The 5′-AMP-activated protein kinase (AMPK) is proposed to be involved in signaling pathways leading to adaptations in skeletal muscle in response to both a single exercise bout and exercise training. This study investigated the effect of endurance training on protein content of catalytic (α1, α2) and regulatory (β1, β2 and γ1, γ2, γ3) subunit isoforms of AMPK as well as on basal AMPK activity in human skeletal muscle. Eight healthy young men performed supervised one-legged knee extensor endurance training for 3 wk. Muscle biopsies were obtained before and 15 h after training in both legs. In response to training the protein content of α1, β2 and γ1 increased in the trained leg by 41, 34, and 26%, respectively (α1 and β2 P < 0.005, γ1 P < 0.05). In contrast, the protein content of the regulatory γ3-isoform decreased by 62% in the trained leg (P = 0.01), whereas no effect of training was seen for α2, β1, and γ2. AMPK activity associated with the α1- and the α2-isoforms increased in the trained leg by 94 and 49%, respectively (both P < 0.005). In agreement with these observations, phosphorylation of α-AMPK-(Thr172) and of the AMPK target acetyl-CoA carboxylase-β(Ser221) increased by 74 and 180%, respectively (both P < 0.001). Essentially similar results were obtained in four additional subjects studied 55 h after training. This study demonstrates that protein content and basal AMPK activity in human skeletal muscle are highly susceptible to endurance exercise training. Except for the increase in γ1 protein, all observed adaptations to training could be ascribed to local contraction-induced mechanisms, since they did not occur in the contralateral untrained muscle.
- 5′-adenosine monophosphate
- acetyl-coenzyme A carboxylase-β
the 5′-amp-activated protein kinase (AMPK) is a heterotrimeric enzyme consisting of a catalytic α-subunit and two regulatory subunits (β and γ). In skeletal muscle, the α- and β-subunits exist in two isoforms (α1, α2, β1, and β2) and the γ-subunit in three isoforms (γ1, γ2, and γ3) (5, 30, 32). AMPK activity is regulated by several mechanisms. A decrease in either the phosphcreatine-to-creatine or ATP-to-AMP ratio causes an allosteric activation of AMPK (26). Furthermore, binding of AMP makes AMPK a better substrate for upstream activating kinases (AMPKK) and a worse substrate for deactivating protein phosphatases (6), determining phosphorylation of the α-subunit on Thr172, which covalently regulates AMPK activity. Taken together, the interplay between allosteric regulation and covalent modification makes AMPK a potent sensor of the cellular energy stress occurring, for example, during exercise, hypoxia, or mitochondrial uncoupling in skeletal muscle (11, 14).
When exercise bouts are repeated regularly, i.e., in exercise training, adaptations occur enabling the muscles to tolerate exercise at higher intensities and/or for longer periods of time. At the cellular level, the improved exercise capacity is related to an altered expression of proteins involved in both substrate transport across the sarcolemma and in substrate metabolism inside the muscle cell. Furthermore, endurance training has been shown to increase glycogen levels and improve insulin sensitivity in human skeletal muscle beyond the isolated effect of the last exercise bout (7, 23). When AMPK is chronically activated in rodents by use of the compound 5-aminoimidazole-4-carboxamide-1-β-d-ribofuranoside (AICAR), a similar adaptive pattern is observed, including increased protein expression of GLUT4, hexokinase, and mitochondrial oxidative enzymes (3, 15, 36) as well as increased glycogen content and improved sensitivity for insulin to stimulate glucose uptake in skeletal muscle (3, 15, 16). Thus evidence is gathering to suggest an implication of AMPK in the adaptive regulation of skeletal muscle in response to exercise training. The idea that AMPK is involved in regulation of protein expression is also supported by the presence of α2-AMPK protein in nuclear preparations from INS-1 cells (29) and skeletal muscle (1, 22) and by the observed AICAR-induced phosphorylation of the transcriptional coactivator p300 in BHK-21 cells (41). Furthermore, the time courses and promoter requirements for AICAR and exercise-induced GLUT4 gene expression in skeletal muscle are strikingly similar (20, 43).
Recently, it has been demonstrated in both rodents and humans that the protein expression of AMPK itself is susceptible to endurance training (8, 17, 24). In two of these studies, an increased protein expression of the catalytic α1-subunit has been observed in human skeletal muscle (17, 24), possibly due to a posttranscriptional event (24). In the study by Nielsen et al. (24), protein expression of all seven isoforms of AMPK was examined. However, in that study an endurance-trained group was compared with an untrained control group, a study design that allows for factors other than exercise training per se to influence the results. To avoid this limitation, the present study was designed to evaluate the isolated effect of endurance training on the protein expression of AMPK in skeletal muscle by means of a longitudinal, one-legged knee extensor training regime. In addition, AMPK activity was measured both before and after training to determine whether changes in protein expression of AMPK coincide with alterations in activity of the enzyme.
MATERIALS AND METHODS
Subjects. Eight healthy young males (Table 1) gave their informed consent before participating in the study, which was approved by the local ethics committee and performed in accordance with the Helsinki declaration. All of the subjects met the criteria of having a maximal oxygen uptake <52 ml O2·kg-1·min-1 as measured during incremental cycling on a cycle ergometer as well as comparable (<5% difference) peak workload (PWL) of the knee extensors in the two legs. The latter was determined in each leg by the one-legged dynamic knee extensor apparatus (2).
Experimental design. On the morning after 2 days of mixed diet (55.5 ± 0.4% carbohydrate, 29.4 ± 0.4% fat, and 15.1 ± 0.2% protein) the subjects arrived at the laboratory in the fasted state. After 30 min of supine rest, needle biopsies were obtained from the vastus lateralis muscle in both legs under local anesthesia with 2% lidocaine. This procedure was repeated following 3 wk of supervised one-legged knee extensor training, with muscle biopsies obtained 15 h after the last exercise bout.
Training. The 3 wk of training consisted of four sessions the 1st wk, five sessions the 2nd wk, and six sessions the 3rd wk. Duration gradually increased from 1 to 2 h per session. To ensure an optimal endurance training regime, the intensity was aimed to vary between 70 and 85% of PWL. This was done in a progressive manner in accordance with an expected increase in PWL during this type of training (27). During each session, the subjects also worked for 5-7 min at 100% of PWL to ensure recruitment of the majority of muscle fibers in the knee extensor region (10). Previous results from our laboratory have shown that the PWL in the untrained leg is unaffected by this protocol (27); thus to evaluate the effect of training on the endurance capacity, a final determination of PWL was made only in the trained leg during the last session of exercise training.
Pulmonary gas analysis. During the incremental cycling test as well as during the PWL tests, expired air was collected in Douglas bags and analyzed for content of oxygen by use of a Servomex paramagnetic analyzer and for content of carbon dioxide by a Beckman infrared carbon dioxide analyzer. Total expired air was determined using a Tissot-type spirometer.
Muscle tissue preparation. The muscle biopsies obtained were frozen in liquid nitrogen within 30 s and stored at -80°C. Frozen muscle was freeze-dried and dissected free of visible fat, blood, and connective tissue before further analyses were performed.
Muscle glycogen. Muscle glycogen content was determined as glycosyl units after acid hydrolysis of freeze-dried muscle tissue by fluorometric methods (18).
Activities of citrate synthase and 3-hydroxyacyl-CoA dehydrogenase. Maximal activity of citrate synthase (CS) and 3-hydroxyacyl-CoA dehydrogenase (HAD) was measured at 25°C in freeze-dried muscle tissue by fluorometric methods (18).
Muscle lysate preparation. Freeze-dried muscle tissue was homogenized (21), and the resultant homogenate was rotated end over end at 4°C for 1 h before being centrifuged for 30 min (17,500 g, 4°C). The supernatant was harvested, frozen in liquid nitrogen, and stored at -80°C. Total protein content in the lysate was determined using the bicinchoninic acid method (Pierce, Rockford, IL).
SDS-PAGE and Western blotting. Muscle lysate proteins were separated using 7.5% or 10% Bis-Tris gels (Invitrogen, Taastrup, DK), and transferred (semidry) to PVDF membranes (Immobilon Transfer Membrane, Millipore, Glostrup, DK). After blocking [TBS + Tween (TBST) + 2% skim milk] the membranes were incubated with primary antibodies (TBST + 2% skim milk) followed by incubation in horseradish peroxidase-conjugated secondary antibody (TBST + 2% skim milk; Amersham Pharmacia Biotech, Buckinghamshire, UK). After detection and quantification using a charge-coupled device image sensor and 1D software (Kodak Image Station E440CF; Kodak, Ballerup, DK), the protein content was finally expressed in arbitrary units relative to a human skeletal muscle standard.
Antibodies used for AMPK subunit isoform detection. The primary antibodies used for detection of AMPK subunit isoforms α1, α2, β2, and γ2 were raised in sheep (5, 8, 40). To detect γ1 and γ3, rabbit antibodies against the NH2-terminal region of the human isoforms were used (Zymed Laboratories, South San Francisco, CA). Finally, an antibody raised in rabbit against an NH2-terminal region identical in both of the human β-isoforms was used for the detection of β1 (Upstate Biotechnology, Lake Placid, NY). This was possible because the two β-isoforms can be distinguished by their different electrophoretic mobility.
For all isoforms of AMPK, electrophoretic mobility corresponded to the expected molecular masses of the proteins. Antibodies used to recognize the seven isoforms of AMPK all detected a polypeptide of the expected mass when CCL-13 cell lysates containing recombinant rat (α1, α2, β1, β2, and γ1) or human (γ2 and γ3) protein isoforms were analyzed.
Antibodies used for detection of α-AMPK and acetyl-CoA carboxylase-β phosphorylation. Phosphorylation of α-AMPK(Thr172) and acetyl-CoA carboxylase-β (ACCβ)(Ser221) was detected using phosphospecific antibodies from Cell Signaling Technology (Beverly, MA) and Upstate Biotechnology, respectively. The ACC phosphospecific antibody is raised against a peptide corresponding to the sequence in rat ACCα containing the Ser79 phosphorylation site, but the antibody most likely also recognizes the human ACCβ when phosphorylated at the corresponding Ser221.
Detection of ACCβ protein. ACCβ contains a biotin moiety recognized by streptavidin, and ACCβ protein was accessed using horseradish peroxidase-conjugated streptavidin (Dako, Copenhagen, DK) (4).
AMPK activity associated with α1 or α2. AMPK activities were measured in immunoprecipitates from 100 μg of whole muscle lysate protein prepared using either anti-α1- or anti-α2-specific antibodies. At saturating AMP concentrations (200 μM), P81 filter paper-based kinase assays were performed using SAMS peptide (HMRSAMSGLHLVKRR) as substrate (200 μM) (39).
Statistics. Values before and after training in both legs were compared using two-way analysis of variance for repeated measures. When a significant main effect was observed, a Student-Newman-Keuls test was used as a post hoc test. A significance level of 0.05 was chosen. Some mean values are presented as percent difference from pretraining values. These values and the corresponding standard error are calculated from the individual percent difference for each subject. All data are expressed as means ± SE.
General adaptations to training. In response to training, PWL of the knee extensors increased in the trained muscle from 47.5 ± 1.6 to 55 ± 1.7 W (P < 0.001). Furthermore, maximal activities of CS and HAD increased on average ∼37 and ∼35%, respectively (both P < 0.005) compared with values before training (Table 2). No effects of training were observed on muscle glycogen content measured 15 h after the last exercise bout (Table 2).
Protein expression of AMPK isoforms. Compared with values before training, protein content of the α1-, β2-, and γ1-isoforms increased by 41 ± 12, 34 ± 9, and 26 ± 9%, respectively (α1 and β2 P < 0.005, γ1 P < 0.05; Figs. 1 and 2). In contrast, protein content of the γ3-subunit decreased by 62 ± 9% (P = 0.01) in response to training. No significant effect of training was seen on the protein content of α2, β1, or γ2. Except for a small but significant increase in γ1 protein (12 ± 9%, P < 0.05) no training-induced changes in the remaining isoforms of AMPK were observed in the contralateral untrained muscle.
α1- and α2-AMPK activities and α-AMPK(Thr172) phosphorylation. α1- and α2-AMPK were immunopurified from muscle lysates, and activities were measured in an in vitro kinase assay. In muscle from the trained leg, activities of α1-and α2-AMPK increased by 94 ± 22 and 49 ± 22%, respectively, (both P < 0.005), whereas no effects of training were observed in the untrained muscle (Fig. 3). In accordance, phosphorylation of α-AMPK (Thr172) increased by 74 ± 10% (P < 0.001) in the trained muscle, with no change in the untrained muscle (Fig. 4A).
ACCβ(Ser221) phosphorylation and ACCβ protein content. ACCβ(Ser221) is a known target for AMPK in human skeletal muscle. The degree of ACCβ(Ser221) phosphorylation is therefore thought to be a marker for AMPK activity in vivo. In muscle of the trained leg, phosphorylation of ACCβ(Ser221) increased by 180 ± 56% (P < 0.001) in response to training, with no change in the untrained muscle (Fig. 4B). In six of eight subjects, total protein content of ACCβ was also higher in trained muscle compared with muscle before training (Fig. 4C), with an average increase of 74 ± 33% (not significant). When ACCβ(Ser221) phosphorylation was expressed in relation to ACCβ protein content, a less pronounced but still highly significant increase in ACCβ(Ser221) phosphorylation was observed in muscle of the trained leg (88 ± 36%, P < 0.01; Fig. 4D).
After 3 wk of one-legged knee extensor endurance training, in vitro AMPK activity associated with both the α1 and α2 catalytic subunit isoforms increased significantly in the trained muscle. In line with this observation, α-AMPK(Thr172) and ACCβ(Ser221) phosphorylation was significantly elevated after training in the trained muscle. The increase in basal AMPK activity coincided with alterations in protein content of several AMPK subunit isoforms. In the trained muscle, protein content of α1, β2 and γ1 increased significantly with training, whereas in contrast protein content of γ3 markedly decreased. Except for the increase in γ1 protein, all observed adaptations to training could be ascribed to local contraction-induced mechanisms, since they did not occur in the contralateral untrained muscle.
When AMPK activity is measured in vitro in cell extracts, only changes that are induced by covalent modification are detected. Exercise training increased resting AMPK activity associated with both the α1 and α2 catalytic subunit isoforms in trained muscle. This is in accordance with the elevated α-AMPK(Thr172) phosphorylation, Thr172 being the principal site of covalent regulation of AMPK activity (12). Changes in protein content might provide a mechanism explaining increased α1-but not α2-AMPK activity in trained muscle, due to increased substrate availability for upstream regulators. However, when α1- and α2-AMPK activities are expressed relative to protein content for each individual α-isoform, in fact the change in response to training is of comparable magnitude (41 ± 16 and 58 ± 26%, respectively). This suggests 1) that the activity of AMPKK or of upstream phosphatases is being regulated by training or 2) that the sensitivity of AMPK to upstream regulation is altered, possibly due to changes in subcellular localization or isoform composition.
Increased phosphorylation of ACCβ(Ser221) was also observed in trained muscle, when expressed both independently and relative to total ACCβ protein levels. Because inactivation of ACCβ via Ser221 phosphorylation is a known response to AMPK activation, a plausible interpretation is that endurance training increases basal AMPK activity in vivo. However, phosphorylation of ACCβ(Ser221) is likely to be regulated by several mechanisms (34), and the interpretation should be viewed with this in mind.
This report is the first to demonstrate that the basal activity of AMPK in human skeletal muscle increases in response to endurance training. Because biopsies were obtained 15 h after the last exercise bout and because muscle glycogen content was restored at this time point (Table 2), the acute effect of exercise on AMPK activity is expected to be reversed (19, 38, 39). These data therefore probably reflect a genuine adaptation to exercise training. To further investigate the time period at which AMPK activity is elevated, a control study including four subjects was conducted. Having performed an identical training regimen, the subjects were examined 55 h (rather than 15 h) after the last exercise bout. Essentially similar responses to training on α-AMPK(Thr172) and ACCβ(Ser221) phosphorylation (+71 and +114%, respectively), α1- and α2-AMPK activity (+67 and +46%, respectively), and AMPK subunit expression (α1 +50, β2 +20, γ1 +23, and γ3 -63%) were observed in this group. Furthermore, the increased AMPK activity was observed despite increased glycogen content (+31%) in the trained muscle at this time point. Together, these observations support the notion that increased basal activity of AMPK after endurance training is a genuine effect of training, as opposed to being a residual effect of the last exercise bout. Furthermore, this effect of training appears to be quite long lasting (>55 h), indicating that AMPK activity is probably chronically elevated in skeletal muscle during periods of regularly repeated endurance exercise.
AMPK is considered a promising candidate to mediate several of the changes in protein expression observed in skeletal muscle in response to endurance training (15, 20, 29, 36). Furthermore, several lines of correlative evidence indicate that activation of AMPK results in improved insulin sensitivity in skeletal muscle. Thus two widely used antidiabetic agents, rosiglitazone and metformin, apart from increasing insulin sensitivity, also activate AMPK in cultured cells (9, 13) and skeletal muscle (9, 44), respectively. Similarly, chronic treatment of rodents with leptin (31, 42) or AICAR (3, 16) increases skeletal muscle sensitivity of insulin to stimulate glucose uptake, concomitant with activating AMPK. After intense bicycle exercise for 60 min, AMPK activity reverses within ∼3 h (39), whereas insulin sensitivity is expected to be improved at this time point (28, 37). This would indicate that AMPK activity is not increased at the time of improved insulin sensitivity after acute exercise. However, this does not rule out an effect of AMPK activation on insulin sensitivity in skeletal muscle; hence, insulin sensitivity might be continuously improved if AMPK activation did not reverse quickly. Data from the present study indicate that AMPK is not only repeatedly activated during periods of endurance training but is also chronically active in the period of recovery. Considering the aforementioned, this observation might be important for understanding why the effect of endurance training on insulin sensitivity extends beyond the acute effect of the last exercise bout.
In the present study, increased protein expression of α1, β2, and γ1 was observed, together with a decrease in γ3 protein content. Increases in expression of α1 protein in vivo have previously been observed in several situations in humans and animals. These involve our cross-sectional (24) and a longitudinal study (17) of training in humans, skeletal muscle of mice in which the α2-subunit of AMPK was knocked out (35), rat cardiac muscle during pressure overload hypertrophy (33), and rat skeletal muscle after 3 wk of treatment with thyroid hormones (25). Taken together, these results indicate that α1 protein expression may be upregulated under conditions where there is a repeated or persistent activation of AMPK due to contractions or other metabolic stress.
In contrast to the present study, no significant differences in muscle protein content of any α- or β-isoforms were observed when trained rats were compared with untrained, although there did appear to be a small increase in α1 protein (8). In addition, in that study, a marked (3-fold) increase in γ3 protein was observed in oxidative fast-twitch fibers. In the present study, a single polypeptide of the expected molecular mass was detected when CCL-13 lysate-containing recombinant human γ3 protein was immunodetected; thus we feel confident that the antibody used targets the correct protein. Although a difference in endurance training stimulus may be a contributing factor to this apparent discrepancy, another explanation could be a species difference in the adaptive response to training. Supporting the present findings in humans is our recent observation (24) that γ3 mRNA is reduced by ∼40% in trained compared with untrained muscle. Also, a similar adaptive response in AMPK protein expression has recently been observed after strength training in elderly subjects. (Wojtaszewski JFP, Birk JB, Frøsig C, Holten M, and Dela F, unpublished observations). Furthermore, in that study, no difference was observed in those isoforms measured (α1, α2, and γ3) when values obtained from whole muscle homogenates were compared with values obtained from muscle lysates. This helps eliminate the concern that changes in AMPK protein with training is caused by translocation of protein from/to the pellet discarded during preparation of lysates.
In most rat tissues, including skeletal muscle, γ3-AMPK complexes contribute with only a small portion (<5%) of the total AMPK activity (5), indicating that the γ3-isoform is less abundant than the other γ-isoforms. An ∼25% increase in γ1 protein content might therefore more than compensate for a ∼60% decrease in γ3 protein, resulting in more total γ protein after training. We therefore propose that training increases total AMPK complex formation in skeletal muscle due to increased protein expression of catalytic as well as regulatory subunits. Furthermore, considering that the observed changes in AMPK subunit expression are highly isoform specific, it seems likely that the functional properties (e.g., subcellular localization, sensitivity toward AMP) of the total population of AMPK complexes are also altered in response to training (5, 29).
In conclusion, the basal AMPK activity increases with endurance training in human skeletal muscle, at least partly due to increased covalent regulation. This effect of training appears to persist for 55 h after the last exercise bout, indicating that AMPK activity is chronically increased during periods of regular, repeated exercise. Protein expression of both catalytic and regulatory subunits of AMPK is also highly susceptible to training. The functional implications of these observations await further investigation, but they might be important for the adaptive response to exercise training with regard to regulation of both metabolism and gene expression. Except for the increase in γ1 protein, all observed adaptations to training could be ascribed to local contraction-induced mechanisms.
The study was supported by grants from the Danish National Research Foundation (no. 504-14), by a Research and Technological Development Projects (QLG1-CT-2001-01488) grant funded by the European Commission, by The Media and Grants Secretariat of the Danish Ministry of Culture, by the Danish Diabetes Association, by the Novo Nordisk Foundation, and by the Danish Medical Research Council. D. G. Hardie was supported by a Programme Grant from Wellcome Trust. J. F. P. Wojtaszewski was supported by a Hallas Møller fellowship from the Novo Nordisk Foundation.
We thank Professor David Carling for the kind donation of recombinant human γ3 protein and Kirsty Mustard for stimulating and helpful scientific discussions. Betina Bolmgren and Jesper Bratz Birk are acknowledged for skilled technical assistance.
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.
- Copyright © 2004 by American Physiological Society