The mammalian target of rapamycin (mTOR) plays crucial roles in the regulation of protein synthesis [6] as a serine/threonine protein kinase of phosphoinositol-3-kinase (PI3K)-related kinase family. Specifically, mTOR interacts with several proteins to form the distinct mTOR complexes 1 (mTORC1) and 2 (mTORC2), which are known to differ in subunit composition, sensitivity to rapamycin, and cell signaling activity [74]. It is accepted that mTORC1 signaling is more closely related to the control of protein synthesis than mTORC2 and is negatively regulated by the tumor suppressor tuberous sclerosis complex 1/2 (TSC 1/2) [146]. Increased activity of mTORC1 stimulates protein translation and synthesis following phosphorylation of key direct downstream effectors, including the 70-kDa ribosomal protein S6 kinase (S6 K1) and eukaryotic initiation factor 4E-binding protein 1 (4E-BP1) [33, 74]. S6 K1 activation also enhances translation elongation via signals that eventually lead to decreased phosphorylation of eukaryotic elongation factor 2 (eEF2) [163]. Moreover, AMPK has been shown to interact with mTOR signaling, and activation by AICAR injection inhibits mTOR signaling and protein synthesis in rat skeletal muscle [7]. AMPK phosphorylates and activates TSC2 [59] to inhibit mTORC1 and S6 K1 activity [58] and directly phosphorylates the mTOR binding partner raptor, leading to binding of 14-3-3 to raptor and allosteric inhibition of mTORC1 [42]. Thus, mTORC1 is a potent intermediate of AMPK-mediated muscle mass regulation.

In a previous study, impairment of muscle hypertrophy during functional overload in aged rats was accompanied by reduced activation of mTOR/S6 K1/eEF2 and mTOR/4E-BP1 signaling, and these responses were negatively correlated with the phosphorylation status of AMPKα Thr¹⁷² [150]. Furthermore, Viollet and colleagues showed that S6 K1 and 4E-BP1 signaling are promoted in AMPKα1 KO mice during functional overload-induced muscle hypertrophy [99]. In addition, increased cell size in AMPKα1/α2-deficient myotubes was prevented by treatments with the mTOR inhibitor rapamycin [73]. Hence, crosstalk between AMPK and mTOR/S6 K1 signaling is likely important for muscle mass control under conditions of hypertrophy.

In contrast with these studies, we found little association of mTOR signaling with skeletal muscle atrophy [22]. Specifically, protein synthesis and phosphorylation status of S6K1 Thr³⁸⁹ were decreased during disuse-induced soleus muscle atrophy. However, suppression of AMPK activity did not rescue these responses [22], indicating that AMPK/mTOR/S6K1 signaling may not be a crucial axis in skeletal muscle atrophy.

Distinct roles of α isoforms (α1 and α2) of AMPK may influence the regulation of muscle mass via mTOR/S6K1 signaling. Accordingly, Viollet and colleagues indicated that AMPKα1-mTOR/S6K1 interactions are important in skeletal muscle hypertrophy. However, another study showed that both skeletal muscle hypertrophy and activation of mTOR/S6K1 signaling remained unchanged in response to functional overload in LKB1 KO mice [89], in which AMPKα2 but not AMPKα1 activity is completely abolished. Consistent with these findings, Viollet and colleagues showed that MyrAkt-induced hypertrophy was similar in muscle fibers lacking AMPKα2 and in control fibers [98]. These results suggest that AMPKα1 is more closely associated with mTOR/S6 K1 activation and skeletal muscle hypertrophy than AMPKα2.

In contrast with these studies, we demonstrated attenuation of unloading-induced skeletal muscle atrophy in AMPK-DN mice, in which AMPKα2 activity was almost completely inhibited but AMPKα1 activity was reduced by only 20% [22]. AMPK is activated in response to energy-depleting stresses [50], and AMPKα2 is more sensitive to energy depletion than AMPKα1 [131, 143]. Mitochondrial dysfunction-mediated impairment of ATP production has been suggested to activate AMPK under conditions of disuse and thereby contributes to muscle atrophy through protein degradation rather than protein synthesis [123]. Taken together, these observations indicate that AMPKα2 regulates skeletal muscle atrophy via mechanisms that are independent of protein synthesis pathways. These findings imply that there are isoform-dependent regulation of muscle mass, although multiple roles of AMPK α, β, and γ isoforms in muscle mass regulation remain unresolved.

The ATP-dependent ubiquitin–proteasome system is the primary protein degradation pathway and intrinsically induces degradation of myofibrillar proteins in skeletal muscle [104]. The ubiquitin ligase enzyme E3 binds its protein substrate and catalyzes the movement of ubiquitin from the ubiquitin-conjugating enzyme (E2) to the substrate. This rate-limiting step of the ubiquitination influences subsequent proteasome-dependent degradation. The human genome contains more than 650 ubiquitin ligases [75], and important skeletal muscle-specific ubiquitin E3 ligases include muscle-specific RING finger protein 1 (MuRF1) and atrogin-1/muscle atrophy F-box (MAFbx). MuRF1 and atrogin-1/MAFbx are both primarily expressed in skeletal muscle [5]. Whereas MuRF1 ubiquitinates several myofibrillar proteins, such as myosin heavy chains [16] and actin [122], atrogin-1/MAFbx targets growth-related proteins such as MyoD [152] and eIF3f [18]. Moreover, MuRF1 and atrogin-1/MAFbx mRNA levels are rapidly upregulated in numerous models of muscle atrophy and play accepted roles in the initiation of atrophy [28].

In the past decade, AMPK has been shown to interact with several E3 ubiquitin ligases. In skeletal muscle, MuRF1 and atrogin-1/MAFbx are potent targets of AMPK-mediated ubiquitination, and MuRF1 and atrogin-1/MAFbx mRNAs were upregulated following treatment with the AMPK activators AICAR [23, 68, 107, 155] and metformin [68] in skeletal muscle cells in vitro. Our recent in vivo studies demonstrated that suppression of AMPK activity attenuates increases in ubiquitination and MuRF1 mRNA expression, thus inhibiting the progress of disuse-induced muscle atrophy in mice soleus muscles [22]. Other studies have also suggested associations between AMPK and E3 ligases during muscle atrophy. Specifically, TNF receptor adaptor protein 6 (TRAF6) was associated with the dimeric ubiquitin-conjugating enzyme 13/ubiquitin-conjugating enzyme variant 1A and promoted the formation of Lys63-linked poly-ubiquitin chains rather than conventional Lys48-linked poly-ubiquitin chains, which are target proteins for degradation [71, 121]. Muscle-specific TRAF6 KO mice were resistant to muscle atrophy following denervation, cancer, or starvation, and TRAF6-mediated ubiquitination was shown to be mediated partly by activation of AMPK [116, 117]. Thus, various relationships between AMPK and the ubiquitin–proteasome system are important in the regulation of skeletal muscle mass.

Autophagy is an important cell proteolytic system that controls protein turnover in skeletal muscle [85] and is regulated by multiple proteins. Among these, Unc-51-like kinase 1 (Ulk1, also known as Atg1) is considered an important serine/threonine protein kinase during the initial stage of autophagosome formation [168]. In subsequent stages, microtubule-associated protein light chain 3 (LC3) is converted to the active form of LC3 (named LC3II) through lipidation and then participates in the formation and elongation of autophagosomes. The presence of LC3 in autophagosomes and the conversion of LC3 to the lower migrating form LC3II have often been used as indicators of autophagic activity [63]. In the final step of autophagy, autophagosomes fuse with lysosomes and their cargoes are degraded [94]. The ubiquitin-binding protein p62, which binds to LC3, is preferentially degraded during autophagy [113], and thus breakdown of p62 is generally used as a marker of autophagy flux [95].

Recently, AICAR-induced AMPK activation was reported to stimulate autophagosome formation in skeletal muscle cells [133]. Accordingly, interactions of AMPK, mTOR, and Ulk1 appear to play crucial roles in the modulation of autophagy [1, 76]. Specifically, mTOR directly binds and negatively regulates Ulk1 activity by phosphorylating the protein at Ser⁶³⁷ and Ser⁷⁵⁷ [35, 140]. In contrast, AMPK binds and phosphorylates Ulk1 at Ser³¹⁷, Ser⁴⁶⁷, Ser⁵⁵⁵, Thr⁵⁷⁵, Ser⁶³⁷, and Ser⁷⁷⁷ and thereby promotes Ulk1 activity and disrupts the Ulk1–mTOR interaction [1]. AMPK also indirectly induces autophagy by inhibiting the mTORC1 complex [35, 76].

Roles of AMPK-mediated autophagy in muscle mass regulation have been reported, and in vitro treatments with the synthetic glucocorticoid dexamethasone have been shown to increase the ratios of LC3II/LC3I dependently of AMPK in skeletal muscle cells [159]. In addition, chronic dexamethasone treatment led to muscle atrophy and AMPK activation in mice, and the muscle atrophy was attenuated by blocking AMPK activation [81]. These results indicate associations of AMPK-mediated autophagy and dexamethasone-induced muscle atrophy.

In a previous study, we demonstrated that increased LC3II/LC3I ratios in mouse soleus muscles during unloading-induced muscle atrophy were attenuated following suppression of AMPK activity [22]. However, unloading was previously shown to result in p62 protein accumulation in atrophic soleus muscles, indicating impairment of autophagic flux [95]. Accordingly, p62 protein expression was increased by unloading in mouse soleus [22], tibialis anterior, and gastrocnemius muscles [80], and no changes in LC3 expression or decreases in other autophagy-related proteins (Atg7 and Beclin-1) were reported after unloading. Taken together, these data warrant further research to precisely determine the roles of AMPK in autophagy activation during disuse-induced muscle atrophy.

AMPK-mediated autophagy has been shown to play critical roles in the regulation of skeletal muscle mass during fasting and/or aging [10]. These investigators showed that myofiber CSA in tibialis anterior muscles was reduced by fasting in wild-type mice but not in muscle-specific AMPK-KO mice and that LC3II/LC3I ratios and phosphorylation states of Ulk1 Ser⁵⁵⁵ after fasting were higher in wild-type than muscle-specific AMPK-KO mice. In addition, AMPK deficiency resulted in increased activation of mTOR and inhibitory phosphorylation of Ulk1 Ser⁷⁵⁷. These findings suggest that induction of muscle atrophy during fasting follows AMPK-mediated activation of autophagy. In addition, they also showed that AMPK deficiency led to greater aging-associated impairments of muscle quality, with increased central nuclei, large round fibers, small angular fibers, and necrotic fibers, and decreased muscle force production from that in wild-type mice. Besides, low muscle quality was reportedly accompanied by p62 accumulation, implying that aging-induced myopathy follows impairments of AMPK-mediated autophagic activity. Collectively, these findings indicate that both activation and inactivation of AMPK-autophagy systems result in muscle loss, and hence activation of autophagy systems by AMPK might be essential for muscle mass homeostasis.

The FoxOs family of transcription factors has been associated with ubiquitin–proteasome and autophagy systems [132, 144]. Among these, FoxO1, FoxO3 (also known as FoxO3a), FoxO4, and FoxO6 have been identified in skeletal muscle, and FoxO1 and FoxO3a are key factors in muscle homeostasis. Akt inhibits FoxO3a by phosphorylating residues Thr³², Ser²⁵³, and Ser³¹⁵ [9], and phosphorylation of FoxO3a at Ser²⁵³ results in exclusion from the nucleus and inhibition of transcription [9]. In contrast, AMPK phosphorylates FoxO3a at six regulatory sites (Thr¹⁷⁹, Ser³⁹⁹, Ser⁴¹³, Ser³⁵⁵, Ser⁵⁸⁸, and Ser⁶²⁶) and promotes FoxO3a transcriptional activity without affecting subcellular localization [41].

Previous studies suggest that AMPK-mediated upregulation and/or nuclear translocation of FoxO1 and FoxO3a contributes to activation of the ubiquitin–proteasome system in skeletal muscle cells [107, 133, 155]. Furthermore, phosphorylation of FoxO3a Ser⁵⁸⁸ by AMPK leads to increased expression of autophagy-related proteins such as beclin and LC3II [133]. In our experiments, decreased phosphorylation of FoxO3a Ser²⁵³ in soleus muscles was accompanied by Akt inactivation during disuse-induced muscle atrophy, but suppression of AMPK activity maintained phosphorylation of FoxO3a Ser²⁵³ at the basal level despite Akt remained inactivated [22]. These data indicate that AMPK affects the phosphorylation status of FoxO3a Ser²⁵³ through an Akt-independent mechanism during muscle atrophy. Thus, although the mechanisms behind the relationship between AMPK and FoxOs activities in muscle mass regulation have not been characterized, AMPK appears to directly or indirectly regulate FoxOs expression and/or nuclear translocation and thereby stimulates ubiquitin–proteasome and/or autophagy systems.

The transcription factor NF-κB is sequestered in the cytoplasm by a family of inhibitory proteins known as IκBα [101]. In this mechanism, the IκB kinase complex phosphorylates IκBα, resulting in its degradation, and thus facilitates nuclear translocation and activation of NF-κB. NF-κB activation has been detected under physiological and pathological atrophic conditions, including unloading, denervation, aging, cancer, sepsis, and diabetes, and pharmacological or genetic inhibition of NF-κB can protect from skeletal muscle atrophy [79, 165]. Moreover, NF-κB signaling was reportedly more important than FoxOs signaling in disuse muscle atrophy, since NF-κB sites, but not FoxOs sites, for transcription of MuRF1 during hindlimb unloading [172]. In our study, the expression of IκBα tended to decrease with muscle atrophy in mice and was higher in AMPK-DN mice than in wild-type mice after hindlimb unloading [22]. Hence, AMPK may regulate NF-κB signaling via the expression of IκBα during disuse-induced muscle atrophy. However, no other reports show associations of AMPK with NF-κB signaling during muscle mass regulation, warranting further research to clarify the precise roles of NF-κB signaling in AMPK-mediated muscle mass regulation.

HSPs are stress-induced molecular chaperones that play crucial roles in maintaining correct the folding and intracellular transport of proteins. Accordingly, as regulators of cell signaling, HSPs have various reported cytoprotective functions [69, 82, 115]. Multiple HSPs have been identified in skeletal muscle, including HSP27, HSP47, HSP60, HSP70, HSP90, and HSP110. Among these, inducible HSP70 (so-called HSP72) is the most widely studied in skeletal muscle, and profound impacts on muscle adaptation have been shown [82]. HSP72 expression is increased in skeletal muscle following exercise [97] and muscle injury [138], and during muscle regrowth and regeneration [135, 138], but is decreased by muscle disuse [22, 137]. Accordingly, overexpression of HSP72 in skeletal muscle prevented immobilization-induced atrophy in rats [137] and improved structural and functional recovery from atrophy of mouse muscles [92]. Moreover, inhibition of FoxOs and NF-κB signaling is considered a potent mechanism by which HSPs regulate skeletal muscle mass [21, 70, 136, 137]. Therefore, HSP72 may be a positive regulator of skeletal muscle mass that downregulates protein degradation pathways.

In our studies, AMPK was implicated in the regulation of HSP72 expression during skeletal muscle hypertrophy and atrophy. Specifically, AICAR-induced AMPK activation decreased HSP72 protein expression in skeletal muscle cells and was not induced following suppression of AMPKα1/α2 [23]. This was the first report to show that AMPK modulates HSP72 expression in skeletal muscle. Furthermore, we recently demonstrated that metformin-induced AMPK activation also downregulated HSP72 mRNA and protein expression in skeletal muscle cells [24], suggesting transcriptional regulation of HSP72 by AMPK. In addition, in vivo experiments showed higher HSP72 expression in intact muscle from AMPK-DN mice than from wild-type mice, and decreased HSP72 expression during unloading-induced muscle atrophy was attenuated in AMPK-DN mice [22]. Taken together, these data suggest an inverse relationship between AMPK and HSP72 in skeletal muscle and indicate that AMPK-mediated regulation of muscle mass may be mediated by HSP72.

The transcription of HSPs is mediated by HSF1 binding to the corresponding regulatory elements [82]. Hence, if AMPK affects the HSF1-mediated transcription of HSPs, several HSPs may be induced by AMPK activation. However, our experiments showed that activation of AMPK by AICAR treatment affected HSP72, but not HSP25 or constitutively expressed HSC70 [23]. Furthermore, HSF1 expression was downregulated following disuse-induced muscle atrophy independently of AMPK activity [22]. These observations indicate that AMPK modulates the expression of HSPs through HSF1-independent mechanisms.

Myostatin is known as growth differentiation factor 8 (GDF-8) and is a member of the transforming growth factor-β (TGFβ) superfamily, which has emerged as central to the regulation of skeletal muscle mass [90]. Myostatin binding to activin type IIB receptor (ActRIIB) leads to the recruitment of activin-like kinase-4 (ALK-4) or ALK-5, and phosphorylation and activation of Smad2 and Smad3, which form complexes with Smad4 [124, 128]. Subsequent nuclear translocation of the Smad2/Smad3/Smad4 complex promotes the transcription of genes involved in proliferation and differentiation and the protein metabolism in skeletal muscle [72, 134, 147]. Activation of Smad2 and Smad3 also impairs Akt activity, leading to inhibition of mTOR-dependent protein synthesis pathways [158] and stimulation of FoxOs-dependent protein degradation pathways [88]. Consequently, myostatin acts as a negative regulator of muscle mass.

Myostatin was shown to be associated with AMPK in skeletal muscle during diabetic muscle atrophy, with concomitant AMPK Thr¹⁷² phosphorylation and increased myostatin expression [55]. Accordingly, some studies have shown that AMPK activation by AICAR stimulates myostatin expression in skeletal muscle cells [19, 77]. These findings suggest that AMPK stimulates myostatin expression and thereby negatively regulates skeletal muscle mass. This hypothesis is supported by our unpublished findings showing that increases in myostatin mRNA expression in atrophic soleus muscles are abolished by suppression of AMPK. We also observed lower plasma myostatin levels in AMPK-DN mice than in wild-type mice (unpublished data). Hence, AMPK may regulate muscle mass by modulating skeletal muscle and/or whole-body myostatin expression. In addition, treatment of skeletal muscle cells with recombinant myostatin reportedly stimulated AMPK by depleting energy stores [14], suggesting that a feed-forward loop between AMPK and myostatin contributes to the control of muscle mass.