PPARβ Activation Induces Rapid Changes of Both AMPK Subunit Expression and AMPK Activation in Mouse Skeletal Muscle

AMP-activated protein kinases (AMPK) are heterotrimeric, αβγ, serine/threonine kinases. The γ3-AMPK subunit is particularly interesting in muscle physiology because 1) it is specifically expressed in skeletal muscle, 2) α2β2γ3 is the AMPK heterotrimer activated during exercise in humans, and 3) it is down-regulated in humans after a training period. However, mechanisms underlying this decrease of γ3-AMPK expression remained unknown. We investigated whether the expression of AMPK subunits and particularly that of γ3-AMPK are regulated by the PPARβ pathway. We report that PPARβ activation with GW0742 induces a rapid (2 h) and sustained down-regulation of γ3-AMPK expression both in mouse skeletal muscles and in culture myotubes. Concomitantly, phosphorylation levels of both AMPK and acetyl-coenzyme A carboxylase are rapidly modified. The γ3-AMPK down-regulation is also observed in muscles from young and adult transgenic mice with muscle-specific overexpression of peroxisome proliferator-activated receptor β (PPARβ). We showed that γ3-AMPK down-regulation is a rapid physiological muscle response observed in mouse after running exercise or fasting, two situations leading to PPARβ activation. Finally, using C2C12, we demonstrated that dose and time-dependent down-regulation of γ3-AMPK expression upon GW0742 treatment, is due to decrease γ3-AMPK promoter activity.

Peroxisome proliferator-activated receptor β (PPARβ, also called PPARδ) is a nuclear receptor, and it is the predominant PPAR isoform in rodent and human skeletal muscle (for review see Refs. 1 and 2). A large body of evidence demonstrated that PPARβ plays a central role in the adaptive responses of skeletal muscle to metabolic challenges. In muscle, its expression is positively regulated by fasting and physical exercise (3, 4). Furthermore, PPARβ agonist treatment is beneficial for many metabolic parameters in rodents, primates, and humans, probably by increasing muscle fatty acid burning (for review see Ref. 5). We reported the effects of either overexpression or pharmacological activation of PPARβ in mouse tibialis anterior (TLA) muscle (6, 7) (and for review see Ref. 2). In both models, we have described a muscle oxidative remodeling characterized by hyperplasia, increase in myonuclear density, and augmentation in oxidative fiber and capillary numbers (6, 8, 9). Moreover, mice overexpressing PPARβ are resistant to a high-fat diet (10). In mice treated with the PPARβ agonist GW0742, these modifications are already detectable after 24 h, and the muscle remodeling is achieved after 48 h (6, 8). The PPARβ-induced oxidative muscle remodeling requires an active calcineurin/nuclear factor of activated T-cells pathway because it is blunted by cyclosporine A cotreatment (6, 8). Because PPARβ activation leads to an angiogenic and oxidative muscle remodeling accompanied by an increased muscle fatty acid burning, it has been proposed that PPARβ activation mimics, at least in part, the muscle adaptive responses to aerobic training exercise (7, 11–13). Recently, Narkar et al. (14) have shown that PPARβ interacts with the AMP-activated protein kinases (AMPK) and that expression of a constitutive form of PPARβ in transgenic mice leads to constitutive AMPK activation.

AMPK (for review see Refs. 15–17) are major players of the muscle metabolic flexibility (18), affecting in some cases fiber typing (19). Indeed, AMPK are 1) activated when the level of intracellular energy is low (increased AMP/ATP ratio) and are thus supposed to play a role of fuel gauge to protect muscle cells against energy deprivation during physical exercise or during other metabolic stress, such as hypoxia (for review see Ref. 20); 2) capable of switching on and off both energy-consuming and -producing pathways, thus controlling the levels of both use and storage of energy substrates (15, 16); 3) sensitive to the quantity of glycogen in the muscular cells (21, 22); and 4) necessary for fiber shift in response to training exercise (19). AMPK are serine/threonine protein kinases and exist as heterotrimeric complexes comprising a catalytic α-subunit and regulatory β- and γ-subunits. Each subunit is expressed as two or more isoforms encoded by distinct genes (23). The α-subunits (α1 and α2) catalytic activities are increased after phosphorylation at the Thr-172 residue within the activation loop (24). Phosphorylation is performed by upstream kinases such as liver kinase B1 and calcium/calmodulin kinase kinase α/β (25–30). The β-subunits (β1 and β2) contain two conserved domains, one required for assembly of the α/β/γ heterotrimer and one that may target AMPK to glycogen particles (31, 32). The γ-subunits (γ1, γ2, and γ3) carry two Bateman domains (33, 34), which bind the regulatory nucleotides AMP and ATP. AMP binding promotes AMPK α-Thr-172 phosphorylation by an allosteric activation of the kinase (33, 35). Of the 12 potential AMPK trimers, only three active heterotrimers, namely α1β2γ1, α2β2γ1, and α2β2γ3 can be evidenced in human skeletal muscle cells (36, 37).

The γ3-AMPK subunit is particularly interesting in muscle physiology. It is the only AMPK subunit expressed specifically in skeletal muscle and in particular in fast and glycolytic fibers (IIb > IIa ≫ I) (38). The γ3-AMPK knockout mice (Prkag3^−/−) are fatigue prone (21, 39). Furthermore, in these animals, specific AMPK responses and functions are lost, suggesting that the other AMPK trimers cannot compensate for γ3-containing trimers. For example, increased glucose uptake in response to the AMPK activator, 5-aminoimidazole-4-carboxamide-1-β-d-ribofuranoside (AICAR), is lost as well as AMPK activation in response to low glycogen concentration (21, 39). Glucose uptake increment in response to hypoxia is also largely down-regulated (40). Surprisingly, overexpression of the γ3-AMPK subunit has only minor consequences in mice, especially in resting conditions (39, 41). Birk and Wojtaszewski (42) have shown recently that α2β2γ3 is the heterotrimer of AMPK, which is activated in human skeletal muscle during short physical exercise, and its activation leads to an increase in fatty acid oxidation through a direct acetyl-coenzyme A carboxylase (ACC) phosphorylation. The paradox of the γ3-AMPK subunit is that its expression is largely down-regulated in normal, obese, or type 2 diabetic men after a few weeks of either endurance or strength training (36, 37). In the same time, γ1-AMPK expression goes up (36, 37). Thus, it was concluded that a switch in abundance from the α2β2γ3 to the α2β2γ1 complexes is occurring in addition to an increased abundance of α1β2γ1 complexes. These results are reinforced by a recent clinical study, on 91 nondiabetic and diabetic twin pairs, showing that individuals with higher aerobic capacity have less γ3-AMPK mRNA, protein expression, and activity independently of sex, age, and obesity (43). In addition, the authors reveal a role of γ3-AMPK containing complexes in down-regulation of insulin-stimulated nonoxidative glucose metabolism possibly through inhibition of glycogen synthase activity (43). It remains unclear whether γ3-AMPK down-regulation is due to a negative feedback on its expression due to repeated α2β2γ3-AMPK activation or is representative of an oxidative muscle remodeling due to regular exercise or is due to other factors.

In this study, we investigated whether the expression of AMPK subunits and particularly that of γ3-AMPK is regulated by the PPARβ pathway, which plays a crucial role in skeletal muscle adaptation to physical exercise. Our data reveal new links between the PPARβ and AMPK pathways that may explain, at least in part, why PPARβ activation induces such rapid (48 h) oxidative muscle remodeling.

Results

PPARβ activation induces a switch between γ3 and γ1-AMPK subunits in mouse skeletal muscles

We have previously shown that a 48-h treatment using the PPARβ agonist GW0742 was sufficient to induce an endurance-like oxidative muscle remodeling (6, 8). In this study, we investigated whether the expression of AMPK subunits are regulated by the PPARβ pathway. We first focused on the subunits composing the three active AMPK heterotrimers in skeletal muscles, namely α1β2γ1, α2β2γ1, and α2β2γ3 (36, 37). Thus, we have determined the expression of the AMPK subunits α1 (Prkaa1), α2 (Prkaa2), β2 (Prkab2), γ1 (Prkag1), and γ3 (Prkag3) in mouse TLA muscles after different times of GW0742 treatment. The expression level of a well-known PPARβ target gene, namely pyruvate dehydrogenase kinase isozyme 4 (PDK4) was also measured to follow PPARβ activation.

As depicted in Fig. 1, standard variations were relatively low in the control group (n =4; injected with vehicle), indicating that the expression level of the different AMPK subunits was fairly homogeneous in TLA muscles. However, statistically relevant changes could be observed after pharmacological activation of PPARβ. In particular, upon PPARβ pharmacological activation, we have observed a switch between γ3- and γ1-AMPK expression. Indeed, γ3-AMPK subunit expression dropped by 40% as soon as 2 h after GW0742 treatment (P = 0.046) and reached a −50% decrease at 5 h, which is maintained at least until the end of the PPARβ agonist treatment (Fig. 1; P < 0.05 at all periods of the time course). The same down-regulation of γ3-AMPK expression was also observed when using RNA from other skeletal muscles of mouse treated with GW0742, i.e. plantaris and vastus lateralis muscles (Table 1). In parallel, γ1-AMPK expression was up-regulated. However, the 18% induction at 24 h did not reach the level of significant variation (P = 0.087). It becomes significant at 48 h (+36%, P = 0.01) after GW0742 treatment (Fig. 1). Western blot analyses confirmed, at the protein level, a switch between γ3 (−20, −31, and −47%) and γ1 (+16, +47, and +55%) AMPK subunits in TLA muscles from mice treated with GW0742 for 8, 24, and 48 h, respectively (Fig. 2, A and B). PPARβ activation using another specific agonist, the GW1516, induced a similar γ3-AMPK down-regulation (data not shown). Thus, a 2-d GW0742 treatment in mice induced almost the same variation pattern of γ1 and γ3-AMPK expression than in skeletal muscles of trained vs. untrained leg in men (Table 1) (36, 37).

Fig. 1.

PPARβ activation induces a rapid switch between γ3- and γ1-AMPK in GW0742-treated mice. Relative quantification of α1 (Prkaa1), α2 (Prkaa2), β2 (Prkab2), γ1 (Prkag1), and γ3 (Prkag3) AMPK subunits and of one PPARβ target gene, namely PDK4, using qPCR analysis and RNA from TLA muscles of 10-wk-old C57BL6 mice treated with GW0742 (1 mg/kg·d) for the indicated times (n = 4 per group) and compared with the control group (Ct treated with vehicle; n = 4). Results were normalized against the 36B4 housekeeping gene. Significant transcriptional repression of γ3-AMPK is observed at all times of GW0742 treatment. Values are means ± sd. *, P < 0.05; **, P < 0.005 when compared with the value of the corresponding AMPK subunit in control vehicle-treated groups (Ct, n = 4).

Fig. 2.

PPARβ activation induces a rapid switch between γ3- and γ1-AMPK protein levels in GW0742-treated mice. a, Representative Western blot analyses of γ1-AMPK, γ3-AMPK, and P85 (loading control) using protein lysates from TLA muscle of mice treated with GW0742 (GW) for the indicated period of time. b, After densitometric analysis, results were normalized against the loading control (P85), pooled, and presented as histograms. Values are means ± sd. *, P < 0.05, when compared with the value of the corresponding AMPK subunit in control groups (vehicle-treated mice) represented by a dotted line.

Table 1.

Comparative variations of γ3- and γ1-AMPK subunit expression

	Human (see Ref. 37), strength training		Mouse
			GW0742 (48 h), n = 4						PPARβ Tg, n = 4 Vast RNA	One night run, n = 6 Vast RNA	4 wk run, n = 4 Vast RNA	Starvation, n = 4 Vast RNA
	Vast RNA	Vast Protein	TLA RNA	TLA Protein	Vast RNA	Plant RNA	C2C12-PPARβ RNA	C2C12-PPARβ Protein
γ1-AMPK (%)	NC	+29a	+36a	+55	NC	NC	NC	NC	NC	NC	NC	+36
γ3-AMPK (%)	−42a	−48a	−50a	−47	−30a	−50a	−60b	−62b	−50a	−30a	−60a	−80a

Wojtaszewski et al. (37) have reported that in normal and type 2 diabetic men, 6 wk of strength training (three sessions per week, session lasting no more than 30 min) leads to modifications of γ3- and γ1-AMPK expression at RNA and/or protein levels. In comparison with their findings, we have summarized here all the variations of γ1- and γ3-AMPK expression observed in mice skeletal muscle [vastus lateralis (Vast), plantaris (Plant), and TLA)] and in C2C12-PPARβ myotubes after 48 h GW0742 treatment, one night of voluntary running (one night run), 4 wk voluntary running (4 wk run), or 48 h starvation and finally in 42-d-old PPARβ transgenic (Tg) mice. NC, No change.

^a

P < 0.05.

^b

P < 0.005.

For other AMPK subunits, we observed that a 24- and 48-h GW0742 treatment led to a significant increase of β2-AMPK (+17%, P = 0.018, and +21%, P = 0.004, respectively), whereas the α2 subunits is modestly but significantly affected only at the 5-h time point (−17%, P = 0.001) (Fig. 1).

Regarding the PPARβ target gene, PDK4, its up-regulation was of the same order of magnitude as the γ3-AMPK down-regulation. Indeed, its expression was up-regulated about two-fold from 2 h to 24 h of GW0742 treatment and it reached a maximum of 3.2-fold after 48 h (P = 0.005).

PPARβ agonist treatment affects both AMPK accumulation and activation levels in mouse skeletal muscle

It is well established that exercise training induces AMPK activation in mammals. Furthermore, it has been shown recently that AMPK is constitutively activated in transgenic mice overexpressing a constitutively active form of PPARβ (14). For these reasons, we have checked, during the 2 d of GW0742 treatment, whether AMPK was activated. Protein lysates from TLA muscle of GW0742-treated mice were used to study α-AMPK (α1 and α2) accumulation and phosphorylation levels. ACC phosphorylation levels have also been analyzed as an indicator for AMPK activation. Results of independent time course experiments (time = 0–8 h and 0–48 h) are depicted with representative Western blots (Fig. 3, A and B) and histograms of quantification analyses (Fig. 3, C–F). AMPK phosphorylation was first reduced between 2 and 4 h and then afterward increased to peak at 8 h (Fig. 3, A, B, and D). The ACC phosphorylation level also peaks at 8 h of GW0742 treatment (Fig. 3, A, B, and F). α-AMPK and ACC phosphorylation remained slightly higher than in controls until the end of the GW0742 treatment (Fig. 3, D and F).

Fig. 3.

PPARβ agonist treatment affects both AMPK accumulation levels and AMPK activation status in mouse muscles. A and b, Representative Western blot analyses of AMPK, P-AMPK, ACC, and P-ACC using proteins obtained from TLA muscle of 10-wk-old mice treated with GW0742 (GW) (1 mg/kg·d) or vehicle [control (Ct)] for the indicated times. C–F, After densitometric analysis, results from short (0–8 h) and long (0–48 h) time-course experiments were normalized against the loading control (P38), pooled, and presented as histograms: C, total α1- and α2-AMPK (AMPK); D, (Thr¹⁷²)-phosphorylated α1- and α2-AMPK (P-AMPK); E, P-AMPK/AMPK ratio; F, (Ser⁷⁹)-phosphorylated ACC (P-ACC). Values are means ± sd. *, P < 0.05 when compared with respective controls.

We also noted that accumulation levels of α-AMPK were significantly reduced in the early time points of GW0742 treatment and then returned to control values (Fig. 3, A–C). For these reasons, the ratio between phosphorylated AMPK and total α-AMPK (P-AMPK/AMPK) is particularly high at 8 h of GW0742 treatment (Fig. 3E). Reduced α-AMPK accumulation was also observed in early times of GW0742 treatment (2–8 h) when analyzing proteins from other mouse muscles such as gastrocnemius, plantaris, and vastus lateralis (data not shown).

PPARβ overexpression leads to a decrease of γ3-AMPK expression in mice skeletal muscle

The PPARβ transgenic mice, constructed in our laboratory, overexpress a wild-type (nonactivated) form of PPARβ in their skeletal muscles and thus required a physiological activation to induce PPARβ activity. However, adult PPARβ transgenic mice present almost the same muscle remodeling than agonist-treated mice, in particular when considering muscle hyperplasia, increase in oxidative fibers, capillaries, and myonuclei numbers (6, 8, 9). Therefore, it was interesting to analyze how AMPK subunits were expressed in skeletal muscles of PPARβ transgenic mice compared with their control littermates. Because the number of glycolytic fibers of PPARβ transgenic mice differs from the one observed in control mice beginning at the age of 2 months (6, 8, 9), we have analyzed the expression of γ1 and γ3-AMPK in the vastus lateralis muscle of growing control (represented by a dotted line) and transgenic mice before and after the PPARβ-induced muscle oxidative remodeling (Fig. 4). We have also analyzed the expression level of PDK4 to follow PPARβ activity. We observed that γ3-AMPK expression is always lower in skeletal muscle from PPARβ transgenic mice than in muscle from control littermates, but its down-regulation became significant only in 42-d-old mice (−45%, P = 0.038). Down-regulation of γ3-AMPK expression is maintained in adult (20 wk old) transgenic mice (−50%, P = 0.032). Interestingly, this down-regulation of γ3-AMPK expression in skeletal muscles of PPARβ transgenic mice is in the same order of values as in muscles of GW0742-treated animals (Table 1). Expression of γ1-AMPK seems to be up-regulated in a manner inversely correlated to γ3-AMPK expression, but due to higher inter-animal variations, it did not reach the level of statistical significance (Fig. 4). Finally, the up-regulation of PDK4 reached significant variation only in adult PPARβ transgenic animals (Fig. 4). Note that in PPARβ transgenic mice, the PDK4 up-regulation (approximately 2-fold) is of the same order of magnitude as the γ3-AMPK down-regulation.

Fig. 4.

PPARβ overexpression leads to a decrease of γ3-AMPK expression in mice skeletal muscle. Relative quantification of γ1 (Prkag1), and γ3 (Prkag3) AMPK subunits as well as of PDK4 and FAT/CD36 using qPCR analysis and RNA from vastus lateralis muscle of PPARβ transgenic (Tg) mice and control animals (HSA-CRE transgenic littermates) at 21, 28, and 42 d and 20 wk (Tg Ad) after birth (n = 4 per group). Results were normalized against the 36B4 housekeeping gene. Values are means ± sd. *, P < 0.05, when compared with the value of the corresponding AMPK subunit in control groups (HSA-CRE transgenic littermates) at the same age (represented by a dotted line).

Decreased γ3-AMPK expression is a physiological response to both voluntary exercise and starvation

To test the effect of voluntary exercise on skeletal muscle expression of γ3- and γ1-AMPK subunits, mice were placed in cages equipped with a running wheel for 4 wk. Total RNA from vastus lateralis muscles of trained or untrained mice were used to perform quantitative PCR (qPCR) analyses. As shown in Fig. 5A, voluntary running exercise induced in skeletal muscles a 60% repression (P = 0.048) of γ3-AMPK mRNA expression but did not affect γ1-AMPK mRNA expression. Concerning the PPARβ target gene, 4 wk of voluntary running led to an up-regulation of PDK4 (+233%, P =0.0024) (Fig. 5A).

Fig. 5.

γ3-AMPK transcriptional repression is a physiological response to both exercise training and fasting. A–C, Relative quantification of γ1-AMPK (Prkag1), γ3-AMPK (Prkag3), and PDK4 obtained by qPCR analysis using RNA from vastus lateralis of 10-wk-old C57BL6 mice after 4 wk of voluntary wheel running (Training; n = 4) compared with untrained mice [control (Ct); n = 4] (A) or one night of voluntary wheel running (1 night run; n = 6) compared with untrained mice (Ct; n = 6) (B) or 24 h of fasting (Fasting; n = 6) compared with control with normal diet (Ct; n = 4) (C). D, Representative Western blot analyses of P-AMPK, PPARβ using lysates obtained from the contralateral vastus lateralis muscle of the starved mice used in C. PPARβ transgenic (Tg) mice were used as positive control for PPARβ Western blotting. After densitometric analysis, results were normalized against the loading control (P85), pooled, and presented as histograms. A–C, Results were normalized against the 36B4 housekeeping gene. Values are means ± sd. *, P < 0.05; **, P < 0.005 when compared with the value of the corresponding gene in control groups (untrained or fed). D, Value are means± sd. *, P < 0.05, when compared with the value of the corresponding protein in the control group (Ct fed).

To test whether the decrease of γ3-AMPK mRNA is representative of muscle remodeling, or reflects another process, we next analyzed its expression in response to shorter challenges. In the first one, mice were placed in cages with a running wheel for only one night (Fig. 5B), and in the other challenges, mice were starved for either 24 h (Fig. 5, C and D) or 48 h (data not shown). Because PPARβ has been shown to be up-regulated and activated during both physical exercise and fasting (3, 4, 9), we also measured in these experiments the expression levels of PDK4. As seen in Fig. 5B, one night of voluntary running was sufficient to induce a significant 30% (P = 0.028) down-regulation of γ3-AMPK expression but did not modify γ1-AMPK expression. Due to high inter-animal variations, this challenge was also not sufficient to induce significant up-regulation of PDK4 expression (Fig. 5B). Starvation also modified skeletal muscle γ3-AMPK expression, but to a higher extent [−72%, P = 0.00016 after 24 h (Fig. 5C), and 80%, P = 0.038 after 48 h (data not shown) of starvation], whereas expression of γ1-AMPK showed only a tendency to be up-regulated (+36%) after 48 h of starvation (data not shown and Table 1). Concomitantly, in muscles of starved mice, expression of PDK4 [+545%, P = 5 × 10⁻⁵ after 24 h (Fig. 5C), and +1673%, P = 0.03 after 48 h (data not shown) of starvation] was highly and significantly up-regulated. Furthermore, we confirmed that in 24-h starved mice, γ3-AMPK down-regulation is related to an increase of both PPARβ mRNA (1.6-fold; P = 0.014) and PPARβ protein (1.95-fold, P = 0.042) levels (Fig. 5, C and D). Fed PPARβ transgenic mice were used as positive control in Western blotting experiments.

PPARβ activation leads to rapid decrease of γ3-AMPK expression in C2C12-PPARβ myotubes

We next analyzed the expression of γ1 and γ3-AMPK in cultured myotubes after PPARβ pharmacological activation. We also quantified the expression level of PDK4 and of another PPARβ target gene, namely FAT/CD36 (fatty acid transporter/cluster of differentiation 36) to follow PPARβ activation. We used the C2C12-PPARβ myoblastic cell line, which overexpresses PPARβ (4). We performed dose-response experiments using increasing concentrations of GW0742 (Fig. 6, A and B). Cells were harvested 48 h after PPARβ pharmacological activation. Our results show that in C2C12-PPARβ differentiated myotubes, γ3-AMPK expression is down-regulated in a dose-dependent manner by GW0742 treatment. Its expression is already reduced at 0.1 nm by 40%, the lowest PPARβ agonist concentration used. Down-regulation reaches a maximum (−70%) at 10 and 100 nm GW0742 (Fig. 6A). On the contrary, no variation in γ1-AMPK expression was observed, whatever the GW0742 concentration used. As expected, PDK4 and FAT/CD36 expression were up-regulated in a dose-dependent manner by PPARβ agonist treatment (Fig. 6B). We then analyzed the time course of this γ3-AMPK down-regulation upon GW0742 treatment (10 nm). As shown in Fig. 6C, γ3-AMPK expression was rapidly and significantly down-regulated upon PPARβ activation, i.e. −30% at 2 h, −40% at 4 and 8 h, and −60% at 24 and 48 h after GW0742 treatment. On the contrary, no statistical variation of γ1-AMPK expression can be evidenced (Fig. 6C and Table 1). Concerning PPARβ target genes PDK4 and FAT/CD36, up-regulation became significant at 2 and 8 h of GW0742 treatment, respectively (Fig. 6D).

Fig. 6.

PPARβ activation leads to rapid transcriptional repression of γ3-AMPK in C2C12-PPARβ myotubes. Relative quantification of γ1-AMPK (Prkag1), γ3-AMPK (Prkag3), FAT/CD36, and PDK4 obtained by qPCR analysis using RNA from C2C12-PPARβ differentiated myotubes after treatment with the PPARβ agonist, GW0742, in a 48-h dose-response analysis (A and B; n = 4 per group) or in a time-course experiment using 10 nm GW0742 (C and D; n = 4 per group). Results were normalized against the 36B4 housekeeping gene. Values are means ± sd. *, P < 0.05; **, P < 0.005 when compared with the value of the corresponding gene in control (Ct) vehicle-treated group.

In C2C12-PPARβ cells treated with GW0742 (10 nm), we have also observed a time-dependent reduction of γ3-AMPK protein level (Fig. 7, A and B). Furthermore, the decrease of γ3-AMPK protein is rapid and followed almost exactly the time-dependent reduction of γ3-AMPK mRNA (compare Figs. 6C and 7B), suggesting a rapid turnover of the protein. Treatment of C2C12-PPARβ cells with GW1516, as specific PPARβ agonist, also induced a time-dependent down-regulation of γ3-AMPK (data not shown). On the contrary, the level of γ1-AMPK protein remains unchanged (Fig. 7, A and B).

Fig. 7.

PPARβ activation leads to rapid reduction of γ3-AMPK protein levels in C2C12-PPARβ myotubes. a, Representative Western blot analyses of γ1-AMPK, γ3-AMPK, and P85 (loading control) using protein lysates from C2C12-PPARβ cells treated with GW0742 (10 nm) for increasing periods of time. b, After densitometric analysis, results were normalized against the loading control (P85), pooled, and represented as histograms. Values are means ± sd. *, P < 0.05; **, P < 0.005 when compared with the value of the corresponding AMPK subunit in control (Ct) groups (cells treated with vehicle).

PPARβ activation leads to down-regulation of γ3-AMPK promoter activity in C2C12-PPARβ myotubes

To investigate how PPARβ activation leads to a decrease of γ3-AMPK mRNA, we constructed the luciferase reporter plasmid, pGL3-B-Prkag3, in which 2.6 kb of the 5′-flanking region of the Prkag3 gene (encoding γ3-AMPK) controls luciferase expression. This DNA region preceding the Prkag3 gene contained both core promoter elements and PPAR-responsive elements. C2C12-PPARβ cells were then transfected with either pGL3-B-Prkag3 or with pGL3-Basic control vector. Due to the myotube differentiation procedure, luciferase activity was measured 7 d after transfection. At d 5 after transfection, cells were treated or not with increasing concentrations of GW0742 (1–100 nm) for 48 h (Fig. 8). We observed a 10- to 100-fold higher luciferase activity in pGL3-B-Prkag3-transfected cells than in cells transfected with pGL3-Basic control vector. This indicates that the 2.6 kb of the 5′-flanking region of the Prkag3 gene contains promoter elements allowing luciferase expression in differentiated myotubes. Remarkably, treatments with increasing concentrations of GW0742 decreased promoter activity in a dose-dependent manner (Fig. 8), suggesting that the observed down-regulation of γ3-AMPK mRNA upon GW0742 treatment is due to a decrease of Prkag3 promoter activity.

Fig. 8.

PPARβ activation leads to a repression of Prkag3 promoter activity. Relative quantification of luciferase and β-galactosidase activity in C2C12-PPARβ cells cotransfected with cytomegalovirus-driven β-galactosidase vector and either pGL3-Basic (pGL3-B) or pGL3-B-Prkag3 vector. Transfected cells were treated (black bars) or not (vehicle only, white bars) with increasing concentrations of GW0742 (1–100 nm). Values are means of luciferase to β-galatosidase ratio ± sd. *, P < 0.05; **, P < 0.005 when compared with the value of pGL3-Basic transfected cells treated with vehicle (DMSO).

Discussion

PPARβ and AMPK are two key actors of muscle adaptation to metabolic challenges, such as endurance physical exercise or fasting (for review see Refs. 17 and 44). These pathways collaborate to promote muscle oxidative phenotype by up-regulating the expression of several genes, including those implicated in fatty acid catabolism. Recently, it has been proposed that both pathways are interconnected, because PPARβ activation promotes AMPK phosphorylation and because AMPK physically interacts with PPARβ and directly regulates its transcriptional activity (14). We report here a new connection between the two regulatory pathways, in which PPARβ activation represses expression of the muscle-specific γ3-AMPK subunit in both cultured myotubes and mouse muscle in vivo, leading to a remodeling of AMPK.

It has been reported that several weeks of either endurance or strength training promote a down-regulation of the γ3-AMPK subunit in human skeletal muscle (36, 37). However, due to both the duration of the exercise training period (i.e. 3–6 wk) (36, 37) and the higher expression of γ3-AMPK in glycolytic fibers (38), decrease of γ3-AMPK expression could result of a transcriptional down-regulation or may reflect an oxidative muscle remodeling. Herein, we report that 4 wk of voluntary exercise also promoted a 60% reduction of γ3-AMPK subunit expression in skeletal muscles of mice kept in cages with a running wheel (mean running distance of 7 km/d) (Fig. 5A and Table 1). Interestingly, this reduction took place very rapidly because γ3-AMPK mRNA was reduced by 30% after only one night of voluntary exercise (Fig. 5B and Table 1). We also observed that starvation seriously down-regulated γ3-AMPK mRNA level in TLA muscle [−72% after 24 h (Fig. 5C) and −80% after 48 h (data not shown and Table 1) of starvation]. Thus, γ3-AMPK down-regulation appears to be a rapid and, however, persistent physiological muscle response to situations requiring increased fatty acid oxidation. Our results are in agreement with clinical data on young and old twins showing that individuals with lower γ3-AMPK levels have higher aerobic capacity (43).

It is tempting to propose a role for PPARβ in such regulation of γ3-AMPK subunit expression in skeletal muscle. Indeed, in these physiological situations, the level of γ3-AMPK down-regulation was well correlated to the activation level of PPARβ, which has been followed by the up-regulation of PDK4, a well-known PPARβ target gene (Fig. 5). Furthermore, we also show that in starved mice, PPARβ expression is increased at both the mRNA and protein level (Fig. 5, C and D). Moreover, we report here that γ3-AMPK mRNA and protein expression is markedly and rapidly down-regulated, when PPARβ is pharmacologically activated with GW0742 or GW1516 in either cellular model or in mouse skeletal muscles (Figs. 1, 2, 6, and 7 and Table 1). In C2C12-PPARβ differentiated myotubes, treatment with GW0742 promotes a dose-dependent γ3-AMPK mRNA reduction that is detectable at the smallest concentration used (40% at 0.1 nm) and reaches a maximum of 70% at 10 nm (Fig. 6A). In mice skeletal muscle, γ3-AMPK mRNA expression is reduced by 50% by either pharmacological activation (Table 1 and Fig. 1) or muscle-specific transgenic overexpression of PPARβ (Table 1 and Fig. 4). Interestingly, this PPARβ-dependent γ3-AMPK down-regulation is in the same order of value as in muscle of both trained mice and men (Table 1) (36, 37).

Moreover, in both myotubes and skeletal muscle, γ3-AMPK down-regulation can be considered as one of the fastest responses to PPARβ activation already detected after 2 h and being maximal within 24 h both in vitro and in vivo (Figs. 1 and 6C). In that respect, this response appeared to be similar to that observed for PDK4 up-regulation (Figs. 1 and 6D) and even faster than FAT/CD36 up-regulation, which was not observed before 8 h in vitro (Fig. 6D). These data indicate that γ3-AMPK down-regulation precedes the PPARβ-promoted muscle oxidative remodeling that was observed after 48 h of treatment (6, 8), excluding the hypothesis that such a down-regulation is related to the decreased number of glycolytic myofibers in GW0742-treated animals. Furthermore, in PPARβ transgenic mice, which overexpress PPARβ in their skeletal muscles, γ3-AMPK expression is also reduced before the reported oxidative muscle remodeling (Fig. 4) (6, 7) and even before significant up-regulation of PDK4 (Fig. 4).

Finally, we found that in both agonist-treated mice and PPARβ transgenic mice, the changes in γ3-AMPK expression are of the same order of magnitude as those seen with PDK4 (2-fold). Note that starvation affected both γ3-AMPK and PDK4 to a higher extent than PPARβ pharmacological activation or muscle-specific transgenic overexpression. Up-regulation of both PPARβ mRNA and protein levels in the skeletal muscle of starved mice (Fig. 5, C and D) represents a possible explanation. However, we cannot exclude that other factors, such as transcription factors or kinases, could act synergistically with PPARβ to modify PDK4 and γ3-AMPK expression.

Our results show that PPARβ overexpression is not sufficient to repress γ3-AMPK expression and that its activation is required. Indeed, PPARβ overexpression start at d 9 postpartum in skeletal muscle of PPARβ transgenic mice (data not shown) and γ3-AMPK down-regulation reaches significant variation only at d 42 when PDK4 showed a tendency to be up-regulated (Fig. 4). Furthermore, γ3-AMPK can easily be detected in C2C12-PPARβ myotubes, and GW0742 treatments highly decrease its expression (Figs. 6 and 7).

A switch between γ3- and γ1-AMPK subunit expression has been reported in vastus lateralis muscle from trained vs. untrained leg in men, with γ3-AMPK being down-regulated at both mRNA and protein levels, whereas γ1-AMPK was increased only at the protein level (Table 1) (36, 37). We showed that up-regulation of the γ1-AMPK subunit upon PPARβ activation is not as straightforward as the down-regulation of γ3-AMPK at least at the mRNA level. Indeed, up-regulation of γ1-AMPK mRNA was observed only in TLA muscle from GW0742-treated mice (Figs. 1 and 2) but not in other skeletal muscle (Table 1) nor in C2C12-PPARβ-treated cells even when high GW0742 concentrations were used (Fig. 6, A and C). Furthermore, in PPARβ transgenic animals, γ1-AMPK mRNA seems up-regulated in an age-dependent manner, but it does not reach significant variation due to high inter-animal variations (Fig. 4). Finally, in mice, exercise training or starvation (Fig. 5, C and D) does not seem to affect γ1-AMPK expression despite PDK4 up-regulation. Thus, our data do not support a role of PPARβ in the regulation of γ1-AMPK expression.

The effects of PPARβ activation on the expression level of the other AMPK subunits does not reveal substantial changes, suggesting a particular response of the γ3-AMPK subunit to activation of the nuclear receptor. How PPARβ activation leads to this rapid down-regulation of γ3-AMPK expression remains an open question. Data reported in Fig. 8 strongly suggest that PPARβ activation represses γ3-AMPK gene transcription. Such a negative action of PPARβ has previously been demonstrated on the WT1 promoter activity (45). However, despite a deletion analysis of the γ3-AMPK promoter, we did not succeed in identifying the PPAR-responsive element that could have been involved. Indeed, the region under study contains elements that are crucial for muscle-specific expression hampering the investigation of the potential repressive effects of PPARβ. Thus, detailed analysis of the corresponding promoter is still needed to explain the exact molecular mechanism involved in this transcriptional repression.

Analysis of the impacts of PPARβ activation on AMPK activity in muscles of mice treated by GW0742 revealed that α-AMPK total protein amount was slightly, but significantly, reduced at 6 and 8 h and returned to control values after 24 h (Fig. 3, A–C). Despite such a reduced amount, the level of phosphorylated AMPK (P-AMPK), the active form of AMPK, is significantly increased at 8 h of treatment (Fig. 3, A, B, and D). Increased AMPK activity is confirmed by the increased levels of P-ACC, a direct target of activated AMPK, at 8 h and to a lesser extent at 24 h of treatment (Fig. 3, A, B, and E). Our results complement previous reports showing AMPK activation both in culture myotubes treated with a PPARβ agonist (46, 47) and in mice skeletal muscle overexpressing a constitutive active form of PPARβ (VP16-PPARβ) (14). In this regard, we report that compared with control fed mice, the basal level of AMPK phosphorylation is also increased in fed transgenic mice overexpressing a wild-type (nonactivated) form of PPARβ in their skeletal muscles (Fig. 5D). Whether this up-regulation of AMPK activity resulted from a direct interaction with PPARβ (14) or was due to a switch in abundance from the α2β2γ3 to the γ1-AMPK containing complexes (α1β2γ1 and α2β2γ1) known to render the AMPK system more sensitive to changes in cellular AMP concentration (48) remains to be determined.

From our data, it can be proposed that activation of the PPARβ pathway results in a very rapid change in AMPK subunit composition, due to the reduction of γ3-AMPK subunit expression, which mimics in part what was observed after exercise training in both men and mice (Table 1). Because no clear physiological functions have been attributed to each AMPK trimer, despite the availability of knockout and transgenic mice models, functional implications of reduced γ3-AMPK expression remains unclear and await further investigations. However, it might be crucial for the adaptive response to exercise training with regard to regulation of both metabolism and gene expression (39, 49, 50). This could be particularly relevant considering the fact that γ3-AMPK protein expression and activity are negatively related to whole-body glucose uptake through the insulin-stimulated nonoxidative glucose metabolism (43). In this study, we show that in addition to PDK4 up-regulation that impairs glucose oxidation by inhibiting pyruvate dehydrogenase activity, FAT/CD36 up-regulation, which augments fatty acid supply and increased expression of other genes such as CPT-1α (data not shown) that favor fatty acid burning, PPARβ activation led to the down-regulation of γ3-AMPK, which is critical for both glycogen ergogenics and coordinated transcription of genes required for lipid and glucose metabolism (21, 51). All together, it is tempting to speculate that rapid changes in both AMPK heterotrimer composition and AMPK activation level in response to PPARβ activation represent leading events for a fuel preference transition from glucose to fatty acids in skeletal muscle.

Materials and Methods

Animals

Animals were maintained in a 12-h light, 12-h dark cycle and received food [UAR (Usine d'alimentation rationnelle), Villemoisson sur Orge, France] and water ad libitum. All experimental procedures were conducted according to French legislation. Ten-week-old males C57BL6J (Janvier, Le Gesnest Saint Isle, France) were used in GW0742 treatments, voluntary running, and starvation experiments. We always used four to six animals per condition to obtain statistical values. Animals were killed after the indicated times by cervical dislocation and muscles from posterior legs (TLA, plantaris, and vastus lateralis) were harvested immediately after killing. Tissues were snap-frozen in liquid nitrogen and stored at −80 C until use.

For voluntary running experiments, animals were placed in a cage equipped with a running wheel (one animal per cage) for either one night or 4 wk. For starvation experiments, animals were fed ad libitum and then were starved for either 24 or 48 h.

Animals overexpressing PPARβ specifically in skeletal muscle were generated in the laboratory as previously described (9). Briefly, B6D2 mice harboring a loxP-stop-loxP-PPARβ-hygromycin cassette were crossed with B6D2 mice expressing Cre recombinase under the human skeletal actin (HSA) promoter (52). All animals were maintained hemizygous for their transgene. Presence of the transgenes was verified by PCR analyses of tail DNA (REDExtract-N-Amp Tissue PCR Kit; Sigma Chemical Co., St. Louis, MO). Animals harboring the two transgenes were used as PPARβ-overexpressing mice, whereas animals harboring the HSA-Cre transgene only were used as controls.

Cell culture

The C2C12-PPARβ cell line was generated in the laboratory as described previously (4). This cell line was chosen in this study because it gives a better and more reproducible response to PPARβ agonist treatment for PDK4 and FAT/CD36 up-regulation than the original C2C12 cell line (4) (data not shown). For myotube differentiation, we proceed as follows: C2C12-PPARβ cells were grown in DMEM culture medium complemented with 1 g/liter d-glucose, 4 mml-glutamine, 25 mm HEPES, 1 mm sodium pyruvate (Invitrogen, Cergy Pontoise, France; no. 22320) and with 8% fetal calf serum (FCS), up to 90% of confluence. Then medium was replaced by a differentiation medium [DMEM complemented with 1 g/liter d-glucose, 4 mml-glutamine, 25 mm HEPES, 1 mm sodium pyruvate (GIBCO; no. 22320), and 2% FCS] for at least 5 d. Differentiated myotubes were finally treated for the indicated period of time with GW0742. Cells were all harvested at the same time corresponding to the 48-h period of treatment.

Cells were incubated at 37 C in a humid atmosphere, 5% CO₂, and culture media were systematically completed with antibiotics (200 IU/ml penicillin G and 50 μg/ml streptomycin) to prevent contamination. Media were renewed every 48 h to avoid exhaustion of constituents.

PPARβ-agonist treatments

Animals

GW0742, a PPARβ-specific agonist (53), was dissolved in vehicle [DMEM/6% dimethyl sulfoxide (DMSO)] and injected to animals sc once a day (0900 h) at 1 mg/kg. Control animals received vehicle (DMEM/6% DMSO) at the same time.

C2C12-PPARβ

Differentiated myotubes were treated in differentiation media with various concentrations of GW0742 (0.1 nm to 100 nm) for various periods of time. In C2C12-PPARβ cells, 10 nm of GW0742 is sufficient to evaluate the effect of PPARβ activation.

Plasmids

Mouse genomic DNA was used to amplify 2.6 kb of the 5′-flanking region of the Prkag3 gene (NT_039170.7 Mm1_39210_37; Mus musculus chromosome 1 genomic contig, strain C57BL/6J). Briefly, a first PCR amplification (GoTaq; Promega, Madison, WI; M3171) was performed using 5′-CACAGGCTGGGCAAGCAC-3′ (sense) and 5′-TCCAGCTCGGGCTCCATG-3′ (reverse) oligonucleotides. The diluted PCR product (1/100) was then used to realize a second PCR amplification using 5′-CTCGGTACCTGTGAGGCATGGAATTCAG-3′ (sense plus KpnI restriction site) and 5′-GAGAAGCTTTCCATGCGGCCAGCTCTAG-3′ (reverse plus HindIII restriction site) oligonucleotides. pGL3-B-Prkag3 was obtained by cloning the 2.6-kb PCR fragment into the KpnI and HindIII restriction site of pGL3-Basic (pGL3-B) vector (Promega; E1751).

Transient transfection experiments

C2C12-PPARβ cells were seeded in 12-well dishes at 50% confluence. The next day, cells were transfected in quadruplicate with 0.1 μg cytomegalovirus-driven β-galactosidase vector and either 0.4 μg per well pGL3-B-Prkag3 or pGL3-Basic vector using Fugene6 according to manufacturer's instructions (Roche, Indianapolis, IN; 11 814 443 001). Cells were left in growing media (8% FCS) until 90% confluence and then in differentiation medium (2% FCS). After 5 d in differentiation medium, cells were treated with increasing concentrations of GW0742 or vehicle in differentiation media for 48 h. Cells were harvested at d 7 after transfection for measurement of luciferase and β-galactosidase activity using a luciferase assay system (Promega; E1500) and β-galactosidase enzyme assay system (Promega; E2000), respectively. Results are expressed as the ratio of luciferase to β-galactosidase activity.

Western blot analysis

Total protein lysates from cells and mice muscles treated with GW0742 or vehicle were prepared in lysis buffer [20 mm Tris (pH 7.4), 137 mm NaCl, 0.5% Nonidet P-40, 0.5% Triton X-100, 10% glycerol, 1 mm Na pyrophosphate, 1 mm orthovanadate, 50 mm β-glycerophosphate, 10 mm EDTA, 1 mm EGTA, 1× Complete protease inhibitor cocktail (Roche; 11 873 580 001)], electrophoresed, and blotted on polyvinylidene difluoride membrane. The following rabbit primary antibodies were used at the same dilution (1:1000) for immunodetection: AMPKα, which detects both the α1- and α20isoforms of the catalytic subunit (Cell Signaling Technology, Danvers, MA; no. 2532); phospho-AMPKα, which detects both α1- and α2-phosphorylated isoforms of the catalytic subunit (Cell Signaling; Thr¹⁷² 40H9 no. 2535); ACC (Cell Signaling; no. 3662); phospho-ACC (UpState Biotechnology, Lake Placid, NY; Ser⁷⁹ no. 07-303); and γ1-AMPK (Epitomics, Burlingame, CA, no. 1592-1). To detect γ3-AMPK, we use the rabbit polyclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA; Sc-20166) at 1:200 dilution. Loading controls P38 and P85 were, respectively, detected with rabbit polyclonal antibody (Cell Signaling; no. 9212) at 1:1000 dilution or mouse monoclonal antibody (UpState; no. 05-217) at 1:500 dilution. PPARβ was detected using a polyclonal antiserum raised against the A/B domain of mouse PPARβ (9). Peroxidase-coupled antirabbit (Vector Laboratories, Burlingame, CA; PI-1000) and antimouse (Vector; PI-2000) secondary antibodies were used at 1:8000 and 1:5000 dilution, respectively. Images were captured using a Kodak (Rochester, NY) Image Station 2000 and quantified using Kodak molecular imaging software.

Quantitative real-time PCR

Reverse transcription were performed using 0.2 μg total RNA from skeletal muscle (TLA, plantaris, or vastus lateralis) or from C2C12-PPARβ and a core kit (RT-RTCK-03; Eurogentec, Seraing, Belgium) according to manufacturer's instructions, and a mix of random primers (9-mers) and oligo-deoxythymidine. qPCR were performed on SDS7900HT (Applied Biosystems, Foster City, CA) using Mesagreen qPCR kit for SYBR (Eurogentec) and the primers listed in Table 2.

Table 2.

List of primer sequences used for quantitative real-time PCR analyses

Gene	Accession number	Sense primer (5′–3′)	Antisense primer (5′–3′)
Prkaa1	NM_001013367.3	TGCTACTCCACAGAGATCGG	GTCTGAGGGCTTTCCTTGAG
Prkaa2	NM_178143.2	CGACTACATCTGCAAACATGG	CAGTAATCCACGGCAGACAG
Prkab2	NM_182997.2	AGATCATGGTGGGGAGCA	GCTGCCAGGGTACAAACTCT
Prkag1	NM_016781.2	TTATCACCGAGTTCCCCAAG	TGCTGTACAAAGATGCCCAG
Prkag3	NM_153744.3	GGTCATCTTTGACACCACGTT	AGAGGAGCTGCCCTCACAC
PDK4	NM_013743.2	GGGTCTCAATAGTGTCACC	GTGGGCCTGGGCATTTAGCA
Fat/CD36	NM_007643.3	TTTCCTCTGACATTTGCAGGTCTA	AAAGGCATTGGCTGGAAGAA
36B4	NM_007475.5	TCCAGGCTTTGGGCATCA	CTTTATCAGCTGCACATCACTCAGA

All reactions of qPCR were performed in triplicate. The relative amount of all mRNAs was calculated using the comparative CT method and 36B4 was used as the invariant gene control for all studies.

Statistical analysis

Data are expressed as means ± sd. Statistical analyses were performed by Student's t tests. A P value <0.05 was considered significant.

†

Annotations provided by Nuclear Receptor Signaling Atlas (NURSA) Bioinformatics Resource. Molecule Pages can be accessed on the NURSA website at www.nursa.org.

Acknowledgments

GW0742 was a generous gift from T. M. Willson (GlaxoSmithKline, Research Triangle Park, NC). We thank M. Aupetit, M. Radjkhumar, F. Millot, J. Paput, S. Smara, G. Manfroni, and G. Visciano for technical assistance. We thank Dr. Kay-Dietrich Wagner, Dr. Nicole Wagner, and Pr. Emmanuel Van Obberghen for critical reading of the manuscript. E.L. declares that she has not participated in animal experiments or in extraction of RNA from animal tissues.

This work was supported by a grant to P.A.G. from Fondation Cœur et Artères. E.L. is the recipient of a fellowship program “For women in Science” of L'Oreal Foundation in partnership with United Nations Educational, Scientific, and Cultural Organization.

Disclosure Summary: The authors have no conflicting financial interest.

Abbreviations

 
• ACC

  Acetyl-coenzyme A carboxylase

 
• AMPK

  AMP-activated protein kinase

 
• DMSO

  dimethyl sulfoxide

 
• FCS

  fetal calf serum

 
• P-AMPK

  phosphorylated AMPK

 
• PDK4

  pyruvate dehydrogenase kinase isozyme 4

 
• PPARβ

  peroxisome proliferator-activated receptor β

 
• qPCR

  quantitative PCR

 
• TLA

  tibialis anterior.