Prior AICAR Stimulation Increases Insulin Sensitivity in Mouse Skeletal Muscle in an AMPK-Dependent Manner

The effect of insulin on skeletal muscle glucose uptake is increased in the period after a single bout of exercise. This phenomenon is observed in muscle from both humans and rodents (1–6) and may persist for up to 48 h after exercise, depending on carbohydrate availability (7–9). Improved muscle insulin sensitivity postexercise is mediated by one or several local contraction-induced mechanisms (10) involving both enhanced transport and intracellular processing of glucose. This period is characterized by increased GLUT4 protein abundance at the plasma membrane and enhanced glycogen synthase activity (11,12). These changes occur independent of global protein synthesis (13), including both total GLUT4 and glycogen synthase protein content (4,11), and are independent of changes in proximal insulin signaling, including Akt activation (3,4,13–17).

AMPK is a heterotrimeric complex consisting of catalytic (α1/α2) and regulatory subunits (β1/β2 and γ1/γ2/γ3). Of the 12 heterotrimeric combinations, only 3 and 5 combinations have been found in the skeletal muscle of human and mouse, respectively (18,19). AMPK is activated in response to various stimuli that increase cellular energy stress (e.g., metformin, hypoxia, hyperosmolarity, muscle contraction, and exercise) (20). With energy stress, intracellular concentrations of AMP and ADP accumulate. This activates AMPK allosterically and decreases the ability of upstream phosphatases to dephosphorylate Thr¹⁷², which further increases AMPK phosphorylation and activity (21). Like exercise, AICAR increases AMPK activity in skeletal muscle (22), which partly mimics the metabolic changes observed during muscle contraction (23).

TBC1D4 is involved in insulin-stimulated glucose transport in skeletal muscle (24) and is regulated via phosphorylation at multiple sites by Akt (25), thereby increasing translocation of GLUT4 to the plasma membrane. AMPK also targets TBC1D4; however, this does not seem to directly affect glucose uptake (26). As insulin (Akt) and exercise/AICAR (AMPK) signaling pathways converge on TBC1D4, this may explain how exercise modulates insulin action to regulate glucose transport in skeletal muscle. Supporting this concept, TBC1D4 phosphorylation is elevated in skeletal muscle several hours after an acute bout of exercise in both rodents and humans, concomitant with increased insulin sensitivity to stimulate glucose uptake in the postexercise period (15,16,27–30).

Prior AICAR stimulation increases skeletal muscle insulin sensitivity (13). However, because AICAR exerts multiple AMPK-independent effects (31), the direct relationship between AMPK and muscle insulin sensitivity has not been established. Thus, the primary purpose of the current study was to determine whether AMPK directly regulates skeletal muscle insulin sensitivity on glucose uptake. We established an ex vivo protocol using mouse muscle to study insulin sensitivity after prior AICAR stimulation and tested the hypothesis that AMPK is necessary for the effect of AICAR to enhance insulin sensitivity. Furthermore, we evaluated TBC1D4 phosphorylation status because this protein is a convergence point for insulin- and exercise-mediated signaling events.

All experiments were approved by the Danish Animal Experimental Inspectorate and the regional animal ethics committee of Northern Stockholm and complied with the European Union Convention for the Protection of Vertebrate Animals Used for Scientific Purposes (Council of Europe 123, Strasbourg, France, 1985). Except for the wild-type (WT) mice (C57BL/6J; Taconic, Ejby, Denmark) used in Figs. 1, 3E, and 8, the animals used in this study were muscle-specific kinase-dead α₂-AMPK (AMPK KD) (32), muscle-specific α₂- and α₁-AMPK double-knockout (AMPK mdKO) (33), and γ₃-AMPK KO mice (34) with corresponding WT littermates used as controls. All mice in this study were female (mean weight 24.3 ± 0.2 g) and 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. Serum was obtained from healthy young men in accordance with a protocol approved by the Ethics Committee of Copenhagen (protocol #H-3–2012–140) and complied with the ethical guidelines of the Declaration of Helsinki II. Informed consent was obtained from all participating subjects before they entered the study.

Fed animals were anesthetized by intraperitoneal injection of pentobarbital (10 mg/100 g body wt) before soleus and extensor digitorum longus (EDL) muscles were dissected and suspended in incubation chambers (Multi Wire Myograph System; DMT, 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, 8 mmol/L mannitol, and 2 mmol/L pyruvate. During the entire incubation period, the buffer was oxygenated with 95% O₂ and 5% CO₂, and maintained at 30°C. After 10 min of preincubation, muscles were incubated for 50 min in the absence or presence of 1 mmol/L AICAR (Toronto Research Chemicals, Toronto, Ontario, Canada) in 100% human serum from healthy overnight-fasted men. The use of serum is necessary to elicit an effect of AICAR on muscle insulin sensitivity (13). Soleus and EDL muscles were allowed to recover in the absence of AICAR in modified KRB supplemented with 5 mmol/L glucose, 5 mmol/L mannitol, and 0.1% BSA for 4 h (soleus muscle) or 6 h (EDL muscle). During recovery, the medium was replaced once every hour to maintain an adequate glucose concentration. Subsequently, paired muscles from each animal were incubated for 30 min in KRB in the absence or presence of a submaximal concentration (100 μU/mL) of insulin (Actrapid; Novo Nordisk, Bagsvaerd, Denmark). The uptake of 2-deoxyglucose was measured during the last 10 min of the 30-min period by adding 1 mmol/L [³H]2-deoxyglucose (0.056 MBq/mL) and 7 mmol/L [¹⁴C]mannitol (0.0167 MBq/mL) to the incubation medium. After incubation, muscles were harvested, washed in ice-cold KRB, quickly dried on filter paper, and frozen in liquid nitrogen.

Muscles were homogenized in 400 μL of ice-cold buffer (10% glycerol, 20 mmol/L sodium pyrophosphate, 1% NP-40, 2 mmol/L phenylmethylsulfonyl fluoride [PMSF], 150 mmol/L sodium chloride, 50 mmol/L HEPES, 20 mmol/L β-glycerophosphate, 10 mmol/L sodium fluoride, 1 mmol/L EDTA, 1 mmol/L EGTA, 10 μg/mL aprotinin, 3 mmol/L benzamidine, 10 μg/mL leupeptin, and 2 mmol/L sodium orthovanadate, pH 7.5) for 2 × 30 s at 30 Hz using steel beads and a TissueLyzer II (QIAGEN, Hilden, Germany). Homogenates were rotated end over end for 1 h before centrifugation at 16,000g for 20 min. The supernatant (lysate) was collected, frozen in liquid nitrogen, and stored at −80°C for later analyses.

Glucose uptake was assessed by the accumulation of [³H]2-deoxyglucose in muscle with the use of [¹⁴C]mannitol (PerkinElmer, Waltham, MA) as an extracellular marker. Radioactivity was measured in 250 μL of lysate by liquid scintillation counting (Ultima Gold and Tri-Carb 2910 TR; PerkinElmer) and was related to the specific activity of the incubation buffer.

Total protein abundance in muscle lysates was determined by the bicinchoninic acid method (ThermoFisher Scientific, Waltham, MA). Muscle lysates were prepared in Laemmli buffer and heated for 10 min at 96°C. Equal amounts of protein were separated by SDS-PAGE on 5% or 7% self-cast gels and transferred to polyvinylidene fluoride membranes using semidry blotting. Membranes were blocked for 5–10 min in 2% skim milk or 3% BSA and probed with primary and secondary antibodies. Proteins with bound antibody were visualized with chemiluminescence (Millipore) using a digital imaging system (ChemiDoc MP System; Bio-Rad). All membranes were stripped with buffer (100 mmol/L 2-mecaptoethanol, 2% SDS, 62.5 mmol/L Tris-HCl, pH 6.7) and reprobed with new primary antibodies for the detection of other phosphorylation sites on identical proteins or the corresponding total proteins. The stripping procedure was verified by reincubating membranes with secondary antibodies for the detection of primary antibodies that were possibly still bound.

The following antibodies were from Cell Signaling Technology (Danvers, MA): anti–phospho-AMPK-Thr¹⁷² (catalog #2531), anti–phospho-acetyl-CoA carboxylase (ACC) Ser⁷⁹ (catalog #3661), anti-Akt2 (D6G4) (catalog #3063), anti–phospho-Akt-Thr³⁰⁸ (catalog #9275), anti–phospho-Akt-Ser⁴⁷³ (catalog #9271), anti–phospho-TBC1D1-Thr⁵⁹⁰ (catalog #6927), anti–phospho-TBC1D4-Ser³¹⁸ (catalog #8619), anti–phospho-TBC1D4-Ser⁵⁸⁸ (#8730), and anti–phospho-TBC1D4-Thr⁶⁴² (catalog #8881). Anti–DYKDDDDK-Tag (FLAG-Tag) (catalog #F1804; Sigma-Aldrich), anti–phospho-TBC1D1-Ser²³⁷ (catalog #2061452; Millipore), anti-TBC1D1 as previously described (35), anti-AS160 (TBC1D4) (catalog #07–741; Millipore), anti–phospho-TBC1D4-Ser⁷¹¹ as previously described (26), and anti–AMPK-α2 (catalog #SC-19131; Santa Cruz Biotechnology). The antibodies used for AMPK activity measurements were anti–AMPK-γ3, anti–AMPK-α1, and anti–AMPK-α2, all of which were provided by Professor D.G. Hardie (University of Dundee, Dundee, Scotland, U.K.).

Five different AMPK trimer complexes have been detected in mouse skeletal muscle: α2β2γ3, α2β1γ1, α2β2γ1, α1β1γ1, and α1β2γ1 (19). α2β2γ3-AMPK activity was measured on γ3-AMPK immunoprecipitates (IPs) from 300 μg of muscle lysate using AMPK-γ3 antibody, G-protein–coupled agarose beads (Millipore) and IP buffer (50 mmol/L NaCl, 1% Triton X-100, 50 mmol/L sodium fluoride, 5 mmol/L sodium-pyrophosphate, 20 mmol/L Tris-base, pH 7.5, 500 μmol/L PMSF, 2 mmol/L dithiothreitol, 4 μg/mL leupeptin, 50 μg/mL soybean trypsin inhibitor, 6 mmol/L benzamidine, and 250 mmol/L sucrose). Samples were treated as previously described (19,36). In short, after overnight end-over-end rotation at 4°C, IPs were centrifuged for 1 min at 2,000g and washed once in IP buffer, once in 6× assay buffer (240 mmol/L HEPES, 480 mmol/L NaCl, pH 7.0), and twice in 3× assay buffer (1:1). The activity assay was performed for 30 min at 30°C in a total volume of 30 μL of kinase mix (40 mmol/L HEPES, 80 mmol/L NaCl, 833 μmol/L dithiothreitol, 200 μmol/L AMP, 100 μmol/L AMARA peptide, 5 mmol/L MgCl₂, 200 μmol/L ATP, and 2 μCi of [γ-33P]-ATP; PerkinElmer). The reaction was terminated by adding 10 μL of 1% phosphoric acid. Twenty microliters of the reaction mix were spotted on P81 filter paper. Filter papers were subsequently washed 4 × 15 min in 1% phosphoric acid. 33P radioactivity was analyzed on dried filter paper using a Storm 850 PhosphorImager (Molecular Dynamics). The combined activity of α2β1γ1 and α2β2γ1 was measured on supernatants from the γ3-AMPK IPs using the AMPK-α2 antibody for a second IP, and the combined activity of α1β1γ1 and α1β2γ1 was measured on supernatants from the α2-AMPK IPs using α1-AMPK antibody for a third IP.

TBC1D4 WT and TBC1D4 T649A and S711A DNA mutant constructs, containing T-to-A and S-to-A point mutations, respectively, were commercially and individually synthesized from the gene encoding mouse TBC1D4 (GeneArt; Life Technologies, Darmstadt, Germany). All three constructs were subsequently subcloned into a p3xflag-cmv-9–10 vector using NotI and KpnI cloning sites before amplification in Escherichia coli TOP10 cells (Invitrogen). Plasmid DNA was extracted using an endotoxin-free Plasmid Mega Kit (QIAGEN) and was diluted in isotonic saline solution to a final concentration of 2 μg/μL. DNA (50 μg) was injected into the tibialis anterior muscle 2 h after treatment with hyaluronidase (Sigma-Aldrich) (one injection of 30 units/muscle, 1 unit/μL), and gene electrotransfer was performed as previously described (24). Seven days after gene electrotransfer, phosphorylation of TBC1D4 Thr⁶⁴⁹ and Ser⁷¹¹ was assessed in the tibialis anterior muscle of anesthetized (8 mg pentobarbital/100 g body wt) animals in response to retro-orbital injection of either saline solution or insulin (10 units/kg). Ten minutes after injection, the tibialis anterior muscle was removed, quickly frozen in liquid nitrogen, and stored at −80°C for subsequent analysis.

Statistical analyses were performed using SigmaPlot version 11.0 (SYSTAT, Erkrath, Germany) and SPSS version 20 (IBM) software. SPSS version 20 was used for three-way ANOVA with repeated measures, while all other analyses were performed using SigmaPlot version 11.0. Data are presented as the mean ± SEM. One-, two-, or three-way ANOVAs with or without repeated measures was used to assess statistical differences, where appropriate. When a three-way interaction occurred (P < 0.05; genotype × AICAR × insulin), a two-way ANOVA with repeated measures was used on each genotype (WT and KD or WT and mdKO) in order to determine the site of interaction between AICAR and insulin (P < 0.05; AICAR × insulin). Any main effects of genotype are included in the figure legends. For post hoc testing, a Student-Newman-Keuls test was used. Correlation analyses were performed by determination of Pearson product moment correlation coefficient. Differences were considered statistically significant at P < 0.05.

Acute AICAR stimulation increased glucose uptake and AMPK phosphorylation in both soleus and EDL muscles (Fig. 1A, B, and F). We then determined the time point at which glucose uptake had reversed to basal levels in order to evaluate the effect of AICAR on insulin sensitivity. In WT soleus and EDL muscle, glucose uptake reversed to basal levels after 4 and 6 h of recovery from AICAR stimulation, respectively (Fig. 1C). Prior AICAR treatment increased the effect of a submaximal insulin concentration (100 µU/mL) to stimulate glucose uptake in the EDL muscle, but not in the soleus muscle of WT mice (Fig. 1D and E). Based on these results, we chose to use only EDL muscle for subsequent experiments.

In order to clarify whether the effect of AICAR on insulin sensitivity is dependent on AMPK, we took advantage of the AMPK KD and AMPK mdKO mouse models in which AMPK activity is decreased or ablated in skeletal muscle. Prior AICAR stimulation increased insulin sensitivity in isolated EDL muscle from WT littermates but failed to increase insulin sensitivity in both transgenic models (Fig. 2A and B). The incremental increase in insulin-stimulated glucose uptake (glucose uptake with insulin minus basal glucose uptake) was significantly higher after prior AICAR stimulation in WT littermates only (Fig. 2C and D).

As AICAR acutely increases phosphorylation of AMPK and the downstream target ACC, we investigated whether this effect was maintained into recovery. Phosphorylation of AMPK and ACC was increased in EDL muscle previously stimulated with AICAR independent of genotype (Fig. 3A–D, H, and I). We assume that the observed increase in ACC phosphorylation in muscle from both transgenic mouse models after prior AICAR treatment corresponds to AMPK-independent effects of AICAR on ACC phosphorylation or AMPK activation in nonmuscle cells. However, prior AICAR treatment increased ACC phosphorylation to a greater extent in muscle from WT littermates than in muscle from both transgenic models (although the increase was significant only in WT mice from the mdKO model), indicating a maintained effect of prior AICAR stimulation on AMPK in muscle cells. Therefore, we measured AMPK activity in WT EDL muscle that had been previously stimulated with AICAR. The combined activity of α1β1γ1 and α1β2γ1 was ∼1.4-fold higher compared with unstimulated control muscle (P = 0.037), while α2β2γ3 activity was ∼2.3-fold higher (P < 0.001) (Fig. 3E). In contrast, the combined activity of AMPK trimer complexes α2β2γ1 and α2β1γ1 was unchanged by prior AICAR treatment. This indicates a persistent effect of prior AICAR stimulation on specific AMPK trimer activity in particular α2β2γ3 activity.

A persistent increase in AMPK α2β2γ3 activity after AICAR stimulation prompted us to test the hypothesis that the effect of AICAR to enhance muscle insulin sensitivity is mediated through the AMPK α2β2γ3 trimer complex. Indeed, prior AICAR stimulation failed to increase muscle insulin sensitivity in whole-body γ₃-AMPK KO mice (Fig. 3F and G). For unknown reasons, prior AICAR treatment still affected basal glucose uptake in WT littermates in this particular experiment, suggesting that the acute effect of AICAR on glucose uptake was not fully reversed.

Prior AICAR stimulation potentially enhances muscle insulin sensitivity to stimulate glucose uptake by regulating proximal insulin-signaling proteins. To investigate this, we measured the phosphorylation of Akt Thr³⁰⁸ and Ser⁴⁷³. Insulin did not further increase the phosphorylation of Thr³⁰⁸ and Ser⁴⁷³ in muscle previously stimulated with AICAR compared with control muscle (Fig. 4A–F).

TBC1D1 is a closely related paralog of TBC1D4 that is regulated by both AMPK and Akt, and regulates glucose transport (37–40). As AMPK increases the phosphorylation of TBC1D1 Ser²³¹ in response to contraction and AICAR, and as Akt increases the phosphorylation of Thr⁵⁹⁰ in response to insulin, we investigated whether changes in TBC1D1 phosphorylation occurred in parallel with the increase in muscle insulin sensitivity after prior AICAR stimulation. Phosphorylation of TBC1D1 Ser²³¹ was markedly increased in WT muscle previously stimulated with AICAR. Prior AICAR stimulation also modestly increased the phosphorylation of TBC1D1 Ser²³¹ in muscle from AMPK KD and mdKO mice (Fig. 5A, B, E, and F). Furthermore, insulin increased the phosphorylation of TBC1D1 Thr⁵⁹⁰ in both mouse models independent of genotype (Fig. 5C–F). However, in AMPK mdKO mice and WT littermates, insulin-stimulated phosphorylation of TBC1D1 Thr⁵⁹⁰ in prior AICAR-stimulated muscle was decreased compared with control muscle (Fig. 5D).

TBC1D4 (like TBC1D1) has been identified as a substrate of both AMPK and Akt in skeletal muscle (25,26), and the phosphorylation of TBC1D4 is critical for insulin-stimulated glucose uptake (24,41). In addition, TBC1D4 is phosphorylated at multiple sites in the postexercise period in parallel with enhanced muscle insulin sensitivity (15,16,27–30). This indicates that the regulation of muscle insulin sensitivity is linked to TBC1D4 phosphorylation. We found an increased effect of insulin on TBC1D4 Thr⁶⁴⁹ and Ser⁷¹¹ phosphorylation in muscle previously stimulated with AICAR compared with control muscle (Fig. 6A–D, I, and J). Furthermore, this effect was dependent on AMPK, because no difference in insulin-mediated phosphorylation was observed between control and prior AICAR-stimulated muscle from either of the two AMPK transgenic models. Importantly, the effect of prior AICAR treatment was site specific, as insulin-stimulated phosphorylation of TBC1D4 Ser³²⁴ and Ser⁵⁹⁵ was unaffected (Fig. 6E–J)

To investigate whether AICAR/AMPK increases muscle insulin sensitivity through TBC1D4, we performed a correlation analysis between Δ values (insulin minus basal) on muscle glucose uptake and TBC1D4 phosphorylation. We found that glucose uptake and phosphorylation of TBC1D4 Ser⁷¹¹ was positively correlated in WT littermates from both AMPK mouse models (P < 0.001 and P < 0.01; Fig. 7A and B, respectively). Correlating data for glucose uptake and phosphorylation of TBC1D4 Thr⁶⁴⁹ revealed a more scattered pattern that was positively correlated in WT littermates from the AMPK KD strain (P < 0.01; Fig. 7C), but was not correlated in WT littermates from the AMPK mdKO strain (P = 0.18; Fig. 7D). In addition, the phosphorylation levels of TBC1D4 Thr⁶⁴⁹ and Ser⁷¹¹ were positively and strongly correlated in WT littermates from both AMPK mouse models (P < 0.001; Fig. 7E and F).

AMPK has been shown to regulate the phosphorylation of TBC1D4 Ser⁷¹¹, and in muscle overexpressing a 4P mutant of TBC1D4 (in which Ser⁷¹¹ is not mutated) the phosphorylation of Ser⁷¹¹ is severely blunted (26). In order to investigate whether changes in phosphorylation level of TBC1D4 Ser⁷¹¹ affect TBC1D4 Thr⁶⁴⁹ phosphorylation and vice versa, TBC1D4-WT, TBC1D4-S711A, and TBC1D4-T649A constructs were expressed in mouse tibialis anterior muscle by gene electrotransfer. Insulin increased the phosphorylation of TBC1D4 Thr⁶⁴⁹ in muscle expressing TBC1D4-WT or TBC1D4-S711A, but Thr⁶⁴⁹ phosphorylation levels were significantly blunted in the latter (Fig. 8A and C). Insulin increased the phosphorylation of TBC1D4 Ser⁷¹¹ in muscle expressing TBC1D4-WT, and this response was completely ablated in muscle expressing TBC1D4-T649A (Fig. 8B and C). Our results suggest that the phosphorylation levels of TBC1D4 Thr⁶⁴⁹ and Ser⁷¹¹ are mutually dependent on each other.

Several lines of evidence imply that AMPK activation regulates skeletal muscle insulin sensitivity. In C₂C₁₂ myotubes, AICAR stimulation or hyperosmotic stress increases insulin sensitivity, which is inhibited by the unspecific AMPK inhibitor compound C (42). Similarly, insulin sensitivity is increased in myotubes transfected with a constitutive active form of AMPKα, which is also suppressed by compound C (43). Furthermore, AICAR fails to increase insulin action in cells transfected with a dominant-negative form of AMPKα (43). Collectively, our data and those obtained in cell culture systems (42,43) suggest that AMPK plays an important role in mediating AICAR-induced increases in skeletal muscle insulin sensitivity to stimulate glucose transport.

AICAR is taken up by the cell, where it acts as an AMP mimetic, thus potentially affecting multiple proteins regulated by AMP. Within recent years, an increased number of AMPK-independent effects of AICAR have been described together with reports identifying AICAR as a modulator of enzymes such as glycogen phosphorylase, glucokinase, and phosphofructokinase (31). However, because AICAR did not increase insulin sensitivity in muscle from AMPK KD or mdKO mice, any possible AMPK-independent effect of AICAR does not seem to account for changes in glucose uptake in response to insulin.

The improved insulin-stimulated glucose uptake in muscle previously stimulated with AICAR occurred independently of changes in proximal insulin signaling (Akt phosphorylation). This is consistent with earlier findings showing that prior AICAR treatment does not increase either Akt phosphorylation or phosphoinositide-3 kinase activity in rat skeletal muscle (13). Similar observations have been made in both human and rodent skeletal muscle after an acute bout of exercise (3,4,13,16,17). Based on this, the mechanism responsible for the AMPK-dependent increase in muscle insulin sensitivity likely involves signal transduction downstream of Akt, implicating a role for TBC1D1 or TBC1D4.

We evaluated the phosphorylation status of key sites on TBC1D1 previously shown to increase in response to AICAR, muscle contraction, exercise, or insulin (16,37,44,45). Phosphorylation of TBC1D1 Ser²³¹ was markedly increased in muscle from WT mice, and only modestly increased in muscle from AMPK KD and mdKO mice 6 h after AICAR treatment. Conversely, phosphorylation of TBC1D1 Thr⁵⁹⁰ was increased in response to insulin independent of genotype. Given that prior AICAR stimulation increased the phosphorylation of TBC1D1 Ser²³¹ in both basal and insulin-stimulated muscle, and insulin increased the phosphorylation of TBC1D1 Thr⁵⁹⁰ in all groups, our results suggest that the phosphorylation of TBC1D1 Ser²³¹ and Thr⁵⁹⁰ is not sufficient for regulating muscle insulin sensitivity in response to prior AICAR treatment. This is supported by findings of identical TBC1D1 Ser²³¹ phosphorylation, and similar increases in insulin-stimulated PAS-TBC1D1 and Ser⁵⁹⁰ phosphorylation in previously rested or exercised muscle from humans and rats (15,16,29).

In addition to TBC1D1, we also analyzed the phosphorylation status of TBC1D4 at multiple sites because it has been suggested to play a prominent role in regulating both insulin-stimulated glucose uptake and insulin action in skeletal muscle after exercise (24,27–29). Recent studies (26,44), using site-specific antibodies, suggest that only the phosphorylation of TBC1D4 Ser⁷¹¹ is increased in mouse skeletal muscle in response to exercise, AICAR, or ex vivo muscle contraction. Because AICAR-mediated phosphorylation of TBC1D4 Ser⁷¹¹ is dependent on AMPK (26), the AMPK-dependent increase in insulin sensitivity after AICAR treatment may be mediated through changes in TBC1D4 Ser⁷¹¹ phosphorylation during acute AICAR stimulation. Our data showing increased insulin action on Ser⁷¹¹ phosphorylation in WT mouse muscle but not in AMPK KD or mdKO mouse muscle previously stimulated with AICAR are consistent with this notion.

In contrast with TBC1D4 Ser⁷¹¹, the phosphorylation of Thr⁶⁴⁹ seems to be important for insulin-stimulated glucose uptake in mouse EDL muscle (26,41). However, this site is not regulated by acute AICAR treatment (26,46). Thus, the potentiated effect of insulin on TBC1D4 Ser⁷¹¹ phosphorylation by prior AICAR treatment appears to mediate an enhanced AMPK-dependent phosphorylation of Thr⁶⁴⁹, which may facilitate the increased effect of insulin on glucose uptake. Such a relationship is supported by the correlative relationship between TBC1D4 Thr⁶⁴⁹ and Ser⁷¹¹ phosphorylation, and in particular by the strong relationship between Ser⁷¹¹ phosphorylation and muscle glucose uptake. In addition, the expression of the S711A TBC1D4 mutant decreased the total phosphorylation of TBC1D4 Thr⁶⁴⁹, whereas the T649A mutant severely decreased the phosphorylation of Ser⁷¹¹ both in the presence or absence of insulin. This clearly indicates interdependence between the two sites and supports a possible mechanism by which AMPK, through TBC1D4 Ser⁷¹¹, regulates insulin action to stimulate glucose uptake.

Previously, it has been shown (47) that discrepancies between Akt and TBC1D4 phosphorylation exist, indicating that only a small fraction of the insulin signal is necessary for mediating full glucose uptake in response to insulin. This is also observed in the current study where phosphorylation of Akt Ser³⁰⁸ and Thr⁴⁷³ does not match either phosphorylation of TBC1D4 Thr⁶⁴⁹ and Ser⁷¹¹ or glucose uptake in prior AICAR-stimulated WT muscle. However, in cases of normal (and perhaps increased) insulin sensitivity there seems to be a good correlation between plasma membrane GLUT4 and TBC1D4 phosphorylation (47). This indicates that glucose uptake and phosphorylation of TBC1D4 are associated, as also indicated by the correlations in the current study.

Besides a change in TBC1D4 phosphorylation, it has been demonstrated that AICAR enhances insulin action in muscle cells by decreasing membrane cholesterol levels in an AMPK-dependent manner (48). This seems plausible since AMPK has been shown to decrease the activity of 3-hydroxy-3-methylglutaryl CoA reductase, the rate-limiting enzyme in cholesterol synthesis (49). We did not measure membrane cholesterol levels, and therefore we cannot rule out that muscle insulin sensitivity after prior AICAR stimulation was affected by a change in membrane cholesterol content.

The enhanced insulin-stimulated glucose uptake after AICAR treatment seems to be dependent on a persistent increase in muscle γ₃-AMPK activity, as evidenced by AMPK activity measurements and insulin-stimulated glucose uptake in γ₃-AMPK KO mice. Furthermore, prior AICAR stimulation failed to increase insulin sensitivity in mouse soleus muscle in which the α2β2γ3 complex represents <2% of all AMPK trimer complexes (19). In both human vastus lateralis muscle (18) and mouse EDL muscle (19), the AMPK α2β2γ3 trimer complex accounts for one-fifth of all AMPK complexes (36). Of interest, AMPK-γ3 protein level is markedly decreased in skeletal muscle from trained humans (50), although insulin sensitivity in general is increased. Conversely, enhanced muscle insulin sensitivity after an acute bout of exercise seems to be lost in the trained state (51). Collectively, these results suggest that prior AICAR stimulation mimics the effect of exercise to enhance skeletal muscle insulin sensitivity, possibly through an AMPK-γ3–dependent mechanism.

In conclusion, prior AICAR stimulation is sufficient to enhance muscle insulin sensitivity. We provide evidence that this effect is likely mediated through AMPK signaling, as AICAR failed to increase insulin sensitivity in skeletal muscle in which AMPK activity was blunted. Although we observed no change in proximal insulin-signaling events, the enhanced insulin-stimulated glucose uptake observed after prior AICAR stimulation was associated and positively correlated with increased TBC1D4 Thr⁶⁴⁹ and Ser⁷¹¹ phosphorylation. This supports the idea that prior activation of AMPK primes a pool of TBC1D4 to potentiate a subsequent effect of insulin to increase GLUT4 translocation to the cell surface and enhance glucose uptake. At present, we have not succeeded in establishing a mouse model for studying insulin sensitivity after prior muscle contraction. Therefore, future studies have to determine whether AMPK is also important for the enhanced insulin action after this intervention. Because TBC1D4 signaling by insulin is potentiated after exercise in both human and rat skeletal muscle (15,16,29,30,52), our hypothesis is that the exercise-induced increase in insulin sensitivity is also regulated via an AMPK-TBC1D4 signaling axis.

Acknowledgments. The authors thank Ann-Marie Petterson, Integrative Physiology Group, Department of Molecular Medicine and Surgery, Karolinska Institutet, Stockholm, Sweden, for her assistance. The authors also thank D.G. Hardie (Division of Molecular Physiology, College of Life Sciences, University of Dundee, Scotland, U.K.) and L.J. Goodyear (Joslin Diabetes Center and Harvard Medical School, Boston, MA) for the donation of antibodies.

Funding. This work was carried out as a part of the research programs “Physical Activity and Nutrition for Improvement of Health” funded by the University of Copenhagen Excellence Program for Interdisciplinary Research and the UNIK project Food, Fitness & Pharma for Health and Disease (see www.foodfitnesspharma.ku.dk), which was supported by the Danish Ministry of Science, Technology and Innovation, and by 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). This study was also funded by the Danish Council for Independent Research Medical Sciences, the Novo Nordisk Foundation, and the Lundbeck Foundation. J.T.T. was supported by a postdoctoral fellowship from the Danish Agency for Science, Technology and Innovation.

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

Author Contributions. R.K. helped to conceive of and design the research, perform the experiments and analysis, and draft the manuscript. J.T.T. and J.F.P.W. helped to conceive of and design the research and draft the manuscript. J.F. helped to perform the experiments. L.L., B.V., J.B.B., P.S., and M.B. helped to perform the analysis. 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.