Lack of Musashi‐2 induces type IIa fiber‐dominated muscle atrophy

Skeletal muscle is a highly plastic tissue, adapting its structure and metabolism in response to diverse conditions such as contractile activity, nutrients, and diseases. Finding a novel master regulator of muscle mass and quality will provide new therapeutic targets for the prevention and treatment of muscle weakness. Musashi is an RNA‐binding protein that dynamically regulates protein expression; it was originally discovered as a cell fate determination factor in neural cells. Here, we report that Musashi‐2 (Msi2) is dominantly expressed in slow‐type muscle fibers, fibers characterized by high metabolism and endurance. Msi2 knockout (KO) mice exhibited a decrease in both soleus myofiber size and number compared to control mice. Biochemical and histological analyses revealed that type IIa fibers, which are of the fast type but have high metabolic capacity, were decreased in Msi2 KO mice. The contraction force of isolated soleus muscle was lower in KO mice, and the expression of the metabolic proteins, cytochrome c oxidase and myoglobin, was also decreased in KO muscle. Our data demonstrate the critical role of Msi2 in the maintenance of normal fiber‐type composition and metabolism.


| INTRODUCTION
Skeletal muscle is composed of a mixture of different types of muscle fibers, which are defined by myosin heavy chain (MHC) isoforms and metabolic activity. Skeletal muscle fibers are classified as either slow-twitch fibers (type I) or one of three types of fast-type fibers (type IIa, IIx, and IIb). While type I fibers exhibit slow contraction speed and fatigue resistance, type II fibers are prone to fatigue but have high contraction speed and strength. When classified by metabolism, type I and type IIa fibers are classed as oxidative fibers because they have high metabolic capacity, while type IIx and IIb fibers are classed as glycolytic. Thus, type IIa fibers exhibit an unusual phenotype, that is, fasttwitch fibers that also have a high metabolic capacity comparable to type I fibers. 1,2 The population of muscle fiber types is inherently determined by species, body part, and genetics, [3][4][5][6] and is also regulated by various stimuli, such as physical activity, aging, diet, and diseases. [7][8][9] Exercise training is one of the most effective ways to induce metabolic adaptation via increase in oxidative capacity in skeletal muscles. Previous research has found that exercise training does not change fiber type from type II to type I fibers, but instead induces a shift within the type II subpopulations. For example, long-term running exercise training induces a fiber type shift from type IIb (IIx) to type IIa without a change in type I fiber number, 10,11 suggesting that the increase in fast oxidative fibers (i.e., type IIa) is what constitutes the adaptation of skeletal muscles to endurance exercise training. In addition, aging mainly induces a decrease in the fiber size and number of type IIa fibers, rather than in type I fibers. 12,13 In this context, understanding the molecular mechanism and finding the factors that lead to the induction of type IIa fiber is an attractive research subject because it contributes to clinical applications to treat metabolic syndrome as well as establishing effective exercise training programs.
Change in cellular phenotype is induced by dynamic regulation of gene expressions. Regulation occurs at all steps of the central dogma; the transcriptional regulations have been studied in depth while investigating the determination of muscle fiber type. [14][15][16] On the other hand, post-transcriptional regulation may also prominently affect gene expression and cell phenotype specification. Musashi (Msi) is an RNA-binding protein that is known to post-transcriptionally regulate genes involved in cell development and differentiation. Msi is a key player in determining cell behavior for development, but its function varies between organs. Msi was originally discovered in neural cells as an essential factor for normal brain development, 17 and has been recognized as a critical element in oncogenesis, regulating self-renewal in cancer stem cells. 18 More recently, two different groups reported that Msi2 is expressed in skeletal muscle cells and acts on myogenesis through miRNA-7 repression. 19,20 However, the function of Msi2 in skeletal muscle is still controversial. The first report of skeletal muscle Msi2 was a study using dystrophic muscle cells isolated from a patient suffering from myotonic dystrophy type 1, a severe muscle wasting disease. This study showed that Msi2 was upregulated in patient-derived muscle cells and that it was a responsible factor for causing muscle atrophy. 20 On the other hand, Yang et al. reported that Msi2 expression increased following muscle differentiation and promoted myogenesis. They showed that disruption of Msi2 prevented muscle differentiation and regeneration, suggesting that Msi2 is a new factor for improving muscle regeneration. 19 Here, we demonstrate that Msi2 positively regulates both skeletal muscle mass and fiber types. Msi2 is abundantly expressed in slow-type muscle fibers, especially in type IIa fibers. Msi2-deficient mice exhibited muscle atrophy caused by reduction of type IIa fibers in soleus muscle. The contraction force of isolated soleus muscle from mice lacking Msi2 was lower than that of control mice, suggesting that Msi2 is necessary to maintain regular function in muscle contraction. Our data suggest that Msi2 is a new element to control the quality of skeletal muscle and could be a drug target to treat metabolic syndrome and muscle atrophy.

| Animals
Mice were bred and maintained in the animal care facilities of Tokyo Metropolitan University. All animal experiments were performed according to protocols approved by the Experimental Animal Care and Use Committee of Tokyo Metropolitan University (permit number A31-22). The Msi2 mutant mice, B6; CB-Msi2Gt(pU-21T)2Imeg (Msi2−/−), were established using gene-trap mutagenesis as described previously. 21 All experiments comparing WT and KO mice were performed simultaneously in sex-and age-matched mice, unless otherwise noted. All mice used in this study were between 8 and 12 weeks of age.

| Myofiber isolation
Extensor digitorum longus (EDL) and soleus muscles were isolated and digested in 0.2% type I collagenase at 37°C for 2 h, as described previously. 22 Muscle tissues were pipetted with a glass Pasteur tube coated with 5% bovine serum albumin (BSA). Myofibers were dispersed by pipetting and confirmed to be thoroughly clean and not to include other cells such as fibroblasts using microscopy under high magnification. The myofibers were washed twice with phosphate-buffered saline (PBS) and used for western blotting.

| Western blotting
Frozen muscle tissue was pulverized and homogenized in lysis buffer 23 and centrifuged at 13 000 g for 15 min at 4°C. Isolated myofibers were sonicated in lysis buffer and centrifuged in the same manner. For MHC detection, samples were prepared by homogenization in SDS buffer as described below. The protein samples were separated using sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) before transferring to polyvinylidene fluoride membranes. The membranes were blocked with Tris-buffered saline containing 5% non-fat dry milk and 0.1% Tween 20. The membranes were incubated overnight with the appropriate primary antibodies, anti-MHC I (Sigma, M8421), anti-MHC IIa (DSHB, sc-71), anti-Msi2 (Abcam, ab76148), anti-COX IV (Cell Signaling Technology, 4844), anti-myoglobin (Abcam, ab77232), anti-βactin (Cell Signaling Technology, 4967), anti-GAPDH (Cell Signaling Technology, 2118), and anti-GLUT4 (Cell Signaling Technology, 2213), before incubating with mouse or rabbit secondary antibodies (GE Healthcare, Chicago, IL) conjugated to horseradish peroxidase. The blots were subsequently developed using Pierce™ ECL (Thermo Fisher Scientific, MA) or ECL plus (PerkinElmer, Waltham, MA) and analyzed using a Luminescent Image Analyzer Im-ageQuant800 (Cytiva). Data were quantified using Im-ageJ software.

RNA sequence analysis
Total RNA was extracted from skeletal muscle tissue using the TRIzol reagent (Invitrogen). RNA was transcribed into cDNA via a standard reverse transcriptase reaction using the PrimeScript™ first-strand cDNA Synthesis Kit (Takara, Shiga, Japan) according to the manufacturer's protocol. These cDNAs were analyzed for mRNA content using conventional PCR and quantitative real-time PCR. Real-time PCR was performed on a 96-well CFX Connect Real-Time PCR Detection System (Bio-Rad Laboratory, Tokyo, Japan) using a THUNDER-BIRD Probe qPCR Mix (TOYOBO, Osaka, Japan). The mRNA levels of each gene were normalized to those of the housekeeping gene GAPDH. The primers used in this study are listed in Table S1. For RNA sequencing, total RNA obtained from each sample was subjected to a sequencing library construction using TruSeq Stranded mRNA Library Prep Kit (Illumina, Inc., San Diego, CA) according to the manufacturer's protocols. The equally pooled libraries of the samples were sequenced using No-vaSeq 6000 (Illumina, Inc.) in 101 base-pair (bp) pairedend reads. All procedures for RNA preparation, RNA sequencing, and data analysis were performed by DNA Chip Research Inc. (Tokyo, Japan).

| Immunostaining
For immunohistochemistry, the calf muscle samples were immediately frozen in 2-methylbutane cooled in liquid nitrogen and stored at −80°C before cryosectioning. Consecutive 10-μm sections were cut using a cryostat (CM 1850; Leica Microsystems, Wetzlar, Germany) and mounted on APS-coated glass slides (Matsunami, Osaka, Japan). The cross sections were fixed with 4% paraformaldehyde, blocked with PBS containing 0.3% Triton X-100 and 5% goat serum, and incubated overnight at 4°C with primary antibodies: anti-Msi2 (Abcam, ab76148), anti-laminin (Sigma, L9393), and anti-neurofilament (Dako, 2F11). All immunostained samples were visualized using appropriate speciesspecific Alexa Fluor 488, 594, or 647 fluorescenceconjugated secondary antibodies (Thermo Fisher Scientific). For staining of acetylcholine receptor, anti-Bungarotoxin conjugate including Alexa Flour 555 (Thermo Fisher Scientific, B13422) was used. Immunostaining of the lumbar spinal cord was performed as described previously. 24 To determine the muscle fiber type, immunofluorescence analysis of MHC expression was performed. Tissue sections were blocked with 10% goat serum in PBS for 60 min at room temperature and incubated with primary antibodies against MHC I (BA-F8), MHC IIa (SC-71), and MHC IIb (BF-F3) for 2 h at room temperature. The sections were incubated with a mixture of second antibodies including Alexa Fluor 350 IgG2b, Alexa Fluor 488 IgG1, and Alexa Fluor 555 IgM for 1 h at room temperature. The fluorescence was visualized using a fluorescence microscope BZ-X810 (Keyence, Osaka, Japan). Cell numbers and area were quantified using the Keyence software BZ-H3C and H3CM.

| SDS-PAGE for MHC isoform separation
Whole soleus muscle samples (100 ng protein) were run on 8% polyacrylamide gels (99:1 of the acrylamide monomer/N,N′-methylenebisacrylamide ratio) as described previously. 8 Briefly, frozen powdered samples were homogenized with pestles in an SDS solution buffer [0.1 M Tris-HCl pH 8.0, 5 mM EDTA, 10%SDS, 40 mM dithiothreitol (DTT) with 1% protease inhibitor cocktail (25955, Nacalai, Japan)] followed by centrifugation at 13 000 g for 15 min. The supernatants were diluted in 2x sample buffer (4% (w/v) SDS, 100 mM DTT, 43% (v/v) glycerol, 0.16 M Tris-HCl buffer (pH 6.8), and 0.2% (w/v) bromophenol blue). Protein concentration was measured by the bicinchoninic acid method standardized with BSA. Samples were incubated at 95°C for 3 min. The loaded protein amount was set at 0.01% of the original muscle tissue and diluted in the same ratio. Samples (5 μL) were run at a constant voltage of 140 V for about 22 h in an incubator set at 4°C with gentle stirring of the pre-cooled lower buffer; gels were silver stained (37937-96, Silver Stain KANTO III; Kanto Chemical, Japan) and scanned using a scanner (CanoScan 9000F Mark II, Canon, Japan) followed by densitometry of MHC isoforms using ImageJ software.

| Mice locomotor activity and glucose tolerance test
Mice were housed individually in cages equipped with infrared sensors for monitoring locomotor activity (ARCO system, Locomotor activity measurement system). After acclimatizing to the cages for 24 h, the amount of locomotor activity was measured for 24 h.
Mice were fasted for over 16 h from 16:00 p.m. Glucose (16806-25, Nacalai Tesque, Japan) was administered intraperitoneally at 1 g/kg body weight; blood was collected from the tail vein at time points 0, 15, 30, 60, 90, and 120 min for the glucose tolerance test. Blood glucose concentration was determined using a glucometer (One-Touch Ultra).

| In vitro glucose transport
After sacrificing the mice, the soleus muscles were rapidly removed and treated for in vitro muscle incubation as previously described. 25 Briefly, both ends of the muscle strips (tendons) were tied with sutures and mounted on an incubation apparatus (Uchida Denshi, Hachioji, Japan). The muscles were preincubated for 30 min in Krebs-Ringer bicarbonate buffer (KRB) containing 2 mM pyruvate with or without 50 mU/mL insulin (Eli Lilly, Indianapolis, USA). The buffers were kept at 37°C throughout the experiment and gassed continuously with 95% O 2 and 5% CO 2 .
For glucose transport, KRB containing 1 mM 2-deoxy-D-glucose (2DG) was added and the muscles were incubated at 37°C for 10 min with or without 50 mU/mL insulin. Glucose transport was terminated by dipping the muscle tissue in ice-cold KRB. It was then weighed and frozen in liquid nitrogen. The muscle tissue was digested in a similar way to the method described in western blotting. To deactivate the proteins, 50 μL of the samples was incubated at 80°C for 15 min and centrifuged at 12 000 g for 20 min at 4°C, and the supernatant was used for measurement of 2DG concentration. The 2DG concentration was measured using the 2DG Uptake Measurement Kit (Cosmo-bio, Japan) following the manufacturer's protocol.

| In vitro contraction force in isolated muscles
The contraction force induced by electrical stimulation of isolated muscles was measured as described previously. 25 The soleus muscles were removed, and both ends of the tendon in the isolated muscle were tied with silken threads and mounted on apparatuses. Muscles were incubated in KRB buffer with 2-mM sodium pyruvate for 20 min at 37°C. The muscles were stimulated to contract for 10 min with the electrodes attached underneath the lid using a pulse generator (Uchida Denshi, Japan) under the following conditions: train rate = 2/min, train duration = 10 s, pulse rate = 100 Hz, duration = 0.1 ms, voltage = 50 V. Muscle tension was monitored using a transducer (Kent Scientific, Torrington, CT) and analyzed using the Power Lab system (AD Instruments, Sydney, Australia). Force generation was plotted using the integrated values for each 10-s increment over a period of 10 min.

| Cloning of the Msi2 gene
Mouse Msi2 cDNA was cloned from the TA muscle using an RT-PCR-based method. The forward primer used for its amplification included the Kozac sequence (CACCATG) before the start codon of the coding sequence for mouse Msi2 (5′-GTCAC CAT GGA GGC AAA TGG GAGCC-3′). The reverse primers included recognition sites to complement the coding sequence for the C-terminal end of mouse Msi2 (5′-TCAGT GGT ATC CAT TTG TAA AGGCC-3′). The PCR was conducted for 28 cycles, and amplified fragments were cloned into the pCR2.1-TOPO cloning vector (Thermo Fisher Scientific), digested with the restriction enzyme EcoRI, and then introduced into the pCAGGS vector (provided by Dr. Miyazaki at Osaka University).

| DNA injection into skeletal muscle and in vivo electroporation
DNA injection and in vivo electroporation were performed according to a previous study. 26 Mice were anesthetized and the plasmid DNA was diluted in 0.9% NaCl saline to a final concentration of 2 μg/μL. An insulin syringe was used to inject 25 μL of DNA solution intramuscularly into the tibialis anterior (TA) muscle along the long axes of the muscle fibers. Immediately after the injection, an electrode and a pair of stainless-steel needles were inserted into the skeletal muscle, which was then stimulated with eight square-wave electric pulses (200 V/cm) using an electrical pulse generator (Uchida Denshi, Japan). Two weeks after electroporation, the muscles were dissected and analyzed.

| Statistical analysis
Data are expressed as the mean ± standard error of the mean (SEM). Two-sided unpaired t-tests were used to compare data between the two groups. Variations were compared using a one-way ANOVA, and a Tukey-Kramer post hoc test was conducted if the ANOVA indicated a significant difference. The level of significance was set to p < .05.

| Msi2 is strongly expressed in slow-type muscles
Msi2 is expressed in different types of stem cells, such as neural stem cells and hematopoietic stem cells, 27,28 and in cultured muscle cells. 19 However, the expression level of Msi2 protein in skeletal muscle cells and the expression pattern depending on the type of skeletal muscles have not yet been fully clarified. Our western blotting analysis demonstrated that Msi2 was expressed in skeletal muscle tissue as well as in the brain ( Figure 1A). Since the homogenates of skeletal muscle tissue include a variety of cells including neural cells, adipocytes, and endothelial cells besides skeletal muscle cells, we isolated single myofibers to eliminate the possibility of Msi2 contamination from non-muscle cells; we demonstrated that the Msi2 protein is evidently expressed in myofibers (skeletal muscle cells) ( Figure 1A).
The expression level of Msi2 protein was compared between EDL and soleus muscles, each of which largely corresponds to fast-and slow-type muscles, respectively ( Figure 1B). Msi2 was strongly expressed in the soleus compared to the EDL in both tissue and myofibers. This means that Msi2 is abundant in slow-type muscles, as shown by the trend in the expression of myoglobin, which is a typical marker for slow-type muscles ( Figure 1C).
To visualize the expression patterns of Msi2 protein in different types of muscles, we performed immunohistochemical analysis using cross sections of calf muscles, which are composed of the gastrocnemius, plantaris, and soleus. Msi2 expression was stronger in the soleus than in the gastrocnemius or plantaris ( Figure 1D). Muscle fiber types were identified on serial sections by staining MHC I, IIa, and IIb using the specific antibodies, confirming that the soleus was composed of type I (blue) and type IIa (green) fibers while gastrocnemius and plantaris were mainly composed of type IIb (red) fibers ( Figure 1D,E). These data supported the results of western blotting that Msi2 is abundant in slow-type muscles. We noticed that the signals of Msi2 vary from cell to cell in soleus tissues, so we checked the muscle fiber types on the serial section. Figure S1A shows typical staining indicating that there were type I and type IIa but few type IIb fibers in soleus. The unstained cells (black) indicated type IIx fibers. The staining signal of Msi2 was quantified in each muscle fiber and compared between fiber types. This showed that Msi2 expression was higher in type IIa fibers compared to type I and type IIx fibers ( Figure S1B).
To validate the staining of Msi2 in muscle sections, we stained the soleus of Msi2 KO mice and confirmed that no Msi2 signal was detected ( Figure S1C). The images showed that Msi2 was strongly expressed on the plasma membrane and weakly distributed throughout the myofibers. For the accuracy of Msi2 localization in myofibers, the section was observed by confocal laser scanning microscopy ( Figure S1D), indicating that Msi2 is localized not only in the plasma membrane but also in the cytoplasm and nucleus.
Msi2 was originally found to be expressed in neuronal cells, and the present study shows that expression in the brain is higher than in skeletal muscle. We also stained motor neurons in skeletal muscle tissue and checked for Msi2 expression. Msi2 expression was observed in all choline acetyltransferase (ChAT)-positive motor neuron somata from fast to slow motor neuron types ( Figure S2A). Figure S2B,C shows images of co-staining of Msi2 with α-Bungarotoxin and Neurofilament, which are markers of acetylcholine receptors and nerve axons, respectively. Msi2 is expressed in motor axon terminals, and the expression level is not low, suggesting that Msi2 detected in skeletal muscle tissue includes Msi2 expressed in motor nerves.

| Skeletal muscle analysis in Msi2 knockout (KO) mice
To investigate the role of Msi2 in skeletal muscle, we analyzed mice in which the Msi2 gene was disrupted using a gene-trap vector. 21 Msi2 KO mice exhibited a significant decrease in body weight (Figure 2A). Although EDL muscle weight was not altered between genotypes, the soleus muscle weight decreased significantly in KO mice compared to wild-type (WT) mice in both females ( Figure 2B) and males ( Figure S3A,B). The deletion of the Msi2 protein in skeletal muscle was determined by western blotting of the soleus ( Figure 2C). Interestingly, the normally red color of the soleus in WT mice changed to white in Msi2 KO mice as shown in Figure 2D. The red color is one of the characteristics of slow-type muscles because there is much myoglobin protein which contains red ferrous ions. The EDL was not atrophied in the KO mice, but the original white color became even whiter ( Figure 2D).
We prepared cross sections of the soleus from WT and Msi2 KO mice, staining them with laminin to measure myofiber size and number. As shown in Figure 2E, the myofiber size was smaller in Msi2 KO mice than in WT mice. We quantified the area and number of all myofibers in the soleus of Msi2 KO mice, confirming a significant decrease in both myofiber size ( Figure 2F,H) and number ( Figure 2G).

KO mice
Based on the observation that the color of the soleus was changed from red to white, we hypothesized that a fibertype shift occurs in the soleus of Msi2 KO mice, from slow to fast. We performed triple staining of MHC I, IIa, and IIb and determined four types of myofibers including MHC IIx, which was not stained and shown as dark regions ( Figure 3A). As shown in Figure 3A, the soleus is composed of mainly type I and IIa fibers, with few type IIb fibers. We counted the numbers of each myofiber, showing that the number of type IIa fibers was lower in Msi2 KO mice than in WT mice ( Figure 3B). Numbers of type I and IIx fibers were not changed ( Figure 3B).
To validate the decrease in type IIa fibers in Msi2 KO muscles, we homogenized the soleus and evaluated expressions of MHC isoforms using SDS-PAGE. The isolated muscles were weighed and homogenized in the same  Figure 1D. Scale bar is 100 μm.
amount of lysis buffer. The homogenates were diluted by the same factor (1/10 000) and loaded for electrophoresis. Since muscle weight was different in each sample, the amount of protein used for SDS-PAGE was different. Figure 4A shows representative silver staining of MHC isoforms in WT and Msi2 KO mice; each MHC protein was identified by the molecular weight. 8 As quantified in Figure 4B, only the expression of MHC IIa including MHC IIx was decreased in Msi2 KO mice. In accordance with immunohistochemical analysis in Figure 3B, the expressions of MHC I and MHC IIb were not changed between the genotypes ( Figure 4B). Since MHC IIa and IIx proteins were not distinguished by the experiment for SDS-PAGE of MHCs, we also quantified the expression levels of MHC isoforms by western blotting using specific antibodies. While the amount of MHC I was not changed in the soleus of the Msi2 KO mice, MHC IIa expression was lower (Figure 4C). The quantification of blots showed that MHC IIa was significantly decreased in the Msi2 KO soleus muscles ( Figure 4D). This multiple analysis strongly demonstrated that the number of type IIa fibers was decreased in the soleus of Msi2 KO mice.

KO mice
Type IIa fibers are classified as fast-type fibers but exhibit a phenotype with high oxidative capacity and endurance because the fibers have many mitochondria. Therefore, we analyzed the metabolic phenotype induced by Msi2 deletion. The locomotor activity was measured for 24 h after adaptation in a metabolic chamber. As shown in Figure 5A, the activity was lower in Msi2 KO mice than in WT controls in both dark and light periods. Total activity as calculated from the area under the activity count curve was significantly decreased in Msi2 KO mice ( Figure 5B). A few days after measuring locomotor activity, we performed a glucose tolerance test to examine glucose metabolism in the whole body. Blood glucose levels after glucose injection were higher in Msi2 KO mice compared to WT mice ( Figure 5C,D), suggesting that glucose disposal capacity is impaired by Msi2 deletion.
To examine whether the impairment of glucose disposal occurred in skeletal muscle, we measured glucose transport in isolated skeletal muscles. Soleus muscles were isolated and incubated in chambers. 2DG was added to the muscles and the effect of insulin (50 mU/mL) on glucose transport was investigated. Glucose transport was calculated by 2DG amount and normalized to protein concentration in homogenized samples. The results showed that both basal and insulinstimulated glucose transport were not changed by Msi2 deletion (Figures 5E and S3C), suggesting that glucose uptake in skeletal muscle was maintained in Msi2 KO mice. We also recalculated glucose transport per muscle tissue rather than per muscle weight. Because the soleus muscle was atrophied, glucose transport per muscle tissue was lower in Msi2 KO mice under both basal and insulin-stimulated conditions ( Figure 5F,G). Glucose transport was also measured in the EDL and found to be unaltered by Msi2 deletion ( Figure S3C). However, the EDL was not atrophied in Msi2 KO mice; glucose transport per tissue was maintained in the muscle ( Figure S3D,E).
To validate the glucose uptake and analyze metabolic capacity in the skeletal muscles, the expressions of key proteins for metabolism were measured by western blotting. Glucose transporter 4 (GLUT4) expression was not decreased but rather increased in Msi2 KO muscle (Figure 6A), supporting the results that glucose uptake was not impaired by Msi2 deletion. However, cytochrome c oxidase IV (COX IV), a protein localized to the inner membrane of mitochondria, was drastically decreased in Msi2 KO muscle ( Figure 6B). Myoglobin, which is an oxygen-binding protein expressed in slow fibers, was also reduced by deletion of Msi2 ( Figure 6B). Since peroxisome proliferator-activated receptor-gamma coactivator (PGC1α) is a key molecule in regulating both fiber type switching and mitochondria biogenesis, we examined whether the PGC1α mRNA expression was changed in Msi2 KO mice by real-time PCR. However, there was no difference in PGC1α expression between WT and KO mice ( Figure S4A).
To investigate whether the changes in metabolismrelated protein expression caused by Msi2 deficiency are muscle-type specific, protein expression levels were also compared genotypically in the EDL. Similar to the soleus, COX IV was also significantly decreased in Msi2 KO EDL muscles, but the degree of decrease was less (42%) than in the soleus, where COX IV expression was decreased by 72% ( Figure S4B). Myoglobin expression was drastically decreased in Msi2 KO EDL muscles ( Figure S4C). GLUT4 expression was not altered in Msi2 KO EDL muscles ( Figure S4D), whereas GLUT4 was increased in Msi2 KO soleus muscles. These data suggest that the metabolic changes caused by Msi2 deletion are not the same in different muscle types.
Since Msi2 is an RNA-binding protein, we examined the mRNA expression level in the soleus of WT and KO mice. Consistent with the change in protein expression, myh2, the gene encoding MHC IIa, was decreased in the Msi2 KO muscles ( Figure S4E). Myoglobin mRNA was significantly decreased in Msi2 KO mice, but cox4 and glut4 did not change between WT and Msi2 KO. To perform a wide analysis of the effect of Msi2 deletion on the soleus muscle, a comprehensive gene expression analysis by RNA sequencing was performed. The genes measured by realtime PCR were confirmed to be decreased in the Msi2 KO muscles. A previous study showed that Msi2 contributes to muscle wasting via hyperactivated autophagy. 20 Our data show that the set of Atg factors involved in autophagosome formation in mammalian cells was increased in the soleus muscle of Msi2 KO mice ( Figure S4F). These results suggest that autophagy can be induced in the absence of Msi2, potentially implicating autophagy in the process of myofiber loss. However, a previous investigation showed contradictory results to the current study, indicating that autophagy was increased in correlation with increasing Msi2 levels. Therefore, a comprehensive investigation of the relationship between Msi2 and autophagy is needed.

| Contraction force of isolated muscles
The changes in fiber type and metabolic capacity potentially affect muscle contraction force and fatigue. To examine this, we isolated soleus muscles from WT and Msi2 KO mice and measured the electrical stimulationinduced muscle contraction force in vitro. For both, the total amount of force generation in the soleus decreased with repeated muscle contractions (Figure 7). However, absolute values of muscle contraction force from the first to fifth contractions were lower in Msi2 KO mice, suggesting a decrease in muscle strength in the mice. Because the Msi2 KO mice had lower body weight and muscle weight, the contractile force was recalculated by normalizing to body weight and myofiber size, respectively (the mean myofiber size from Figure 2F was used). Contractile force normalized to body weight remained low in Msi2 KO mice ( Figure S5A). On the other hand, the contractile force corrected for the myofiber area was no longer different between WT and Msi2 KO mice ( Figure S5B), indicating that the specific muscle force, which is the tension of the myofiber itself, was maintained in Msi2 KO mice. These results suggest that the decrease in contractile force in Msi2 KO mice was caused by a decrease in myofiber number and atrophy of myofibers.

| Effect of Msi2 overexpression in skeletal muscle on metabolism-related protein expression
To determine whether increased Msi2 expression in adult WT mice increases type IIa fibers and metabolism-related proteins, we performed gene transfer of Msi2 into mouse skeletal muscle by electroporation. We originally isolated Msi2 cDNA from skeletal muscle and then introduced it into the pCAGGS expression vector. The cloned Msi2 F I G U R E 5 Effect of Msi2 deletion on whole-body and skeletal muscle glucose disposal. Locomotor activity was monitored in wild-type (WT) and Msi2 knockout (KO) mice for 24 h. Data were recorded every 3 h as a time series (A) and over 12 h as a total (B) for dark and light periods, respectively. *p < .05. Values are presented as mean ± SEM (n = 4). (C, D) Glucose tolerance test results. Mice were injected i.p. with glucose (1 g/kg), and blood from the tail vein was provided for measuring glucose. Plasma glucose concentration (C) and respective areas under the curves (AUC) above baseline (D) in WT and KO mice are given. *p < .05. Values are presented as mean ± SEM (n = 4). (E) Glucose transport induced by insulin in the soleus muscle in WT and KO mice. Isolated soleus muscles were incubated with or without insulin (50 mU/mL) for 10 min. Glucose transport was normalized to protein concentration in homogenized samples, and the graph shows the fold increase over the basal value for glucose transport. Insulin-induced glucose transport was observed in WT and KO muscles, but there were no significant differences between genotypes. *p < .05. Values are presented as mean ± SEM (n = 4). (F) The glucose transport data in Figure  5E were recalculated and expressed as the amount of glucose per tissue in both basal (F) and insulin-stimulated (G) conditions. Because the value was not normalized to the muscle tissue, the glucose transport in the muscle of KO mice was decreased due to the lower muscle weight. *p < .05. Values are expressed as mean ± SEM (n = 4).
cDNA was sequenced and confirmed to be identical to Msi2 (NM_054043.3). Two weeks after electroporation, the TA muscles overexpressing Msi2 and contralateral TA muscles injected with an empty vector control were isolated and analyzed. Western blotting showed that MHC IIa protein, which is a marker of type IIa fibers, was not significantly increased by Msi2 overexpression, although the mean value tended to be higher ( Figure 8A,B). On the other hand, the levels of myoglobin and COX IV protein significantly increased in the TA muscles overexpressing Msi2 (Figure 8C,D). Electroporation of skeletal muscle enables a strong increase in protein expression in the injected cells, but the successful transduction of genes is limited to a small subset of cells. Although overexpression of Msi2 did not change the muscle fiber type significantly, the expression of proteins involved in metabolism increased, suggesting that Msi2 in skeletal muscle directly regulates muscle metabolism.

| DISCUSSION
Our data demonstrated that Msi2 regulates not only skeletal muscle mass but also quality in mouse skeletal muscles. Msi2 was enriched in slow-type muscle soleus compared to fast-type muscles, and histological analysis F I G U R E 7 Contraction force of isolated muscles from WT and Msi2 knockout mice. The contraction force of the isolated soleus in wild-type (WT) and Msi2 knockout (KO) mice was evaluated. Electrical stimulations were applied for 10 s and repeated 10 times with a 50-s rest period. Force generation was measured with a strain gauge and recorded. Contraction force was lower in KO muscles than in WT muscles, from the first to fifth contractions. *p < .05. Values are presented as mean ± SEM (n = 4).

F I G U R E 8
Effect of Msi2 overexpression in TA muscle by electroporation on protein expression. Msi2 or an empty vector was introduced into the tibialis anterior (TA) muscle by electroporation. Two weeks after electroporation, the TA muscles were lysed and used for western blotting. Msi2 increased in Msi2-electroporated muscles compared to empty vector-injected muscles (A). The expression of MHC IIa, a marker of type IIa fibers, was not significantly increased by Msi2 electroporation (B). Myoglobin (C) and COX IV (D) protein levels, which were decreased in Msi2 KO muscles, increased following Msi2 overexpression. *p < .05. Values are presented as mean ± SEM (n = 6). revealed that Msi2 was strongly expressed in slow-type fibers, especially in type IIa fiber. The mice lacking Msi2 have fewer type IIa fibers and show muscle atrophy as well as decreased strength in soleus muscles. The dysfunctions in whole-body glucose tolerance and fatigue resistance in Msi2 KO mice might be induced by the decrease in type IIa fibers since the fibers have high oxidative capacity and endurance capacity. This is supported by previous reports that the decrease in type IIa fibers and the increase in type IIb fibers were associated with lower endurance performance and diabetes. 29,30 Type IIa fibers exhibit a rare combination of both endurance and explosive power; they are highly plastic, increasing with exercise training and decreasing with inactivity. Therefore, elucidation of the molecular function of Msi2 will contribute to an understanding of muscle adaptation mechanisms and the prevention of muscle deterioration. This report is the first step in proving the significance of Msi2 in skeletal muscle function even though the downstream signaling of Mis2 was not clarified.
Dysfunctions in whole-body glucose tolerance are usually induced by impaired glucose disposal in skeletal muscles. However, insulin-stimulated glucose transport in isolated soleus muscles was maintained in Msi2 KO mice in our experiments. One potential reason is that the loss of muscle mass reduces glucose utilization throughout the body. Whole-body glucose disposal is determined by the integrated actions of multiple tissues, but skeletal muscle primarily accounts for insulin-mediated blood glucose clearance. 31 The body weight loss in Msi2 KO mice (Figure 2A) indicates that skeletal muscle mass was decreased throughout the body, not only in leg muscles. Even though glucose transport capacity was not changed significantly, total glucose disposal capacity in skeletal muscle was decreased by the loss of muscle mass in Msi2 KO mice. Indeed, glucose transport per soleus tissue was decreased in the KO mouse under both basal and insulin-stimulated conditions, respectively ( Figure 5F,G). Another is the compensation effect of Msi2 deletion, with increased expression of GLUT4, a glucose transporter, which translocates to the plasma membrane in response to insulin, in Msi2 KO mice. Although the metabolic capacity in skeletal muscle was decreased by Msi2 deletion as is clear from the data showing decreased COX IV protein, the glucose transport capacity was maintained by an increase of GLUT4 expression.
In this study, we used whole-body KO mice, which is a limitation in directly demonstrating that Msi2 in skeletal muscle is solely responsible for inducing the observed phenotype. Given Msi2 expression in the brain and in multiple other tissues of the body, it is important to consider the potential effects of Msi2 deletion in multiple other tissues. We confirmed that Msi2 is expressed in motor neurons in skeletal muscle and could not deny the possibility that a loss of Msi2 in motor neurons contributes to muscle atrophy and weakness in Msi2 KO mice. However, if a Msi2 deficiency in the motor nerve causes muscle atrophy, then the effects of Msi2 deficiency should be prominent in the EDL, which has more muscle fibers and is innervated by many motor neurons compared to soleus, 32,33 because Msi2 expression was observed in all types of motor neuron ( Figure S2). The phenotype induced by Msi2 deletion was observed mainly in the soleus, suggesting that the contribution of Msi2 in the motor neuron to muscle atrophy is small.
To confirm that the metabolic change in Msi2 KO mice was induced by a loss of Msi2 in skeletal muscle and not by a systemic effect of a global loss of Msi2, muscle-specific conditional KO mice are needed. Instead, we showed that transient overexpression of Msi2 in skeletal muscles increased key metabolic proteins, which were decreased in Msi2 KO mice. Due to the limited nature of gene transfer by electroporation, which only affects specific myofibers within the muscle rather than the entire tissue, the changes observed in myofiber type and metabolism were not as clear-cut as the phenotype observed in the KO mice. Nevertheless, these results support the hypothesis that the phenotype observed in Msi2-deficient mice is primarily due to the absence of Msi2 in skeletal muscle. Homeostasis of skeletal muscle mass is controlled not only by myofibers but also by factors around cells, such as satellite cells, mesenchymal progenitors, endothelial cells, and neurons. Further studies are needed to elucidate the mechanism by which Msi2 expression in these cells coordinates the maintenance of skeletal muscle mass.
The elucidation of molecular pathways by which Msi2 regulates muscle fiber type and metabolism is the next theme in our research. Notably, our data showed that expressions of PGC-1α were not changed in Msi2 KO muscles although COX IV protein, a mitochondria marker, was decreased ( Figure 6). PGC-1α is a key molecule in inducing fiber-type change and increases mitochondrial metabolic capacity in skeletal muscle, mediating exerciseinduced adaptation. Based on these data, Msi2 acts downstream of the PGC-1α pathways. It is necessary to clarify to which genes Msi2 binds and regulates translation in skeletal muscle cells by comprehensive gene analysis.

AUTHOR CONTRIBUTIONS
Yasuro Furuichi conceived and designed the research; Yasuro Furuichi, Ayana Furutani, and Kotaro Tamura conducted the research and acquired the data; Yasuro Furuichi, Ayana Furutani, Kotaro Tamura, Yasuko Manabe, and Nobuharu L. Fujii analyzed and interpreted the data. All the authors were involved in drafting and revising the manuscript.