Myogenin is an essential regulator of adult myofibre growth and muscle stem cell homeostasis

Growth and maintenance of skeletal muscle fibres depend on coordinated activation and return to quiescence of resident muscle stem cells (MuSCs). The transcription factor Myogenin (Myog) regulates myocyte fusion during development, but its role in adult myogenesis remains unclear. In contrast to mice, myog-/-zebrafish are viable, but have hypotrophic muscles. By isolating adult myofibres with associated MuSCs, we found that myog-/- myofibres have severely reduced nuclear number, but increased myonuclear domain size. Expression of fusogenic genes is decreased, Pax7 upregulated, MuSCs are fivefold more numerous and mis-positioned throughout the length of myog-/-myofibres instead of localising at myofibre ends as in wild-type. Loss of Myog dysregulates mTORC1 signalling, resulting in an ‘alerted’ state of MuSCs, which display precocious activation and faster cell cycle entry ex vivo, concomitant with myod upregulation. Thus, beyond controlling myocyte fusion, Myog influences the MuSC:niche relationship, demonstrating a multi-level contribution to muscle homeostasis throughout life.


Introduction
Maintenance of adult skeletal muscle depends on the ability of multinucleated myofibres to grow and regenerate, thereby ensuring optimal functionality throughout life. To facilitate this homeostasis, adult vertebrate muscle contains a specialised population of precursors cells, muscle stem cells (MuSCs), also termed satellite cells, located between the basal lamina and sarcolemma of most adult myofibres (Mauro, 1961). Like other stem cells, MuSCs are mitotically dormant in normal circumstances but poised to respond to functional demand for new myonuclei throughout adult life (Relaix and Zammit, 2012;Purohit and Dhawan, 2019). The transcription factor Pax7 is considered a canonical quiescent MuSC marker across several vertebrate species and its expression is maintained during the progression to activation and proliferation, before downregulation at the onset of myogenic differentiation (Berberoglu et al., 2017;Buckingham and Relaix, 2015;Chen et al., 2006;Hammond et al., 2007;Hollway et al., 2007;Kawakami et al., 1997;Seale et al., 2000;Seger et al., 2011;Zammit et al., 2006;Olguin and Olwin, 2004). Null mutations for Pax7 severely affects MuSC maintenance and muscle regeneration in amniotes, amphibia and teleosts (Berberoglu et al., 2017;Chen et al., 2006;Oustanina et al., 2004;Relaix et al., 2006;Seale et al., 2000). In adult zebrafish, Pax7 + MuSCs contribute to regeneration of myofibres upon muscle damage, being the functional counterparts of MuSCs in mammals (Berberoglu et al., 2017;Hollway et al., 2007;Pipalia et al., 2016).
Proper function of the MuSC pool depends on a dynamic balance between quiescence and activation and is sustained by feedback signalling from surrounding muscle (Mashinchian et al., 2018;Forcina et al., 2019). In response to stimuli, quiescent MuSCs activate and become muscle progenitor cells (MPCs), which proliferate, differentiate and fuse to contribute myonuclei to preexisting multinucleated myofibres: a process that resembles aspects of myogenesis by embryonic myoblasts and relies on an intricate molecular network comprising both extrinsic and intrinsic mechanisms (Buckingham and Relaix, 2015). Among many converging factors, the members of the Myogenic Regulatory Factor (MRF) family of transcription factors, Myod, Myf5, Mrf4 and Myogenin (Myog), are key regulators of vertebrate muscle gene expression during both early and adult myogenesis (Zammit, 2017;Hernández-Hernández et al., 2017). In adult muscle, Myod and Myf5 are mainly expressed in MuSC and are crucial for efficient activation and proliferation (Cooper et al., 1999;Kuang et al., 2007;Megeney et al., 1996;Soleimani et al., 2012), whereas Mrf4 accumulates in mature myofibres and contributes to ensure their homeostatic size regulation (Moretti et al., 2016;Voytik et al., 1993).
Knockout of the Myog gene in mouse leads to severe muscle deficiencies and, in contrast to the other MRFs, is neonatal lethal (Hasty et al., 1993;Nabeshima et al., 1993), thus making it difficult to investigate Myog function in detail. As in embryogenesis, Myog expression is upregulated in proliferating MuSCs that then rapidly undergo myogenic differentiation (Zammit et al., 2004) and its expression appears reciprocally balanced with that of Pax7 (Olguin et al., 2007;Riuzzi et al., 2014;Zammit et al., 2006;Olguin and Olwin, 2004). However, depletion of mouse Myog in myoblasts does not block accumulation of differentiation markers in vitro, but cell biological aspects such as myotube formation were not explored (Meadows et al., 2008). We recently expanded this observation, reporting that, despite being dispensable for myogenic differentiation, Myog is essential for most myocyte fusion and its functional depletion leads to formation of mononucleated myofibres and reduced myotome growth during zebrafish embryonic/larval stages (Ganassi et al., 2018). Moreover, modulation of rodent Myog expression shifts muscle enzyme activity towards oxidative metabolism, alters exercise capacity and is required for neurogenic atrophy Meadows et al., 2008;Moresi et al., 2010;Ekmark et al., 2003;Hughes et al., 1999), suggesting a broad Myog role in regulating adult muscle homeostasis. Nevertheless, whether Myog acts on myofibre, MuSC or both has remained elusive. Importantly, in contrast to mouse, zebrafish myog -/mutant adult fish are alive, albeit displaying hypotrophic muscle (Ganassi et al., 2018). Congruent with the fish phenotype, almost complete depletion of mouse Myog after birth results in decreased muscle size (Knapp et al., 2006;Meadows et al., 2008), thereby confirming a conserved vertebrate role for Myog in regulating bulk muscle growth and maintenance.
Here, to explore the role of Myog in adult muscle, we analysed myog -/mutant zebrafish and deployed our recently developed method to isolate viable single myofibres with associated MuSCs, allowing culture of muscle progenitor cells (MPCs). Morphometric analysis showed that myog -/muscle has reduced myofibre growth and nuclear accretion compared to siblings (sibs), revealing increased myonuclear domain size. Adult myog -/muscle exhibited upregulation of pax7a and pax7b expression along with supernumerary MuSCs, that were randomly distributed throughout myofibre length compared to sib MuSCs, which localise mainly at myofibre ends, suggesting an altered MuSC niche. Ex vivo analysis of myog -/-MuSCs revealed elevated phosphorylation of ribosomal protein S6, concomitant with reduced expression of the mTOR inhibitors tsc1a and tsc1b, indicating enhanced mTORC1 activity, which marks MuSCs that have left deep quiescence to enter into an 'alert' pseudo-activated state (Rodgers et al., 2014). Congruently, cultured myofibres from mutants yielded increased numbers of MPCs with an accelerated proliferative phase, marked by upregulated expression of pax7a, pax7b, and myod. Together, our study demonstrates that Myog plays a crucial role in adult muscle growth and MuSC homeostasis.
(L) Increase in Surface Area (SA) from juvenile to adult stage (=Ad_SA -Juv_SA) indicates reduced growth rate in myog -/-. Data from Figure 1-figure supplement 1E, unpaired t-test. All graphs report mean ± SEM. Statistical significance within (coloured p) or between (black p) genotypes is indicated. Figure 1 continued on next page in myog -/fish compared to co-reared sib controls ( Figure 1A,B) in line with the nonsense-mediated decay reported previously (Ganassi et al., 2018). Levels of mRNA encoded by the terminal myogenesis genes mymk (tmem8c), mymx and mrf4 (myf6) were downregulated by 52%, 38% and 68%, respectively, in myog -/compare to the relative expression in sib, as during the embryonic stage, suggesting a continuous muscle defect in adulthood ( Figure 1B). To explore further the muscle defect at a cellular level, we isolated single myofibres from juvenile (1 mpf; months post-fertilisation) and adult (8 mpf) myog -/and sibs ( Figure 1C). Both juvenile and adult mutant myofibres were on average 20% and 50% shorter than those in control, respectively, demonstrating an early onset growth deficit in myog -/-, with lifecourse worsening ( Figure 1C,D and Figure 1-figure supplement 1A,B). Despite significant 50% reduction in overall fibre length, sarcomere length was unaffected in adult myog -/myofibres, as in embryonic myofibres ( Figure 1E; Ganassi et al., 2018). In contrast, myofibres isolated from the adult hypomorphic myog fh265 mutants Ganassi et al., 2018) were indistinguishable from those of age-matched sibs ( Figure 1-figure supplement 1C). The number of nuclei associated with each isolated myofibre appeared greatly reduced in myog -/mutants compared to sibs ( Figure 1F and Figure 1-figure supplement 1D). Indeed, nuclear numbers per unit length were reduced fourfold in mutants ( Figure 1G). Congruently, calculation of the average number of nuclei associated with each adult myofibre by multiplying the number of nuclei/unit length by the length revealed an approximate 90% reduction, from an average of 95 nuclei per myofibre in sibs to nearly 12 nuclei per myofibre in myog -/-( Figure 1H). Number of total nuclei in myog -/myofibres was reduced similarly at juvenile stage compared to sibs ( . However, although myofibre SA and number of nuclei positively correlated in myog -/-, we observed a reduced nuclear accretion as myofibres acquired larger area compared to sib, hence highlighting a significantly different growth mode between the two genotypes (different trend slopes) ( Figure 1K). Coherently, the incremental increase in myofibre SA over the juvenile-to-adult time-frame was significantly reduced in myog -/-, confirming a persistent deficit in muscle growth ( Figure 1L). Thus, loss of Myog function affects normal muscle growth, impinging on size and nuclear accretion of myofibres.

Altered MuSC:niche relation in myog -/mutant myofibres
Myofibres from myog -/showed a striking set of alterations in MuSC distribution. In control sib, MuSCs were found mostly localised close to the myofibre end, whereas in myog -/the supernumerary MuSCs were randomly distributed along the myofibre length ( Figure 2H and Around 60% of myog -/myofibres bore two or more MuSCs, whereas only 9% of sib myofibres had two or more MuSCs, and none had over three ( Figure 2G). Strikingly, 21% of sib myofibres had no GFP +ve cells, compared to only 5% of myog -/myofibres ( Figure 2G and

Lack of Myog enhances mTORC1 signalling in adult muscle and MuSCs
To investigate further how Myog may influence the MuSC niche, we assessed the expression of factors that contribute to muscle growth via regulation of MuSC activation. Whole muscle analysis revealed 130% increased level of igf1 mRNA and nearly 50% downregulation of myostatin b (mstnb) mRNA, both regulators of muscle growth and the mTORC1 (mechanistic Target Of Rapamycin Complex 1) pathway (Yoon, 2017;Trendelenburg et al., 2009), suggesting that lack of Myog triggers a growth signalling response ( Figure 3A,B). In contrast, expression of the zebrafish IGF1 receptors, igfr1a and igfr1b was unaltered (Figure 3-figure supplement 1) indicating that myog -/myofibres, and/or MuSCs, may respond to the higher IGF1 level similarly to sib. IGF1 signalling involves phosphorylation of AKT that activates mTORC1 resulting in downstream events such as promotion of muscle growth, accompanied by phosphorylation of the ribosomal protein S6 (RPS6) ( Figure 3C and Yoon, 2017;Ruvinsky et al., 2009). Western blot analysis of whole muscle extracts revealed that whereas AKT phosphorylation at Ser473 (pAKT) appeared unchanged, downstream phosphorylation of RPS6 at Ser240/244 (pRPS6) was enhanced fourfold in myog -/muscle compared to sib ( Figure 3D). Phosphorylation of RPS6 (pRPS6) marks the onset of MuSC activation in vivo (Rodgers et al., 2014). Thus, we analysed pRPS6 level in MuSCs on freshly isolated myofibres from pax7a:GFP;myog -/or control pax7a:GFP;myog +/sib. Myog -/-MuSCs showed a robust twofold increase in pRPS6 level ( Figure 3E,F), in line with quantification in whole muscle. No change was detected in myofibres themselves. As pRPS6 was increased but pAKT was not similarly altered, we reasoned that lack of Myog might control the expression level of factors that inhibit the mTORC1 pathway. Tsc1 and Tsc2 form a stabilised GTPase complex that represses the mTOR cascade ( Figure 3C; Nobukini and Thomas, 2004). Levels of tsc1a and tsc1b mRNAs were reduced by 51% and 57%, respectively, in myog -/muscle, whereas tsc2 mRNA was unchanged compare to sib ( Figure 3G), consistent with increased pRPS6. We conclude that lack of Myog triggers activation of the mTORC1 pathway in MuSCs, concomitant with tsc1a/b repression, leading to precocious MuSC activation.
Adult myog -/-MuSC-derived myoblasts exhibit faster transition to proliferation To assess MuSC activation and proliferation potential, myofibres bearing GFP +ve cells, from either pax7a:GFP;myog -/or control pax7a:GFP;myog +/sib were individually plated, cultured for 4 days in growth medium and the number of associated myogenic cells counted (Figure 4-figure supplement 1A). Myog -/mutants yielded at least three times as many mononucleate cells than did sibs ( Figure 4-figure supplement 1A), paralleling the increased number of MuSCs on freshly-isolated myofibres. The vast majority (80-90%) of cells from fish of either genotype was GFP +ve , so bona fide MPCs. Myog -/myofibres also displayed an average threefold increased GFP +ve cell yield compared to sibs ( Figure 4A-C). Within the 4-day culture period, sib myofibres (70% of which had only a single GFP +ve MuSC) yielded around 40 MPCs, reflecting a mean doubling time of around 20 hr. Mutants yielded a slightly higher proportion of GFP +ve MPCs than sibs ( Figure 4C), perhaps reflecting the increased number of MuSCs on each myofibre and the upregulation of pax7a mRNA in myog -/muscle. Importantly, although both genotypes yielded a GFP -ve cell population, all cells immunoreacted for Desmin, confirming their myogenic identity ( Figure 4-figure supplement 1A). Notably, the relative MPC proliferation rates were not significantly different among genotypes, calculated as the ratio of the number of GFP +ve MPCs obtained after 4 days in culture to the average number of GFP +ve MuSC per myofibre at the time of isolation ( Figure 4D). We conclude that myog -/mutant and sib MuSCs have similar proliferative potentiality when assayed in culture.
To explore proliferation dynamics further, MPCs were pulsed with EdU at either 2, 3 or 4 days ex vivo after myofibre plating in growth medium ( Figure 4E). To increase MPC yield for downstream analyses, 90-100 fibres were plated per well ( Figure 4E and Figure 4-figure supplement 1B). As expected, myog -/myofibres yielded more MPCs after 2 days in culture compared to sibs, and both genotypes produced mainly Desmin +ve muscle lineage cells (Figure 4-figure supplement 1B,C). An EdU pulse 2 days after plating revealed striking differences between myog -/-MPCs, 60% of which were EdU +ve , compared to sib MPCs, only 20% of which incorporated EdU ( Figure 4E,F). A day later (on day 3), the difference in proliferation had reversed; sib MPCs were almost 60% EdU +ve , whereas incorporation into myog -/-MPCs was significantly reduced compared to both sibs on day 3 and myog -/-MPCs on day 2 ( Figure 4E,F). This difference in S-phase labelling between myog -/and sib MPCs persisted on day 4 ( Figure 4E,F). Statistical analysis of overall EdU incorporation dynamics as Area Under Curve (AUC) revealed slightly higher proliferation rate of sib MPCs ( Figure 4E). Although we have not analysed EdU incorporation at day 1 (when many MPCs are still on their myofibres which, being often lost during staining, prevented analysis), these data clearly reveal that proliferation is not increased in mutant MPCs, simply activated earlier. We conclude that MPCs in myog -/mutants are more readily driven into proliferation upon release from their in vivo environment. We next explored MuSC dynamics by analysing MPCs mRNA levels at day 2 and day 3 of culture. Both pax7a and pax7b were more abundant, fourfold and threefold, respectively, in myog -/-MPCs at 2 days compared to sib ( Figure 4G), consistent with the higher percentage of MuSCs observed ( Figure 4C). Strikingly, while pax7a mRNA reduced with time, pax7b mRNA was maintained at similar level between day 2 and day 3 across genotypes, despite confirming overall increased level in myog -/-MPCs ( Figure 4G). Moreover, myod mRNA was fifteenfold more abundant in myog -/-MPCs compared to sib, again suggesting an earlier entry into activation phase. Notably, higher myod expression in myog -/-MPCs lasted into day 3. In contrast, myf5 mRNA level appeared similar in myog -/and sib MPCs at both 2 and 3 days, although tending to decrease at the later timepoint in sibs ( Figure 4G), perhaps reflecting their slower activation. The data suggest that myog -/-MPCs show faster activation and greater recovery from a stem-like state than MPCs from sibs.
The decreased proliferation but enhanced myod mRNA in cultured myog -/compared to sib MPCs also suggested that the mutant cultures might be entering the terminal differentiation program despite the presence of growth medium, presumably due to their rapid proliferation resulting in the formation of confluent cell clusters that consequently begin myogenic terminal differentiation prematurely. We assessed onset of differentiation by analysing expression of mef2d, mylpfa (encoding a fast myosin light chain), flnca and mybpc1, crucial players in myogenesis, which revealed significantly higher levels of all the assessed mRNAs in myog -/relative to sib MPCs on day 3, with an approximate upregulation of mef2d (twofold), mylpfa (fortyfold), flnca (fivefold) and mybpc1 (thirtyfold) ( Figure 4H). Among the four, only mylpfa showed accumulation in myog -/-MPCs already at day 2. In addition, mutant MPCs significantly accumulated cdkn1a (p21) mRNA (fourfold) between day 2 and day 3, in line with reduced proliferation and myogenic progression, whereas sib MPCs did not (Figure 4-figure supplement 1D). In contrast, level of cdkn1ca (p57), which promotes terminal myogenesis (Osborn et al., 2011), was unaltered in both myog -/and sib MPCs (Figure 4-figure supplement 1D), highlighting the immature state of myogenic progression. However, despite the continued exposure to growth medium, immunostaining confirmed accumulation of sarcomeric myosin heavy chain (MyHC) and significantly higher differentiation index in myog -/compared to sib MPCs at day 3, ( Figure 4I). As reported previously (Ganassi et al., 2018), upon culture under differentiation conditions, sib MPCs underwent cell fusion forming visible multinucleated myotubes, whereas differentiated myog -/-MPCs remained mostly mononucleated (Figure 4-figure supplement 1E). Altogether, our data indicate that whereas Myog function is dispensable for MPC proliferation and terminal differentiation, its lack either prevents MuSCs achieving full quiescence or accelerates MuSC transition into the proliferation phase, suggesting that Myog contributes to maintaining the MuSC niche.

Discussion
Here, we describe a novel function for the transcription factor Myog in regulating adult skeletal muscle growth rate and MuSC dynamics through four major findings. First, Myog influences MuSC number. Second, Myog is required for MuSCs to adopt their normal niche position. Third, Myog contributes to MuSC deep quiescence, regulating expression of genes involved in mTORC1 signalling. Lastly, Myog is required for proper myofibre growth and myonuclear accretion throughout life.

Myog controls MuSC number
We show that myog -/mutation causes supernumerary MuSCs/MPCs, paralleled by increased pax7a and pax7b mRNA. In addition to the fivefold increase in MuSC/adult myofibre and 50% reduction in myofibre length reported here, we previously described that adult myog -/mutants have unaltered body length and numbers of myofibre profiles/transverse body section roughly equal to their nonmutant siblings (Ganassi et al., 2018). Taken together, these data suggest that total number of Pax7 + MuSCs is increased around 10-fold in adult myog -/mutant myotome. In mouse, expression of Pax7 and Myog in MuSC appear to be mutually exclusive and controlled through reciprocal inhibition in in vitro studies (Olguin et al., 2007;Riuzzi et al., 2014). Persistent Pax7 expression delays Myog accumulation in cultured myoblasts and ex vivo in MuSCs on myofibres, whereas silencing of Myog can result in retention of Pax7 expression in differentiation (Olguin et al., 2007;Zammit et al., 2006). As Myog is required for adult myoblast fusion in vitro (Ganassi et al., 2018) and mymk and mymx mRNAs are reduced in adult myog -/muscle, one must consider the possibility that some pax7a:GFP +ve mononucleate cells are not MuSCs but differentiated myocytes that retain the pax7a reporter and are unable to fuse. However, we detect Pax3/7 protein in MuSCs, showing that strong GFP accumulation is not just perdurance of earlier signal. Moreover, most myog -/-MuSCs enter into S-phase after 2 days ex vivo. Subsequently, myog -/-MPCs quickly upregulate the differentiation marker mylpfa fortyfold, suggesting that an insignificant fraction was previously terminally differentiated. In addition, myog -/muscle expresses more myf5, a MPC marker. In post-natal rodent muscle, Myf5 is expressed in MuSCs, not myofibres (Beauchamp et al., 2000). Like our finding, murine Myog-knockout also led to Myf5 accumulation in neonatal limb muscle, even though onethird of the total nuclei are still MuSCs/MPCs in wild type neonates (Cardasis and Cooper, 1975;Rawls et al., 1998). Taken together, these data argue that the abundant pax7a:GFP +ve cells are MuSCs or MPCs. We consider two possible hypotheses for the increased number of MuSC/MPCs in myog -/mutants; differentiation failure or active accumulation. We do not favour the idea that absence of Myog prevents MPCs undergoing terminal differentiation leading to the accumulation of cells blocked in differentiation for three reasons. First, myog -/-MPCs readily undergo differentiation in culture and during myogenesis in the embryo (Ganassi et al., 2018). Second, although muscle is reduced in size, there is enough formed to support life and sarcomere length is unaffected, so terminal differentiation is fairly efficient without Myog during larval and adult growth and any required muscle repair. Third, both in fish and amniotes, MPCs ready to differentiate express high levels of Myod, which acts in a feedforward mechanism to trigger cell cycle exit and Myog expression (Weintraub et al., 1991;Hinits et al., 2009). We do not, however, observe increased levels of myod mRNA in myog -/mutant muscle. Nevertheless, when myog -/-MPCs are placed in culture they dramatically upregulate myod, arguing that Myod upregulation is also a characteristic of adult zebrafish MPC myogenic progression. Active accumulation of MuSCs, on the other hand, is suggested by the increase in pax7a, pax7b and myf5 mRNAs in myog -/mutant muscle. Below, we raise the hypothesis that the abundance of myofibres with low numbers of nuclei may trigger MuSC accumulation as a homeostatic response to tissue insufficiency.

-test. (F) Representative images of sib and myog -/-MPC numbers and EdU incorporation at indicated times in culture, showing detection of EdU (red) and nuclei (blue). (G) qPCR analysis
showing expression dynamics of pax7a, pax7b, myod and myf5 mRNAs in sib and myog -/-MPCs cultured for 2 (n = 4) or 3 days (n = 3), unpaired t-test. (H) qPCR analysis indicating upregulation of mef2d, mylpfa, flnca and mybpc1 mRNAs in myog -/-MPCs cultured in growth medium for 2 (n = 4) or 3 days (n = 3), unpaired t-test. (I) Representative immunodetection of MyHC (red) and nuclei (white) on sib or myog -/-MPCs at 3 days in growth medium for 3 days (left). Extent of differentiation (Index (%), see Materials and methods) is higher in myog -/than in sib MPCs (right). All graphs report mean ± SEM. Statistical significance within (coloured p) or between (black p) genotypes is indicated. Scale bars = 100 mm. The online version of this article includes the following figure supplement(s) for figure 4:

Myog is required for MuSCs to adopt their normal niche position
In wild-type fish 21% of myofibres have no associated MuSCs, despite having around 95 myofibre nuclei. This finding strongly argues that, in order to grow or repair damage, MuSC/MPCs must migrate across the basal lamina between fibres, as occurs in developing rodent muscle (Hughes and Blau, 1990). We observe 0.92 MuSCs/wild-type myofibre, on average. If one assumes random MuSC distribution amongst myofibres, the Poisson distribution predicts that 40% of myofibres should have no associated-MuSCs. The lower proportion of myofibres lacking MuSCs suggests that MuSCs are not distributed randomly. Conversely, Poissonian distribution would predict that 37% of myofibres should have a single MuSC. The observed value of 70% strongly suggests that MuSCs actively disperse between myofibres. Nevertheless, the system is imperfect, as the 21% of myofibres lacking MuSCs shows. Despite the fivefold increase in MuSCs in myog -/mutant, around 5% of myofibres still have no associated MuSC, which is far more than expected by the Poisson distribution (1.5%). Similarly, the 33% of myog -/myofibres with only a single MuSC compares with an expected value of 6%. In contrast, myog -/myofibres with over 10 MuSCs are over-represented, perhaps due to rare local regenerative events. In the absence of Myog, MuSCs tend to cluster on some myofibres, while also showing a tendency to be present at just one per myofibre more often than expected. These observations raise the possibility that myofibres in both wild-type and myog -/mutant contain a single specific niche for MuSCs.
Most MuSCs associated with isolated myofibres in zebrafish are located near myofibre ends, which would be attached to the vertical myoseptum tendon-like structure in vivo. We hypothesise that the fibre end provides a special niche in which MuSCs accumulate, perhaps formed in association with the myotendinous junction (MTJ). Strikingly, the preferred location of MuSCs in wild-type myofibres is between 10 and 20 mm from the myofibre end, suggesting that MuSCs locate not on the MTJ surface itself, but on the immediately adjacent cylindrical sarcolemma, retaining a potential physical contact with the MTJ. Moreover, this suggested location of the MuSC niche may relate to evidence that addition of new nuclei occurs preferentially at the ends of myofibres during growth (Kitiyakara and Angevine, 1963;Aziz and Goldspink, 1974). Upon explant ex vivo, MuSCs activate and begin to migrate away from myofibre ends. In myog -/mutants, the increased number of MuSCs are almost randomly distributed along the fibre, but the slightly higher abundance at myofibre ends might suggest that cells disperse once the MTJ niche is filled. However, when the fraction of myofibres in myog -/mutants that have only a single MuSC were analysed separately, significant dispersal was still observed. An additional possibility is that loss of MTJ contact by the short myofibres in myog -/mutants contributes to niche disruption. We conclude that Myog is required for assembly of the myofibre end MuSC:niche complex.
Mammalian MuSCs are generally not associated with myofibre ends, perhaps because myofibres are longer. Zebrafish myofibres are on average 1 mm long and rarely exceed 2 mm. In contrast, in the murine muscles most frequently analysed, those of the limbs, myofibres are up to 1 cm long. We suggest that the more numerous nuclei and larger number of MuSCs required for efficient growth and repair have allowed (or selected for) the MuSC niche to disperse along the myofibre.

Myog contributes to MuSC deep quiescence
Numerous lines of evidence argue that MuSCs in myog -/zebrafish fail to enter full quiescence. First, MuSCs are more numerous in myog -/mutant adults. Second, they are often not in their normal niche. Third, they contain more pRPS6 upon acute isolation. Fourth, they activate and proliferate more rapidly upon ex vivo culture. Three hypotheses could explain the MuSC myog -/phenotype. (1) Lack of Myog function in early development. However, MuSCs are not increased in mutant larvae (Ganassi et al., 2018) so it seems the defect worsens over the life course, suggesting a continuous need for Myog. (2) Lack of Myog within myofibre nuclei, which might lead to signals promoting MuSC proliferation (see below). (3) Myog could also function within MuSC themselves to control the balance between quiescence and activation. Whereas most literature describes Myog expression at the differentiation step following MuSC proliferation, various reports that noted Myog + MuSCs support a Myog MuSC-specific function. An early study investigating differential expression of Myod and Myog in rat muscle observed rare but intense accumulation of Myog mRNA along myofibre edges, anatomically reminiscent of MuSC niche (Hughes et al., 1993). Notably, Myog expression predominates in oxidative myofibres also characterised by higher MuSC density (Gibson and Schultz, 1982;Hughes et al., 1993). Rantanen et al. identified a population of dormant Myog + MuSCs which may contribute to muscle repair bypassing the proliferation step (Rantanen et al., 1995). Co-expression of Pax7 and Myog was observed in a satellite-cell study in transverse sections of both mouse soleus and EDL muscles (Schultz et al., 2006) and in rare cells in fish dermomyotome (Devoto et al., 2006). Likewise, immunohistochemistry on human resting muscle biopsies found Myog expression in MuSC but not in myonuclei (Lindströ m et al., 2010). More recently, the Blau group reported a low but detectable level of Myog protein in both resting murine MuSCs and in MPCs returning to quiescence after recovery from injury (Porpiglia et al., 2017). In addition, datasets show that Myog mRNA is present in quiescent MuSCs and rapidly decreases significantly during the 3 hours of myofibre isolation (Machado et al., 2017). Interestingly, Mrf4 expression in MuSC was downregulated in the same time frame, concomitant with exit from quiescence marked by the upregulation of Myod, demonstrating the presence of two well-known muscle 'differentiation-specific' markers in dormant stem cells. Together, such observations prompt re-evaluation of an intrinsic Myog function in quiescent MuSC.

Myog is required for proper myofibre growth
MRFs including Myog are required in the adult to restrict murine myofibre size (Moresi et al., 2010;Moretti et al., 2016). We find that, in zebrafish, Myog is required both for myocyte fusion that permits increase in nuclear number and consequent growth to the normal fibre size. We also show that Myog limits myonuclear domain size both in juvenile and adult muscle. Conditional depletion of Myog in adult mouse muscle decreased myofibre cross-sectional area, although further morphometric analysis of myofibre nuclear domains was not reported (Meadows et al., 2008). It therefore seems that, in addition to its role in promoting MPC differentiation and fusion, Myog restricts the volume of sarcoplasm supported by each myonucleus, both in denervated (Moresi et al., 2010) and innervated (this study) adult muscle. Both adult mouse and zebrafish Myog mutants show reduction in Mrf4 mRNA (Meadows et al., 2008;Venuti et al., 1995;Knapp et al., 2006;Rawls et al., 1998;Ganassi et al., 2018). Mrf4 itself restricts murine myofibre growth (Moretti et al., 2016), although no myofibre size change was reported in the various Mrf4 knockout mice (Zhang et al., 1995;Patapoutian et al., 1995). So Myog may either restrict myonuclear domain size directly or by regulating Mrf4 expression.
Unlike fish, which grow to some extent throughout life, mice do not require MuSCs to maintain adult muscle size (Keefe et al., 2015). Nevertheless, during murine adult myogenesis myonuclear accretion from MuSC-derived MPC fusion is required for myofibre growth and regeneration, as in fish (Pallafacchina et al., 2013). Congruently, expression of fusogenic genes myomixer (mymx) and myomaker (mymk) is reduced in both myog -/adult and embryonic muscles (Ganassi et al., 2018). Depletion of zebrafish Mymk leads to fewer myofibre nuclei in the surviving adults (Shi et al., 2018). Likewise, Mymk-KO in murine MuSCs at P0 led to 75% reduction of myonuclear number in extensor digitorum longus (EDL) myofibres measured at P28 (Nikolaou et al., 2019). Although severely reduced, accretion of myonuclei does occur in growing myog -/zebrafish, thus confirming the persistence of a Myog-independent pathway to fusion (Ganassi et al., 2018), which appears to allow a fraction of MPCs to sustain some muscle growth.
Addition of myonuclei to growing muscle relies on interactive signals between MuSCs and myofibres. Fish lacking Myog have increased igf1 and decreased mstnb expression. Both changes are predicted to enhance muscle size (Gao et al., 2016;Powell-Braxton et al., 1993;Coleman et al., 1995;Durieux et al., 2007;Trendelenburg et al., 2009;Zimmers et al., 2002;McPherron et al., 1997), and both may be important in our myog mutants. As we observe in myog -/fish, loss of Mstn increases MuSCs in amniotes (McCroskery et al., 2003;Manceau et al., 2008). MuSC contribution is, however, dispensable for myofibre hypertrophy in Mstn knockout mice, in which growth derives from increased myonuclear domain size. Mstn knockouts thus resemble both the MuSC increase and myofibre domain size increase in myog mutant fish. On the other hand, IGF1 has been shown to increase MPC proliferation and differentiation and thereby enhance myofibre growth, but without change in nuclear domain size (Fiorotto et al., 2003;McCall et al., 1998). Igf1 increase might thus enhance the number of MuSCs in myog mutants.
IGF1 and Myostatin signalling also affect MuSC activation status in mice (Cornelison and Wold, 1997;McCroskery et al., 2003;Zhang et al., 2010;Chakravarthy et al., 2001) and other fish (Garikipati and Rodgers, 2012). Both can act through the mTORC1 pathway (Trendelenburg et al., 2009;Latres et al., 2005). The IGF1 signal cascade phosphorylates AKT and activates mTORC1 leading to phosphorylation of RPS6 and eIF4EBPs, with downstream effects on both MuSCs and adult myofibres (Schiaffino and Mammucari, 2011). Loss of Mstn also activates the mTORC1/RPS6 pathway (Trendelenburg et al., 2009). Our data show enhanced RPS6 phosphorylation (pRPS6) both in myog -/muscle and MuSC analysis ex vivo. Interestingly, mTORC1 controls the transition from deep G 0 quiescence to an intermediate pseudo-activated state defined as G alert , in which MuSCs that accumulate pRPS6 are primed for activation compared to their quiescent counterparts (Rodgers et al., 2014;Porpiglia et al., 2017). Similar to zebrafish myog -/-MuSCs, 'alerted' mouse MuSCs display accelerated transition to proliferation ex-vivo compared to the quiescent population (Rodgers et al., 2014). As we detected no difference in Akt activation and unaltered levels of the receptors igfr1a and igfr1b, we deduce that increased phosphorylation of RPS6 is unlikely to result from enhanced upstream IGF1 cascade. However, myog -/muscle showed significant downregulation of tsc1a and tsc1b mRNAs, that encode functional orthologues of mammalian TSC1, a mTORC1 inhibitor highly conserved from fly to human (DiBella et al., 2009;Nobukini and Thomas, 2004) that promotes stemness in various tissues (Chen et al., 2008;Quan et al., 2013). Mouse Tsc -/-MuSCs display enhanced pRPS6 and accelerated entry into proliferation in vivo and in vitro (Rodgers et al., 2014). Murine Myog can bind to conserved regions in Tsc1 gene, whereas Myod does not, thus suggesting direct transcriptional regulation (Barbara Wold laboratory, Caltech; https://www.encodeproject.org/experiments/ENCSR000AID/) (ENCODE Project Consortium, 2012). These observations raise the hypothesis that loss of Myog leads to mTORC1 activation in MuSCs, exit from G 0 into G alert and subsequent increase in MuSC number. Taken together, our findings suggest that loss of Myog acts in adult animals to influence both MuSCs and muscle fibres, thus acting as a coordinator of tissue homeostasis. Continued on next page myog fh265 allele. For myofibre dissociation, trunk muscle was incubated in 0.2% Collagenase (C0130, Sigma Aldrich), 1% Penicillin/Streptomycin DMEM supplemented with 50 mg/ml gentamycin (Thermo Fisher) at 28.5˚C for at least 2 hours. Single muscle myofibres were released by trituration using heat-polished glass pipettes and washed three times with DMEM Glutamax High Glucose (Gibco). Myofibres were then imaged for total length measure or plated on Matrigel (Invitrogen) coated 96or 24-well plates and cultured in growth medium (20% Fetal Bovine Serum in 1% Penicillin/Streptomycin/DMEM GlutaMAX High Glucose supplemented with 10 mg/ml gentamycin). At indicated time points, MPCs were washed twice with PBS to remove plated myofibres, EdU pulsed (10 mM, Invitrogen Life Technologies) for 8 hr in fresh media and then fixed with 4% PFA for 15 min. For immunostaining myofibres were fixed in 4% PFA in PBS immediately after dissociation to reduce processing time and avoid MuSC activation.

Myofibre morphometry and growth mode analysis
Absolute myofibre length was measured for viable (i.e. non-fixed) myofibres imaged immediately after isolation on Leica M stereo-microscope using x1.6 magnification. Measurement of Surface Area (SA) and number of nuclei was performed on fixed myofibres at x10 or x20 magnification using an Axiovert 200M microscope (Zeiss). Myofibre diameter was measured at two locations over total myofibre length. Morphometric calculation were carried out as described in Brack et al., 2005. Briefly, SA/unit length = myofibre segment length x p x mean myofibre diameter; surface area domain size (SADS) = SA/myofibre nuclei in segment (Hoechst +ve ); Myofibre volume/unit length = myofibre segment length x p x (radius) 2 ; myonuclear domain = myofibre segment length x p x (radius) 2 /myonuclei (i.e. GFP -ve nuclei) in segment. MuSC (GFP +ve cell) distance to nearest myofibre-end was measured as schematised in Figure 2-figure supplement 1G. To graph frequency distribution, absolute nearest myofibre-end distances were normalised for myofibre length difference across genotype. Each myofibre was virtually divided in tenths where 1/10 and 5/10 corresponded to nearest segment to myofibre-end or segment at the myofibre half-length, respectively. Each tenth measured either 100 mm in sib or 50 mm in myog -/-. Generalised linear models (GLS) were used to explore growth assessing genotype differences in the relationship between SA and number of nuclei per 100 mm length (NoN), using the GLS function of NMLE package in R version 3.6.1 'Action of the Toes' (https://www.R-project.org / Pinheiro et al., 2020). SA was log e -transformed prior to analysis to generate a linear relationship with NoN, in a dataset pooling both juvenile and adult data. The dependent variable was set as NoN and the model included main effects of log e (SA) and genotype, as well as an interaction term. The dataset was then split by genotype and separate GLS models were run to analyse the relationship between log e (SA) and NoN for each genotype. t, p, and DF values were extracted for these models with the 'summary' function and were as follow: . Interaction between genotype and log e (SA): t = 7.332, p<0.0001, DF = 1164 . Siblings only, effect of log e (SA): t = 11.448, p<0.0001, DF = 1,97 . Mutants only, effect of log e (SA): t = 4.515, p<0.0001, DF = 1,67 Detailed steps of the GLS analysis are reported: . Decision to pool age groups Initial analyses showed no effect of age on the relationship between log e (SA) and NoN, therefore different age groups were pooled to study the effect of mutation. Note that these initial analyses suffered from high confounding correlation between SA and age in siblings; however, visual analysis of the few datapoints where juvenile and adult SA values overlapped gave no indication that age was affecting the NoN directly.
. Decision to use Generalised Linear Model instead of General Linear Model Significant heteroscedasticity was identified in simple linear models using R package lmtest (Zeileis, 2002), primarily because variance in NoN increased with log e (SA). GLS models were used to weight explained variance by the value of the dependent variables, controlling for this heteroscedasticity. For each step of analysis, every combination of weighting (weighting by log e (SA), genotype, both, or neither) was compared and the model with the lowest AIC was selected.
. Issue of Correlation between log e (SA) and Genotype In the final model, there was a significant correlation between explanatory variables log e (SA) and genotype, which may have confounded interpretations. This correlation was removed by subsetting with the 'Matchit' and 'Matching' R packages, and qualitatively identical results were found by rerunning the analysis using this smaller dataset (Ho et al., 2011;Sekhon, 2011).

RNA extraction, RT-PCR and qPCR
Genotyped myog kg125/+ (sib) or myog kg125 (myog -/-) adult trunk muscles were triturated and processed for RNA extraction using RNA Purification Plus Kit (Norgen) following supplier's instructions. Total RNA was reverse transcribed using Superscript III reverse transcriptase (Invitrogen) following supplier's instructions. For MPCs, RNA extraction was performed at indicated time point using micro-RNA kit (Qiagen). Prior to collection, culture wells were washed three times with PBS to remove myofibres, followed by MPCs detachment using Accutase (Sigma Aldrich) for 10 min at 28.5C . QPCR on technical triplicates for each sample was performed on 5 ng of relative RNA using Takyon Low ROX SYBR 2X MasterMix blue dTTP (Takyon) on a ViiAseven thermal cycler (Applied Biosystems). Ct values of all genes analysed were normalised to the geometrical mean of Ct values of three housekeeping genes (actinb2, sep15 and b2m) and fold changes were calculated using DDCt method (Livak and Schmittgen, 2001). Results are presented as mean value ± SEM of fold changes from independent experiments as indicated. Used primers, purchased from Sigma-Aldrich (KiCqStart SYBR Green Primers Predesigned, Sigma Aldrich) are listed in Supplementary file 1.

Statistical analyses
Quantitative analysis on images was performed with Fiji (NIH, www.Fiji.sc) and ZEN (2009ZEN ( + 2012 software. Statistical analyses were performed using GraphPad (Prism 6 or 8) or Microsoft Excel (v16) for unpaired or paired two-tailed Student's t-test as indicated and one-way ANOVA followed by Holm-Sidak's posthoc (growth mode analysis) or Bonferroni (MuSC analysis). 2 test was performed on raw data and used to assess difference between distributions. . Transparent reporting form Data availability All data generated or analysed during this study are included in the manuscript and supporting files.