-Related Myopathy describes a group of muscle diseases, clinically homogeneous although with variable severity, all caused by mutations in the gene encoding selenoprotein N (SelN). The main characteristic features are prevalent atrophy and weakness of trunk muscles, which are present from infancy but lead to complications such as scoliosis and respiratory failure later in life. To date, no curative treatment is available, and the pathophysiological mechanisms remain largely unknown. In this context, development of animal models reproducing the biochemical defect of the human condition is an important step. In this report, we describe for the first time a mammalian model for SelN deficiency, obtained by inactivation of the Sepn1
gene in mice. Sepn1−/−
mice were healthy and macroscopically indistinguishable from wild-type littermates. More surprisingly, we detected only subtle alterations in the morphology, ultrastructure and contractility of the mutant muscles. Together with our previous studies on embryonic development 
, these observations demonstrate that SelN is dispensable for survival, muscle development and muscle maintenance in mice, at least under basal housing conditions.
This absence of a major phenotype contrasts with the marked muscle defects observed in zebrafish morpholino mutants. In SelN-depleted zebrafish embryos, major developmental defects and important alteration of muscle architecture were observed 
. It was notably shown that SelN plays an important role for proper development of the slow muscle fiber lineage. This defect in the establishment and maintenance of slow fibers was not observed in the Sepn1−/−
mouse. In addition, an alteration in the myoseptum structures was seen in mutant zebrafish, suggesting a possible function for SelN in the establishment of myotendinous junctions, the mammalian analogs of myosepta. This hypothesis could be excluded in our mouse model since EM analyses showed normal myotendinous junctions in all muscles examined (data not shown). In addition, a functional link between SelN and RyR was proposed, consistent with the alterations of calcium dynamics and triad structures detected in the zebrafish mutants 
. Here we show that RyR1 and RyR3 expression and localization, as well as the structure of the triadic junctions, are not affected by SelN depletion in mice. RyR activity was not addressed in this study, but considering the preserved contractility of the Sepn1−/−
muscles, it is unlikely that SelN deficiency drastically impairs RyR channel regulation, which would result in a severe perturbation of excitation-contraction coupling. One possible explanation is that the divergent phenotypes observed in zebrafish and mice may reflect existing differences in muscle development and physiology between the two species, and/or that SelN plays distinct functions in the two species. Similarly, the differences between the basal phenotype of our murine model and the symptoms described in SEPN1
-RM patients is intriguing, but not exceptional: many mutations causing genetic neuromuscular diseases in humans lead to a mild phenotype once transposed to a mouse model 
. The absence of basal phenotype in the Sepn1−/−
mice might result from the obvious differences existing between mice and human, including differential use of specific muscles, such as postural muscles. The human myopathy involves predominantly muscles which are constantly active, particularly postural back muscles whose tonic activity maintains the upright position, a situation clearly different for rodents. In addition, SEPN1
-RM patients are ambulant and usually lead an active, close-to-normal life before the full phenotype develops in puberty 
. In contrast, standard mouse housing conditions provide a stable environment, with restricted physical activity and controlled conditions of temperature or other potential stressors, which could contribute to protect these mice from spontaneous development of the phenotype.
Interestingly, while running tests failed to demonstrate or to induce any defects in SelN-deficient muscles, homozygous Sepn1−/−
animals developed a pronounced phenotype when submitted to repeated forced swimming tests (FST). We decided to challenge our model with FST because this test associates physical exercise with a combination of behavior and environmental stresses. Indeed, FST is a widely used procedure to assess depression in mice, because it provides a context for global stress induction 
. Moreover, during the swimming sessions, mouse back and neck muscles are strongly solicited, since they are submitted to constant isometric contractions in order to maintain the swimming posture. After repeated swimming sessions, Sepn1−/−
mice developed symptoms reminiscent of the clinical spectrum described in SEPN1
-RM patients, with a predominant effect on trunk muscles leading to a marked deformation of the spine and, at the cellular level, an abnormal switch toward smaller and slower fibers. Overall, this reveals a striking sensitivity of exercised SelN-deficient muscle to a recurrent stress context. In addition, our results demonstrate the important involvement of SelN in muscle physiology, since no other obvious defects were observed in any of the other organs analyzed.
As mentioned previously in patients, clinical signs are highly homogeneous and recognizable, but the severity and spectrum of histopathological presentations are variable 
. In a recent publication, Cagliani and collaborators 
described three families harboring SEPN1
mutations with symptoms ranging from the severe early-onset rigid spine syndrome requiring early spinal fusion and assisted ventilation to more benign forms without major scoliosis. We also previously reported one patient with a mild phenotype first referred in her 30s, despite the almost complete absence of SelN expression 
. This variability in the clinical severity, regardless of the nature and position of the mutation, suggests that other elements modulate the degree of the phenotype in humans; these could be unknown genetic variants but also non-genetic factors, in particular involvement of environmental and/or physiological stresses in the aggravation of the pathology. In this context, our animal model will be useful to evaluate the role of exercise and additional factors, such as nutritional or pharmacological, in the pathogenic mechanism of the disease.
We recently showed that SelN plays a major role in satellite cell maintenance in adult skeletal muscles, a mechanism essential for muscle regeneration under basal conditions and even more in an acute stress context, i.e. following cardiotoxin-induced necrosis/injury 
. Hence, after two rounds of injury, a major impairment of the regeneration process was observed with drastic atrophy of the muscle. The absence of a basal phenotype in the Sepn1−/−
mouse model indicates that this defect in satellite cell dynamics is nevertheless compatible with correct development, growth and maintenance of skeletal muscle tissues. In the FST conditions, we found that satellite cells were activated neither in wild-type nor in mutant mice (data not shown), suggesting that the abnormal phenotype developed by mutant mice is independent from the satellite cell defect. These observations suggest that SelN participates in at least two distinct (but not mutually exclusive) processes, namely, muscle progenitor maintenance and mature fiber homeostasis.
In the Sepn1−/−
mouse model, the phenotype observed in the FST conditions developed after several repetitions of the swimming session, suggesting a cumulative process that progressively impairs muscle function. Previous studies based on ex vivo
analyses of human cells indicated that SelN is involved in oxidative homeostasis in cells 
. Accordingly, Oxyblot analysis revealed a significant increase in the overall content of carbonylated proteins in SelN-deficient quadriceps, but not in paravertebral muscles, under basal conditions (Figure S4
). In the FST context, one explanation for the muscle alterations in mutants is that training induces an elevated tissue oxidation process, due to a combination of increased activity and stress. Interestingly, while FST induced an increase in oxidized protein content in quadriceps from both wild-type and mutant mice, in paravertebral muscles increased protein oxidation was observed in the mutant mice exclusively (Figure S4
). This result suggested a different ability of mutant and wild-type paravertebral muscles to respond to stress and exercise. One remaining question is how the abnormal oxidative conditions resulting from FST would induce the alterations observed in the Sepn1−/−
muscles, including the tubular aggregates and the fiber hypotrophy, notably in paravertebral muscles. It has been established that free radicals generated by oxidative stress in muscle disuse context participate in proteolysis induction and muscle atrophy 
. Moreover, disequilibrium in the redox potential within the muscle fiber could result in reticular stress and calcium homeostasis dysregulation 
. Thus, it seems possible that in the absence of SelN and in a situation of exercise and/or stress, altered oxidative conditions may lead to perturbed RyR activity. The subsequent altered calcium dynamics could then be responsible for atrophy, by activation of Ca2+
-dependent proteases, and fiber type switch, by modifying the calcineurin pathway 
. Furthermore, it was shown that sarcoplasmic reticulum stress is responsible for tubular aggregate accumulation as an adaptive response to increased calcium influx 
Most members of the selenoprotein family, at least those with known functions, are enzymes involved in oxidation-reduction reactions and have been implicated in oxidative stress regulation 
. Some selenoproteins, such as glutathione peroxidase 1 (GPx1), thioredoxin reductase (TrxR) and possibly selenoprotein P (SelP), are involved in the detoxification of reactive oxygen species. Others like methionine sulfoxyreductases (Msr) or phospholipid hydroperoxide glutathione peroxidase (GPx4) are involved in oxidative damage repair 
. The molecular function of SelN is still elusive; whether it contributes to the maintenance of the redox homeostasis within the endoplasmic reticulum or controls more specific redox-regulated targets remains to be determined. One possible hypothesis, which we investigated in light of the mild muscle phenotype obtained in the Sepn1−/−
mice, was a compensatory expression of other selenoprotein(s). Our data revealed no major change in the expression of selenoprotein transcripts, suggesting no functional redundancy between SelN and another protein of this family (Figure S1
). Other enzymes or molecules involved in oxidative stress response may rather compensate the absence of SelN. One interesting candidate is vitamin E or α-tocopherol, a liposoluble antioxydant known to be involved in oxidative stress protection. Indeed, combined selenium and vitamin E nutritional deficiency has been linked to several muscle syndromes in both human and cattle 
; reviewed in 
. Moreover, N
-acetylcysteine (NAC), a thiol donor that increases glutathione synthesis, was recently shown to reduce protein oxidation and protect SelN-deficient fibroblasts from cell death under oxidative stress conditions. Therefore, NAC has been envisioned as a potential treatment for individuals affected with SEPN1
. The Sepn1−/−
mouse model submitted to the exercise conditions described here will be useful to test this strategy.
In conclusion, we propose that despite the lack of a spontaneous phenotype, the Sepn1 knock-out mice presented in this study constitute a useful tool to clarify the function(s) of SelN and to better understand the pathophysiological mechanisms underlying SEPN1-RM. While there was no obvious phenotype in normal housing conditions, we showed mild hypotrophy and absolute maximal force reduction in TA, limited fiber size modification in paravertebral muscles and an increase in oxidized protein levels; these data suggest that the pathophysiological process may already be on-going although not readily observable. In addition, specific experimental conditions demonstrated a particular susceptibility of Sepn1−/− muscles to exercise in a stress context, and led to a clear muscle phenotype whose distribution is comparable to that of SEPN1-RM patients. Therefore, further studies in this animal model will provide important clues to understand the influence of environmental factors in the disease pathogenesis, severity and progression, and might be helpful to identify and investigate potential therapeutic targets.