Adiponectin Increases Fatty Acid Oxidation in Skeletal Muscle Cells by Sequential Activation of AMP-Activated Protein Kinase, p38 Mitogen-Activated Protein Kinase, and Peroxisome Proliferator–Activated Receptor α

Induction of myogenic differentiation was carried out as previously described (13). In brief, mouse C2C12 myoblasts were maintained in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal bovine serum. To induce differentiation, confluent cells were incubated in DMEM supplemented with 2% horse serum. The cells differentiated into multinucleated myotubes in 6–7 days.

For viral infections, differentiated C2C12 myotubes were infected with either null or dominant-negative AMPK (T172A) (31) expressing adenovirus (300 PFU/cell). After 4 h, the medium was replaced with fresh medium containing 2% horse serum.

Recombinant adiponectin protein was purified using a modified version of a previously described protocol (32). In brief, HEK293 cells were transfected with plasmid vectors expressing myc-His–tagged adiponectin (pcDNA3.1-adiponectin-myc-His). After transfection, the cells were incubated in serum-free DMEM with 0.1 g/l l-ascorbic acid for 48 h. The medium was then collected and centrifuged at 3,000 rpm for 10 min to remove cellular debris. After adding saturated ammonium sulfate to 40% (wt/vol) and incubating at 4°C for 4 h with shaking, the protein fraction was precipitated by centrifugation at 3,000 rpm for 3 h at 4°C. The precipitate was resuspended in 10 mmol/l HEPES (pH 8.0) buffer containing 50 mmol/l NaCl and 1 mmol/l CaCl₂ and bound to Ni-NTA agarose beads. The beads were washed twice with washing buffer (50 mmol/l NaH₂PO₄, 300 mmol/l NaCl, and 20 mmol/l imidazole), and adiponectin was collected with elution buffer (50 mmol/l NaH₂PO₄, 300 mmol/l NaCl, and 250 mmol/l imidazole). The pH of the washing and elution buffers was adjusted to 8.0 with NaOH before use. Only batches containing adiponectin >95% purity, which was determined by Coomassie staining, were used for subsequent experiments.

HEK293 cells were grown in DMEM containing 10% bovine calf serum. Luciferase reporter constructs were cotransfected with PPARα, retinoid X receptor α, and/or adiponectin expression vectors by using the calcium phosphate method. After transfection, the cells were serum-starved for 18 h and incubated with 10 μmol/l Wy14643 (Biomol), 1 mmol/l araA (Sigma), 5 μmol/l SB203580 (A.G. Scientific), or vehicle (DMSO) for an additional 24 h before lysis. The luciferase activity was normalized to β-gal activity.

Cells were lysed in 1× radioimmunoprecipitation assay buffer (1% Triton X-100, 0.1% sodium deoxycholate, and 140 mmol/l NaCl). The lysates were separated by SDS-PAGE and transferred to polyvinylidene difluoride membranes (Millipore). Following transfer, the membranes were blocked with milk and probed with primary antibodies against phospho-AMPK (Cell Signaling), AMPK (Upstate), phospho-p38 MAPK (BD Transduction Laboratories), p38 MAPK (Santa Cruz), phospho-ACC (Upstate), and β-actin (Sigma). The results were visualized using horseradish peroxidase–conjugated secondary antibodies (Bio-Rad) and enhanced chemiluminescence.

AMPK activity assays were carried out as previously described (33). In brief, total cell extracts were prepared from differentiated C2C12 myotubes in 1× radioimmunoprecipitation assay buffer and precipitated with saturated ammonium sulfate solution (final concentration 35%). Assays were performed at 30°C for 10 min in 25 μl reaction mixtures containing 5 μg protein extracts in a reaction buffer (40 mmol/l HEPES [pH 7.0], 80 mmol/l NaCl, 5 mmol/l magnesium acetate, 1 mmol/l dithiothreitol, 200 μmol/l each of AMP and ATP, and 2 μCi [γ-³²P] ATP) with or without 200 μmol/l SAMS peptide (Upstate). For immunoprecipitated kinase assays, cell lysates were immunoprecipitated with anti–AMPK-α1 or -α2 antibodies (Upstate), washed with 40 mmol/l HEPES (pH 7.0), and suspended in 20 μl reaction buffer. The reaction mixtures were spotted onto P81 cation exchange papers, washed three times with 1% phosphoric acid, and measured using a scintillation counter. AMPK activity was expressed as [³²P] incorporated per microgram of protein.

Total cell extract (500 μg) was immunoprecipitated with anti-p38 MAPK antibodies and washed with 20 mmol/l HEPES (pH 7.4). Kinase assays were performed for 30 min at 30°C using 2 μg glutathione S-transferase–activating transcription factor 2 (GST-ATF2) as a substrate in the reaction buffer (20 mmol/l HEPES [pH 7.4], 10 mmol/l MgCl₂, 12 mmol/l β-glycerophosphate, 1 mmol/l Na₃VO₄, 2 mmol/l dithiothreitol, 0.5 mmol/l phenylmethylsulfonyl fluoride, and 1 μCi [γ-³²P] ATP). The reactions were terminated using 5× SDS sample buffer, and the products were resolved by SDS-PAGE followed by transfer to polyvinylidene difluoride membranes. The level of incorporated [³²P] was detected by autoradiography, and the membranes were immunoblotted using anti-p38 MAPK antibodies as loading control.

Total RNA was isolated with Trizol (Invitrogen) according to the manufacturer’s protocol. Total RNA, 5 μg, was reverse transcribed with oligo dT using RevertAid M-MuLV Reverse Transcriptase (MBI Fermentas, ON, Canada). cDNA levels were quantified by real-time PCR with SYBR Green using the MyiQ single-color real-time PCR detection system (Bio-Rad). The primer sequences used were as follows: acyl-CoA oxidase (ACO) forward, 5′-TGT TAA GAA GAG TGC CAC CAT-3′ and ACO reverse, 5′-ATC CAT CTC TTC ATA ACC AAA TTT-3′; CPT1 forward, 5′-ACT CCT GGA AGA AGA AGT TCA-3′ and CPT1 reverse, 5′-AGT ATC TTT GAC AGC TGG GAC-3′; FABP3 forward, 5′-CCC CTC AGC TCA GCA CCA T-3′ and FABP3 reverse, 5′-CAG AAA AAT CCC AAC CCA AGA AT-3′; and GAPDH forward, 5′-CAG AAC TAC ATC CCT GCA TCC-3′ and GAPDH reverse, 5′-CCA CCT TCC TGA TGT CAT CA-3′.

Fatty acid oxidation assays were carried out as previously described (34). In brief, differentiated C2C12 myotubes were incubated in α–minimum essential medium containing 0.1 mmol/l palmitate (9,10-[³H]palmitate, 5 μCi/ml) and 2% BSA. After incubation, the medium was recovered and precipitated twice with an equal volume of 10% trichloroacetic acid. The supernatants were transferred to open 1.5-ml microcentrifuge tubes placed in a scintillation vial containing 0.5 ml unlabeled water and incubated at 50°C for 18 h. Following this vapor-phase equilibration step, the tubes were removed and scintillation enhancer fluid (PerkinElmer Life and Analytical Sciences) added to the vial. ³H₂O content was measured in a scintillation counter.

One of the most profound characteristics of adiponectin is its ability to stimulate fatty acid oxidation in peripheral tissues (6). To understand how adiponectin achieves this, we examined the changes in the expression level of genes involved in fatty acid oxidation. When differentiated C2C12 cells were treated with purified adiponectin, the mRNA levels of ACO and CPT1 were increased in a dose-dependent manner (Fig. 1A). Because these genes are known to be transcriptionally regulated by PPARα, we then performed reporter assays to determine whether PPARα transcription activity can be regulated by adiponectin. As shown in Fig. 1B, cotransfection with adiponectin increased the transcriptional activity of PPARα on a reporter construct driven by multiple PPAR response elements (DR-1). This stimulatory effect was observed both in the presence and absence of Wy14643, a synthetic ligand of PPARα, indicating that regulation of PPARα by adiponectin appeared to be ligand independent. This result is consistent with a previous report in which adiponectin was shown to enhance the activation of UAS-luciferase reporter by GAL4-PPARα fusion protein (13,14). These results imply that the stimulation of fatty acid oxidation by adiponectin in muscle cells could occur through the activation of PPARα.

Since AMPK is a well-established mediator of adiponectin action, we tested whether direct activation of AMPK could similarly stimulate PPARα transcription activity. 5-Aminoimidazole-4-caroxamide-1-β-d-ribofuranoside (AICAR), which activates AMPK after being metabolized to 5-aminoimidazole-4-caroxamide-1-β-d-ribofuranoside-5-monophosphate in the cells, also increased the transcriptional activity of PPARα at the promoter of muscle-type carnitine palmitoyl transferase (Fig. 2A). Cotreatment with araA, an inhibitor of AMPK, potently inhibited this effect, confirming that the stimulation of PPARα activity by AICAR was the result of AMPK activation (Fig. 2A). Furthermore, araA also reduced the stimulation of PPARα activity by adiponectin, demonstrating that AMPK is indeed involved in the activation of PPARα by adiponectin (Fig. 2B). Since it was recently reported that adiponectin activates p38 MAPK in addition to AMPK in C2C12 cells (14), we tested whether p38 MAPK might also be involved in this pathway. As shown in Fig. 2B, inhibition of p38 MAPK using SB203580 also significantly inhibited the effect of adiponectin on PPARα activity.

To confirm the roles of AMPK and p38 MAPK in the activation of PPARα, we examined the effects of araA and SB203580 on the mRNA levels of PPARα target genes in C2C12 myotubes. Consistent with the results shown in Fig. 1A, adiponectin increased the expression levels of ACO, CPT1, and FABP3 (Fig. 3A). However, pretreatment of cells with araA or SB203580 abolished the adiponectin-dependent increase in PPARα target gene expression. Furthermore, dominant-negative AMPK similarly repressed the stimulation of PPARα target gene expression by adiponectin (Fig. 3B). Together, these data indicate that the activation of PPARα by adiponectin occurs through a pathway that requires activation of both AMPK and p38 MAPK.

The results from the reporter assays and gene expression analyses suggested that AMPK might be able to stimulate fatty acid oxidation by activating PPARα in muscle cells. To test this possibility, we performed in vitro fatty acid oxidation assays in C2C12 cells using radioisotope-labeled palmitate. As expected, fatty acid oxidation was enhanced by the activation of PPARα or AMPK using Wy14643 or AICAR, respectively (Fig. 4A, lanes 2 and 3). Notably, we observed that cotreatment of Wy14643 and AICAR further elevated fatty acid oxidation (Fig. 4A, lane 4).

Thus far, stimulation of fatty acid oxidation by AMPK was believed to be primarily mediated by phosphorylation of ACC (35). The ACC-dependent regulation of fatty acid oxidation does not require synthesis of new proteins, while PPARα exerts its effect by increasing the expression of its target genes. We took advantage of this difference to determine whether AMPK can indeed stimulate fatty acid oxidation by activating PPARα. When cells were treated with cycloheximide to block protein synthesis, AICAR-induced fatty acid oxidation was reduced by ∼30% (Fig. 4B). This reduction can probably be attributed to the inhibition of effect of PPARα activity, which would act downstream of AMPK to induce the expression of genes involved in fatty acid oxidation.

The roles of AMPK and p38 MAPK in the stimulation of fatty acid oxidation by adiponectin were also examined using fatty acid oxidation assays. As previously reported (10), adiponectin increased fatty acid oxidation by ∼40% (Fig. 4C). This stimulation was suppressed by araA or SB203580 (Fig. 4C, lanes 3 and 4), and cotreatment with both araA and SB203580 did not result in additive or synergistic effects (Fig. 4C, lane 5). Taken together, these data reveal the presence of a pathway through which adiponectin stimulates fatty acid oxidation by activating PPARα.

To get a clearer picture of the adiponectin signaling pathway, we examined the changes in the level of AMPK and p38 MAPK phosphorylation induced by adiponectin. Consistent with previous reports (14,34), adiponectin enhanced the phosphorylation of AMPK, p38 MAPK, and ACC in C2C12 myotubes (Fig. 5A). However, when AMPK was chemically inhibited with araA, the stimulation of AMPK, p38 MAPK, and ACC phosphorylation was attenuated. On the other hand, inhibition of p38 MAPK with SB203580 did not affect adiponectin-induced phosphorylation of AMPK and ACC. Similar results were obtained when AMPK was directly activated using AICAR (Fig. 5B and D). AICAR increased the phosphorylation and kinase activities of AMPK and p38 MAPK, and these effects were inhibited by araA. Likewise, when AMPK was blocked using dominant-negative AMPK, both adiponectin and AICAR did not induce p38 MAPK phosphorylation (Fig. 5C). Conversely, insulin, which activates p38 MAPK in muscle cells, did not affect AMPK phosphorylation or kinase activity (online appendix Fig. 1A and C [available at http://diabetes.diabetesjournals.org]). Therefore, these results indicate that p38 MAPK might function as a downstream signaling molecule of AMPK in the adiponectin signaling pathway.

In an attempt to determine which catalytic subunit of AMPK, α1 or α2, mediates adiponectin signaling in muscle cells, we performed in vitro kinase assays using antibodies specific for each isoform. The kinase activity of both α1 and α2 isoforms was increased by adiponectin treatment, the α1 isoform by ∼50% and the α2 isoform by over twofold (Fig. 5D). Thus, it is possible to speculate that AMPKα2 might be the major mediator of adiponectin function in muscle cells, although the role of α1 cannot be ruled out.

Previously, it has been demonstrated that the transcriptional activity of PPARα can be stimulated by p38 MAPK through the phosphorylation of three serine residues in the AF-1 domain (serines 6, 12, and 21) (36). In addition, we found three putative AMPK phosphorylation sites (serines 167, 373, and 453) in PPARα through sequence analysis. To test whether AMPK might directly modulate PPARα activity, we performed reporter assays using PPARα mutants in which these serine residues were substituted with alanine. Mutation at the p38 MAPK phosphorylation sites (S6/12/21A) resulted in the loss of stimulation of transcriptional activity by AICAR that is observed in wild-type PPARα (Fig. 6, lanes 3 and 11). In contrast, the S167/373/453A mutant retained its sensitivity to AICAR (Fig. 6, lanes 3 and 7). These results indicate that p38 MAPK, but not AMPK, is directly involved in regulating the transcriptional activity of PPARα.

The regulation of fatty acid oxidation by adiponectin through PPARα would be expected to have a longer-lasting effect compared with the pathway involving ACC phosphorylation. To analyze the short- and long-term effects of adiponectin on fatty acid utilization, we compared the time-dependent changes in the phosphorylation of signaling molecules, expression of PPARα target genes, and fatty acid oxidation. As shown in Fig. 7A, phosphorylation of AMPK was transient, lasting for only 30 min after induction by adiponectin. Phosphorylation of p38 MAPK and ACC lasted longer and was detected at 60 min. During this short period, adiponectin would transiently induce fatty acid oxidation by phosphorylating and inhibiting ACC, which is confirmed by the fatty acid oxidation assay results (Fig. 7D). Meanwhile, the expression of PPARα target genes, specifically ACO and CPT1, peaked at 3 h after adiponectin treatment and slowly receded to basal levels after 24 h (Fig. 7B). Interestingly, the rate of fatty acid oxidation constantly remained higher in adiponectin-treated cells throughout the entire duration (Fig. 7C and D). Therefore, adiponectin seems to be able to increase fatty acid oxidation for a prolonged period by inducing the synthesis of proteins involved in fatty acid oxidation, in addition to the transient stimulation via ACC phosphorylation.

In this study, we have discovered a new mechanism by which adiponectin can stimulate fatty acid oxidation in muscle cells via sequential activation of AMPK, p38 MAPK, and PPARα. These results represent the first detailed characterization of the adiponectin signaling pathway that stimulates fatty acid oxidation via the activation of PPARα.

Accumulating results have unveiled adiponectin as a promising candidate for the treatment of obesity-associated metabolic syndromes. In animal models and human subjects with obesity and type 2 diabetes, adiponectin stimulates fatty acid oxidation, decreases lipid accumulation in muscles, lowers plasma free fatty acid levels, and enhances insulin sensitivity (10,37). Adiponectin has been shown to increase fatty acid oxidation via activation of AMPK and phosphorylation of ACC (35). However, the phosphorylation of ACC induced by adiponectin is short lived (35) (Fig. 7A). Therefore, this pathway cannot fully explain the long-term effects of adiponectin on weight loss and fatty acid oxidation. On the other hand, activation of PPARα would be a promising pathway by which adiponectin can trigger long-term stimulation of fatty acid oxidation, even after the initial signaling is gone, because this would promote the amounts of relatively long-lived enzymes involved in fatty acid metabolism. Indeed, we found that adiponectin, via AMPK, enhanced the transcriptional activity of PPARα and the expression of its target genes (Figs. 1 and 2). This result is supported by the fact that activation of AMPK has a potent lipid-lowering effect that is equivalent to the effect of PPARα agonists in obese animal models. In vivo infusion of AICAR significantly decreases plasma free fatty acid levels and the triglyceride concentration in both lean and obese Zucker rats (38,39).

Using dominant-negative AMPK and chemical inhibitors of AMPK and p38 MAPK, we revealed that AMPK and p38 MAPK are sequentially activated (Fig. 5 and online appendix Fig. 1). Previous studies have also suggested that the activation of p38 MAPK is associated with the stimulation of AMPK in glucose metabolism. For example, in skeletal muscle, AICAR increases glucose uptake by activating p38 MAPK, which is involved in the activation of GLUT4 translocation to the plasma membrane (40). In the same manner, p38 MAPK acts downstream of AMPK in cardiomyocytes, and the inhibition of the AMPK/p38 MAPK signaling pathway partially abolishes the stimulation of glucose uptake in response to hypoxia (41). Therefore, p38 MAPK and AMPK appear to be components in the adiponectin signaling pathway in muscle cells that work to increase insulin sensitivity by activating both glucose uptake and fatty acid utilization. However, treatment with adiponectin did not elevate fatty acid oxidation in primary hepatocytes (11,42) (unpublished data), suggesting that muscle and liver tissues may respond differently to adiponectin in terms of fatty acid oxidation.

Adiponectin circulates in serum as trimers and hexamers of relatively low molecular weight (LMW) and middle molecular weight and larger multimeric forms of high molecular weight (HMW) (32,42,43). Several studies have pointed to a greater role of HMW adiponectin in improving insulin sensitivity (31,43). Moreover, thiazolidinediones increased the level of circulating HMW adiponectin (44,45). When we compared different fractions of adiponectin containing either the HMW or LMW forms, the HMW adiponectin containing fractions exhibited pronounced effect on both PPARα target gene expression and AMPK and p38 MAPK phosphorylation, while LMW adiponectin exerted only marginal effects (online appendix Fig. 2). This is in concert with a previous study demonstrating a higher correlation between HMW adiponectin and whole-body fat oxidation than LMW adiponectin in human and rodents (46).

The mutation of putative AMPK phosphorylation sites on PPARα did not affect its responsiveness to AICAR. In contrast, mutant PPARα in which the potential p38 MAPK phosphorylation sites at serine residues were substituted to alanine (S6/12/21A) was unresponsive to AMPK activation (Fig. 6). This implies that AMPK controls the transcriptional activity of PPARα through p38 MAPK and not by direct modification. Of note, p38 MAPK is known to phosphorylate PPARα and increase its association with a coactivator PPARγ coativator-1α (36). However, the possibility of the presence of other AMPK phosphorylation sites in addition to the serine residues examined in our experiments cannot be ruled out. In addition, it has been proposed that adiponectin can enhance PPARα activity through the production of its ligands (47). These results suggest the presence of multiple pathways that could mediate the effect of adiponectin on fatty acid metabolism in muscles. Further research is required to validate these ideas.

Collectively, we have provided evidence that adiponectin, through the sequential activation of AMPK and p38 MAPK, stimulates PPARα activity in muscle cells (Fig. 8). There still remains missing links to be connected in this pathway, such as how adiponectin activates AMPK or what factors mediate the signaling pathways between AMPK and p38 MAPK. Nonetheless, our results have provided a possible explanation as to how adiponectin triggers sustained increases in fatty acid oxidation in muscle cells. Intervention of this pathway may provide a means by which obesity-related metabolic disorders may be treated.

This study was supported in part by grants from the Center for Biological Modulators of the 21C Frontier R&D Program, Molecular and Cellular Biodiscovery Research Program, and the National Research Laboratory Program. M.J.Y., J.J.C., Y.H.A., S.H.H., and J.B.K. were supported by the BK21 Research Fellowship from the Ministry of Education and Human Resources Development.

We thank D.P. Kelly for the muscle-type carnitine palmitoyl transferase promoter-luciferase vector and K.U. Lee and J.Y. Park for the adenovirus expressing dominant-negative AMPK. We are grateful to T.W. Nam and Y.J. Seok for assistance with the gel filtration assay.