Physical exercise has potent therapeutic and preventive effects against metabolic disorders. A number of studies have suggested that 5′-AMP-activated protein kinase (AMPK) plays a pivotal role in regulating carbohydrate and lipid metabolism in contracting skeletal muscles, while several genetically manipulated animal models revealed the significance of AMPK-independent pathways. To elucidate significance of AMPK and AMPK-independent signals in contracting skeletal muscles, we conducted a metabolomic analysis that compared the metabolic effects of 5-aminoimidazole-4-carboxamide-1-β-d-ribonucleoside (AICAR) stimulation with the electrical contraction ex vivo in isolated rat epitrochlearis muscles, in which both α1- and α2-isoforms of AMPK and glucose uptake were equally activated. The metabolomic analysis using capillary electrophoresis time-of-flight mass spectrometry detected 184 peaks and successfully annotated 132 small molecules. AICAR stimulation exhibited high similarity to the electrical contraction in overall metabolites. Principal component analysis (PCA) demonstrated that the major principal component characterized common effects whereas the minor principal component distinguished the difference. PCA and a factor analysis suggested a substantial change in redox status as a result of AMPK activation. We also found a decrease in reduced glutathione levels in both AICAR-stimulated and contracting muscles. The muscle contraction-evoked influences related to the metabolism of amino acids, in particular, aspartate, alanine, or lysine, are supposed to be independent of AMPK activation. Our results substantiate the significance of AMPK activation in contracting skeletal muscles and provide novel evidence that AICAR stimulation closely mimics the metabolomic changes in the contracting skeletal muscles.
- 5′-AMP-activated protein kinase
- muscle contraction
- isolated skeletal muscle
the number of patients exhibiting symptoms of metabolic disorders is pandemically increasing nowadays (1, 2, 21, 46). Endurance exercise along with diet therapy has been known to improve and prevent such metabolic diseases as diabetes, obesity, and hyperlipidemia, and therapeutic exercise has been the authorized first-line therapy by official guidelines (3, 7, 23, 43, 56). Thus physical exercise is important to ameliorate carbohydrate and lipid metabolism in the treatment of metabolic disorders preventing such complications as cardiovascular diseases.
A series of studies from us and others have provided much evidence that 5′-AMP-activated protein kinase (AMPK) plays a significant role in metabolic changes in contracting skeletal muscles including enhancement of glucose uptake activities (6, 17–19, 22, 26, 31, 34, 35, 37, 47, 48, 51, 57). Recent studies have established concepts of AMPK as a “fuel gauge” to monitor the status of intracellular energy and a key enzyme in regulating metabolism. AMPK is allosterically activated in response to the elevation in AMP-to-ATP ratio, and phosphorylation of threonine 172 residue of the α-subunit by upstream kinases such as LKB1 or calcium/calmodulin-dependent kinase kinase-β increases the activity of AMPK (5, 10, 12, 14–16, 45, 52). The AMPK activation leads to a wide range of metabolic responses including glucose uptake, fatty acid oxidation, and insulin sensitization, etc.; thus it has been attracting attention as a potential therapeutic target molecule. (17, 19, 32–34, 37, 49, 54). Furthermore, AMPK has been shown to be an important mediator of metabolic actions of some kinds of hormones and drugs including leptin, adiponectin, metformin, and so on (32, 33, 36, 54, 58).
On the other hand, several studies have demonstrated that AMPK activation is not necessary for metabolic changing in contracting skeletal muscles. We indeed found some discrepancies in glycogen metabolism between AICAR-induced AMPK activation and electrical muscle contraction (34). In addition, recent studies using genetically manipulated animals have suggested the significance of AMPK-independent pathways in contraction-evoked glucose metabolism. While exercise tolerance and glucose clearance during exercise are impaired in striated muscle-specific transgenic mice expressing dominant negative AMPK-α2 in vivo, muscle contraction-induced glucose uptake activities are not or only partially inhibited in skeletal muscles from genetically AMPK-impaired mice (11, 24, 27, 35). Since AMPK is an important regulatory molecule, it may be possible that genetically manipulated animals acquire compensatory mechanisms during the process of growth. Therefore, a different approach from using genetic model animals should lead to further investigating metabolic significance of AMPK in contracting skeletal muscles.
Furthermore, recent rapid advance in analytical technologies including capillary electrophoresis and mass spectrometry has enabled us more easily to obtain comprehensive information on small metabolites or “metabolome” as it is called. This new metabolomic approach should become a powerful tool in investigating the whole effects of multipotent stimuli such as muscle contraction. However, there have been almost no reports demonstrating changes in the metabolic status within contracting skeletal muscles using metabolomic methods.
Therefore, here we performed comparative metabolome analysis using a popular AMPK activator, 5-aminoimidazole-4-carboxamide-1-β-d-ribonucleoside (AICAR), and electrical contraction ex vivo in isolated rat epitrochlearis muscles to elucidate the comprehensive metabolic effects in contracting skeletal muscles and the dependence and independence of AMPK there.
MATERIALS AND METHODS
Materials and reagents.
All reagents were analytic grade and obtained from Wako Pure Chemical (Osaka, Japan), Tokyo Chemical Industry (Tokyo, Japan), or Sigma-Aldrich (St. Louis, MO) unless otherwise stated. All radioactive materials were obtained from Perkin Elmer Japan (Yokohama, Japan). Protein A-Sepharose and p81 paper were from GE Healthcare Biosciences (Little Chalfont, UK).
Male Sprague-Dawley rats (120–130 g, Japan SLC, Hamamatsu, Japan) were housed in an animal facility maintained at 23°C with a 12-h light-dark cycle and fed standard laboratory chow and water ad libitum. The rats were randomly assigned to experimental groups. All the experimental procedures were approved by Animal Research Committees of the Kyoto University or the University of Tokushima.
C2C12 myoblast cells were obtained from American Type Culture Collection. The cells were grown in DMEM containing 10% fetal bovine serum in a humidified atmosphere with 5% CO2. The myoblasts were induced to differentiate to myotube by 4-day culture in DMEM containing 2% horse serum. The cells were stimulated in Krebs-Ringer bicarbonate (KRB) buffer containing 2 mM sodium pyruvate (KRB-P) with 0.1% bovine serum albumin and lysed in 5% meta-phosphoric acid for determining reduced glutathione levels.
Skeletal muscle sample preparations: pharmacological stimulation by AICAR and ex vivo electrical contraction of isolated epitrochlearis muscles.
Skeletal muscle samples were prepared as previously described (18, 30, 34). In brief, epitrochlearis muscles were rapidly isolated from rats just after cervical dislocation and the muscles were preincubated for 40 min in KRB-P followed by treatment with 2 mM AICAR for 40 min or electrical stimulation to contract ex vivo in KRB-P (contraction) (1/min train rate, 10-s train duration, 100-Hz pulse rate, 0.1-ms pulse duration, 100 V, for 10 min). Control samples (none) were incubated in KRB-P for 40 min after the preincubation. The buffers were continuously gassed with 95% O2-5% CO2 and maintained at 37°C before and during the stimulation.
Isoform-specific activity of AMPK.
Isoform-specific activities of AMPK were determined as described previously (33, 34). In brief, frozen samples were homogenized in HEPES/Triton-based lysis buffer and then centrifuged. The supernatants were immunoprecipitated with protein A-Sepharose and isoform-specific antibodies raised against AMPK-α1 or -α2 (47). Kinase activities in the immune complex were detected by the phosphorylation of the SAMS peptide using [γ-32P]ATP.
3-O-methylglucose uptake activity.
3-O-methylglucose (3MG) uptake activity was determined as an index of glucose uptake activity as we have described elsewhere (30, 34). After the incubation period, the isolated skeletal muscles were incubated in KRB containing 1 mM 3MG, 1.5 μCi/ml [3H]3MG, 7 mM mannitol, and 0.3 μCi/ml [14C]mannitol at 30°C for 10 min. The muscles were digested in 1 M NaOH, which was followed by neutralization with HCl. The radioactivity in the aliquots of the samples was determined by a liquid scintillation counter (Aloka, Tokyo, Japan).
We analyzed metabolomic profiles of the samples using capillary electrophoresis time-of-flight mass spectrometry (CE-TOFMS; Agilent Technologies, Santa Clara, CA, and Human Metabolome Technologies, Yamagata, Japan; Ref. 44). The stimulated isolated rat epitrochlearis muscles (n = 4, each group) were homogenized in methanol, ultrafiltrated (UltrafreeMC cutoff; 5 kDa; Merck-Millipore) and then subjected to the CE-TOFMS in cation and anion modes. The samples were assayed by mixture due to the fact that the multivariant statistical analyses cannot deal with statistical errors of each sample. That is the methodological limitations of the analyses. The detected peak areas were calculated and quantified using internal controls. The peaks were annotated by m/z values and migration time according to an annotation database including amino acids, organic acids, nucleic acids, and carbohydrate phosphate (Human Metabolome Technologies).
Reduced glutathione levels.
Intracellular reduced glutathione levels were determined by colorimetric detection using a ready-to-use reaction mix solution containing 4-chloro-1-methyl-7-trifluoromethyl-quinolinium methylsulfate (OxisResearch, Burlingame, CA).
Comparison of the data on AMPK activities and 3MG uptake were evaluated by Tukey-Kramer's test. Comparison of the data on glutathione levels were evaluated by Student's test. Values were expressed as means ± SE. P < 0.05 was considered statistically significant.
Correlation analyses were performed on the logarithmically transformed values of fold change on each molecule by AICAR stimulation (x-axis) and muscle contraction (y-axis). The molecules involved in each metabolic pathway are listed on Table 1.
Principal component analysis (PCA) was performed using software, SampleStat ver3.13 (Human Metabolome Technologies). All the data were normalized averages to zero and standard deviations to one according to the standard procedures of PCA. The amount of molecules of which peaks were lower than detection levels was considered zero.
Factor analysis was performed by adopting a Varimax-rotation and a maximum likelihood method with one factor extracted. Quatimax rotation, another orthogonal rotation method, yielded the same results. All data were normalized as described above.
AICAR treatment and electrical contraction stimulates AMPK activities and glucose uptake in isolated rat epitrochlearis muscles.
To elucidate significance of AMPK-dependent and -independent metabolic changes in the following metabolomic analyses, we first determined experimental conditions under which AICAR stimulation and electrical contraction caused the same extent of AMPK activation. As we have reported previously, in isolated rat epitrochlearis muscles, 2 mM AICAR stimulation for 40 min and tetanic electrical stimulation stated in methods led to the same extent of activation in both AMPK-α1 and -α2 (Fig. 1, A and B) (34). In accordance, these stimulations caused the same level of increase in glucose uptake (Fig. 1C). Thus we proceeded to compare the metabolic profiles using the above experimental conditions.
Comparison of metabolomic profiles between AICAR-stimulated and electrical contracting rat epitrochlearis muscles.
We detected 99 peaks in the cation mode and 85 peaks in the anion mode, respectively, and successfully annotated 132 molecules from the metabolomic analysis using CE-TOFMS (listed in the Table 1). Changes in all the molecules were highly correlated between AICAR stimulation and electrical contraction as shown in Fig. 2A. We performed a subanalysis classified by each metabolic pathway to clarify which metabolic pathway is dependent on or independent of AMPK. The molecules related to glucose metabolism including glycolysis and tricarboxylic acid cycle (Fig. 2B), nicotinamide adenine dinucleotide (NAD) metabolism (Fig. 2C), nucleic acids metabolism (Fig. 2D), and ureic acid metabolism inclusive of amino acids such as glutamate, glutamine, histidine, proline, and arginine (Fig. 2E) also exhibited high correlation between the stimulations. As for other amino acids, molecules in the metabolic pathways related to glycine, serine, and cysteine were also significantly correlated (Fig. 2H), whereas the metabolites related to branched amino acids (Fig. 2F) as well as aspartate, alanine, lysine (Fig. 2G), and aryl amino acids (Fig. 2I) were not.
PCA on the metabolomic profile revealed high similarity between AICAR stimulation and electrical contraction.
PCA extracted primary and secondary principal components (PC1 and PC2) of which proportions were 63.8 and 36.2%, respectively. The principal component scores of each status were quiescent control; 12.47, 0.80, AICAR stimulation; −5.32, −8.53, and electrical contraction; −7.15, 7.73 (PC1, PC2, respectively; Fig. 3A). As apparently shown in Fig. 3A, we found that PC1 distinguished quiescent control from the other two stimulations, while PC2 characterized the difference between AICAR stimulation and electrical contraction. We listed the top 51 molecules that had PC1 factor loadings >97% of the highest one (carboxymethyllysine) and the top 22 molecules that had PC2 factor loadings >90% of the highest one (ADP; Fig. 3, B and C). Figure 3D visualizes all molecules and their changes in level, whereas Fig. 3, E and F, shows the molecules highly contributing to PC1 or PC2, respectively.
Factor analysis revealed a common metabolic factor between AICAR stimulation and electrical contraction.
We further investigated the metabolomic profiles by factor analysis to extract a common metabolic factor underlying the metabolome-wide modulation by AICAR stimulation and electrical contraction. One common factor was tried to be extracted from the normalized metabolomic profiles using a maximum likelihood method according to the standard procedure, and we successfully determined the primary common factor of which proportion is as high as 81.0%. Furthermore, we show the factor scores (Fig. 4A) and changes in the levels (Fig. 4B) of the top 25 molecules with factor loadings >90% of the highest one (glutathione).
AICAR stimulation leads to decrease in reduced glutathione levels.
We determined the effects of AICAR stimulation on reduced glutathione levels using cultured myotube cells. The reduced glutathione levels in AICAR-stimulated cells were ∼10% lower than those in the control cells (Fig. 5).
This is the first report demonstrating the metabolomic profiles on small molecules in contracting skeletal muscle at the level of isolated tissue ex vivo. We are also the first to elucidate the metabolomic high resemblance between AICAR stimulation and electrical contraction and distinguish the AMPK-independent metabolomic changes from AMPK-dependent ones.
The correlation analyses revealed that AICAR stimulation closely mimics electrical contraction in isolated rat epitrochlearis muscle in terms of metabolome (Fig. 2A). As clearly shown in Fig. 3A, the major PC1 distinguished both AICAR and electrical contraction stimulations from the quiescent control, while the minor PC2 characterized difference between AICAR stimulation and electrical contraction without being affected by potential common factors. The proportion of PC1 provided by PCA, a well-established nonbiased statistical analysis method for dealing with many parameters, was as high as 63.8%, substantiate the metabolomic similarity (Fig. 3, A and B). These results suggest quite high significance of AMPK activation on metabolic changes in contracting skeletal muscles. AICAR stimulation should qualify as a suitable mimetic of muscle contraction even though AICAR may not necessarily be specific to AMPK as we have demonstrated that AICAR-derived ZMP is a potent activator of glycogen phosphorylase (34).
Furthermore, the secondary principal component, PC2, of which values in AICAR-stimulated and contracting muscles were −8.53 and 7.73, respectively, in contrast to just 0.80 in quiescent status, well characterized the difference between effects by AICAR-stimulation and electrical stimulation (Fig. 3, A and C). Taken together, PC1 and PC2 should successfully represent AMPK-activated and AMPK-independent metabolic effects in contracting muscles. These principal components will provide us a novel way to evaluate potentials and a mode of actions of such compounds harboring similar metabolic effects to physical exercise and/or AMPK activators.
The factor analysis revealed reduced glutathione as a most potent molecule affecting the common factor between AICAR stimulation and electrical contraction (Fig. 4A). Of particular interest was that the reduced glutathione had the second largest impact on the PC1 (Fig. 3B). The reduced glutathione levels in AICAR-stimulated and contracting muscles were 0.24 and 0.16 times as much as of that in control samples, suggesting the common effects of oxidative stresses (Fig. 2A). It is common to read in reports that oxidative stress resulting from physical exercise is a potent activator of AMPK (41, 42, 47, 59). For example, we have demonstrated that hydrogen peroxide activate AMPK, preferably α1-isoform, with an increase in the oxidized glutathione (47). Supplementation of reduced glutathione inhibits AMPK phosphorylation, which is an index of the activity, evoked by antimycin A-induced oxidative stresses in embryo (53). However, to the contrary, the effects of AMPK activation on the redox status have not been clarified well. In general, AMPK activation, which may be occasionally caused by oxidative stresses as aforementioned, is believed to counteract the oxidative stresses. A yeast homolog of AMPK, Snf1, has been shown to protect yeast cells against oxidative stimuli (38). On the other hand, there seems no report showing that AMPK activation negatively regulates reduced glutathione except for the article describing inhibition of compensatory increase in reduced glutathione for tert-butylhydroxyquinone-induced decrease by metformin or AICAR stimulation (4). Therefore, we investigated whether AICAR-induced AMPK activation decreases reduced glutathione using another system, cultured myotube cells, and found that AICAR stimulation decreases the level (Fig. 5). The change in redox status is supposed to underlie the common metabolic effects as the PCA and the factor analysis indicated (Figs. 3B and 4A). We also believe that we are the first to report that AMPK activation leads redox status to be oxidative.
Much evidence demonstrating the significance of AMPK in metabolic effects of muscle contraction has been provided in these two decades (6, 17–19, 22, 26, 31, 34, 35, 37, 47, 48, 51, 57). Nevertheless, several studies using genetically manipulated animals demonstrated that AMPK activation is not necessarily required for metabolic effects in contracting skeletal muscles. Muscle contraction-induced glucose uptake activity is only partially inhibited in skeletal muscles from transgenic mice overexpressing kinase-dead AMPK, while AICAR-induced glucose uptake activity is completely inhibited in the transgenic mice (35). Fujii et al. (11) also observed partial inhibition of contraction-induced glucose uptake in extensor digitorum longus muscles from muscle-specific transgenic mice expressing inactive AMPK-α2, but they concluded that the glucose uptake activity is unaffected by the transgenic AMPK inhibition after adjustment of muscle force generation as muscle force is impaired in AMPK-α2-deficient mice. Muscle contraction-evoked glucose uptake activities in skeletal muscles from AMPK-α2-knockout mice are not lower than those in wild-type mice (24). However, these studies cannot fully exclude the possibility of influences by acquired compensatory mechanisms. LKB1 is an upstream kinase of AMPK, and LKB1 dominantly regulates AMPK and AMPK-related kinases (8, 52). Sakamoto et al. (40) demonstrated that contraction-enhanced glucose uptake is lost in skeletal muscles from muscle-specific LKB1 knockout mice, and Koh et al. (25) reported that contraction-induced glucose uptake activity is lower in tibialis anterior muscles overexpressing mutant SNARK, a member of the AMPK-related kinase family, than in those transfected with empty vector, suggesting possibilities that other AMPK-related kinases may compensate the role of AMPK when impaired. Our present study should also be of significance as a novel approach without such disturbances.
There had been no reports on metabolomic analysis in contracting skeletal muscles electrically stimulated ex vivo. However, only a few studies have dealt with “omics” technologies to investigate metabolic effects of physical exercise or muscle contraction. Goto-Inoue et al. (13) very recently reported lipidomic analysis adopting a matrix-assisted laser desorption/ionization imaging mass spectrometry (MALDI-IMS) in mice transverse abdominal muscles electrically stimulated to contract in vivo under anesthesia, revealing a significant change not in the total amount but in the composition of lipids. Secretome profiles from skeletal muscles after semichronic training were demonstrated by the differential display method (39). In humans, metabolome and lipidome not in skeletal muscles but only in plasma have been investigated (28, 29, 55). Our present study directly analyzed skeletal muscles and adopted electrical stimulation ex vivo to induce contraction, which enables us to reveal the pure results of muscle contraction itself without being affected by secondary effects such as changes in blood flow or hormones. It should be highly advantageous especially in comprehensive analyses like the metabolomics. Further study comparing the electrical contraction ex vivo with physical exercises will provide us the information on such secondary effects, which is significant components of metabolic influences by physical exercise, on the metabolome. Metabolomic analysis on the genetically manipulated animals will also be informative in spite of the aforementioned problems.
There have been lots of studies on the significance of AMPK on carbohydrate metabolism, and it is reasonable that quite high correlation was observed in the subanalysis related to glycolysis and tricarboxylic acid cycles (Fig. 2B) (6, 17–20, 22, 26, 31, 34, 35, 37, 47, 48, 50, 51, 57). As for NAD metabolism, Canto et al. (9) showed AMPK activation increases NAD production leading to sirt1 activation of which molecular mechanism has not been clarified well. The enhancement of NAD production seems to require >4 h. In accordance, we did not find increase in the NAD in our acute experimental conditions (0.98-fold for AICAR stimulation, 1.14-fold for electrical contraction to the control; Fig. 2C). NADP levels were also almost unchanged (0.87-fold for AICAR stimulation; 1.00-fold for electrical contraction to the control), while nicotinamide was highly decreased (0.53-fold for AICAR stimulation, 0.59-fold for electrical contraction to the control). The common influences by AMPK activation on the redox status may partially explain the similarity in NAD metabolism. The metabolites related to nucleic acid and ureic acid metabolisms also exhibited high similarity between AICAR stimulation and electrical contraction (Fig. 2, D and E). AICAR stimulation did not change AMP concentration while electrical contraction strikingly increased it (0.90-fold for AICAR stimulation, 2.00-fold for electrical contraction to the control). ATP levels were markedly increased to the same extent, which are supposed to be due to AMPK activation-stimulated compensation of ATP levels (2.03-fold for AICAR stimulation, 1.77-fold for electrical contraction to the control). It is interesting that ADP of which levels were changed only ∼10% (0.90-fold for AICAR stimulation, 1.07-fold for electrical contraction to the control) is the most influential factor consisting of PC2 (Fig. 3C). It may be due to the discrepancy in AMP and ATP levels.
In spite of the overall high correlation, discrepancies were observed as for amino acids metabolism (Fig. 2, F–I). Among the 20 fundamental amino acids, only cystein was not detected in the muscle samples. There was no correlation in the remaining 19 amino acids (r = −0.357, P = 0.133, data not shown). However, 16 amino acids except for aspartate, alanine and lysine exhibited striking correlation between the stimulations (r = 0.776, P < 0.001). Therefore, metabolic pathways related to aspartate, alanine, or lysine are supposed to be the significant signals characterizing difference between AICAR stimulation and electrical contraction, i.e., AMPK-independent metabolome.
Muscle contraction consumes much ATP and the resultant AMP leads to AMPK activation. In contrast, AICAR stimulation activates AMPK without affecting energy demand and this difference in energy status will influence the demands of the metabolites. In contrast to the effects of muscle contraction, in which metabolic turnover is accelerated, the metabolic changes evoked by AMPK activation without increase in energy demand may not last long because the metabolic pathways will eventually turn off due to negative feedback from the accumulated products. We admit the possibility that such kinds of disturbances in addition to the off-target effects of AICAR may be involved in our observations. However, we found it noteworthy that AICAR stimulation acutely exhibits very close similarity to muscle contraction in many aspects of metabolic results despite the big differences in the energy status.
In conclusion, our results provided novel evidence that AICAR stimulation closely mimics the effects of electrical contraction metabolome widely. AMPK should be a central player at regulating a large part of the metabolome in contracting skeletal muscles. We also successfully distinguished the major AMPK-dependent metabolic effects from independent ones, which are minor, in contracting skeletal muscles, where AMPK is activated. Further studies conjunct with enzymatic analyses will provide us more profound understanding of metabolomic regulation in contracting skeletal muscles and should be helpful in developing more efficient therapeutic and preventive strategies against metabolic disorders such as diabetes.
This work was supported by research grants from the Nakatomi Foundation, Urakami Foundation for Food and Food Culture, Japan Society for the Promotion of Science, and Japan Vascular Disease Research Foundation.
No conflicts of interest, financial or otherwise, are declared by the author(s).
Author contributions: L.M. conception and design of research; L.M., T.E., R.O., E.K., and Y.T. performed experiments; L.M. analyzed data; L.M. interpreted results of experiments; L.M. prepared figures; L.M. drafted manuscript; L.M. edited and revised manuscript; L.M., K.T., and T.H. approved final version of manuscript.
We thank Mayumi Ozawa, Ikuko Uemura, and Mary Miller-Maka for checking the manuscript; Dr. Taro Toyoda, Dr. Kazuhiko Tsutsumi, Dr. Satoko Nakanishi, and Atsushi Nagashima for helpful advice; and Radioisotope Research Center of Kyoto University and Kyoto University Research Center for Low Temperature and Materials Sciences for instrumental supports.
- Copyright © 2013 the American Physiological Society