Bi-allelic expression of the RyR1 p.A4329D mutation decreases muscle strength in slow-twitch muscles in mice

Mutations in the ryanodine receptor 1 ( RYR1 ) gene are associated with several human congenital myopathies, including the dominantly inherited central core disease and exercise-induced rhabdomyolysis, and the more severe recessive phenotypes, including multiminicore disease, centronuclear myopathy, and congenital fiber type disproportion. Within the latter group, those carrying a hypomorphic mutation in one allele and a missense mutation in the other are the most severely affected. Because of nonsense-mediated decay, most hypomorphic alleles are not expressed, resulting in homozygous expression of the missense mutation allele. This should result in 50% reduced expression of the ryanodine receptor in skeletal muscle, but its observed content is even lower. To study in more detail the biochemistry and pathophysiology of recessive RYR1 myopathies, here we investigated a mouse model we recently generated by analyzing the effect of bi-allelic versus mono-allelic expression of the RyR1 p.A4329D mutation. Our results revealed that the expression of two alleles carrying the same mutation or of one allele with the mutation in combination with a hypomorphic allele does not result in functionally equal outcomes and impacts skeletal muscles differently. In particular, the bi-allelic

Mutations in the ryanodine receptor 1 (RYR1) gene are associated with several human congenital myopathies, including the dominantly inherited central core disease and exercise-induced rhabdomyolysis, and the more severe recessive phenotypes, including multiminicore disease, centronuclear myopathy, and congenital fiber type disproportion. Within the latter group, those carrying a hypomorphic mutation in one allele and a missense mutation in the other are the most severely affected. Because of nonsense-mediated decay, most hypomorphic alleles are not expressed, resulting in homozygous expression of the missense mutation allele. This should result in 50% reduced expression of the ryanodine receptor in skeletal muscle, but its observed content is even lower. To study in more detail the biochemistry and pathophysiology of recessive RYR1 myopathies, here we investigated a mouse model we recently generated by analyzing the effect of bi-allelic versus mono-allelic expression of the RyR1 p.A4329D mutation. Our results revealed that the expression of two alleles carrying the same mutation or of one allele with the mutation in combination with a hypomorphic allele does not result in functionally equal outcomes and impacts skeletal muscles differently. In particular, the bi-allelic RyR1 p. A4329D mutation caused a milder phenotype than its monoallelic expression, leading to changes in the biochemical properties and physiological function only of slow-twitch muscles and largely sparing fast-twitch muscles. In summary, bi-allelic expression of the RyR1 p.A4329D mutation phenotypically differs from mono-allelic expression of this mutation in a compound heterozygous carrier.
Muscle contraction is brought about by a massive release of Ca 21 from the sarcoplasmic reticulum (SR) throughout the entire length of the muscle fiber via a process called excitationcontraction coupling (ECC) (1,2). Calcium release from the SR via the calcium release channel ryanodine receptor is initiated by a voltage-dependent orthograde signal delivered by the dihydropyridine receptor (DHPR), the voltage sensor localized in the transverse tubule membrane (T-tubules). Muscle relaxation is brought about by calcium reuptake into the SR by the sarco (endo)plasmic CaATPase (SERCA). ECC occurs at the contact region between the T-tubules and the SR membrane, within a domain encompassing a macromolecular complex formed by the ryanodine receptor 1 (RyR1), DHPR, and the calcium-buffering protein calsequestrin (3)(4)(5).
Mutations in RYR1, the gene encoding RyR1, have been identified in patients with a variety of neuromuscular disorders (6)(7)(8), including malignant hyperthermia susceptibility (MIM number 145600) (9), central core disease (CCD; MIM number 11700) (10), specific forms of multiminicore disease (MmD; MIM number 255320) (11), centronuclear myopathy (CNM; MIM number 255320), and congenital fiber type disproportion (12). A great deal of data has shown that recessive RYR1 mutations are mostly linked to cases of MmD, CNM, or congenital fiber type disproportion; patients with recessive mutations may carry homozygous mutations (i.e. the same mutation in both alleles) or two different mutations, one inherited from the paternal allele and the other from the maternal allele. In many patients carrying compound heterozygous mutations, one of the mutations causes a premature termination of the RyR1 protomer, leading to the mono-allelic and homozygous expression of a missense mutation. One of the molecular signatures of the mono-allelic expression of a missense mutation is a dramatic decrease of the RyR1 protein content in total muscle homogenates from patients and the overexpression of chromatin-modifying enzymes, such as histone deacetylases (HDACs) and DNA methyltransferases (DNMTs) (13).
We have recently developed a mouse model carrying compound heterozygous recessive Ryr1 mutations isogenic with those identified in severely affected MmD patients (14), namely, the frameshift RyR1 p.Q1970fsX16 mutation in exon 36 plus the missense mutation RyR1 p.A4329D in exon 91. The focus of this study is to determine whether phenotypic differences result from the presence of a homozygous Ryr1 mutation and the mono-allelic expression of the same Ryr1 mutation on the genetic background of a compound heterozygous carrier (with the other allele carrying a null mutation). In particular, in the present report we have examined the phenotype of mice carrying the homozygous mutation in Ryr1 exon 91 (referred to as HO91) compared with that of the mouse model carrying the compound heterozygous mutation in exon 36 and exon 91 (15) (referred to as DKI) and that of a mouse heterozygous for the expression of the exon 91 mutation (referred to as Het91). Our results show that at variance with compound heterozygous RyR1Q1970fsX161A4329D mice, the homozygous expression of the RyR1 p.A4329D mutation (referred to as RyR1A4329D mice) mainly affects slow-twitch muscles, with negligible effects on fast-twitch muscles. In particular, our results show that the homozygous expression of RyR1 p.A4329D causes a decrease of the RyR1 protein content in total muscle homogenates from soleus muscles only. The reduction of RyR1 protein is associated with a decrease of the (i) total running distance, (ii) the peak twitch and tetanic force in soleus muscles, and (iii) the peak twitch and tetanic calcium transient in soleus muscles.

In vivo phenotype of homozygous and heterozygous RyR1A4329D mice
The presence of the homozygous missense mutation in Ryr1 Ex91 of the mice used for the experiments was confirmed by diagnostic digestion of PCR-amplified DNA with the restriction enzyme PvuII (Fig. S1). RyR1A4329D homozygous (HO91) mice did not exhibit a postnatal lethal phenotype or defects in embryonic development and were indistinguishable from WT or Het91 Ryr1 mutant carrier littermates. Analysis of the growth curves performed during the observation period of 18 weeks indicates that both male and female HO91 mice exhibit a body weight similar to that of WT littermates (Fig. 1A). The growth curve of Het91 mice was similar to that of their WT littermates.
The presence of the homozygous RyR1 p.A4329D mutation functionally impacts in vivo muscle function in mice We used the voluntary running wheel setup (16) to investigate the in vivo skeletal muscle function of RyR1A4329D mutant mice. Mice had free access to the running wheel at any time of day; this experimental approach avoids potential problems linked to the effect of circadian rhythm on animal activity and/or animal compliance in performing nonvoluntary motor activity. The total dark phase running distance of three-month-old HO91 and Het91 mice was compared with that of their WT littermates (Fig. 1B). In the first week, the total running distance of WT and mutant mice was similar. 2 weeks of training improved skeletal muscle performance in all groups. After 3 weeks of training, the total running distance of the WT and Het91 mice was ;34% longer than that of HO91 mice (Mann-Whitney two-tailed test, calculated over the running period of 21 days; p , 0.05, n = 12 WT, n = 8 heterozygous, and n = 7 homozygous). The shorter running distance was also associated with a lower median cruise speed of the HO91 mutant mice than WT littermates ( Fig. 1C; p , 0.05 by Mann-Whitney test, two-tailed test). Interestingly, both HO91 and Het91 mice spent significantly less time on the running wheel than their WT littermates (Fig. 1D).
The homozygous RyR1 p.A4329D mutation decreases isometric force development only in soleus muscles The reduced speed and shorter total running distance of the HO91 mice may result from reduced muscle strength caused by alterations of the macromolecular complexes involved in ECC. We examined such a possibility by investigating the mechanical properties of intact extensor digitorum longus (EDL) and soleus muscles upon delivery of a single action potential or by a train of pulses of 1.0-ms duration delivered at 50, 100, and 150 Hz for 400 ms in EDL muscles and at 50, 100, and 120 Hz for 1100 ms in soleus muscles ( Fig. 2) (16). The averaged specific twitch peak force induced by a single action potential in EDL from HO91 mice was similar to that of EDL from WT mice (  mm 2 , n = 9, respectively; values are expressed as mean 6 S.D., p , 0.05). On the other hand, soleus muscles from HO91 mice exhibited a 36% decrease of twitch specific peak force compared with that of WT littermates (34.20 6 15.57 mN/mm 2 , n = 9, versus 52.75 6 11.15 mN/mm 2 , n = 10, respectively; values are expressed as mean 6 S.D.; p , 0.05), as well as a 30% decrease in the maximal specific tetanic force (200.72 6 32.82 mN/mm 2 , n = 10, versus 143.66 6 55.96 mN/mm 2 , n = 8, in WT and HO91, respectively; values are expressed as mean 6 S. D., p , 0.05). The specific force (twitch and tetanic) of EDL and soleus muscles from the Het91 mice were not different from those of littermate WT mice (Fig. S2).
The decrease in force output cannot be accounted for by fast-to-slow fiber transition, because there were no changes in the expression of myosin heavy-chain (MyHC) isoforms in EDL and soleus muscles ( Fig. 3A and B). Similarly, the reduced absolute force developed cannot be attributed to a major loss of contractile proteins, as the wet weight of the EDL and soleus muscles from HO91 and WT mice were similar (muscle wet weight of EDL was 12.79 6 1.54 mg, n = 11, versus 11.21 6 1.10 mg, n = 9, in WT and HO91 mice, respectively). Values are expressed as mean 6 S.D., p , 0.05. Wet weight of soleus was 12.37 6 2.06 mg, n = 10, versus 12.16 6 0.93 mg, n = 9, in WT and HO91, respectively; values are expressed as mean 6 S.D., p . 0.05. On the other hand, soleus muscles from HO91 mice show a small decrease of the minimal Feret's fiber diameter, indicating a slight atrophy of the slow-twitch fibers (Fig. 3A, C, and D), which may explain, at least in part, the decrease of muscle strength. No histological changes were present in EDL and soleus muscles from Het91 mice (results not shown).
The homozygous RyR1 p.A4329D mutation decreases RyR1 and Ca v 1.1 content in sarcotubular membranes from slow-twitch muscles The decreased muscle strength of RyR1A4329D mice was apparently not because of fast-to-slow fiber type transition but rather may be linked to atrophy of slow-twitch muscles, to alterations of the expression of ECC components, or to a combination of both. Thus, we investigated the protein content and transcripts encoding the major protein components of the macromolecular complex involved in skeletal muscle ECC (17). Transcript quantification was assessed by real-time quantitative PCR (qPCR) comparing WT and HO91 samples. No changes in the expression levels of Cacna1s, Stac3, and Ryr1 transcripts were apparent in EDL and soleus muscles from HO91 mice (Fig. 4A). We also investigated if the levels of histone deacetylases as class II HDACs are elevated in muscles of patients carrying recessive RYR1 mutations (13) as well as in the DKI mice (15). The expression of Hdac9 (a class IIa deacetylase) was elevated in soleus muscles from HO91 mice, whereas the levels of expression of the other HDACs were similar in WT and HO91 muscles (Fig. 4A). The contents of the main protein components of sarcotubular membranes in total homogenate from EDL and soleus muscles from WT and HO91 mice were determined by quantitative Western blot analysis. Figure S3 shows the specific immunoreactivity of the antibodies that were used for Western blot analysis. No significant changes in Ca v 1.1, RyR1, Stac3, SERCA1, SERCA2, calsequestrin, calreticulin, and JP45 protein content were observed in EDL total homogenates from HO91 mice (Fig. 4B). Interestingly, however, RyR1 and Ca v 1.1 protein content was significantly reduced in soleus muscles from HO91 mice compared with that of WT littermates (RyR1 content was 64.0 6 19.8% of WT [mean 6 S.E.], n = 13, p , 0.01, and Ca v 1.1 content was 44.9 6 7.9% of WT [mean 6 S.E.], n = 7, p , 0.05). The content of the other SR proteins analyzed was unchanged (Fig. 4B). ECC protein content was unaltered in muscles from Het91 mice (results not shown).
The decrease of RyR1 and Ca v 1.1 observed in soleus muscles should affect the amount of calcium released during ECC, and this was directly investigated by measuring calcium transients in single EDL and soleus fibers loaded with the low-affinity calcium indicator MagFluo4.

Calcium transients in isolated EDL and soleus fibers from WT and RyRA4329D mice
The resting calcium concentration ([Ca 21 ] i ) was measured with the ratiometric Ca 21 indicator Fura-2, and there were no differences between EDL and soleus fibers from HO91 and WT mice (Fig. 5A). In line with previous results (18), the resting [Ca 21 ] i of soleus fibers was higher than that of EDL fibers. In particular, the mean 6 S.E. resting [Ca 21 ] i in soleus was 86.5 6 6.3 nM (n = 29) and 82.3 6 4.5 nM (n = 29), and in EDL  it was 71.6 6 6.6 nM (n = 24) and 68.2 6 3.5 nM (n = 32) in WT and HO91 mice, respectively (Fig. 5A).
In the presence of 1.8 mM Ca 21 in the extracellular solution, the averaged peak Ca 21 transient induced by a single action potential in EDL fibers from WT mice was similar to that of HO91 mice (Fig. 5B). The mean (6S.D.) DF/F 0 was 1.14 6 0.52, n = 41, versus 1.17 6 0.36, n = 50, respectively (Mann-Whitney test, p . 0.05). In contrast, the peak Ca 21 transients elicited in HO91 soleus fibers were decreased by 28% compared with those of the WT (Fig. 5B) (Table S3). In soleus fibers from HO91 mice, the peak calcium transients induced by tetanic stimulation (Fig. 5B) was approximately 30% lower than that of WT mice (the mean 6 S.D. DF/F 0 was 1.12 6 0.13, n = 24, versus 1.60 6 0.11, respectively; Mann-Whitney test, p , 0.05) ( Table S3). Analysis of calcium homeostasis in flexor digitorum brevis (FDB) fibers from Het91 mice revealed no differences (i) in the resting [Ca 21 ] i , (ii) in the twitch Ca 21 transient, and (iii) in the Ca 21 transient elicited by tetanic stimulation between WT and Het91 mice (Fig. S4).

Discussion
Here, we investigated the skeletal muscle phenotype of a mouse model carrying the p.A4329D missense mutation in ex91 of the Ryr1 gene. This study shows that the bi-allelic expression of the RyR1 p.A4329D mutation is phenotypically different from the mono-allelic expression of the same Ryr1 mutation on the genetic background of a compound heterozygous carrier. In particular, the presence of the bi-allelic expression of the RyR1 p.A4329D mutation causes a mild muscle phenotype affecting mostly slow-twitch muscles, with minor, if any, effects on fast-twitch EDL muscles. Slow-twitch muscles from RyR1A4329D homozygous mice show a 60% reduction of RyR1 and Ca v 1.1 protein content and no changes in Stac3 content, whereas no changes were apparent in total homogenates of fast-twitch EDL muscles. The reduction of RyR1 content in slow-twitch muscles is accompanied by (i) a small reduction of minimal Ferret's diameter of soleus fibers and a 30% reduction . RyR1 protein and transcript levels are altered in soleus muscles from HO91 mice but not in EDL muscles. A, quantitative real-time PCR of the indicated transcripts in skeletal muscles from HO91 mice relative to WT littermates (the latter were set as 1). Transcript levels were normalized to Actn2 as a muscle-specific gene using the DDC T method. *, p , 0.05 (Student's t test). B, representative immunoblots of total homogenates of EDL and soleus muscles from WT and HO91 mice probed with the indicated antibodies. of (ii) peak twitch and tetanic force development and (iii) peak tetanic calcium transients.
Fiber type specificity of the muscle phenotype associated with the bi-allelic expression of the RyR1 p.A4329D mutation The basic mechanism of ECC is similar between different fiber types; however, there are functional and structural heterogeneities within the calcium homeostasis toolkit between fastand slow-twitch muscles controlling the kinetics of the calcium transients leading to muscle contraction (19). The ECC macromolecular complex of slow-twitch muscles shows distinct qualitative and quantitative features compared with those of fasttwitch muscles that might be important and account for the fiber type specificity of the muscle phenotype of the HO91 mice. Indeed, in slow-twitch muscles the membrane density of the DHPR and RyR1 is ;2to 3-fold lower than that present in fast-twitch muscles (20)(21)(22), and we hypothesize that a further reduction in their content, below a critical level, impacts ECC, as seen in soleus muscles. As to the cause of the decreased RyR1 content in HO91 soleus muscles, it cannot be because of protein instability, as it would have occurred in all muscle types. In this context, it is interesting that the mechanical properties of extraocular muscles from HO91 mice were similar to those of WT littermates, as was the RyR1 protein level (23). The reduction in RyR1 and Ca v 1.1 may be caused at least in part by alterations of transcriptional regulation brought about by epigenetic modifications. Compatibly, in soleus muscles (but not in EDL muscles), HDAC9 levels were increased. HDAC9, like HDAC4, HDAC5, and HDAC7, is a member of the class II HDAC family (24), and these enzymes are involved in chromatin modification and regulation of mef-2-dependent transcription (25,26). Furthermore, HDAC4 and HDAC5 levels are increased in the muscles of patients carrying recessive RYR1 mutations (13). Calcium-induced calcium release in mammalian skeletal muscles: fast-twitch versus slow-twitch muscles The domain between rabbit RyR1 residues 4274 and 4535 is a regulatory domain (27). Additionally, we have shown that the RyR1 p.A4329D mutation decreases the sensitivity of RyR1 to calcium and caffeine (15). The A4329 residue is distant from the amino acid residues that have been proposed to come from the RyR1 calcium binding sites in the linear RyR1 sequence (28). High-resolution cryo-EM of RyR1 has shown that E3893, E3967, and T5001, plus amino acids H3895 and Q3970, make up the high-affinity calcium binding site, which is likely to be involved in calcium-dependent RyR1 activation. The resolution of the amino acid sequence encompassing residue A4329 of the available RyR1 3D structure is poor, and presently it is still not possible to establish exactly whether A4329 interacts with any of the above-mentioned key residues to destabilize the conformation of the calcium binding site. As to the mechanism by which the low-calcium and caffeine sensitivity of RyR1 p. A4329D channels contribute to the muscle phenotype of the HO91 mice, it may relate to transcriptional regulation. Indeed, it is possible that the downregulation of calciuminduced calcium release in slow-twitch fibers affects the activity of calcium-dependent transcription factors such as MEF2 and NFAT (18,29), which, in turn, might be involved in the expression of RyR and Cav1.1 in muscles. Our results suggest that the calcium-dependent regulation of the RyR1 is relevant, at least in part, in determining the molecular and functional signature of slow-twitch muscles (30, 31) but have a negligible effect on the characteristics of fast-twitch muscles.
Bi-allelic expression of the RyR1 p.A4329D mutation versus mono-allelic expression of the RyR1 p.A4329D mutation on the genetic background of a compound heterozygous carrier Comparison of compound heterozygous RyR1Q1970fsX161 A4329D mice with age-matched homozygous RyR1A4329D mice shows that the former exhibit a defect of postnatal body weight gain. The weight gain defects affect both male and female RyR1Q1970fsX161A4329D double knock-in mice. In the present paper, we show that the bi-allelic expression of the single RyR1 p.A4329D mutation does not affect the body weight growth curve or the phenotype of fast-twitch muscles. Therefore, the impairment of the early postnatal development of the compound heterozygous RyR1Q1970fsX161A4329D mice might affect the maturation of fast-twitch muscles, leading to a decrease of RyR1 expression and the appearance of a muscle phenotype in EDL muscles.
In conclusion, our data are consistent with the hypothesis that intracellular pathways that are activated by the bi-allelic expression of the RyR1 p.A4329D mutation are not identical to those activated by the mono-allelic expression in a compound heterozygous genetic background, and that such a divergence accounts for the diversity of the muscle phenotype of the mice carrying these recessive RYR1 mutations.

Compliance with ethical standards
All experiments were carried out on 8-12-week-old male mice unless otherwise stated. All experimental procedures were approved by the Cantonal Veterinary Authority of Basel Stadt (BS Kantonales Veterinäramt permit numbers 1728 and 2115).
Genotyping RyR1A4329D mice and real-time qPCR PCR amplification of Ryr1 exon 91 was performed on genomic DNA of WT and HO91 mice using specifically designed primers (Table S1), as previously described (15). GoTaq® DNA polymerase (Promega) was used in the amplification of the product, as previously described (15). The amplified DNA products were digested with PvuII (R0151L BioLabs) for 1 h at 37°C, separated on a 7.5% polyacrylamide gel, and stained with ethidium bromide to visualize DNA bands. For quantitative real-time PCR (qPCR), cDNA was prepared from total RNA extracted from frozen muscles using TRIzol ® (Ther-moFischer; 15596026) according to the manufacturer's instructions. DNA was removed using DNase I (Invitrogen; 18068-015), and 1000 ng RNA was reverse transcribed into cDNA using the high-capacity cDNA reverse transcription kit (Applied Biosystems; 4368814). The cDNA was amplified by qPCR using the primers listed in Table S1, and transcript levels were quantified using Power SYBR ® Green reagent master mix (Applied Biosystems; 4367659), using the Applied Biosystems 7500 Fast real-time PCR system running 7500 software, version 2.3, as previously described (15). Transcript quantification was based on the comparative DDC T method. Each reaction was performed in duplicate, and results are expressed as relative gene expression normalized to that of actinin (Actn2).

In vivo muscle strength assessment
Mice were individually housed in cages equipped with a running wheel carrying a magnet, as previously described (15). Wheel revolutions were registered by reed sensors connected to an I-7053D digital-input module (Spectra), and the revolution counters were read by a standard laptop computer via an I-7520 RS-485-to-RS-232 interface converter (Spectra). Digitized signals were processed by the ''mouse running'' software developed at Santhera Pharmaceuticals. Total running distance (meters) and speed (km/h) were evaluated.

In vitro muscle strength assessment
To test muscle force in vitro, EDL and soleus muscles were dissected from 12-week-old mice and mounted onto a muscle force-transducing setup (Heidelberg Scientific Instruments), as previously described (15). Muscle force was digitized at 4 kHz by using an AD Instrument converter and stimulated with 15-V pulses for 1.0 ms. EDL tetanus was recorded in response to a train of pulses of 1.0-ms duration delivered at 50/100/150 Hz and 50/100/120 Hz for EDL and soleus muscle, respectively, for 400 ms and 1100 ms. Specific force was normalized to the muscle cross-sectional area [wet weight (mg)/length (mm) 3 1.06 (density, mg/mm 3 )].
For electrically evoked Ca 21 transients, single fibers were incubated for 10 min at 19°C in Ringer's solution containing 10 mM low-affinity calcium indicator Mag-Fluo-4 AM (Thermo-Fischer), 50 mM N-benzyl-p-toluene sulfonamide (BTS, Tocris). Fibers were rinsed twice with fresh Tyrode's solution, and measurements were carried out in Tyrode's solution containing 50 mM BTS as previously described (15). Measurements were carried out with a Nikon Eclipse inverted fluorescent microscope equipped with a 203 PH1 DL magnification objective. The light signals from a spot of 1-mm diameter of the magnified image of single fibers were converted into electrical signals by a photomultiplier from Myotronic (Heidelberg, Germany) connected to a PowerLab Chart 5 analog digital interface. Fluorescent signals were analyzed using PowerLab Chart 5 and Origin 6.0 software. Changes in fluorescence were calculated as DF/F 0 = (F max 2 F rest )/(F rest ). Kinetic parameters were analyzed using PowerLab Chart5 software. Fibers were excited at 480 nm and then stimulated with a single pulse of 50 V of 1-ms duration or with a train of pulses of 50 V of 1-ms duration delivered at 100 Hz for 300 ms. Fibers were isolated from 3-5 mice, and the results were pooled.

Histological examination
Muscles from WT and HO91 mice were isolated and mounted for fluorescence microscopy imaging. Changes in muscle fiber type and calculation of the minimal Feret diameter, the closest possible distance between the two parallel tangents of an object, were determined as described previously (32). Images were obtained using an Olympus IX series microscope and analyzed using CellP software.

Biochemical analysis of total muscle homogenates
Total muscle homogenates were prepared from EDL and soleus muscles of WT and HO91 mice. SDS-PAGE and Western blotting of total homogenates were carried out as previously described (13,15). Western blots were stained with the primary antibodies listed in Table S2, followed by peroxidase-conjugated protein G (Sigma, P8170; 1:130,000) or peroxidase-conjugated anti-mouse IgG (Fab-specific) antibody (Sigma, A2304; 1:200,000). The immunopositive bands were visualized by chemiluminescence using the WesternBright ECL-HRP substrate (Witec AG). Densitometry of the immune-positive bands was carried out using Fusion Solo S (Witec AG). A representative immunoblot of each antibody on total muscle homogenates from WT mice is shown in Fig. S3.

Statistical analysis
Statistical analysis was performed using Student's unpaired t test for normally distributed values and the Mann-Whitney U test when values were not normally distributed. A p value of ,0.05 was considered significant.

Data availability
All data are contained within the manuscript.