The metabolic role of 5′AMP-activated protein kinase (AMPK) in regulation of skeletal muscle metabolism in humans is unresolved. We measured isoform-specific AMPK activity and β-acetyl-CoA carboxylase (ACCβ) Ser221 phosphorylation and substrate balance in skeletal muscle of eight athletes at rest, during cycling exercise for 1 h at 70% peak oxygen consumption, and 1 h into recovery. The experiment was performed twice, once in a glycogen-loaded (glycogen concentration ∼900 mmol/kg dry wt) and once in a glycogen-depleted (glycogen concentration ∼160 mmol/kg dry wt) state. At rest, plasma long-chain fatty acids (FA) were twofold higher in the glycogen-depleted than in the loaded state, and muscle α1 AMPK (160%) and α2 AMPK (145%) activities and ACCβ Ser221phosphorylation (137%) were also significantly higher in the glycogen-depleted state. During exercise, α2 AMPK activity, ACCβ Ser221 phosphorylation, plasma catecholamines, and leg glucose and net FA uptake were significantly higher in the glycogen-depleted than in the glycogen-loaded state without apparent differences in muscle high-energy phosphates. Thus exercise in the glycogen-depleted state elicits an enhanced uptake of circulating fuels that might be associated with elevated muscle AMPK activation. It is concluded that muscle AMPK activity and ACCβ Ser221phosphorylation at rest and during exercise are sensitive to the fuel status of the muscle. During exercise, this dependence may in part be mediated by humoral factors.
- acetyl-CoA carboxylase
- fatty acids
in humans, 5′amp-activated protein kinase (AMPK) is stimulated in skeletal muscle during exercise, and the degree of activation is dependent on the exercise intensity (5, 14, 32, 59). In rodent muscle, pharmaceutical activation of AMPK by 5-aminoimidazole-4-carboxamide (AICA)-riboside leads to GLUT4 recruitment to the surface membrane and a stimulation of glucose transport that is additive to the effect of insulin but not to contraction (22, 28). In addition, contraction-stimulated glucose transport and AMPK activity covary with time in incubated skeletal muscle (33). On the basis of these findings, it is hypothesized that AMPK is a mediator of contraction-stimulated glucose transport (16, 22, 28). However, recent studies in muscle from mice that express a dominant inhibitory mutant of α-AMPK show that a substantial part of contraction-induced glucose transport in vitro is AMPK independent, whereas AICA-riboside- and hypoxia-induced transport is totally dependent on AMPK activation (31, 45). In accordance, rodent studies have revealed conditions under which muscle glucose transport by these stimuli is increased in response to contraction without concurrent AMPK activation (10). Furthermore, energy-depleting stimuli, such as 2,4-dinitrophenol and hyperosmolarity, increase AMPK activity and glucose transport in L6 cells and skeletal muscle (21,40). However, the increase in glucose transport, but not the AMPK activation, is inhibited by calcium chelation and protein kinase C inhibition (40). Thus the exact role of AMPK in the regulation of muscle glucose uptake during exercise is not clear.
Studies indicate that carnitine palmitoyltransferase I (CPT I) is released from inhibition when the cellular concentration of malonyl-CoA decreases. This increases the capacity for mitochondrial transport and oxidation of long-chain fatty acids. β-Acetyl-CoA carboxylase (ACCβ) is the enzyme that converts acetyl-CoA to malonyl-CoA. ACCβ and malonyl-CoA decarboxylase (MDC) are possibly important regulators of the cellular malonyl-CoA level in skeletal muscle (44,53). In the resting muscle, the level of malonyl-CoA regulates CPT I activity (1, 7, 28), and during euglycemic hyperinsulinemic clamp conditions, long-chain fatty acid oxidation is diminished, possibly due to inhibition of CPT I by increasing malonyl-CoA concentrations (3, 43, 47). ACCβ is inactivated when phosphorylated by AMPK on Ser221 in human muscle (8, 17). This, rather than activation of MDC, is possibly the main reason why AICA-riboside-induced AMPK activation leads to a fall in the malonyl-CoA concentration in muscle (18,24). Because exercise activates AMPK and leads to phosphorylation and inactivation of ACCβ, it is tempting to speculate that this leads to decreased malonyl-CoA levels and increased potential for fatty acid oxidation during exercise, as indicated by studies in rodents (28, 44) but so far not by studies in humans (9, 36, 37).
The cellular AMP, ATP, and creatine phosphate (CrP) levels are identified as major regulators of AMPK activity. When the AMP-to-ATP (AMP/ATP) and Cr-to-CrP (Cr/CrP) ratios increase, AMPK becomes activated allosterically (19, 54). AMPK activity also depends on phosphorylation by upstream kinases (19, 54), which are also regulated by the AMP/ATP ratio. In addition to several isoforms of the two regulatory subunits (β, γ) of AMPK, two isoforms of the catalytic subunit, α1 and α2, have been identified in mammalian cells (6, 49, 51). Manipulating glycogen in rat muscle by exercise followed by chow feeding, or by exercise followed by fat or carbohydrate feeding, leads to enhanced activation of AMPK by muscle contraction and AICA-riboside in the glycogen-depleted situations (10, 25, 57). When α-isoform-specific AMPK activity is measured, the basal level of α2, and to a lesser extent α1, activity is dependent on whether the muscles are loaded or depleted of glycogen (57). In two of the studies mentioned above (10, 57), adenine nucleotide content was measured and apparently was similar in glycogen-depleted and glycogen-loaded muscle under all conditions (basal, contraction, AICA-riboside). This suggests that “fuel status” of rodent muscle might modulate AMPK activity independently of energy status, as reflected in normal adenine nucleotide and CrP concentrations. The capacity for glycogen storage is markedly higher in human than in rodent muscle, and whether a similar fuel-sensing mechanism is operating in human skeletal muscle is unknown.
The present study was undertaken to investigate the role of AMPK activity in muscle carbohydrate and fat metabolism during exercise in healthy, well-trained men. To clarify the regulatory role of muscle fuel status on these cellular processes, the subjects were studied under conditions in which muscle glycogen content was either low or high. This design allowed an investigation of AMPK and ACCβ under conditions of either predominantly high-carbohydrate or high-fat oxidation, respectively.
Eight young (28 ± 1 yr), healthy, well-trained men gave their written informed consent to participate in the study, which was approved by the Copenhagen Ethics Committee and conformed to the Declaration of Helsinki. Body height, weight, and body mass index were 184 ± 2 cm, 81 ± 3 kg, and 24 ± 1 kg/m2, respectively. The subjects participated in physical exercise training 5–8 times per week (2–3 h per bout), for the most part performing cycle spinning, but also performing some resistance training. One to two weeks before the experiments, peak pulmonary oxygen consumption (V˙o 2 peak) was determined during an incremental cycling ergometer test (V˙o 2 peak = 65 ± 1 ml · kg−1 · min−1).
The subjects randomly underwent two experimental trials separated by 2 wk. The subjects were instructed to eat a mixed diet and to avoid exercise training for 2 days before each experiment. At noon on the day before an experiment, the subjects underwent a glycogen-depleting cycle exercise protocol. The subjects performed cycle exercise at 80%V˙o 2 peak for 60 min followed by arm cranking at 60 W for 15 min. Then a prolonged period of intermittent exercise was performed consisting of 1.5-min cycle exercise bouts at workloads of 100 and 50%V˙o 2 peak, respectively. When the subjects were unable to maintain the high intensity, the workload was lowered to 95% V˙o 2 peak. This protocol was followed until the high intensity reached 80%V˙o 2 peak. The subjects then performed exercise in 5-min intervals for 45 min, alternating between cycle exercise (75% V˙o 2 peak) and arm cranking (45 W). Finally, the subjects performed five full cycling sprints for 1 min with a 3-min rest period between them. The glycogen-depleting exercise on average lasted for ∼5 h and was well tolerated by all of the subjects. The subjects then left the laboratory and were instructed to eat a specified, controlled, isoenergetic diet for dinner and during the evening consisting of 80% of energy (E%) from carbohydrate, 7 E% from fat, and 13 E% from protein (total energy intake: 17.4 MJ) in one of the trials, and in the other trial, 2 E% from carbohydrate, 86 E% from fat, and 12 E% from protein (total energy intake: 16.3 MJ). The subjects were allowed to drink unlimited amounts of water and to eat the specified diet until 11:00 PM.
On the day of the experiment, the subjects ingested a light breakfast of 75 E% from carbohydrate, 8 E% from fat, and 17 E% from protein (total energy intake: 713 KJ) and, using a minimum of physical activity, arrived at the laboratory 2 h later. After they had rested for 45 min, Teflon catheters were inserted under the inguinal ligament in one femoral artery and one vein, as described previously (42). When 4 h had elapsed since breakfast, blood samples were drawn simultaneously from the arterial and venous catheters, and leg blood flow was measured using the bolus infusion thermodilution method as previously described (2). Then a needle biopsy was taken from the vastus lateralis under local anesthesia, and resting pulmonary V˙o 2 was measured using an online gas and airflow analyzer (Medgraphics, Medical Graphics, St. Paul, MN). The subjects then performed exercise on a cycle ergometer for 1 h at a relative workload of 70%V˙o 2 peak. Subsequently, the subjects rested in the supine position for 1 h. Biopsies were taken at 10 and 60 min of exercise and at 60 min into recovery. The biopsies were taken from one leg during the first trial and from the contralateral leg during the second trial. The biopsies were taken through two incisions spaced 5–6 cm apart. Blood samples were drawn, andV˙o 2 was measured at 10, 20, 30, 45, and 60 min of exercise and at 30 and 60 min during recovery. During exercise and recovery, blood flow was measured using the constant and bolus infusion thermodilution method, respectively (2,42). During exercise, the degree of perceived exertion (PE) was evaluated in accordance with the method described by Borg (4).
Hormones and metabolites in blood and plasma.
Blood glucose and lactate concentrations were measured using a dual-channel glucose-lactate analyzer (YSI-2700 Select, Yellow Springs Instruments, Yellow Springs, OH). Plasma free fatty acid (FA) concentrations were measured using an automatic spectrophotometer (Cobas FARA 2, Roche Diagnostic, Basel, Switzerland) in accordance with the method described by Shimizu et al. (48). Plasma concentrations of insulin, norepinephrine, and epinephrine were measured using radioimmunoassay kits (Insulin RIA 100, Pharmacia, Stockholm, Sweden, and KatCombi, Immuno-Biological Laboratories, Hamburg, Germany).
Muscle biopsy handling.
Biopsies were quickly frozen in liquid nitrogen while still situated in the biopsy needles. After termination of exercise, <30 s elapsed before the “exercise” biopsy was frozen. The frozen biopsies were freeze-dried and dissected free of visible blood, fat, and connective tissue before any analysis was performed.
Muscle nucleotides, Cr, and CrP.
Muscle specimens were extracted with perchloric acid, neutralized, and analyzed for nucleotides. Content of ATP, ADP, AMP, and IMP were determined by reverse-phase HPLC according to a previously described method (52). Muscle Cr and CrP content was measured fluorometrically, as previously described (26). Because of technical difficulties, content of adenine nucleotides, Cr, and CrP was obtained in only five of the eight subjects.
Glycogen content was determined as glycosyl units after acid hydrolysis (26).
Muscle AMPK activity and ACCβ Ser221phosphorylation.
Muscle AMPK activities and ACCβ Ser221 phosphorylation were measured in muscle lysates prepared as described previously (27).
α-Isoform-specific AMPK activity was measured in immunoprecipitates from 200 μg of muscle lysate protein with anti-α1 or anti-α2 antibodies. A p81 filter paper assay, using SAMS-peptide (HMRSAMSGLHLVKRR) (200 μmol/l) as substrate, was used to measure AMPK activity, as previously described (59).
ACCβ Ser221 phosphorylation was measured by the method described by Chen et al. (5). Succinctly, ACC from lysates containing 200 μg of protein was affinity purified by incubation with monomeric Avidin agarose beads overnight (Sigma, St. Louis, MO). Beads were washed twice in buffer [50 mM HEPES (7.5), 100 mM NaF, 2 mM Na3VO4, and 1% Triton X-100] and boiled in Laemmli buffer before subjected to SDS-PAGE (5%) and Western blotting. A phosphospecific ACCα Ser79 antibody was used for blotting, and enhanced chemiluminescence was used as a detecting system. A single band was revealed with the expected molecular mass of ∼260 KDa when analyzed using a CCD-image sensor (Image Station, E440CF, Kodak, Glostrup, Denmark). This antibody probably recognizes the equivalent Ser221 in human ACCβ.
Calculations and statistics.
Control samples were added to all activity assays, and assay-to-assay variation was accounted for by expressing the data relative to these samples. Net FA and glucose uptake across the limb were calculated by multiplying arteriovenous differences with either plasma or blood flow, respectively. Net FA and glucose clearance were calculated by dividing the uptake by the arterial concentration of the two metabolites, respectively. Carbohydrate and lipid utilization was based on indirect calorimetry by use of V˙o 2 and the pulmonary respiratory exchange ratio (RER) data, without correction for protein oxidation (12). Because protein and ketone oxidation, as well as ketone production, was probably significant during exercise in the low-glycogen condition, the calorimetry data can only be evaluated as rough estimates. Data are expressed as means ± SE. Statistical evaluation was performed by two-way analysis of variance (ANOVA) with repeated measures. When ANOVA revealed significant differences, a post hoc test was used to correct for multiple comparisons (the Student-Newman-Keuls test). Differences between groups were considered statistically significant whenP < 0.05.
On the day before the experiments, the subjects underwent a prolonged glycogen-depleting exercise protocol. On average, the subjects performed cycle exercise for 5 h before the predefined state of exhaustion was reached. In two of the eight subjects, biopsies were obtained before and immediately after this glycogen-depleting exercise. In these subjects, muscle glycogen content decreased from ∼630 to 70 mmol/kg dry wt. After exercise, the subjects ingested either a carbohydrate-poor [low-glycogen (LG)] or carbohydrate-rich [high-glycogen (HG)] diet. On the following day, muscle glycogen was low (163 ± 12 mmol/kg dry wt) or was supercompensated (909 ± 75 mmol/kg dry wt; Fig. 1).
During the 60 min of cycle exercise, muscle glycogen content decreased progressively in the HG trial only, but at all time points it was higher than in the LG trial (Fig. 1). During the 1-h recovery, only water was given to the subjects, and no glycogen resynthesis was observed. The RER at rest was higher in the HG compared with the LG trial (0.84 ± 0.007 vs. 0.81 ± 0.003, respectively,P < 0.05; Table 1). RER increased upon exercise to a level of 0.93 ± 0.01 in the HG trial, whereas it decreased during exercise to a level of 0.73 ± 0.01 in the LG trial. In recovery, RER remained low in the LG trial and decreased toward resting levels in the HG trial (Table 1). During the two trials, the exercise load was exactly the same (256 ± 9 W) but elicited a slightly higher pulmonary oxygen uptake during the LG trial compared with the HG trial (average during exercise 47 ± 0.3 vs. 45 ± 0.3 ml O2 · min−1 · kg−1, respectively, P < 0.05; Table 1). This difference in pulmonary oxygen uptake was not present before or after exercise.
Despite the light breakfast given 4 h before exercise was initiated, arterial plasma FA concentration at rest was higher in the LG compared with the HG trial (P < 0.01; Fig.2 A). Arterial plasma FA concentration remained significantly higher during the LG compared with the HG trial (P < 0.001). An initial small drop in arterial plasma FA concentrations was observed during exercise, but for the remaining period of exercise, the concentrations increased to preexercise levels and continued to rise (P < 0.01) above resting levels throughout recovery in both trials. The pretreatment protocol did not affect resting arterial blood glucose levels on the day of the experiment, although it tended to be lower in the LG trial (4.5 ± 0.2 vs 4.8 ± 0.1 mM). During exercise, arterial blood glucose concentration decreased in the LG trial only [60 min value: 3.4 ± 0.3 mM (LG) vs. 4.8 ± 0.2 mM (HG) (P < 0.01; Fig. 2 B)]. During recovery, blood glucose was maintained at the end-exercise levels in both the LG and HG trials.
Leg blood flow was similar in the two trials at rest (average: 0.4 ± 0.03 l/min) and during recovery (average30 and 60 min: 0.8 ± 0.04 l/min). Blood flow increased rapidly during exercise in both trials to steady-state levels, which were slightly higher in the LG trial (8.2 ± 0.2 l/min) compared with the HG trial (7.5 ± 0.2 l/min; P < 0.02). In the resting state, glucose and net FA uptake (arteriovenous differences times blood and plasma flow, respectively) were not different in the two trials (Fig. 3, A and B). Despite similar workloads, the exercising leg in the LG trial took up approximately twice the amount of glucose and FA as in the HG trial (Fig. 3, A and B). Glucose clearance was even more markedly elevated in the LG compared with the HG trial (Fig.4 B). On the other hand, net FA clearance was the same under the two conditions (Fig. 4 A). On the basis of the pulmonary oxygen uptake and the RER, we roughly estimated the amount of energy that was covered by oxidation of carbohydrate and fat, respectively. Thus, in the LG and HG trials, carbohydrate oxidation covered 41 ± 1 and 52 ± 2% at rest and 10 ± 2 and 79 ± 1% at the end of exercise, respectively. Total energy production was not significantly different between the two conditions at rest (LG 72 ± 3 and HG 66 ± 3 J · min−1 · kg body wt−1) or at the end of exercise (LG 975 ± 6 and HG 892 ± 13 J · min−1 · kg body wt−1).
Analysis of muscle biopsies taken before, during, and 1 h after termination of exercise showed that α2 AMPK activity was increased in response to exercise and returned toward resting levels during recovery (main effect, P < 0.01; Fig.5). Statistical analysis revealed that the increase in α2 activity was exclusively present in the LG trial. Interestingly, before and during (P < 0.001) but not after exercise (P = 0.17), α2 activities were significantly higher during the LG trial compared with the HG trial. Similarly, α1 AMPK activity was in general higher in the LG compared with HG trial (P = 0.03) but did not increase during exercise.
Because AMPK activity is measured using immunoprecipitates with in vitro assays, the AMPK activity data may not reflect the full activity in vivo. To evaluate this, we measured the degree of phosphorylation of the AMPK phosphorylation site Ser221 in ACC. In accordance with the AMPK activity, phosphorylation of ACC was also higher in the LG compared with the HG trial (Fig. 6; P < 0.001). ACC Ser221phosphorylation was induced by exercise in both trials (P < 0.001), in contrast to the in vitro-measured unchanged AMPK activity in the HG trial (Fig. 5). Nevertheless, two measures of AMPK activity, the in vitro activity assay and the ACC phosphorylation (reflecting both covalent and allosteric modifications), revealed that AMPK activity at rest and the degree of activation by exercise are enhanced when muscle is glycogen depleted and are blunted when glycogen levels are high.
The prevailing glycogen content did not affect the muscle concentrations of AMP, ATP, Cr, CrP, or the respective ratios of these at rest (Table 2). In both trials, exercise also did not induce changes in the muscle concentrations of these metabolites (Table 2). Muscle lactate content did not change significantly with time in either trial but was significantly lower in the LG compared with the HG trial (P < 0.03). Probably this difference could be attributed to a tendency for a higher lactate concentration during exercise in the HG compared with the LG trial (4.4 ± 1 vs. 2.7 ± 0.6 mmol/kg dry wt, values obtained at 60 min) rather than at rest (2.6 ± 0.4 vs. 2.7 ± 0.6 mmol/kg dry wt) and recovery (2.8 ± 0.7 vs. 2.8 ± 0.8 mmol/kg dry wt).
The catecholamine concentrations increased during exercise and, despite similar work performed, the increase was higher in the LG compared with the HG trial (P < 0.05; Table3). Heart rate was similar at rest but increased significantly to higher values in the LG compared with the HG trial (Table 1). Plasma insulin concentrations decreased ∼50% in both trials and tended to return to resting values in recovery (Table3). The average PE level was markedly higher during exercise in the LG compared with the HG trial (Table 1).
To investigate the regulatory role of muscle fuel status on carbohydrate and fat metabolism, we measured muscle-signaling responses and substrate utilization during exercise in well-trained males under conditions in which muscle glycogen content was either low or high. In the resting state, AMPK activity and ACCβ Ser221phosphorylation were lower in glycogen-loaded compared with glycogen-depleted muscles. This is in accordance with our previous findings in perfused rodent skeletal muscle (57). In that study and in the present one, concentrations of creatine phosphate and adenine nucleotides in resting muscles were not altered by the glycogen manipulation. This would suggest that fuel-dependent mechanisms independent of energy status may regulate AMPK signaling. This idea concurs with recent studies in which metformin and hyperosmolarity also seem to regulate AMPK signaling independently of changes in energy status (13, 20). In the present study, several factors besides muscle glycogen content are different in the two trials, even at rest, and evaluation of the role of each factor in the regulation of AMPK signaling is not possible from the present data alone. A recent study in vitro indicates that the presence of palmitate for 5 h during incubation does not increase AMPK activity in isolated rat skeletal muscle (38). This suggests that the increased AMPK activity at rest in the present study is not induced by the elevated plasma FA concentration in the glycogen-depleted trial. Besides in vitro data (10, 25, 57), other studies are supportive of the view that the muscle glycogen content per se is a regulator of muscle AMPK activity. Thus covariation of AMPK activation and glycogen degradation exists during short-term (20- to 30-min) cycling exercise performed at high and moderate (80 and 65%V˙o 2 peak) exercise intensities as well as during long-term (4-h) cycling exercise at low intensity (45%V˙o 2 peak) (34, 50, 58). On the other hand, in patients with McArdle's disease, the high muscle glycogen content does not affect AMPK activity compared with control subjects (35). In these patients, AMPK is also activated during exercise despite the absence of glycogen breakdown (35), indicating that factors other than glycogen also regulate AMPK activity during exercise. Another characteristic of the present study, and of studies that manipulate glycogen in rodents, is that glucose availability has either been limited or is plentiful in the period after the glycogen-depleting exercise (10, 25,57). The idea that glucose availability influences AMPK activity is also suggested from a range of studies in yeast, hepatocytes, pancreatic β-cells, and apparently also myoblasts (19, 46, 56,60, 61). Thus, although not proven by the present study, the carbohydrate availability during the pretreatment may have induced the differences in AMPK signaling on the following day during resting conditions.
Adenine nucleotide concentrations did not change during exercise. Somewhat more surprising was the observation that creatine phosphate also did not change during exercise, even in the LG trial. Due to loss of samples, these measurements were obtained from only five of the eight subjects, and this significantly weakens the power of the statistics performed. But evaluation of the individual data does not reveal any trend for such changes, and if we have missed a true difference, it most likely has been small in magnitude. Previous measurements suggest that our methods are indeed able to detect differences during exercise. For example, in muscle biopsies taken and handled in the same manner as in the present study, we recently obtained data showing decreased creatine phosphate in untrained (∼70%) and in trained (∼35%) subjects during exercise at somewhat higher intensity (80% V˙o 2 peak) than in the present study (34). The lack of changes in the present study could be due to the extreme fitness of the subjects. The fact that these subjects undertook spinning (indoor cycle) exercise training 5–8 times per wk for 2–3 h per bout may have accustomed them so that energy balance is better maintained even during cycling exercise demanding 70%V˙o 2 peak . In addition, IMP concentrations were very low, if detectable at all (average value ∼0.07 mmol/kg dry wt), and were not affected by exercise or glycogen content (data not shown). Because IMP may be a reflector of the free AMP concentration over time, these data also suggest that energy status has been well maintained during exercise. Nevertheless, it cannot be excluded that localized changes within the muscle cells were missed or that small changes had occurred that were normalized in the period between the end of exercise and the time that the biopsy was frozen (<30 s).
The present data do not prove that the altered fuel supply directly regulates muscle AMPK activity. During exercise, but not at rest, the catecholamine concentrations were different in the two trials. Thus, during the glycogen-depleted trial, both epinephrine and norepinephrine concentrations were elevated compared with the glycogen-loaded trial. This also most likely caused the differences in heart rate and is in line with the observation that the subjects felt the same work output to be heavier in the glycogen-depleted situation. Interestingly, several findings suggest that adrenergic stimulation may stimulate AMPK activity. Thus, in rat fat cells, short-term administration of the β-adrenergic agonist isoproterenol in vitro increased total AMPK activity threefold (30). Apparently similar findings have been observed in incubated skeletal muscle, as communicated in preliminary form by Park et al. (39). Also, in mice, leptin administration leads to an activation of muscle AMPK signaling in part through an α-adrenergic-dependent mechanism (29). Although muscle contractions certainly can activate AMPK by local mechanisms, the findings just mentioned may indicate that the catecholamine response to exercise in vivo is an additional regulator of muscle AMPK signaling. This view would also be in line with the observation that AMPK activation during exercise in vivo is markedly dependent on the exercise intensity (5, 14, 32,59), as is the adrenergic response. Additional experiments are needed to elucidate this possibility, because it may also provide an explanation for the observed AMPK activation in nonmuscle tissue after exercise in rodents (39). In this context, it is important to note that Ser79 on ACCα, or Ser221 on ACCβ, is possibly the most important phosphorylation site for the AMPK-induced regulation of ACC activity in vitro (8, 17). However, other sites on ACC can be phosphorylated by, for example, the cAMP-dependent protein kinase (PKA) (8, 17, 55), but the muscle isoform of ACC is not deactivated by PKA-induced phosphorylation (55). Thus, even though catecholamine concentrations are higher and PKA may be activated more during the LG than during the HG trial, this would not confound the measurements of Ser221phosphorylation on ACCβ in the present study.
As evident from the higher RER values and the larger glycogen breakdown, the energy utilized during exercise in the glycogen-loaded trial was derived mainly from carbohydrate metabolism. As previously observed in both rodents and humans, muscle glucose uptake was higher in the glycogen-depleted compared with the glycogen-loaded trial (15, 23, 41). The mechanism behind this phenomenon includes higher muscle glucose 6-phosphate content, thereby inhibiting glucose phosphorylation (15, 23) as well as diminished GLUT4 recruitment to the cell surface membrane when muscle glycogen is high (11). The present observation of an enhanced uptake of glucose despite decreasing plasma glucose concentrations supports the idea of enhanced cellular glucose uptake capacity in the glycogen-depleted trial (glucose transport and/or glucose phosphorylation). Several studies have indicated a regulatory role for AMPK in muscle glucose transport (22, 28), and it is tempting to speculate whether this might also be the case in the present study, because AMPK is activated to a greater extent in the glycogen-depleted trial. However, the covariation seen in the present study does not bring further evidence to this hypothesis, because many other factors might be involved and might be responsible for the covariance observed. In fact, a recent study using trangenic mice strongly suggests that contraction-stimulated glucose transport is only to a limited extent dependent on signals through the AMPK-signaling cascade (31).
In conclusion, AMPK activity in resting human muscle and the degree of activation during exercise are dependent on the fuel status of the muscle cells; i.e., AMPK activity is elevated in muscle low in glycogen. At rest, this dependency is independent of the energy status of the cells. During exercise, the low-glycogen state was also accompanied by increased leg glucose and net FA uptake as well as increased plasma concentrations of catecholamines compared with the glycogen-loaded state. These findings raise the possibility that AMPK activity may have a regulatory role in substrate utilization during exercise in humans. Furthermore, the data suggest that AMPK activity during exercise may be regulated by fuel availability as well as humoral factors.
We thank Betina Bolmgren and Karina Olsen for skilled technical assistance.
The study was supported by grants from the Danish National Research Foundation (no. 504–14), The Media and Grants Secretariat of the Danish Ministry of Culture, the Danish Diabetes Association, the Novo Nordisk Foundation, and by a Research and Technological Development Project (QLG1-CT-2001–01488) funded by the European Commission. J. F. P. Wojtaszewski was supported by a postdoctoral fellowship from the Danish Medical Research Council. A Programme Grant from Wellcome Trust supported D. G. Hardie. B. E. Kemp was supported by the National Health and Medical Research Council of Australia, the National Heart Foundation, the Wellcome Trust, and the Australian Research Council.
Address for reprint requests and other correspondence: J. F. P. Wojtaszewski, Copenhagen Muscle Research Centre, Institute of Exercise and Sport Sciences, August Krogh Institute, 13 Universitetsparken, Univ. of Copenhagen, 2100 Copenhagen, Denmark (E-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
First published December 17, 2002;10.1152/ajpendo.00436.2002
- Copyright © 2003 the American Physiological Society