Long noncoding RNA SAM promotes myoblast proliferation through stabilizing Sugt1 and facilitating kinetochore assembly

The functional study of lncRNAs in skeletal muscle satellite cells (SCs) remains at the infancy stage. Here we identify SAM (Sugt1 asssociated muscle) lncRNA that is enriched in the proliferating myoblasts. Global deletion of SAM has no overt effect on mice but impairs adult muscle regeneration following acute damage; it also exacerbates the chronic injury-induced dystrophic phenotype in mdx mice. Consistently, inducible deletion of SAM in SCs leads to deficiency in muscle regeneration. Further examination reveals that SAM loss results in a cell-autonomous defect in the proliferative expansion of myoblasts. Mechanistically, we find SAM interacts and stabilizes Sugt1, a co-chaperon protein key to kinetochore assembly during cell division. Loss of SAM or Sugt1 both disrupts kinetochore assembly in mitotic cells due to the mislocalization of two components: Dsn1 and Hec1. Altogether, our findings identify SAM as a regulator of SC proliferation through facilitating Sugt1 mediated kinetochore assembly during cell division.

S keletal muscle has a robust regenerative capacity, which mainly relies on the activation of resident muscle stem cells, termed satellite cells (SCs). These cells are uniquely marked by the expression of paired box 7 (Pax7) protein and normally lie in a niche beneath the basal lamina of myofibers in a quiescent stage. Upon injury, they are rapidly activated to enter the cell cycle and undergo proliferative expansion as myoblast (MB) cells which then differentiate and fuse to form multinucleated myotube (MT) cells; these myotubes further mature into myofibers to restore the damaged muscle. Meanwhile, a subset of SCs exit the cell cycle and return to the quiescent stage for replenishing the adult stem cell pool. Fine-tuned regulation of cell cycle is thus essential to ensure appropriate progression through the various overlapping states: activation, proliferation, differentiation, and self-renewal/returning to quiescence.
The cell cycle involves DNA replication and subsequent chromosome separation. The faithful chromosome segregation relies on the assembly of mitotic kinetochore on centromeric chromatin to mediate its interaction with spindle microtubules 1 . In vertebrates, the kinetochore is a multilayered disc structure that contains more than a hundred of proteins components 2 . CCAN, the constitutive centromere-associated network, is restricted to the centromeres throughout the cell cycle forming a major component of inner kinetochore whereas KMN network, including the KNL1 complex (containing Knl1 (kinetochore scaffold 1), and Zwint (ZW10 interactor)), the MIS12 complex (containing Mis12, Dsn1, Pmf1 (polyamine-modulated factor 1) and Nsl1), and the NDC80 complex (containing Ndc80 (also called Hec1), Nuf2, Spc24, and Spc25), is recruited to the centromere by the CCAN during specific stages of mitosis, forming prominent subunits of outer kinetochore 3,4 . Among the KMN complexes, Mis12 complex is the keystone to serve as a protein interaction hub which assembles outer kinetochore and links to the inner kinetochore 5 . Ndc80 complex directly interacts with microtubules through its component Hec1 6 . Given the large number of kinetochore components, its proper assembly is a dynamic and highly orchestrated process. Any error in kinetochore assembly such as improper targeting or turnover of any component can affect the progression of mitosis, leading to disrupted microtubule attachment, improper chromosomal segregation, the formation of multipolar spindles, mitotic delay or aneuploidy, etc. [7][8][9] . It is thus important to elucidate the regulatory mechanisms facilitating kinetochore assembly, which has not been done in SCs. It is known that SGT1, suppressor of G2 allele of SKP1 (S. cerevisiae) (Sugt1) is a highly conserved protein involved in kinetochore assembly 10 . As a co-chaperone for Hsp90 protein, mammalian Sugt1 ensures efficient formation of microtubule-binding sites by recruiting Mis12 complexes to kinetochore 11 . Reduction of Sugt1 in Hela cells leads to destabilization and mis-localization of Dsn1 and Hec1, thus causing inefficient formation of high-affinity kinetochore-microtubule attachment sites and a mitotic delay 10,11 . A recent study also showed that a regulatory phosphatase PHLPP1 dephosphorylates Sugt1 thereby prevents Sugt1 from associating with E3 ligase in turn, countering Sugt1 ubiquitination and degradation during kinetochore formation 12 .
Long non-coding RNAs (lncRNAs) are emerging as a family of gene regulators of skeletal muscle regeneration and SC activities. Thousands of lncRNAs have been identified in skeletal muscle cells but our understanding of lncRNA participation in skeletal myogenesis is still at the infancy stage with only a handful of reports from our group and others [13][14][15][16][17][18] . Most efforts concentrated on illuminating their regulatory mechanisms in the transition of MB into MT using a mouse MB line, C2C12 [13][14][15] ; it remains largely uncharacterized whether lncRNAs can regulate other states of SCs. In terms of underpinning molecular mechanisms, lncRNAs are best known for engaging in transcriptional and epigenetic regulation on chromatins, usually through their interaction with chromatin regulators 19 ; other unique mechanisms are also being uncovered to explain the diversified modes of lncRNA actions. For example, recently, lncRNAs generated from the repeat region of centromere in Drosophila and human, were found to bind to the kinetochore component CENP-C, adding lncRNA to the complex epigenetic marks at centromeres 20,21 . Still, it is not known whether noncentromeric lncRNAs exist to interact with proteins involved in kinetochore assembly. Additionally, in vivo functional analysis is in general lacking for most lncRNAs studied so far despite a wealth of knowledge accumulated from using in vitro cell culture; to date there have been only a few lncRNA genetically studied using knockout (KO) animals 22,23 .
Here, we have identified one lncRNA, SAM, as a regulator of MB proliferation. Its expression is evidently upregulated when SCs undergo active proliferation; knockdown of SAM in vitro delays proliferative expansion of cells. To further investigate its function in vivo, we generated a KO mouse of SAM using KOfirst strategy; loss of SAM does not cause overt phenotype but indeed leads to impaired regeneration after acute injury. Consistently, inducible deletion of SAM in SCs also delays the process of acute injury-induced muscle regeneration. Moreover, deletion of SAM in a dystrophic mdx mouse exacerbates the chronic injury-induced dystrophic phenotype. Further examination reveals that SAM deletion results in the cellautonomous defect in MB proliferation, pointing to SAM as a promoting factor of MB proliferation. High throughput identification of SAM interacting protein partners reveals that it can specifically bind to Sugt1 and stabilizes its protein level in MBs; loss of SAM causes increased ubiquitination of Sugt1. Mechanistically, SAM facilitates Sugt1-mediated kinetochore assembly. Loss of SAM or Sugt1 both causes disrupted chromosome alignment and microtubule attachment, which is likely a result of mis-localization of Dsn1 and Hec1 proteins in centromere. Altogether our findings have identified SAM as a regulator of MB proliferation through its synergistic action with Sugt1 to promote kinetochore assembly during cell division.

Results
SAM is enriched in MB and promotes cell proliferation. Previously we have defined dozens of uncharacterized lncRNAs from C2C12 MB vs. MT cells through de novo discovery approach integrating RNA-seq and ChIP-seq datasets 13 . One lncRNA, Gm11974, named as Sugt1 Associated Muscle (SAM) lncRNA in the present study, displayed relatively high expression and unexplored function in MB cells (Fig. 1a). It localizes on mouse chromosome 11, in the intervening region of Myo1g (Myosin IG) and Ccm2 (Cerebral cavernous malformation 2) protein-coding genes (Fig. 1b), with well-defined gene structure and a binding peak of myogenic master transcription factor, MyoD on its promoter region (Fig. 1a). A human homolog of this gene, SNHG15, has been studied in cancer, showing upregulated expression in multiple tumor tissues or cells [24][25][26] and it promotes cancer cell proliferation and migration by serving as a sponge for miR-NAs [27][28][29] . Through rapid amplification with cDNA ends (RACE), one dominant isoform was cloned from C2C12 MB cells, which was 592 bp long with four exons (Fig. 1c). It was predicted as a non-coding RNA by iSeeRNA 30 (Supplementary Fig. 1a), consistent with its annotation in the RefSeq (Accession no. NR 045893). SAM was readily detected in C2C12 MBs and downregulated when the cells underwent differentiation to form MTs ( Supplementary Fig. 1b) primary MBs isolated from the skeletal muscle compared with the whole muscle tissue (Fig. 1d). To further examine its expression dynamics during SC lineage progression, freshly isolated SCs (FISCs) from limb muscles of Pax7-nGFP mice 31 were cultured with growth medium to become activated (ASCs or MBs) which were further cultured to differentiate (DSCs); SAM level was evidently induced (4.7 fold) in ASCs vs. FISCs but decreased sharply (72.71%) in DSCs vs. FISCs (Fig. 1e). Interestingly, SAM expression appeared not to be heterogeneous in SCs, since no significant difference was detected in the isolated Pax7 High and Pax7 Low subpopulations 32 of FISCs or ASCs ( Supplementary  Fig. 1c, d). The above results suggested that SAM might promote MB proliferation. RNA fluorescence in situ hybridization (RNA-FISH) analysis revealed that SAM transcripts mainly distributed in the cytoplasm of SC (Fig. 1f); a stronger signal was detected in ASC vs. FISC or DSC. Similarly, the predominant cytoplasmic localization was also observed in C2C12 MB but decreased in MT ( Supplementary Fig. 1e). Consistently, cellular fractionation assay in ASCs (Fig. 1g) or C2C12 ( Supplementary Fig. 1f) also showed that SAM transcripts were enriched in cytoplasmic extracts, in a similar pattern as Gapdh transcripts, whereas lncRNA Malat1 was only found in nuclear extracts 16 . The unique cytoplasmic localization of SAM suggested that its function may be distinct from many lncRNAs that are involved in transcriptional regulation in myogenesis 17 , which therefore triggered our further investigation.
To test if SAM is required for efficient MB proliferation, we knocked down SAM expression in ASCs with two different siRNA oligos (22.98% and 33.27% reduction, respectively) ( Fig. 1h). Forty-eight-hour post-transfection, SCs were stained for Pax7 and MyoD to evaluate the degree of proliferation. It is known that fully activated SCs are marked by both Pax7 and MyoD while self-renewing SCs express Pax7, but not MyoD; In DSCs, Pax7 expression is lost while Myogenin (MyoG) expression increases 33 . Indeed, the percentage of Pax7+MyoD + cells was markedly reduced upon siSAM knockdown (14.43% and 29.85%, respectively) ( Fig. 1i, Supplementary Fig. 1g and h). This was further confirmed by staining for Ki67; the percentage of Ki67+ SCs was decreased upon SAM loss (48.11% and 55.25%) (Fig. 1j). On the contrary, when overexpressing SAM by transfecting a SAM-expressing plasmid (Fig. 1k), an increase in the percentage of Pax7+MyoD+ (7.16%) or Ki67+ (74.89%) cells was observed (Fig. 1l, m, Supplementary Fig. 1i and j). Altogether, the above results from loss and gain-offunction assays in vitro on SCs demonstrated that SAM promotes MB proliferation. When performing similar assays using C2C12 MB cell line with stable SAM knockdown by a shRNA, the same overall conclusions were reached (Supplementary Fig. 1k-p). In addition, by cell cycle analysis of synchronized cells, shSAM cells displayed a higher percentage of cells in G1 phase at both 12 h and 24 h compared with control (Ctrl) cells, suggesting SAM loss caused cell cycle arrest at G1 phase ( Supplementary Fig. 1q); nevertheless, SAM expression did not show dynamic pattern during the cell cycle progression ( Supplementary Fig. 1r). Next, we also examined whether the loss of SAM has any effect on MB differentiation. Single myofibers were isolated from extensor digitorum longus (EDL) muscle of mouse and transfected with siSAM. Staining with MyoG and MyoD 72 h post-transfection revealed that the percentage of MyoD+MyoG+ cells was significantly increased (16.43% and 19.53%) upon SAM knockdown (Fig. 1n, Supplementary Fig. 1s and t), implying that these cells may have precocious differentiation potential. Collectively, these findings from the in vitro cultured cells indicate that SAM is necessary for maintaining proper myogenic proliferation and preventing precocious differentiation.
SAM deletion in mouse impairs muscle regeneration. To further elucidate the functional roles of SAM in vivo, we generated a KO mouse. Given that lncRNA locus may function as an enhancer region to regulate gene expression 34 and active enhancer mark, H3K27ac, was indeed found on SAM locus (Fig. 1a), we thus employed a KO-first strategy that ablates gene function by inserting RNA processing signals without deletion of the target locus. As illustrated in Fig. 2a, the KO-first allele was generated by inserting a splicer acceptor (SA)-internal ribosomal entry site (IRES)-LacZ cassette and a Neo-polyadenylation (pA) signal into the intron 2, thus achieving the disruption of SAM transcription. The insertion was flanked by FRT sites that will allow Flippase recombinase to remove the gene-trapping cassette, hereby converting the KO to a conditional allele with loxP sites flanking exons 3-4. DNA genotyping confirmed the insertion of the SA-IRES-LacZ-pA cassette in the KO mouse genome ( Supplementary  Fig. 2a, b). Three qRT-PCR primers targeting different regions (exons 1-2, exons 2-3, and exons 3-4) were used to detect possible transcription (Fig. 2a); and no transcripts were detected with any pair of primers in the isolated SCs (Fig. 2b) or tested tissues ( Supplementary Fig. 2c). It is interesting that no truncated transcript from exons 1-2 was detected despite the PolyA was inserted after exon 2. To test if non-sense-mediated RNA decay had possibly led to degradation of the transcript, we found treatment with cycloheximide (CHX), which is known to reverse non-sense-mediated RNA decay 35 induced the appearance of the truncated transcript from exons 1-2 ( Supplementary Fig. 2d). Examining the adult mouse phenotype, we found that the KO mice were viable, fertile without overt morphological deformities (Fig. 2c); consistently, the size and weight of the KO mice were comparable with the WT littermates (Fig. 2c, d). Histological analyses of adult tissues including liver, spleen, lung, kidney, and ovary also revealed no overt differences between the KO and WT littermates ( Supplementary Fig. 2e); similarly, when examining the adult skeletal muscle at 8 weeks, the fibers also appeared normal in size and pattern (Fig. 2e), showing the deletion of SAM may not have any impact on the adult muscle development. In addition, the number of Pax7+ quiescent SCs (QSCs) was not changed in the muscles of the KO mice (Fig. 2f), indicating that SAM may not be required for maintenance of the SC pool. Lastly, examining muscle formation at embryonic (E18.5) (Supplementary Fig. 2f, g) or postnatal (P7) days ( Supplementary Fig. 2h) revealed no overt changes in muscle morphology and the number of Pax7+ cells in WT vs. KO mice, suggesting that SAM may not play a role in embryonic or postnatal myogenesis.
Considering the promoting function of SAM that was uncovered above in proliferating MB in vitro ( Fig. 1), we speculated that loss of SAM may have an impact on muscle regeneration in vivo. To test this notion, BaCl 2 was injected into the tibialis anterior (TA) muscles of 8-9 weeks old mice to induce massive myofiber necrosis followed by immune cell infiltration, activation, and proliferation of SCs, which then formed new myofibers to repair the damaged fibers within 3-4 weeks post injection. The newly formed myofibers were normally characterized by centrally localized nuclei (CLN) and expression of embryonic MyHC (eMyHC) protein. The above injected muscles were collected 4, 7, 14, and 28 days after the injury for evaluation of the degree of muscle regeneration (Fig. 2g). Indeed, by H&E staining, the number of CLN+ fibers per field was evidently decreased (32.29%) in KO vs. WT mice 4 days after the injury (Fig. 2h); consistently, the number of eMyHC+ fibers was also decreased by 20.16% (Fig. 2i). Nevertheless, by day 7, no significant difference was found in KO vs. WT mice; by day 28, the damaged muscle fibers were fully regenerated in both mice (Fig. 2h, i). The above results indicate that SAM deletion causes a delay but not a blockage in injury-induced muscle regeneration.
In addition, we found that Pax7+ or MyoD+ cells were both reduced significantly (39.44% and 38.81%, respectively) on the KO muscles compared with WT (Fig. 2j, k) 3 days after injury, suggesting a decline in the expansion of SC progeny during the regeneration process. Lastly, we quantified the number of Pax7+ cells one month after injury when SCs were expected to return to quiescence ( Supplementary Fig. 2i); no significant difference was observed from the injured muscles of KO vs. WT mice ( Supplementary Fig. 2j), suggesting SAM ablation may not exert apparent influence on SC self-renewal during muscle regeneration.
Inducible ablation of SAM in SC delays muscle regeneration. To further pinpoint that the above described regeneration phenotype is attributed to the loss of SAM in SCs, we further generated a SC-specific inducible knockout (iKO) mouse. As illustrated in Fig. 2l, SAM floxed mice (SAM fl/fl ) were created by crossing the KO with a FLPeR recombinase mouse, which led to the excision of the SA-IRES-LacZ-pA cassette flanked by FRT sites ( Supplementary Fig. 2k, l). Further breeding with a Pax7 creER mouse 36 to generate Pax7 creER/+ ; SAM fl/fl mouse (termed SAM iKO) led to permanent deletion of exons 3-4 of SAM in the adult Pax7+ SCs following five consecutive doses of tamoxifen (TM) injection in 2-month-old mouse (Fig. 2m); the successful elimination of exons 3-4 of SAM was confirmed (Fig. 2n); interestingly, a truncated transcript was generated from exons 1-2 ( Fig. 2n). Consistent with what was observed in the KO mouse ( Fig. 2h, i), the iKO mouse also displayed impaired regenerative ability after BaCl 2 induced muscle injury as assessed by a 21.95% decreased number of eMyHC+ 4 days after injury (Fig. 2o, p). Taken together, findings from both KO and iKO mice solidify our thinking that SAM is necessary for the timely repair of damaged skeletal muscle tissue after acute injury.
SAM deletion aggravates dystrophic phenotype in mdx mouse. Besides acute injury by BaCl 2 injection, innate genetic defects can also provoke chronic injury-induced muscle regeneration. For example, in the widely used mouse model for Duchenne muscular dystrophy (DMD), mdx mouse displays extensive muscle degeneration and regeneration as early as~3 weeks of age; repetitive degeneration/regeneration cycles lead to the eventual loss of SC regenerative capacity and fatty fibrosis in old mdx mouse 37,38 . To examine whether SAM loss may affect chronic injury-induced regeneration in DMD, we generated SAM; dystrophin double KO (dKO) mouse by crossing the SAM KO first mouse with mdx mouse (Fig. 3a). As expected, SAM expression was completely depleted in freshly sorted SCs of dKO vs. control (Ctrl) mdx mice (Fig. 3b). The dKO mouse displayed no overt difference from the Ctrl mouse ( Fig. 3c) with comparable body weight during the course of 27 weeks ( Supplementary Fig. 3a); TA and gastrocnemius (GAS) muscles also showed comparable weight at 8 weeks ( Supplementary Fig. 3b). However, when examined closely, smaller myofibers were more frequently observed in the TA muscles of 8 weeks old dKO mouse as measured by the cross-sectional area (CSA) of individual fiber (Fig. 3d). Moreover, histological examination revealed increased size of unrepaired areas (Fig. 3e), an increased number of eMyHC+ myofibers (Fig. 3f) and increased infiltration of CD68+ macrophages (Fig. 3g) in dKO mice, suggesting loss of SAM delays the muscle regeneration in limb muscles. Compared to limb muscles, mdx diaphragm (Dia) muscle is known to exhibit a more severe dystrophic phenotype manifested by fibrosis and fatty infiltration that worsens as mice age 33,39 . Expectedly, the dKO mice at 6-8 months displayed the exacerbation of fibrosis as evidenced by increased Collagen I or Trichrome staining ( Fig. 3h-j). Taken together, the above results suggest that loss of SAM aggravates dystrophic phenotype of mdx mice.

Loss of SAM leads to SC autonomous defects in proliferation.
To further elucidate the impact of SAM loss on SC activities, we tested whether SAM loss impaires MB proliferation in the KO mouse ( Fig. 4). First, in vivo EdU labeling after BaCl 2 injury indeed revealed a reduced percentage (12.33%) of proliferating MBs in KO vs. WT littermates ( Fig. 4a and Supplementary  Fig. 4a); similar reduction (14.67%) was also observed when the assay was performed in iKO vs. Ctrl littermates (Fig. 4b). To further elucidate whether this proliferative defect is cell-autonomous, FISCs from KO or WT mice were cultured for 2 days and a lower percentage of EdU+ cells in KO (57.01%) vs. WT (66.91%) cells was detected (Fig. 4c). Furthermore, a significant reduction of the percentage of Pax7+MyoD+ cells was observed in KO (85.61%) vs. WT (91.78%) cells, suggesting a decline in the proliferative capacity of MBs (Fig. 4d). Consistently, when performed on SCs isolated from iKO mouse, the same conclusion was reached; a reduced percentage of EdU+ (25.61%) or Pax7+MyoD+ cells (5.59%) was found in iKO vs. Ctrl cells (Fig. 4e, f). Moreover, we also isolated single myofibers and performed the above assays in SCs associated with the cultured myofibers. Again, the percentage of EdU+ or Pax7+MyoD+ cells was significantly reduced in KO (10.88% and 5.35%, respectively) vs. WT cells ( Fig. 4g and Supplementary Fig. 4b). In addition, MTS assay also revealed that SCs from KO muscle displayed a declining proliferating rate compared with WT control (Supplementary Fig. 4c). Of note, the impaired proliferation in KO SCs was rescued by re-expressing a SAM plasmid ( Supplementary  Fig. 4d, e and Fig. 4h), pinpointing loss of SAM as the cause of the deficient proliferation.
The above findings supported SAM loss inhibits proliferation in MBs. To further investigate if it also has any impact on other aspects of SC activities. We first found that within 30 h after isolation, the percentages of EdU+ and Pax7+ MyoD+ cells were reduced 36.62% and 17.91%, respectively in KO vs. WT ( Fig. 4i, j), indicating a possible defect at the very early activation stage. Further assessing the differentiation ability, we found the percentage of MyoG+MyoD+ cells was increased (25.72%) in KO vs. WT SCs cultured for 3 days or myofiber-associated SCs cultured for the same period (19.87%) ( Fig. 4k and Supplementary Fig. 4f). This was further substantiated by measuring the fusion index by MF20 staining after 2 days in DM; KO cells showed a higher fusion ability (2.7 fold) than WT cells (Fig. 4l), indicating SAM loss leads to an increased propensity for differentiation, which was consistent with the finding from Fig. 1n. Lastly, the TUNEL assay did not detect differences in KO vs. WT (Supplementary Fig. 4g) cells cultured for 2 days, suggesting SAM loss may not have caused SC apoptosis. Fig. 2 Constitutive or inducible SAM deletion impairs muscle regeneration. a Schematic illustration of wild-type (WT) and SAM knockout-first (KO) mouse alleles. SA splice acceptor, IRES internal ribosome entry site, Neo neomycin, pA polyadenylation signal. Arrows indicate the locations of primers used for qRT-PCR. b qRT-PCR detection of SAM in FISCs. c Representative images of adult WT and KO mice. d Body weight from age-matched WT and KO mice. e Hematoxylin and Eosin (H&E) staining of tibialis anterior (TA) muscle from WT and KO mice. f IF staining for Pax7 and Laminin on the above muscles. Pax7+ SCs per 100 fibers were quantified. g The scheme for BaCl 2 injection induced TA muscle injury and subsequent analyses. h H&E staining was performed on the above muscles. The centrally localized nuclei (CLN) fibers were quantified at day 4 post-injury. i eMyHC and Laminin immunostaining was performed in g harvested muscles. eMyHC+ fibers were quantified at day 4 and 7 post-injury. j IF staining for Pax7 and Laminin or k MyoD and Laminin was performed in the TA muscles 3 days post-injury. Positively stained cells were quantified. l Breeding scheme for generating Control (Ctrl) and inducible SAM knockout mice (iKO). m Schematic of Tamoxifen (TM) injection, BaCl 2 injection, and SC collection in Ctrl or iKO mice. n qRT-PCR detection of SAM in FISCs 3 days after TM injection. o H&E staining was performed in the TA muscles 4 days after injury. p IF staining for eMyHC and Laminin in the above muscles was performed and eMyHC+ fibers were quantified. The data are presented as mean ± SD in b, d, f, h-k, n, and p. The p values by two-tailed unpaired t test are used for comparing two groups, ns not significant. The total number of mice used are indicated in b, d, f, h-k, n, and p. Scale bars: 50 µm (e, f, j, and k), 100 µm (h, i, o, and p). Source data are provided as a Source Data file.
Altogether, the above results demonstrate that SAM deletion causes a delay in SC activation and proliferation but increases the propensity for precocious differentiation.
Lastly, the above phenotypical changes in cells were also substantiated when RNA-seq was performed to assess transcriptomic changes caused by SAM loss. The knock-down of SAM led to 250 genes down-regulated and 167 genes up-regulated in MBs ( Supplementary Fig. 4h). Gene ontology (GO) cluster analysis revealed that the down-regulated genes were enriched for GO terms including cell cycle, M phase, microtubule-based process, chromatin assembly, etc. (Supplementary Fig. 4i), in line with the above uncovered function of SAM in promoting cell proliferation. The up-regulated genes were, on the other hand, enriched for skeletal system development, muscle cell differentiation, etc. (Supplementary Fig. 4j), which was consistent with the precocious differentiation phenotype upon SAM loss.
SAM interacts with Sugt1 in MBs. To dissect the molecular mechanism underlying SAM function in MBs, we sought to identify the interacting protein partners of SAM considering the well-known protein-binding ability of lncRNAs that endows themselves with many regulatory capacities 40 . To this end, we conducted RNA-pull down assay followed by mass spectrometry (MS) in C2C12 MBs using in vitro transcribed biotin-labeled SAM or GFP transcripts 13 (Fig. 5a, b). A list of proteins was identified as potential interacting partners of SAM, among which Sugt1 caught our attention because of its known function in kinetochore assembly and cell mitosis 10,41 . Next, we confirmed the SAM/Sugt1 association by Western blotting following RNA pull-down. Indeed, an evident amount of Sugt1 was captured by SAM, but not GFP transcripts (Fig. 5c). No interaction was detected between SAM and a few other known RNA-binding proteins, Hnrnpl 42 , Dnmt3a, and Dnmt3b 15 , suggesting the specificity of the SAM/Sugt1 association. To further confirm their interaction, native RNA immunoprecipitation (RIP) assay was performed using an antibody against Sugt1 (Fig. 5d). A higher level (3.1 fold) of SAM was pulled down by the Sugt1 antibody vs. IgG control (Fig. 5d) while several control transcripts including Gapdh, β-Actin mRNAs, and lncRNA Dum 15 were not retrieved. Consistently, the co-labeling of SAM by RNA-FISH and Sugt1 protein by immunofluorescence (IF) revealed an evident colocalization of SAM with Flag-labeled Sugt1 in MBs (Fig. 5e). Altogether the above results substantiated that SAM specifically interacts with Sugt1 protein in MBs. In addition, we generated a series of deletion fragments of SAM, F1-F5 according to the predicted secondary structure by RNAfold ( Supplementary  Fig. 5a) and performed RNA-pulldown assay ( Supplementary  Fig. 5b)  Interestingly, both F1 and F5 fragments of SAM retrieved comparable amounts of Sugt1 with the full-length transcripts. Nonetheless, the truncated transcript of exons 1-2 (containing F1 and F2) did not seem to be functional in muscle regeneration (Fig. 2n-p).
To further understand how Sugt1 and SAM together partake in the regulation of ASC proliferation, we found that similar to SAM, Sugt1 expression was also up-regulated upon SC activation at 24 h but down-regulated in differentiated cells at 96 h (Fig. 5f) siRNA oligos (Fig. 5g), the proliferative ability of ASCs was reduced as shown by a decreased percentage of EdU+ cells (16.53% and 13.82%) compared to controls (Fig. 5h), phenocopying the effect of SAM loss. Moreover, the overexpression of Sugt1 (Supplementary Fig. 5c) fully rescued the deficient proliferation of the KO ASCs (Fig. 5i). Altogether the above results demonstrated the functional synergism of SAM/Sugt1 in regulating SC proliferation. The conclusion was also substantiated when the expression dynamics and loss-of-function assays were performed using C2C12 MBs (Supplementary Fig. 5d-g). Interestingly, unlike SAM loss, Sugt1 knockdown did not seem to accelerate differentiation; instead, its loss may have delayed differentiation as assessed by the reduced number of MyoD+MyoG+ cells compared to control cells ( Supplementary Fig. 5h).
To further ask how SAM association regulates Sugt1, we found that SAM depletion in SC did not alter the mRNA level of Sugt1 (Fig. 5j) or its proper localization at kinetochores in prometaphase ( Supplementary Fig. 5i). Furthermore, it did not appear to alter the basal level of Sugt1 protein (Fig. 5k, left two lanes). However, treatment with a protein biosynthesis inhibitor, CHX, caused a marked decrease (31.4%) of Sugt1 protein in KO (Fig. 5k, lane 4 vs. 2) whereas only 18% in WT (Fig. 5k, lane 3 vs. 1), suggesting lower stability of Sugt1 in KO cells. Consistently, in a 20 h long CHX chase experiment, the half-life of Sugt1 protein was reduced at a faster rate upon SAM knockdown, confirming SAM is required for maintaining the protein stability of Sugt1 (Fig. 5l). Moreover, the decreased Sugt1 upon SAM depletion was blocked in the presence of a proteasome inhibitor, MG132 (Fig. 5m, lane 6 vs. 4), suggesting SAM may stabilize Sugt1 through preventing its ubiquitination. To further test this notion, HA-tagged ubiquitin protein was over expressed in Ctrl or shSAM MBs together with Sugt1 protein; we found an increased accumulation of poly-ubiquitinated Sugt1 in shSAM cells (Fig. 5n). Consistently, the stability of Sugt1 protein was rescued after restoring SAM expression in the presence of CHX without changing its RNA level ( Supplementary Fig. 5j and k). To examine if SAM stabilizing Sugt1 protein specifically occurs in MB cells, we found no decrease in Sugt1 level in primary hepatocytes isolated from WT vs. KO mouse with or without CHX treatment (Supplementary Fig. 5l). Lastly, to further strengthen that SAM promotes MB proliferation through stabilizing Sugt1 protein, we found that overexpressing the WT or a stable mutant of Sugt1 (Sugt1-4A) 12 fully rescued the deficient proliferation of SAM KO cells while over-expressing a highly unstable mutant of Sugt1 (Sugt1-4E) 12 failed (Fig. 5o).
SAM/Sugt1 regulate kinetochore assembly in MBs. Since Sugt1 is critical for proper kinetochore assembly during cell division 10,41 , we next tested whether SAM/Sugt1 together regulate SC proliferation through modulating kinetochore assembly. By staining chromosomes with DAPI, centromeres with ACA and spindles with α-Tubulin, in control cells, a robust spindle structure was preserved in metaphase cells, and bundles of microtubules were observed to terminate in kinetochores (Fig. 6a). In contrast, cells with Sugt1 knockdown exhibited disorganized spindle structures with multipolar spindles and fragmented spindle poles frequently observed (Fig. 6a). The above phenomena were also observed in C2C12 MBs when Sugt1 was decreased ( Supplementary Fig. 6a). Altogether, our data demonstrate Sugt1 is important for proper chromosomal alignment and spindle organization and thus timely mitotic division of MBs. Next, to demonstrate that SAM functions synergistically with Sugt1, we found SAM KO cells displayed evident defects in chromosome alignment and mitotic spindle formation (Fig. 6b) (Supplementary movies 1-4). Again, this was also more frequently observed in C2C12 MBs with SAM knockdown vs. control cells (Supplementary Fig. 6b), confirming SAM is needed for proper chromosomal alignment and mitotic division. To further determine if the above observed mitotic defects in SAM-depleted cells were due to kinetochore abnormalities, we examined kinetochore-microtubule (kt-mt) attachments under cold treatment considering the loss of cold stable kt-mt attachments is commonly used as an indicator of kinetochore defects 43 . WT and KO SCs were treated on ice for 10 min followed by α-Tubulin staining of microtubules (Fig. 6c); the fluorescence intensity was markedly decreased (20.76%) in KO vs. WT SCs, suggesting SAM loss led to increased instability of microtubules due to decreased kinetochores attaching. Consistently, when performed on C2C12, the same conclusion was reached; a reduced fluorescence intensity of microtubules was found in shSAM MBs under cold treatment ( Supplementary Fig. 6c).
To further pinpoint the defect in kinetochore assembly upon SAM loss, we examined the localization of Mis12 complex, since it is known as a client of Hsp90-Sugt1 to be stabilized and targeted to the kinetochore 11 . As expected, by IF a lower level of fluorescent signals of Dsn1 subunit was observed at the kinetochores in the mitotic KO vs. WT cells (Fig. 6d). We next examined the localization of Hec1 giving that as a so-called keystone complex Mis12 contributes to the localization of Ndc80 complex 44 ; interestingly, an over-accumulation of Hec1 kinetochore signals was detected in the KO vs. WT SCs (Fig. 6e). Similar phenomena were also observed in C2C12 MBs with decreased SAM knockdown (Supplementary Fig. 6d) despite the total level of Hec1 protein was largely unaltered (Supplementary Fig. 6e). Altogether, the above results confirmed the importance of SAM/ Sugt1 in the proper localization of kinetochore components. Lastly, to pinpoint it is the defective kinetochore assembly that mediates SAM KO phenotype, we found that knockdown of Dsn1 or Hec1 in SCs also delayed cell proliferation as assessed by a decreased percentage of EdU+ cells (Figs. 6f, g and Supplementary Fig. 6f, g). Meanwhile, since Akt is a known client of Sugt1 and its phosphorylation at position 473 by Sugt1 can promote Fig. 4 SAM loss in SCs leads to cell-autonomous defects in proliferation. a Top: the experimental scheme for in vivo EdU assay in WT and KO or b Ctrl and iKO mice. Bottom: The percentage of EdU+ SCs was quantified. c Top: the experimental scheme for in vitro EdU assay in 48 h-cultured SC isolated from WT and KO mice. Bottom: The percentage of EdU+ cells was quantified. d The above cultured cells were stained for Pax7 and MyoD. The percentage of double-positive cells was quantified. e and f The above assays were performed in SCs isolated from Ctrl and iKO mice. g EdU assay was performed on single myofibers isolated from WT and KO mice. The percentage of EdU+ SCs was quantified. h EdU assay in ASC transfected with a Vector or SAM expressing plasmid. The percentage of EdU+ SCs was quantified. The center line is represented as mean. i EdU assay in 30 h-cultured SC isolated from WT and KO mice. The percentage of EdU+ SCs was quantified. j Pax7 and MyoD staining in 20h-cultured SCs isolated from WT and KO mice. The percentage of double-positive cells was quantified. k MyoD and MyoG staining in 3 days-cultured SCs isolated from WT and KO mice. Quantification of the doublepositive cells was performed. l MF20 staining in 4 days-cultured SCs isolated from WT and KO mice. The fusion index of myotubes (≥2 nuclei)/total MF20+ cells) was quantified. DM differentiation medium. The data are presented as mean ± SD in a-g and i-l. The p values by two-tailed unpaired t test are indicated in a, d, h, and l and by two-tailed paired t test are indicated in b, c, e, f, g, i, j, and k. The total number of mice used in a-g, i-k and biologically independent samples in h and l are indicated. Scale bars: 100 µm a-l. Source data are provided as a Source Data file.
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16553-6 ARTICLE NATURE COMMUNICATIONS | (2020) 11:2725 | https://doi.org/10.1038/s41467-020-16553-6 | www.nature.com/naturecommunications cancer cell proliferation 45 , we tested if it could also mediate SAM effect but found that Akt p473 level was not decreased in KO vs. WT cells (Supplementary Fig. 6h). Altogether, the above findings demonstrate that SAM and Sugt1 together facilitate the assembly of kinetochore complex to ensure proper microtubule attachment in mitotic MBs. SAM deletion disrupts kinetochore assembly and thus delays the cell proliferation. Lastly, since it is believed that kinetochore disruption results in the mitotic arrest which is often followed by cell death 46,47 or induces mitotic slippage accompanied by the production of aneuploid and cell senescence 48,49 , we examined the consequence of such kinetochore defects in MBs and indeed detected an increased number of aneuploidy cells in KO vs. WT ASCs (Fig. 6h). However, no sign of cell apoptosis was detected earlier (Supplementary Fig. 4g); SA-β-Gal staining also revealed no indication of cellular senescence ( Supplementary  Fig. 6i).

Discussion
In this study, we identified and characterized the functional role of a lncRNA, SAM, in regulating SC activity and muscle regeneration. Collectively, our findings suggest a model, in which SAM regulates SC proliferation by binding with co-chaperon protein Sugt1 to facilitate the kinetochore assembly during mitosis, thereby governing the fidelity of cell division (Fig. 7). We infer that SAM stabilizes Sugt1 protein through direct association; it thus facilitates the correct localization of Mis12 complex which is required for proper assembly of kinetochore and microtubule attachment during the mitotic progression of MB cells. Loss of SAM in SCs leads to disrupted cell division and delayed proliferation, thus impairs muscle regeneration after acute or chronic muscle injuries. Although initially identified in C2C12 muscle cells through integrating RNA-seq and ChIP-seq analyses, we expanded our study to SCs to show that SAM is highly enriched in activated SCs. Moreover, gain or loss of function of SAM in both C2C12 cells and ASCs altered cell proliferation. Extending the in vitro cell culture-based investigation, we provided extensive mouse genetic evidence to characterize SAM function in vivo utilizing three KO mouse models: whole-body KO, SC-specific inducible KO (iKO) and mdx; SAM dKO mice. Results from using all three models consistently supported a role for SAM in regulating muscle regeneration after acute and chronic injuries. The KO-first strategy allowed us to delete SAM without major disruption of the genomic region, therefore, avoiding the complication of disrupting a potential enhancer in this region. Analyzing the KO mice led to the observation that SAM is not essential for mouse survival and fertility. However, the regeneration process of skeletal muscle after acute injury by BaCl 2 injection was evidently impaired in both KO and iKO mice. Nonetheless, in both models, the injured muscle eventually recovered completely from the injury, indicating that loss of SAM delays but does not block the regeneration of skeletal muscle. In the third model, the dKO mice displayed much more severe dystrophic phenotypes characterized by extensive fibrosis compared to the mdx controls; this could be caused by the amplified proliferative defect due to repeated cycles of degeneration-regeneration that is typical of dystrophic muscles. Taken together, findings from using the three mouse models solidified the role of SAM in regulating skeletal muscle regeneration in vivo, which adds genetic evidence for the functionality of lncRNAs in vivo.
Through identifying its interacting protein partners, we gained mechanistic insights into how SAM regulates SC proliferation. Sugt1 was identified as a specific interacting partner with SAM in MBs; their association is supported by results of RNA pull-down, native RIP, and co-localization assays. Furthermore, we showed that association with SAM probably serves to stabilize Sugt1 as SAM loss appeared to increase the ubiquitination level of Sugt1.
Consistently, a recent report 12 demonstrated that an E3 ligase, RNF41 regulates the ubiquitination of Sugt1 in a phosphorylation-dependent manner and PHLPP1 dephosphorylates Sugt1 to prevent it from associating with RNF41. In the future it may be worthy of the efforts to further investigate if SAM may facilitate the homodimerization of Sugt1 or involve in the dephosphorylation of Sugt1 in MBs.
At the cellular level, we showed Sugt1 is required for kinetochore assembly as loss of Sugt1 in MBs led to typical defects associated with cell mitosis; for example, cells presented pronounced defects in kinetochore-microtubule attachment, spindle formation and chromosome misalignments, which is in line with what was observed in Hela cells. Thus, in both Hela and SCs, Sugt1 appears to exert a conserved function of regulating kinetochore assembly. Similarly, loss of SAM largely photocopied the kinetochore abnormalities observed in Sugt1-depleted cells, leading us to conclude that SAM and Sugt1 synergistically regulate Mis12 targeting and kinetochore assembly to control MB proliferation. Expectedly, Dsn1 kinetochore signals were significantly decreased in SAM KO cells, in line with what was observed in Hela cells when Sugt1 was depleted. According to Davies et al. 11 the degradation of Dsn1 in Hela cells is dependent on Skp-Ub ligase thus suggesting this Ub pathway may be well functional in MB cells. Nevertheless, we observed increased accumulation of Hec1 protein at the kinetochores upon SAM loss, indicating Hec1 may not be subject to Skp-Ub degradation in MBs. Still it was shown that over-accumulation of Hec1 in mouse MEF cells caused aberrant spindle phenotype 9 , suggesting the mislocalization of Hec1 is indeed detrimental to the assembly of kinetochore and microtubule attachment in the SAM KO cells.
It is also interesting to ponder on the fate of the MBs with the abnormality in cell division. In many studies, kinetochore defective cells will show arrest or delay in metaphase via the spindle assembly checkpoint (SAC), or perhaps slip out of mitosis with chromosome segregation errors increasing the frequency of senescence or apoptosis 47,49 . Indeed, an increased number of aneuploidy cells was detected upon SAM loss, which did not lead to evident cell apoptosis or senescence; Unlike SCs deficient in SAC which resisted differentiation 50 , SAM loss did not seem to impede MB differentiation propensity. In fact, precocious differentiation was observed in SAM KO cells, which seems to suggest that the aneuploidy MBs in SAM KO may have eventually undergone premature differentiation. Coincidentally, Gogendeau et. al. 51 has described similar consequences in neural stem cells (NSCs) and intestine stem cells (ISCs), where they found that aneuploid NSCs do not die by apoptosis, instead, they display G1 lengthening and undergo premature differentiation 51 . Intriguingly, Sugt1 loss did not cause premature differentiation. We suspect it is possible that Sugt1 depletion had caused much more severe defects so that the cells eventually underwent apoptosis without being able to differentiate; this needs to be further investigated using a Sugt1 KO mouse model. Alternatively, it is also likely that the differentiation function of SAM may not be fully dependent on Sugt1. Altogether, our findings add to the growing list of cellular mechanisms studied in the proliferation and differentiation of muscle stem cells and their progeny, thus enhancing our knowledge in the broad field of muscle stem cell and regeneration. Additionally, to the lncRNA field, this will add in vivo genetic evidence for lncRNA involvement in muscle regeneration and provides new insights into the mechanisms of lncRNA action to the growing list of lncRNA functions. Single myofibers isolation and culture. Briefly, EDL muscles were dissected and digested with Collagenase II (800 units ml −1 ) in DMEM medium at 37°C for 75 min. Single myofibers were released by gentle trituration with Ham's F-10 medium -- Fig. 7 Schematic model depicting the functional role of SAM during SC activation/proliferation. In WT mice, SAM regulates SCs proliferation by binding with co-chaperon protein Sugt1 to facilitate the kinetochore assembly during mitosis, thereby governing the fidelity of cell division. In KO mice, loss of SAM induces degradation of Sugt1 by ubiquitination then disrupts kinetochore assembly in mitotic cells due to mislocalization of Dsn1 and Hec1, delays proliferation and impairs muscle regeneration after acute or chronic muscle injuries. SA-β-galactosidase staining. Cellular senescence was evaluated by βgalactosidase activity using β-galactosidase Senescence Kit (#9860, Cell Signaling Technology). Briefly, cells were fixed for 15 min followed by washing in PBS twice. Then fixed cells were incubated with β-galactosidase staining solution at 37°C in a dry incubator (no CO 2 ) at least overnight. The cells were then observed under a microscope for the development of blue color.

Methods
Chromosome spread assay. Cells were cultured for 3 days and treated with 100 ng ml −1 nocodazole for 3 h before harvesting. Trypsinized cell pellets were resuspended in pre-warmed hypotonic solution (75 mM KCl) and incubated for 20 min at 37°C followed by collecting by centrifugation for 5 min at 500 × g and gently resuspended with freshly prepared fixative solution (methanol/glacial acetic acid 3:1). Cells were fixed for 30 min. Two or three drops of suspended cells were released to pre-cold slides. The slides were then air-dried, and chromosomes were stained with DAPI.
Isolation of mouse primary hepatocytes. Liver tissue was isolated from mice and finely minced followed by digestion with collagenase II (400 U ml −1 ) in water bath with shaking at 37°C for 30 min. Digested tissue was mixed with a 10 ml serological pipette. The solution was triturated for 10-15 times or until the suspension traveled up and down the pipette smoothly without clogging. The cell suspension was then filtered through 70 µm cell strainer and centrifuged by 1300 rpm for 5 min. Cell pellet was washed twice in PBS and resuspended in culture medium (DMEM supplemented with 10% FBS, 100 U ml −1 penicillin and 100 IU ml −1 streptomycin). Primary hepatocytes were seeded on dishes and incubated at 37°C with 5% CO 2 for 3 h. After cells had adhered (3-4 h) media was removed and replaced with fresh culture medium and continued to culture for 3 days.
In vivo ubiquitination assay. C2C12 cells were transfected with HA-ubiquitin and flag-Sugt1 plasmids. 38 h after transfection, cells were treated with MG132 (10 µM) for 10 h. The whole-cell extracts prepared by lysis buffer were subjected to immunoprecipitation of Sugt1 protein. The levels of ubiquitinated protein were then detected by immunoblotting with HA antibody.
IF staining and image acquisition. For IF staining, cells were fixed in 4% PFA for 15 min and permeabilized with 0.5% NP-40 for 10 min. Then cells were blocked in 5% BSA for 1 h followed by incubating with primary antibodies overnight at 4°C and secondary antibodies for one hour at RT. For kinetochore protein staining, cells need be pre-permeabilized in 1% Triton X-100 in PHEM buffer (60 mM Pipes, 25 mM HEPES, 10 mM EGTA, and 2 mM MgCl 2 , pH 6.9) for 5 min before cells were fixed with 3.7% formaldehyde (Sigma) for 20 min. After fixation, cells were proceeded as described above. For cold-stable microtubule analysis, cells were incubated on ice for indicated times followed by fixation with PHEM buffer containing 3.7% formaldehyde and 0.2% Triton X-100 for 10 min on ice and cells then were stained as above. Antibodies and dilutions were used as following: For measurements of fluorescence intensities, 10 optical slices were acquired at 0.3 μm intervals. Measurements of tubulin, Hec1, Dsn1, and Sugt1 intensities were conducted with maximum intensity projections of images by in house program written in MATLAB (R2014b) language. Exposure settings were held constant within each group of experiments.
Immunohistochemistry 53 . In brief, slides were fixed with 4% PFA for 15 min at RT and permeabilized in ice cold menthol for 6 min at −20°C. Heat-mediated antigen retrieval with a 0.01 M citric acid (pH 6.0) was performed for 5 min in a microwave. After 4% BBBSA (4% IgG-free BSA in PBS; Jackson, 001-000-162) blocking, the sections were further blocked with unconjugated AffiniPure Fab RNA fluorescence in situ hybridization 54 . The Stellaris™-type oligonucleotides targeting SAM were modified with Biotin. Probe sequences are shown in Supplementary Table 2. Briefly, For SAM FISH, cells were fixed with 3.7% formaldehyde for 10 min at RT and permeabilized in 70% ethanol overnight at 4°C and hybridized with probes in buffer (2× SSC, pH = 7.0, 10% formamide, 2 mM VRC, 0.2 mg ml −1 BSA, 1 mg ml −1 yeast tRNA, and 100 mg ml −1 dextran sulfate) for overnight at 37°C. After washing, cells were blocked with 4% BSA and then incubated with Cy3-streptavidin antibody (Jackson, ref: 016-160). Prolong Gold antifade reagent was applied to mount the slides for DAPI. Images were taken with a ×63 NA 1.4 oil objective on the laser scanning confocal microscope (Carl Zeiss LSM 880). For FISH and flag-Sugt1 IF co-staining, prior to the hybridization, cells fixed in 3.7% formaldehyde and stored in 70% ethanol were permeabilized with 0.5% Triton x-100 for 10 min at RT. After washing cells were proceeded with the FISH protocol as described above. The following antibodies and dilutions were used. Mouse anti-flag (1:200, Sigma). Goat anti-mouse IgG Alexa Fluor 488 (1:200, Invitrogen).
RNA-seq and data analysis. For library construction, we used a protocol as described before 13,14 . The purified library products were evaluated using a Bioanalyzer (Agilent) and SYBR qPCR and sequenced on an Illumina Hi-seq 2000 sequencer (pair-end with 50 bp). Sequenced fragments were mapped to reference mouse genome (mm9) using TopHat2 54 . Cufflinks 55 was then used to estimate the relative abundance of transcripts in RNA-Seq experiments. Abundances were reported in fragments per kilobase per million (FPKM), which is conceptually analogous to the reads per kilobase per million (RPKM) used for single-end RNA-seq. Differentially expressed genes were identified if the fold change ≥ 1.5 by comparing siSAM and siNC samples.
Statistics and reproducibility. Data were analyzed using GraphPad Prism (version 8; GraphPad Software, San Diego, CA). Data were represented as the average of at least three biologically independent samples ± SD or ±SEM unless indicated. The statistical significance was assessed by the Student's two-tailed paired and unpaired t-test. ns, not significant. Representative images of at least three independent experiments were shown in Fig. 5a, c, d, e, k, m, n. and Supplementary  Figs. 2b, e, f, l; 4g; 5d, e, 6e and h. Representative images of two independent experiments were shown in Supplementary Fig. 5b, k, and l.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
The data supporting the findings of this study are available from the corresponding author on reasonable request. RNA-seq data have been deposited in the Gene Expression Omnibus under the accession code GSE126423. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE 56