Molecular Mechanism by Which AMP-Activated Protein Kinase Activation Promotes Glycogen Accumulation in Muscle

d-[U-¹⁴C]glucose-1-phosphate was purchased from Hartmann Analytic GmbH (Brunswick, Germany). All other radiochemicals were from Perkin Elmer (Buckinghamshire, U.K.). Human insulin (Actrapid) was from Novo-Nordisk (Bagsværd, Denmark). AICAR was from Toronto Research Chemicals (Ontario, Canada). Wortmannin was from Tocris (Avonmouth, U.K.). Horseradish peroxidase-conjugated secondary antibodies were from Thermo Fisher (Northumberland, U.K.). All other chemicals, unless specified otherwise, were obtained from Sigma (Poole, U.K.).

Animal studies were approved by the University of Dundee Ethics Committee and performed under a U.K. Home Office project license. C57BL/6 J mice were obtained from Harlan (Leicestershire, U.K.). Generation of the muscle GS^R582A/R582A knock-in (22) and transgenic AMPK kinase dead (K45R) (23) mice has been previously described.

Total acetyl CoA carboxylase (ACC), phospho ACC1 (Ser⁷⁹, also detects equivalent site on mouse ACC2 at Ser²¹²), total raptor, phospho raptor (Ser⁷⁹²), total glycogen synthase (GS), phospho GS (Ser⁶⁴¹), total AMPKα, phospho AMPKα (Thr¹⁷²), and phospho protein kinase B (PKB; Thr³⁰⁸) antibodies were from Cell Signaling Technology (Beverly, MA). Hexokinase II antibody (C-14) was obtained from Santa Cruz Biotechnology (Santa Cruz, CA). The phospho GS antibodies (Ser⁸ and Ser^8/11) (24) and the total AMPKα1 and AMPKα2 antibodies used for immunoprecipitation were generated and donated by Professor D. Grahame Hardie (University of Dundee). GLUT4 antibody was generated and donated by Professor Geoffrey Holman (University of Bath). Generation of the TBC1D1 (total [S279C, 1st bleed] and antiphospho Ser²³¹ [S131C, 2nd bleed]) (25), phosphorylase (total [S956A, 2nd bleed] and antiphospho Ser¹⁵ [S960A, 1st bleed]), and total PKBα (S742B, 2nd bleed) antibodies (26) has been previously described.

Mice were killed by cervical dislocation, and extensor digitorum longus (EDL) muscles were rapidly removed and mounted on an incubation apparatus as described (26). Muscles were incubated in 2 mL Krebs-Ringer bicarbonate (KRB) buffer (in mmol/L: 117 NaCl, 4.7 KCl, 2.5 CaCl₂, 1.2 KH₂PO₄, 1.2 MgSO₄, and 24.6 NaHCO₃, pH 7.4) containing 5.5 mmol/L d-glucose for 40 min at 37°C gassed continuously with 95% O₂ and 5% CO₂. At the end of the incubation period muscles were frozen in liquid nitrogen and stored at −80°C.

EDL muscles were isolated and 2-deoxy-[³H]glucose uptake was measured as described (27). Briefly, muscles were incubated in 2 mL KRB containing 2 mmol/L pyruvate for 40 min at 37°C. Muscles were transferred to 2 mL KRB containing 1 mmol/L 2-deoxyglucose (1.5 mCi/mmol 2-deoxy-d-[1,2-³H(N)]glucose) and 7 mmol/L d-mannitol (0.064 mCi/mmol d-[1-¹⁴C]mannitol) and incubated for an additional 10 min at 30°C. Muscles were frozen in liquid nitrogen and acid hydrolysates subjected to scintillation counting (27).

d-[U-¹⁴C]glucose incorporation into muscle glycogen was determined as previously described (26). Briefly, EDL muscles were incubated in 2 mL KRB containing 5.5 mmol/L d-glucose and 0.1 mCi/mmol d-[U-¹⁴C]glucose for 40 min at 37°C. Muscles were frozen in liquid nitrogen and [¹⁴C]glycogen extracted by ethanol precipitation from KOH digests as described (26).

The rate of glycolysis was determined by the detritiation of [5-³H]glucose as described (22). Briefly, muscles were incubated in 2 mL KRB containing 5.5 mmol/L glucose and 0.5 mCi/mmol [5-³H]glucose for 40 min at 37°C. Muscles were frozen in liquid nitrogen, and the rate of [5-³H]glucose incorporation into glycogen was determined as described for d-[U-¹⁴C]glucose. [³H]H₂O was isolated from conditioned KRB by borate complex ion exchange chromatography and measured by scintillation counting.

Muscle lysates were prepared as described (26). Homogenates were clarified at 3,000g for 10 min at 4°C, and protein concentration was estimated using Bradford reagent and bovine serum albumin (BSA) as standard. Lysates were frozen in liquid nitrogen and stored at −80°C.

Muscle extracts (20–30 μg) were denatured in SDS sample buffer, separated by SDS-PAGE, and transferred to polyvinylidene fluoride membrane. Membranes were blocked for 1 h in 20 mmol/L Tris-HCl (pH 7.5), 137 mmol/L NaCl, and 0.1% (v/v) Tween-20 (TBST) containing 5% (w/v) skimmed milk. Membranes were incubated in primary antibody prepared in TBST containing 5% (w/v) BSA overnight at 4°C. Detection was performed using horseradish peroxidase-conjugated secondary antibodies and enhanced chemiluminescence reagent.

Muscle homogenates (25 μg) were assayed for glycogen synthase and phosphorylase activity (reverse direction) by measuring the incorporation of UDP-[U-¹⁴C]glucose and [U-¹⁴C]glucose-1-phosphate respectively into glycogen, as described (22). Results are expressed as the activity ratio in the absence and presence of 10 mmol/L G6P (glycogen synthase) or 2 mmol/L AMP (phosphorylase).

AMPK was immunoprecipitated from 30 μg lysate with antibodies against the α1 and α2 subunits and assayed for phosphotransferase activity toward AMARA peptide (AMARAASAAALARRR) using [γ-³²P]ATP, as previously described (28).

Frozen muscles were digested in 100 μL of 1 mol/L KOH for 20 min at 80°C. The pH was adjusted to 4.8 with 50 μL of 4 mol/L acetic acid and 250 μL of 4 units/mL amyloglucosidase (Aspergillus niger) in 0.2 mol/L sodium acetate (pH 4.8 added). Samples were incubated for 2 h at 40°C, clarified at 16,000g for 10 min, and neutralized with NaOH. Glucose released from glycogen was determined using a commercial hexokinase/G6P dehydrogenase (G6PDH) coupled assay (Amresco, Solon, OH) using d-glucose as a standard.

G6P was assayed fluorometrically in HClO₄ extracts of EDL muscle, as previously described (22).

Data are expressed as means ± SEM. Statistical analysis was performed by unpaired, two-tailed Student t test or one-way or two-way ANOVA with Dunnett post hoc test. Differences between groups were considered statistically significant when P < 0.05.

We first measured the effect of the pharmacologic AMPK activator, AICAR, on AMPK activity in EDL muscle (composed mainly of fast-twitch, glycolytic fibers) isolated from male C57BL/6 mice. EDL was used because the effects of AICAR on AMPK activity and glucose uptake are reported to be robust in this muscle (29). As previously reported (27), incubation of EDL ex vivo with AICAR promoted robust phosphorylation of the AMPKα catalytic subunit at Thr172 in the T-loop and activation of both AMPKα1 (∼twofold) and AMPKα2 (∼fourfold) compared with unstimulated muscle (Fig. 1A and B). Consistent with these observations, AICAR treatment increased phosphorylation of known AMPK substrates such as ACC2 at Ser212, raptor at Ser792 (30), and TBC1D1 at Ser231 (25,31) (Fig. 1B).

We next assessed the effect of AICAR on phosphorylation and activity of GS in isolated EDL muscle. Muscles were incubated in parallel with GS-activating (insulin) and GS-inhibiting (adrenaline) hormones as controls. AICAR treatment resulted in a modest but significant decrease in GS activity (∼10%) (Fig. 2A). However, in contrast to previous studies (13,14), no detectable change in the phosphorylation of GS at Ser8, an inhibitory site that is targeted by AMPK (12), was observed (Fig. 2B). Because phosphorylation of Ser8 is known to promote subsequent phosphorylation of Ser11 by casein kinase 1, which cannot be detected by Ser8 phospho-specific antibody, we also monitored dual phosphorylation of Ser8 and Ser11 by a phospho-specific antibody that detects only when both sites are phosphorylated. AICAR failed to induce a detectable change in the phosphorylation of Ser8 and Ser11, whereas adrenaline robustly increased phosphorylation of these sites, correlating with a marked decrease in GS activity (Fig. 2A and B). AICAR did not modify phosphorylation of other key regulatory sites such as Ser641, whereas insulin promoted dephosphorylation of this site likely via the PKB/GS kinase 3 (GSK3) pathway (Fig. 2B), resulting in a ∼1.7-fold increase in GS activity (Fig. 2A and B), as previously reported (32).

A well-established physiologic role of AMPK in muscle is to stimulate glucose transport (5). Incubation of EDL with AICAR or insulin significantly increased 2-deoxy-[³H]glucose uptake by ∼twofold and ∼1.5-fold, respectively (Fig. 3A). AICAR also caused an increase (2.5-fold) in the levels of G6P compared with resting EDL (Fig. 3B). Although chronic/continuous AICAR treatment is known to promote glucose transport and phosphorylation at least partially through an increase in the levels of GLUT4 and hexokinase II- (16,17) short-term incubation (40 min) did not cause detectable changes in the amount of these proteins (Fig. 3E). We next measured glycogen synthesis by monitoring incorporation of d-[¹⁴C-U]glucose into glycogen in response to AICAR or insulin. Both AICAR and insulin stimulated glycogen synthesis by ∼1.6-fold (Fig. 3C), which was associated with a significant increase (1.4-fold) in muscle glycogen concentration (Fig. 3D).

The effect of AICAR on both glucose uptake and glycogen synthesis was distinct from those by insulin, and the phosphoinositide-3 kinase pathway as wortmannin, a selective phosphoinositide-3 kinase inhibitor, abolished insulin-stimulated glucose transport and glycogen synthesis, but not in response to AICAR (Fig. 4A and B).

Net glycogen content and incorporation of [¹⁴C]glucose are dependent on both glycogen synthesis and degradation. Consequently, the increased glycogen content observed in AICAR-treated muscles could be partly due to a decrease in glycogen degradation. To rule out this possibility, we assessed the activity of muscle glycogen phosphorylase, a rate-limiting enzyme in glycogen degradation. Phosphorylase activity and phosphorylation at Ser15, an activating residue that is catalyzed by phosphorylase kinase, were both unchanged in response to AICAR or insulin (Fig. 5A and B). To confirm that our assay was sensitive enough to detect changes in phosphorylase activity, isolated EDL muscles were also incubated in the presence of adrenaline, an activator of phosphorylase through a phosphorylase kinase-dependent pathway. As expected, adrenaline caused a significant activation of phosphorylase (twofold) that was accompanied by an increase (∼50%) in phosphorylation of Ser15 (Fig. 5B).

To establish that AICAR-stimulated glycogen synthesis is mediated through AMPK and not by off-target action(s) of the compound, we assessed the effect of AICAR on glycogen synthesis in EDL from mice expressing catalytically inactive/kinase dead (KD) AMPKα2 (23). Immunoblot analysis using anti-pan-AMPKα antibody confirmed that KD AMPK, epitope-tagged with Myc, displayed a slightly slower mobility and endogenous AMPKα was absent (Fig. 6D). As previously reported, AMPKα2 activity was largely eliminated, and AMPKα1 was also substantially reduced at rest in EDL from AMPK KD animals (Fig. 6A), an effect likely due to the displacement of endogenous α2 and also α1 from βγ heterotrimers by over-expressed KD α2, as has previously been suggested (23). Consistent with previous studies (33), AICAR significantly stimulated both AMPKα1 and AMPKα2 activity in wild-type mice, whereas neither AMPKα1 nor α2 activity was significantly increased in muscles from AMPK KD mice (Fig. 6A). AICAR caused a robust increase in ACC2 phosphorylation, which was partially suppressed (∼30–40%) in AMPK KD mice (Fig. 6D), as previously reported (33).

We next measured the effect of AICAR on GS activity in EDL isolated from AMPK KD or wild-type littermate control animals. In unstimulated muscles, GS activity was significantly higher (about twofold) in AMPK KD compared with wild-type mice (Fig. 6B), which was associated with a modest decrease in GS phosphorylation at both Ser8 and Ser641 (Fig. 6D). AICAR caused a modest (∼10%) decrease in GS activity in wild-type mice but not in AMPK KD animals, demonstrating that AMPK catalytic activity, most likely α2 activity (13), is required for GS inactivation by AICAR (Fig. 6B). AICAR did not significantly promote single Ser8 and dual Ser8/11 phosphorylation in EDL from wild-type or AMPK KD mice. Insulin promoted GS dephosphorylation on Ser641 and activation in muscles of wild-type and AMPK KD mice to a relatively similar degree (Fig. 6B and D). Insulin did not cause a significant change in Ser8 phosphorylation, although some muscle showed a slightly reduced phospho signal on this site (Fig. 6D). We next measured glycogen synthesis and observed that AICAR-stimulated increases were abolished in AMPK KD animals, whereas insulin stimulated glycogen synthesis in both genotypes to the same extent (Fig. 6C). Total protein expression and phosphorylation of phosphorylase was similar between wild-type and AMPK KD muscles in both resting and AICAR-treated mice (Fig. 6D).

To establish that AICAR-stimulated glycogen synthesis is mediated through allosteric activation of GS by G6P, we have used G6P-insensitive GS knock-in mice in which a critical G6P permissive residue (Arg582) is changed to Ala (GS^R582A/R582A). We have recently reported that mutant GS derived from GS^R582A/R582A mouse skeletal muscle is completely resistant to G6P but retains its capacity to be activated through dephosphorylation by GSK3 inhibition in response to insulin (22).

As reported previously, glycogen synthesis in resting EDL was comparable between wild-type and GS^R582A/+ mice, and there was a substantial decrease (∼70–80%) in glycogen synthesis in GS^R582A/R582A animals compared with wild-type or GS^R582A/+. AICAR stimulated glycogen synthesis in wild-type and GS^R582A/+ mice by ∼twofold and ∼1.5-fold, respectively, whereas AICAR had no effect on glycogen synthesis in GS^R582A/R582A animals (Fig. 7A). There was a trend that AICAR modestly inhibited glycogen synthesis in GS^R582A/R582A mice, possibly through deactivation of GS or stimulation of phosphorylase. To rule out the possibility that the reduced glycogen synthesis observed in GS^+/R582A and GS^R582A/R582A mice was caused by reduced glucose transport or altered AMPK activity, or both, these parameters were measured in isolated EDL muscle in the presence or absence of AICAR. We observed no difference in basal or in AICAR-stimulated glucose uptake across all genotypes (Fig. 7B). We also confirmed that resting and AICAR-stimulated AMPK activity (α1 and α2) were similar between wild-type and GS^R582A/R582A mice (Fig. 7C).

Interestingly, despite the significant impairment in glycogen synthesis, overall glucose utilization, as determined by metabolism of [5-³H]glucose was also similar between wild-type and GS^R582A/R582A mice (Fig. 7D). The decrease in glycogen synthesis in GS^R582A/R582A mice was compensated by a significant increase in the rate of glycolysis in both resting and AICAR-stimulated muscles. Consistent with this finding, the rate of lactate release was also elevated under resting and AICAR-stimulated conditions in GS^R582A/R582A mice compared with wild-type (Fig. 7E). We confirmed that GS activity in resting EDL muscles was comparable between wild-type and GS^R582A/R582A mice when assayed in the absence of G6P, whereas the robust G6P-mediated activation observed in wild-type mice was completely abolished in the muscle of GS^R582A/R582A mice (Fig. 7F). AICAR modestly deactivated muscle GS in both wild-type and GS^R582A/R582A animals (Fig. 7F). Potentially, hypophosphorylation of GS by inhibition of GSK3 may compensate for the lack of allosteric activation by G6P in GS^R582A/R582A mice. Accordingly, we co-incubated EDL muscle with AICAR and the GSK3 selective inhibitor, CT99021, and observed a significant, albeit modest increase in glycogen synthesis compared with muscle treated with AICAR alone (Supplementary Fig. 1).

We (8) and others (16,17) previously reported that AICAR treatment caused an increase in muscle glycogen levels; however, whether this was mediated through AMPK has not been clearly demonstrated. AICAR (Z-riboside) is a widely used AMPK activator that is taken up into cells by adenosine transporters and converted by adenosine kinase to the monophosphorylated derivative ZMP (34). ZMP binds to the γ subunit of AMPK and mimics the effect of AMP on the allosteric activation of the kinase and also on inhibition of dephosphorylation (35). However, because AICAR has been found to produce AMPK-independent effects due to binding of ZMP to other AMP/ZMP-regulated enzymes and also to unknown off-target effects (34), we wished to establish that AICAR-stimulated glycogen synthesis occurs in an AMPK-dependent manner.

Abbott Laboratories has recently identified a direct and specific activator of AMPK, the thienopyridine A-769662, which is a useful tool to understand the physiologic consequences of AMPK activation in animals (36–38). However, Scott et al. (39) have demonstrated that this compound selectively activates β1-containing AMPK trimeric (αβγ) complexes, but not β2-complexes in cell-free assays. They further investigated the selectivity of A-769662 in vivo and showed that A-769662 failed to stimulate AMPK in AMPKβ1-deficient mouse tissues (39). The two β1-containing AMPK heterotrimers (α1β1γ1 and α2β1γ1), as well as activity associated with these complexes, appear to constitute only a small fraction of the total pool of AMPK trimeric complex/activity (i.e., <5%) in mouse skeletal muscle (40). A-769662 treatment resulted in only a modest activation of AMPK (37,39) and failed to promote AMPK-dependent increases in glucose uptake in isolated mouse skeletal muscles (40). Therefore, A-769662 would not be a suitable tool to study the effect of AMPK on glucose/glycogen metabolism in mouse skeletal muscle. As an alternative approach, we used a genetically engineered mouse model, AMPK KD, in which AMPK is inactivated by transgenic over-expression of a dominant inhibitory AMPKα2 mutant in muscle cells (6), and established that catalytic activity of AMPK is necessary to elicit AICAR-stimulated glycogen synthesis most likely by increased glucose transport and subsequent accumulation of intracellular G6P.

Consistent with previous studies, AICAR modestly decreased the muscle GS activity ratio (13,15), a measure of phospho-dependent activity, in cell-free assays in wild-type animals, and this effect was lost in AMPK KD mice. However, we failed to detect a consistent elevation in GS phosphorylation at site 2 (Ser8) or site 2 + 2a (Ser8 and Ser11) in EDL incubated with AICAR ex vivo. These antibodies have been stringently validated (24), so it would appear that the effect of AICAR on site 2 phosphorylation is modest (as evidenced by only ∼10% decrease in GS activity), and in the current study, it failed to reach statistical significance.

This modest effect of AICAR on GS phosphorylation probably explains why a previous study failed to observe significant GS inhibition in skeletal muscle (8). A more robust effect of AICAR on GS activity and phosphorylation has been observed in rat skeletal muscle, although the reason for this is unclear (13). We tested if G6P directly affected GS phosphorylation by AMPK in response to AICAR. AMPK-mediated Ser8 phosphorylation of GS was not significantly altered in the presence of G6P in cell-free assays, although supraphysiologic concentrations of G6P (2 mmol/L) modestly reduced GS Ser8 phosphorylation by AMPK (Supplementary Fig. 2). Although we cannot completely rule out the possibility that AMPK inactivates GS through phosphorylation at other site(s), apart from Ser8, a previous in vitro phospho-mapping study did not identify additional phosphorylation sites (12). In support of this, there were no other obvious AMPK target site(s) (the typical consensus motif is φX[B, X)XX(Ser/Thr]XXXφ, where φ is a hydrophobic residue, B is basic, and X is any residue) (1) on GS when motif scan analysis (Scansite, http://scansite.mit.edu/) was performed. Regardless of the mechanism, the modest inactivation of GS by AMPK was overridden by the allosteric stimulator, G6P, resulting in elevated GS activity in vivo as evidenced by an increase in [¹⁴C]glucose incorporation into glycogen.

Gain-of-function mutation in AMPKγ1 (R70Q) (41) and naturally occurring mutations in γ3 (R200Q) originally identified in the Hampshire pig (Sus scrofa domesticus; RN-) (42), are known to promote marked glycogen accumulation in skeletal muscle. In addition, mutations in the AMPKγ2 subunit (encoded by the PRKAG2 gene) have been implicated in the human cardiomyopathy, Wolff-Parkinson-White syndrome, which is characterized by ventricular pre-excitation, and in certain cases, cardiac hypertrophy (18). Notably, several genetic studies have revealed that γ2 mutations cause excess myocardial glycogen accumulation (43–45), which is hypothesized to cause conduction system abnormalities by unknown mechanism(s) (18,46). Interestingly, Luptak et al. (9) demonstrated that transgenic mice over-expressing one of the γ2 mutations (N488I) in cardiomyocytes displayed aberrant high activity of AMPK resulting in elevated intracellular [G6P] due to increased glucose uptake, which serves as both the carbon skeleton for glycogen synthesis and the allosteric stimulator of GS. It would be of major interest to cross AMPKγ2 mutant transgenic animals with G6P-resistant GS^R582A/R582A knock-in mice and investigate if abnormal cardiac glycogen accumulation and the associated pathologies can be rescued.

In summary, we provide genetic evidence that AMPK-mediated glycogen synthesis occurs by increased activity of GS through its allosteric stimulator, G6P, and we propose the following model (Fig. 8): elevated glucose transport promoted by increased AMPK activity causes an accumulation of intracellular G6P. This leads to allosteric activation of GS, which overrides the inhibitory effect of AMPK on GS resulting in a net increase in GS activity and excess glycogen storage in muscle cells. Our work is of particular importance when considering AMPK as a target for treating metabolic disorders such as type 2 diabetes (47) because chronic/persistent activation of AMPK may have adverse consequences on cardiac function.

This study was supported by Diabetes UK (07/0003529 to K.S.), Dundee and district of Diabetes U.K. volunteer group, British Heart Foundation (PG/09/059 to K.S.), the U.K. Medical Research Council, Danish Medical Research Councils, the Novo Nordisk Research Foundation, and the Lundbeck Foundation. This work was done as a part of the research program of the UNIK: Food, Fitness & Pharma for Health and Disease (see www.foodfitnesspharma.ku.dk). The UNIK project is supported by the Danish Ministry of Science, Technology and Innovation.

This study also received support from AstraZeneca, Boehringer Ingelheim, GlaxoSmithKline, Merck–Serono, and Pfizer to K.S. No other potential conflicts of interest relevant to this study were reported.

R.W.H. designed and conducted most of the experiments, analyzed data, and wrote the manuscript. J.T.T. and J.F.P.W. conducted experiments, analyzed data, reviewed and edited the manuscript, and contributed to discussion. K.S. designed and conducted experiments, analyzed data, wrote the manuscript, and also supervised the project.

The authors thank Morris Birnbaum (University of Pennsylvania) for the donation of AMPK kinase dead animals, Gail Fraser and members of the resource unit for genotyping and technical assistance, and the antibody purification teams (Division of Signal Transduction Therapy, University of Dundee, Dundee, Scotland, U.K.), coordinated by Hilary McLauchlan and James Hastie.