Down‐regulation of the mitochondrial i‐AAA protease Yme1L induces muscle atrophy via FoxO3a and myostatin activation

Abstract Muscle atrophy is closely associated with many diseases, including diabetes and cardiac failure. Growing evidence has shown that mitochondrial dysfunction is related to muscle atrophy; however, the underlying mechanisms are still unclear. To elucidate how mitochondrial dysfunction causes muscle atrophy, we used hindlimb‐immobilized mice. Mitochondrial function is optimized by balancing mitochondrial dynamics, and we observed that this balance shifted towards mitochondrial fission and that MuRF1 and atrogin‐1 expression levels were elevated in these mice. We also found that the expression of yeast mitochondrial escape 1‐like ATPase (Yme1L), a mitochondrial AAA protease was significantly reduced both in hindlimb‐immobilized mice and carbonyl cyanide m‐chlorophenylhydrazone (CCCP)‐treated C2C12 myotubes. When Yme1L was depleted in myotubes, the short form of optic atrophy 1 (Opa1) accumulated, leading to mitochondrial fragmentation. Moreover, a loss of Yme1L, but not of LonP1, activated AMPK and FoxO3a and concomitantly increased MuRF1 in C2C12 myotubes. Intriguingly, the expression of myostatin, a myokine responsible for muscle protein degradation, was significantly increased by the transient knock‐down of Yme1L. Taken together, our results suggest that a deficiency in Yme1L and the consequential imbalance in mitochondrial dynamics result in the activation of FoxO3a and myostatin, which contribute to the pathological state of muscle atrophy.

inhibited. In contrast, the phosphorylation of FoxO3a via AMPK or JNK promotes the activity of FoxO3a. Inhibition of the IGF-1 (insulin-like growth factor-1)/Akt/mTOR pathway and elevation of AMPK together facilitate FoxO3a activity, which concomitantly leads to muscle atrophy.
Growing evidence has shown that mitochondrial dysfunction is one of the important factors in muscle atrophy. 6 Mitochondria are dynamic organelles, and their functions are maintained by mitochondrial quality control systems which comprise mitochondrial biogenesis, fusion, fission, proteostasis and mitophagy. A balance between these procedures is important for the maintenance of mitochondrial homeostasis, and conditions that shift the balance towards mitochondrial fission lead to mitochondrial fragmentation. 7  In humans, m-AAA proteases are homo-or hetero-oligomeric proteins composed of ATPase family gene 3-like protein 2 (AFG3L2) and paraplegin (Spg7), whereas i-AAA proteases consist of homo-oligomeric Yme1L proteins. 11 Together with AAAs, the ATP-independent peptidase Oma1, also located in the IM, participates in mitochondrial dynamics through the processing of Opa1. 12 In this study, we demonstrated that muscle atrophy was closely related to an imbalance in the mitochondrial quality control system.
In particular, a deficiency in Yme1L accelerated this imbalance towards mitochondrial fission and mitophagy, subsequently activating FoxO3a and myostatin, which are essential for the degradation of muscle proteins. Apparently, Yme1L plays an important role in the maintaining healthy mitochondria in the muscle and loss of Yme1 induces the pathological state of muscle atrophy.

| Animal studies
Male C57BL/6 mice (7 weeks old) were purchased from Koatech (Pyeongtaek, Korea) and were housed at a constant room temperature (23°C) with a 12 hours light/dark cycles and free access to food and water. After a week of adaptation, the animals were randomly assigned to two groups: Control group (n = 8) and hindlimb immobilization (disuse) group (n = 8). For the induction of muscle atrophy, the hindlimbs of the mice were immobilized by stapling the foot to the limb using a surgical stapler (Visistat Skin Stapler 35W, WECK, Teleflex), as previously described. 13 After 5 days of immobilization, the animals were anaesthetized, and then, the gastrocnemius muscles (GA) were dissected and immediately frozen in liquid nitrogen for subsequent analyses. All animal procedures were performed according to the Korean National Institutes of Health Animal Care and Use Committee (Permit Number: KCDC-061-18-2A), and the animals were cared for in accordance with the Korea National Institute of Health guidelines.

| Grip strength and rotarod test
Five days after immobilization, forelimb skeletal muscle strength was measured using a digital grip strength metre (Bioseb). The mean of three consecutive trials was considered an index of grip strength.
For the rotarod test (Med-associated), the mice were placed on a rotating rod that accelerated from 4 to 40 rpm over 5 minutes. 14 The mean time to fall from the rod (s) in three different trials was analysed individually.

| Measurement of myosin heavy chain distribution
Serial paraffin-sections of GA muscle were deparaffinized and permeabilized with 1% Triton X-100 for 5 minutes before the samples were blocked with 10% normal goat serum. Then sections were stained with the following antibodies: SC-71 for the Type 2a MyHC Fluorescence was detected using an FV3000 laser scanning confocal microscope (Olympus Corporation).

| Cell culture and Yme1L small interfering RNA transfection
Murine C2C12 myoblasts were purchased from the American Type Culture Collection (ATCC) and maintained in medium consisting of Dulbecco's modified Eagle's medium (DMEM; Gibco; Thermo Fisher Scientific, Inc) supplemented with 1% penicillin-streptomycin (Gibco)/ 10% FBS (Gibco) in 5% CO 2 at 37°C. For the induction of myogenic differentiation, the cells were differentiated into myotubes with DMEM supplemented with 2% horse serum for 5 days. To induce mitochondrial dysfunction, the C2C12 myotubes were treated with 2,

| Determination of mitochondrial membrane potential
The mitochondrial membrane potential of C2C12 myotubes was assessed with a JC-1 mitochondrial membrane potential assay kit  and Drp1 (611738) were purchased from BD Bioscience.

| Quantitative PCR analysis
Total RNA was extracted from differentiated C2C12 myotubes or gastrocnemius muscle tissue using the RNeasy Mini kit (Qiagen) according to the manufacturer's instructions. cDNA was synthesized using SuperScript III reverse transcriptase (Invitrogen; Thermo Fisher Scientific, Inc). Quantitative PCR was performed using SYBR Green Master Mix (Thermo Fisher Scientific) in a total volume of 20 µL with the QuantStudio 6 Flex system (Thermo Fisher Scientific). The expression of the target genes was normalized to GAPDH expression, and relative quantification was evaluated with the 2 −ΔΔCt method.
The primer pairs for the specific target genes were designed as listed in Table S1.

| Statistical analysis
All results are expressed as the means ± standard error of the mean (SEM). Statistical analysis was performed using Student's t tests with GraphPad Prism software (GraphPad). For multiple comparison, one-way ANOVA (analysis of variance) followed up by Bonferroni's multiple comparison test were used to analyse statistical differences. Values of P < .05 were considered statistically significant.

| Muscle atrophy due to hindlimb immobilization is correlated with mitochondrial dysfunction
To elucidate the underlying mechanisms of muscle atrophy, we first established a mouse model of skeletal muscle atrophy, which was generated by hindlimb immobilization (disuse) for 5 days. As expected, grip strength, endurance on the rotarod and muscle fibre diameter were decreased in the hindlimb-immobilized mice compared with the control mice ( Figure 1A,B). There are four distinct fibre types in mammalian skeletal muscles: oxidative slow-twitch type 1 fibres, oxidative fast-twitch type 2A fibres and glycolytic fast-twitch type 2X fibres and 2B fibres, each containing different myosin heavy chain (MyHC) isoforms, referred to as MyHC1, MyHC2a, MyHC2x and MyHC2b, respectively. 15 Campos's group and Bloemberg's group showed that type 2B fibres are predominant in the gastrocnemius (GA) muscle in mice. 16,17 Coincidently, when we observed muscle fibre distribution in the cross-sectional area (CSA) of GA muscle by fluorescent staining, MyHC2b was the predominant isoform and was significantly reduced by hindlimb immobilization ( Figure 1C).
The MyHC1 and MyHC2a levels in GA muscle were low and did not change significantly following immobilization. This result was consistent with the significant reduction in the MyHC2b isoform expression in the GA muscles from the hindlimb-immobilized mice ( Figure 1D).
A previously report showed that interleukin-1 beta (IL-1β) and interleukin-6 (IL-6) are involved in the muscle protein degradation. 18 Consistent with this finding, muscle immobilization resulted in an increase in the expression of IL-1β and IL-6 in the GA muscle ( Figure 1E).
Additionally, the MuRF1, MAFbx/atrogin 1 and FoxO3a levels were markedly increased in the GA muscles of hindlimb-immobilized mice, and p97/VCP, an ATPase complex known to facilitate myofibrillar protein degradation, 19 was slightly elevated ( Figure 1F). Moreover, we observed a significant increase in the serum myostatin levels in disuse mice compared with control mice ( Figure 1G).
To confirm whether muscle atrophy induced by immobilization is accompanied by mitochondrial dysfunction, we next performed Western blot analysis to measure the expression of different mitochondrial oxidative phosphorylation (OXPHOS) complex subunits.
As shown in Figure 2A . Serum myostatin levels (n = 8/ group) were quantified by using a GDF8/ myostatin Quantikine ELISA kit (R&D Systems) according to the manufacturer's instructions. All results are representative of more than three independent experiments. The data represent the mean ± SEM. P values < .05 obtained from Student's t tests were considered significant as "atrogins". 20 Indeed, these electron transport chain complex proteins were shown to be atrogins that are negatively correlated with muscle atrophy.
Mitochondrial function is maintained by a mitochondrial quality control system. Therefore, we next measured whether proteins related to the mitochondrial quality control systems were changed.
Mfn1, Mfn2 and the long isoform of OPA1 were significantly decreased ( Figure 2B). Consistent with these findings, the levels of Drp1 and autophagy-related genes, such as LC3B, SQSTM1/p62 and Beclin 1, were strongly up-regulated in immobilized muscles.
Additionally, activating transcription factor 4 (ATF4), a transcription factor that responds to mitochondrial stress 21 ; Bcl-2/adenovirus E1B 19 kD-interacting protein 3 (Bnip3); and GABA receptor-associated protein-like 1 (Gabarapl1), which are associated with mitophagy, were significantly induced in disuse mice ( Figure 2C). Based on these results, we verified that the reduction in mitochondrial functions is closely related to muscle wasting.

| CCCP-induced mitochondrial dysfunction accompanies muscle atrophy in C2C12 myotubes
To understand the molecular mechanisms by which mitochondrial dysfunction causes muscle atrophy, we used fully differentiated C2C12 myotubes that were treated with CCCP. As expected, the mitochondrial membrane potential and total ATP level were substantially reduced along with the reduction in the mitochondrial electron transport chain complex proteins in CCCP-treated myotube cells (Figure 3 A,B). Mitochondrial dysfunction is known to increase intracellular ROS generation. 22,23 As shown in Figure 3C, the intracellular ROS levels were drastically elevated and the reduced by pretreatment with the antioxidant N-acetylcysteine (NAC) in C2C12 myotubes. Next, we measured the proteins related to the mitochondrial dynamics and found that the levels of Mfn1, Mfn2 and the long form of OPA1 gradually decreased, while the expression of Fis1 increased upon treatment with CCCP and these effects were rescued by pretreatment with the antioxidant N-acetylcysteine (NAC) in C2C12 myotubes ( Figure 3D,E).
Notably, the abnormal elevation of ROS shifted the mitochondrial dynamic balance towards mitochondrial fission.
We next examined whether the mitochondrial dysfunction as a result of CCCP induces muscle atrophy. As shown in Figure 4A and 4, the diameter of myotubes and the expression of MyHC were decreased by CCCP treatment. Coincidently, the levels of AMPK, MuRF1 and FoxO3a activity were significantly increased in the cells treated with CCCP ( Figure 4C,E). In addition, we found that atrogins related to muscle atrophy, such as ATF4, myostatin and Gabarapl1 were significantly activated, and all these effects were attenuated by NAC ( Figure 4D,E). To clarify the effects of mitochondrial dysfunction and the subsequent elevation of intracellular ROS on muscle atrophy, we specifically pretreated C2C12 myotubes with the mitochondria-targeted antioxidant Mito Q.
F I G U R E 2 Disturbance in the balance of mitochondrial dynamics by hindlimb immobilization. Western blot analysis of mitochondrial OXPHOS complex subunits (A) and mitochondrial fusion and fission marker proteins (B) in the GA muscle. C, Relative mRNA levels of ATF4, BNIP3 and Gabarapl1 in the GA muscle (n = 8). All results are representative of more than three independent experiments. The data represent the mean ± SEM. P values < .05 obtained from Student's t tests were considered significant Pretreatment with Mito Q rescued mitochondrial fusion (reduction in mitophagy), and the subsequent reduction in muscle atrophy markers including FoxO3a, MuRF1, atrogin-1 and myostatin ( Figure 4F,G and H). As shown in Figure 4G, not only NAC but also the mitochondria-specific antioxidant Mito Q could effectively ameliorate muscle atrophy induced by CCCP treatment. Therefore, the abnormal elevation of ROS as a result of mitochondrial dysfunction seems to be associated with muscle atrophy.

| A reduction in mitochondrial Yme1L specifically induces muscle atrophy
We further examined the mechanisms by which elevated ROS causes muscle atrophy. The mitochondrial chaperones and proteases are essential to maintain proper mitochondria functions under stress conditions. 9 Therefore, we examined the relationships between the proteases involved in mitochondrial quality control and muscle atrophy. As shown in Figure 5A, the expression levels of LonP1 and Hsp60 were almost unchanged in our experiment; however, the expression of Yme1L was significantly reduced in CCCP-treated myotubes. In addition, these alterations were abolished by NAC pretreatment. To further directly analyse the effects of Yme1L depletion, we transfected siRNA specific for Yme1L into C2C12 myotube cells. We first showed that the expression of Yme1L was almost completely diminished ( Figure 5B) and that the depletion of Yme1L increased the short form of Opa1, which was accompanied by an elevation in Oma1 expression, as previously described ( Figure 5C). 12 To determine whether the loss of Yme1L is related to muscle atrophy, we investigated alterations in proteins involved in muscle wasting. Intriguingly, we observed that Yme1L knock-down substantially induced the activation of AMPK and FoxO3a and concomitantly increased MuRF1 expression. Oma1 is known to be responsible for the formation of s-Opa1 when Yme1L is absent. 24,25 Accordingly, the loss of Oma1 significantly reduced not only the formation of the short form of Opa1 but also level of MuRF1; however, the levels of Yme1L itself were not altered in siOma1-transfected cells ( Figure 5D).
To test whether the loss of Yme1L specifically induces muscle atrophy, we transfected CCCP-treated myotubes with siRNA targeting LonP1. In contrast to the loss of Yme1L, the depletion of LonP1 did not alter the expression of MuRF1 ( Figure 5E). To confirm the relationship between the loss of Yme1L and muscle atrophy, we examined the expression level of Yme1L in the disused mice. As shown in Figure 5F, Yme1L was drastically reduced: consequently, Oma1 was significantly increased, whereas LonP1 was not altered in immobilized mice.

| Yme1L loss of function results in the upregulation of myostatin in myotubes
To further elucidate the role of Yme1L in muscle atrophy, we determined the morphological changes in myotubes after the transient knock-down of Yme1L. As shown in Figure 6A and 6, the F I G U R E 3 CCCP treatment causes an imbalance in mitochondrial dynamics in C2C12 myotubes. A, The mitochondrial membrane potential was evaluated as described in the Materials and Methods (left). Intracellular ATP levels were measured in C2C12 myotubes treated with 10 μmol/L CCCP in the presence or absence of 2 mmol/L NAC (right). ATP levels were estimated by using an ATP colorimetric/fluorometric assay kit (Abcam, ab83355) according to the manufacturer's instructions (B) Changes in the mRNA levels of mitochondrial OXPHOS complex subunits after CCCP treatment (C) Measurement of ROS in C2C12 myotubes as described in the Materials and Methods. Scale bars = 50 μm. D, Changes in the expression of proteins related to mitochondrial fusion and fission. E, Immunoblot analyses of proteins related to mitochondrial dynamics in C2C12 myotubes. All results are representative of more than three independent experiments. The data represent the mean ± SEM. P value < .05 obtained by one-way ANOVA was considered statistically significant thickness of the myotubes and the myotube fusion index were significantly reduced by the depletion of Yme1L. We next analysed the expression patterns of MyHC isoforms in siYme1L-transfected cells. Consistent with the morphological changes, the expression of MyHC2a was significantly decreased, consistent with the observed results in the CCCP-treated myotubes ( Figure 6C).
Interestingly, myostatin expression but not GDF15 expression was strongly increased by the transient knock-down of Yme1L ( Figure 6D). These findings were consistent with the down-regulation of genes responsible for mitochondrial biogenesis and muscle protein synthesis, such as peroxisome proliferator-activated receptor-gamma coactivator 1 alpha (PGC1α), PGC1α4 and peroxisome proliferator-activated receptors delta (PPARδ), observed in myotubes following Yme1L knock-down ( Figure 6E). Previously, McFarlane et al 26 reported that myostatin causes muscle atrophy via the repression of the IGF-1/PI3K/Akt pathway. Thus, we next investigated whether the depletion of Yme1L affects Akt signalling. As shown in Figure 6F, Yme1L knock-down inhibited Akt phosphorylation in differentiated myotubes. Intriguingly, the depletion of Yme1L also inhibited insulin-induced insulin receptor signalling and reduced insulin receptor substrate-1 (IRS-1) expression ( Figure 6F). These results suggest that the ablation of Yme1L alone causes muscle atrophy and disrupts insulin signalling, which may consequently lead to insulin resistance in atrophy-induced muscle cells.

| D ISCUSS I ON
In recent decades, the importance of mitochondrial quality control in muscle atrophy has increased. Notably, the decline of mitochondrial function and the accompanying elevation of ROS have been  12 We showed that the loss of Yme1L specifically results in muscle atrophy via the elevation of Oma1 and the concomitant accumulation of the short form of Opa1; however, the depletion of LonP1 in C2C12 myotubes did not directly induce muscle atrophy ( Figure 5E). LonP1 is a major protein in the mtUPR and is important for mitochondrial biogenesis. 27 Previously, Wagatsuma A et al 28 reported that LonP1 is reduced in hindlimb unloading mice; nevertheless, in our work the loss of LonP1 did not trigger muscle atrophy.
LonP1 is likely reduced when mitochondria are massively decreased by prolonged muscle unloading; however the reduction in LonP1 itself does not accompany muscle atrophy. In contrast, reduction in Yme1L seems directly to induce protein breakdown in muscle cells via an imbalance in mitochondrial dynamics. Indeed, Yme1L participates in various processes to maintain healthy mitochondria, including the regulation of mitophagy by controlling Oma1 and s-Opa1. 29 Yme1L is involved in the assembly of mitochondrial respiratory chain subunits, including F1F0-ATPase, Cox4 and ND1, 30 and it removes damaged or misfolded mitochondrial inner membrane proteins. 31 Recently, Yme1L was shown to participate in the degradation of F I G U R E 5 Knock-down of Yme1L specifically causes muscle atrophy. A, Reduction of Yme1L following CCCP treatment. C2C12 myotubes were treated with 10 μmol/L CCCP in the presence or absence of 2 mmol/L NAC for 24 h. Western blot analyses are shown. B, Yme1L mRNA and protein levels were determined in 6-day differentiated myotubes after transfection. C, The expression of proteins related to mitochondrial dynamics was analysed by Western blot analyses of 6-day differentiated myotubes after transfection. (D and E) The expression of proteins related to mitochondrial fusion and muscle atrophy pathways was analysed by Western blots after siOma1 (D) or siLonP1 (E) transfection. F, Immunoblot analyses of Yme1L, Oma1 and LonP1 in the GA muscle. All results are representative of more than three independent experiments. The data represent the mean ± SEM. P values < .05 obtained from one-way ANOVA were considered significant Tom22, a receptor involved in mitochondrial protein import that is located in the mitochondrial outer membrane. 32   Myostatin is a negative regulator of muscle mass, and its overexpression reduces muscle mass in adult mice and rats 34 as well as in humans. 35 However, the regulation of myostatin expression has not been clearly elucidated. Notably, our data showed that the loss of Yme1L drastically increases myostatin in C2C12 myotubes ( Figure 6D). The activation of myostatin following Yme1L depletion seems to be the important mechanisms by which an imbalance in mitochondrial dynamics induces muscle atrophy, but more studies are required to understand the precise mechanisms by which mitochondrial dynamics regulates myostatin expression.
In summary, our results demonstrated the correlations between balanced mitochondrial dynamics and the maintenance of mass and skeletal muscle function, and we suggest that Yme1L may play a pivotal role in the regulation of muscle mass by controlling mitochondrial dynamics. The loss of Yme1L activates AMPK but represses IGF-1/Akt signalling, in turn activating FoxO3a. In addition, elevation of myostatin together facilitates muscle fibre breakdown which concomitantly leads to muscle atrophy. Further studies are required to elucidate whether Yme1L is decreased in sarcopenic atrophy in mice and humans. F I G U R E 6 A reduction in Yme1L increases myostatin and perturbs insulin signalling. A, A reduction in MyHC following the depletion of Yme1L. Immunofluorescence staining for the myosin heavy chain in siRNA-transfected cells (left). Scale bars = 50 μm. Analysis of the diameter of the myotubes (right). B, Calculation of the myotube fusion index as a percentage of the nuclei within a differentiated myotube containing more than three nuclei among the total nuclei (n = 4). C, Quantitative real-time PCR analysis of MyHC isoforms (2A, 2B and 2X). D, Knock-down of Yme1L elevated myostatin expression. Relative mRNA levels of myostatin and GDF15 in siRNA-transfected C2C12 myotubes. E, Quantitative real-time PCR analysis of PGC1α, PGC1α4 and PPARδ. F, Impairment of insulin activity by Yme1L depletion. Phosphorylated and total IR and Akt protein levels were determined by Western blotting (left). Relative mRNA levels of IRS1 (right). All results are representative of more than three independent experiments. The data represent the mean ± SEM. P values < .05 obtained from Student's t tests were considered significant