Activation of the AMPK-FOXO3 Pathway Reduces Fatty Acid–Induced Increase in Intracellular Reactive Oxygen Species by Upregulating Thioredoxin

Human aortic endothelial cells (HAECs) (Cell Applications, San Diego, CA) were cultured in EGM-2 media (Cambrex, East Rutherford, NJ), which contained endothelial cell basic media, 2% FBS, hydrocortisone, fibroblast growth factor 2, vascular endothelial growth factor, IGF-I, epidermal growth factor, ascorbic acid, GA-1000, and heparin. The cells were transfected with small interfering RNAs (siRNAs) or plamid DNAs or treated with AICAR or palmitic acid at various concentrations and for the time indicated.

Saturated palmitic acid was used in this study. We prepared lipid-containing media by conjugating palmitic acid to BSA using a modification of the method described previously (19). Briefly, palmitic acid was dissolved in ethanol at 200 mmol/l and then combined with 10% FFA-free, low-endotoxin BSA, giving a final concentration of 1 to 5 mmol/l. The pH of all solutions was adjusted to ∼7.5, and the stock solutions were filter-sterilized and stored at −20°C until used. Control solutions containing ethanol and BSA were prepared similarly. Working solutions were prepared fresh by diluting the stock solution (1:10) in 2% FCS–endothelial cell basic media. All palmitic acid media contained 1% BSA; however, the palmitic acid–to–BSA ratio varied with the palmitic acid concentration.

Gene expression was silenced with specific siRNAs, including AMPK siRNA (Santa Cruz Biotechnology, Santa Cruz, CA), Trx siRNA (Santa Cruz Biotechnology), and forkhead transcription factor 3 (FOXO3) siRNA (Dharmacon, Chicago, IL). Various FOXO3a plasmids (Addgene, Cambridge, MA), including wild type (HA-FOXO3a WT), constitutively active (HA-FOXO3a TM), and dominant-negative (HA-FOXO3a TM deltaDB) DNAs were used in this study. Transfection of HAECs or human smooth muscle cells (HSMCs) with siRNAs or plasmid DNAs was carried out with Lipofectamine 2000 (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. Transfected cells were then treated with palmitic acid and AICAR at the designated concentrations for the indicated amount of time. The efficiency of transfection was confirmed by Western blot.

Intracellular ROS levels were detected with the oxidant-sensitive fluorogenic probes 5- (and -6) -chloromethyl-2′,7′-dichlorodihydrofluorescein diacetate, acetyl ester (CM-H₂DCFDA). Treated cells on the coverslip were incubated with 5 μmol/l CM-H₂DCFDA in serum-free medium for 30 min at 37°C. The slides were examined under a Leica DMLS epifluorescence microscope (Leica Microsystems, Bannockburn, IL), the images were captured with a Leica DC 100 digital camera using identical acquisition settings, and the data were analyzed with Image-Pro Plus V4.5 software (Media Cybernetics, Bethesda, MD). Fluorescence was detected and normalized to cell number. The mean fluorescent intensity was calculated randomly from five visual fields per coverslip. Relative ROS levels were compared and expressed as the percentage of the nontreatment control subjects.

Tissue ROS levels were detected with DHE. Fresh segments of thoracic aorta were frozen in optimal cutting temperature compound. Cryosections (6 μm) were equilibrated with Krebs-HEPES buffer (130 mmol/l NaCl, 5.6 mmol/l KCl, 2 mmol/l CaCl₂, 0.24 mmol/l MgCl₂, 11 mmol/l glucose, and 8.3 mmol/l HEPES; pH 7.4) at 37°C for 30 min. Cryosections were incubated with 2 μmol/l DHE at 37°C for 30 min and stained with the nuclear counterstain DAPI (0.1 μg/ml) at room temperature for 5 min. Fluorescence was detected and all images were captured with identical acquisition parameters. Values of red ethidium fluorescence were normalized to blue DAPI fluorescence. The mean fluorescent intensity randomly counted from three visual fields per vessel was calculated.

Cells were grown on glass coverslips and treated with palmitic acid and AICAR. Treated cells were washed with PBS, fixed with 4% paraformaldehyde for 10 min, and permeabilized with 0.2% Triton X-100 for 5 min. The coverslips were blocked with 1% BSA, incubated with the primary antibody, washed with PBS, incubated with Texas Red–labeled secondary antibody, and then stained with 0.1 μg/ml DAPI at room temperature for 5 min. Fluorescence was detected.

For immunohistochemical analysis, formalin-fixed, paraffin-embedded aortic sections were deparaffinized and rehydrated before antigen retrieval in citrate buffer (92–98°C for 12 min). Endogenous peroxidase activity was quenched by incubating the slides with 3% hydrogen peroxide for 10 min, and nonspecific staining was reduced by blocking with 5% normal blocking horse serum. The sections were incubated with the primary antibodies at 4°C overnight and then incubated with second antibody and detected with 3,3-diaminoben zidine (DAB) using the VECTASTAIN ABC kit (Vector Laboratories, Burlingame, CA). Nuclei were counterstained with hematoxylin. Slides treated only with normal IgG were used as negative controls. The images were captured and analyzed with Image-Pro Plus V4.5 software (Media Cybernetics). The signal density was normalized to vascular area. The mean intensity was calculated randomly from three visual fields per vessel.

Treated cells were collected and lysed as described previously (20). The NE-PER nuclear and cytoplasmic extraction kit (Thermo Fisher Scientific, Rockford, IL) was used to separate and prepare nuclear and cytoplasmic proteins from cultured HAECs. Protein samples (15 μg per lane) were subjected to SDS-PAGE and transferred to polyvinylidene fluoride (PVDF) membranes. The membranes were blocked, incubated with primary antibody, washed, and incubated with the secondary HRP-labeled antibody. Bands were visualized with enhanced chemiluminescence (Amersham Biosciences, Piscataway, NJ). Protein bands, including β-actin, were quantified by densitometry with the Quantity One imaging program (Bio-Rad, Hercules, CA). The relative protein levels were normalized to β-actin and expressed as the percentage of the nontreatment control subjects.

Total RNA from treated cells was extracted with Trizol (Invitrogen) according to the manufacturer's protocol. The mRNAs were reverse-transcribed with the iScript cDNA synthesis kit (Bio-Rad). Quantitative RT-PCR (qRT-PCR) was performed with the iCycler iQ RT-PCR detection system (Bio-Rad). Primers were designed with Beacon Designer 2.0 software (Premier Biosoft International, Palo Alto, CA). We used the following primers for human Trx: forward 5′-GCCTTGCAAAATGATCAAGC-3 and reverse 5′-TTGGCTCCAGAAAATTCACC-3′. mRNA levels were acquired by normalizing the threshold cycle (Ct) of Trx to the Ct of β-actin. The relative levels of mRNA were compared and expressed as the percentage of the nontreatment control subjects.

We used the chromatin immunoprecipitation (ChIP) assay kit (Upstate Biotechnology, Lake Placid, NY) as described previously (20). The immunoprecipitated DNA and the input DNA were quantified with the qRT-PCR detection system (Bio-Rad). The relative levels of DNA were normalized to the input DNA and expressed as the percentage of the nontreatment control. The PCR products were also separated on a 1.5% agarose gel. The following primers were used for the FOXO binding site in the 5′-flanking region of the human Trx gene: for site 6 forward primer 5′-CCGCACTAAACCGCTGTGTC-3′ and reverse primer 5′-CTCCGGAATTTACCGTGACC-3′.

Immunoprecipitation was conducted as described previously (21). Treated cells were lysed for 60 min in ice-cold extraction buffer containing 50 mmol/l Tris-Cl (pH 7.5), 100 mmol/l NaCl, 1% Triton X-100, 1 mmol/l dithiothreitol, 1 mmol/l EDTA, 1 mmol/l EGTA, 2 mmol/l Na₃VO₄, 50 mmol/l β-glycerophosphate, and a protease inhibitor mixture (Amersham Biosciences). Cleared cell lysates were incubated with the appropriate antibody precoupled to protein A/G-agarose beads (Santa Cruz Biotechnology) at 4°C overnight. The beads were washed twice with extraction buffer and then twice with extraction buffer containing 0.5 mol/l LiCl. Proteins were eluted either in kinase buffer for the kinase assay or in SDS sample buffer for Western blot analysis.

AMPK was precipitated from cell lysates with an anti-AMPK antibody. AMPK-containing beads were incubated with recombinant FOXO3 in kinase assay buffer supplemented with 100 μmol/l ATP for 20 min at 30°C. Samples were separated on a 10% SDS-PAGE and transferred to PVDF membranes. Anti-serine and anti-threonine antibodies were used to detect phosphorylated serines and threonines incorporated into FOXO3.

Trx activity was measured with the insulin disulfide reduction assay as described elsewhere (22). Total cellular protein was extracted with lysis buffer (20 mmol/l HEPES pH 7.9, 100 mmol/l KCl, 300 mmol/l NaCl, 10 mmol/l EDTA, 0.1% Triton X-100, 1 mg/ml Protease Inhibitor Cocktail III). Cellular protein extracts were incubated with buffer (50 mmol/l HEPES pH 7.6, 1 mmol/l EDTA, 1 mg/ml BSA, 2 mmol/l DTT) at 37°C for 15 min before they were incubated with Trx reductase (Sigma, St. Louis, MO) in the reaction buffer (0.3 mmol/l insulin, 200 μmol/l NADPH, 1 mmol/l EDTA, and 20 mmol/l HEPES pH 7.6) at 37°C for 20 min. The reaction was terminated by adding stop mix (6 mol/l guanidine HCl and 1 mmol/l DTNB in 0.2M Tris-HCl, pH 8.0), and the absorption at 412 nm was measured. Relative Trx activities were quantified after normalization with total protein and expressed as the percentage of the nontreatment control subjects.

Four-week-old apolipoprotein-E knockout (ApoE−/−) male mice (The Jackson Laboratory, Bar Harbor, ME) were fed a high-fat diet (Research Diets, New Brunswick, NJ) for 4 weeks and then subcutaneously injected with AICAR (0.5 mg/g body weight per day) or an equivalent volume of normal saline for 2 days. Mice were killed 24 h later. The aortas were irrigated with PBS, collected, and preserved at −80°C until used; alternatively, the aortas were fixed in 4% paraformaldehyde for the immunochemistry assay or optimal cutting temperature compound for ROS detection. All experiments were approved by the Animal Care Research at Baylor College of Medicine.

All quantitative variables are presented as the means ± SE from three separate experiments. We compared the differences of three or more groups with one-way ANOVA. Two-tailed P < 0.05 was considered statistically significant.

We first tested whether activation of the AMPK pathway could reduce FFA-induced ROS production. HAECs were incubated with increasing amounts of AICAR in the presence or absence of palmitic acid; ROS levels were detected in the treated cells. As shown in Fig. 1,A, AICAR treatment alone had minimal effects on basal ROS levels. Palmitic acid significantly increased intracellular ROS levels, an observation consistent with previous reports (23). The palmitic acid–induced increase in intracellular ROS levels was reduced by AICAR in a dose-dependent manner with up to a 60% reduction at the highest dose (500 μmol/l). This result indicates that activation of AMPK can reduce intracellular ROS levels. Additionally, suppression of AMPK by specific siRNAs not only increased basal ROS levels, but also augmented the palmitic acid–induced increase in ROS levels (Fig. 1 B). Furthermore, the AICAR-induced reduction in ROS levels was abolished by AMPK siRNA. These data suggest that the AMPK pathway is capable of reducing intracellular ROS levels under basal conditions and when induced by palmitic acid.

We investigated whether the AMPK pathway could reduce ROS levels through Trx. First, we examined whether the AMPK pathway could regulate Trx expression. As shown in Fig. 2,A, activation of the AMPK pathway by AICAR significantly upregulated expression of Trx in the absence and presence of palmitic acid. Although palmitic acid itself transiently increased Trx expression (data not shown), prolonged palmitic acid exposure decreased Trx expression. Importantly, knockdown of AMPKα by its specific siRNA inhibited both basal and AICAR-induced Trx expression (Fig. 2,B), implicating that the AMPK pathway is involved in upregulating Trx. Consistent with this expression pattern, AICAR significantly increased total cellular Trx activity in the absence and presence of palmitic acid (Fig. 2 C). Taken together, these data suggest that activation of AMPK increases Trx expression.

We determined whether Trx is involved in the AMPK-induced reduction in ROS levels. Trx expression was silenced by Trx siRNA, and the effect of AMPK on ROS levels in Trx knockdown cells was examined. As shown in Fig. 3, Trx siRNA not only increased basal ROS levels, but also amplified the palmitic acid–induced increase in ROS levels. Furthermore, Trx siRNA prevented the AICAR-induced reduction in ROS levels. These data suggest that Trx is capable of reducing intracellular ROS levels and is involved in the APMK-mediated reduction in ROS levels.

To further explore the mechanisms of AMPK-induced upregulation of Trx expression, we examined whether AICAR could affect Trx transcription. Using qRT-PCR, we found that AICAR significantly increased Trx mRNA in a dose-dependent manner (Fig. 4,A). Knockdown of AMPKα by its specific siRNA reduced basal Trx expression (Fig. 4 B). Moreover, AICAR-induced Trx mRNA was reduced in the presence of AMPK siRNA. These results indicate that AMPK increases Trx expression at the mRNA level.

We investigated the mechanisms responsible for the AMPK-mediated upregulation of Trx. The 5′ flanking region of the human Trx gene contains consensus-binding sites for many transcription factors. We identified FOXO3 as one of these transcriptional factors that may mediate AMPK-induced Trx transcription. Silencing FOXO3 with siRNA significantly prevented the AICAR-induced expression of Trx at both the protein (Fig. 5,A) and mRNA level (Fig. 5,B), indicating that FOXO3a is involved in the AMPK-induced upregulation of Trx. Furthermore, overexpression of constitutively active FOXO3a (FOXO3a CA) significantly increased Trx expression in the absence or presence of AICAR, but domain-negative FOXO3a (FOXO3a DN) dramatically decreased Trx expression (Fig. 5 C), further suggesting that FOXO3a is capable of upregulating Trx expression. Together, these data support the critical role of FOXO3 in the AMPK-induced upregulation of Trx.

To examine whether FOXO3 directly induces Trx transcription, we examined whether FOXO3 binds to the Trx promoter in vivo. FOXO binds to the consensus site 5′-(C/G)(A/T)AAA(C/T)A-3′ (24). The promoter region in the Trx gene contains six putative FOXO binding sites (tgAAAGAgtga at −1346/−1342, tgAAAGAagga at −1339/−1335, gaAAACAcaga at −1236/−1232, caAAATAccgc at −859/−855, ggAAACActga at −807/−803, and tgAAAGAacag at −613/−609) (Fig. 6,A). Results of the ChIP assay performed with a FOXO3 antibody showed that FOXO3 strongly bound to site 6 (Fig. 6,B). Importantly, the binding of FOXO3 to the Trx promoter was significantly increased by AICAR treatment (Fig. 6 B, P < 0.001), further suggesting that FOXO3 may mediate AMPK-induced Trx transcription.

We examined how the transcription factor FOXO3 binds to the Trx promoter and induces gene expression. One possible mechanism is chromatin remodeling in which transcription factors bind to a promoter and recruit histone acetylases (e.g., p300) that acetylate histones, unwind chromatin, and induce gene transcription. To test whether this mechanism applied to the FOXO3-mediated transcription activation of the Trx gene, we first examined whether p300 could bind to the Trx promoter. By using the ChIP assay, we found that AICAR treatment significantly increased p300 binding to the Trx promoter (Fig. 6,C). We then examined whether FOXO3 was associated with p300 in vivo. By using a coimmunoprecipitation assay, we observed that FOXO3 was associated with p300 and that this association was increased by AICAR treatment (Fig. 6,D), indicating that p300 recruitment to the Trx promoter may be mediated at least in part by FOXO3. Furthermore, the double-ChIP assay (Fig. 6 E) showed that FOXO and p300 were in the same transcription complex in the Trx promoter, and this association was increased by AICAR treatment. Together, these results suggest that activated FOXO3 may recruit p300 and form a transcription activation complex in the Trx promoter, a process that can be promoted by the AMPK pathway.

We investigated the potential mechanisms by which AMPK regulates FOXO3. The immunostaining assay (Fig. 7,A) and Western blot (Fig. 7,B) showed that AICAR significantly increased the translocation of FOXO3 from the cytoplasm to the nucleus, which was prevented by AMPKα siRNA. We also examined whether AMPK could directly phosphorylate FOXO3. The in vitro kinase assay, with purified AMPK as the kinase and recombinant FOXO3 as the substrate, showed that AMPK directly phosphorylated FOXO3 at serine and threonine sites and that AICAR increased threonine phosphorylation of FOXO3 (Fig. 7 C). These results indicate that FOXO3 may be phosphorylated by AMPK and subsequently translocate into the nucleus where it binds the Trx promoter and increases Trx transcription.

Finally, we examined whether AMPK activation could affect the expression of Trx and, thus, reduce ROS levels in vivo. We used ApoE−/− mice fed a high-fat diet, a model that can produce metabolic disturbances, ROS overproduction, and vascular changes similar to those seen in metabolic syndrome. These mice were injected with either saline or AICAR (0.5 mg/g body weight per day) for 2 days; ROS levels and Trx expression in the aorta were compared. As shown in Fig. 8, the AICAR injection decreased ROS levels (Fig. 8,A) and increased expression of Trx in the aortic wall (Fig. 8 B), suggesting that activation of the AMPK pathway may enhance Trx expression and subsequently reduce ROS levels in the vascular wall.

In the present study, we showed that activation of AMPK reduces ROS levels by inducing expression of the antioxidant Trx. The transcriptional factor FOXO3 mediated the induction of transcription. AMPK activates FOXO3 by promoting its nuclear translocation, Trx promoter binding, and subsequently transcription complex formation. Based on these findings, we propose a pathway of the AMPK-mediated reduction of intracellular ROS (Fig. 8 C).

Fatty acids are fuels that are used to efficiently generate ATP primarily through β-oxidation. However, when fatty acids are present in excessive amounts, along with increased oxidation and energy generation, they produce increased ROS, which contribute significantly to the pathogenesis of microvascular and macrovascular complications in diabetes (25,26). Strategies to decrease intracellular ROS levels and oxidative damage may have therapeutic potential in treating diabetes and its complications.

Our finding that activation of the AMPK pathway reduced intracellular ROS levels is consistent with previous reports (7,,–10). The AMPK pathway acts as a fuel gauge by switching on catabolic pathways for ATP generation when energy is depleted, a process coupled to the increase in ROS production. It has been shown that ROS can activate the AMPK pathway (5,–7). The ability of AMPK to reduce ROS levels counterbalances the overproduction of ROS during fatty acid consumption. Reducing fatty acid–induced increases in ROS levels in endothelial cells may be an important mechanism in AMPK-mediated cardiovascular protection. Additionally, AMPK regulates endothelial function (2), angiogenesis (27), and the cell cycle (28). Moreover, AMPK also inhibits vascular inflammation (3), prevents endothelial injury induced by hyperglycemia and FFAs (4), and reduces myocardial infarction (29). Thus, upregulating this pathway may provide therapeutic benefits by not only reducing lipid storage and insulin resistance, but also preventing cardiovascular complications in metabolic syndrome.

We studied the mechanisms involved in the AMPK-mediated reduction in ROS. Decreasing intravascular ROS levels can be achieved by preventing the generation of or removing excess reactive species. Previous studies suggest that activation of the AMPK pathway normalizes hyperglycemia-induced ROS production by inducing manganese superoxide dismutase (8,30). We have shown for the first time that the AMPK pathway can decrease fatty acid–induced increases in intracellular ROS levels by upregulating Trx, a novel additional mechanism that explains AMPK's effects on reducing intracellular ROS. Trx is ubiquitously expressed in endothelial cells and protects the cells from ROS-induced cytotoxicity (15). Trx can also bind and inhibit apoptosis signal-regulating kinase 1 (31), an upstream kinase in the cellular stress–sensitive pathways (i.e., JNK and p38 pathways). Thus, upregulation of the Trx system by the AMPK pathway may be an important protective mechanism against excessive oxidative stress and the activation of stress-signaling pathways in the body.

Regarding how the AMPK pathway induces Trx expression, our study suggests that FOXO3 may be a target transcription factor that mediates AMPK's effects on Trx expression and ROS reduction. FOXO transcription factors are important regulators of metabolism, cell-cycle progression, apoptosis, and oxidative stress resistance. Recent findings suggest that ROS can activate FOXO (33,,,–37). Although FOXOs mediate ROS-induced apoptosis (37,38) under lethal conditions, they can increase cell survival in response to physiologic oxidative stress (32,39,,–42), a function that is required for long-term regenerative potential and cell longevity (41,43).

The mechanisms whereby FOXO3 reduces ROS levels are not well defined. It has been shown that FOXO3 may be involved in the induction of catalase (44). Our study shows that FOXO3 reduces intracellular ROS levels by directly inducing the antioxidant Trx. When activated by AMPK, FOXO3 directly binds to the Trx promoter and forms a transcriptional complex on the Trx promoter, which may lead to activation of Trx transcription. However, further site-directed mutagenesis studies are needed to determine whether FOXO3 indeed targets site 6 and induces the transcriptional complex formation in the Trx promoter and whether this site is important for FOXO3-mediated Trx promoter transactivation. Together, these findings indicate the importance of FOXO3 in reducing ROS levels and protecting cells.

Increasing evidence suggests that AMPK can directly phosphorylate FOXO3, which mediates AMPKs ability to reduce cell stress and increase cell survival (45,46). Greer and colleagues have recently shown that AMPK directly phosphorylates at least six residues in the C-terminal domain of FOXO3, which activates the FOXO3 transcription factor (45). Our results support their findings. We showed that activation of AMPK by AICAR induced the nuclear translocation of FOXO3 and the binding of FOXO3 to the Trx promoter. Further studies will be necessary to define the detailed mechanisms of FOXO3 regulation by the AMPK pathway in response to metabolic stress.

In summary, using both in vitro and in vivo experiments we have shown that activation of the AMPK pathway significantly reduced palmitic acid–induced intracellular ROS levels by increasing the expression of the antioxidant Trx. The transcriptional factor FOXO3 mediated AMPK's effect on Trx expression. AMPK upregulated Trx transcription by increasing the nuclear translocation of FOXO3 and by promoting its binding to the Trx promoter. The AMPK-FOXO pathway has protective effects against cellular superoxide levels induced by metabolic stress and could be a therapeutic target when treating cardiovascular diseases in metabolic syndrome.

This study was supported by American Heart Association Grants AHA-TX 0565134Y (to Y.H.S.) and AHA-0730190N (to Y.H.S.), National Institutes of Health Grant R01-HL071608 (to X.L.W.), and Natural Science Foundation of China Grants 2006CB503803 and 2006AA02A406 (to Y.Z.).

No potential conflicts of interest relevant to this article were reported.

Parts of this study were presented in abstract form at the 68th Scientific Sessions of the American Diabetes Association, San Francisco, California, 6–10 June 2008.

We acknowledge Hilary D. Marks, PhD, of the Texas Heart Institute at St. Luke's Episcopal Hospital, for her editorial assistance.