Smad2/3 Proteins Are Required for Immobilization-induced Skeletal Muscle Atrophy*

Skeletal muscle atrophy promotes muscle weakness, limiting activities of daily living. However, mechanisms underlying atrophy remain unclear. Here, we show that skeletal muscle immobilization elevates Smad2/3 protein but not mRNA levels in muscle, promoting atrophy. Furthermore, we demonstrate that myostatin, which negatively regulates muscle hypertrophy, is dispensable for denervation-induced muscle atrophy and Smad2/3 protein accumulation. Moreover, muscle-specific Smad2/3-deficient mice exhibited significant resistance to denervation-induced muscle atrophy. In addition, expression of the atrogenes Atrogin-1 and MuRF1, which underlie muscle atrophy, did not increase in muscles of Smad2/3-deficient mice following denervation. We also demonstrate that serum starvation promotes Smad2/3 protein accumulation in C2C12 myogenic cells, an in vitro muscle atrophy model, an effect inhibited by IGF1 treatment. In vivo, we observed IGF1 receptor deactivation in immobilized muscle, even in the presence of normal levels of circulating IGF1. Denervation-induced muscle atrophy was accompanied by reduced glucose intake and elevated levels of branched-chain amino acids, effects that were Smad2/3-dependent. Thus, muscle immobilization attenuates IGF1 signals at the receptor rather than the ligand level, leading to Smad2/3 protein accumulation, muscle atrophy, and accompanying metabolic changes.

Skeletal muscle atrophy promotes muscle weakness, limiting activities of daily living. However, mechanisms underlying atrophy remain unclear. Here, we show that skeletal muscle immobilization elevates Smad2/3 protein but not mRNA levels in muscle, promoting atrophy. Furthermore, we demonstrate that myostatin, which negatively regulates muscle hypertrophy, is dispensable for denervation-induced muscle atrophy and Smad2/3 protein accumulation. Moreover, muscle-specific Smad2/3-deficient mice exhibited significant resistance to denervation-induced muscle atrophy. In addition, expression of the atrogenes Atrogin-1 and MuRF1, which underlie muscle atrophy, did not increase in muscles of Smad2/3-deficient mice following denervation. We also demonstrate that serum starvation promotes Smad2/3 protein accumulation in C2C12 myogenic cells, an in vitro muscle atrophy model, an effect inhibited by IGF1 treatment. In vivo, we observed IGF1 receptor deactivation in immobilized muscle, even in the presence of normal levels of circulating IGF1. Denervation-induced muscle atrophy was accompanied by reduced glucose intake and elevated levels of branched-chain amino acids, effects that were Smad2/3-dependent. Thus, muscle immobilization attenuates IGF1 signals at the receptor rather than the ligand level, leading to Smad2/3 protein accumulation, muscle atrophy, and accompanying metabolic changes.
Skeletal muscle homeostasis is maintained as a balance of anabolic and catabolic signals (1,2). Insulin or growth hormone signals activate anabolic signals, whereas starvation, cachexia, prolonged disuse/unloading, or steroid treatment promotes muscle atrophy/wasting (3). Loss of muscle volume and power in the elderly frequently hampers activities of daily living, resulting in further muscle wasting and sarcopenia, which is characterized by further loss of muscle volume and power (4). These conditions in the elderly are also associated with falls, which cause fractures and potentially contribute to dementia. Defining mechanisms underlying muscle atrophy could encourage development of therapies to prevent these complications.
Younger individuals also can exhibit muscle atrophy from long bed rest, cachexia resulting from malignancies, high dose or long term steroid treatment, or even prolonged gravity-free conditions associated with space flight. Local atrophy can occur in muscles immobilized following injuries such as muscle/ligament/tendon ruptures or bone fracture. Local muscle atrophy, however, is usually not accompanied by systemic muscle atrophy and is instead thought to be driven by signals derived in the immobilized region.
Here, we report that Smad2/3 proteins but not mRNA accumulate in immobilized skeletal muscle following either denervation or limb fixation, an outcome required for immobilization-induced muscle atrophy, as lack of Smad2/3 completely abrogated the denervation-induced atrophy phenotype. Interestingly, lack of myostatin in adult mice did not prevent denervation-induced muscle atrophy, and Smad2/3 protein still accumulated in atrophic muscles of these mice. In normal cells, Smad2/3 proteins were continuously destabilized by IGF1 signaling, although muscle immobilization blocked IGF1-receptor activation, even in the presence of the IGF1 ligand in vivo. Finally, we show that muscle immobilization resulted in reduced glucose uptake but increased levels of branched-chain amino acids (BCAA) 2 in muscle.

Experimental Procedures
Mice-Mstnfl/fl mice carrying loxP-flanked Mstn alleles as well as Ckmm Cre mice were purchased from The Jackson Laboratory (Bar Harbor, ME). Inducible conditional Mstn knockout mice (Mx1 Cre; Mstnfl/fl) were generated by crossing Mstnfl/fl with Mx1 Cre transgenic mice. For denervation-induced muscle atrophy, 8-week-old mice were anesthetized, and then a 1-mm portion of sciatic nerve was removed to denervate the gastrocnemius muscle. For fixation-induced atrophy, 8-weekold mice were anesthetized, and the hindlimb was held in a fixed position by stapling (Autoclip, BD Biosciences). Thus, immobilization on one side was achieved by either denervation or fixation, and sham surgery was performed on the other hindlimb. Animals were maintained under specific pathogenfree conditions in animal facilities certified by the Keio University Animal Care Committee. Animal protocols were also approved by the committee.
Histology-Harvested muscles were fixed in 10% neutralbuffered formalin, embedded in paraffin, cut into 4-m sections, and stained with hematoxylin and eosin. Fiber cross-sectional area (CSA) was analyzed in H&E-stained gastrocnemius muscle sections. For each muscle, CSA values were calculated by analyzing at least 1000 myofibers using BioRevo (Keyence, Osaka, Japan).
ELISA-Serum IGF1 levels were measured by ELISA (R&D Systems, Minneapolis, MN) according to the manufacturer's instructions.
Metabolome Analysis-Frozen tissue (ϳ40 mg) was added to methanol (500 l) containing internal standards (20 mol/liter each of methionine sulfone and D-camphor-10-sulfonic acid) and homogenized using a cell disrupter. Then chloroform (500 l) and Milli-Q water (200 l) were added to the homogenate. The solution was thoroughly mixed and then centrifuged at 4600 ϫ g for 15 min at 4°C, and the aqueous fraction was centrifugally filtered through a Millipore 5-kDa cutoff filter to remove large molecules. The filtrate was dried using an evacuated centrifuge and dissolved in Milli-Q water (50 l) containing 200 mol/liter reference compounds (3-aminopyrrolidine and trimesic acid) prior to capillary electrophoresis/mass spectrometry analysis. Capillary electrophoresis/mass spectrome-try-based metabolomic profiling and data analysis were performed essentially as described (32)(33)(34)(35)(36).
Statistical Analysis-Statistical analysis was performed using the unpaired two-tailed.

Smad2/3 Protein Accumulates in Atrophic Muscle-Because
immobilization-induced skeletal muscle atrophy occurs only in directly affected muscle, we hypothesized that atrophic signals are regulated cell-intrinsically rather than systemically. Thus, we examined changes in levels of intracellular proteins in skeletal muscle following unilateral sciatic nerve denervation in 8-week-old mice. Significantly reduced skeletal muscle volume was evident in gastrocnemius muscle 1 week after denervation compared with the sham-operated side (Fig. 1A), and expression of mRNAs encoding atrophy-associated ubiquitin ligases such as Atrogin-1 and MuRF1 increased in atrophic muscle compared with non-atrophic muscle (Fig. 1B). Western analysis showed that Smad2/3 proteins, which are downstream effectors of myostatin-ActRIIB signaling, accumulated significantly in atrophic muscle (Fig. 1C). The fact that our model enables comparison between denervation-induced atrophic and sham- operated non-atrophic muscle in the same mouse meant that levels of circulating factors were comparable, supporting the idea that Smad2/3 accumulation in atrophic muscle is regulated cell-intrinsically. Furthermore, neither Smad2 nor Smad 3 mRNA levels were up-regulated in atrophic compared with non-atrophic muscle (Fig. 1D). Nerve activity may function to maintain muscle homeostasis; however, accumulation of both total Smad2/3 and phospho (pSmad2/3)-protein in the absence of mRNA up-regulation was also seen in a different immobilization-induced model (a fixation model) without denervation (Fig. 2), suggesting that Smad2/3 accumulation is a universal mechanism underlying immobilization-induced muscle atrophy. Thus, we utilized the denervation model for the following in vivo immobilization-induced muscle atrophy model.
Smad2/3 Protein Accumulation Is Required for Skeletal Muscle Atrophy-To determine how Smad2/3 protein accumulation functions in skeletal muscle atrophy, we generated skeletal muscle-specific Smad2/3 doubly deficient (DKO) mice by crossing Smad2 flox/flox /Smad3 null mice with creatine kinase muscle Cre mice (Fig. 3A). Smad2 mRNA expression was significantly down-regulated in DKO mice (data not shown). Interestingly, Smad2/3 DKO mice were resistant to denervation-induced muscle atrophy shown by reduced muscle weight and CSA of fibers. These observations support the idea that Smad2/3 proteins are required for muscle atrophy (Fig. 3, B and C). Significant up-regulation of Atrogin-1 and MuRF1 expression seen in atrophic relative to control non-atrophic muscles was not evident in Smad2/3 DKO mice (Fig. 3D), supporting the idea that Smad2/3 protein accumulation is required for both atrophy of immobilized muscle and atrogene expression. The bone morphogenetic protein-Smad1/5/8 axis is reportedly activated by BMP7 to promote muscle hypertrophy (37), and Smad1/5/8 are required for myostatin-induced muscle hypertrophy (38). Indeed, we detected Smad1/5/8 activation in muscle of WT, Smad2 cKO, and Smad3 KO mice following denervation; however, Smad1/5/8 activation was less robust in DKO than other mice (Fig. 3E). Akt protein accumulated following loss of either Smad2 or Smad3 in denervated muscle (Fig. 3E). This finding suggests that hypertrophic signals via Smad1/5/8 are activated following denervation and are Smad2/3-dependent mechanisms.
Myostatin Is Dispensable for Smad2/3 Protein Accumulation and Skeletal Muscle Atrophy-Mice deficient in myostatin exhibit elevated muscle volume (28), and anti-myostatin or anti-ActRIIB antibodies have been developed as human therapies for muscle atrophy (27). Smad2/3 is reportedly activated following myostatin stimulation. To analyze how changes in myostatin expression might affect skeletal muscle atrophy in adult animals, we generated global but inducible myostatin knock-out mice by crossing myostatin flox/flox mice with interferon-inducible Mx-1 promoter-driven Cre (Mx-Cre) mice (Mstn cKO; Mx;myostatin flox/flox ). Myostatin cKO was induced at 8 weeks of age, and then denervation was performed. Following unilateral denervation, we observed denervation-induced muscle atrophy in both Mstn cKO and control mice (Fig. 4A). In addition, pSmad2/3 and Smad2/3 total protein accumulation in atrophic relative to non-atrophic muscle occurred in both Mstn cKO and control mice (Fig. 4B). Expression of both Atrogin-1 and MuRF1 significantly increased in atrophic compared with non-atrophic muscle of Mstn cKO and control mice following denervation (Fig. 4C), suggesting that myostatin is dispensable for immobilization-induced pSmad2/3 and total Smad2/3 protein accumulation and skeletal muscle atrophy.
IGF1 Signals Block Smad2/3 Protein Accumulation in Muscle-Next we employed serum starvation of C2C12 myogenic cells, an in vitro muscle atrophy model (39), to examine Smad2/3 effects in vitro. Serum-starved C2C12 cells showed Smad2/3 accumulation at protein but not mRNA levels as well as elevated expression of Atrogin-1 and MuRF1, effects seen in vivo (Fig. 5, A and B). Serum starvation conditions differ from immobilization; however, we utilized this model because Smad2/3 protein but not mRNA accumulation occurred concomitantly with the elevated Atrogin-1 and MuRF1 levels in myogenic cells in vitro as observed in vivo immobilization mod-

Smad2/3-dependent Muscle Atrophy by Immobilization
in serum-starved C2C12 cells, an effect reversed by either LY294002 or U0126 treatment (Fig. 5E). Interestingly, the ability of IGF1 to block starvation-induced Smad2/3 protein accumulation was abolished by treating cells with the proteasome inhibitor MG132 (Fig. 5F), suggesting that Smad2/3 proteins are regulated at steady state by the ubiquitin/proteasome pathway in skeletal muscle and that IGF1 loss promotes Smad2/3 protein stabilization and concomitant accumulation followed by skeletal muscle atrophy.
IGF1 Receptor Activation Is Abrogated by Muscle Immobilization-Next, we asked how muscle immobilization impairs IGF1 signaling. Recently, the E3 ubiquitin ligases Cbl-b or Fbxo40 were shown to play a crucial role in inducing muscle atrophy by degrading IRS1, an IGF1 signal transducer (10,42). However, we did not detect significant elevation of mRNAs encoding either factor in atrophic compared with non-atrophic muscle in our immobilization-induced atrophy models (data not shown). Interestingly, however, IGF1 receptor phosphorylation, as analyzed by Western blot, decreased in denervationinduced atrophic compared with non-atrophic muscle (Fig.  6A), and comparable changes were seen in fixation-induced atrophic muscle (data not shown). It is noteworthy that in this experimental system impaired IGF1 receptor activation occurs only atrophic muscle, despite the fact that both atrophic and non-atrophic muscle are exposed to equivalent levels of circulating IGF1. We, in fact, confirmed that denervated and sham-operated mice showed similar levels of circulating IGF1 (Fig. 6B).
Finally, we evaluated potential metabolic changes underlying muscle atrophy. Like insulin, IGF1 promotes cellular glucose uptake (43)(44)(45). Given that IGF1 receptor phosphorylation (and hence activation) is impaired in atrophic muscle, we analyzed glucose levels in immobilized muscle and found that they were significantly lower in denervation-induced atrophic than in sham-operated non-atrophic muscle (Fig. 6C). Lower glucose levels were accompanied by lower ATP levels in atrophic compared with non-atrophic muscle (Fig. 6C). By contrast, levels of the BCAAs leucine, isoleucine, and valine were significantly elevated in atrophic versus non-atrophic muscle (Fig.  6D), an effect indicative of protein degradation. Reduced glucose uptake into muscle was also seen in Smad2/3 DKO denervated relative to non-denervated muscles following denervation (Fig. 6E), although elevation of BCAA production in muscles by denervation was altered in DKO mice (Fig. 6F). These results suggest that muscle atrophy is accompanied by altered energy homeostasis.

Discussion
Skeletal muscle volume is controlled by a balance of anabolic and catabolic signals (1,2). Thus, decreased muscle volume occurs via either down-regulated anabolic signals, which occur in conditions such as starvation or elevated catabolic signals, which are activated by prolonged inactivity, limb fixation, muscle disuse, steroid use, or mechanical unloading (3). Here, we found that muscle immobilization by either denervation or limb fixation impairs IGF1 receptor activation based on observed changes in phosphorylation levels, followed by accumulation of Smad2/3 proteins (Fig. 6G). Because we detected phosphorylation of Smad2 and Smad3 in muscle after immobilization by either denervation or fixation, we conclude that activation and accumulation of Smad2/3 protein is required for muscle atrophy. Smad2/3 protein phosphorylation was detected in muscle following denervation even in myostatin cKO mice (Fig. 4B). Thus, further studies are needed to identify activators upstream of Smad2/3 after muscle immobilization. Interestingly, IGF1 receptor activation was inhibited specifically in atrophic but not in non-atrophic muscles in the same animal, in the presence of comparable levels of circulating IGF1 ligand. This finding means that muscle atrophy seen in immobilized muscle is not driven systemically. Smad2/3 are reported downstream effectors of myostatin-ActRIIB signaling (46); however, we found that myostatin was dispensable for immobilization-induced Smad2/3 activation. By contrast, lack of Smad2/3 in muscles resulted in significant resistance to immobilization-induced skeletal muscle atrophy and induced atrogene expression. These results suggest that Smad2/3 protein accumulation is regulated cell intrinsically via a localized mechanism rather than systemically.
We also show that muscle atrophy is closely related to energy metabolism and that BCAA generation accompanied by muscle atrophy is Smad2/3-dependent. Because BCAA cannot be synthesized de novo, elevated BCAA levels in atrophic muscles are likely due to Smad2/3-dependent protein degradation. Overall,  JUNE 3, 2016 • VOLUME 291 • NUMBER 23

Smad2/3-dependent Muscle Atrophy by Immobilization
our study suggests that Smad2/3 could be considered potential therapeutic targets to prevent skeletal muscle atrophy.
IGF1 signaling in muscle reportedly inhibits muscle atrophy through Akt-dependent phosphorylation and cytosolic sequestration of FOXOs, which are upstream activators of atrophyrelated genes (7,12). One of those atrophy-related genes, the ubiquitin ligase Cbl-b, is required to promote unloading-induced atrophy by inhibiting IGF1 signaling via ubiquitinating and destabilizing IRS1, an IGF1 signal transducer, and activating FoxO3 (10). Thus, in a muscle atrophy model, IGF1 signal-ing is impaired at the signal transducer level. However, in our immobilization-induced muscle atrophy models, we did not detect significant activation of Cbl-b expression in atrophic muscle (data not shown) but rather observed impaired IGF1 signaling at the receptor level, even under normal circulating IGF1 concentrations. We also showed that IGF1 signals suppress Smad2/3 protein accumulation following serum starvation in C2C12 cells, effects rescued by treatment with Akt or ERK inhibitors or with the proteasome inhibitor MG132. These results suggest that muscle movement may activate the Akt,
An interesting phenomena reported here is that IGF1 receptors in immobilized muscle are insensitive to circulating IGF1 ligands, whereas IGF1 receptors in non-atrophic muscle are activated in a normal fashion, a finding consistent with the report that IGF1 overexpression in skeletal muscle does not prevent unloading-induced muscle atrophy (47). Smad2/3 protein accumulation in muscle atrophy in our models did not depend on circulating myostatin. Myostatin deficiency reportedly results in muscle hypertrophy (48), but we found that myostatin inhibition in adult animals did not increase muscle vol-ume (data not shown). Moreover, IGF1 receptor deficiency in muscle reportedly significantly reduces muscle volume (49). These results suggest that muscle volume is likely regulated cell intrinsically in adult animals. At present, how immobilization promotes IGF1 receptor insensitivity in muscle remains unclear, but our results clearly show that muscle movement maintains IGF1 receptor activation as a means of achieving muscle homeostasis.

Smad2/3-dependent Muscle Atrophy by Immobilization
the mechanisms identified here, namely IGF1 receptor insensitivity and Smad2/3 protein accumulation, are applicable to other types of muscle atrophy, although common mechanisms may underlie various forms of muscle atrophy. For example, the atrogenes Atrogin-1 and MuRF1 are reportedly universally up-regulated in atrophic muscle following denervation, immobilization, or unloading; furthermore, Atrogin-1 or MuRF1 deficiency promotes partial but significant resistance to denervation-induced muscle atrophy (8). Unloading also activates FoxO3, which is upstream of Atrogin-1 and MuRF1, in atrophic muscle, and Cbl-b deficiency inhibits both FoxO3 activation and Atrogin-1/MuRF1 expression in an unloading-induced muscle atrophy model (10). By contrast, in a steroid-induced muscle atrophy model, glucocorticoid receptor activity reportedly directly induces expression of KLF15, a factor required to induce Atrogin-1 and MuRF1 expression, following dexamethasone stimulation of myogenic cells (53). KLF15 knockdown blocks the ability of dexamethasone to induce Atrogin-1 and MuRF1 (53). We found that targeting Smad2/3 in muscle promoted resistance to denervation-induced muscle atrophy and inhibited Atrogin-1/MuRF1 expression. Recently, factors other than Atrogin-1/MuRF1, such as Traf6, Musa1/Fbxo30, Smart/ Fbxo21, and Fbxo31, were reported to be up-regulated in denervation-induced muscle atrophy (54,55); however, we did not detect differences in expression of those genes in muscle of control and DKO mice after denervation (data not shown). Reasons for these discrepancies remain unknown, but they could be due to differences of experimental design. For example, Traf6 is reportedly up-regulated in tibialis anterior muscles 4 days after denervation, whereas Musa1/Fbxo30, Smart/Fbxo21, and Fbxo31 were shown to be up-regulated 3 days after denervation in gastrocnemius muscle (54,55). Taken together, although their activity may differ in specific contexts, activators of Atrogin-1/MuRF1 could serve as therapeutic targets to prevent the muscle atrophy.
We also found that deactivation of the IGF1 receptor by muscle immobilization significantly reduced intracellular glucose levels relative to normal muscle, possibly as an energy-saving mechanism. Dysfunctional muscle may instead act as a BCAA reservoir to drive the tricarboxylic acid cycle. Increasing levels of circulating BCAAs are also reported in steroid-induced muscle atrophy models (53). We did not detect BCAA elevation in sera following sciatic nerve denervation (data not shown), but that is likely due to the small amount of atrophic muscle compared with that seen in steroid-induced systemic atrophy. Nonetheless, BCAA generation following muscle atrophy may occur universally. Following denervation, muscle atrophy and concomitant increases in BCAA in that muscle were seen in both Smad2 cKO and Smad3 KO mice, but both were abrogated in Smad2/3 DKO mice (Fig. 6F). These data suggest that Smad2 and Smad3 have a redundant function in immobilization-induced muscle atrophy and BCAA production in muscle. Nonetheless, the absence of muscle atrophy plus the lack of BCAA generation seen following denervation in Smad2/3 DKO mice suggest that Smad2/3 are required for degradation of muscle proteins to BCAA in conditions of atrophy. By contrast, glucose uptake likely depends on movement-induced IGF1 receptor activation rather than Smad2/3; thus reduced glucose uptake was seen in both Smad2/3 DKO and control mice after denervation. IGF1 receptor inactivation was not evident in vitro in C2C12 cells, even under serum starvation conditions, and Smad2/3 accumulation in serum-starved myogenic cells was antagonized by IGF1 signaling in an Akt-dependent manner.
Today, sarcopenia is a serious health concern in the elderly. At present, there is no medicine available to block muscle weakness and deterioration, but understanding mechanisms underlying immobilization-induced muscle atrophy may shed light on "disuse"-induced, age-related muscle atrophy. Further studies are needed to elucidate whether mechanisms underlying immobilization-induced muscle atrophy described here are applicable to age-related sarcopenia in human subjects.