Enhanced Muscle Insulin Sensitivity After Contraction/Exercise Is Mediated by AMPK

Skeletal muscle from human, sheep, dog, and rodents demonstrates increased insulin-stimulated glucose uptake in the period after a single bout of exercise (1–10). This phenomenon is observed in both healthy and insulin-resistant muscle (11–13) and has been suggested to involve an increased abundance of GLUT4 at the plasma membrane (14). Moreover, changes in muscle insulin sensitivity occurs independent of changes in protein synthesis (15), indicating involvement of posttranslational mechanisms. Interestingly, studies of human and rodent muscle suggest that prior exercise does not improve the ability of insulin to stimulate components of the proximal insulin signaling cascade, including the insulin receptor, insulin receptor substrate 1, PI3K, and Akt (5,15–18). This supports the notion that improved insulin sensitivity after exercise is not caused by enhanced delivery of insulin to the muscle and indicates an important role for more distal intramyocellular signaling events.

AMPK is a heterotrimeric complex containing catalytic α and regulatory β and γ subunits, of which several isoforms exist (α1, α2, β1, β2, γ1, γ2, and γ3) (19). In human and mouse skeletal muscle, three (α2β2γ1, α2β2γ3, and α1β2γ1) and five (α2β2γ1, α2β2γ3, α2β1γ1, α1β2γ1, and α1β1γ1) heterotrimeric combinations have been found, respectively (20,21). Interestingly, mouse skeletal muscle contains two β1-associated complexes that are not found in human skeletal muscle. Furthermore, in mouse extensor digitorum longus (EDL) and human vastus lateralis muscle, the α2β2γ3 complex represents ∼20% of all AMPK heterotrimer complexes, whereas in mouse soleus (SOL) muscle, it comprises <2% (20,21). AMPK is considered an important sensor of cellular energy balance, and in skeletal muscle, AMPK is activated during conditions of cellular stress, such as muscle contraction and hypoxia (22). When activated, AMPK stimulates ATP-generating processes (e.g., glucose uptake and lipid oxidation) while inhibiting ATP-consuming processes (e.g., protein and lipid synthesis) in an attempt to restore cellular energy homeostasis (22,23).

TBC1D4 is phosphorylated by multiple kinases (including Akt) during insulin stimulation (24,25). This modification has been suggested to be important for insulin-stimulated glucose uptake (26). AMPK is also upstream of TBC1D4, and both contraction- and AICAR–induced AMPK activation increase phosphorylation of TBC1D4 (25). Within recent years, TBC1D4 has emerged as a likely candidate for mediating the insulin-sensitizing effect of prior exercise on skeletal muscle glucose uptake. In support of this, phosphorylation of TBC1D4 is elevated in prior exercised human and rat muscle, concomitant with enhanced insulin sensitivity (11,17,18,27–29).

Prior AICAR stimulation increases insulin sensitivity to stimulate glucose transport in rat muscle (15), and we have recently provided evidence that this is mediated by AMPK in muscle of mice (30). We also reported a positive association between insulin-stimulated glucose uptake and phosphorylation of regulatory sites on TBC1D4 (30). This suggests a mechanism by which AICAR, through AMPK, potentiates a subsequent effect of a submaximal concentration of insulin on TBC1D4, leading to improved insulin-stimulated glucose uptake.

During AICAR stimulation, cells maintain energy and fuel homeostasis. In contrast, the myocyte is subjected to energy and fuel disturbances during exercise/contraction, which likely contributes to AMPK activation. Furthermore, while AMPK regulates muscle glucose uptake, fatty acid uptake, gene activation, and mitochondrial protein content in response to AICAR treatment (31–34), activation of AMPK is not necessary for inducing such effects in response to exercise/contraction (31–36). Hence, little evidence exists to support the assumption that AICAR- and exercise/contraction-induced biological responses are equally dependent on AMPK activation.

Since the first proposal of an insulin-sensitizing effect of prior exercise by Bergström and Hultman (37) and the subsequent proof of this in rat and human skeletal muscle (1,2), an ongoing search for molecular interactions between exercise and insulin signaling has occurred. To further study this, we established an experimental protocol in which mouse muscle displays enhanced insulin sensitivity to stimulate glucose uptake after in situ contraction. We used this model to provide genetic evidence for the hypothesis that AMPK acts as a molecular transducer between exercise and insulin signaling and, thus, is necessary for the ability of prior contraction/exercise to increase muscle insulin sensitivity.

All experiments were approved by the Danish Animal Experiments Inspectorate (2014-15-2934-01037 and 2013-15-2934-00911), as well as the regional ethics committee of Northern Stockholm, and complied with the EU convention for protection of vertebra animals used for scientific purposes (Council of Europe, Treaty 123/170, Strasbourg, France, 1985/1998). Animals used in this study were AMPKα1α2 muscle-specific double-knockout (mdKO) and whole-body AMPKγ3 KO female mice with corresponding wild-type (WT) littermates as controls (35,36,38). Animals (16 ± 5 weeks [means ± SD]) were maintained on a 12:12 light-dark cycle (6:00 a.m. to 6:00 p.m.) with unlimited access to standard rodent chow and water.

For all experiments, fed mice were anesthetized by an intraperitoneal injection of pentobarbital (10 mg/100 g body weight) before both common peroneal or tibial nerves were exposed. Subsequently, an electrode was placed on a single common peroneal or tibial nerve followed by in situ contraction of EDL or SOL muscle, respectively. The contralateral leg served as a rested control. The contraction protocol consisted of 0.5-s trains (100 Hz, 0.1 ms, 2–5 V) repeated every 1.5 s for 10 min. To determine glucose uptake during in situ contraction, tail blood was collected at time points 0, 5, and 10 min. After the first blood sample, a bolus of [³H]2-deoxyglucose (12.3 MBq/kg body weight) dissolved in isotonic saline was injected retroorbitally. After the last blood sample, EDL or SOL muscles were rapidly dissected and frozen in liquid nitrogen. Uptake of [³H]2-deoxyglucose into muscle was assessed based on accumulated [³H]2-deoxyglucose-6-phosphate and tracer-specific activity in plasma as previously described (39).

For measurements of insulin sensitivity after in situ contraction, electrodes were connected to either the common peroneal nerve (EDL) or the tibial nerve (SOL) of both legs of the anesthetized animals. One-half of the animals served as sham-operated controls. Immediately after in situ contraction of EDL or SOL, muscles were dissected and suspended at low tension (∼1 mN) in incubation chambers (model 610/820M; Danish Myo Technology, Aarhus, Denmark) containing Krebs-Ringer buffer (KRB) (117 mmol/L NaCl, 4.7 mmol/L KCl, 2.5 mmol/L CaCl₂, 1.2 mmol/L KH₂PO₄, 1.2 mmol/L MgSO₄, 0.5 mmol/L NaHCO₃ [pH 7.4]) supplemented with 0.1% BSA, 5 mmol/L mannitol, and 5 mmol/L d-glucose. During the entire incubation period, the buffer was oxygenated with 95% O₂ and 5% CO₂ and maintained at 30°C. SOL and EDL muscles were allowed to recover for 2 and 3 h, respectively. These time points were selected based on measurements demonstrating reversal of muscle glucose uptake after in situ contraction. During recovery, the incubation medium was replaced once every 30 min to maintain an adequate glucose concentration. Subsequently, basal, submaximal (100 μU/mL/694.5 pmol/L), and maximal (10,000 μU/mL/69,450 pmol/L; only EDL) insulin-stimulated 2-deoxyglucose uptake was measured during the last 10 min of a 30-min stimulation period by adding 1 mmol/L [³H]2-deoxyglucose (0.028 MBq/mL), 7 mmol/L [¹⁴C]mannitol (0.0083 MBq/mL), and 2 mmol/L pyruvate to a glucose-free incubation medium. 2-Deoxyglucose uptake was assessed by the accumulation of [³H]2-deoxyglucose into muscle with the use of [¹⁴C]mannitol as an extracellular marker (30). Radioactivity was measured on 200 μL lysate by liquid scintillation counting (Ultima Gold and Tri-Carb 2910 TR; PerkinElmer) and related to the specific activity of the incubation media.

All mice were acclimatized to treadmill running on five consecutive days. The acclimatization consisted of a 2-min warm up (0–10.2 m/min) followed by 5 min of running at 10.2 m/min and 0° incline. Two days after the acclimatization, mice were subjected to a graded maximal running test, as previously described (36). For insulin tolerance tests (ITTs), mice were fasted in single cages for 2 h (∼8:00–10:00 a.m.) before performing a single bout of treadmill exercise (30 min, 15° incline, and 55% of maximal running speed). Resting control mice were left in the cage. After exercise, mice were returned to their individual cage without access to food for 1 h, after which they were administered with either 0.3 units (2.09 nmol) or 0.4 units (2.78 nmol) of insulin per kilogram body weight intraperitoneally, respectively. Throughout the ITT, blood was collected from the tail vein at 0, 20, 40, 60, 90, and 120 min, and blood glucose concentration was determined using a glucometer (Contour XT; Bayer, Leverkusen, Germany). Area over the curve values were calculated from time points 0–40 min, since changes in blood glucose concentrations at later time points may largely reflect the ability to counteract hypoglycemia rather than peripheral glucose disposal. Three to four weeks after the last ITT, in vivo muscle glucose uptake during the first 40 min of an ITT (0.3 units/kg insulin) was measured in the same mice 1 h after treadmill exercise. All mice received an intraperitoneal injection of insulin dissolved in isotonic saline (10 µL/g body weight) containing 0.1 mmol/L [³H]2-deoxyglucose (1.78 MBq) 1 h after rest and exercise. Immediately before, as well as 20 and 40 min into the ITT, blood glucose concentration was measured from the tail vein and blood samples were obtained for determination of radioactivity. After the last blood sample, mice were euthanized by cervical dislocation and tissues were rapidly dissected and frozen in liquid nitrogen. Uptake of [³H]2-deoxyglucose into muscle was assessed as previously described (39).

Whole-body AMPKγ3 KO mice were anesthetized and EDL muscles were isolated and preincubated in KRB (40 min) before they were stimulated to contract (10 min), as previously described (40).

Muscles were homogenized in 400 µL ice-cold homogenization buffer (30) and rotated end-over-end for 1 h at 4°C. Part of the homogenate was centrifuged at 16,000g for 20 min at 4°C, after which lysate (supernatant) was collected and frozen in liquid nitrogen for later analyses. Total protein abundance in muscle lysate and homogenate was determined by the bicinchoninic acid method (Thermo Fisher Scientific, Waltham, MA).

Muscle glycogen synthase (GS) activity was measured in 75 µg muscle homogenate using 96-well microtitre plates as previously described (11,41). Samples were assayed in triplicate in the presence of 0.02 and 8.0 mmol/L glucose-6-phosphate and presented as percent glucose-6-phosphate independent activity (GS_0.02 * 100/GS_8.0; %I-form) and total GS activity (GS_8.0; total), respectively.

Heterotrimer-specific AMPK activity in mouse skeletal muscle was determined as previously described (30). AMPK activity was measured on 300 µg of muscle lysate protein using AMPK-γ3, -α2, and -α1 antibodies for three consecutive immunoprecipitations.

Muscle glycogen content was measured on 200 µg of muscle protein homogenate after acid hydrolysis, as previously described (36).

Muscle lysates and homogenates were boiled in Laemmli buffer for 10 min before being subjected to SDS-PAGE and immunoblotting as previously described (30). Quantification of protein phosphorylation has not been related to protein abundance since abundance of all measured proteins did not change in response to any specified intervention. Small differences in total abundance were observed for some proteins between genotypes; however, this did not affect phosphorylation dynamics or interpretation of data.

Antibodies against phospho-AMPK-Thr¹⁷², phospho-ACCα/β-Ser^79/212, Akt2, phospho-Akt-Ser⁴⁷³, phospho-Akt-Thr³⁰⁸, phospho-TBC1D1-Thr⁵⁹⁰, phospho-TBC1D4-(Ser³¹⁸, Ser⁵⁸⁸, Thr⁶⁴²), phospho-ERK1/2-Thr²⁰²/Tyr²⁰⁴, and hexokinase II (HKII) were purchased from Cell Signaling Technology (Danvers, MA). Antibodies against phospho-TBC1D1-Ser²³¹ and AS160 (TBC1D4) were from Millipore (Temecula, CA), and antibodies against AMPKα2 and GLUT4 were purchased from Santa Cruz Biotechnology (Dallas, TX) and Thermo Fisher Scientific, respectively. ACC protein was determined using horseradish peroxidase–conjugated streptavidin from Dako (Glostrup, Denmark). TBC1D1, pyruvate dehydrogenase (PDH), AMPKα1, and GS protein, as well as phosphorylation of TBC1D4-Ser⁷¹¹, GS site 2+2a and site 3a+3b, were determined using specific antibodies as previously described (11,36,41). Antibodies used for AMPK activity measurements were against AMPKα2 (Santa Cruz Biotechnology), AMPKγ3 (provided by D.G. Hardie, University of Dundee, Scotland, U.K.), and AMPKα1 (purchased from GenScript Jiangning, Nanjing, China).

Statistical analyses were performed using SigmaPlot (version 13.0; Systat, Erkrath, Germany). Two-way ANOVA with or without repeated measures and paired/unpaired Student t tests were used to assess statistical differences within and between genotypes, where appropriate. A three-way ANOVA was used to assess differences in total muscle protein abundance between genotypes. The Student-Newman-Keuls test was used for post hoc testing, and all main effects have been indicated by lines. Correlation analyses were performed by calculating Pearson product moment correlation coefficient. Data are expressed as the means ± SEM unless stated otherwise. Differences were considered statistically significant at P < 0.05.

During in situ contraction, glucose uptake in EDL and SOL muscle increased similarly in AMPK WT and mdKO mice (Fig. 1A). Furthermore, in situ contraction decreased muscle glycogen content (Fig. 1B) and increased Erk1/2 Thr²⁰²/Tyr²⁰⁴ phosphorylation (Fig. 1C) to an extent that did not differ between genotypes. This suggests that the electrical stimulation protocol induced similar changes in WT and mdKO muscle. In situ contraction markedly increased phosphorylation of AMPK Thr¹⁷² (Fig. 1D) and downstream targets ACC Ser²¹² (Fig. 1E), TBC1D1 Ser²³¹ (Fig. 1F), and TBC1D4 Ser⁷¹¹ (Fig. 1G) in EDL and SOL muscle from WT mice, whereas only minor, if any, changes were seen in EDL and SOL muscle from mdKO mice. Contraction did not alter total protein abundance of Erk1/2, AMPKα1, AMPKα2, ACC, TBC1D1, and TBC1D4 in either EDL or SOL muscle (Fig. 1H). As expected, EDL and SOL muscle from AMPK mdKO mice showed a substantial loss of AMPKα1 and AMPKα2 protein abundance. Identical to previous observations (36), ACC and TBC1D1 protein abundance was decreased in AMPK mdKO skeletal muscle compared with WT littermates. Intriguingly, protein abundance of Erk1/2 was increased (∼35%, P < 0.01) whereas TBC1D4 protein abundance was decreased (∼20%, P < 0.05) in SOL muscle from mdKO mice compared with WT mice (Fig. 1H).

To test whether the effect of muscle contraction on insulin sensitivity is dependent on AMPK, we measured submaximal insulin-stimulated glucose uptake ex vivo after in situ contraction. Three hours after in situ contraction, “basal” glucose uptake was not significantly different between prior contracted and rested EDL muscle (Fig. 2A). However, prior in situ contraction increased submaximal insulin-stimulated glucose uptake in isolated EDL muscle from WT mice but failed to do so in EDL muscle from AMPK mdKO mice (Fig. 2A). Maximal insulin-stimulated glucose uptake was similar between prior contracted and rested EDL muscle in both genotypes (Fig. 2A). The incremental increase in submaximal insulin-stimulated glucose uptake (delta insulin: submaximal insulin-stimulated glucose uptake minus basal glucose uptake) was significantly higher after prior in situ contraction in WT mice only (Fig. 2B). Interestingly, prior in situ contraction did not increase submaximal insulin-stimulated glucose uptake ex vivo in WT SOL muscle (Fig. 2C and D). On the basis of these results, we performed the subsequent in situ experiments in EDL muscle from WT and mdKO mice with SOL muscle from WT mice as a negative control.

To evaluate the involvement of skeletal muscle AMPK in regulating whole-body insulin sensitivity after an acute exercise bout in vivo, we performed intraperitoneal ITTs on AMPK WT and mdKO mice at rest and 1 h after a single bout of acute treadmill exercise. Prior exercise enhanced the blood glucose–lowering response to a submaximal concentration of insulin (0.3 units/kg body weight) (Fig. 2E) and improved insulin tolerance by ∼250% in WT mice (Fig. 2G). In contrast, prior exercise did not induce a greater insulin response to lower blood glucose concentrations in AMPK mdKO mice (Fig. 2F and G). Furthermore, insulin-stimulated glucose uptake during the first 40 min of an ITT (0.3 units/kg body weight) was significantly improved 1 h after exercise in m. tibialis anterior from WT mice but not in muscle from AMPK mdKO mice (Fig. 2E and F).

Prior contraction and submaximal insulin did not alter total protein abundance of Akt2, HKII, GS, ACC, PDH, TBC1D1, TBC1D4, and GLUT4 in EDL muscle within either genotype (Fig. 3A). Total protein abundance of ACC, TBC1D1, HKII, and PDH was significantly lower (∼20–25%; n = 12–13, P < 0.05–0.001) in EDL muscle from AMPK mdKO mice compared with WT mice. In contrast, Akt2 muscle protein level was ∼33% higher (n = 12–13, P < 0.001) in EDL muscle from mdKO mice compared with WT mice. No differences in GS, TBC1D4, and GLUT4 muscle protein level were observed between genotypes. Like in EDL muscle, prior contraction and submaximal insulin did not alter total protein abundance of Akt2, TBC1D1, TBC1D4, ACC, and AMPKα2 in WT SOL muscle (Fig. 3B)

Previous studies of human, sheep, rat, and mouse investigating muscle insulin sensitivity in the postexercise state suggest that the increased ability of insulin to stimulate glucose uptake occurs independent of enhanced proximal insulin signaling (IR, IRS1, PI3K, and Akt) (5–7,15–18). In accordance, in the current study, prior contraction also did not affect submaximal insulin-induced phosphorylation of Akt Thr³⁰⁸ and Ser⁴⁷³ in EDL (Fig. 4A and B) or SOL muscle (Fig. 4C and D).

TBC1D4 is phosphorylated by Akt and AMPK (24–26). TBC1D4 has been suggested to regulate muscle insulin sensitivity, as insulin-stimulated phosphorylation of TBC1D4 is enhanced during conditions in which muscle displays increased insulin sensitivity after exercise and AICAR treatment (27–30). In the current study, prior contraction of EDL muscle increased the effect of submaximal insulin stimulation on TBC1D4 Thr⁶⁴⁹ and Ser⁷¹¹ phosphorylation compared with rested control muscle from WT mice (Fig. 5A and B). Interestingly, this effect was dependent on AMPK since insulin-induced phosphorylation of TBC1D4 was similar between rested and prior contracted EDL muscle from AMPK mdKO mice. Submaximal insulin-stimulated phosphorylation of TBC1D4 Ser³²⁴ and Ser⁵⁹⁵ was unaffected by prior muscle contraction of EDL muscle (Fig. 5C and D), suggesting a highly selective interaction between these stimuli. Correlation analyses revealed that delta insulin for glucose uptake and delta insulin for phosphorylation of TBC1D4 Thr⁶⁴⁹ and Ser⁷¹¹ in EDL muscle were positively correlated in WT mice (r = 0.43–0.65, P < 0.05–0.001) (Fig. 5E–G). Interestingly, prior contraction did not affect insulin-stimulated phosphorylation of TBC1D4 Thr⁶⁴⁹ and Ser⁷¹¹ in WT SOL muscle (Fig. 5H and I) in parallel with unchanged insulin sensitivity.

As muscle contraction acutely increases AMPK activity, we investigated whether this effect was maintained in EDL muscle recovered for 3 h ex vivo. In muscle from AMPK mdKO mice, phosphorylation of AMPK Thr¹⁷² and downstream target ACC Ser²¹² was reduced by ∼80–90% compared with WT littermates (Fig. 6A and B). Phosphorylation of AMPK Thr¹⁷² and ACC Ser²¹² had returned to resting levels 3 h after contraction. Submaximal insulin did not affect phosphorylation of AMPK Thr¹⁷² but induced a minor increase in ACC Ser²¹² phosphorylation in EDL muscle from AMPK mdKO mice (Fig. 6A and B). When measuring AMPK heterotrimer complex activity 3 h into recovery, we found that AMPKγ3-associated activity was increased in prior contracted WT EDL muscle (P < 0.01) (Fig. 6C), whereas no significant differences for the remaining AMPKα2- and AMPKα1-associated activities were found between rested and prior contracted muscle (Fig. 6D and E). Phosphorylation of AMPK Thr¹⁷² and ACC Ser²¹² was similar between prior rested and contracted WT SOL muscle (Fig. 6F and G). Also, no significant differences in AMPK activity were observed between prior contracted and rested WT SOL muscle (Fig. 6H). Taken together, this demonstrates that AMPKγ3-associated activity is increased concomitant with enhanced muscle insulin sensitivity. To elucidate a possible role of AMPKγ3 to enhance muscle insulin sensitivity after contraction, we investigated phosphorylation of TBC1D4 Ser⁷¹¹ in EDL muscle from AMPKγ3 KO mice. Interestingly, ex vivo contraction increased phosphorylation of TBC1D4 Ser⁷¹¹ in WT muscle but not in muscle from AMPKγ3 KO mice (Fig. 6I).

Phosphorylation of TBC1D1 has been proposed to regulate muscle glucose uptake in response to insulin and contraction (42,43). In the current study, submaximal insulin stimulation increased phosphorylation of TBC1D1 Thr⁵⁹⁰ in EDL muscle of AMPK mdKO and WT mice, with no significant differences between rested and prior contracted muscle (Fig. 7A). Furthermore, phosphorylation of TBC1D1 Ser²³¹ had returned to resting levels 3 h after contraction and did not respond to submaximal insulin stimulation (Fig. 7B). Also, phosphorylation of TBC1D1 Ser²³¹ was reduced in EDL muscle from AMPK mdKO mice compared with WT mice (Fig. 7B). Phosphorylation of TBC1D1 Thr⁵⁹⁰ and Ser²³¹ in WT SOL muscle (Fig. 7C and D) was similar to findings in WT EDL muscle, indicating that TBC1D1 is not involved in regulating muscle insulin sensitivity after contraction.

To determine whether GS was secondarily affecting 2-deoxyglucose uptake in skeletal muscle, we measured GS phosphorylation and activity. In WT mice, basal GS activity (%I-form) was similar between prior rested and contracted EDL muscle (Fig. 8A). In contrast, GS activity was significantly higher in previously contracted EDL muscle compared with rested muscle in AMPK mdKO mice. Submaximal insulin stimulation significantly increased GS activity in both rested and prior contracted EDL muscle independent of genotype (Fig. 8A). Total GS activity was similar between genotypes and did not change in response to prior contraction or submaximal insulin stimulation (Fig. 8B). GS activity increases by dephosphorylation (44). In the current study, phosphorylation at COOH-terminal GS residues (3a+3b) decreased similarly in rested and prior contracted EDL muscle during submaximal insulin stimulation in both genotypes (Fig. 8C). No significant differences in phosphorylation at N-terminal GS residues (2+2a) were observed in response to prior contraction or submaximal insulin stimulation (Fig. 8D). Notably, EDL muscle from AMPK mdKO mice had an ∼40% reduction in GS site 2+2a phosphorylation in line with the notion that AMPK is a kinase for GS site 2 (45,46). This may explain the higher GS activity observed in EDL muscle of the AMPK mdKO mouse (Fig. 8A).

In skeletal muscle, insulin-stimulated glucose uptake is suggested to be regulated by glycogen content per se (47). Three hours after in situ contraction, glycogen content was lower in previously contracted EDL muscle compared with rested muscle (Fig. 8E). Interestingly, glycogen content was significantly lower in prior contracted EDL muscle from AMPK mdKO mice compared with EDL muscle from WT littermates, suggesting that any potential influence of glycogen per se is not a factor explaining the loss of contraction-induced insulin sensitization in muscle from AMPK mdKO mice. Notably, glycogen content was similar between prior contracted and rested WT SOL muscle in parallel with normal insulin sensitivity (Fig. 8F).

The current study represents a further contribution to the search of molecular transducers involved in the insulin-sensitizing effect of exercise. Here, we provide evidence to support that AMPK is necessary for increasing insulin sensitivity to stimulate glucose uptake in EDL muscle after in situ contraction, as well as enhancing whole-body insulin sensitivity and insulin-stimulated muscle glucose uptake after a single bout of acute exercise. We establish a causal link between a contraction-regulated signal and the subsequent improvement in muscle insulin sensitivity. On the basis of our findings, we propose that contraction-induced activation of AMPK potentiates the ability of insulin to increase phosphorylation of TBC1D4 leading to enhanced muscle glucose uptake.

Theoretically, synthesis of new proteins involved in muscle glucose uptake may mediate improvements in skeletal muscle insulin sensitivity after contraction. However, we found that greater insulin-stimulated glucose uptake after contraction occurred without an increase in the abundance of multiple proteins involved in insulin-mediated signaling, as also supported by findings from others (12,15) and our previous observations in man (5,11,28). We found that the HKII protein level was decreased in EDL muscle from AMPK mdKO mice compared with WT mice. However, since maximal insulin-stimulated glucose uptake was similar between genotypes, this indicates that lower HKII protein abundance in EDL muscle from AMPK mdKO mice is not rate limiting for the ability of insulin to stimulate glucose uptake after contraction.

Similar to previous findings in humans and rodents (4–6,17), the increase in skeletal muscle insulin sensitivity after contraction was not associated with enhanced proximal insulin signaling measured at the level of phosphorylated Akt Thr³⁰⁸ and Ser⁴⁷³. This further supports the notion that the intracellular mechanism responsible for increasing muscle insulin sensitivity after contraction is located downstream of Akt or involves insulin-regulated parallel pathways converging with elements regulating glucose transport.

Improved insulin sensitivity after exercise is associated with enhanced translocation of GLUT4 to the cell surface membrane in skeletal muscle (14). Contraction-induced phosphorylation of TBC1D1, as well as insulin-induced phosphorylation of TBC1D4 regulates glucose uptake and GLUT4 translocation in skeletal muscle (26,41,42,48), indicating a role of these proteins in enhancing skeletal muscle insulin sensitivity after exercise/contraction. Since phosphorylation of TBC1D1 was similar between genotypes (WT vs. mdKO) and muscle type (EDL vs. SOL), this indicates that TBC1D1 is not involved in the insulin-sensitizing effect of prior contraction. This is supported by several other studies in humans and rats (11,17,18,49).

In contrast to TBC1D1, we observed an increased effect of a submaximal dose of insulin to stimulate phosphorylation of TBC1D4 Thr⁶⁴⁹ and Ser⁷¹¹ in prior contracted WT EDL muscle but not in EDL muscle from AMPK mdKO mice. Furthermore, insulin-stimulated phosphorylation of TBC1D4 was not enhanced in prior contracted WT SOL muscle. We hypothesize that the potentiating effect of AMPK activation by prior contraction on insulin-stimulated phosphorylation of TBC1D4 Ser⁷¹¹ induces a subsequent increase in TBC1D4 Thr⁶⁴⁹ phosphorylation, which may facilitate the enhanced effect of insulin on glucose uptake. These observations are supported by positive and significant correlations between TBC1D4 Thr⁶⁴⁹ and Ser⁷¹¹ phosphorylation, as well as between glucose uptake and phosphorylation of TBC1D4 Thr⁶⁴⁹/Ser⁷¹¹, of which Thr⁶⁴⁹ has previously been reported to be important for muscle glucose uptake in response to insulin (50). Moreover, these observations are fully in line with our previous findings in prior AICAR-stimulated EDL muscle (30).

In the current study, contraction-induced phosphorylation of TBC1D4 Ser⁷¹¹ is dependent on AMPKα1α2 and more specifically on AMPKγ3. This suggests that the increase in muscle insulin sensitivity after prior contraction may be mediated through increased AMPKγ3 activity during and/or after contraction. This idea is supported by observations in the AMPKγ3-scarce SOL muscle (21) in which prior in situ contraction failed to enhance insulin sensitivity, AMPKγ3 activity, and phosphorylation of TBC1D4 Ser⁷¹¹. It is also in line with our previous findings showing that the increased effect of insulin on muscle glucose uptake after prior AICAR stimulation is dependent on AMPKγ3 (30). Studies in humans and rats have not found evidence to support increased AMPK activity at time points of enhanced muscle insulin sensitivity after exercise/contraction (11,18,29). This may be related to measures of AMPK phosphorylation or surrogate measures of this (e.g., pACC and pTBC1D1) that potentially conceal differential regulation among the AMPK heterotrimers (51,52), as also evident in the current study.

The amount of muscle glycogen consumed during an exercise bout may play a role in regulating postexercise insulin sensitivity (53). However, we found that the electrical stimulation protocol decreased glycogen content to similar levels in muscle from both genotypes, indicating that glycogen depletion per se does not mediate changes in muscle insulin sensitivity as also previously suggested (54). Skeletal muscle glycogen content and insulin-stimulated glucose uptake display an inverse relationship (47,55). Thus, glycogen levels at the time of insulin stimulation (rather than immediately after contraction) may be of importance for muscle insulin sensitivity. However, immediately before insulin stimulation, muscle glycogen content was lower in prior contracted mdKO muscle compared with WT muscle, indicating that this is not the reason for the lost ability of prior contraction to enhance muscle insulin sensitivity in AMPK mdKO mice. Since glycogen content was lower in prior contracted muscle compared with rested muscle in WT mice, we cannot rule out a functional role of decreased glycogen content for mediating the insulin-sensitizing effect of prior contraction. This may be supported by observations in prior contracted WT SOL muscle in which glycogen content had returned to resting levels concomitant with normal insulin sensitivity. In fact, it may be hypothesized that reduced glycogen levels signal via AMPK to enhance muscle insulin sensitivity after exercise/contraction. At present we do not possess solid evidence to support this idea, and studies using AICAR (15,30) indicate at least that it is possible to bypass this association.

In human and rodent skeletal muscle displaying increased insulin sensitivity after exercise, GS activity is higher in prior exercised muscle compared with nonexercised muscle (1,5,9,11). Thus, higher GS activity may be needed for the prior exercised muscle to handle the increased amount of glucose taken up during insulin stimulation. Because ∼30% of 2-deoxyglucose taken up by muscle is incorporated into glycogen during insulin stimulation (56, and J.R. Hingst and J.F.P.W., unpublished observations in mouse muscle), our findings on 2-deoxyglucose uptake may be influenced by possible dysregulation of GS activity in skeletal muscle of AMPK mdKO mice. However, we found that in vitro GS activity and phosphorylation were regulated similarly in muscle of AMPK WT and mdKO mice in response to insulin. In fact, GS activity was elevated in prior rested and prior exercised muscle from AMPK mdKO mice compared with WT mice. This suggests that elevated GS activity is not a primary driver for improvements in muscle insulin sensitivity at the level of glucose uptake.

In conclusion, we provide evidence to support that prior contraction increases insulin sensitivity in EDL muscle to stimulate glucose uptake by an AMPK-dependent mechanism. Since the relative distribution of AMPK heterotrimeric complexes in human vastus lateralis greatly resembles that of mouse EDL muscle (20,21), our findings may be of high relevance for glucose metabolism in human skeletal muscle as well. Furthermore, we recently found intact regulation of the AMPK signaling network in skeletal muscle of patients with type 2 diabetes (51), and several findings of enhanced insulin-stimulated phosphorylation of TBC1D4 in prior exercised human muscle (11,28) support the notion of an AMPK-TBC1D4 signaling axis regulating muscle insulin sensitivity. Altogether, we provide basic insight to a physiological role of AMPK in skeletal muscle, strengthening the idea of AMPK being a relevant target for physiological and pharmacological interventions in the prevention and treatment of muscle insulin resistance in various conditions.

Acknowledgments. The authors thank Betina Bolmgren, Irene Bech Nielsen, and Jeppe Kjærgaard Larsen (Department of Nutrition, Exercise and Sports, Faculty of Science, University of Copenhagen) for their skilled technical assistance.

Funding. This study was supported by grants from the Danish Council for Independent Research, Medical Sciences; the research programme (2016) “Physical activity and nutrition for improvement of health” funded by the University of Copenhagen; the Lundbeck Foundation; the Novo Nordisk Foundation; and the Novo Nordisk Foundation Center for Basic Metabolic Research. The Novo Nordisk Foundation Center for Basic Metabolic Research is an independent research center at the University of Copenhagen that is partially funded by an unrestricted donation from the Novo Nordisk Foundation (www.metabol.ku.dk).

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

Author Contributions. R.K. designed and performed the experiments, analyzed the data, and wrote the manuscript. N.M.-H. performed the experiments and analyzed the data. J.B.B. performed biochemical assays and analyzed the data. M.F., B.V., M.B., and J.R.Z. provided founder mice for the study cohort. J.T.T. and J.F.P.W. designed the experiments and wrote the manuscript. All authors interpreted the results, edited and revised the manuscript, and read and approved the final version of the manuscript. J.F.P.W. 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.