Analysis of muscle fibre input dynamics using a myog:GFP transgenic trout model

Skeletal muscle formation involves the determination of mesodermal cells into myoblasts and the differentiation of myoblasts into myofibres. These steps depend on the activity of transcription factors in the basic-helix-loop-helix (bHLH) protein family that include MyoD, myf5, myogenin and mrf4 (Weintraub, 1993). These four myogenic regulatory factors (MRFs), initially discovered in mammals, were later identified in many fish species, in which they are often found in two paralogues as a result of the early teleost whole-genome duplication (3R) (Jaillon et al., 2004). Thus, gilthead sea bream retain two paralogues of myod (myod1 and myod2) (Tan and Du, 2002). By contrast, the salmonid lineage lost myod2 but, as a result of additional duplications, gained three myod1 paralogues (Macqueen and Johnston, 2006) that have distinct developmental expression patterns (Delalande and Rescan, 1999; Macqueen and Johnston, 2006). Only a single myogenin gene has been identified to date in teleosts (Rescan et al., 1995; Weinberg et al., 1996; Macqueen et al., 2007; Codina et al., 2008). Myogenin expression in teleosts, as in mammals, starts just before muscle differentiation both in vivo and in vitro (Rescan et al., 1995; Weinberg et al., 1996; Delalande and Rescan, 1999; Codina et al., 2008; García de la Serrana et al., 2014). Emphasising the major role myogenin plays in muscle development, targeted inactivation of myogenin has been shown to be lethal in mice as a result of the failure of myoblasts to terminally differentiate (Hasty et al., 1993; Nabeshima et al., 1993). Knock down of myogenin alone has little effect on muscle phenotype in zebrafish, whereas knock down of both myogenin and Myod prevents fast muscle fibre differentiation. This led to the suggestion that myogenin cooperates with MyoD in fast myogenesis (Hinits et al., 2011).

Muscle fibre formation generally occurs in three phases in fish species (Rowlerson and Veggetti, 2001; Steinbacher et al., 2007; Johnston et al., 2011). During the first phase, distinct slow- and fast-muscle cell populations differentiate within the embryonic somite to form the superficial narrow band of slow muscle fibres and the deep large mass of fast muscle fibres, respectively. In the second phase, which occurs late in embryogenesis and/or in larvae, new fibres are formed in a discrete continuous layer at the surface of the embryonic myotome. During the third phase of myogenesis, termed mosaic hyperplasia, new small-diameter muscle fibres become inserted between pre-existing fibres throughout the whole myotome. Mosaic hyperplasia is generally thought to occur far into adulthood (Rowlerson and Veggetti, 2001; Johnston et al., 2011). Cell lineage tracings in zebrafish have shown that transient subcompartments of the early somite are sources of distinct waves of myogenic precursors necessary for the different phases of myogenesis. The first wave of myogenesis comes from medial MyoD-positive presomitic cells, known as adaxial cells. After somite formation, the adaxial cells undergo a radial migration from their original medial location to form a superficial layer of embryonic slow muscle fibres (Devoto et al., 1996). Coincident with migration of the slow-fibre precursors, the posterior domain of the early somite, which expresses MRFs, differentiates to form the fast muscle fibres of the embryonic myotome while the anterior domain of the early somite gives rise to cells that migrate to the outer lateral surface of the somite to form the dermomyotome-like epithelium surrounding the superficial layer of embryonic slow muscle fibres (Devoto et al., 2006; Hollway et al., 2007; Stellabotte et al., 2007). The dermomyotome-like epithelium, which does not express MRFs, provides myogenic progenitor cells necessary for the second and probably third phase of myogenesis (Hollway et al., 2007; Stellabotte et al., 2007; Steinbacher et al., 2008). The pattern of muscle growth varies between fish species. For example, the third phase of myogenesis follows the second phase of myogenesis in zebrafish and occurs after hatching (Patterson et al., 2008), while in brown trout, the second and the third phase of myogenesis begin simultaneously and vigorously at an early point of myotome development, well before hatching (Steinbacher et al., 2007). How muscle fibre recruitment is achieved in fish throughout post-embryonic life is poorly known. Given that the expression of myogenin marks new small-diameter muscle fibres in the fish myotome (Patterson et al., 2008; Rescan, 2008), we generated transgenic trout, Oncorhynchus mykiss (Walbaum 1792), carrying green fluorescent protein (GFP) cDNA driven by the myogenin promoter to visualise muscle fibre production throughout the trout's lifespan and in various physiological contexts, including muscle regeneration and prolonged fasting.

The proximal region of the promoter was found to include two putative E-box sites (CANNTG) in positions −175 and −192 relative to the translation initiation codon. A MEF2 (myocyte enhancer factor 2) binding site (TAAATTTA) and a MEF3 binding site (TCAGGTTT) were also present in positions −244 and −266, respectively, relative to the translation initiation codon (Fig. 1A) (GenBank accession number KM015458).

GFP expression in F1 offspring was first detectable around the 50 somite stage (12 days post-fertilisation, dpf) in the most rostral somites. GFP fluorescence progressed caudally as somites formed in a rostral-to-caudal wave. Transverse sections showed that GFP expression within somites progressed medio-laterally, starting deep in the somite and then extending to more lateral regions (Fig. 2A–C). At the end of somitogenesis, GFP fluorescence was observed within the whole myomeric axial musculature including the deep fast-twitch and peripheral slow-twitch muscle fibres (Fig. 2C). The myotome was entirely fluorescent up to pre-hatching stages, hampering further discrimination between fluorescent primary myofibres and small (presumably new) fluorescent secondary myofibres. Nevertheless, at the onset of free swimming, GFP fluorescence was predominantly observed in small-diameter myofibres at the periphery of the fast-twitch myotome, especially in dorsal and ventral extremes and in regions next to the horizontal myoseptum (Fig. 2D). The first occurrence of discriminable GFP-positive small-diameter muscle fibres scattered throughout the fast myotome was evidenced about 45 dpf (Fig. 2E). In fry stages, muscle hyperplasia was only of the mosaic type in fast muscle domains (Fig. 2F). Additionally, strong fluorescence was observed in the superficial slow-twitch muscle fibres of the growing myotome (Fig. 1D,F). Fluorescence in slow-twitch myofibres persisted throughout the life of the trout.

Fast-twitch muscle constitutes the majority of the myotomal musculature. We examined GFP expression in fast myotomal muscle from 8, 12, 15, 18 and 24 month old trout weighing 120, 450, 1200, 2200 and 2600 g, respectively. Strongly fluorescent small (new) muscle fibres (<25 µm), which give the muscle a mosaic appearance, were abundant in the fast muscle of 8 month old trout, but became progressively more scarce in the fast muscle of 12 and 15 month old trout (Fig. 3A–C). Large (old) myofibres did not display fluorescence, whereas myofibres with intermediate size showed residual fluorescence in which the intensity was inversely correlated with their size (Fig. 3A–C). No fluorescence was detected in the fast muscle of 18 and 24 month old trout (Fig. 3D,E), indicating that muscle fibre production ceases from 18 months post-fertilisation onwards. Concomitant with the increasing scarcity of fluorescent small-diameter muscle fibres during ageing, there was a regular increase in the average size of the muscle fibres, indicating that hypertrophy became the predominant growth mode as the trout grew older (Fig. 3A–E).

During normal development, the maximum fibre number in trout myotomal muscle was reached approximately 18 months following fertilisation. The cessation of muscle hyperplasia in aged trout may result either from the lack of recruitable myogenic cells or from changes in the microcellular environment that inhibit myotube production from activable myogenic cells. To test the hypothesis that recruitable myogenic cells are still present in the myotomal muscle of aged trout, we mechanically made lesions in the muscle of a 24 month old transgenic trout. Twenty days post-lesion, at the sites of lesion, muscle fibre organisation was strongly altered with a significant deposit of connective tissue (Fig. 3Fii), and the de novo appearance of regenerating, fluorescent, small-diameter muscle fibres (Fig. 3Fi). By contrast, fluorescence was not detected in the contralateral side of the lesioned trout, or at sites distant from the lesion areas.

Many fish species, including trout, have adapted to long-term food deprivation by mobilising energy materials stored in their tissues, particularly skeletal muscle proteins, which results in muscle atrophy. To examine whether small myofibres are still produced in atrophying muscle, we subjected juvenile trout to food deprivation. In our experiment, the mean body mass of the 6 month old trout was about 30 g prior to food deprivation. At the end of a 45 day fasting period, the mean body mass of the trout decreased to 23.1±2.2 g (mean and s.e., N=5) whereas well-fed trout exhibited a mean body mass of 75.8±10 g (mean and s.e., N=5). Transverse sections showed that the distribution of GFP-expressing small muscle myofibres in the muscle of the fasted trout was close to that observed in the muscle of the well-fed trout (Fig. 4A,B). To confirm this unexpected observation, we performed real-time PCR analysis and, in agreement with the GFP pattern, observed no significant difference in the mRNA level of both myogenin-driven GFP and myogenin transcripts between muscles from fasted and well-fed transgenic trout (Fig. 4C,D). In contrast, Sqstm1 (an autophagy-related gene) (Bonaldo and Sandri, 2013) and FBXO32/atrogin1 (an atrophy-related gene) (Bodine and Baehr, 2014) were strongly up-regulated (10- and 1000-fold, respectively) in muscle from fasted trout (Fig. 4E,F). Overall, our observations indicate that, in juvenile trout, myogenin-positive muscle fibres are still produced in atrophied muscle from fasted trout.

To describe muscle fibre input dynamics in trout (O. mykiss), we generated a stable transgenic line carrying GFP cDNA driven by the myogenin promoter. Sequencing the trout myogenin promoter revealed the presence of two putative E-box sites, a MEF2 site and a MEF3 site in the proximal domain. These motifs, which are highly conserved among fish species including zebrafish (Du et al., 2003), flounder (Xu et al., 2007), sea bream (Codina et al., 2008) and Senegalese sole (Campos et al., 2013), are essential for myogenin promoter activity as shown by site-directed mutagenesis (Du et al., 2003; Codina et al., 2008). Interestingly, mutation of the MEF3 site in the mouse myogenin promoter also abolishes correct expression in somites (Spitz et al., 1998). This indicates that the MEF3 motif, which binds Six/sine oculis homeoproteins, is an essential evolutionary conserved cis-regulatory element of myogenin transcription.

During development, GFP fluorescence was initially detected in the deep part of the somite and then extended laterally as the somite matured. Additionally, fluorescence was found to progress caudally as somites matured in a rostral-to-caudal wave. Thus, spatial myog:GFP expression is consistent with endogenous myogenin expression (Delalande and Rescan, 1999), suggesting that the promoter fragment we used contains the regulatory sequence necessary to recapitulate the endogenous expression pattern. However, embryonic GFP fluorescence was only detected at 11 dpf, whereas evidence of myogenin transcripts was found earlier, at approximately 8 dpf (Delalande and Rescan, 1999). This discrepancy probably relates to the fact that detection of GFP fluorescence in our transgenic trout involves not only the production of the myogenin-driven GFP-encoding transcript but also its translation into proteins and their accumulation to generate enough fluorescence to be imaged. Because of a persistent accumulation of GFP protein throughout the whole myotome, the restricted expression of GFP in areas of the myotome that are known to be stratified growth zones was not visualised before pre-hatching stages. In the brown trout, Steinbacher and collaborators have shown, on the basis of morphological analysis, that myotome expansion by stratified growth of the second-phase myogenesis occurs before the completion of somitogenesis (Steinbacher et al., 2007). These authors further reported that fast muscle domains are interspersed with undifferentiated cells around the end of somitogenesis, suggesting that mosaic hyperplasia in trout begins at an early point of myotome development (Steinbacher et al., 2007). In our model, mosaic hyperplasia as revealed by the presence of small GFP-positive myofibres scattered throughout the fast myotome, was not apparent before the free-swimming stage.

We further examined GFP expression at post-larval stages. Strikingly, throughout a trout's life, GFP was expressed strongly in all fibres forming the peripheral slow muscle. This contrasts with in situ hybridisation data in hatching trout embryos showing that myogenin expression in the slow muscle area next to the horizontal myoseptum is confined to small new fibres (Steinbacher et al., 2007). However, the persistent high GFP fluorescence level in slow muscle fibres is in line with previous expression analysis showing a strong accumulation of myogenin transcripts only in slow muscle fibres of adult trout (Rescan et al., 1995; Rescan 2001). Myogenin is also predominantly expressed in the slow myofibres of rodents, and its expression increases oxidative metabolism in muscles (Hughes et al., 1999; Ekmark et al., 2003). By contrast, in the bulk fast muscle, robust expression of GFP was restricted to nascent small-diameter muscle fibres and was therefore indicative of muscle fibre recruitment (muscle hyperplasia). Using this model, we found that muscle hyperplasia is prevalent in juvenile life, decreases progressively in later stages and then ceases approximately 18 months post-fertilisation. The fact that muscle hyperplasia stops in aged trout raises the question of how muscle hyperplasia is regulated throughout the course of a fish's life. Here, we report that muscle fibres are produced de novo in lesioned muscles of aged trout. This suggests that the cessation of muscle hyperplasia in aged trout is not attributable to a lack of recruitable myogenic cells, but rather to changes in the aged skeletal muscle niche that repress myogenic cell activation. In a previous study, using laser capture microdissection and microarray analysis, we reported the identification of genes that are overexpressed in hyperplastic regions of the late embryo myotome compared with adult myotomal muscle (Rescan et al., 2013). In light of the results obtained here, it would also be of interest to reciprocally examine genes that are overexpressed in adult myotomal muscle compared with the hyperplastic subdomain of a late-stage embryo in an attempt to identify genes that could contribute to the repression of myogenic cell activation.

Fish are well adapted to long-term fasting in their natural habitat. Fasting induces a loss of muscle mass resulting from the breakdown of muscle proteins, which involves several proteolytic systems such as the autophagy–lysosome and the ubiquitin–proteasome systems (Bonaldo and Sandri, 2013; Bodine and Baehr, 2014). Some studies have addressed changes in transcriptional regulation associated with muscle atrophy (Salem et al., 2007; Rescan et al., 2007), but less is known about concomitant changes in muscle fibre production. In this study, we show that GFP-expressing small myofibres are still produced in the muscle of fasted trout and are as numerous as they are in the muscle of well-fed trout. In agreement with this observation, PCR experiments showed that levels of myogenin promoter-driven GFP transcript and myogenin transcripts were similar in muscle from fasted and fed trout. Similarly, it has been shown that myogenin expression is not affected after fasting in juvenile gilthead sea bream (Garcia de la Serrana et al., 2014). In line with the lasting production of myofibres in atrophied muscle, it is interesting to note that the catabolic response induced by short-term calorie restriction is favourable to myogenic cell activation and muscle repair in mouse (Cerletti et al., 2012). Furthermore, the induction of autophagy in muscle satellite cells is crucial to meet bioenergetic demands during the transition from quiescence to activation (Tang and Rando, 2014). The production of nascent GFP-expressing myofibres in the muscle of fasted trout paradoxically suggests that amino acids released from protein breakdown during muscle atrophy are in part reused for building new muscle fibres from persistently activated myogenic cells. It could be speculated that the lasting formation of small fibres in atrophied muscle from starved fish may allow, after refeeding, rapid compensatory growth by hypertrophy of small myofibres, whose number is not severely altered during fasting.

In conclusion, our model provides insights into muscle fibre input in fish, and could help to identify factors that influence muscle growth dynamics and their underlying mechanisms.

All experiments in this study were carried out in the rainbow trout O. mykiss. They were performed in accordance with the recommendations of the Comité National de Reflexion Ethique sur l'Experimentation Animale of the Ministry of Higher Education and Research and approved by the Local Animal Care and Use Committee. Fish and their progeny were maintained at 10°C in recirculating water under artificial light–dark conditions, mimicking annual photoperiod variations. Chemical parameters of the water were monitored regularly. Oxygen levels always remained above 98% saturation. Prior to sampling, myog:GFP transgenic post-larval fish were rapidly anaesthetised with phenoxyethanol (Sigma; 4 ml/10 l fresh water). Animals were killed by decapitation and transverse slices of fast muscle situated just beneath the dorsal fin were removed for GFP fluorescence observations and RNA extraction. For the fasting experiment, myog:GFP trout were fed to satiation with a commercial diet until they weighed approximately 30 g. Then, the trout were separated into two groups: a fasted group (N=5) that was composed of fish deprived of food for 45 days and a control group (N=5) that was composed of trout fed to satiation. Animals from both groups were killed and sampled 45 days after the trial began. Lesions were made in 24 month old anaesthetised transgenic myog:GFP trout by inserting and withdrawing a syringe needle (1.2×40 mm) into trunk muscle, 1 cm beneath the dorsal fin. After a period of 20 days, the trout were killed by anaesthetic (2-phenoxyethanol) overdose and decapitated. Blocks of muscle close to the site of the lesions were then cut and embedded for transverse sectioning as indicated below.

To isolate the trout myogenin proximal promoter, a 5.3× genome coverage rainbow trout bacterial artificial chromosome (BAC) library (Palti et al., 2004) was screened by PCR using primers designed from the myogenin cDNA sequence (Rescan et al., 1995). DNA from positive BACs was isolated using the Nucleobond BAC Maxi kit (BD Bioscience). The primer walking method was then used to obtain the genomic sequence directly from the selected BACs.

The 5′-end of the transgene construct used in this study contained 1.3 kb of trout myogenin promoter, the 0.8 kb of the GFP coding region and an 860 bp fragment including the small t intron and the polyadenylation signal from SV40 (Luckow and Schütz, 1987) (Fig. 1B). DNA fragments used for the transgene construct were amplified by PCR using primers containing restriction sites for ligations. All subcloning procedures were performed in a modified pBluescriptII SK+ carrying I-SceI sites at both ends of the polylinker to enable I-SceI meganuclease-mediated transgenesis (Thermes et al., 2002).

Eggs were collected by gentle manual stripping performed under anaesthesia (0.05% 2-phenoxyethanol). The DNA construct was microinjected after fertilisation as previously reported (Gabillard et al., 2010). The injected solution contained 30 ng μl⁻¹ DNA, I-SceI buffer (0.5×, New England Biolabs, Ipswich, MA, USA), meganuclease I-SceI (New England Biolabs) at a rate of 1 unit μl⁻¹, and 0.1% Phenol Red (Thermes et al., 2002). After injection of the myog:GFP construct, approximately 30% of the surviving hatched embryos (F0) displayed mosaic muscle fluorescence. The strongest fluorescing GFP-positive embryos were raised to sexual maturity and outcrossed to the wild-type strain. In this way, three founder fish were obtained as shown by the presence of GFP expression in the F1 offspring. Transgenic founders had a germline transmission rate ranging from 5% to 30%.

Confocal microscopy of F1 fluorescent embryos was performed using a Leica TCS SP8 MP microscope with integrated vibratome sectioning and LAS AF software. Images were obtained with a 25× water immersion objective (Leica HCX IRAPO L, NA 0.95) and adjusted for contrast, brightness and dynamic range using ImageJ software. For histological examination of fluorescent trout at post-embryonic stages, samples were embedded in 30% ovalbumin, 0.5% gelatine and 1% glutaraldehyde in PBS. Blocks were then sectioned at 40 μm on a Leica vibratome (Leica Microsystem, Germany), and sections were examined using a Nikon ECLIPSE 90i microscope.

Myogenin and myogenin-driven GFP transcription was examined in muscle from fasted and well-fed trout using real-time RT-PCR. Total RNA from a transverse slice of fast muscle situated just beneath the dorsal fin was purified using TRIzol reagent. RNA integrity and concentration were controlled and calculated using an Agilent Bioanalyser. Total RNA (0.5 µg) was reverse transcribed according to the manufacturer's recommendations (High Capacity cDNA Reverse Transcription Kit, Applied Biosystems, Foster City, CA, USA). Expression levels were determined using the I-cycler IQ instrument (Bio-Rad, Hercules, CA, USA). Briefly, reverse transcription products were diluted to 1/20, and 5 µl was used for each real-time PCR reaction, which was performed using a real-time PCR kit provided with a SYBR Green fluorophore (Fast Sybr Green Master Mix, Applied Biosystems). Primer combinations were as follows: sense primer (AGCAGGAGAACGACCAGGGAAC) and antisense primer (GTGTTGCTCCACTCTGGGCTG) for myogenin; sense primer (GACGTAAACGGCCACAAGTTCAG) and antisense primer (CAGATGAACTTCAGGGTCAGCTT) for GFP; sense primer (GTGCGATCAAATGGATTCAAAC) and antisense primer (GATTGCATCATTTCCCCACT) for FBXO32; sense primer (AGCCCACTGGGTATCGATGT) and antisense primer (GGTCACGTGAGTCCATTCCT) for Sqstm1. The amount of target RNA was determined by comparison with a standard curve generated using a serial dilution of RT reactions. This dilution curve was used to ensure that PCR efficiency ranged from 90% to 100% and that amplification was linear within the sample set. The level of 18S RNA in each sample was also measured by real-time RT-PCR and used for target gene abundance normalisation within the sample set. Data on gene expression analysis are expressed as means±s.d. (N=5) and were analysed by one-way ANOVA followed by the Student–Newman–Keuls test. For all statistical analyses, the level of significance was set at P<0.05.

We would like to thank Cecile Melin and Jean-Luc Thomas for obtaining and rearing the trout embryos, Dr Catherine Labbé for cryopreservation of sperm from transgenic founders, Dr Y. Palti and Dr C. E. Rexroad III for sharing the rainbow trout BAC library, and Kamila Canale Tabet for screening the rainbow trout BAC library.