AMP-activated protein kinase (AMPK) has been extensively studied in whole muscle biopsy samples of humans, yet the fiber type-specific expression and/or activation of AMPK is unknown. We examined basal and exercise AMPK-α Thr172 phosphorylation and AMPK subunit expression (α1, α2, and γ3) in type I, IIa, and IIx fibers of human skeletal muscle before and after 10 days of exercise training. Before training basal AMPK phosphorylation was greatest in type IIa fibers (P < 0.05 vs. type I and IIx), while an acute bout of exercise increased AMPK phosphorylation in all fibers (P < 0.05), with the greatest increase occurring in type IIx fibers. Exercise training significantly increased basal AMPK phosphorylation in all fibers, and the exercise-induced increases were uniformly suppressed compared with pretraining exercise. Expression of AMPK-α1 and -α2 was similar between fibers and was not altered by exercise training. However, AMPK-γ3 was differentially expressed in skeletal muscle fibers (type IIx > type IIa > type I), irrespective of training status. Thus skeletal muscle AMPK phosphorylation and AMPK expression are fiber type specific in humans in the basal state, as well as during exercise. Our findings reveal fiber type-specific differences that have been masked in previous studies examining mixed muscle samples.
- muscle fiber
the enzyme amp-activated protein kinase (AMPK) is an αβγ heterotrimer ubiquitously expressed in skeletal muscle of a number of species (8, 10, 24) and has been postulated as a potential therapeutic target for the treatment of type 2 diabetes (22, 36). This hypothesis is primarily driven by the finding that physiological activation of AMPK (e.g., moderate intensity exercise), as well as pharmacological activation of AMPK via agents such as metformin and rosiglitazone, are both associated with increased metabolic flux (i.e., glucose uptake and fatty acid oxidation), increased GLUT-4 and hexokinase II levels, and increased markers of mitochondrial biogenesis in skeletal muscle (5, 15, 17, 21, 23, 37). To date, human studies assessing the role of AMPK in physiological processes have examined mixed muscle biopsy samples from the vastus lateralis muscle. The vastus lateralis muscle contains a mixture of slow oxidative (type I), fast oxidative (type IIa), and fast glycolytic (type IIx) fibers [∼47%, 37%, and 16%, respectively in untrained individuals (25)], all of which differ in regard to their contractile and metabolic capabilities (9, 32). Thus a caveat of human studies examining AMPK within mixed muscle biopsy samples is that one cannot determine whether AMPK activation is fiber type specific. Indeed, it is well known that moderate-intensity exercise as well as prolonged low-intensity exercise preferentially activates AMPK-α2 in human vastus lateralis muscle (5, 39), yet the fiber type-specific activation of AMPK-α2, if any, is unknown.
In rodents, one can indirectly examine AMPK activation within different fibers because some skeletal muscle groups (e.g., soleus, gastrocnemius, red and white quadriceps) have greatly differing proportions of type I, IIa, IIb, and IId/x fibers (2, 7). In rats, acute exercise increases total AMPK-α activity to the greatest extent in red quadriceps (contains primarily type IIa and IId/x fibers) and to a lesser extent in soleus (primarily type I fibers) and shows no change in white quadriceps (purely type IIb fibers) (8, 29). Exercise also increases AMPK-α1 and -α2 activity in mouse red and white gastrocnemius muscle (15). In humans, prolonged submaximal exercise results in a greater energy deficit within type II compared with type I skeletal muscle fibers (31, 34). Given that cellular stress is a major activator of AMPK (11), it could be predicted that activation of AMPK would occur to the greatest extent in type II fibers during exercise, although no study has examined the fiber type-specific activation of AMPK during exercise in humans or any other species.
In mixed muscle biopsy sample analysis we recently found (20) that the ∼8- to 10-fold increase in AMPK activation during prolonged moderate-intensity exercise in untrained men was abolished after 10 days of exercise training. Durante et al. (8) examined the effects of 7 wk of treadmill training on AMPK activity in rat skeletal muscle and found that the exercise-induced increase in AMPK activity was greater in soleus muscle after training, yet there was no increase in red quadriceps and a decrease in white quadriceps AMPK activity after training. This suggests a possible fiber type-specific response in regard to AMPK activation during exercise performed after exercise training, although whether this occurs in human skeletal muscle is unknown.
Given that human skeletal muscle fiber type distribution is heterogeneous, and our understanding of fiber type-specific AMPK activation in human skeletal muscle is unknown, we determined whether the activation of AMPK-α was differentially regulated between type I, IIa, and IIx human skeletal muscle fibers at rest and during exercise performed before and after 10 days of exercise training. It was hypothesized that an acute bout of moderate-intensity exercise would increase AMPK-α activation (assessed via AMPK-α Thr172 phosphorylation) to the greatest extent within type II fibers. It was also hypothesized that AMPK activation would be suppressed within all fibers during posttraining exercise compared with pretraining exercise. Finally, since short-term exercise training increases basal AMPK activity in whole muscle of humans (10, 20), we hypothesized that basal AMPK activity would be increased within all skeletal muscle fibers after short-term exercise training.
MATERIALS AND METHODS
Six healthy, nonsmoking men provided informed written consent to participate in this study, which was approved by the Monash University Standing Committee on Ethics in Research involving Humans and the Human Research Ethics Committee of the University of Melbourne. Participants were all considered to be aerobically untrained [i.e., peak oxygen consumption (V̇o2peak) ≤ 45 mg·kg−1·min−1; Table 1]. Three of these participants were a subset from a previous study that examined the effect of a low- or high-carbohydrate (CHO) diet on AMPK signaling during posttraining exercise (17). The remaining three participants performed an protocol identical to that described previously (17). The participants' diets differed in terms of CHO (n = 4 for low CHO; n = 2 for high CHO), although energy intake was similar before each experimental trial as determined with Foodworks Professional 2003 (Xyris Software). We showed previously (17) that dietary manipulation, in terms of either a high- or a low-CHO diet, does not affect whole muscle AMPK signaling at rest and during exercise either before or after 10 days of exercise training.
The experimental design for this study has been previously described (20). Briefly, 5–7 days after a V̇o2peak test, participants completed the first experimental trial (pretraining), which involved cycling at 115 ± 12 W for a maximum of 120 min (115 ± 5 min) at 65 ± 2% of pretraining V̇o2peak. Expired air was collected into Douglas bags to measure oxygen consumption (V̇o2) and carbon dioxide production (V̇co2), and whole body rates of CHO and fat oxidation were calculated with nonprotein respiratory exchange ratio (RER) (27). Muscle biopsies were performed at rest, after 30 min of exercise, and at the end of exercise, and samples were frozen in liquid nitrogen within 4–6 s of insertion of the biopsy needle at rest and within 8–12 s of the participant stopping exercise. A standard 60-s period was allowed for completion of the biopsy and taping of the area before recommencing exercise after the 30-min muscle sample was obtained. After the pretraining exercise trial, participants performed 10 days of exercise training over a 2-wk period, which involved 1 day of cycling for 45–90 min at 75% V̇o2peak (150 ± 18 W) alternated with 1 day of interval training (6 × 5-min work bouts) at 90–100% V̇o2peak (185 ± 22 W). Twenty-four to forty-eight hours after the 10th training session, participants performed a posttraining exercise trial, which involved cycling at the same absolute workload as before training, equivalent to 58 ± 2% of posttraining V̇o2peak.
Triple-labeled immunohistochemistry, using a modified method of Russell et al. (30), was used to gain quantitative data with respect to the phosphorylation and expression of AMPK within type I, IIa, and IIx fibers of human skeletal muscle as previously described (19). Briefly, pre- and posttraining muscle sections were incubated overnight at 4°C in a solution of 50% preimmune sheep serum (SS) and 3% bovine serum albumin (BSA) in phosphate-buffered saline (PBS). Type I and IIa fibers were detected with monoclonal antibodies (A4.840, mouse IgM and N2.261, mouse IgGγ1, respectively) obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the National Institute of Child Health and Human Development and maintained by the Department of Biological Sciences, University of Iowa (Iowa City, IA) (14, 35).
In conjunction with the A4.840 and N2.261 monoclonal antibodies, one of the following affinity-purified polyclonal rabbit IgG antibodies was added: AMPK-α1 (1:50 dilution; amino acid sequence 373–390 of rat AMPK-α1), AMPK-α2 (1:200; 490–516 of rat AMPK-α2), AMPK-γ3 (1:50; RWTRQKSVEEGEPPGQG of human AMPK-γ3). In rodents, positive correlations between AMPK-α Thr172 phosphorylation and AMPK-α activity have been observed in skeletal muscle (26), and thus a phospho-AMPK-α Thr172 antibody (1:50; KDGEFLRT172SCGSPNY of rat AMPK-α) was also used. The dilutions used for each antibody were within the predetermined linear range. Negative sections were incubated in 10% SS and 1% BSA in PBS, in the absence of primary antibodies. The primary antibodies were visualized with goat anti-rabbit IgG secondary antibody conjugated with Alexa Fluor 488 (1:2,000; Molecular Probes, Eugene, OR), goat anti-mouse IgGγ1 conjugated with Alexa Fluor 350 (1:1,000; Molecular Probes), or goat anti-mouse IgM conjugated with Texas red (1:1,000; Southern Biotechnology Associates, Birmingham, AL). A total of 314 ± 71 type I fibers, 329 ± 79 type IIa fibers, and 50 ± 13 type IIx fibers were counted for each protein, with type I, IIa, and IIx fibers constituting 42 ± 3%, 45 ± 3%, and 13 ± 1% of the total, respectively. Fluorescence staining in all images was quantified by semiquantitative densitometric analysis using Imaging Research software (MCID/AIS). Specifically, regions of interest were defined, and the relative fluorescence was expressed in arbitrary units as PSL per square millimeter (pixel intensity as a measure of density per unit squared).
In preliminary experiments, preincubation of primary antibodies with their corresponding peptide resulted in no detectable labeling of human skeletal muscle sections (see Supplemental Fig. S1), demonstrating that the primary antibodies used were not binding nonspecifically to the muscle sections.1 Incubation of rabbit IgG (Santa Cruz Biotechnology, Santa Cruz, CA) in concentrations identical to that of each primary rabbit IgG antibody did not result in preferential labeling of type I, IIa, or IIx fibers (see Supplemental Fig. S2), demonstrating that any fiber type differences in AMPK expression and/or phosphorylation were not due to rabbit IgG per se. Finally, the primary antibodies used specifically targeted the protein of interest, as assessed by Western blotting of whole cell lysates (see Supplemental Fig. S3). We did not assess AMPK-β1, -β2, -γ1, or -γ2 expression in skeletal muscle fibers, because testing of these antibodies revealed nonspecific binding.
Whole muscle protein expression.
Whole muscle homogenates (40 μg) were subjected to SDS-PAGE. AMPK-α1, -α2, and -γ3 protein expression was detected with the same antibodies described above for immunohistochemistry, while β-actin was detected with a rabbit polyclonal antibody (Abcam, Cambridge, UK). Binding of these antibodies was detected with an anti-rabbit IRDye 800-labeled secondary antibody (LICOR Biosciences, Lincoln, NE). Fluorescence was detected and quantified with the Odyssey infrared imaging system (LICOR Biosciences), and protein expression was normalized to β-actin.
A paired student t-test was used to assess whole muscle protein expression before compared with after exercise training. When comparing basal protein expression within type I, IIa, and IIx fibers before and after training, a two-factor (trial × fiber type) repeated-measures ANOVA was performed with the statistical package SPSS. A three-factor (trial × time × fiber type) repeated-measures ANOVA was used to compare AMPK-α Thr172 phosphorylation in fibers over time before and after training. If the ANOVA was significant (P < 0.05), specific differences were located with Fisher's least significant difference (LSD) test.
Exercise training significantly increased V̇o2peak (in l/min and ml·kg−1·min−1) by ∼9% (P < 0.05; Table 1). V̇o2 averaged 65 ± 2% of pretraining V̇o2peak during exercise and was significantly reduced during posttraining exercise (62 ± 2% of pretraining V̇o2peak; P < 0.05). Given that participants' V̇o2peak was greater after training, the posttraining exercise was equivalent to 59 ± 3% of posttraining V̇o2peak (P < 0.001 vs. pretraining). RER and CHO oxidation were both significantly reduced during posttraining exercise (P < 0.05), whereas fat oxidation was significantly elevated (P < 0.05) (Table 1).
AMPK subunit expression.
Figure 1 shows representative immunohistochemistry images for AMPK-α1 (Fig. 1A), AMPK-α2 (Fig. 1B), and AMPK-γ3 (Fig. 1C) subunit expression in type I, type IIa, and IIx fibers before and after short-term exercise training. AMPK-α1 and -α2 subunit expression was not significantly different among type I, IIa, or IIx fibers either before or after exercise training (Fig. 2, A and B), findings that were also observed in whole muscle samples (Fig. 3). In contrast, AMPK-γ3 subunit expression varied significantly among fibers, with an ordered expression of type IIx > type IIa > type I (P < 0.01; Fig. 1C and Fig. 2C). The pattern of AMPK-γ3 subunit expression after exercise training was similar to that observed before training (i.e., type IIx > type IIa > type I), and exercise training did not alter the abundance of AMPK-γ3 either within fibers (P = 0.2) (Fig. 2C) or within whole muscle (Fig. 3). Exercise training did not alter whole muscle β-actin levels (data not shown).
AMPK-α Thr172 phosphorylation.
Before exercise training, basal AMPK-α Thr172 phosphorylation was greatest in type IIa fibers (Fig. 2D). Exercise performed before training resulted in a significant increase in AMPK-α Thr172 phosphorylation in all fibers (Fig. 4). However, the greatest increase in AMPK-α Thr172 phosphorylation was observed within type IIx fibers (Fig. 4B). Short-term exercise training increased basal AMPK-α Thr172 phosphorylation in all fibers by an average of 50 ± 10% (P < 0.01 vs. pretraining; Fig. 2D). During posttraining exercise, AMPK-α Thr172 phosphorylation significantly increased from rest after 30 min in all fiber types (Fig. 4B). By the end of exercise AMPK-α Thr172 phosphorylation had returned to resting levels in type I and IIa fibers, but not type IIx fibers (Fig. 4B). When compared with pretraining exercise, the alterations in AMPK-α Thr172 phosphorylation during posttraining exercise were significantly attenuated in all fibers (Fig. 4B).
AMPK has been described as a regulator of skeletal muscle metabolism during exercise; however, skeletal muscle is not homogeneous but rather comprised of different fiber types. Given that type I and II fibers are recruited to differing degrees during submaximal exercise in humans (31), we examined AMPK activation in type I, IIa, and IIx fibers of human skeletal muscle at rest and during exercise performed before and after 10 days of exercise training. We show that basal AMPK-α Thr172 phosphorylation, and thus AMPK activation (21), is greatest within type IIa fibers, a phenomenon that is not altered by 10 days of exercise training in previously untrained individuals. Another important finding was that acute exercise increases AMPK phosphorylation in all skeletal muscle fiber types, with type IIx fibers eliciting the greatest increase. Furthermore, after short-term exercise training, basal AMPK phosphorylation is increased in all skeletal muscle fiber types, while there is less of an increase in AMPK phosphorylation during posttraining exercise in all fiber types. These findings extend results derived from mixed muscle biopsy samples by highlighting novel fiber type-specific differences in AMPK phosphorylation in human skeletal muscle.
Previously, we found (20), using the same muscle samples, that 10 days of exercise training suppressed whole muscle AMPK-α1 and -α2 activation, as well as AMPK-α2 Thr172 phosphorylation during exercise, and it is now apparent that this observation was due to suppressed AMPK activation within all muscle fiber types (see Fig. 4). The most likely reason for this finding is the tighter skeletal muscle metabolic control that we observed in the whole mixed muscle samples during exercise performed after 10 days of exercise training (20). We are unaware of any study that has examined metabolic changes within specific muscle fibers after exercise training. It is also possible that the lower relative exercise intensity after training (∼58% V̇o2peak vs. ∼65% V̇o2peak during pretraining exercise) contributed to the suppressed AMPK activation. However, we have previously observed (5) an increase in AMPK-α2 and -α1 activity in human skeletal muscle during an acute bout of exercise at ∼59% V̇o2peak.
The reason(s) accounting for the greater increase in AMPK-α Thr172 phosphorylation within type IIx fibers than type I and IIa fibers during exercise is unclear but may relate to AMPK-γ3 expression. In human skeletal muscle ∼60% of all phosphorylated AMPK-α is associated with AMPK-γ3 during submaximal exercise (3), and in the present study we show that AMPK-γ3, like AMPK-α Thr172 phosphorylation, is most abundant within type IIx fibers both before and after exercise training (see Figs. 1C and 2C). Rodent studies support our findings, as it has been shown previously that rodent muscle groups expressing the greatest amount of AMPK-γ3 also have augmented AMPK activation in response to contraction (8, 18). In rat brain extracts the AMPK-γ3 subunit has a low AMP dependence (6), and it is thus intriguing that only AMPK α2/γ3 complex activities are increased in response to submaximal cycling exercise in human skeletal muscle (3). This would suggest that AMPK α2/β2/γ3 complexes are more sensitive to AMP (unlike rat brain), or more sensitive to upstream regulators compared with α2/β2/γ1 complexes.
An interesting observation from the present study was the preferential localization of the AMPK-α1 and -α2 subunits to the subsarcolemmal region (see Fig. 1, A and B). Findings from rat skeletal muscle support our observation, because AMPK-α1 and -α2 expression also appears to be greater at the subsarcolemmal region in individual skeletal muscle fibers (2, 28). From a protein-protein perspective, localization of AMPK within the vicinity of the plasma membrane would be advantageous in that it would facilitate the phosphorylation of the skeletal muscle isoform of neuronal nitric oxide synthase (nNOSμ), a downstream target of AMPK (4). Indeed, we have shown that nNOSμ protein expression is elevated in all skeletal muscle fiber types after 10 days of exercise training (19), as was observed for AMPK-α Thr172 phosphorylation in the present study. Likewise, it has been shown that acetyl-CoA carboxylase (ACC)β, another downstream target of AMPK (4), is localized to the mitochondria (1). Thus it is also possible that AMPK localized to the vicinity of the plasma membrane facilitates interactions with ACCβ located within subsarcolemmal mitochondria.
In humans, exercise training for 10–15 days has been shown to increase basal AMPK-α1 activity and total Thr172 phosphorylation in skeletal muscle (10, 20). Here we show that this increase is due to a uniform increase in AMPK-α Thr172 phosphorylation within all fiber types (see Fig. 2D). It is not possible to state whether this increase was due to increases in AMPK-α1 and/or AMPK-α2 activity because the amino acid sequence surrounding AMPK-α Thr172 is conserved between these subunits (12). AMPK phosphorylates several enzymes in skeletal muscle (4, 11) including nNOSμ (2). Interestingly, we have shown (19) that, like basal AMPK-α Thr172 phosphorylation, nNOSμ protein expression is also increased in type I, IIa, and IIx fibers of individuals after 10 days of exercise training. This supports our present data regarding AMPK-α Thr172 phosphorylation.
The present study also demonstrated that 10 days of exercise training does not alter the protein expression of AMPK-α1 and -α2 either within whole muscle (see Fig. 3) or within different muscle fiber types (see Figs. 1 and 2). Consistent with this, 3–6 wk of single-legged exercise training in humans does not alter AMPK-α2 protein expression; however, it does significantly increase AMPK-α1 protein expression in the trained leg (10, 16, 38). This suggests that training periods >10 days are required to elicit increases in AMPK-α1 protein expression, which is consistent with the finding of elevated AMPK-α1 protein levels in well-trained compared with untrained individuals (24).
We found that there were differing levels of skeletal muscle AMPK-γ3 protein expression between muscle fiber types in untrained individuals. Ten days of exercise training tended to alter AMPK-γ3 expression (∼10% lower), but this was not statistically significant. Previously, it has been shown that both single-legged short-term endurance training (3 wk) and short-term strength training (6 wk) decrease AMPK-γ3 expression in mixed muscle samples of humans (10, 38). Therefore, had the exercise training regimen employed in the present study been extended to >10 days, it is possible that we also would have observed a significant decrease in AMPK-γ3 expression within muscle fibers. However, AMPK-γ3 expression does not differ in mixed muscle samples of well-trained versus untrained individuals (24), while 7 wk of endurance training in rodents increases AMPK-γ3 expression in red quadriceps muscle (8).
It is clear from the present findings that potential fiber type-specific responses need to be considered in human studies. It has recently been shown that individuals who are obese, or who are obese with type 2 diabetes, have suppressed activation of AMPK during moderate-intensity exercise (33). Interestingly, there is evidence to suggest that skeletal muscle fiber type composition is altered in these individuals (13), although whether AMPK activation is also altered within different fibers during exercise is unknown. Likewise, exercise represents just one known mechanism that activates AMPK, and as such it will also be important to determine whether other mechanisms that activate skeletal muscle AMPK, such as metformin (23) and rosiglitazone (17), also activate AMPK in a fiber type-specific manner. Finally, Birk and Wojtaszewski (3) have demonstrated that only 3 of 12 possible different AMPK heterotrimers exist in mixed muscle samples of humans. Thus, while technically difficult, it will be important for future studies to examine the expression of AMPK heterotrimers in type I, IIa, and IIx fibers of skeletal muscle. Understanding the fiber type-specific expression and/or activation of AMPK will assist in pharmacological targeting of AMPK in human skeletal muscle.
In conclusion, we have shown that there are fiber type-specific differences in AMPK activation within human skeletal muscle at rest, during an acute bout of exercise, and after short-term exercise training. AMPK-γ3 is also expressed in a fiber type-specific fashion, which is not altered by short-term exercise training. Thus immunohistochemistry is a valuable tool to elucidate fiber type-specific differences in enzyme expression and activation that would otherwise be masked in studies examining whole muscle.
This work was supported by a grant from the National Health and Medical Research Council (NHMRC) of Australia (no. 237002) and Diabetes Australia (G. K. McConell).
The authors thank the participants for taking part in this study, Dr. Aaron Russell for technical advice, Prof. Bruce Kemp for the AMPK antibodies and peptides, and Prof. David Wasserman for helpful discussions and constructive criticism during the preparation of this manuscript.
Present address of D. E. Myers: Dept. of Medicine, Royal Melbourne Hospital, University of Melbourne, Parkville, VIC, Australia.
↵1 The online version of this article contains supplemental material.
- Copyright © 2009 the American Physiological Society