AMPK and PPARβ positive feedback loop regulates endurance exercise training-mediated GLUT4 expression in skeletal muscle

Abstract

The objective of this study is to determine whether AMP-activated protein kinase (AMPK), peroxisome proliferator-activated receptor gamma coactivator 1-α (PGC-1α), or peroxisome proliferator-activated receptor β (PPARβ) can independently mediate the increase of glucose transporter type 4 (GLUT4) expression that occurs in response to exercise training. We found that PPARβ can regulate GLUT4 expression without PGC-1α. We also found AMPK and PPARβ are important for maintaining normal physiological levels of GLUT4 protein in the sedentary condition as well following exercise training. However, AMPK and PPARβ are not essential for the increase in GLUT4 protein expression that occurs in response to exercise training. We discovered that AMPK activation increases PPARβ via myocyte enhancer factor 2A (MEF2A), which acted as a transcription factor for PPARβ. Furthermore, exercise training increases the cooperation of AMPK and PPARβ to regulate glucose uptake. In conclusion, cooperation between AMPK and PPARβ via NRF-1/MEF2A pathway enhances the exercise training mediated adaptive increase in GLUT4 expression and subsequent glucose uptake in skeletal muscle.

INTRODUCTION

Exercise has a major impact on whole-body glucose metabolism. Exercise improves glucose control by increasing skeletal muscle glucose uptake (13, 15), by both insulin-dependent and -independent mechanisms (13, 15). It has been shown that the acute increase in glucose uptake that occurs in response to muscle contractions is mediated by the translocation of intracellular vesicles containing glucose transporter type 4 (GLUT4) proteins to the cell membrane (13, 15). In addition to the acute effect on the movement of GLUT4 within the cell, exercise training (Ext) increases GLUT4 protein expression in skeletal muscle, which is a major determinant of maximal glucose transport response achieved by either contraction (30, 32) and/or via insulin-stimulated pathway (11, 19).

GLUT4 transcription in skeletal muscle is mediated by myocyte enhancer factor 2A (MEF2A) transcription factor (25, 27). Exercise increases an MEF2A binding to GLUT4 promoter in skeletal muscle (16, 33, 34, 36), and exercise-induced MEF2 is necessary for GLUT4 gene expression in skeletal muscle (16, 36). Ext is also known to activate AMP-activated protein kinase (AMPK), peroxisome proliferator-activated receptor β/δ (PPARβ) and peroxisome proliferator-activated receptor gamma coactivator 1-α (PGC-1α), and each play an important role in regulating glucose uptake (8, 14). Each one of these factors increases glucose uptake by increasing GLUT4 expression via MEF2A activity or expression (8, 14). Moreover, increased expression of PPARβ, a transcription regulator in skeletal muscle (3, 37), also enhances GLUT4 expression, glucose metabolism (8), and insulin sensitivity (24, 37). Because PPARβ overexpression can increase AMPK activity (29) and PGC-1α protein (9), and AMPK activation in turn can increase PPARβ (6, 20) and PGC-1α (17), it is unclear whether any single factor or molecular regulatory point is critical for enhancement of GLUT4 expression. Therefore, the objective of the current study was to determine whether AMPK, PGC-1α, or PPARβ independently mediate the increase of GLUT4 expression that occurs in response to Ext or whether the coordinated activity of AMPK, PGC-1α, and PPARβ are required. Knowing what factors are necessary for the increase in GLUT4 is key to identifying potential therapeutic targets to improve systemic glucose management.

MATERIALS AND METHODS

Animals.

This study was approved by the Animal Studies Committee of Washington University (St. Louis, MO). All animal studies were performed at the Washington University of St. Louis. Male C57BL/6J mice weighing ~20 g mice were purchased from Jackson Laboratory (Bar Harbor, ME). Mice were housed three animals per cage and provided with a diet of Purina chow and water ad libitum. Male Wistar rats weighing ~150 g were purchased from Charles River Laboratories (New York, NY), singly housed, and provided with Purina chow and water ad libitum.

Cell lines.

C₂C₁₂ mouse muscle cell (source from ATCC, cat. no. CRL-1772, female), HEK293 cell (source from ATCC, no. CRL-1573, female) were obtained from ATCC (Manassas, VA). Cells were maintained in 5% CO₂ at 37°C and grown in DMEM (Sigma-Aldrich, cat. no. D-5796, St. Louis, MO) containing 10% fetal bovine serum (Sigma-Aldrich, cat. no. F-2442), penicillin (100 U/ml), and streptomycin (100 µg/ml) (Thermo Fisher Scientific, cat. no, 15140122, Waltham, MA). C₂C₁₂ cells were differentiated into myotubes by changing to the media containing 2% horse serum (Thermo Fisher Scientific, cat. no. 26050070), penicillin (100 U/ml), and streptomycin (100 µg/ml) (Thermo Fisher Scientific, cat. no. 15140122).

DNA constructs.

The Myc-PPARβ cDNA was PCR cloned (9) and inserted into pcDNA 3.1 vector (Thermo Fisher Scientific, Houston, TX). The pcDNA3.1-Flag-MEF2A was generated as previous published (31, 39). The constitutively active AMPK construct (CA-AMPK), dominant negative (DN)-AMPK, and short hairpin PPARβ RNA-silencing constructs (shPPARβ) were gifted by John O. Holloszy (Washington University School of Medicine). The MEF2 binding region of the PPARβ promoter was PCR cloned (Forward: 5′-GAGA GGGGAGGGGTCATCTTT-3′; Reverse 5′-TGCTCCTCCCTTAGCTGCTA-3′) using human genomic DNA (Clontech, Mountain View, CA) and ligated into a pZsGreen 1.1 vector (Clontech) to construct a PPARβ promoter reporter. MEF2A plasmid was inserted into pCMX-Gal4 vector (Addgene, Cambridge, MA) to make MEF2A-Gal4 vector, and Gal4 response element (Addgene) was inserted into pZsGreen 1.1 vector (Clontech) to make Gal4-green fluorescence protein (GFP).

Muscle electroporation.

Transfection of plasmid DNA into mouse tibialis anterior (TA) muscle, rat epitrochlearis (Epi), or triceps muscle, mainly glycolytic fiber type, was performed using an electric pulse-mediated gene transfer technique (31, 39). Mice were anesthetized with isoflurane gas. A TA muscle was injected with 50 µg of plasmid DNA containing either myc-PPARβ-GFP or empty vector (EV) in 50 µl with GFP saline using a 27-gauge needle at a rate of 20 µl/min. Rats were anesthetized, and an Epi muscle was injected with 10 µg of plasmid DNA containing either a shPPARβ or an EV in 10 µl saline using a 27-gauge needle at a rate of 10 µl/min. Epi muscle in the rat was used for exercise; AMPK or PPARβ lacking from skeletal muscle can decrease exercise capacity. The gene lacking Epi has few effects on exercise capacity because Epi is small muscle attached triceps (31, 39). A triceps was injected with 80 μg of plasmid DNA containing either empty PPARβ with or without DN-AMPK or EV in 80 μl GFP saline using a 27-gauge needle at a rate of 20 μl/min. After injection an electric field, generated with S88 stimulator (Grass Instrument, West Warwick, RI), was applied to the muscle (12). Animals were anesthetized with an intraperitoneal injection of sodium pentobarbital and muscles were dissected, frozen in liquid nitrogen, and kept at −80°C until analyzed. Anesthetized mice were euthanized by exsanguination.

Exercise training.

Rats were subjected to a 2-wk Ext program. The protocol consisted of 4 bouts of swimming per day in steel barrels filled with temperature controlled (~35°C) water to a depth of ~60 cm (4, 28). Each bout lasted for 30 mins, with a 5-min rest period between bouts. After the first bout, a weight equal to 1.5% of body weight was tied to the base of the rat’s tail. The animals swam with the weight attached for the remaining three bouts. Animals were acclimatized to swimming for 10 min per day during the week before the exercise bout. The Epi muscles were studied ~24 h after the exercise.

Western blot analysis.

Frozen muscles were homogenized, and Western blots were prepared as previously described (21). Briefly, frozen tissues were powdered and then homogenized in a 15:1 (vol/wt) ratio of ice-cold RIPA buffer supplemented with protease inhibitor cocktail tablet (Santa Cruz Biotechnology, cat. no. SC-29131, Santa Cruz, CA). Protein concentration was measured as previously described (23). Antibodies used are as follows: PGC-1α (cat. no. 516557) from Millipore (Billerica, MA); MEF2A (cat. no. 9736), c-Myc-Tag (cat. no. 2276), AMPK (cat. no. 2793), and phospho-AMPK (cat. no. 2541) from Cell Signaling Technologies (Danvers, MA); PPARβ (cat. no. PA5-29678) from Thermo Fisher Scientific (Houston, TX); NRF-1 (cat. no. SC-23624) from Santa Cruz Biotechnology (Santa Cruz, CA); calcium/calmodulin-dependent protein kinase kinase β (CaMKKβ; cat. no. A304-009A) from Bethyl (Montgomery, TX); β-actin (cat. no. A5441) from Sigma Aldrich; GLUT4 (cat. no. ab-654), sodium potassium ATPase (ab-76020) from Abcam (Cambridge, MA). Secondary antibodies such as donkey anti-mouse (cat. no. 715-035-150), donkey anti-rabbit (cat. no. 711-035-152), and streptavidin (cat. no. 016-030-084) were obtained from Jackson Immunoresearch Laboratories (West Grove, PA). Antibody-bound protein was detected by Clarity Western ECL Substrate (Bio-Rad, cat. no. 170-5060, Hercules, CA). Signal was visualized using a C-DiGit blot scanner (Li-COR Bioscience, cat. no. 3600-00, Lincoln, NE).

Myc-PPARβ immunoprecipitation.

Myc-PPARβ immunoprecipitation (IP) was performed as previously described (21). Briefly, 200 mg protein contained cell extracts of myotubes were rotated with myc-antibody (Cell Signaling) at 4°C overnight. The following morning magnetic beads (Millipore) were added and the samples were rotated at 4°C for 4 h. Western blotting was used to measure AMPK with an anti-AMPK antibody (Cell Signaling).

Semiquantitative RT-PCR.

mRNA was determined as previously described (35). Total RNA from cells were isolated using TRIzol reagent (Thermo Fisher Scientific, Houston, TX). DNase-treated total RNA (1 g) was reverse transcribed into cDNA by using random primer and Im Prom-II Reverse Transcriptase (Promega, Madison, WI). Aliquots of each reverse-transcribed reaction were added to a PCR master mix (Promega) mixture containing Taq DNA polymerase, deoxynucleoside triphosphates, MgCl2, reaction buffers at optimal concentrations for efficient amplification of DNA templates by PCR, and 10 pmol of both sense and antisense primers (forward 5′-TGGAGCTCGATGACAGTGAC-3′, reverse 5′-GTACTGGCTGTCAGGGTGGT-3′); (forward 5′-GACCATAAACGATGCCGACT-3′, reverse 5′-AGACAAATCGCTCCACCAAC-3′). The reaction medium was subjected to PCR amplification. After the lid was warmed at 94°C for 2 min, the PCR mixtures were subjected to a 27-cycle profile, including denaturation for 30 s at 94°C, hybridization for 60 s at 58°C, and elongation for 60 s at 72°C. 18S rRNA expression was simultaneously measured as an internal standard by using a QuantumRNA 18S Internal Standard Kit (Ambion, Austin, TX). PCR products were separated by electrophoresis on 2% agarose, stained with SYBR Green (Molecular Probes, Eugene, OR), photographed, and analyzed by densitometry. The ratio of uncoupling protein 3 to 18S rRNA standard band densities was then calculated.

Gal4-MEF2A Activity.

pCMX-Gal4-MEF2A and Gal4-GFP were cotransfected with DNA plasmid such as CA-AMPK, PPARβ, DN-AMPK and shPPARβ using Lipofectamine 2000 (Thermo Fisher Scientific, Houston, TX) in C₂C₁₂ cells. Two days after transfection, cells were harvested and GFP and DS-Red fluorescence intensities were measured in cell extracts using a microplate reader (Biotek, Winooski, VT). GFP fluorescence intensity was corrected for DS-Red fluorescence intensity and normalized to the value of EV control.

PPARβ and MEF2A promoter activity assay.

To determine the effect of MEF2A overexpression on PPARβ promoter transcriptional activity, HEK cells were cotransfected with a PPARβ promoter-GFP reporter construct and a cytomegalovirus promoter driven DS-Red-Express 2-C1 vector (Clontech, Mountain View, CA) and either MEF2A or EV using Lipofectamine 2000 (Thermo Fisher Scientific, Houston, TX). Thirty-six hours posttransfection, cells were harvested, and GFP and DS-Red fluorescence intensities were measured in cell extracts using a microplate reader (Biotek, Winooski, VT). GFP fluorescence intensity was corrected for DS-Red fluorescence intensity and normalized to the value of EV control.

Immunofluorescence analysis for GLUT4 localization.

Thirty minutes after insulin intraperitoneal injection, dissected muscles were embedded in OCT compound (Tissue-Tek, VWR, IL), frozen immediately in isopentane, which was cooled liquid nitrogen and stored at −80°C. Muscle was cut into 10-μm thick cryosections with cryostat. Multicolor immunofluorescence was conducted (2) to determine GLUT4 localization. For immunofluorescence staining, frozen muscles were cryosectioned using a microtome within a cryostat (Thermo Electronic) to a thickness of 10 µm onto uncoated glass microscope slides. GLUT4 antibody (Abcam) was applied to the sections for 2 h at room temperature. GLUT4 primary antibody was combined with either dystrophin (Sigma Aldrich) antibodies. Secondary antibodies were applied to sections for 30 min at room temperature. GLUT4 antibody was targeted with goat anti-rabbit IgG 488 (Thermo Fisher Scientific, Houston, TX) and dystrophin with goat anti-mouse IgG2b 594 (Thermo Fisher Scientific, Houston, TX). Slides were visualized with Nikon eclipse te2000-u (Nikon, Melville, NY) and X-cite 120 PC (Excelitas, Waltham, MA). The microscope was equipped with a red (Excitation: BP 545/25 nm; Emission BP 605/70 nm) and green (Excitation: BP 470/40 nm; Emission BP 525/50 nm) filters. The image was captured with MetaMorph software (Molecular Devices, Sunnyvale, CA).

Chromatin IP assay.

Chromatin IP (ChIP) assays were performed using an EZ-ChIP kit (Upstate Biotechnology, Lake Placid, NY). To determine MEF2A binding to PPARβ promoter, MEF2A was overexpressed in HEK cells by Lipofectamine 2000 (Thermo Fisher Scientific, Houston, TX). ChIP assay protocol was previously published (21). Briefly, 10-cm plates of the HEK cells were cross-linked in PBS containing 1% formaldehyde. HEK cells were extracted, and DNA was sheared by sonicating. After centrifugation, supernatant containing 0.1 mg protein was diluted to 0.1 ml with dilution buffer (Upstate Biotechnology, Lake Placid, NY). Five micrograms of anti-Flag antibody was added per sample and incubated overnight at 4°C. Streptavidin beads 88816 (Thermo Fisher Scientific, Waltham, MA) solution was added, and the sample was mixed for 4 h at 4°C. Precipitated complexes were eluted in 100 µl of elution buffer (Upstate Biotechnology) and cross-linking was reversed by the addition of 8 µl of 5 M NaCl per sample followed by incubation at 65°C overnight. Coimmunoprecipitated DNA was purified according to the manufacturer’s directions. Primers were designed to amplify the MEF2A-binding region of the PPARβ promoter. The following primers were used: forward, 5′-GGTTGGGATAATAGGTCC CTTCC-3′ reverse, 5′-AAAAAGGGCTTTCCGCCTTC-3′. PCR products were loaded onto 2% agarose gel, electrophoretically separated, and densities of the bands quantified. Purified DNA from the input sample that was not immunoprecipitated was PCR amplified and used to normalize signals from the ChIP assays.

Glucose consumption.

Each gene (CA-AMPK, PPARβ, DN-AMPK, and shPPARβ) was transfected in C₂C₁₂ using lipofectamine 2000 (Thermo Fisher Scientific, Houston, TX). Glucose level in remained cell medium was determined using assays kit obtained from BioVision (Milpitas, CA). Glucose consumption was calculated as total glucose concentration minus glucose that remained in the media.

Extraction of plasma membrane protein.

We used a commercially available plasma membrane protein extraction kit (BioVision) and followed manufacturer’s directions. Cells in t75 flask were collected and then spun down. Cells were washed with PBS and were homogenized with homogenization buffer on ice, and then samples were centrifuged at 700 g for 10 min at 4°C to collect the supernatant. The supernatant was centrifuged at 10,000 g for 30 min at 4°C to collect the pellet. The pellet was resuspended in 200 µl of the upper phase solution, and then 200 µl of lower phase solution was added and was mixed well, followed by incubation on ice for 5 min. Another fresh phase tube without sample was prepared with 200 µl of upper phase solution and 200 µl of lower phase solution followed by centrifugation 1,000 g for 5 min. An upper phase was collected and then was diluted with 5 volumes of water followed by incubation 5 min on ice. Sample was spun at 15,000 g for 10 min at 4°C and then we used the pellet that is the plasma membrane protein.

Quantification and statistical analysis.

Statistical procedures were indicated in figure legends. Briefly, statistical analyses were conducted using SigmaPlot v.12 software (Systat, San Jose, CA). Normally distributed data were analyzed utilizing standard parametric statistics including Student’s t-test, one-way ANOVA or two-way ANOVA, and post hoc analysis was conducted with Fisher least significant difference. Data are expressed as means ± SE, and statistical significance was accepted at P < 0.05.

RESULTS

AMPK plays an important role in maintaining endurance training-induced GLUT4 expression in skeletal muscle.

It is well known that AMPK activation by muscle contraction leads to increased glucose uptake via an insulin-independent manner. Ablation of AMPKα2 activity in skeletal muscle exacerbates the development of insulin resistance induced by high-fat feeding (7). AMPKα is also essential for the acute exercise induced increase in GLUT4 mRNA expression (5) but not for Ext-induced GLUT4 protein expression in mouse skeletal muscle (5). Hence, the role of AMPK in inducing and maintaining GLUT4 expression in skeletal muscle by endurance Ext is still controversial.

To evaluate whether AMPK is essential for Ext-mediated induction of GLUT4 expression in skeletal muscle, we expressed a DN-AMPK or EV in left or right Epi of rat, respectively (Fig. 1A). Animals were randomly assigned to either a 2-wk sedentary or Ext program. Twenty four hours after the Ext intervention, DN-AMPK overexpression in skeletal muscle resulted in a reduction in MEF2A and GLUT4 protein content when compared with EV, in both sedentary and Ext groups (Fig. 1A). These results showed that AMPK is essential for maintaining normal physiological levels of GLUT4 expression with or without Ext. Ext did result in a significant increase in GLUT4 expression in both the EV and DN-AMPK muscles (Fig. 1A). This result indicates that other factors induced by Ext can mediate an increase in GLUT4 expression. Thus, although AMPK appeared to be critical for the maintenance of normal levels of GLUT4 (in both sedentary and following Ext), it was not essential for the increase in GLUT4 expression that occurred in response to Ext (Fig. 1A).

Fig. 1.

AMPK maintains and increases GLUT4 expression in skeletal muscle by endurance exercise training (Ext). Depletion of AMPK activity decreases a normal adaptive response for GLUT4 expression in rat skeletal muscles by endurance Ext (A). Values are means ± SE [n = 6 epitrochlearis (Epi) per group]. *P < 0.05 vs. EV, #P < 0.05 vs. sedentary (Sed). Significance was determined using two-way ANOVA and Tukey test. Depletion of AMPK activation reduces GLUT4 expression in rat skeletal muscles via NRF-1/MEF2A induced by PPARβ (B). Values are means ± SE (n = 6 triceps per group). *P < 0.05 vs. EV, #P < 0.05 vs. PPARβ. Significance was determined using one-way ANOVA and Tukey test. AMPK, AMPK kinase; DN, dominant negative; EV, empty vector; GLUT4, glucose transporter type 4; MEF2A, myocyte enhancer factor 2A; PPARβ, peroxisome proliferator-activated receptor β.

Because AMPK is not essential for this increase in GLUT4 expression, we also tested PPARβ and NRF-1, because we previously showed that PPARβ can bind and increase NRF-1 promoter activity (1, 21, 31), which regulates MEF2A transcription (1, 21, 31) for GLUT4 expression. Thus, PPARβ and NRF-1 are upstream regulators of GLUT4 expression. We showed that AMPK is directly involved in the regulation of both PPARβ and NRF-1 as the ablation of AMPK activation results in a significant reduction in of PPARβ and NRF-1 expression, when compared with EV, in both sedentary and Ext groups (Fig. 1A). Furthermore, as noted above Ext-induced GLUT4 levels were significantly lower in DN-AMPK compared with EV (Fig. 1A). Despite the ablation of AMPK in DN-AMPK overexpressed muscles, Ext still caused an increase in the expression of PPARβ, NRF-1, MEF2A, and GLUT4 in these muscles compared with sedentary intervention, albeit at a much lower expression level. Thus, other factors induced by Ext can affect PPARβ, NRF-1, MEF2A, and GLUT4 expression.

To define the role of PPARβ in regulating GLUT4 expression we overexpressed PPARβ with or without DN-AMPK in sedentary rat triceps. We found that PPARβ overexpression with functioning AMPK results in an increase in pAMPK, NRF1, MEF2A, and GLUT4 expression (Fig. 1B). However, PPARβ overexpression with DN-AMPK did not increase MEF2A and GLUT4 expression (Fig. 1B). Thus, AMPK activation is necessary for PPARβ overexpression to increase GLUT4.

Importance of PPARβ for Ext-mediated induction of GLUT4 expression in skeletal muscle.

Previous studies have shown that PPARβ regulates glucose metabolism and insulin sensitivity (22), and muscle-specific overexpression of PPARβ increases GLUT4 protein expression (8). To determine if PPARβ is essential for the adaptive increase in GLUT4 expression following Ext, PPARβ expression was suppressed in Epi muscle via PPARβ shRNA overexpression (Fig. 2A). Reduced PPARβ expression resulted in a reduction in CaMKKβ, MEF2A, and GLUT4 in muscles from sedentary animals, indicating that PPARβ has an important role in maintaining normal levels of these proteins (Fig. 2B). In addition, although total AMPK protein levels were not different with a reduction in PPARβ expression, pAMPK was reduced in muscles from sedentary animals (Fig. 2B). As expected, Ext resulted in an increase in CaMKKβ, pAMPK, and GLUT4 (Fig. 2B). PPARβ shRNA overexpression did not prevent an increase in CaMKKβ, pAMPK, MEF2A, and GLUT4 in response to Ext, but the levels were significantly reduced compared with the EV control (Fig. 2B). These results indicate that PPARβ is important for the maintenance of GLUT4 expression in both the sedentary condition and following Ext (Fig. 2). Because an increase in GLUT4 that occurs in response to Ext was not prevented when PPARβ was suppressed, this suggests other factors appear to be important in effecting an increase in CaMKKβ, AMPK activation, and MEF2A, leading to an increase in GLUT4 expression following Ext.

Fig. 2.

PPARβ maintains and increases GLUT4 expression in skeletal muscle by endurance exercise training (Ext). A: PPARβ expression in rat skeletal muscles by shPPARβ overexpression. B: depletion of PPARβ expression reduces a normal adaptive response for GLUT4 expression in rat skeletal muscles by endurance Ext. Values are means ± SE (n = 6 Epi per group). *P < 0.05 vs. EV, #P < 0.05 vs. sedentary (Sed). Significance was determined using two-way ANOVA and Tukey test. Epi, epitrochlearis; EV, empty vector; GLUT4, glucose transporter type 4; MEF2A, myocyte enhancer factor 2A; PPARβ, peroxisome proliferator-activated receptor β; shPPARβ, short hairpin PPARβ.

PPARβ increases GLUT4 expression without PGC-1α.

PGC-1α has been shown to coactivate MEF2A (10) and also increase MEF2A protein expression by activating NRF-1 (1, 31). PPARβ overexpression in skeletal muscle has also been shown to cause an increase in NRF-1 expression by activating the NRF-1 promoter before an increase in PGC-1α protein expression (21). Because MEF2A has binding site for NRF-1, we hypothesized that PPARβ can directly increase GLUT4 expression via NRF-1 activation. To test this, we generated muscl-specific PGC-1α knockout mice (Fig. 3A) and overexpressed PPARβ in TA muscle (Fig. 3B). PPARβ overexpression in PGC-1α-negative TA muscle caused a significant increase in the expression of NRF1, MEF2A, GLUT4, and AMPK (total and pAMPK) (Fig. 3B). These results demonstrate that PPARβ can enhance GLUT4 expression independent of PGC-1α expression.

Fig. 3.

PPARβ increases GLUT4 expression via NRF-1/MEF2A without PGC-1α. A: PGC-1α expression in PGC-1α knock out (KO) muscle. B: PPARβ overexpression in PGC-1α KO mouse muscle results in an increase GLUT4 via NRF-1/MEF2A. Values are means ± SE [n = 4 tibialis anterior (TA) muscles per group]. *P < 0.05, **P < 0.01, ***P < 0.001 versus empty vector (EV). Significance was determined using student’s t-test. GLUT4, glucose transporter type 4; MEF2A, myocyte enhancer factor 2A; PGC-1α, peroxisome proliferator-activated receptor gamma coactivator 1-α; PPARβ, peroxisome proliferator-activated receptor β.

AMPK activity and PPARβ are essential for MEF2A activation.

Here, we show that AMPK and PPARβ independently play important roles in the maintenance of normal GLUT4 expression in skeletal muscle either in the sedentary condition or following Ext (Figs. 1 and 2). However, AMPK and PPARβ are not essential for the relative increase in GLUT4 expression observed in response to Ext (Figs. 1 and 2). Thus, we sought to determine whether AMPK activation or PPARβ can independently or cooperatively activate MEF2A. To determine this, Gal4 green fluorescence reporter was overexpressed in C₂C₁₂ with or without CA-AMPK and/or PPARβ (Fig. 4A). We found that MEF2A activity was significantly increased in response to either CA-AMPK or PPARβ overexpression in C₂C₁₂ myocytes. Furthermore, an increase in MEF2A activity did not occur with overexpression of CA-AMPK if PPARβ expression was knocked down or with PPARβ overexpression if AMPK activity was limited (DN-AMPK) (Fig. 4). Thus, AMPK activity or PPARβ is essential for activation of MEF2A. Because overexpression of both CA-AMPK and PPARβ does not synergistically increase MEF2A activity, this result suggests that CA-AMPK and PPARβ cooperate to activate MEF2A via the same pathway (Fig. 4B). Finally, the changes observed in MEF2A promoter activity with AMPK and PPARβ expression were closely reflected in glucose consumption patterns (Fig. 4C). These results support the hypothesis that AMPK and PPARβ cooperate to increase glucose uptake via MEF2A/GLUT4 pathway (Fig. 4C).

Fig. 4.

AMPK activity and PPARβ is essential for MEF2A activation and glucose uptake. A: GFP represents MEF2A activity and DS-red is for normalization for GFP. B: AMPK activity and PPARβ is essential for MEF2A activity. C: AMPK and PPARβ cooperate for glucose uptake into the cell. Glucose consumption was calculated as total glucose present minus remained glucose in media. Values are means ± SE (n = 8 C₂C₁₂ cells per group). *P < 0.05 vs. EV. Significance was determined using one-way ANOVA and Tukey test. AMPK, AMP kinase; CA, constitutively active; DN, dominant negative; EV, empty vector; GFP, green flourescense protein; GLUT4, glucose transporter type 4; MEF2A, myocyte enhancer factor 2A; PPARβ, peroxisome proliferator-activated receptor β; shPPARβ, short hairpin PPARβ.

MEF2A is a transcription factor for PPARβ.

Previous work from our group has shown that exogenous overexpression of PPARβ increases endogenous expression of PPARβ (21). Therefore, we wanted to determine if MEF2A, which increases with Ext, PPARβ overexpression, and AMPK activation, may also regulate the transcription of PPARβ. Thus we mapped the PPARβ promoter region to a MEF2A binding site (Fig. 5A). Furthermore, when MEF2A was overexpressed this significantly increased PPARβ promoter activity (Fig. 5B).

Fig. 5.

AMPK-PPARβ positive feedback loop for GLUT4 expression. A: MEF2A binds to PPARβ promoter. Values are means ± SE (n = 4 HEK cell per group). ***P < 0.001 vs. IgG. Significance was determined using student’s t-test. B: MEF2A activates PPARβ promoter. Values are means ± SE (n = 6 HEK cell per group). ***P < 0.001 vs. EV. C: activated AMPK binds to PPARβ. D: CA-AMPK increases PPARβ mRNA in mouse skeletal muscle. Values are means ± SE (n = 6 TA muscle per group). ***P < 0.001 vs. EV. E: schema of the positive feedback loop between activated AMPK and PPARβ. AMPK, AMP kinase; CA, constitutively active; EV, empty vector; GLUT4, glucose transporter type 4; MEF2A, myocyte enhancer factor 2A; PPARβ, peroxisome proliferator-activated receptor β; shPPARβ, short hairpin PPARβ; TA, tibialis anterior; WT, wild-type.

AMPK and PPARβ positive loop increase GLUT4 localization.

It has been shown that activated AMPK can regulate MEF2A transcription (26). Our study shows that MEF2A can also regulate PPARβ transcription by binding to its promoter and increasing transcription (Fig. 5, A and B). As has been reported previously, AMPK can bind directly to PPARβ (8), but the relationship between this cooperative binding and the regulation of GLUT4 is not clear. Expressing CA-AMPK in HEK cells, we found that activated AMPK has higher binding capacity to PPARβ than nonactivated AMPK (Fig. 5C). In addition, CA-AMPK overexpression in skeletal muscle also increased PPARβ transcription (Fig. 5D).

As noted above, activated AMPK and PPARβ cooperate to control glucose uptake (Fig. 4, B and C). To show more clearly that this is because of an increase in GLUT4 in the membrane, we measured GLUT4 localization in PPARβ-overexpressed and -activated AMPK muscle. Our immunofluorescence results showed that GLUT4 localization in PPARβ-overexpressed muscle is higher than that of EV muscle (Fig. 6A). We also confirmed that PPARβ increases GLUT4 localization to the plasma membrane in C₂C₁₂ cells (Fig. 6B). Finally, we have previously shown that PPARβ overexpression can increase AMPK activation via NRF-1/CaMKKβ pathway (21). The results in this study support the hypothesis that a positive feedback loop exists between AMPK activation, PPARβ, and MEF2A (Fig. 5E).

Fig. 6.

PPARβ overexpression increases GLUT4 localization. A: insulin treatment in PPARβ overexpressed muscle increases GLUT4 localization to the plasma membrane compared with EV muscle. Scale bar, 100 µm. B: PPARβ increases GLUT4 localization to the plasma membrane. Values are means ± SE (n = 3 C₂C₁₂ t75 flask per group). *P < 0.05 vs. EV. Significance was determined using student’s t-test. EV, empty vector; GLUT4, glucose transporter type 4; PPARβ, peroxisome proliferator-activated receptor β.

DISCUSSION

GLUT4 is a primary factor determining the rate of glucose uptake in skeletal muscle. A key adaptation to endurance Ext is an increased expression of GLUT4 in skeletal muscle, allowing greater capacity for glucose transport and better systemic glucose control. The molecular mechanisms responsible for this adaptation to Ext are not fully understood. Three key factors (AMPK, PPARβ, and PGC-1α) known to be activated by Ext are likely important in the adaptive increase in GLUT4 seen with Ext. Overexpression of each of these factors individually has been shown to increase GLUT4 expression. The key findings in this study are that AMPK activation and PPARβ expression induced by endurance Ext cooperate to enhance GLUT4 expression and glucose uptake in skeletal muscle, and that PGC-1α is not required for this effect to occur.

Because PPARβ overexpression increases GLUT4 (Fig. 1A) and also increases PGC-1α by a posttranscription mechanism (21), it was unclear whether PPARβ, PGC-1α or both proteins were primarily responsible for the regulation of GLUT4 expression in response to endurance Ext. We found PPARβ overexpression increased GLUT4 along with NRF-1 and MEF2A without PGC-1α (Fig. 3). This indicates that PGC-1α is not essential for PPARβ dependent regulation of GLUT4 expression (Fig. 3). This finding is consistent with other work that has found PGC-1α is not required for normal muscle glucose disposal and insulin sensitivity (38). Although PGC-1α is not essential to regulate GLUT4 expression (Fig. 3B), increased PPARβ expression by Ext has been shown to increase PGC-1α protein level, and suppression of PPARβ (shPPARβ) markedly reduced the adaptive response to PGC-1α expression by Ext (21). Additionally, because PGC-1α is a coactivator for MEF2A (10) and also increases MEF2A expression by coactivating NRF-1 (1, 31), it is likely that the increases in PGC-1α along with PPARβ in response to Ext synergistically increase GLUT4 expression and enhances the capacity for glucose uptake.

The role of AMPK in inducing GLUT4 mRNA and protein in response to Ext is controversial. Earlier studies showed that Ext increases AMPK activity and MEF2A binding to GLUT4 promoter, resulting in increased transcription and translation (26, 33, 34). However, Ext studies in AMPK knockout mice (5, 18) suggested that AMPK is not required to induce GLUT4 mRNA and protein in response to Ext. In our study, the muscles where AMPK activation was dramatically limited (DN-AMPK) Ext still resulted in a clear increase in GLUT4 expression [2.43-fold higher than sedentary control (Fig. 1)], which was similar to the increase in GLUT4 expression that occurred in the control muscles (EV; 2.09 fold higher GLUT4 expression in Ext group compared with sedentary control). This result clearly indicates that other factors regulated by Ext can cause an increase in GLUT4 expression. Thus AMPK appears to be essential in maintaining a certain level of GLUT4 expression in both the sedentary condition and following Ext (Fig. 1A).

The transcription factor MEF2A is an important regulator of GLUT4 expression and is activated by AMPK (26). When AMPK activation was limited (DN-AMPK) in our study, NRF-1 and MEF2A expression was only partially blunted (Fig. 1A). Thus, other factors are likely important to regulate an increase in MEF2A. Because PPARβ is a transcription factor for NRF-1, which is upstream for MEF2A, we hypothesized that AMPK cooperates with PPARβ to amplify the normal adaptive increase in GLUT4 protein seen with Ext. We demonstrated that PPARβ is not required for an adaptive increase in GLUT4 expression with Ext. When muscle PPARβ expression was suppressed (shPPARβ), Ext caused a 2.39-fold increase in GLUT4 protein content. This increase was similar, if not greater, than the increase in GLUT4 observed in the corresponding control muscles (2.11-fold increase in response to Ext in EV-treated muscle; Fig. 2B). Although the increase in response to Ext still occurred in the absence of normal PPARβ, the overall expression of GLUT4 and MEF2A was lower in both the Ext and sedentary muscles compared with the corresponding control (EV) (Fig. 2B). This result indicates that other factors regulated by Ext can cause an adaptive increase in PPARβ-dependent GLUT4 expression and like AMPK, PPARβ appears to be essential in maintaining a certain level of GLUT4 expression both in the sedentary condition and following Ext (Fig. 2B).

The expression of PPARβ and the activation of AMPK both play cooperative roles in MEF2A promoter activation. We found that PPARβ overexpression increases AMPK activation, and when AMPK activation was limited by the overexpression of DN-AMPK, PPARβ also was decreased (along with NRF-1, MEF2A, and GLUT4) (Fig. 1A). Moreover, AMPK activation, or an increase in PPARβ expression, causes an increase in MEF2A promoter activity and glucose uptake (Fig. 4). Interestingly, we found that if PPARβ expression was suppressed (shPPARβ), AMPK activation did not result in increased MEF2A promoter activity (Fig. 4) or glucose uptake. Likewise, when AMPK activation was limited (DN-AMPK), overexpression of PPARβ did not increase MEF2A promoter activity activation and increase glucose uptake (Fig. 4). Finally, we did not observe any additive effect of PPARβ overexpression with CA-AMPK on MEF2A promoter activity or glucose uptake (Fig. 4). All of these results are strong evidence that AMPK and PPARβ cooperate to regulate GLUT4 expression via NRF-1/MEF2A pathway.

Previous studies have shown that PPARβ overexpression in muscle increases pAMPK via NRF-1/CAMKKβ as well as PPARβ expression itself (8, 21, 29). Increased AMPK activity has also been shown to mediate an increase in PPARβ mRNA and protein expression (6, 20). Thus, AMPK actvity can modulate PPARβ expression and PPARβ has been shown to influence AMPK activity. The mechanism by which these two proteins cooperate to modulate GLUT4 has not been clear. In the present study, we found that MEF2A binds to the PPARβ promoter, and p-AMPK/PPARβ bind and increase MEF2A promoter activity (Fig. 5), further regulating PPARβ protein expression. In total, these results suggest the existance of a positive feedback loop involving PPARβ and AMPK activation. This would help explain the cooperative roles for these proteins in regulating MEF2A promoter activity and GLUT4 expression (Fig. 5E).

In conclusion, we demonstrated the both AMPK and PPARβ are essential in maintaining GLUT4 expression in skeletal muscle in both the sedentary condition and following Ext. In addition, the regulation of GLUT4 expression involves a positive feedback loop between activated AMPK and PPARβ, and this loop regulates Ext-induced GLUT4 expression. Our data suggest that activated AMPK acts as a switch to mediate GLUT4 expression because AMPK and PPARβ cooperate to increase MEF2A activation and glucose uptake.

GRANTS

Support for this work was provided by National Institute of Health grant no. AG-00425 (to J. Holloszy), David H. Murdock-Dole Food Company Professorship (to K. Nair), and the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (grant no. NRF-2019R1A2C1006334). This study was also supported by a National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIT) (grant no. NRF-2019R1A2C1006334).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

J.-H.K. performed experiments; J.-H.K., D.-H.H., and S.D. analyzed data; J.-H.K., J.O.H., and S.D. interpreted results of experiments; J.-H.K. and S.D. prepared figures; J.-H.K. drafted manuscript; J.-H.K., C.R.H., K.S.N., and S.D. edited and revised manuscript; J.-H.K., K.S.N., and S.D. approved final version of manuscript; J.O.H. conceived and designed research.