The m6A reader YTHDC1 regulates muscle stem cell proliferation via PI4K–Akt–mTOR signalling

Abstract Muscle stem cells are required for the homeostasis and regeneration of mammalian skeletal muscles. It has been reported that RNA N6‐methyladenosine (m6A) modifications play a pivotal role in muscle development and regeneration. Nevertheless, we know little about which m6A reader regulates mammalian muscle stem cells. Here, we discovered that the m6A reader Ythdc1 is indispensable for mouse skeletal muscle regeneration and proliferation of muscle stem cells. In the absence of Ythdc1, Muscle stem cells in adult mice are unable to exit from quiescence. Mechanistically, Ythdc1 binds to m6A‐modified Pi4k2a and Pi4kb mRNAs to regulate their alternative splicing and thus PI4K–Akt–mTOR signalling. Ythdc1‐null muscle stem cells show a deficiency in phosphatidylinositol (PI) 3,4,5‐trisphosphate, phospho‐Akt and phospho‐S6, which correlates with a failure in exit from quiescence. Our findings connect dynamic RNA methylation to the regulation of PI4K–Akt–mTOR signalling during stem cell proliferation and adult tissue regeneration.


| INTRODUCTION
Skeletal muscle stem cells, also known as satellite cells (SCs), are responsible for maintenance and repair of muscle tissue throughout lifespan. 1 Muscle stem cells are located between sarcolemma of myofibers and the surrounding basal lamina in muscle tissue. 2 Adult SCs which remain mitotically quiescent under normal condition characteristically express a paired box transcription factor Pax7. 3 Responding to exercise and injury, the quiescent SCs is first activated and then enter cell cycle to proliferate which results in an increase in number. 4 Following proliferation, the offspring of SCs differentiate and fuse to form multinucleated myotubes or fuse to damaged myofibers to repair injured muscle tissue. 4,5 Besides, a minority of the proliferating SCs return to dormant state to replenish the stem cell pool. 5 It was documented that SCs fate is governed by a number of intrinsic and extrinsic signals including Notch, Wnt, mTOR, p38/MAPK, insulin-like growth factor, fibroblast growth factor (FGF) and others. [6][7][8][9][10][11] However, regulation of SCs fate is not fully understood.
Dynamic and reversible N6-methyladenosine (m6A) is one of the most prominent post-transcriptional modifications on the RNA molecule in eukaryotic cells. 12 The m6A modification is catalysed by a methyltransferase complex consisting of Mettl3 and other cofactors (writer). 12 The reversible RNA modification can be erased by demethylases, Fto and Alkbh5 (eraser). 12 Readers of m6A-modified RNA (reader) preferentially bind to target mRNA to control cellular fate of modified mRNA, including gene expression, mRNA stability, mRNA translation, splicing and nuclear export of mRNA. 13 Similar to epigenetic modifications, m6A methylation of mRNA is involved in a variety of biological processes such as cell expansion, cell differentiation, cell apoptosis and cell migration and invasion. 14 A number of studies showed that m6A modification is important for cell fate decision of somatic stem cells and embryonic stem cells, maintenance and development of cancer stem cells. [15][16][17][18][19][20][21] In mouse skeletal muscle, the global levels of m6A of regenerating muscle tissue decline during the transition of SCs from proliferation to differentiation. 22 Mettl3 regulates cell differentiation of C2C12 myoblast. 22,23 In SCs-specific Mettl3 knockout (KO) mice, regenerative capacity of skeletal muscle is impaired severely due to the failure in quiescence exit of SCs. 24 Fto is demonstrated to be involved in myogenesis which depends on its demethylase activity. 25 All of these research results indicate that m6A plays a critical role in myogenesis of SCs and myoblast.
Ythdc1 is an identified reader of m6A-modified RNA which is localized in the nucleus. 26 Depletion of Ythdc1 in mice leads to embryonic lethality. It was documented that Ythdc1 is involved in mRNA splicing, polyadenylation, nuclear export and degradation. [27][28][29] It plays a pivotal role in mouse oocyte development, maintenance of mouse ES cells, normal and malignant haematopoiesis in a m6Adependent manner. 27 Figure S1A). Ythdc1 was nearly undetectable in both mRNA and protein level in SCs in KO mice ( Figure 1B,C). RNA sequencing (RNA-seq) data also confirmed that Exons 5-7 of Ythdc1 was depleted ( Figure 1A). To further assess the function of Ythdc1 in SCs in vivo, acute muscle injury-regeneration model was employed ( Figure 1D). At Days 7 and 28 post-injury, the regenerated TA muscle was smaller, and mass of regenerated TA decreased in KO mice compared with WT littermates ( Figures 1E and S4A F I G U R E 2 Legend on next page. were observed than that in WT mice ( Figures 1F and S4C). Consistently, immunofluorescence (IF) staining of eMHC (embryonic myosin heavy chain), which is expressed in newborn myofibers, manifested that eMHC + regenerating myofibers were rarely seen in KO mice 5 days after injury ( Figure 1G). Taken together, our data demonstrate that deletion of Ythdc1 in SCs leads to severe impairment of skeletal muscle regeneration in vivo.

| Ythdc1 is required for the proliferation of SCs in vivo and in vitro
To explore the reason of regeneration failure in KO mice, we quantified the number of SCs in injured mice muscle by FACS. Three and a half days after injury, we observed dramatic decrease in frequency of SCs in injured hind limb muscle in Ythdc1 KO mice ( Figure 2B). And the expression of genes related to myogenesis in Ythdc1-null SCs, such as MyoG, Tnni2, Acta1 and Myh3, was much lower than that in WT SCs after injury ( Figure S1C).
As we know, in response to acute muscle damage, the quiescent SCs exit dormancy, undergo proliferation, and then differentiate and fuse to form newborn multinucleated myotubes. 4 Thus, at first, we examined 5-ethynyl-2 0 -deoxyuridine (EdU) incorporation in vivo upon injury. Nearly, 40% of SCs isolated from injured WT mice were labelled by EdU, whereas only about 5% of Ythdc1-null SCs were labelled ( Figure 2D). Ki67 staining of SCs from injured mice confirmed cell cycle arrest in Ythdc1-null SCs, more than 80% of SCs in injured WT mice expressed Ki67 protein highly, <20% of SCs from KO mice did so ( Figure 2E). Furthermore, the size of SCs isolated from WT mice was much bigger than that isolated from KO mice upon injury ( Figure 2C).
In addition, SCs were isolated freshly from uninjured mice by FACS at Day 9 post-TMX treatment, and cultured in growth medium to determine EdU incorporation and ki67 expression in vitro. First, we confirmed that almost 97.1% cells sorted by FACS are Pax7-positive SCs ( Figure S1D). The results showed that the number of EdU + cells in Ythdc1-null SCs decreased remarkably (WT [55.4%] vs. KO [6.8%]; Figure 2F). When stained for Ki67 in cultured SCs, the abundance of Ki67 + SCs were declined dramatically in Ythdc1-null SCs (WT [58.9%] vs. KO [6.5%]; Figure 2G). By Hoechst 33342 and Pyronin Y staining for DNA/RNA content, the results showed that much more cultured SCs isolated from KO mice were blocked in G0 state with low RNA content compared with that from WT mice ( Figure 2H,I). Taken together, our data above indicate that Ythdc1 is indispensable for proliferation of SCs in vivo and in vitro.

| Ythdc1 regulates splicing and promotes intron retention in SCs
To dissect the underlying mechanism, RNA-seq was performed with RNAs extracted from SCs isolated from uninjured mice. Given that the key role of Ythdc1 is to regulate alternative splicing of its target mRNA, 32 we applied rMATS (replicate multivariate analysis of transcript splicing) to detect differential alternative splicing in the absence of Ythdc1 in SCs. 33 We identified a total of 3437 significantly differential alternative splicing events (ASEs) including SE, RI, MXE, A5SS and A3SS ( Figure 3A,B). We found that 26.19% of detected retained intron events showed significant differences in SCs between WT mice and KO mice ( Figure 3B). And in Ythdc1-null SCs, inclusion level of intron decreased distinctly, whereas other types of ASE did not  between Exons 9 and 10 of Pi4k2a preferred to retain in WT SCs ( Figure 4B). Taken together, Ythdc1 regulates mRNA splicing and promotes intron retention in SCs.

| Akt-mTOR pathway is interfered in Ythdc1-null SCs
PI lipids have been shown to be crucial signal molecules in response to extracellular growth factor. PI and its derivatives involve diverse cellular processes including cell proliferation, cell differentiation, cell death, membrane trafficking, cell-matrix adhesion and gene expression. 34,35 PI 4-kinase (PI4K) played a critical role in PI metabolism which had an effect on the production of PIP3 ( Figure 4C). 36,37 Our data showed that the transcription initiation site in the second exon of Pi4kb was skipped, which may give rise to loss of function of Pi4kb.
Intron retention of Pi4k2a altered 3 0 untranslated region, which may regulate Pi4k2a feature. Thus, PIP3 content in Ythdc1-null SCs may be lower than that in normal SCs. Without enough PIP3, activation of Akt-mTOR would be disturbed, which leads to failure in quiescence exit in SCs. 11,[38][39][40] To test whether Akt-mTOR was interfered, we determined phosphorylation of Akt in SCs by FACS at Day 2 post-BaCl 2 -induced muscle injury in vivo. The results showed that activated Akt in Ythdc1-null SCs, phosphorylated Akt on serine 473 (pAkt-473), was less than that in control cells ( Figure 4D,E). In cultured SCs isolated from mice, pAkt-473 was also downregulated in the absence of Ythdc1 in SCs ( Figure S2G). Next, we quantified phosphorylated ribosomal S6 protein (p-S6), a direct target of Akt-mTOR pathway, by FACS at same time point. The protein level of pS6 of Ythdc1-null SCs was distinctly lower than that in control cells in vivo ( Figure 4F,G). Consistently, immunofluorescence staining and western blot analysis of p-S6 in cultured SCs for 36 h in vitro confirmed interference of Akt-mTOR pathway ( Figure S2E,F). Taken together, activation of Akt-mTOR pathway in Ythdc1-null SCs is disturbed, which is responsible for failure in quiescence exit.
To verify the hypothesis that depletion of Ythdc1 in SCs decreased the production of PIP3 via regulation of Pi4kb and Pi4k2a directly, at first, we determined the abundance of PIP3 in SCs via immunofluorescence staining. The results revealed that PIP3 content in Ythdc1-null SCs was significantly reduced compared to that in control cells ( Figure 4H,I). Next, we performed RNA immunoprecipitation (RIP)-RT-qPCR assay with antibody against Ythdc1 to examine the interaction between Ythdc1 protein and target mRNAs. The results validated that Ythdc1 markedly enriched Pi4kb and Pi4k2a mRNA compared with the immunoglobulin G (IgG) pull-down control in C2C12 myoblast ( Figure 4J), demonstrating that Ythdc1 interacted with Pi4kb and Pi4k2a mRNA. Furthermore, we performed m6A methylation peak calling using algorithm exomePeak with MeRIP-seq data from previous study. 23,41 We uncovered that m6A modification of Pi4kb and Pi4k2a enriched in the exon, nearby which ASEs could be found ( Figure 4K,L). In addition, overexpression of WT Ythdc1, not the YTH-domain-mutants which impede recognition and binding of m6A modified RNA and Ythdc1, rescued the proliferation deficiency of Ythdc1-null SCs, which implicated that Ythdc1 regulates the function of SCs in a m6A-dependent manner (Figures S5A,B). Taken together, we demonstrate that Ythdc1 binds to m6A-modified Pi4kb and Pi4k2a to regulate production of PIP3, which leads to failure in activation of Akt-mTOR pathway in Ythdc1-null SCs.

| DISCUSSION
In this study, we used conditional KO mice to uncover that Ythdc1, one of the readers of RNA m6A modifications, is required for SC proliferation to repair damaged skeletal muscle tissues. Our results are consistent with conclusions from previously published reports, 24,25 which showed that the m6A writer Mettl3 regulates the proliferation of muscle stem cells during mouse muscle regeneration, and that the m6A eraser Fto controls mouse muscle development. Although previous studies implicated m6A modification in myogenesis, we did not observe obvious developmental defects in SC-specific Ythdf1-null mice. Our previous study implicated that loss of Ythdf2 enhances regenerative capacity of adult HSC, 42,43 but showed no difference in muscle regeneration. It is also unclear exactly how Fto regulates the mTOR pathway in myogenesis. 25 Our results suggest at least one possible mechanism is that m6A-modified Pi4k2a and Pi4kb mRNAs regulate PI4K-Akt-mTOR signalling. These results indicate that a sophisticated system exists to mediate m6A modifications' effects on biological processes. More studies will be needed to dissect the role of writers, erasers and readers of m6A modifications in different contexts, including skeletal muscle development and regeneration.
Ythdc1 is a reader protein of m6A modification which is localized in the nucleus. It is involved in mRNA transcription, splicing, nuclear export, polyadenylation and degradation. 27,28,32 In this study, we found that Ythdc1 promotes intron retention in SCs in general. Specifically, for Pi4k2a and Pi4kb, Ythdc1 binds directly to their m6Amodified mRNAs to regulate their alternative splicing. PI4K is a key regulator of PI metabolism, 36 and indeed we observed that PIP3 levels are deficient in Ythdc1-null SCs. It is known that PIP3-associated metabolites play a central role in signal transduction in cellular processes. 44 Importantly, disorder of PIP3 metabolism correlates with a variety of human pathologies including neurodegeneration, primary immunodeficiencies and cancer. 35,45 Thus, it is possible that Ythdc1 regulation of PI4K alternative splicing and PIP3 metabolism could play important roles in other contexts as well, and serve as a potential therapeutic target for PIP3-related diseases.
In summary, we discovered that Ythdc1 is indispensable for mouse muscle regeneration. Upon loss of Ythdc1, SCs are unable to exit from quiescence. Ythdc1 binds to m6A-modified Pi4k2a and Pi4kb mRNA to regulate their alternative splicing and thus PI4K activity and PIP3 metabolism. Depletion of Ythdc1 in SCs results in decreased production of PIP3. PIP3, a critical second messenger, triggers Akt phosphorylation. Therefore, Ythdc1 regulates SCs' exit from quiescence to proliferate during muscle regeneration, by modulating PI4K alternative splicing, PIP3 levels, and Akt-mTOR signalling. Our

| Muscle injury and regeneration model
Adult mice were anaesthetized with isoflurane during surgery. Muscle injury and regeneration was induced by intramuscular injection of 50 μL of 1.2% (wt/vol) BaCl 2 solution into TA muscles. After injury, TA muscles were harvested at various time points for further analysis.

| Isolation of adult muscle SCs by FACS
Adult muscle SCs were isolated as previously described. 46 Briefly, Hind limb muscles dissected from mice were gently minced, digested by collagenase II (800 U/mL) in washing medium (Ham's F10 with 10% horse serum) for 90 min in a shaking water bath at 37 C. Then,  Table S1.  Resources Table). The membranes were then incubated with secondary antibodies (Key Resources Table) for 1 h at RT, and detected on an Amersham Imager 600 System (GE).

| H&E staining and immunofluorescence staining
TA muscle tissues from WT or KO mice were isolated and dehydrated in 30% (wt/vol) sucrose solution. Then, the dehydrated TA muscles were frozen in optimal cutting temperature (OCT) compound. Cross sections (10 μm) of frozen TA muscle were used for H&E staining and immunofluorescence staining. For H&E staining, the slides were first stained in haematoxylin for 8 min, washed in running tap water for 5 min and then counterstained in eosin-phloxine solution for 1 min.
The specimens were dehydrated in graded ethanol (70%-100%) and Xylene, and then covered with mounting medium. For immunofluorescence staining of cryo-sections, immunofluorescence staining was performed as described in the article. 47 Immunofluorescence staining of Ki67 and Pax7, cells were fixed with 4% paraformaldehyde (PFA) for 15 min at RT. Then cells were permeabilized with 0.5% Triton X-100 in phosphate buffer solution (PBS) for 20 min, blocked with 3% IgG-free bovine serum albumin (BSA) and 0.1% Triton X-100 in PBS for 1 hour (blocking solution) at RT, followed by incubation with antibody against Ki67 (Abcam) or Pax7 (DSHB) at 4 C overnight. After washing with PBS, the samples were incubated with secondary antibodies (Key Resources Table) for 1 h at RT. Immunofluorescence staining of pS6 (CST) was performed according to the manufacturer's instructions. The nuclei were counterstained with DAPI.

| EdU assay in vivo and in vitro
In vivo, by Day 2 post-injury, mice were injected intraperitoneally with EdU (Beyotime, 0.05 mg/g body weight), followed by FACS iso-

| Validation of ASEs
For validation of ASEs, RT-PCR were conducted. The PCR cycles varied for each ASE because of the different abundance of transcript.
Primers used for RT PCR are shown in Table S1.

| Analysis of protein or PIP3 level by flow cytometry
Satellite cells were isolated as described above. FITC anti-Sca1 instead of PE/Cy7 anti-Sca1 and APC-Cy7 streptavidin instead of PE streptavidin were used for the staining. Then immunofluorescence staining of intracellular phosphor-Akt (Ser473; CST) and pS6 (CST) for flow cytometric analysis were performed according to the manufacture's protocol. Immunofluorescence staining for PIP3, cells stained with above antibodies were fixed in 4% formaldehyde for 20 min at RT, washed three times with PBS, and then permeabilized with 0.2% Triton X-100 in PBS for 10 min at RT. Samples were blocked with 3% IgG-free BSA and 0.1% Triton X-100 in PBS for 1 h at RT, followed by incubation with antibody against PIP3 (Echelon Biosciences) or anti mouse IgG isotype control (Santa Cruz) at RT for 90 min. After washing with PBS, the samples were incubated with secondary antibodies (Key Resources Table) for 30 min at RT followed by FACS analysis.

| RIP real-time qPCR
C2C12 myoblast cultured in DMEM with 10% FBS, 1% penicillin/ streptomycin were collected for further study. All the specific manipulations were performed according to the protocol of Imprint RIP Kit (RIP-12RXN, Sigma-Aldrich). The samples were incubated with antibody against Ythdc1 (CST) or anti Rabbit IgG isotype control (CST).
The primers for RIP real-time qPCR were listed in Table S1.

| Analysis of m6A methylation sites by exomePeak algorithm
First, we downloaded the online raw data from DNA Data Bank of Japan database (accession no: DRA005057). Raw sequencing reads were mapped to reference mouse genome (mm10) using Hisat2 software. PCR duplication and low-quality reads were removed by using SAMtools software. Mapped and filtered reads were subjected to the methylation peak caller exomePeak to predict m6A methylation sites.

| RNA-seq and data analysis
Adult Ythdc1 flox/flox mice and their Ythdc1 flox/flox : Pax7CreERT2 littermates were treated with TMX for 5 consecutive days. Nine days after the last TMX injection, SCs were isolated by FACS as described above.
RNA extraction and sequencing libraries were prepared as described in Zhencan Shi et al. 48 Sequencing was performed on NovaSeq platform (pair-end with 150 bp). Sequencing reads were mapped to the mouse genome assembly (mm10) using Hisat2 software. Replicate multivariate analysis of transcript splicing (rMATS) was applied to RNA-seq data to detect differential ASEs. ASEs (false discovery rate (FDR) <0.05) with j IncLevelDiff j (KO-WT) > 0.1 were considered significant difference.
Sequencing data of the experiment are deposited at NIH's Sequence Read Archive and have been assigned a BioProject accession number PRJNA776399.

| QUANTIFICATION AND STATISTICAL ANALYSIS
All data represent mean ± SD. Statistical analysis was performed using unpaired two-tailed Student's t-test. For all tests, p-values < 0.05 were considered significant.