Notchless defines a stage-specific requirement for ribosome biogenesis during lineage progression in adult skeletal myogenesis

In all cell types, protein synthesis is carried out by ribosomes, which are composed of proteins and RNA in two major subunits (Fromont-Racine et al., 2003; Peña et al., 2017). Although this process was thought to be highly conserved in all eukaryotic cells, recent studies have shown that regulation of protein synthesis is an important mitigator of cell fate control (Buszczak et al., 2014; Signer et al., 2014; Zhang et al., 2014). For example, in the haematopoietic system, modulation of the phosphorylation of the eukaryotic factors 4E-BP1 (also known as EIF4EBP1) and 4E-BP2 (EIF4EBP2), which inhibit the cap-dependent translation initiation factor EIF4E, is crucial for tissue homeostasis and cell fate control (Signer et al., 2016). The regulation of mRNA translation has also been reported to be essential to regulate the quiescent-to-activation states of skeletal muscle stem cells (Zismanov et al., 2016).

Recent studies have shown that the role of ribosomal proteins is not limited to generic protein synthesis factories, but can be implicated in active regulation of spatiotemporal protein expression (Lindstrom, 2009; Xue and Barna, 2012). In addition, heterogeneities in ribosome composition point to another level of translational control of protein synthesis. Several reports have indicated that rRNAs or ribosomal proteins in diverse organisms and tissues are under temporal or cell-type regulation for ribosome biogenesis (Zhou et al., 2015). For example, during anterior-posterior patterning of the embryo, RPL38-associated ribosomes have an affinity for translation of a subset of Hox mRNAs through a specific internal ribosome entry site sequence (Kondrashov et al., 2011; Xue et al., 2015).

In addition to these new regulatory roles for ribosomes, attention has focused on ribosomopathies, which are a collection of disorders that involve ribosomal proteins or ribosomal components. Surprisingly, several ribosomal protein mutations produce tissue-specific phenotypes, but not a broad deficiency and loss of all cells, as the highly conserved structure of ribosomes might suggest (Armistead and Triggs-Raine, 2014; Mills and Green, 2017). Among these pathologies, several impact on red blood cell formation, such as Diamond-Blackfan anaemia or 5q-Syndrome, whereas others result in craniofacial abnormalities or spleen defects (Ross and Zarbalis, 2014; Zhou et al., 2015).

Ribosome biogenesis is a highly regulated complex process involving ∼200 different proteins that act in the folding and processing of rRNAs, ribosome sub-unit assembly and nuclear export (Peña et al., 2017). Several quality control checkpoints are involved (Pechmann et al., 2013; Peña et al., 2017) and errors in this process can trigger cell growth arrest or apoptosis, through a mechanism referred to as the ribosomal stress pathway (Golomb et al., 2014; Pechmann et al., 2013). Notchless (Nle; Nle1) is a pro-ribosomal protein that participates in the assembly of the pre-60S ribosomal unit in nucleoli (de la Cruz et al., 2005). In yeast, its orthologue Rsa4 is involved in the last steps of pre-60S formation in nucleoli before its export to the cytoplasm. Nle orchestrates nuclear export with other partners (Nerurkar et al., 2015; Peña et al., 2017). The ATPase Rea1 (Mdn1) allows release of Rsa4 from the 80S subunit in a GTPase-dependent manner, which only then can bind export proteins, such as the Crm1 protein that will direct the large subunit to the cytoplasm (Baßler et al., 2014; Matsuo et al., 2014). Although the precise function of Nle in this last step remains unclear, it has been proposed to promote a conformation change that permits quality control to be passed before export (Matsuo et al., 2014). In all cases, Rsa4 or Nle deficiency in yeast or mammals, respectively, is lethal (de la Cruz et al., 2005; Cormier et al., 2006). Conditional ablation of Nle in adult mouse haematopoietic stem cells (HSCs) results in their loss (Le Bouteiller et al., 2013) yet deletion of Nle in B lymphocytes showed that this regulator is not required for their maintenance. In another context, conditional ablation of Nle with an inducible Villin-CreERT2 resulted in the loss of intestinal stem and progenitor cells (Stedman et al., 2015).

Skeletal muscle satellite (stem) cells are quiescent in adult mice and can be identified by several markers, including the paired/homeodomain transcription factor Pax7 (Relaix and Zammit, 2012; Rocheteau et al., 2012; Seale et al., 2000). Following muscle injury, the cells undergo lineage progression from quiescence to proliferation and generate myoblasts that will differentiate to form new myofibres. Each cell state can be identified by a combination of markers. A small proportion of satellite cells return to their niche and self-renew during the regeneration process. In homeostatic adult mouse muscle, the turnover of satellite cells can vary in different muscles, but is considered to be relatively low (Keefe et al., 2015; Pawlikowski et al., 2015). Studies involving the incorporation of radiolabelled elements have shown that ribosomal proteins turnover to a certain extent in proliferating and differentiating muscle cells, with a reduced accumulation of ribosomes in myofibres compared with myoblasts (Agrawal and Bowman, 1987; Jacobs et al., 1985). Although these studies showed some heterogeneity in ribosomal proteins between proliferating and differentiating cells, there were discrepancies concerning the turnover of ribosomal proteins during differentiation. The role of ribosome biogenesis during muscle stem cell quiescence and activation has not been addressed to date.

Using a specific Cre inducible mouse model, we evaluated the requirement of Nle-mediated ribosome biosynthesis in distinct cell states during lineage progression in quiescent, proliferating and differentiated myogenic cells. We show that, in contrast to the transit amplification phase of myoblast production, which is dependent on the formation of newly synthesised ribosomes in an Nle-dependent manner, quiescent and differentiated cells appear to be independent of Nle-mediated ribosome formation. These findings point to distinct regulatory functions for Nle-dependent ribosome biogenesis in different myogenic cell states, and reveal a hitherto unappreciated level of regulation that can contribute to heterogeneity in the skeletal muscle lineage.

Fluorescent in situ hybridisation (FISH), with rRNA probes hybridising to nucleolar precursors (Fig. 1A), was used to quantify the levels of rRNA present within the nucleolus of individual cells. The detection of interspaced transcriptional sequences (its) present in long transient nucleolar rRNA species before being fully processed was used as a readout of ribosome biogenesis taking place in a given cell (its1 and its2 probes detect rRNA precursors of the small and large subunit, respectively). Freshly isolated single myofibres (T0 h) were used to evaluate ribosome biogenesis in quiescent satellite cells and myonuclei simultaneously. As all satellite cells express Pax7 in homeostatic muscle, Tg:Pax7-nGFP mice (Sambasivan et al., 2009) were used to identify quiescent satellite cells located on the myofibre surface. Quantification of its1 and its2 FISH signals showed a significantly lower expression in quiescent satellite cells compared with the robust levels detected in all myonuclei (Fig. 1B,C). When myofibres are cultured in suspension, satellite cells activate to generate myoblasts (>80% divide within 24 h) and express the myogenic commitment gene Myod (Myod1) within 24 h (Rocheteau et al., 2012; Siegel et al., 2011; Zammit and Beauchamp, 2001). Quantification of FISH signals in the Pax7-nGFP⁺ cells at 48 h showed a marked increase of the its1 and its2 signals, reflecting the increased activity of ribosome biogenesis in these proliferating myogenic cells (Fig. 1D,E). This increase in rRNA precursor levels was also detected by RT-qPCR on quiescent and in vitro activated Pax7-nGFP⁺ cells that had been isolated by fluorescence-activated cell sorting (FACS) (Fig. 1F). Accordingly, the level of 18S and 28S rRNAs was high in activated satellite cells (Fig. S1A). These results show that the demand on ribosome biogenesis is low in satellite cells, as is expected owing to their quiescent state, then the demand significantly increased during proliferation and was sustained in differentiated myofibres.

Consistent with these dynamics, Nle expression was low in quiescent satellite cells, it then increased upon activation and decreased in differentiated myofibres (Fig. 1G). A time course of in vitro cultured primary cells showed an increase in Nle transcript and protein levels as early as 3 h after their isolation by FACS (Fig. 1G,H). This increase correlated with the disappearance of the Notch-effector gene HeyL, which is a marker of quiescent satellite cells (Fig. 1I) (Mourikis et al., 2012). Nle protein can be identified in vitro and in vivo in proliferating myogenic cells (Fig. 1H). In addition, antibody staining of cultured wild-type single myofibres (Fig. 1J) or cells cultured for 24 h (Fig. S1B) showed nucleolar staining for Nle protein within Myod⁺ activated satellite cells, in contrast to quiescent satellite cells that had no detectable protein by immunofluorescence (Fig. 1J, Fig. S1B). Therefore, dynamic expression of rRNAs and Nle occurs in quiescent and activated satellite cells, and in differentiated myonuclei.

To address the functional consequence of Nle deficiency in myogenic cells, we used transgenic Tg:Pax7-CRE^ERT2 mice that allow inducible Cre-mediated recombination in quiescent and activated satellite cells (Mourikis et al., 2012). Crosses with mice carrying the Nle^flox (Le Bouteiller et al., 2013) and Nle^null (Cormier et al., 2006) alleles led to the generation of Nle^flox/null;Tg:Pax7-Cre^ERT2;Tg:Pax7-nGFP (Nle cKO) and Nle^flox/+;Tg:Pax7-Cre^ERT2;Tg:Pax7-nGFP (Cont) mice (Fig. S2A). Cre-mediated deletion of Nle was induced in satellite cells by providing tamoxifen in food to mice for 2 weeks, followed by a single tamoxifen injection to increase the number of recombined cells. Genomic polymerase chain reaction (PCR) showed an efficient deletion of up to 95% of the Nle allele in quiescent satellite cells under this regime (Fig. S2B,C). Accordingly, immunofluorescence staining of cultured cells showed that ∼95% of satellite cells lacked nucleolar staining for Nle protein (Fig. S2D,E).

To examine the requirement for Nle in proliferating myogenic cells, Nle cKO satellite cells were isolated by FACS 2 weeks after the last tamoxifen injection and cultured in growth medium for 20 h or 44 h (Fig. 2A). Interestingly, these cells adhered to the culture plate, but they did not spread as control cells (Fig. 2B). Tibialis anterior (TA) muscles were then injured by injection of cardiotoxin (CTX; a peptide from snake venom that induces myofibre lysis; Hardy et al., 2016) and pulsed with 5-ethynyl-2′-deoxyuridine (EdU) for 4 h before FACS at 20 h post-injury (pi). Nle cKO cells were significantly compromised in their ability to uptake EdU compared with control cells (Fig. S3A). mRNA was extracted directly from Pax7-nGFP⁺ primary myogenic cells isolated using FACS at 20 h and 44 h pi. Quantitative real-time PCR (RT-qPCR) showed a decline in HeyL transcript levels in Nle cKO cells, which indicates that they had exited quiescence as control cells (Mourikis et al., 2012); however, there was a marked loss of the cell cycle regulators cyclinE (Ccne1/Ccne2) and cyclinD1 (Ccnd1) (Fig. 2C, Fig. S3B).

Absence of transcripts and protein for the differentiation marker myogenin (Myog) indicated that the loss of proliferation was not a consequence of precocious differentiation (Fig. 2C; data not shown). In addition, after cell activation, Myod and Myog transcript levels were lower than in the control, whereas Myf5 transcripts were unaffected (Fig. 2C, Fig. S3C). Clonal analysis of single cells cultured for 4 days and 3 weeks, as well as failure to uptake EdU in vivo, confirmed the compromised proliferative potential of Nle cKO cells (Fig. 2D, Fig. S3A). Genomic PCR confirmed the high recombination frequency in control cells and showed that rare large differentiated clones obtained from Nle cKO mice after 3 weeks in culture arose from non-deleted escaper cells (Fig. 2D, Fig. S3D).

As indicated above, the transition between the quiescent and proliferation state is associated with increased ribosome biogenesis and Nle expression (Fig. 1C,F). In the absence of Nle, we observed the accumulation of its2-containing rRNA precursors in vivo at 44 h after injury (Fig. 3A), consistent with a role for Nle in ribosomal large subunit maturation and export. Contrary to other cell types (Le Bouteiller et al., 2013; Stedman et al., 2015), Nle deficiency in satellite cells not only resulted in the accumulation of pre-60S rRNA precursors, but also a significant decrease in 28S rRNA transcript levels, and to some extent 18S rRNA transcripts, albeit the latter was not significant (Fig. 3A). FISH analysis of rRNA transcripts on plated satellite cells that were maintained for 24 h in culture showed a clear diminution of mature rRNA within the cytoplasm of the Nle cKO cells (Fig. 3B). Quantification of the intensity of fluorescence from FISH confocal images was performed on the nucleus and the cytoplasm for the 18S and 28S rRNAs. Measures of the surface area of the nuclear and cytoplasmic compartments highlight the failure of the Nle null cells to spread, activate and correctly increase nuclear and cytoplasmic volume compared with control cells (Fig. 3C). The reduction of ribosomal 18S and 28S rRNAs was observed mainly in the cytoplasm in Nle cKO cells (Fig. 3D).

Defective ribosome biogenesis is linked to the upregulation of the p53 (Trp53) signalling pathway, which leads to various cellular responses including cell cycle arrest and apoptosis (Golomb et al., 2014; Zhang et al., 2003). We used isolated single myofibres from control and cKO extensor digitorum longus (EDL) muscles to evaluate p53 expression in quiescent and activated satellite cells. No p53 protein was detected in the quiescent cells (data not shown). After 24 h of culture, myofibre-associated satellite cells (GFP⁺) were activated, as shown by the presence of Myod protein, and these cells lacked p53 protein (Fig. 3E,F). Strikingly, 80% of activated Nle cKO satellite cells expressed p53 protein at this stage (Fig. 3E,F). Collectively, our data show that, upon activation, Nle-deficient satellite cells display defective ribosomal large subunit production, which leads to a reduced quantity of cytoplasmic mature ribosomes and stabilisation of p53.

To confirm these observations, we examined satellite cells after their in vivo activation with CTX. Notably, the stabilisation of p53 corresponded to upregulation of several p53 target genes in Nle cKO cells, such as the cell cycle regulator p21 (Cdkn1a) and the pro-apoptotic genes Bax and Noxa (Pmaip1) (Fig. S4A,B). Apoptosis did not appear to be a major response to p53 stabilisation, as the proportion of Nle cKO satellite cells that were positive for activated caspase 3 was less than 2%, even after 72 h in culture (Fig. S4C). We then performed live videomicroscopy of cultured Nle cKO and control satellite cells to monitor their behaviour dynamically. In keeping with our findings, Nle cKO cells did not spread on the surface of the culture plate, they had limited motility, they did not undergo cell division and they displayed no significant cell death (Movie 1, Fig. S4D). In contrast, Cont cells were highly motile and performed multiple cell divisions during the same period (Movie 2, Fig. S4D). Nuclear and nucleolar staining performed either by FISH or immunostaining show distinct nucleoli and do not suggest a compromised integrity of this structure (Fig. 3B, Figs S4E,F).

We next investigated whether the defect in cell division was associated with reduced growth of cKO cells. We monitored the phosphorylation status of the S6-ribosomal protein RPS6, which is associated with cell growth and protein synthesis. In controls, phosphorylated (p)-RPS6 levels were low in quiescent satellite cells but increased after isolation and culture for 72 h (Fig. 3G,H,I). Although similar levels of p-RPS6 were noted in quiescent Nle cKO cells compared with the control, the mutant cells failed to upregulate p-RPS6 levels during this period, consistent with their compromised proliferation status. Accordingly, Myog protein was not detected in Nle cKO cells grown for 4 days in culture (Fig. S4F).

The regeneration of Nle cKO TA muscle was evaluated by histology 8 days after CTX-induced injury. Newly regenerated myofibres with centrally located myonuclei, a hallmark of regenerated fibres, were observed throughout the control tissue sections (Fig. 4A). The density of mononucleated cells was increased in the mutant and necrotic myofibres persisted in some areas at 8 days pi (Fig. 4A). Quantification of regeneration showed that Nle cKO muscles had significantly fewer regenerated myofibres with centrally located myonuclei, which were reduced in cross-sectional area (Fig. 4B). Sirius Red staining on muscle sections showed that Nle-mutant TA muscles had higher levels of collagen deposition in the intercellular space, compared with the control (Fig. 4A). However, by 4 weeks pi histological analysis of the TA muscle showed that early regeneration defects were largely restored (Fig. 4C), with a number of myofibres almost equivalent to controls; however, these myofibres showed a smaller cross-sectional area compared with the control (Fig. 4D). Genomic PCR analysis on self-renewed Pax7-nGFP⁺ cells isolated by FACS from 4 weeks pi TA muscle showed that these cells were not recombined for the Nle^flox allele and thus corresponded to escaper cells (Fig. 4E). Therefore, we propose that the rescued regeneration observed 4 weeks pi was due to escaper cells and that null cells do not contribute to muscle repair.

We then investigated whether Nle was required for the maintenance of quiescent satellite cells during homeostasis by quantifying the number of cells at different time points following tamoxifen injection. No change in the number of Pax7⁺ cells was observed on freshly isolated EDL myofibres 9 days or 21 days post-tamoxifen treatment (Fig. 5A) or frozen sections of TA muscle stained for Pax7-nGFP⁺ cells (Fig. S5A). The number of myofibres, and the number of myofibres with centrally located myonuclei, was not significantly altered between control and Nle mutant mice by 21 days (Fig. S5A). Given that satellite cells have a relatively low turnover (Chakkalakal et al., 2012; Keefe et al., 2015; Pawlikowski et al., 2015) we quantified, using FACS, the number of Pax7-nGFP⁺ satellite cells at 12 weeks post-tamoxifen treatment in different muscles. No significant loss of satellite cells was observed between mutant and control TA, quadriceps and abdominal muscles (Fig. 5B). However, at 15 months post-tamoxifen treatment, a dramatic loss of satellite cells was observed in the Nle mutant TA, quadriceps and abdominal muscles (Fig. 5C) without a loss in muscle mass (Fig. S5B). Therefore, we propose that the spontaneous exit of satellite cells from quiescence over the 15 month period resulted in their loss, and that the satellite cell niche was not replenished. Genomic PCR on the remaining Pax7-nGFP⁺ cells showed an absence of the deleted band and Nle mRNA expression was equivalent to the control (Fig. S5C), indicating that the remaining cells were not Nle-deleted and were likely cells that escaped the recombination process. Consistent with these results, a reduced number of centrally located myonuclei was observed in Nle-deficient TA muscles compared with 15-month-old control mice (Fig. S5D), suggesting that Nle-deleted satellite cells did not contribute to myofibres, whereas this occurs periodically in control mice during this period. Taken together, these findings suggest that, in the absence of Nle, satellite cells that spontaneously exit quiescence during homeostasis are not replenished by self-renewal of the pool over extended periods.

To investigate in more detail the phenotype of Nle-deficient quiescent satellite cells, we reasoned that the loss of Nle and ribosome biogenesis could affect their metabolic activity. Therefore, we quantified the mitochondrial content in satellite cells at 10 days (Fig. S5E) and up to 7 weeks post-tamoxifen treatment (Fig. 5D). Quantification of MitoTracker staining revealed no significant differences in mitochondrial content between Nle mutants and control quiescent cells (Fig. 5E, Fig. S5E). However, in vitro activated Nle cKO cells showed a 50% reduction in their mitochondrial content at 40 h after plating (Fig. 5F, Fig. S5E), and this reduction reached 80% when the distinction between Nle-deficient cells and escaper cells was taken into account (Fig. 5G, Fig. S5E). Indeed, genomic PCR analysis of the two populations (R6 and R7 populations) obtained by FACS of the 40 h activated Nle cKO cells clearly demonstrated that the more abundant population with the lower mitochondria content corresponded to Nle-deficient cells, whereas the minor population with the mitochondrial content similar to control cells corresponded to escaper cells (Fig. S5F).

To assess the function of Nle in differentiated muscle cells, we used the transgenic alpha-skeletal actin-Cre (Tg:HSA-Cre) mouse to delete Nle specifically in differentiated myofibres (HSA-Nle KO mutants) throughout prenatal and postnatal development, but not in quiescent or proliferating myoblasts (Miniou et al., 1999). Recombination of the Nle flox allele (Nle^Δ) was readily observed by genomic PCR from whole abdominal muscle extracts (Fig. 6A). The efficient deletion of Nle in Cont and mutant differentiated muscle cells was also observed in 2-month-old EDL isolated myofibres (Fig. 6A). Consequently, Nle transcripts were absent in myofibres (Fig. 6B), but mRNA for the structural protein troponin was detectable (Fig. 6B). Although ribogenesis was sustained to a high level in differentiated myofibres (Fig. 1B,C), deletion of Nle from embryogenesis to adulthood resulted in no overt phenotype, as total body weight and TA muscle weight were similar between HSA-Nle KO mutants and controls (Fig. 6C). Furthermore, the cross-sectional area of individual TA myofibres did not show significant differences even in 4-month-old HSA-Nle KO mutants compared with control animals (Fig. 6D). In addition, FISH and RT-qPCR analysis showed no differences in the expression and content of 18S and 28S rRNA between 4-month-old mutant and control muscles (Fig. 6E-G). Taken together, these findings indicate that Nle does not play a major role in differentiated muscle fibres.

The differential regulation of translation in distinct contexts provides another level of control for modulating cell fate choices. The commitment and differentiation decisions of mammalian skeletal muscle stem cells have been studied extensively, yet the specific role of ribosome biogenesis in translation control has only recently begun to be explored. Our study reveals that Nle, an important factor for the maturation of the large ribosomal subunit, is essential for the activation and proliferation of myogenic cells, but it appears to be dispensable for the maintenance of quiescent stem cells and the generation of differentiated myofibres.

First, our results show that its1 and its2, which correspond to non-transcribed spacer regions that are present only in unprocessed forms of pre-rRNAs, are present in quiescent satellites at greatly reduced levels compared with activated or differentiated myonuclei, indicative of a reduced production of new ribosomes. This suggests that there is a low turnover of ribosomes. Similarly, decreased ribosome biogenesis was found in yeast that was forced to adopt a quiescent state following nutrient starvation (Gray et al., 2004). Maintenance of a low rate of protein synthesis in stem cells was also shown in quiescent and dividing HSCs through the control of the initiation factor 4E-BP1 (Buszczak et al., 2014; Signer et al., 2014). In quiescent satellite cells, phosphorylation of EIF2A has been reported to result in a general inhibition of mRNA translation, despite maintaining a high expression of some quiescent-specific mRNAs (Zismanov et al., 2016). However, in the nucleoli of in vivo and in vitro activated satellite cells and differentiated myonuclei, its1 and its2 are highly expressed, consistent with a higher demand for protein synthesis.

Furthermore, the demand for ribosome biogenesis to sustain protein synthesis in differentiated cells varies in different tissues. In vitro pulse-chase experiments with ¹⁴C and ³H radiolabelled amino acids in rat L6 (Jacobs et al., 1985) and MM14 mouse (Agrawal and Bowman, 1987) myoblasts and differentiated myofibres showed that the proportion of ribosomal proteins and rRNAs were reduced by half in the transition from proliferating myoblasts to differentiated myotubes in culture. How these findings correlate with in vivo myofibres that are subjected to constant muscle contractions is not clear. Our FISH analysis of its1 and its2 on freshly isolated myofibres provides some insights into in vivo rRNAs synthesis and ribosome formation. In humans, ribosome biogenesis has been coupled with hypertrophy following high intensity training or the use of anabolic agents (Chaillou et al., 2014). Similarly, this inability to increase ribosome formation has been linked to the diminished capacity of aged muscle to hypertrophy (Stec et al., 2015). Interestingly, during denervation-induced atrophy, the ubiquitin and autophagy pathways are upregulated, leading to a loss of muscle mass, but the proteomic analysis revealed that ribosomal components were concomitantly upregulated (Lang et al., 2017). In future studies, it would be interesting to use proteomic approaches to examine ribosome turnover in these different cell states.

Ribosome assembly is a multistep process comprising hundreds of proteins, some of which are thought to be involved in mRNA translational control in a spatiotemporal manner. Perturbations in this process can result in tissue-specific defects (Kondrashov et al., 2011). Among these, Nle participates in the formation of the large 60S subunit and has specific roles in proliferating and differentiating cells (Le Bouteiller et al., 2013; Stedman et al., 2015). Here, we show that Nle expression is rapidly upregulated at the transcriptional and protein level, during the quiescence-to-activation transition of satellite cells following muscle damage. Loss of function of Nle showed that active Notch signalling declines as expected during this transition, an indication that Nle is not required for exit from the quiescent state. However, these cells fail to transit through S-phase, as shown by their inability to increase their cell size, express cyclinE, incorporate EdU and restore myofibres following muscle damage. In addition, Myf5 transcription that is normally upregulated during the activation of satellite cells (Gayraud-Morel et al., 2012) remained unchanged in the Nle mutant cells. Although a requirement for Nle during the myogenic cell expansion was expected, the lack of phenotype in Nle-deficient myofibres was surprising. Both of these cell states require a sustained demand for protein synthesis. In skeletal muscle, the turnover of myofibrillar proteins is constant and is significantly increased upon physical activity, whereas it is maintained at a low level in resting muscle (Norton and Layman, 2006). Accordingly, we showed the presence of a high level of nucleolar rRNAs in myoblasts and differentiated myofibres, suggesting constant ribosome biogenesis. These results suggest that, as in HSCs (Le Bouteiller et al., 2013), Nle is dispensable for ribosome formation in differentiated myogenic cells. Studies in yeast have shown that the Nle homolog is crucial for nuclear export of the pre-60S subunit following its release by Rix-Rea1 (Peña et al., 2017). Therefore, our studies indicate that an Nle-independent mechanism acts in quiescent cells and differentiated myofibres for export of this ribosomal subunit.

Moreover, the absence of Nle normally triggers a p53-dependent ribosomal stress pathway in which impairment of ribosome formation promotes ribosomal proteins to bind to MDM2 and block MDM2-mediated p53 degradation, thereby inducing p53-dependent cellular responses including apoptosis and cell cycle arrest (Golomb et al., 2014; Zhang et al., 2003). During the quiescence-to-activation transition, p53 was detected in the majority of Nle null satellite cells, but not in control cells, consistent with the notion that ribosome biogenesis is disrupted during exit from quiescence in the absence of Nle. However, this upregulation of p53 did not correlate with the induction of apoptosis, but instead with a disruption of cell growth, as shown by the reduction of the phosphorylated form of RPS6. Taken together, these findings show that Nle plays stage-specific roles during lineage progression of satellite cells, as has been reported in highly proliferative tissues such as blood and intestine (Le Bouteiller et al., 2013; Stedman et al., 2015). However, in contrast to quiescent HSCs, Nle is dispensable for quiescent satellite cells.

Another unexpected observation from our results was the finding that the 28S rRNA was lost during the transition from quiescence to activation, a phenotype that was not observed previously for blood and intestine (Le Bouteiller et al., 2013; Stedman et al., 2015). This may be because of our ability to examine this transition with somewhat greater precision in muscle; however it remains unclear whether some 28S rRNA are degraded during this transition. The observed diminution of the 18S might be a consequence of the downregulation of 28S, as Nle is known to be mainly involved in large subunit formation. These observations raise the possibility that the transition state is dynamic for rRNA turnover and might reflect a previously unappreciated regulatory step.

In summary, our findings point to ribosome biogenesis, and specifically Nle, as crucial regulators of cell state transitions in skeletal muscle. Remarkably, the functional role of Nle is restricted to the proliferation state, but not the quiescent and differentiated states during lineage progression. These observations provide a framework for further investigations into the role of specific regulators of ribosome biogenesis in different physiological states.

Nle deletion was obtained by crossing Nle^flox/+ and Nle^flox/null mice (Cormier et al., 2006) to Tg:Pax7-Cre^ERT2 (Mourikis et al., 2012). In addition, the mice were crossed to transgenic Tg:Pax7-nGFP mice that allow the specific detection of the satellite cell population (Sambasivan et al., 2009). As a result, Nle cKO and Cont mice correspond to the following genotypes: Tg:Pax7nGFP;Tg:Pax7Cre^ERT2;Nle^flox/null and Tg:Pax7nGFP;Tg:Pax7Cre^ERT2;Nle^flox/+, respectively. Unless specified, mice were fed with tamoxifen-granules (Harlan Laboratories) for 2 weeks and received tamoxifen by intraperitoneal (IP) injection (20 mg/ml tamoxifen solution diluted in corn oil) on the 15th day. After tamoxifen induction, the genotypes of mutant and control mice were: Tg:Pax7nGFP;Tg:Pax7Cre^ERT2;Nle^Δ/null and Tg:Pax7nGFP;Tg:Pax7Cre^ERT2;Nle^Δ/+, respectively. Nle deletion in differentiated myofibres was carried out by crossing Nle^flox/+ and Nle^flox/null mice to transgenic Tg:HSA-Cre mice (Miniou et al., 1999). The genotype of control and mutant mice were: Tg:HSA-Cre;Nle^Δ/+ and Tg:HSA-Cre;Nle^Δ/null. For proliferation experiments, mice were injected with EdU (Invitrogen) 4 h before sacrifice. Tails or cells were lysed overnight at 56°C in lysis buffer. Primers and strategy for genotyping by PCR have previously been described (Le Bouteiller et al., 2013). Animals were handled according to European Community guidelines and the ethics committee of the Institut Pasteur (CETEA) approved protocols.

Total RNA was extracted with RLT buffer (Qiagen RNeasy Micro Kit) or Tri Reagent (Sigma-Aldrich) for isolated single myofibres. cDNAs were prepared using reverse transcription (Super Script III, 18080044, Invitrogen) and real-time PCR was performed using SYBRGreen Universal Mix (13608700, Roche) or Taqman (04673417, Roche). Primers are listed in Table S1. Data were analysed using StepOnePlus RT PCR software v2.1 (Thermo Fisher Scientific) and Microsoft Excel. TBP transcript levels were used for normalisation of each target (=ΔC_T). RT-qPCR C_T values were analysed using the 2^−(ΔΔCT) method to calculate the fold expression (Livak and Schmittgen, 2001).

For indirect immunofluorescence, cells were plated and grown using the Permanox Nunc LabTek Chamber Slide System (177445, Thermo Scientific). Cells were then fixed with 4% paraformaldehyde (PFA; Electron Microscopy Sciences) for 5 min at room temperature (RT), rinsed 3× with PBS, permeabilised with 0.5% Triton X-100 for 10 min at RT and blocked for 20 min with 20% goat serum (Gibco) at RT. Primary antibodies were incubated overnight at 4°C, washed 3× with PBS and incubated for 2 h with Alexa-conjugated secondary antibody and Hoechst, then washed 3× with PBS. Slides were mounted in PBS/80% glycerol before observation. Single myofibre preparations have been described (Gayraud-Morel et al., 2007). For detection of nascent RNA, 1 mM of 5-ethynyl uridine (Invitrogen) was added to the growth medium for 4 h before fixation and detection with Click-iT Reaction Kit (Invitrogen). Samples were observed using an inverted Observer microscope (Zeiss) or an SP5 confocal microscope (Leica). For histology, TA muscles were snap-frozen in liquid nitrogen-cooled-isopentane for 1 min and stored at −80°C before cryosectioning. For histology, cryosections (12 µm) were rinsed with PBS and fixed in 10% formalin for 3 min before staining with Haematoxylin and Eosin or with Sirius Red.

Quiescent and activated satellite cells were isolated from total limb muscles or injured TA muscles, respectively (Gayraud-Morel et al., 2017). Briefly, dissected muscles were collected into cold Dulbecco's Modified Eagle Medium (DMEM), minced with fine scissors and transferred into a 50 ml Falcon tube containing 20 ml of DMEM, 0.1% collagenase D (1088866, Roche), 0.25% trypsin (15090-046, Gibco), 0.1% DNAse1 10 mg/ml (11284932001, Roche) solution. The tube was gently agitated in a 37°C water bath for 30 min and kept standing without agitation for the last 5 min to allow muscle fragments to sediment. The supernatant was collected in a 50 ml Falcon tube containing 10% foetal bovine serum (FBS; Gibco) and kept on ice. A total of four rounds of digestion were required to dissociate the tissue. The supernatants were successively filtered through a 100 µm and a 70 µm cell strainer (Miltenyi Biotec), before being pooled and centrifuged at 4°C at 50 g for 10 min to clear larger debris (this step was omitted for satellite cell preparation from injured muscles). The pellet was discarded and the supernatant was then centrifuged at 500 g for 15 min to pellet the satellite cells. The supernatant was discarded and the pellet washed twice with 40 ml of cold DMEM. Before the last centrifugation, the cell suspension was filtered through a 40 µm cell strainer. The final pellet was suspended in 1 ml of DMEM/2% FBS, before cell sorting using a MoFlow Legacy FACS (Beckman Coulter Life Sciences). Sorted cells were collected in DMEM/20% FBS/Penicillin-Streptomycin or directly into Buffer RLT (Qiagen) for RNA extraction.

For cell culture, satellite cells were plated on 1 mg/ml Matrigel- (354234, Corning) coated dishes and grown in DMEM/F12 (Gibco) containing 20% FBS, 2% Ultroser-G (15950-017, Pall Biosepra) and Penicillin-Streptomycin (Gibco). For clonal analysis, a single Pax7-nGFP cell was distributed per well, using MoFlow Legacy FACS, on Matrigel-coated 96-well plates containing growth medium. Cells were maintained on growth medium for a period of 4 days or 3 weeks. Before cell analysis, the cells were fixed with 2% PFA and rinsed with PBS. Counting of cells/well was performed using the OPERA system.

Fixed muscles that had been prepared for immunohistochemistry were sectioned and used for FISH staining. Plated cells or myofibres were fixed for 30 min with 4% PFA, rinsed with PBS and placed in 70% ethanol. Samples were rehydrated with 2× saline-sodium citrate buffer containing 10% formamide and processed as previously described (O'Donohue et al., 2010). Fluorochrome-conjugated probes (Eurogentec) are as follows: cy5-its1, 5′-TAGACACGGAAGAGCCGGACGGGAAAGA-3′; cy3-its2, 5′-GCGATTGATCGTCAACCGACGCTC-3′; cy5-18S, 5′-TTTACTTCCTCTAGATAGTCAAGTTCGACC-3′; cy3- or Alexa 488-28S, 5′-CCCGTTCCCTTGGCTGTGGTTTCGCTGGATA-3′; cy3-5′ets, 5′-AGAGAAAAGAGCGGAGGTTCGGGACTCCAA-3′.

Cells that had been freshly isolated by FACS directly in DMEM without serum, or cells that had been cultured in vitro, were incubated in 200 nM Deep Red MitoTracker staining solution for 30 min at 37°C, washed with PBS and fixed in 2% PFA for 5 min at RT before analysis by sorting on a MoFlow Legacy cell sorter.

Quiescent and injury-activated sorted satellite cells or in vitro cultured cells were collected in DMEM and counted with a Malassez cell counter; 50,000 cells for each condition were centrifuged and resuspended in radioimmunoprecipitation assay buffer plus protease inhibitors (Sigma-Aldrich). Cell lysates were run on 10% or 4-12% pre-casted gels (Novex). The gel was blotted onto a polyvinylidene difluoride membrane for 2 h at 200 mA at 4°C, and then saturated with 10% fat-free milk in Tris-buffered saline for 30 min at 4°C. The membrane was incubated with primary anti-Nle antibody (1:500) overnight at 4°C and with a secondary anti-rabbit HRP antibody (1:10,000) for 2 h (31460, Thermo Scientific Pierce Antibodies), then signal was revealed by Ecl-Plus (Pierce).

Primary antibodies used included: anti-GFP (chicken polyclonal, Abcam, ab13970, 1:1000), anti-laminin (rabbit polyclonal, Sigma-Aldrich, L-9393, 1:400), anti-Nle (rabbit polyclonal, 1:100, Le Bouteiller et al., 2013), anti-myogenin (mouse monoclonal, Developmental Studies Hybridoma Bank, F5D, 1:50), anti-Pax7 (mouse monoclonal, Developmental Studies Hybridoma Bank, Pax7, 1:50), anti-Myod (mouse monoclonal, Dako, Clone 5.8A, M3512, 1:100), anti-phospho-S6 ribosomal protein (rabbit polyclonal, Cell Signaling, 2F9, 4856, 1:2000), anti-S6 ribosomal protein (rabbit polyclonal, Cell Signaling, 5G10, 2217, 1:2000). Secondary antibodies used included: anti-rabbit HRP antibody (31460, Thermo Scientific Pierce Antibodies, 1:10,000), anti-rabbit Alexa 488 (A-A11070, Invitrogen, 1:1000), anti-mouse Alexa 555 (A-21425, Invitrogen, 1:1000) and anti-chick 488 (A11039, Invitrogen, 1:1000).

Confocal images (×40 field) were acquired on a SPE confocal microscope (Leica). Images were reconstructed using z-stack projection using FIJI software. FISH signal quantification was performed on raw images by manually delimiting nucleoli, nucleus, whole cell or whole myofibres as individual regions of interest (ROI). The fluorescence intensity provided by FIJI corresponds to raw integrative density for each ROI. Cross-sectional areas were defined using FIJI on TIFF images of anti-laminin immunostained cryosections. Images of cell staining of p-RPS6 were performed on Zeiss Observer microscopes. The fluorescence intensity provided by FIJI corresponds to the integrative density for each ROI. Approximately 100 cells were quantified for each time point; n-values are specified for each experiment in the figure legends or figures.

Cells isolated by FACS were plated on Matrigel-coated 24-well glass-bottom plates (P24G-0-10F; MatTek Corporation) with growth medium. The plate was then incubated for 20 h at 37°C, 5% CO₂, and 3% O₂ (Zeiss). Cells were then imaged up to 30 h with a Zeiss AxioObserver Z1 with an LCI PlnN 10×0.8 W DICII objective and AxioCam camera. An area of interest was cropped, and four to eight representative cells were tracked manually using TrackMate on FIJI.

For comparison between two groups, the two-tailed Student's t-test was performed to calculate P values and to determine statistically significant differences (*P<0.05; **P<0.01; ***P<0.001). Between three and five mice were analysed for most experiments. Data are presented as mean±s.d.

We thank Patricia Flamant and Clémire Cimper for the histology and cryosections and Jean-Yves Tinevez at the Imaging Platform for assistance with cell tracking. We also thank Sophie Novault (Flow Cytometry Platform of the Technology Core at the Center for Translational Science at Institut Pasteur).