Unacylated Ghrelin Reduces Skeletal Muscle Reactive Oxygen Species Generation and Inflammation and Prevents High-Fat Diet–Induced Hyperglycemia and Whole-Body Insulin Resistance in Rodents Free

Clustered metabolic abnormalities, including excess reactive oxygen species (ROS) generation and inflammation activation, are proposed contributors to the onset of skeletal muscle insulin resistance (1–5). Excess muscle ROS production and inflammation are indeed linked at the level of inhibitor of κB (IκB)/nuclear factor-κB (NF-κB) activation and may cause insulin resistance by inhibiting insulin signaling downstream of insulin receptor (2,3,5). Ghrelin is a peptide hormone predominantly secreted by the stomach, and its acylated form (AG) is a major hypothalamic orexigenic signal (6,7). Sustained AG administration causes weight gain and hyperglycemia despite enhanced muscle mitochondrial oxidative capacity (8,9) by increasing food intake, hepatic gluconeogenesis, and fat deposition in rodents (10,11). A comprehensive understanding of the metabolic impact of ghrelin, however, has recently been allowed by reports of independent, more favorable effects of its unacylated form (UnAG). Although no specific UnAG receptor has yet been identified, UnAG counteracts glucogenic effects of AG as well as AG-induced hyperglycemia (10), and negative associations have been reported between circulating UnAG and markers of whole-body insulin resistance in humans (12,13). Of interest, emerging antioxidant effects have been reported for UnAG in various cell types (14–17), and we have demonstrated that UnAG stimulates autophagy in rodent muscle, thereby also potentially lowering muscle oxidative stress through disposal of damaged mitochondria (18). No information is available, however, on 1) the impact of UnAG on skeletal muscle ROS generation, inflammation, and insulin action and 2) whether UnAG prevents altered oxidative stress, inflammation, and insulin action in obesity and diabetes.

We therefore studied lean rats and a transgenic mouse model of systemic UnAG overproduction (19) to test the hypothesis that UnAG 1) lowers mitochondrial ROS production and inflammation and enhances insulin action in lean rodent muscle and 2) normalizes high-fat diet (HFD)–induced muscle metabolic alterations, whole-body insulin resistance, and hyperglycemia. In addition, effects of UnAG were verified in vitro in myotubes, where we also mechanistically tested the hypothesis that UnAG activities are at least partly mediated by positive modulation of autophagy.

Experiments were approved by the Animal Studies Committee at the University of Trieste (Trieste, Italy). Twenty 12-week-old male Wistar rats (Harlan-Italy, San Pietro-al-Natisone, Udine, Italy) were housed for 2 weeks in individual cages with a 12-h light/dark cycle at the University of Trieste Animal Facility, with ad libitum access to water and standard chow (Harlan 2018, 14.2 kJ/g). Animals were then randomly assigned to 4-day, twice-daily 200-μg subcutaneous injections of UnAG (n = 10; Bachem, Bubendorf, Switzerland) or vehicle (control [Ct]) (n = 10; NaCl 0.9% weight for volume). UnAG dose was based on previous studies in which equimolar AG modulated the same parameters (8). Body weight and food intake were monitored daily; after the last injection, food was removed for 3 h followed by anesthesia (tiobutabarbital 100 mg/kg, tiletamine/zolazepam [1:1] 40 mg/kg i.p.). Gastrocnemius and extensor digitorum longus muscles were then surgically isolated, and blood was collected by heart puncture.

Generation and characteristics of transgenic mice overexpressing UnAG (Tg Myh6/Ghrl) were previously described (19). Selective ghrelin overproduction in the heart, characterized by negligible acylating activity, results in a 40-fold increment in circulating UnAG without AG modification. Fourteen Tg Myh6/Ghrl and 14 matched wild-type male mice underwent 16-week standard or HFD feeding (10% or 60% calories from fat; Research Diets, New Brunswick, NJ) and were killed as described previously. Insulin tolerance tests (ITTs) were performed at 15 weeks of treatment by intraperitoneal insulin injection (Humulin R 3 nmol/kg; Eli Lilly, Indianapolis, IN) after a 4-h fast. Blood glucose was measured from tail blood (ACCU-CHEK Active; Roche, Basel, Switzerland) immediately before injection and at 20, 40, 60, and 80 min.

C2C12 myoblasts were differentiated in myotubes (20). After a 4-day incubation with differentiation medium and an 18-h starvation, cells were treated with AG or UnAG (0.1, 0.5, 1 μmol/L) for 48 h, collected, and processed. In additional experiments, the potential role of autophagy in effects of UnAG was investigated by genomic silencing of the autophagy mediator ATG5 (18). Small interfering RNA (siRNA) knockdown of ATG5 was performed by reverse transfection at final 25 nmol/L concentration with mouse ATG5 siRNA (M-064838-02-0005; Dharmacon) or with nontargeting control siRNA #4 (NT4) (D-001210-04-20; Dharmacon) using Lipofectamine RNAiMAX (Life Technologies). Twenty-four hours after transfection, culture medium was replaced, and after 36 h, it was differentiated, treated, and processed as aforementioned. ATG5 protein levels were quantified by Western blot.

Plasma insulin concentration was measured by ELISA (Ultrasensitive ELISA; DRG, Springfield, NJ). Plasma glucose and nonesterified fatty acid (NEFA) levels were determined by standard enzymatic colorimetric assays (21,22).

Mitochondrial H₂O₂ production was assessed in isolated intact mitochondria from tissues and cells by using the Amplex Red (10 μmol/L; Invitrogen, Carlsbad, CA) horseradish peroxidase method, modified as previously reported and normalized by citrate synthase activity in the same mitochondrial preparation (22,23). Assay substrate concentrations (mmol/L) were 8 glutamate, 4 malate (GM); 10 succinate (S); 4 glutamate, 2 malate, 10 succinate (GMS); and 0.05 palmitoyl-l-carnitine, 2 malate (PCM). Superoxide anion production sources in gastrocnemius muscle whole-tissue homogenate were assessed by using the lucigenin chemiluminescence method as previously described (22) and normalized by protein concentration (BCA Protein Assay Kit; Pierce, Rockford, IL). The impact of subsequent addition of specific inhibitors on specific substrate-stimulated production rates was used to evaluate relative superoxide production from each source (mitochondria: 5 μmol/L CCCP on succinate; nitric oxide synthase [NOS]: 10 mmol/L l-N^G-nitro-l-arginine methyl ester on 10 mmol/L l-arginine; NADPH oxidase: 200 μmol/L diphenylene iodonium on 1 mmol/L NADPH; xanthine oxidase: 200 μmol/L oxypurinol on 500 μmol/L xanthine) as referenced.

Total and oxidized glutathione were determined as referenced (24) on ∼50 mg gastrocnemius cleaned and homogenized in ice-cold 5% (weight for volume) metaphosphoric acid (20 mL/g tissue). Reduced glutathione (GSH) was calculated as total minus oxidized fraction (GSSG). Commercial kits were used to measure catalase (Amplex Red Catalase Assay Kit; Invitrogen) and glutathione peroxidase activities (Abcam, Cambridge, U.K.).

Cytokine profile and insulin signaling protein phosphorylation (p) at insulin receptor (IR)^Y1162/Y1163, IRS-1^S312, AKT^S473, GSK-3β^S9, PRAS40^T246, and P70S6K^T421/S424 levels were measured by xMAP technology (MAGPIX; Luminex Corporation, Austin, TX) using commercial kits, validated by the manufacturer for multiplexing profiling (LRC0002M, LHO0001M, LHO0002; Life Technologies, Carlsbad, CA). MILLIPLEX Analyst software (Millipore, Billerica, MA) was used for interpolating data to a standard curve. Phosphorylation of each protein is expressed as phosphoprotein units per total in picograms.

Western blots were performed as previously described (21,22,25). Equal loading was checked by Ponceau S staining and GAPDH reprobing. Primary antibody dilutions were anti-Mn superoxide dismutase (SOD) and anti-CuZnSOD (Stressgen, Ann Arbor, MI) 1:5,000 and 1:1,000, respectively; anti-IκB (Cell Signaling Technology, Beverly, MA) 1:500; anti-pIRS-1^Y612 (Abcam) 1:500; anti-ATG5 (Cell Signaling Technology) 1:2,000; anti-LC3B (Sigma) 1:1,500; anti-β-actin (Sigma) 1:25,000; and anti-GAPDH (Santa Cruz Biotechnology, Dallas, TX) 1:1,000.

NF-κB binding activity was assessed by nonradioactive electrophoretic mobility shift assay (22) with modifications. Equal amounts of nuclear protein were loaded for each sample. After incubation with poly(deoxyinosinic-deoxycytidylic) acid (0.05 μg/μL) and double-stranded 3′-biotinylated DNA probe, electrophoretic separation of nuclear extracts was performed in 0.8% agarose gel. Band specificity evaluation and identification were performed by running a pooled sample preincubated for 20 min with excess unlabeled probe (1,000×), anti-p65 (2 μg; Millipore), or anti-p105/p50 (2 μg; Abcam) antibody. Results were calculated from optical density of NF-κB–specific bands.

Tissue glucose uptake was measured ex vivo with nonradioactive 2-deoxyglucose (2-DG) (26). Extensor digitorum longus muscle is largely metabolically similar to gastrocnemius muscle (27) and was used due to smaller diameter and better exchange with incubation buffer (28). Two muscle sections were incubated for 30 min at 37°C under constant oxygenation with or without insulin (Humulin R 600 pmol/L) in isotonic buffer (pH 7.4) in BSA (1 mg/mL) and pyruvate (2 mmol/L). After 20-min incubation with pyruvate substituted with 2-DG (1 mmol/L), samples were snap frozen and kept at −80°C. After homogenization in ultrapure water followed by NaOH addition (0.07 N), enzymes and endogenous NAD(P)H and NAD(P) were inactivated by 45-min incubation at 85°C. Equinormal quantities of HCl were then added, and samples were cleared of debris by centrifugation (10,000g for 5 min) and transferred to 96-well microplates for incubation (37°C for 60 min) in assay buffer with β-NADP (0.1 mmol/L) and glucose-6-phosphate (G6P) dehydrogenase (20 units/mL) from Leuconostoc mesenteroides (buffer C) or with β-NAD (0.1 mmol/L) and G6P dehydrogenase (0.3 units/mL) (buffer D). Concentrations of G6P and G6P + 2-deoxyglucose-6-phosphate (2-DG6P) were quantified by fluorimetrically measuring (Infinite F200; Tecan, Männedorf, Switzerland) conversion of resazurin to resorufin in buffers C and D, respectively. Values related to 2-DG6P were calculated by subtraction, interpolated on a standard curve of 2-DG6P, and normalized by protein concentration in sample homogenate. Tissue 2-DG6P uptake was expressed in micromoles of 2-DG per milligram protein in 30 min.

The ATP synthesis rate in tissues and cells was measured ex vivo in freshly isolated mitochondria by using a luciferin-luciferase luminometric assay (22). Integrity of mitochondria isolated by gentle homogenization was tested by comparison of citrate synthase measurements in samples before and after membrane disruption (29). After signal stabilization and addition of excess substrates, a first 10-min kinetic read was performed followed by addition of 100 μmol/L ADP and a 20-min read. Final respiration substrates composition and reaction concentrations (mmol/L) were 0.25 pyruvate, 0.0125 palmitoyl-l-carnitine, 2.5 α-ketoglutarate, 0.25 malate (PPKM); 0.025 palmitoyl-l-carnitine, 0.5 malate (PCM); 20 succinate, 0.1 rotenone (SR); and 10 glutamate, 5 malate (GM). The impact of complex-related energy flux on ATP synthesis was calculated as the difference in production rate induced by the addition, in a subsequent 20-min read, of a complex-specific inhibitor during state 3 respiration on excess complex-specific substrate. For complex I–related ATP synthesis, substrate and inhibitor were GM and rotenone (2 μmol/L), whereas for complex II, these were SR and malonate (1 mmol/L). Mitochondrial functional integrity in each preparation was confirmed by a >80% and >95% decrease in state 3 ATP synthesis after addition of CCCP 30 μmol/L and oligomycin 2 μg/μL, respectively. Values were then normalized by ATP synthesis rate with the nonspecific substrate PPKM, and data are presented as the ratio between values obtained for complex I–related over complex II–related results.

Groups were compared by using Student t test or one-way ANOVA followed by appropriate post hoc tests. Bonferroni correction for multiple comparisons was applied. P < 0.05 was considered statistically significant.

In lean adult rats, exogenous 4-day UnAG did not modify body weight (Ct 319.6 ± 3.6 g, UnAG 324.1 ± 6.1 g), weight gain during treatment (Ct 13.0 ± 1.4 g, UnAG 11.6 ± 1.1 g), or caloric intake (Ct 76.9 ± 2.3 kcal/day, UnAG 73.5 ± 1.8 kcal/day). Plasma glucose (Ct 118.6 ± 6.0 mg/dL, UnAG 120.5 ± 7.5 mg/dL), insulin (Ct 12.8 ± 2.1 μU/mL, UnAG 14.3 ± 2.9 μU/mL), and NEFA (Ct 0.27 ± 0.06 mmol/L, UnAG, 0.21 ± 0.03 mmol/L) concentrations were comparable between groups.

UnAG lowered gastrocnemius H₂O₂ and superoxide anion production rate, and this effect involved mitochondrial respiration-dependent ROS generation (Fig. 1A–C). NOS-dependent but not xanthine- or NADPH oxidase–dependent superoxide production was also reduced by UnAG (Fig. 1D–F). UnAG-treated rats also had lower muscle oxidized-to-total glutathione, a marker of tissue redox state (Fig. 1G and H). Conversely, tissue protein levels of SOD isoforms and activities of antioxidant catalase and glutathione peroxidase were not modified by UnAG (Fig. 1I–L).

Protein expression of the NF-κB inhibitor IκB was higher in UnAG- compared with saline-treated rats, with parallel reduction of proinflammatory NF-κB p65/p50 nuclear binding activity (Fig. 2A and B). UnAG also increased p50/p50 homodimer binding activity (Fig. 2B), a transcription activator for anti-inflammatory interleukin-10 (IL-10) (30). UnAG treatment consistently resulted in anti-inflammatory changes in muscle cytokine patterns, with higher IL-10 expression and lower proinflammatory IL-1α and tumor necrosis factor-α (TNF-α) (Fig. 2C–G).

UnAG also led to insulin signaling activation with increased phosphorylation of AKT^S473, GSK-3β^S9, PRAS40^T246, and P70S6K^T421/S424 (Fig. 3A–F), consistent with the activation of the kinase activity of both mTORC (mammalian target of rapamycin complex) complexes. Changes in insulin signaling were paralleled by higher insulin-stimulated muscle glucose uptake (Fig. 3G). These effects were further associated with enhanced IRS-1^S312 phosphorylation (Fig. 3B), an mTORC-dependent negative feedback mechanism and marker for enhanced insulin signaling (31). To determine whether activating IRS-1 phosphorylations were also enhanced, we measured pIRS-1^Y612 and found no stimulation in UnAG-treated animals (Supplementary Fig. 1A), further indicating that UnAG-associated activation of insulin signaling occurs downstream of mTORC complexes but not at the IR–IRS-1 level.

In liver tissue, a nonstatistically significant reduction in mitochondrial superoxide production was observed. This relatively minor change was not associated with altered redox state, inflammation markers, or insulin signaling (Supplementary Figs. 2A–I and 3A–F), as previously shown (32,33).

Upregulation of circulating UnAG by myocardial overexpression of the ghrelin gene (Tg Myh6/Ghrl) (19) did not modify body weight (control diet: Ct 31.0 ± 2.1 g, Tg Myh6/Ghrl 28.7 ± 2.1 g; HFD: Ct 37.9 ± 3.0 g, Tg Myh6/Ghrl 36.6 ± 1.1 g) or caloric intake (control diet: Ct 13.5 ± 0.1 kcal/day, Tg Myh6/Ghrl 14.6 ± 0.6 kcal/day; HFD: Ct 17.6 ± 0.1 kcal/day, Tg Myh6/Ghrl 17.9 ± 0.5 kcal/day) under any dietary regimen (19). Blood glucose (control diet: Ct 106.0 ± 7.3 mg/dL, Tg Myh6/Ghrl 98.2 ± 8.0 mg/dL), plasma insulin (control diet: Ct 13.3 ± 1.8 μU/mL, Tg Myh6/Ghrl 14.5 ± 3.1 μU/mL), and NEFA (control diet: Ct 0.32 ± 0.06 mmol/L, Tg Myh6/Ghrl 0.38 ± 0.08 mmol/L) were also comparable among lean groups. In contrast, blood glucose (HFD: Ct 161.9 ± 30.7 mg/dL, Tg Myh6/Ghrl 102.7 ± 11.6 mg/dL, P < 0.05 Ct vs. Tg Myh6/Ghrl) and plasma insulin (HFD: Ct 25.1 ± 2.2 μU/mL, Tg Myh6/Ghrl 16.5 ± 1.6 μU/mL, P < 0.05 Ct vs. Tg Myh6/Ghrl) but not NEFA (HFD: Ct 0.27 ± 0.05 mmol/L, Tg Myh6/Ghrl 0.32 ± 0.10 mmol/L, P = not significant [NS]) were lower in HFD-obese Tg Myh6/Ghrl compared with both wild-type HFD animals and lean groups (P = NS, HFD Tg Myh6/Ghrl vs. lean groups).

Consistent with exogenous UnAG administration, circulating UnAG upregulation in Tg Myh6/Ghrl was characterized by lower muscle oxidized-to-total glutathione, lower proinflammatory tissue cytokine profile, and more pronounced phosphorylation of AKT^S473, GSK-3β^S9, PRAS40^T246, and P70S6K^T421/S424 (Fig. 4A–M). These effects were associated with higher insulin sensitivity by area under the curve for ITT-induced blood glucose changes (Fig. 4N–P). Obese wild-type animals were, as expected, hyperglycemic and insulin resistant (Fig. 4N–P). The obese wild-type group also had higher oxidized-to-total glutathione, higher proinflammatory cytokine profile, and reduced phosphorylation of AKT^S473 and GSK-3β^S9 in gastrocnemius (Fig. 4A–M). Compared with lean Tg Myh6/Ghrl, obese Tg Myh6/Ghrl animals had moderately higher muscle oxidized-to-total glutathione and TNF-α, which remained, however, lower (P < 0.05) than in obese and comparable (P = NS) to lean wild-type animals (Fig. 4B and E). In addition, UnAG upregulation prevented obesity-associated increments (P < 0.05 vs. obese wild type) in muscle proinflammatory cytokines IL-1α and IL-1β with lower IL-6, and resulted in normalized activating phosphorylation at AKT^S473 and GSK-3β^S9 levels (P = NS vs. lean Tg Myh6/Ghrl). Of note, obese Tg Myh6/Ghrl were protected from obesity-induced hyperglycemia and whole-body insulin resistance (Fig. 4N–P), with both parameters superimposable to lean wild-type animals. Insulin signaling proteins upstream of AKT were not activated in Tg Myh6/Ghrl, with patterns of pIRS-1^S312 and pIRS-1^Y612 comparable to those observed in exogenously treated animals (Fig. 4H and I and Supplementary Fig. 1B).

Forty-eight–hour UnAG treatment of C2C12 myotubes lowered mitochondrial ROS generation, with largely dose-dependent effects (Fig. 5A). Consistent with in vivo data, UnAG treatment also resulted in increased activating phosphorylation of mTORC complexes–dependent insulin signaling proteins AKT^S473, GSK-3β^S9, PRAS40^T246, and P70S6K^T421/S424. Patterns of pIR^Y1162/Y1163 and pIRS-1^S312 were also comparable in C2C12 and in vivo experiments, supporting the lack of activation of IR–IRS-1 (Fig. 5B–G).

C2C12 myotubes do not appear to express the AG receptor growth hormone secretagogue receptor 1 (GHSR1) (34). To further exclude the possibility that UnAG-induced changes result from nonspecific activation of additional AG-regulated pathways, C2C12 experiments were performed with equimolar AG concentrations. Forty-eight–hour AG incubation failed to inhibit ROS production and to activate insulin signaling, except for less pronounced enhancement of pGSK-3β^S9 (Fig. 5A–G).

In additional experiments, C2C12 myotubes were incubated with UnAG after genomic silencing of the autophagy mediator ATG5. ATG5 silencing abolished UnAG activities on both mitochondrial ROS production and insulin signaling (Fig. 6A–H). Levels of the autophagy activation marker LC3II/LC3I were also higher in HFD-obese Tg Myh6/Ghrl than wild-type mice (Fig. 6I).

Consistent with previous results (22,35), UnAG-induced changes in redox state, inflammation, and insulin signaling were associated not with an enhanced but, rather, with a lower or an unchanged ATP production rate in vivo and in vitro, respectively (Fig. 7A–C). Higher skeletal muscle ATP production was observed in obese mice compared with lean counterparts, but UnAG upregulation was also associated with lower ATP production rates in obese animals (Fig. 7B). UnAG modified muscle respiratory chain complex–related ATP production by shifting ATP synthesis toward complex I over complex II both in vivo and in vitro (Fig. 7D and E). Differently from UnAG, AG enhanced ATP production in C2C12 myotubes (Fig. 7C). Liver ATP production was not modified by UnAG (Supplementary Fig. 3G).

These studies led to several findings. First, sustained UnAG administration in vivo leads to 1) lower muscle ROS production and less oxidized tissue redox state, 2) anti-inflammatory changes in tissue NF-κB activation and cytokine patterns, and 3) enhanced mTORC-dependent insulin signaling with higher insulin-stimulated muscle glucose uptake. Second, muscle effects of UnAG are reproduced in a model of systemic circulating UnAG upregulation with HFD-induced obesity, resulting in prevention of obesity-associated hyperglycemia and whole-body insulin resistance. Third, UnAG effects are dose-dependently confirmed in myotubes; differential effects of AG and UnAG are observed in vitro, indicating that UnAG acts at least partly directly and independently of AG-regulated pathways. Finally, UnAG effects in vitro are abolished by autophagy inhibition, thereby indicating mechanistic involvement of autophagy in UnAG activities.

The results show that UnAG negatively regulates skeletal muscle ROS production and inflammation, and these effects are indirectly supported by previous in vitro observations in nonmuscle cells (14,16,19). In another study, UnAG reduced endothelial oxidative stress in models of peripheral artery disease by restoring SOD expression (15,16). Skeletal muscle SOD expression and antioxidant enzyme activities were, however, unchanged by UnAG in the current model, indicating lower mitochondrial ROS generation rather than enhanced antioxidant defenses as a key mediator of UnAG-induced muscle antioxidant activity. Since recently identifying UnAG as a potent inducer of autophagy in cardiomyocytes and myotubes (35), we speculate that enhanced removal of dysfunctional mitochondria could have contributed to lower tissue oxidative load in the current experimental setting. This hypothesis was confirmed in myotube experiments using siRNA-mediated autophagy inhibition. Among less quantitatively relevant ROS sources (36), UnAG selectively inhibited NOS-dependent superoxide production. This finding is consistent with emerging colocalization and functional interactions among NOS, nitric oxide (NO), and muscle mitochondria (37,38). In particular, NO production has been reported to enhance mitochondrial ROS generation (37), whereas UnAG was reported to reduce NO production induced by proinflammatory cytokines in various settings (39). Potential interactions among UnAG, NO, and mitochondrial ROS generation should be directly investigated in future studies.

Sustained UnAG administration enhanced skeletal muscle insulin signaling downstream of mTORC complexes while not at the IR–IRS-1 level, and these effects were paralleled by increased insulin-stimulated muscle glucose uptake. These changes agree with and provide a molecular basis for clinical observations linking UnAG with preserved whole-body insulin sensitivity in humans (12,13). Of note, autophagy inhibition in vitro abolished UnAG activities on both mitochondrial ROS production and insulin signaling. These observations provide further strong support for a causal negative impact of mitochondrial ROS production on AKT-dependent insulin signaling, which agrees with previous observations (1–5). UnAG effects were associated with enhanced inhibitory IRS-1^S312 phosphorylation; this seemingly paradoxical observation is, however, consistent with reports of IRS-1^S312 phosphorylation as a physiological negative feedback modulation following downstream signaling activation (31).

Results in Tg Myh6/Ghrl mice with chronic systemic UnAG overexposure (19) confirmed effects of exogenous UnAG administration, and these results are supported by higher insulin sensitivity in a lean UnAG adipose transgenic model (17). Because plasma AG and IGF-I are unchanged in Tg Myh6/Ghrl (19), the current findings further confirm that UnAG effects are independent of changes in AG and its potential impact on growth hormone-IGF-I through GHSR1 (7,19). Most importantly, circulating UnAG upregulation prevented HFD-induced hyperglycemia and systemic insulin resistance, whereas muscle oxidative stress markers, inflammation, and impaired insulin signaling were preserved overall at levels comparable with lean Tg Myh6/Ghrl or wild-type animals. UnAG-dependent stimulation of muscle autophagy was also confirmed in vivo in HFD-fed obese Tg Myh6/Ghrl by a higher LC3II/LC3I ratio and, therefore, could have potentially directly contributed to beneficial effects of UnAG overexpression (40). Of note, obese Tg Myh6/Ghrl animals showed no increments in pIRS-1^S312 compared with wild-type counterparts, and lack of effect was associated with lack of stimulation of insulin signaling activation at P70S6K levels. These combined observations are consistent with the hypothesis that IRS-1^S312 phosphorylation is at least partly mediated by this feedback loop (31). Potential mechanisms underlying differential regulation of PRAS40^T246 and P70S6K^T421/S424 phosphorylation in obese versus lean models of UnAG exposure should be investigated in future studies. Overall, results in the HFD-obesity model demonstrated that effects of UnAG translate into beneficial metabolic changes in a clinically relevant model of dietary-induced insulin resistance and hyperglycemia, thereby providing a strong rationale for therapeutic strategies to increase UnAG availability in obese, insulin resistant, and type 2 diabetic conditions.

Myotube experiments agreed with in vivo studies by showing superimposable effects of UnAG on ROS production and insulin signaling that were not induced by equimolar AG concentrations. These observations strongly indicate that UnAG directly stimulates skeletal muscle insulin signaling and are consistent with previously reported UnAG signaling and antiatrophic activities in skeletal muscle of both wild-type and GHSR1 knockout mice (19). These findings overall provide strong support to the hypothesis that UnAG effects in skeletal muscle are independent of GHSR1 and are mediated by alternative, yet unidentified, UnAG receptors. Both AG and UnAG also stimulate differentiation of C2C12 myoblasts (34), and both ghrelin forms enhance mTORC2-mediated antiatrophic signaling under acute experimental conditions in C2C12 myotubes as well as in vivo in skeletal muscle of GHSR knockout mice (19,32). In the current studies with prolonged hormone incubation, highest AG doses selectively induced a moderate increase of GSK-3β^S9 phosphorylation but failed to reduce ROS generation and to enhance downstream insulin signaling. Also consistent with these findings, AG is a weaker autophagy inducer than UnAG and fails to stimulate both mitophagy (35) and ischemia-induced skeletal muscle regeneration (15). On the basis of the available knowledge, differential muscle effects of ghrelin forms may depend on still uninvestigated acylation-selective and time-dependent AG activities. Overall, differential effects of ghrelin forms on muscle insulin signaling are fully consistent with clinical observations linking UnAG, but not AG, to whole-body insulin sensitivity in humans (12,13).

Finally, UnAG-induced lower ROS production, lower inflammation, and enhanced insulin signaling were associated with reduced or unchanged ATP production. In agreement with previous studies, HFD-fed animals conversely showed higher mitochondrial ATP production despite higher oxidative stress markers and insulin resistance (32), and this alteration could involve enhanced substrate availability through feed-forward mechanisms (32). The current results, therefore, provide further evidence against a role for low mitochondrial function to primarily cause insulin resistance (41–45), conversely indicating UnAG as a novel modulator of muscle mitochondrial activity with a negative impact on both ATP and ROS production in vivo. Unchanged mitochondrial ATP production in vitro, however, does not support a direct role of UnAG to inhibit mitochondrial function, while it further indicates that reduced mitochondrial function is not a prerequisite for reduced ROS generation. Of note, UnAG modified complex-related ATP production by favoring complex I– over complex II–related synthesis in vitro and in vivo, potentially reflecting preferential glucose- over fat-derived substrate oxidation (46). Because glucose-related substrate oxidation may lower mitochondrial ROS generation (47,48), this mechanism could also contribute to inhibiting ROS production. The current results warrant further studies on interactions between UnAG and muscle mitochondrial function.

In conclusion, these studies demonstrate a novel role of UnAG to modulate skeletal muscle redox state, inflammation, and insulin signaling. UnAG-treated rat muscle is characterized by lower mitochondrial ROS production, lower inflammation, and enhanced insulin signaling and action. These effects are tissue specific, appear to be direct and independent of acylated hormone, and could be at least partly mediated by UnAG-dependent stimulation of autophagy (Fig. 8). UnAG overexpression also prevents obesity-associated hyperglycemia and systemic insulin resistance as well as muscle oxidative stress, inflammation activation, and impaired insulin signaling. The current findings collectively indicate UnAG as a potential novel treatment for obesity-associated metabolic alterations.

Acknowledgments. The authors thank Marco Stebel and Davide Barbetta (University of Trieste Animal Facility, Trieste, Italy) and Margherita De Nardo and Lorenza Mamolo (Department of Medicine, Surgery and Health Sciences, University of Trieste) for assistance with the in vivo procedures. Manuela Boschelle and Chiara Matilde Boccato (Department of Medicine, Surgery and Health Sciences, University of Trieste) are acknowledged for technical assistance in ex vivo and in vitro experiments.

Funding. This work was funded by the European Society for Clinical Nutrition and Metabolism (ESPEN) through a research fellowship to G.G.C.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. G.G.C. contributed to the study design, experiments, data research and analysis, and writing and final approval of the manuscript. M.Z. contributed to the discussion and review, editing, and final approval of the manuscript. A.S. contributed to the experiments, data analysis, discussion, and final approval of the manuscript. P.V. and G.G. contributed to the discussion and final approval of the manuscript. G.R. and A.F. contributed to the experiments, discussion, and final approval of the manuscript. N.F. generated the transgenic mice and contributed to the discussion and final approval of the manuscript. A.G. generated the transgenic mice and contributed to the discussion and review, editing, and final approval of the manuscript. M.G. contributed to discussion and review, editing, and final approval of the manuscript. R.B. contributed to the study design, discussion, and writing and final approval of the manuscript. R.B. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.