in developing embryo and stem cells, including mesenchymal stem cells, embryonic stem cells and spermatocytes (Ronfani et al., 2001; Seyedin and Kistler, 1979). In addition, HMGB2 has been shown to be downregulated significantly within the first 24 h of myogenic differentiation (Rajan et al., 2012). In accordance with these observations, our results suggest that HMGB2 is highly expressed in activated satellite cells and C2C12 myoblasts, but decreases significantly during myogenesis and is scarcely expressed in mature myofibers (Fig. 2). Therefore, we speculate that HMGB2 functions in myoblasts to maintain proliferation and keep stemness. This was well confirmed by our data showing that knockdown of HMGB2 impedes progression of myoblasts cell cycle both in vitro and in vivo.

Myf5 is required for muscle lineage determination in the early phase of muscle development and is considered as a marker of committed myogenic progenitors (Bryson-Richardson and Currie, 2008). Myf5 is also implicated in myoblast proliferation (Apponi et al., 2011). It is well known that cyclin A2 controls both S phase and G2/M transition. Thus, in our study, the si-HMGB2-mediated S and G2/M phase arrest can be explained by the simultaneous reduction in Myf5 and cyclin A2 proteins.

By contrast, cell cycle withdraw is required for the initiation of myogenic differentiation. We revealed that HMGB2 depletion induced withdraw from cell cycle and enhanced myotube formation in C2C12 cells. Contradictorily, a decrease of HMGB2 attenuated muscle repair, which might be explained by an insufficient satellite cell population due to impair of myoblast amplification when injury occurred. Experiments in vitro showed that both knocking down and overexpressing HMGB2 did not influence Pax7 expression in C2C12 myoblasts, implying that HMGB2 does not function by regulating Pax7 (Fig. 4C,E). However, depletion of HMGB2 in tibialis anterior resulted in the reduction of Pax7 protein and Pax7-positive satellite cells (Fig. 4H,I), which might result from the decreased number of proliferating satellite cells (Fig. 4I). Interestingly, we found that the expression of HMGB2 increased immediately after injury, which is similar to what occurs with Myf5 and MyoD (Tidball and Villalta, 2010). This further indicated that HMGB2 affects muscle regeneration by stimulating satellite cell expansion mainly in the early stage of injury.

It has been well-established that HMGB proteins can regulate gene-specific transcription through interaction with transcription factors, as well as remodeling chromatin structure (Laurent et al., 2010; Ueda and Yoshida, 2010). Transcription factors identified as the partner of HMGB protein have been well reviewed (Agresti and Bianchi, 2003). Notably, interactions between Oct4 and HMGB2 regulates the Akt pathway and embryonic stem cells pluripotency (Campbell and Rudnicki, 2013). Noboru Taniguchi and his colleagues have reported that HMGB2 coordinates with the Runx2–LEF1–β-catenin complex on the Runx2 proximal promoter to control Runx2 transcription and influence chondrogenic differentiation (Taniguchi et al., 2011). We found that HMGB2 specifically transactivates IGF2BP2 in muscle stem cells or progenitors (Fig. 5), implying that a potential cooperation between HMGB2 and specific transcription factors on IGF2BP2 promoter probably exists.

The HMGA2–IGF2BP2 regulatory axis during muscle development, embryogenesis and tumorigenesis has been well characterized in numerous publications (Cleynen et al., 2007; Li et al., 2012). Here, we also showed that HMGB2 mediates IGF2BP2 transcription both in vivo and in vitro during myogenesis (Fig. 5). Furthermore, HMGB2 was confirmed to positively regulate the expression of HMGA2 (Fig. S2C–E). This implies that HMGB2 might control the expression of IGF2BP2 through a HMGA2-dependent mechanism. RNA-binding proteins (RBPs) could bind to specific sequences of target mRNAs (RNA-binding domains), allowing the control of localization, stability, degradation and translation of mRNAs (Kelly and Corbett, 2009; Wilusz and Wilusz, 2010). In this regard, IGF2BP2 is a key RBP and targets various mRNAs (Dai et al., 2011; Gong et al., 2015; Li et al., 2012). Importantly, strong evidence has been recently provided to prove that IGF2BP2 binds to proliferation-relevant mRNAs, including Myc, Sp1, Igf1r, Ccng1 and Nras, and then controls their stability and/or translation (Gong et al., 2015; Li et al., 2012). Similarly, we identified Myf5 and cyclin A2 as two new mRNA targets of IGF2BP2 in myoblasts (Fig. 6C), and also demonstrated that IGF2BP2 maintains the stability of cyclin A2 but not Myf5 mRNA (Fig. 6H). Further observation indicated that the increased Myf5 protein generation after IGF2BP2 binding to its mRNA can likely be attributed to the enhanced translation (Fig. 6I). Taken together, the specific binding to both mRNAs results in increased protein levels and proliferation maintenance in myoblasts.

The canonical Wnt/β-catenin pathway is required for muscle development and satellite cell proliferation, which is responsible for muscle repair (Otto et al., 2008; von Maltzahn et al., 2012). Strong evidence has confirmed the interactions between HMGB2 and Wnt/β-catenin pathway in mesenchymal stem cells and the superficial zone of articular cartilage (Taniguchi et al., 2011, 2009). This guides us to hypothesize that HMGB2 also influences myoblasts proliferation and muscle regeneration through Wnt/β-catenin pathway.

In summary, the data present here highlights the crucial role of the HMGB2–IGF2BP2 axis in myogenesis and early muscle regeneration (Fig. 7E). HMGB2 induces proliferation of muscle stem cells or progenitors through IGF2BP2, and is necessary for proper muscle regeneration. The insights into these distinct mechanisms provide potential therapeutic approaches to enhance the regenerative ability of muscle stem cells, thereby possibly treating muscle diseases such as muscular dystrophies.

8-week-old male Mus musculus (C57BL/6) were housed in a specific-pathogen-free (SPF) facility with a 12-h dark and 12-h light cycle. All animal experiments were approved by the Animal Care and Use Committee of Guangdong Province and carried out in accordance with ethical standards.

C2C12 cells, provided by ATCC, were cultured in in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal bovine serum (FBS) (GIBICO, Shanghai, China) (growth medium) until confluence. When cells had reached 100% confluence (day 0), C2C12 cells were switched into DMEM with 2% horse serum (GIBICO) (induction medium). All cells are maintained at 37°C in a humidified incubator with 5% CO2.

Satellite cells were isolated from the extensor digitorum longus (EDL) of adult mice. The EDL was isolated from 8-week-old mice and digested with 0.2% Collagenase type I (Sigma, Shanghai, China) solution at 37°C until sufficient myofibers were released. Dead myofibers were removed by transferring the digestion products into new prewarmed DMEM (10% horse serum) for at least three times. Single myofibers were maintained in DMEM (10% horse serum, 0.5% chick embryo extract) for 1 day and then switched to proliferation medium (20% FBS, 10% horse serum, 2% chick embryo extract in DMEM). When primary myoblasts grew and migrated out from the basal lamina, cells were cultured in F10 (GIBICO) medium [20% horse serum, 2.5 ng/ml bFGF (Invitrogen, Shanghai, China), 1% Gluta-Max (GIBCO), 1% penicillin-streptomycin (GIBICO)].

A set of three Stealth RNAi™ siRNAs against mouse HMGB2 and against mouse IGF2BP2 (Invitrogen, Shanghai, China) were purchased from Life Technologies. Their sequences are provided in Table S2 and Table S3, respectively. Stealth siRNA Negative Control containing medium GC content (Invitrogen, Shanghai, China) was used as the negative control. The CDS sequences encoding mouse HMGB2 or IGF2BP2 were inserted into a PCDNA3.1 plasmid backbone (Invitrogen). C2C12 cells were seeded in the 6-well or 12-well plates 1 day before treatment. The expression plasmids or siRNAs were transfected into cells with Lipofectamine^® 2000 (Invitrogen, Shanghai, China) as per the manufacturer's instruction. The medium with transfection mixture was replaced with fresh growth medium at 6 h after transfection.

Total RNA was extracted from cells or tibialis anterior muscle by using TRIzol^® Reagent (Invitrogen, Shanghai, China), and then cDNA was synthesized using a reverse transcription kit (Promega, Beijing, China). The real-time quantitative PCR was performed using a SYBR Green qPCR Kit (Genestar, Beijing, China) and detected in the LightCycler 480 II system (Roche, Basel, Switzerland). All results were normalized to that of GAPDH. The primers used for qPCR were given in Table S1.

Following transfected with si-IGF2BP2 or si-NT for 24 h, C2C12 cells were harvested directly or treated with 5 μM actinomycin D (Act D) and then harvested at the indicated time points. Total RNA extraction, reverse transcription and qPCR were performed as above indicted to detect mRNA decay.

Cells or tibialis anterior muscle were incubated in cell lysis buffer on ice to completely release total protein. Total protein was separated by SDS-PAGE and transferred onto PVDF member (Bio-Rad, Shanghai, China). Then, immunoblotting for target proteins were carried out by specific antibodies as described in Table S4. α-Tubulin was used as the internal control.

In adult mice, right tibialis anterior muscles were administrated with 100 μl of 10 mM CTX and an equal volume of NaCl was injected into left tibialis anterior muscles as a control.

Reagent A was prepared by mixing 12.5 μg si-NT or the target siRNA with 12.5 μl physiological saline solution. Reagent B was prepared by mixing 6.25 μl Entranster-in vivo (Engreen, Beijing, China) with 18.75 μl physiological saline solution. Then, reagent B was added to reagent A and mixed completely. The mixture was incubated at room temperature for 15 min before injecting into tibialis anterior muscles with a syringe.

The RIP experiment was performed as described previously (Keene et al., 2006). In brief, first, prepare the following solutions: polysome lysis buffer, 100 mM KCl, 5 mM MgCl2, 10 mM HEPES (pH 7.0), 0.5% NP40, 1 mM DTT, 100 units/ml RNase, 400 µM vanadyl ribonucleoside complexes (VRC; added before use), protease inhibitor cocktail (added before use); and NT2 buffer, 50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1 mM MgCl₂ and 0.05% NP40. C2C12 cells in growth medium were collected and broken down in polysome lysis buffer on ice for 5 min. To coat antibody on Protein-A/G Magnetic beads (Millipore, Shanghai, China), IgG or IGF2BP2 Ab was added into the bead slurry and incubated for 18 h followed by washing four or five times with ice-cold NT2 buffer. Then, messenger ribonucleoprotein (mRNP) lysate was added into antibody–Protein-A/G bead mixture and incubated for 4 h at 4°C. After washing the beads with ice-cold NT2 buffer, proteinase K was used for 30 min at 55°C to release the RNP components. Finally, RNA was isolated from the immunoprecipitated pellet by TRIzol^® Reagent and reverse-transcribed as described above. The enrichment of target mRNAs was detected by qPCR.

Tibialis anterior muscles were isolated from adult mice and fixed in 4% paraformaldehyde for 12 h at 4°C. Then, samples were subjected to dehydration using graded ethanol and paraffin embedding. Paraffin-embedded samples were cut into 4-μm sections. Paraffin sections were then analyzed by H&E staining or immunostaining with HMGB2 antibody using a cell and tissue staining kit (rabbit kit HRP-DAB system; R&D, Shanghai, China) as per the manufacturer's instruction.

Tibialis anterior muscles were immediately frozen in liquid nitrogen upon being isolated from adult mice and embed in O.C.T. compound. Cryostat sections (10 μm) were prepared on a cryostat microtome. Cells cultured in 12-well plates or on cryostat slides were fixed on ice for 10 min using 4% paraformaldehyde, followed by permeabilization with PBST solution (0.5% Triton X-100 in PBS) for 15–20 min. After blocking with goat serum, immunostaining with specific antibodies (mentioned in Table S4) was performed, followed by counterstaining with DAPI.

Intraperitoneal injection of EdU (Sigma, Shanghai, China) in PBS was performed 4 h before killing the mice (50 mg/kg). Tibialis anterior muscles were then isolated, frozen and sectioned at 10 µm. Cryostat slides were fixed and permeabilized as described above. Finally, EdU detection was performed using an in vivo EdU Click Kit 488 (Sigma, Shanghai, China) as per the manufacturer's instruction.

C2C12 cells were digested with tripsin and fixed in 70% ethanol overnight. After three washes in PBS, fixed cells were treated with propidium iodide solution (10 mg propidium iodide, 0.5 ml Triton X-100, 200 mg sodium citrate and 129.6 ml PBS in 200 ml, pH 7.2–7.6) for 30 min in the dark. Then, cells were dissociated by pipetting up and down gently, and analyzed using a BD FACSCalibur system (BD Biosciences, Franklin Lakes, USA).

Cell proliferation was evaluated by an impedance-based RTCA xCELLigence DP system (ACEA Biosciences, CA) that can monitor real-time cell proliferation. Cells were seeded into an E-Plate 16 (ACEA Biosciences) with 200 μl growth medium and allowed to grow for 62 h at 37°C in a 5% CO₂ atmosphere. Cellular impedance was detected every 6 h. The data collected from cell-electrode impedance reflects the cell proliferation index.

Total RNA was extracted from CTX-injured tibialis anterior muscles using TRIzol^® reagent as per the manufacturer's instruction. Sequencing analysis was performed with a Illumina Genome Analyzer (Illumina, CA) according to the manufacturer's instructions.

All antibodies used in current study are provided in Table S4. Previous studies have demonstrated that all of them are validated for use (Aguilera et al., 2011; Averous et al., 2012; Della Pietra et al., 2015; Dixon et al., 2011; Gong et al., 2015; Kim et al., 2011; Li et al., 2012; Murphy et al., 2011; Nissar et al., 2012; Rao et al., 2014; Sarkar and Zohn, 2012; Takeda et al., 2014; Tan et al., 2011; Taniguchi et al., 2011; Wu et al., 2010; Zhang et al., 2011).

All data are presented as the mean±s.e.m.; significance of differences in comparisons were determined by a Student's t-test. Values of P<0.05 were considered as statistically significant.

Tong Jiang is acknowledged for qPCR experiments and vector construction.