GCN2 deficiency protects mice from denervation-induced skeletal muscle atrophy via inhibiting FoxO3a nuclear translocation

Several recent clinical studies have indicated that dietary supplementation with branched-chain amino acids (BCAA), particularly with leucine, is an effective anti-atrophic therapy (Bauer et al., 2015; Tsien et al., 2015; English et al., 2016). In animal models, BCAA can prevent denervation (Ribeiro et al., 2015), hindlimb suspension (Maki et al., 2012; Jang et al., 2015) or dexamethasone-induced (Yamamoto et al., 2010) muscle atrophy. General control nonderepressible 2 kinase (GCN2) is a well-known amino-acid sensor. Under conditions of amino-acid deprivation, the increased level of uncharged transfer RNA (tRNA) activates GCN2 through binding to the histadyl-tRNA synthetase-like domain (Wek et al., 1995). Upon activation, GCN2 phosphorylates eukaryotic initiation factor 2 alpha at Ser51, which leads to translational arrest and restoration of amino acid homeostasis (Wek et al., 1995; Sood et al., 2000). As amino acids are potent modulators of protein turnover in skeletal muscle, we proposed that GCN2 may affect denervation-induced muscle atrophy, but the detail mechanism remains unclear. To investigate the impact of GCN2 on the development of muscle atrophy, we performed sciatic denervation procedure on hindlimb muscles in wild-type (WT) and Gcn2 mice. On day 7 after denervation, WT tibial anterior (TA) muscle mass decreased to 70.9% ± 1.8% of the contralateral level, while GCN2-deficient TA muscle mass remained at 83.1% ± 1.6% of its contralateral level (Fig. 1A). GCN2 deficiency also significantly attenuated the muscle mass loss in atrophied gastrocnemius (GAS) and extensor digitorum longus (EDL) muscles. Similar results were observed on day 14 after denervation (Fig. S1). Wheat germ agglutinin (WGA) staining of muscle cryosections demonstrated that GCN2-deficient TA muscles had a better preservation of myofiber size in response to denervation (Fig. 1B). After denervation, myofiber size distribution calculated from WT TA muscles showed a leftward shift from its contralateral conditions. However, such shift was delayed in atrophied GCN2-deficient TA muscles (Fig. 1C). To further verify whether activation of GCN2 contributes to muscle atrophy, we overexpressed GCN2 in flexor digitorum brevis (FDB) muscles using in vivo electroporation. The transfection efficiency was confirmed by Western blot (Fig. S2). After denervation for ten days, the diameter of WT FDB myofiber decreased from 37.2 ± 0.9 μm to 24.2 ± 0.7 μm, while the diameter of Gcn2 FDB myofiber was maintained at 32.3 ± 0.7 μm. Overexpression of GCN2 resulted in a further reduction in the diameter of FDB myofibers in both WT and Gcn2 mice. However, the reduction was more dramatically in Gcn2 FDB muscle and the significant difference in the diameter of atrophied FDB myofiber between WT and Gcn2 mice was diminished after transfected with the GCN2 plasmid (Fig. 1D). Emerging evidence suggests that the protein degradation in muscle atrophy is mediated by FoxO3a, an important regulator of Atrogin-1, MuRF-1 and LC3 (Sandri et al., 2004; Zhao et al., 2007; Guo et al., 2016). Upon atrophy stimuli, FoxO3a shuffled into the myofiber nucleus, which leads to transcriptional activation (Sandri et al., 2004). Through Western blotting and immunostaining, we observed that the phosphorylation level of FoxO3a at Ser207 were increased in TA muscles of WT mice after denervation, and FoxO3 was mainly located in the nucleus. However, the denervationinduced phosphorylation and nuclear accumulation of FoxO3a were significantly less in GCN2-deficient muscles compared with that in WT muscles (Fig. 2A–B). We also observed significant increases in protein expression of MuRF-1 and Atrogin-1, as well as the ratio of conversion of LC3 into the activated forms (LC3-II/I) in WT atrophic TA muscles. However, increases in E3 ubiquitin ligases expression and activation of autophagy were statistically less remarkable in atrophic Gcn2 muscles than in WT (Fig. S3). To investigate whether GCN2 activation directly causes FoxO3a nuclear translocation in muscle atrophy, we generated a stable C2C12 cell line (mGCN2-C2C12) with doxycycline (Dox)-controlled expression of flag-tagged mouse


Mice and Denervation-Induced Muscle Atrophy
Male C57BL/6J (HFK Bioscience Co., Beijing) and GCN2 -/mice (Harding et al., 2000) (congenic with the C57BL/6J strain, kindly provided by Dr Yingjie Chen from University of Minnesota), 8-10 weeks of age, were used. Animal studies were performed in accordance with the principles of laboratory animal care (NIH publication no. 85-23, revised 1985) and with approval by the University Of Chinese Academy of Science Animal Care and Use Committee. The denervation procedure was performed on WT (n=20) and GCN2 -/mice (n=23) as previously described (Wei et al., 2013;Tang et al., 2015). Muscle samples were harvested at 7 days or 14 day after the denervation surgery.

Cross-Sectional Area Assessment
As described previously (Wei et al., 2013), frozen muscle sections (8 μm) were stained wheat germ agglutinin (WGA) and the cross-sectional area (CSA) was quantified using NIH Image J software (Bethesda, Maryland, USA). At least 5 mice/group were used for these experiments. The results were expressed as the mean CSA±S.E. and as the percentage of fibers distributed.
Briefly, mice were first anaesthetized and received a single injection of 0.4 U of hyaluronidase (Sigma) in 25 μl PBS into the flexor digitorum brevis (FDB) muscle.
After 1 h, 10μg of control plasmid pIRS2-EGFP or pIRS2-EGFP-GCN2 was injected into each FDB, followed by 20 electrical pulses with a duration of 20 ms to ensure the absorption of the plasmid. The electrical field intensity was 100 V/cm. Ten days after electroporation, mice were euthanized and muscles were collected.

Cell Culture
C2C12 myoblasts were obtained from the American Type Culture Collection (ATCC, Manassas, VA) and grown in DMEM supplemented with 10% FBS and 1% penicillin and streptomycin. Myoblast fusion and differentiation was induced in subconfluent cells by replacing the medium with DMEM supplemented with 2% horse serum. To generate a stable, doxycycline-inducible GCN2 overexpression cell line, 5×10 5 exponentially growing cells were transfected with pLVX-Tet3G-GCN2 lentivirus for 24 h, followed by puromycin selection (1μg/mL) for 3 weeks.

Immunoprecipitation
The mouse FoxO3a cDNA in the pEGFP-C2 vector was transfected into mGCN2-C2C12 cells using Lipofectamine 2000 (Invitrogen, Grand Island, NY, USA) according to the manufacturer's instruction. Doxycycline was added to the cells 12 h post-transfection to inducing GCN2 expression. After transfection for 48 h, the cells were collected and lysed with the EBC lysis buffer, which contained 50 mmol/L Tris(pH8.0), 120 mmol/L NaCl, 0.5% NP-40 and protease inhibitor cocktail (Roche, Switzerland). Immunoprecipitation was performed as previously reported (Bi et al., 2010). Briefly, lysates were precleared with protein A/G Plus-agarose beads (Santa Cruz, Dallas, Texas, USA) at 4 °C for 20 min. Following the removal of the beads by centrifugation, lysates were incubated with anti-Flag or anti-GFP antibodies in the presence of 15 μl of protein A/G Plus-agarose beads overnight at 4 °C. After washing four times, the immunoprecipitates were subjected to immunoblotting. At least 3 independent experiments were performed.

Western Blotting
Freshly isolated TA muscle (10-20 mg) was homogenized in buffer (50 mM Tris-Cl, 150mM NaCl, 100 μg/ml phenylmethylsulfonyl fluoride, protease and phosphatase inhibitor cocktail from Roche and 1%Triton X-100) on ice for 30 min. After centrifugation at 12,000 ×g for 20 min at 4 °C, the supernatant was used for western blot analysis as previously described (Guo et al., 2016). Primary antibodies used in this study were as follows: the antibodies against GCN2 and LC-3 were from Cell Signaling Technology (Danvers, MA, USA); the antibodies against Flag, GFP and phospho-FoxO3a ser207 were from Invitrogen (Grand Island, NY, USA); and the antibodies against Atrogin-1, MuRF-1 and β-actin were from Abcam (Cambridge, UK).
Data and Statistical Analysis. All values are expressed as the mean ± standard error.
Statistical significance was defined as p< 0.05. One-way or two-way analysis of variance (ANOVA) was used to test each variable for differences among the treatment groups with StatView (SAS Institute Inc). If ANOVA demonstrated a significant effect, pair-wise post hoc comparisons were made with Fisher's least significant difference test.