MEF2A Binding to the Glut4 Promoter Occurs via an AMPKα₂-Dependent Mechanism

Abstract

Purpose: 

The role of AMP-activated protein kinase α₂ (AMPKα₂) in regulating MEF2A nucleus translocation, nuclear histone deacetylase 5 (HDAC5) association with MEF2, HDAC5 nuclear export, MEF2A binding to the Glut4 promoter, and GLUT4 expression was investigated.

Methods: 

This was investigated in muscles from AMPKα₂ overexpression (OE) mice, AMPKα₂ knockout (KO) mice, and corresponding wild-type (WT) mice that had undertaken a 28-d program of treadmill training by: 1) AMPKα-Thr¹⁷² phosphorylation by Western blot, 2) total and nuclear MEF2A by Western blot, 3) nuclear HDAC5 association with MEF2 by coimmunoprecipitation, 4) total and nuclear HDAC5 by Western blot, 5) bound MEF2A at the Glut4 MEF2 cis-element by chromatin immunoprecipitation, and 6) GLUT4 expression by real-time polymerase chain reaction and Western blot.

Results: 

OE or KO of AMPKα₂ isoform heightened or attenuated the training-induced increase in nuclear MEF2A content, Glut4 promoter-bound MEF2A. However, OE or KO of the AMPKα₂ isoform did not have any effect on the content of nuclear HDAC5 association with MEF2 after 28 d of exercise training, although 35% lower nuclear HDAC5 protein content was found in α₂-OE training muscles. Lastly, OE of the α₂-isoform was associated with 120% and 155% higher GLUT4 protein and mRNA in training muscles. However, the training-induced increases of GLUT4 protein and mRNA contents were normal in α₂-KO muscles despite the reduced AMPK signaling.

Conclusions: 

Exercise training increases the nuclear MEF2A content and binding of MEF2A to their binding sites on the Glut4 gene by an AMPKα₂-dependent mechanism, but intracellular signaling molecules other than AMPKα₂ are important in regulating training-induced HDAC5 nuclear export. Furthermore, although AMPKα₂ mediates the training-induced increase in Glut4 promoter-bound MEF2A, the present data do not support an essential role of AMPKα₂ in regulating training-induced GLUT4 expression in skeletal muscle.

Numerous studies have shown that regular exercise increases the content of lucose transporter 4 (GLUT4) protein and enhances glucose transport capacity in skeletal muscle (^10,29). Experiments using transgenic mice containing various constructs of the rat, human, or mouse Glut4 promoter, fused to a reporter gene, have shown that the myocyte enhancer factor 2 (MEF2) binding site is necessary for GLUT4 transcription in skeletal muscle (^12,22,30), and mutations of the MEF2 binding site resulted in a nearly complete loss of reporter gene expression (⁵). Likewise, a mutation of the MEF2 binding site in the Glut4 promoter region in C2C12 cells results in a loss-of-function and decreased GLUT4 mRNA expression (¹²). Using gel mobility shift assays, McGee et al. (¹⁵) found that the amount of bound MEF2A increased twofold after 60 min of exercise. Using an in vivo DNA binding assay, Smith et al. (²⁷) demonstrated that exercise increased the binding of MEF2A to the Glut4 promoter in rat skeletal muscle. In both studies, exercise also increased GLUT4 expression. Results from these studies provide strong support for the hypothesis that GLUT4 transcription after exercise is mediated by increases in the binding of MEF2 to their binding sites on the Glut4 gene. However, the mechanism involved remains to be elucidated.

Recent experiments have indicated that AMP-activated protein kinase (AMPK) might be involved in regulating MEF2A binding to the Glut4 gene. First, Zheng et al. (³⁷) found that the mRNA level of chloramphenicol acyltransferase reporter gene, which is driven by 1154 or 895 bp (both of these constructs contain MEF2 sites) of the human GLUT-4 proximal promoter, was increased in 5-aminoimidazole-4-carboxamide-1-β-d-ribofuranoside (AICAR)-treated transgenic mice and that nuclear extracts from AICAR-treated mice bound to the consensus sequence of MEF2 (from 2473 to 2464) to a greater extent than from saline-injected mice. Second, Holmes et al. (⁹) showed that, although AMPK does not seem to phosphorylate MEF2A in vitro, MEF2 binding to DNA was increased twofold at 2 h after AICAR treatment. Because AMPK activity is elevated in skeletal muscle during exercise (^7,34), it seems reasonable to speculate that the observed increase in binding of MEF2A to the Glut4 gene after exercise might be mediated by AMPK. In addition, previous study showed that increased binding of MEF2A to its consensus sequence on the Glut4 gene can occur because of increased translocation of MEF2 to the nucleus (^9,27). Evidence that exercise increases nuclear abundance of MEF2A was reported previously (¹⁵), and this would be also mediated by AMPK because MEF2 protein increased in the nucleus at 2 h after AICAR treatment (⁹).

In the basal state, MEF2 is physically associated with members of the class IIa histone deacetylases (HDAC) (^8,17), and this ensures that transcriptional coactivators and other enzymes required for transcription are denied access to the surrounding region of DNA, resulting in repression of MEF2-dependent transcription (¹⁸). Initial characterization of HDAC5, a class IIa HDAC highly expressed in skeletal muscle, showed that HDAC activity is highly controlled by phosphorylation. Specifically, phosphorylation of S259 and S498 on HDAC5 dissociates HDAC5 from MEF2 and provides binding sites for 14-3-3 chaperone proteins that export HDAC5 from the nucleus (¹⁷). Gene reporter assays using the human Glut4 promoter region tagged to the firefly luciferase gene showed that HDAC5 represses MEF2A-activated Glut4 reporter gene activity (¹⁶). After 60 min of cycling at 70% of V˙O_2peak, HDAC5 is dissociated from MEF2 and is exported from the nucleus (¹⁴). This is also associated with an increase in GLUT4 gene expression immediately after exercise. These data suggest that exercise may disrupt MEF/HDAC5 complexes and increase the accessibility of MEF2 transcription factors to their binding domains on DNA. In vitro phosphorylation assays revealed that AMPK does phosphorylate HDAC5 and that mutation of serines 259 and 498 to alanine residues blocked this effect. Furthermore, activation of AMPK in culture by AICAR results in the phosphorylation of HDAC5 and association with decreased abundance of HDAC5 found at the MEF2 binding region on the Glut4 promoter (¹⁶). Further evidence that AMPK phosphorylates HDAC5 to release HDAC5 from an HDAC5/MEF2 complex comes from observations that AMPK phosphorylation and HDAC5 Ser498 phosphorylation both increase in response to exercise (³²). However, it is still too early to establish that the observed decrease in nuclear HDAC5 association with MEF2 and increase in HDAC5 nuclear export after exercise was mediated by AMPK.

The relationship between MEF2 nucleus translocation, nuclear HDAC5 association with MEF2, HDAC5 nuclear export, MEF2A binding to the Glut4 promoter, and GLUT4 expression remains obscure. Furthermore, the role that AMPKα₂ plays in these events is not well defined. The purpose of the present study was to explore these issues, and this was investigated in muscles from AMPKα₂ overexpression (OE) mice, AMPKα₂ knockout mice (KO), and corresponding wild-type (WT) mice that had undertaken a 28-d program of treadmill training. We report here that exercise training increases binding of MEF2A to the Glut4 promoter by increasing the nuclear and total MEF2A protein and disrupting MEF/HDAC5 complexes. OE or KO of the AMPKα₂ isoform heightened or attenuated the training-induced increase in nuclear MEF2 content and MEF2A binding to the Glut4 promoter. However, OE or KO of the AMPKα₂ isoform did not have any effect on the content of nuclear HDAC5 association with MEF2 after 28 d of exercise training, although a 45% lower nuclear HDAC5 protein content was found in α₂-OE training muscles. Lastly, OE of the α₂-isoform was associated with higher GLUT4 expression in training muscles; however, the training-induced increases of GLUT4 protein and mRNA were normal in α₂-KO muscles despite the reduced AMPK signaling.

METHODS

Generating transgenic mouse with OE of AMPKα₂.

To generate the transgenic mouse with OE of AMPKα₂, the 1.7-kb cDNA encoding human PRKAA2 [GeneID: 5563] (protein kinase, AMP-activated, α₂ catalytic subunit) was ligated into the BamHI and EcoRI sites of the pcDNA3.1+ vector under the CMV promoter (Fig. 1A). A linear 7128-bp DNA fragment containing the entire CMV promoter, the complete human PRKAA2 cDNA ORF, and hGH 3′UT was released by digesting with PvuI and was microinjected into male pronuclei of fertilized mouse oocytes, which were then implanted into pseudopregnant females to generate the transgenic mouse line (¹). Genotyping was performed by genomic polymerase chain reaction (PCR) analysis of genomic DNA using the primers 5′-GCCACAAAGAAGAGCATACAAG-3′ and 5′-TGGACGGGTTGAAGAGATG-3′. The PCR was conducted using 30 cycles of the following conditions: denaturation at 94°C for 30 s, annealing at 58°C for 30 s, and extension at 72°C for 30 s to amplify the desired 307-bp fragment of the PRKAA2 transgene (Fig. 1B). The expression of the target gene was analyzed by Western blot (Fig. 1C) using antibodies to PRKAA2 (Santa Cruz Biotechnology, Santa Cruz, CA). The transgenic mice with OE of AMPKα₂ (AMPKα₂-OE) were maintained on a C57BL/6J genetic background.

FIGURE 1: Generation of transgenic mice. The PRKAA2 transgenic construct was generated by inserting the target genes under the control of CMV promoter (A) and the transgenic mice were created by microinjection. The transgenic mice were screened by genomic PCR for the presence of PRKAA2 gene (B) with a 307-bp PCR product, and lanes 6, 7, and 8 were the transgenic mice. Lane 4 is the molecular weight marker; lane 1, blank control; lane 2, negative control; lane 3, positive control of the transgenic construct DNA. Lanes 5, 9, 10, and 11 are the nontransgenic mice. The two lines, F₀33 and F₀24, with PRKAA2 OE were selected by Western blot (C).

Animals.

The study protocol adhered to the American College of Sports Medicine's animal care standards and was approved by the Animal Care and Use Committee of Beijing Sport University. WT mice were provided by the Institute of Laboratory Animal Science of Peking Union Medical College. Transgenic mice with AMPKα₂-OE were generated as indicated. AMPKα₂ KO mice specimen were kindly provided by Benoit Viollet (Department of Endocrinology, Metabolism and Cancer, Institute Cochin, University Paris Descartes, Paris, France) and bred by the Institute of Laboratory Animal Science of Peking Union Medical College. All mice with a mean body weight of 20 ± 2 g were housed with controlled room temperature and lighting (20°C-25°C and 12:12-h light-dark cycle) and free access to food and water.

Experimental protocol.

After allowing acclimatization to their housing and the treadmill, WT mice (n = 20), AMPKα₂ OE mice (n = 20), and AMPKα₂ KO mice (n = 20) were randomly subdivided into two groups: the control group (C) and training group (T), with 10 mice for each group. Mice of training groups ran on the treadmill at the speed of 12 m·min⁻¹ with a slope of 0°, 60 min·d⁻¹ for 28 d (⁶). Mice were anesthetized with an intraperitoneal injection of pentobarbital, 5 mg·100 g⁻¹ of body weight, at 12 h after the last training. To verify that running increased muscle AMPK activity, an additional acute exercise group (E; 10 mice for each gene-type mice) was performed. Mice were kept in the same environment as the control group and were killed immediately after they completed 60 min of continuous running (same speed as the training group) on the experimental day. Quadriceps femoris muscles were excised, cleaned of blood and connective tissue, quick-frozen with aluminum tongs, precooled in liquid nitrogen, and stored at −80°C.

Real-time PCR.

Total RNA was isolated from ∼30 mg of crushed muscle tissue using the TRI regent according to the manufacturer's instructions. Reverse transcription (RT) of total RNA to cDNA was performed using the AMV Reverse Transcriptase Kit (A3500; Promega, Madison, WI) in a Perkin-Elmer DNA Thermal Cycler (Perkin-Elmer, Applied Biosystems, Foster City, CA). First-strand cDNA was synthesized from 1 μg of total RNA in a 20-μL reaction using random hexamers as primers. RT-PCR was performed in a GeneAmp 7300 Real-Time PCR System using the GeneAmp SYBR^® Green Kit (Perkin-Elmer, Applied Biosystems) with the previously synthesized cDNA as template in a 20-μL reaction. A no-template control was included in all experiments. Forward and reverse primers complementary to the mouse Glut4 gene (GenBank NM_009204) were designed using Primer Express software (Perkin-Elmer, Applied Biosystems). The GLUT4 forward primer sequence (5′ to 3′) was CTT GGC TCC CTT CAG TTT GG, whereas the reverse primer sequence was CTA CCC AGC CAC GTT GCA TT. The expression of GLUT4 was compared with the expression of β-actin, which was continuously expressed at constant amounts in cells. The primers for β-actin were AGG CAA ACC GTG AAA AGA TG (forward) and CAC AGC CTG GAT GGC TAC GT (reverse). The difference in expression between control and experimental was calculated as 2^−ΔΔCT, as described previously (¹³).

Chromatin immunoprecipitation assays.

Approximately 50 mg of frozen triceps muscle was ground in liquid nitrogen and cross-linked using 1% formaldehyde. Chromatin was sheared to fragments 300-800 bp by sonication and centrifuged, and the supernatant was precleared with salmon sperm DNA/protein A agarose to yield input sample. Chromatin from 250 μL of the input sample was immunoprecipitated by incubating with protein agarose A and antibodies directed against MEF2A or with immunoglobulin G (IgG). Precipitated complexes were reverse cross-linked in 0.5 mol·L⁻¹ NaCl at 65°C for 6 h, and the coimmunoprecipitated DNA was purified by phenol-chloroform extraction and resuspended in dH₂O. A 270-bp fragment corresponding to nucleotides −336 to −604 of the mouse Glut4 promoter containing the MEF2 binding site was amplified by real-time PCR using the following primers: 5′-CAG GCA TGG TCT CCA CAT ACA C-3′ (forward) and 5′-GGT AAC TCC AGC AGG ATG ACA-3′ (reverse). The following pair of primers was used as a negative control: 5′-GACGGACACCTTCTCTCTTAGC-3′(forward) and 5′-CCACAGCCTAGCCACAACAC-3′ (reverse). These primers amplify a 283-bp fragment corresponding to nucleotides −4620 to −4903 relative to the start of transcription, which does not contain the MEF2 binding sequence. DNA from 10 μL of input sample that did not undergo chromatin immunoprecipitation (ChIP), but was reverse cross-linked and purified as described above, was also PCR-amplified using the same set of primers. The cycle threshold (CT) of MEF2 form MEF2 ChIP was subtracted from the CT of MEF2 form input giving the ΔCT. The value of the control ΔCT (WT) was set to 1, and the control ΔCT (WT) was subtracted from the experimental ΔCT yielding the ΔΔCT. The difference between control and experimental was calculated as 2^−ΔΔCT, as described (¹³).

Immunoblot analysis.

Total proteins were isolated from ∼30 mg of muscle using T-PER tissue protein extraction reagents (Pierce, Rockford, IL). Nuclear proteins were isolated from ∼40 mg of muscle using NE-PER nuclear and cytoplasmic extraction reagents (Pierce). Protein concentration was measured using the BCA protein assay kit (Pierce). After SDS-PAGE, immunoblotting was done using the following primary antibodies: anti-AMPKα-Thr¹⁷² phospho (Cell Signaling Technology, Beverly, MA), anti-AMPKα₂ (Santa Cruz Biotechnology, Santa Cruz, CA), anti-MEF2A (Abcam, Cambridge, UK), anti-HDAC5 (Cell Signaling Technology), and anti-GLUT4 (Santa Cruz Biotechnology). Secondary antibodies used were all species-specific horseradish peroxidase-conjugated immunoglobulins (Cell Signaling Technology, Danvers, MA). Immunoreactive bands were highlighted by electrochemiluminescence (ECL) technology, exposed to light-sensitive film for 15-60 s, and quantified by densitometry using image analysis software (Kodak Digital Science, New York, NY). To confirm equal protein loading and transfer, membranes were stripped and reprobed with anti-β-actin or anti-Histone H1 (Santa Cruz Biotechnology). The individual values were originally expressed as a percentage of a standard (β-actin/Histone H1 content) and then expressed as a percentage of the control group (WT) value.

Coimmunoprecipitation.

Nuclear protein (500 μg) was made up to 500 μL in immunoprecipitation wash buffer (50 mmol·L⁻¹ Tris, pH 7.5, 1 mmol·L⁻¹ EDTA, 1 mmol·L⁻¹ EGTA, 10% glycerol, 1% Triton X-100, 50 mmol·L⁻¹ NaF, 5 mmol·L⁻¹ Na pyrophosphate, 1 mmol·L⁻¹ dithiothreitol, and 1 mmol·L⁻¹ phenylmethylsulfonyl fluoride). Samples were precleared with 50 μL of protein A Sepharose beads (Santa Cruz Biotechnology) before being incubated with 5 μL of anti-MEF2 (Santa Cruz Biotechnology) antibody overnight at 4°C. Samples were again incubated with 50 μL of protein A Sepharose beads for 2 h while rotating at 4°C. The Sepharose-bound immune complex was pelleted by centrifugation and washed four times with 1 mL of immunoprecipitation wash buffer. After boiling in SDS sample buffer for 5 min, immune complexes were resolved by SDS-PAGE, and the amounts of HDAC5 present in the complex were determined by probing with HDAC5 antibody. A control lane of beads only was included on all gels to ensure that the signals observed were specific to the immune complex and not the beads themselves. Membranes were reprobed with the MEF2 antibody to ensure that equal amounts of MEF2 protein had been pulled down.

Statistical analysis.

All values are reported as means ± SE. Mean differences from each experiment were analyzed by Student's t-test or one- or two-way ANOVA, and a Tukey post hoc test was used when significance was found. Statistical significance was set at P ≤ 0.05.

RESULTS

Exercise increases AMPKα-Thr¹⁷² phosphorylation (P) in muscle.

The acute experiments were performed to verify that AMPK was activated by the interventions and that the α₂-KO or α₂-OE was associated with a changed AMPK signaling. Mice were killed immediately after they completed 60 min of continuous running on the experimental day. As an indication of AMPK activity, the content of AMPKα that was phosphorylated at Thr¹⁷² was measured by Western blot, using phosphospecific antibodies. AMPKα phosphorylation was increased in response to running by 40% compared with nonexercised controls (Fig. 2B). OE of the α₂-isoform, which resulted in a higher signal than that in WT mice when using the α₂-antibody in muscle lysates (Fig. 2A), was associated with 150% higher AMPKα-P in exercised muscles (Fig. 2B). KO of the α₂-isoform, which resulted in no detectable signal when using the α₂-antibody in muscle lysates (Fig. 2A), was associated with 56% and 38% lower AMPKα-P activity in resting and exercised muscles, respectively, compared with the WT muscles (Fig. 2B). Collectively, these results show that exercise-induced increase in AMPKα phosphorylation can be effectively affected by OE and KO of the AMPKα₂-isoform.

FIGURE 2: AMPKα₂ protein content (A) and AMPKα-Thr¹⁷² phosphorylation (P) activity (B) in muscle lysates of WT, α₂-OE, and α₂-KO mice after 60 min of treadmill running as well as in corresponding controls. OE of the α₂-gene resulted in higher signal than that in WT mice, and KO of the α₂-gene resulted in no detectable signal when using the α₂-antibody in muscle lysates. Protein phosphorylation was assessed by immunoblotting, using phosphospecific antibodies. n = 8-10. *Significantly different from the control group (P < 0.05). **Significantly different from the control group (P < 0.01). ##Significantly different from the WT group (P < 0.01). Data are presented as means ± SE relative to WT control.

Exercise training increases MEF2A binding to the Glut4 promoter by AMPKα₂-dependent mechanism.

ChIP assays were used to measure the amount of MEF2A that was bound to the Glut4 promoter. DNA-bound MEF2A was coimmunoprecipitated with an antibody against MEF2A and PCR-amplified using primers spanning the MEF2 site on the Glut4 promoter. When the assay was conducted with a nonspecific (IgG) antibody (Figs. 3B, C) or when primers that amplify a region in the Glut4 gene that does not contain the MEF2 site (Fig. 3D) were used, no signal was obtained. These control experiments demonstrate the specificity of the ChIP assays in assessing the amount of MEF2A that is bound to its cis-element on the Glut4 promoter. MEF2A that was bound to the Glut4 promoter was 2.7-fold higher after 28 d of exercise training compared with controls (Fig. 3A). This result demonstrates that the exercise protocol was of sufficient intensity to activate the signaling pathways that cause Glut4 promoter-bound MEF2A up-regulation. OE of the α₂-isoform was associated with 171% higher Glut4 promoter-bound MEF2A in training muscles (Fig. 3A). KO the α₂-isoform significantly reduced but did not abolish the training-induced increase in Glut4 promoter-bound MEF2A, which remained twofold higher compared with controls (Fig. 3A). These results demonstrate that increased binding of MEF2A to the Glut4 promoter after exercise requires AMPK activation.

FIGURE 3: A, The amount of MEF2A that was bound to the Glut4 promoter in WT, α₂-OE, and α₂-KO mice muscles after 28 d of exercise training as well as in corresponding controls. This was assessed by ChIP assays. **Significantly different from control group (P < 0.01). ##Significantly different from WT group (P < 0.01). Data are presented as means ± SE relative to WT control. B-D, Control experiments. B, Gels showing PCR products when an anti-MEF2A antibody or a mouse IgG was used. Chromatin from an aliquot of sample that did not undergo immunoprecipitation was also PCR-amplified and used for comparison (Input). C and D, PCR, using primers that amplify a region in the Glut4 gene that does not contain an MEF2 cis-element (C) and primers that amplify the MEF2 site on the Glut4 promoter (D), was performed on DNA that was coimmunoprecipitated with an anti-MEF2A antibody (MEF2A ChIP) or from Input sample that did not undergo immunoprecipitation (Input).

Increased nuclear MEF2A protein after exercise training occurs via an AMPKα₂-dependent mechanism.

To further investigate the mechanism by which exercise increases the binding of MEF2A to the Glut4 gene, the content of total and nuclear MEF2A was measured by Western blot. Twenty-eight days of exercise training increased total and nuclear MEF2A by 34% and 55% in WT muscles, respectively, compared with controls (Figs. 4A, B). OE of the α₂-isoform was associated with 120% higher nuclear MEF2A protein in training muscles, and KO of the α₂-isoform was associated with 23% lower nuclear MEF2A protein in training muscles (Fig. 4B). However, the total MEF2A protein was normal in α₂-OE/KO muscles despite the changed AMPK signaling (Fig. 4A). These results imply that the increased nuclear MEF2 after exercise training occurs via an AMPKα₂-dependent mechanism.

FIGURE 4: Total (A) and nuclear MEF2 (B) protein content in WT, α₂-OE, and α₂-KO mice muscles after 28 d of exercise training as well as in corresponding controls. Protein content was assessed by immunoblotting. n = 8-10. **Significantly different from the control group (P < 0.01). #Significantly different from the WT group (P < 0.05). Data are presented as means ± SE relative to the WT control.

Coimmunoprecipitation analysis confirms that intracellular signaling molecules other than AMPKα₂ are important in regulating training-induced HDAC5 nuclear export.

HDAC5 association with MEF2 decreased 35% after 28 d of exercise training, suggesting that HDAC5 does regulate MEF2 in human skeletal muscle and is sensitive to exercise (Figs. 5C, D). Also observed was a 27% decrease in nuclear HDAC5 protein (Figs. 5B, D), whereas there was no change in total HDAC5 protein (Figs. 5A, D), suggesting that HDAC5 was exported from the nucleus. Although OE of the α₂-isoform was associated with 35% lower nuclear HDAC5 protein content (Figs. 5B, D), nuclear HDAC5 association with MEF2 (Figs. 5C, D) and total HDAC5 (Figs. 5A, D) were not different between α₂-OE and WT mice after 28 d of exercise training. The total HDAC5 protein (Figs. 5A, D), nuclear HDAC5 association with MEF2 (Figs. 5C, D), and nuclear HDAC5 protein contents (Figs. 5B, D) were normal in α₂-KO muscles despite the changed AMPK signaling. These results imply that intracellular signaling molecules other than AMPK are important in regulating training-induced HDAC5 nuclear export.

FIGURE 5: Content of total (A)/nuclear MEF2 (B) protein, MEF2-associated HDAC5 (C) in WT, α₂-OE, and α₂-KO mice muscles after 28 d of exercise training as well as in corresponding controls. Protein content was assessed by immunoblotting. MEF2-associated HDAC5 was measured by Coimmunoprecipitation. n = 8-10. *Significantly different from the control group (P < 0.05). **Significantly different from the control group (P<0.01). #Significantly different from the WT group (P < 0.05). Data are presented as means ± SE relative to the WT control. D. Representative blot. IB, immunoblot; IP, immunoprecipitate.

AMPKα₂ is not indispensable in regulating the training-induced GLUT4 expression.

Twenty-eight days of exercise training increased GLUT4 protein and mRNA by 48% and 100% in WT muscles, respectively, compared with controls (Figs. 6B, A). As expected, OE of the α₂-isoform was associated with 120% and 155% higher GLUT4 protein and mRNA in training muscles (Figs. 6B, A). However, the training-induced increases of GLUT4 protein and mRNA were normal in α₂-KO muscles despite the reduced AMPK signaling (Figs. 6B, A). Collectively, these results demonstrate that AMPKα₂ is not indispensable in regulating the training-induced GLUT4 expression in mouse skeletal muscle.

FIGURE 6: GLUT4 mRNA (A) and protein (B) content in WT, α₂-OE, and α₂-KO mice muscles after 28 d of exercise training as well as in corresponding controls. GLUT4 mRNA was measured by RT-PCR, and GLUT4 protein was measured by Western blot. **Significantly different from the control group (P < 0.01). #Significantly different from the same WT group (P < 0.05). ##Significantly different from the WT group (P < 0.01). Data are presented as means ± SE relative to the WT control.

DISCUSSION

Regular exercises may protect against the development of type 2 diabetes or delay its onset in individuals who are genetically predisposed to the disease, partly because it increases the content of GLUT4 protein in the skeletal muscle (^10,29). Studies of the Glut4 promoter have clearly demonstrated that GLUT4 expression is regulated by multiple transcription factors, including MEF2 (^12,22,31). Three isoforms of the MEF2 protein, namely, MEF2A, MEF2C, and MEF2D, are expressed in the skeletal muscle (^3,19). Previous studies have revealed that MEF2A/MEF2D heterodimer binds to the MEF2 binding site (^12,19,30) and that the amount of bound MEF2A increases after an acute bout of exercise (^15,27). In the present study, the amount of bound MEF2A was also threefold higher 12 h after the last training session compared with controls. Although these observations provide strong evidences that exercise/exercise training increases the binding of MEF2A to its cis-element on the Glut4 promoter, little is known about the intracellular signaling pathways involved in eliciting this response. The observations that the AMPK is activated in skeletal muscle during exercise (^7,34) and stimulation of AMPK with AICAR increases the binding of MEF2A to the Glut4 gene (⁹) suggest that AMPK is a likely candidate. To explore whether AMPKα₂ was involved in the regulation of MEF2A binding to the Glut4 promoter after exercise training, we have used AMPKα₂-OE/KO mice. In characterizing these mice, OE mice showed higher signal than that in WT mice, and KO mice showed no detectable signal when using the α₂-antibody in muscle lysates. OE, KO, and WT mice underwent a 28-d program of treadmill training, and the content of DNA-bound MEF2A was measured by ChIP. Findings indicate that OE of the α₂-isoform was associated with 170% higher and KO of the α₂-isoform was associated with 61% lower Glut4 promoter-bound MEF2A in training muscles, implying that exercise training increases MEF2A binding to the Glut4 promoter through a mechanism that is mediated by AMPKα₂.

A previous study has shown that increased binding of MEF2A to its consensus sequence on the Glut4 gene can occur because of increased translocation of MEF2 to the nucleus, with or without a modest increase in total MEF2A content. A recent study demonstrated that an acute bout of exercise increases nuclear abundance of MEF2A in human skeletal muscle (¹⁵). In the study of Smith et al. (²⁷), total MEF2A levels after an acute bout of exercise were not elevated at the time when binding increased. Collectively, these experiments suggest that the acute exercise-induced increases of Glut4 promoter-bound MEF2A were the result of increased nuclear MEF2A content but not the total MEF2A content. Because total MEF2A content was also increased after 4 wk of exercise training, our data favored the hypothesis that, unlike in acute exercise, training-induced increases of Glut4 promoter-bound MEF2A were the result of both the increased total and nuclear MEF2A content. In the present study, OE of the α₂-isoform was associated with 120% higher and KO the α₂-isoform was associated with 23% lower nuclear MEF2A protein in the training muscles. The total MEF2A protein was normal in α₂-OE/KO muscles, indicating that the increased translocation of MEF2A to the nucleus after exercise training occurs via an AMPKα₂-dependent mechanism. This conclusion is supported by previous findings from the laboratory of Holmes et al. (⁹) that MEF2 protein increased in the nucleus at 2 h after AICAR treatment.

The increased binding of MEF2A to the Glut4 promoter after exercise may also be because of increased accessibility of the transcription factor to their binding sites. In the basal state, MEF2 is physically associated with members of the class IIa HDAC (^8,17). Previous studies show that exercise may disrupt MEF/HDAC5 complexes and cause nuclear export of the deacetylase (¹⁴). The liberated MEF2 is then able to associate with cofactors having histone acetyltransferase activity, which modify histone tails and cause chromatin relaxation (³⁵). These modifications would conceivably increase the accessibility of MEF2 transcription factors to their binding domains on DNA to increase MEF2/DNA interactions (²). Consistent with this theory, we observed that HDAC5 association with MEF2 decreased 35% after 28 d of exercise training. Also observed was a 27% decrease in nuclear HDAC5 protein, whereas there was no change in total HDAC5 protein, suggesting that HDAC5 was exported from the nucleus. In vitro phosphorylation assays established that AMPK does phosphorylate HDAC5 and that S259 and S498 were required for this response. Furthermore, this seems to be a functional relationship because AMPK and HDAC5 associate in vivo and treatment of human primary myotubes with AICAR induces HDAC5 nuclear export. In addition, the increase in AMPK activity associated with AICAR treatment decreased the abundance of HDAC5 found at the MEF2 binding region on the GLUT4 promoter (¹⁶). Collectively, these data suggest that AMPK is an HDAC5 kinase. However, in the present study, OE or KO of the AMPKα₂-isoform did not have any effect on the content of nuclear HDAC5 association with MEF2 after 28 d of exercise training, although 35% lower nuclear HDAC5 protein content was found in α₂-OE training muscles. The present data do not support an essential role of AMPKα₂ in regulating training-induced HDAC5 nuclear export and we reasoned that the observed decrease in nuclear HDAC5 association with MEF2 was likely the result of other factors.

Mukwevho et al. (²⁰) reported that incubated C2C12 myotubes with caffeine reduced nuclear HDAC5 content, and this effect was attenuated in the inclusion of dantrolene or KN93 together with caffeine. Furthermore, using immunocytochemical analysis, they found that incubation with caffeine caused a decrease in intranuclear but an increase in perinuclear HDAC5 density. Also, preincubation with KN93 or dantrolene before caffeine administration blocked the effect of caffeine. These results indicate that caffeine causes HDAC5 export from the nucleus via a calcium-dependent CaMKII mechanism. In addition to CaMKII, recent reports found that protein kinase D (PKD), which was renamed when it was found that its catalytic domain shares high homology with the CAMK (²⁵), directly phosphorylates HDAC5, and PKD activity in contracted skeletal muscle is higher in AMPK α₂ kinase dead (KD) mice (⁴). Because CAMK II (^24,27) activity is increased in the skeletal muscle during exercise, and PKD is potently activated in response to oxidative stress (²⁵), it seems reasonable to speculate that the observed decreased nuclear HDAC5 association with MEF2 might be mediated by CaMKII and PKD. However, we did not measure the CAMK II and PKD activity. As stated above, the increased binding of MEF2A to the Glut4 promoter after exercise may be due to increased accessibility of the transcription factor to its binding sites. In this case, the observation that KO of the AMPKα₂-isoform did not completely prevent the increase in binding of MEF2A to the Glut4 promoter by exercise training was not surprising because the nuclear HDAC5 association with MEF2 and nuclear HDAC5 protein contents in KO mice was significantly decreased after 28 d of exercise training.

It is now well established that GLUT4 transcription after exercise is mediated by increases in the binding of MEF2 to its binding sites on the Glut4 gene (^5,15,27). In the present report, although OE of AMPKα₂, which was associated with higher AMPKα-P in exercised muscles, heightened the training-induced increase in MEF2A binding to the Glut4 promoter and GLUT4 expression, the finding that training-induced increases of GLUT4 protein and mRNA contents were normal in α₂-KO muscles despite the reduced AMPK signaling and Glut4 promoter-bound MEF2A provides evidence that AMPKα₂ is not indispensable in regulating training-induced GLUT4 expression in the skeletal muscle. There is evidence that CaMK is also required in regulating GLUT4 expression. A study by Ojuka et al. (²¹) showed that GLUT4 content in L6 myotubes increased when intercellular calcium was raised using caffeine and that the caffeine-induced increase in GLUT4 was removed when the CaMK inhibitor, KN93, was included in the medium. However, Smith et al. (²⁸) recently demonstrated that the exercise-induced increases in CaMKII phosphorylation, MEF2A binding, and GLUT4 expression were attenuated or abolished when KN93 was administered to rats before exercise and concluded that CaMKII activation by exercise increases GLUT4 expression via the increased accessibility of MEF2A to its cis-element on the gene. Although we did not measure the CaMKII activity, the finding that Glut4 promoter-bound MEF2A in KO training muscles was significantly lower than that in WT mice suggests that the observed increase in GLUT4 expression in AMPKα₂ KO mice was unlikely mediated by CaMKII because regulation by CaMKII would be expected to generate a normal Glut4 promoter-bound MEF2A content in AMPK α₂ KO mice. For the same reason, the possibility that p38 involved in this process was also ruled out. P38, which is activated by exercise, has been shown to phosphorylate threonine residues in the transactivation domains of MEF2A and increase its transcriptional activity (^14,26,33,36). Clearly, further studies are needed to elucidate the role of other signaling molecules, including in GLUT4 expression. Indeed, many studies have indicated that the Glut4 promoter is also regulated by additional transcription factors such as GEF (^11,23). Other signals that regulate this transcription factor may therefore also increase GLUT4 expression.

In summary, we have provided data showing that exercise training increases binding of MEF2A to the Glut4 promoter by increasing the nuclear and total MEF2A protein and disrupting MEF/HDAC5 complexes. In addition, the finding that nuclear MEF2A protein content and binding of MEF2A to the Glut4 promoter increased after exercise training in a manner that were both AMPK activity-dependent supports the hypothesis that exercise training increases the nuclear MEF2A content and binding of MEF2A to their binding sites on the Glut4 gene via an AMPKα₂-dependent mechanism. However, the finding that HDAC5 association with MEF2 was normal in α₂-OE/KO training muscles despite the changed AMPK signaling may suggest that intracellular signaling molecules other than AMPKα₂ are important in regulating training-induced HDAC5 nuclear export. Lastly, training-induced GLUT4 expression was normal in α₂-KO muscles despite the reduced AMPK signaling and Glut4 promoter-bound MEF2A suggests that AMPKα₂ is not indispensable in regulating training-induced GLUT4 expression in the skeletal muscle.

This study was supported by the National Natural Science Foundation of China (30971412), the Industry Foundation of Ministry of Health, China (200802036), and the Natural Science Foundation of Beijing (5102024).

The authors thank Mr. Benoit Viollet from the Department of Endocrinology, Metabolism and Cancer, Institute Cochin, University Paris Descartes, France, for being so kind for providing us two AMPKα₂ knockout mice. In addition, the technical assistance of Mr. Zhang Lianfeng at the Institute of Laboratory Animal Sciences, Chinese Academy of Medical Sciences & Peking Union Medical College, is greatly appreciated.

The results of the present study do not constitute endorsement by the American College of Sports Medicine.