Activation of Protein Phosphatase 2A by Palmitate Inhibits AMP-activated Protein Kinase*

Elevated levels of free fatty acids contribute to cardiovascular diseases, but the mechanisms remain poorly understood. The present study was aimed to determine if free fatty acid inhibits the AMP-activated kinase (AMPK). Exposure of cultured bovine aortic endothelial cells (BAECs) to palmitate (0.4 mm) but not to palmitoleic or oleic acid (0.4 mm) for 40 h significantly reduced the Thr172 phosphorylation of AMPK-α without altering its protein expression or the phosphorylation of LKB1-Ser428, a major AMPK kinase in BAECs. Further, in LKB1-deficient cells, palmitate suppressed AMPK-Thr172 implying that the inhibitory effects of palmitate on AMPK might be independent of LKB1. In contrast, 2-bromopalmitate, a non-metabolizable analog of palmitate, did not alter the phosphorylation of AMPK and acetyl-CoA carboxylase. Further, palmitate significantly increased the activity of protein phosphatase (PP)2A. Inhibition of PP2A with either okadaic acid, a selective PP2A inhibitor, or PP2A small interference RNA abolished palmitate-induced inhibition on AMPK-Thr172 phosphorylation. Exposure of BAECs to C2-ceramide, a cell-permeable analog of ceramide, mimicked the effects of palmitate. Conversely, fumonisin B1, which selectively inhibits ceramide synthase and decreases de novo formation of ceramide, abolished the effects of palmitate on both PP2A and AMPK. Inhibition of AMPK in parallel with increased PP2A activity was founded in C57BL/6J mice fed with high fat diet (HFD) rich in palmitate but not in mice fed with HFD rich in oleate. Moreover, inhibition of PP2A with PP2A-specific siRNA but not scrambled siRNA reversed HFD-induced inhibition on the phosphorylation of AMPK-Thr172 and endothelial nitric-oxide synthase (eNOS)-Ser1177 in mice fed with high fat diets. Taken together, we conclude that palmitate inhibits the phosphorylation of both AMPK and endothelial nitric-oxide synthase in endothelial cells via ceramide-dependent PP2A activation.

Evidence accumulated over the past several years indicates that the AMP-activated protein kinase (AMPK) 2 may be a ther-apeutic target for treating insulin resistance and type 2 diabetes (1). AMPK is a heterotrimeric protein formed by an ␣ subunit, which contains the catalytic activity, and by the ␤ and ␥ regulatory subunits important in maintaining stability of the heterotrimer complex (2). AMPK belongs to a family of energy-sensing enzymes functioning as a "fuel gauge" that monitors changes in the energy status of a cell (3,4). When activated, AMPK shuts down anabolic pathways and promotes catabolism in response to an elevated AMP/ATP ratio by down-regulating the activity of several key enzymes of intermediary metabolism (4). Two primary acute consequences of AMPK activation are 1) an increase in glucose uptake by induction of glucose 4 transporter microvesicle cytoplasm to membrane translocation and fusion and 2) an increase in fatty acid oxidation by phosphorylation and inactivation of acetyl-CoA carboxylase (ACC), the rate-limiting enzyme in fatty acid synthesis (5). Therefore, the AMPK signal pathways are thought to play a central role in the regulation of cellular glucose and lipid homeostasis.
The control of AMPK activity is complex, and the classic view is that AMPK is activated allosterically by an increase in the intracellular AMP/ATP ratios and/or by the phosphorylation of threonine 172 within the ␣ subunit. Several protein kinases responsible for this phosphorylation have been identified. They include Peutz-Jeghers syndrome kinase LKB1 (LKB1) (6), and the Ca 2ϩ /calmodulin-dependent protein kinase kinase (7). Protein phosphorylation signal transduction systems are balanced and regulated delicately by both phosphatase and kinase. Since AMPK is activated by (a) protein kinase(s) at the threonine 172 residue, one can easily assume that AMPK can be regulated negatively by (a) serine/threonine phosphatase(s). To date, a wide range of physiological stressors, pharmacological agents, and hormones associated with increase in the intracellular AMP/ATP ratios have been demonstrated to activate AMPK (8). AMPK is also thought to be regulated by glycogen (9), which is the major cellular storage form of carbohydrates and thus, an additional indicator of cellular energy status. Lipids are the other major energy source for cellular metabolism. Recent studies (10,11) in heart and liver have revealed that AMPK may be sensitive to the "lipid status" of a cell, and activation may be influenced by intracellular fatty This article has been withdrawn by the authors. The Journal raised questions regarding Figs. 1A, 4C, 5A, and 8 (A and C). Twelve years after the publication, the authors were able to locate some, but not all, of the original data and were able to locate some repeated experiments performed at the time of the original work, which the authors state support the conclusions of the paper. The authors state that the results of this paper are confirmed acid availability independent of the cellular AMP levels. Indeed, a reduced fat oxidative capacity has been reported in type 2 diabetic patients (12). Also, high fat diet feeding significantly decreases phospho-AMPK in the liver and muscles of rodents (13,14). AMPK activity is reduced in aortic endothelium or skeletal muscle of obese rats compared with lean animals (15,16). These data raise the possibility that chronic exposure of fatty acids to cells inhibit AMPK activation. However, the mechanisms by which fatty acids inhibit AMPK are poorly understood.
In an attempt to understand the mechanism underlying which chronically increased FFA inhibits AMPK, we examined the effects of the saturated fatty acid palmitate, which makes up 30 -40% of plasma FFA, in cultured endothelial cells and in mice. Here we report that palmitate inhibited AMPK via ceramide-dependent PP2A activation in vivo.
Fatty acids were added to the culture medium as a fatty acid-BSA complex. Briefly, a stock solution of fatty acids was directly prepared in 95% ethanol and kept at 4°C. The fatty acids were added to the cell cultures coupled to fatty acid-free bovine serum albumin (BSA) in the ratio of 2 mol of fatty acid to 1 mol of albumin. These complexes were constituted in stopper-covered flasks by adding the appropriate volume of the ethanolic FFAs solution to the albumin previously dissolved in culture medium containing no fetal bovine serum (here termed SF-EBM). These solutions were gently stirred and sterilized by filtration through 0.2-m filters. Final concentrations of ethanol in the culture medium were kept to Ͻ0.1%. The concentrations of endotoxins in BSA were very low, as assessed by the supplier (3 enzyme units/mg BSA versus ϳ30 -60 enzyme units/mg of BSA for standard albumin preparations).
Fatty acid-BSA complexes were added to culture dishes at a final concentration of 0.4 mM fatty acid. Controls were incubated with equal concentrations of FFA-free albumin as present in fatty acid-treated cells. In some experiments, BAECs were incubated in the presence or absence of either 2-bromopalmitate (2-BrP, 0.4 mM), C 2 -ceramide (15 M), fumonisin B1 (15 M), AICAR (2 mM), or okadaic acid (OA, 2 nM), and peroxynitrite (ONOO Ϫ , 50 M). C 2 -ceramide and fumonisin B1 were first dissolved in prewarmed 37°C Me 2 SO (Sigma) at a concentration of 15 mM and were simply added to the cell culture media resulting in a 1000-fold dilution. The final Me 2 SO concentration is 0.1%. For control experiments, BAECs were exposed to solvent alone (0.1% Me 2 SO).
Cell Viability and Apoptosis-To exclude the potential contribution of cell death to the effects of FFA on AMPK phosphorylation in our experimental conditions, we first verified cell viability after 40 h of culture in the presence of either 0.4 mM palmitic or oleic acid. Cells were rinsed with phosphate-buffered saline, trypsinized, washed with medium, centrifuged, and resuspended in phosphate-buffered saline. Next, cells were mixed with the same volume of 0.25% trypan blue and transferred to a slide for 3 min. A total of 300 cells was microscopically counted using a hemocytometer to determine the dead cell (stained blue) rate. The experiments were performed in triplicate. Compared with the control (cells not exposed to FFA), no significant differences were detected for cell viability after exposing the cells for 40 h to the treatment conditions. Apoptosis was measured using the Cell Death Detection ELISA (Roche Diagnostics, Mannheim, Germany) to detect cytoplasmic histone-associated DNA fragments (mono-and oligonucleosomes) according to the manufacturer's protocol. Data were normalized for comparison by total protein concentration (Bio-Rad).
Animals-Male C57BL/6J mice were purchased from the Jackson Laboratory. The animals were maintained in a temperature-controlled room (22°C) on a 12-h light-dark cycle. The study was approved by the Institutional Animal Care and Use Committee at University of Oklahoma Health Sciences Center. One week after arrival, mice were divided into three groups and were fed with normal chow, diet rich in palmitic acid (palmitate-HFD, high palmitate, 16:0 ϭ 50% of total fatty acids) or diet rich in oleic acid (oleate-HFD, high oleic, 18:1 ϭ 80% of total fatty acids, Research Diets, New Brunswick, NJ) for 3 months.
Food intake and body weight were measured once a week. The intraperitoneal glucose tolerance test was performed after 12 weeks in 6 mice from each dietary group. Briefly, mice fasted for 12 h and blood was drawn from the tail vein at 0, 5, 15, 30, 60, and 120 min after intraperitoneal injection of glucose (2 g/kg body weight). The blood glucose was assayed using a blood glucose meter (LifeScan, Inc.). The trapezoid rule was used to determine the area under the curve (AUC) for glucose concentrations in each animal.
Finally, after 12 weeks, blood samples were taken from the intraorbital, retrobulbar plexus from non-fasted, anesthetized mice to measure basal plasma levels of glucose and insulin. Insulin was determined using a Mouse Insulin Elisa Kit (Linco Research, St. Charles, MO). The mice were sacrificed, and thoracic aortas were immediately isolated and snap frozen in liquid nitrogen. The manipulation times were reduced to the minimum between aorta isolation and storage in liquid nitrogen. To avoid additional phosphorylation/dephosphorylation, Krebs-Ringer bicarbonate solution, which was used to rinse isolated aortas, was insufflated with 95% O 2 and 5% CO 2 , and the mixtures were placed on ice to prevent tissue hypoxia. Mouse aortas were subsequently homogenated for determination of AMPK, ACC, and eNOS. In some experiments, the adventitia was removed from the isolated aortas. Aortas were then cut open along the ordinate axis, and the endothelium was removed by gentle rubbing of the intima with curved forceps.
Immunohistochemistry-Ceramide formation was determined using immunohistochemistry. After incubation with media containing palmitic, palmitoleic, or oleic acid, the cells were fixed with 100% methanol for 3 min at Ϫ20°C and then blocked with 1% normal goat serum in phosphate-buffered saline. Ceramide was detected using a specific anti-ceramide antibody (Alexis, Carlsbad, CA; clone MID 15B4, 1:200 dilutions in blocking buffer) and assayed using an Alexa 488-labeled anti-mouse antibody (Molecular Probes, 1:200 dilution in blocking buffer). Cells were examined under a fluorescence microscope (Olympus IX71) and pictures were taken for analysis.
siRNA Construction and Infection-Double-stranded RNA with sequence 5Ј-CCAUUCUUCGAGGAAAUCAtt-3Ј was designed to target the open reading frame of the bovine PP2A catalytic subunit C␣ (GenBank TM accession number M16968) and purchased from Ambion (Austin, TX). A scrambled sequence served as control. Transient infection of siRNA was carried out using Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol. Cells were plated in a 6-well plate at 2.0 ϫ 10 5 cells/well in 1.5 ml of EBM supplemented with 10% serum without antibiotics. BAECs were transfected with either scrambled or PP2A-specific siRNA oligonucleotides at a final concentration of 80 nM. No cell toxicity was observed at the concentrations of siRNA used. The PP2A expression was verified in Western blots. We found the maximal inhibition of PP2A was achieved after 2 days infection, and cells were treated on day 3 post infection.
For mice, Double-stranded RNA, with sequence 5Ј-GC-CUCUUGUCAUCAACAGCtt-3Ј, was designed to target the open reading frame of the mouse PP2A catalytic subunit C␣ (GenBank TM accession number NM019411) and was purchased from Ambion. 50 g of siRNAs was mixed with in vivo-jetPEI TM (Qbiogene, Carlsbad, CA) at an N/P ratio of 5 at room temperature for 15 min. For intravenous administration, 200 l of the mixture containing the indicated amounts of siRNA was injected retro-orbitally. After 48 h, the mice were anesthetized and sacrificed. Aortas were isolated for biochemical assays.
Assay of Phosphatase Activities-PP2A activity was measured by using threonine phosphopeptide (KRpTIRR) as the substrate with the PP2A Immunoprecipitation Phosphatase Assay Kit (Upstate Biotechnology). The cells were lysed in lysis buffer (50 mM HEPES (pH 7.5), 150 mM NaCl, 1 mM EGTA, 10% glycerol, 1.5 mM magnesium chloride, 1% Triton X-100, 1 g of leupeptin/ml, 50 units of aprotinin/ml, 1 mM phenylmethylsulfonyl fluoride). Clarified supernatants were incubated with anti-PP2A antibody and protein A-agarose for 2 h at 4°C. As for mice, PP2A immunoprecipitates in total aorta were prepared as described previously (18). The washed immunoprecipitates were resuspended in p-nitrophenyl phosphate Ser/Thr assay buffer, provided by the kit, and incubated for 2 h at 4°C. After washing the beads three times, the diluted phosphopeptide (750 M) and Ser/Thr assay buffer were added, and the mixture was incubated for 10 min at 30°C followed by addition of the Malachite Green Phosphate Detection Solution. PP2A activity in the reactive system was determined by measuring the absorbance at 650 nm and comparing absorbance values of samples to negative controls containing no enzyme.
The activity of PP2C in cell culture with p-nitrophenyl phosphate as substrate was essentially assayed as previously described (19). The initial rate of liberation of p-nitrophenol was determined spectrophotometrically at 405 nm.
Western Blot Analysis-BAEC, HeLa-S3, A549, and VSMC cells were washed twice with cold phosphate-buffered saline and lysed in cold radioimmune precipitation assay buffer containing 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, and a mixture of protease inhibitors (Roche Applied Science). The protein concentrations were determined with a bicinchoninic acid protein assay system (Pierce). Proteins were subjected to Western blots. The antibody bindings were detected by using ECL-Plus, as described previously (20).
Quantification of Western Blot-The intensity (area ϫ density) of the individual bands on Western blots was measured by densitometry (model GS-700, Imaging Densitometer, Bio-Rad). The background was subtracted from the calculated area.
Expression of Data and Statistics-Data are expressed as mean Ϯ S.E. Intergroup comparisons were performed by Student's paired t test. p Ͻ 0.05 was considered significant.

Effects of Palmitate on Phosphorylation of AMPK and Its
Downstream Enzyme, ACC-Ser 79 -Previous studies had demonstrated that palmitate activates AMPK in rats heart (10) and FIGURE 2. Inhibition of AMPK by palmitate is LKB1-independent. A, phosphorylation of LKB1 Ser 428 was not affected by chronic incubation of 400 M palmitate in BAECs. Confluent BAECs were exposed to different treatment factors indicated and phosphorylated LKB1-Ser 428 was detected by the phosphospecific antibody in Western blots. The blot is a representative of three blots obtained from three independent experiments. B, summary data (n ϭ 3; *, p Ͻ 0.05 versus control). C and E, LKB1 is not required for palmitate-induced AMPK inhibition. Confluent LKB1-deficient HeLa-S3 (C) or A549 cells (E) were exposed to 400 M palmitate for 40 h, and AMPK phosphorylation was assayed as described under "Experimental Procedures." The blot is a representative of three blots obtained from three independent experiments. The lower panels (D and F) give the summary data (n ϭ 5; *, p Ͻ 0.05 versus control).
skeletal muscle cells (21) after acute treatment. To investigate whether or not palmitate activated AMPK in endothelial cells, BAECs were treated with various concentrations of palmitate for 1-40 h. AMPK activation was monitored in Western blots by staining with a specific antibody against phosphorylated Thr 172 of AMPK, which is essential for its activity (22). AICAR is a compound taken up by the cells and phosphorylated to the monophosphate form ZMP, which can accumulate in the cell, mimicking the effect of AMP on AMPK phosphorylation and activation (4,23). Thus, AICAR (2 mM, 2 h) was used as a positive control for AMPK activation. Because ACC is a substrate for AMPK (23), the determination of ACC phosphorylation also served as an indicator of AMPK activity. As shown in Fig. 1, AICAR elicited a 2.8-to 3.3-fold increase in both AMPK and ACC-Ser 79 , respectively. As depicted in Fig. 1A, exposure to low concentrations of palmitate (0.05-0.2 mM) for 40 h had no effect on the basal levels of both AMPK-Thr 172 and ACC-Ser 79 in BAECs. However, exposure of BAECs to palmitate at high concentrations (0.4 and 0.6 mM) for 40h significantly suppressed the levels of both AMPK-Thr 172 and ACC-Ser 79 (Fig. 1,  A and B).
We next examined the time dependence of palmitate on both AMPK-Thr 172 and ACC-Ser 79 . Interestingly, acute exposure of BAECs to palmitate for 1 h caused a transient increase in both AMPK-Thr 172 and ACC-Ser 79 phosphorylation. The levels of AMPK and ACC were reduced to the basal levels at 5-h incubation. Prolonged incubation of BAECs with palmitate (20 -40 h) caused a significant reduction of both AMPK-Thr 172 and ACC-Ser 79 (Fig. 1, C and D). In addition, palmitate did not affect the total content of both AMPK and ACC (Fig. 1), suggesting that palmitate-suppressed phosphorylation of both AMPK and ACC was not due to altered expression of AMPK and ACC. As depicted in Fig. 1 (E and F), 0.4 mM palmitate also reduced both basal and AICAR-stimulated AMPK and ACC phosphorylation at 40 h. Because 0.4 mM palmitate caused the greatest extent of inhibition on the phosphorylation of AMPK and ACC at 40 h, we exposed BAECs to 0.4 mM palmitate in the rest of the study.
Inhibition of AMPK by Palmitate Is LKB1-independent-Two recent studies (6,24) had identified that LKB1 acts as one of the AMPK upstream kinases, AMPK kinase. We then determined whether or not palmitate-induced AMPK inhibition was due to its inhibition on an AMPK kinase such as LKB1. We determined if palmitate altered the phosphorylation of LKB1 at Ser 428 , which is essential for ONOO Ϫ -induced AMPK activation (25). As depicted in Fig. 2A, ONOO Ϫ significantly increased the levels of LKB1-Ser 428 phosphorylation compared with its control. In contrast, palmitate did not alter the levels of LKB1-Ser 428 .
We next determined if palmitate inhibited AMPK in either HeLa-S3 or A549, two LKB1-deficient cells. Similar to that seen in BAECs, palmitate (0.4 mM) inhibited the Thr 172 phosphorylation of AMPK in both HeLa-S3 and A549 by 60 and 70%, respectively (Fig. 2, C and E). Because palmitate inhibited AMPK Thr 172 phosphorylation in both BAEC and LKB1-depleted cells, these results suggest that palmitate-induced reduction of AMPK-Thr 172 might be independent of its upstream kinase, LKB1.
Inhibition of AMPK by Palmitate Requires Palmitate Activation but Is ␤-Oxidation-independent-The inhibitory effects of saturated FFA on AMPK phosphorylation could either be due to a direct action or due to products generated by its metabolism. To distinguish between direct and indirect effects of FFA we determined the effects of a non-metabolizable analog of palmitate, 2-BrP, on AMPK and ACC phosphorylation. As shown in Fig. 3, 2-BrP, unlike palmitate, did not suppress the phosphorylation of both AMPK and ACC, suggesting that the effect of palmitate was dependent on its metabolism in endothelial cells.
AMPK, as defined by its name, is regulated by the ratios of AMP to ATP. Increased oxidation of palmitate might inhibit AMPK by increasing ATP via enhanced mitochondrial oxidation of palmitate. To this end, BAECs were preincubated with 2-BrP (0.4 mM) 2 h prior to the addition of palmitate. 2-BrP inhibits long-chain fatty acid ␤-oxidation and consequent ATP  MARCH 30, 2007 • VOLUME 282 • NUMBER 13 production by irreversibly binding to carnitine-palmitoyltransferase 1 (26). As shown in Fig. 3 (A and B), 2-BrP did not alter palmitate-induced inhibition on both AMPK and ACC phosphorylation. These results suggest that increased oxidation of FFA was unlikely a contributor to palmitate-induced AMPK inhibition.

Palmitate Inhibits AMPK
Inhibition of AMPK by Palmitate Is Independent of Apoptosis-There is evidence that palmitate at high concentrations increases apoptosis. We next determined if AMPK inhibition by palmitate was due to accelerated apoptosis. As shown in Fig. 3C, exposure of BAECs to 0.4 mM palmitate did not increase endothelial cell apoptosis. Compared with control groups, elevated concentrations of palmitate (0.6 -0.8 mM) significantly increased endothelial apoptosis (2.6-and 3.2-fold, respectively, p Ͻ 0.01, data not shown). In addition, 2-BrP had no effect on apoptosis in palmitate-treated cells. Thus, palmitate-induced AMPK inhibition was not likely to be related to palmitate-induced apoptosis.
PP2A Mediates Palmitate-induced Decrease in AMPK Phosphorylation-PP2A is a major protein serine/threonine phosphatase that regulates many signaling pathways in mammalian cells (27). To determine whether the inhibitory effect of palmitate on AMPK phosphorylation was due to the activation of PP2A, we applied a cell-permeable inhibitor of PP2A, OA, which inhibits PP2A at a very low concentration (2 nM) (28). As depicted in Fig. 4 (A and B), OA significantly attenuated the inhibitory effects of palmitate on both AMPK and ACC phosphorylation. Furthermore, OA also enhanced both AMPK and ACC phosphorylation in both control and BAECs treated with AICAR (Fig. 4, A and B), suggesting that PP2A might negatively regulate AMPK in BAECs.
To further examine if palmitate inhibited AMPK by activating PP2A in BAECs, we suppressed the expression of PP2A by applying PP2A siRNA in BAECs. As shown in Fig. 4C, PP2A siRNA suppressed the expression of PP2A by 68%. As expected, suppression of PP2A by PP2A siRNA increased both AMPK and ACC phosphorylation in both unstimulated and AICAR-treated BAECs, confirming that PP2A might negatively regulate AMPK phosphorylation in BAECs in basal conditions. Further, infection of PP2A siRNA but not scrambled siRNA reversed the inhibitory effect of palmitate on AMPK and ACC, implying that palmitate-induced AMPK inhibition might be due to the activation of PP2A (Fig. 4D).

Inhibition of AMPK by Palmitate but Not by Palmitoleic Acid or
Oleic Acid-There is evidence that ceramide is a potent activator of PP2A (29). We next determined whether ceramide was involved in palmitate-induced AMPK inhibition. Because palmitate (16:0), but not its monounsaturated counterpart palmitoleate (16:1) or oleate (18:1), is correlated with de novo synthesis of ceramide (30, 31), we employed both fatty acids to test if ceramide participates in palmitate-induced AMPK inhibition. As shown in Fig. 5 (A and B), chronic incubation of BAECs with 0.4 mM palmitate attenuated the levels of AMPK-Thr 172 , whereas neither 0.4 mM palmitoleate nor oleate had effects. Because palmitate but not the monounsaturated palmitoleic and oleic acid increases the levels of ceramide, these results suggest that palmitate might inhibit AMPK by increasing ceramide synthesis.
Ceramide-dependent AMPK Inhibition by Palmitate-We next determined if ceramide mimicked the effects of palmitate on AMPK phosphorylation. To this end, BAECs were exposed to exogenous C 2 -ceramide, a cell-permeable ceramide analog. Compared with control, C 2 -ceramide (15 M) significantly suppressed the levels of AMPK Thr 172 phosphorylation without altering the expression of AMPK ␣ expression (Fig. 5, C and D). In contrast, the metabolically inactive ceramide analog, C 2 -dihydroceramide, did not alter palmitate-induced AMPK phosphorylation (data not shown).
We next assayed if selective inhibition of ceramide synthesis attenuated the effects of palmitate on AMPK phosphorylation. Fumonisin B1, a specific inhibitor of ceramide synthase, can inhibit ceramide de novo formation. Concurrently, administration of fumonisin B1 (15 M) with 0.4 mM palmitate for 40 h significantly reduced the inhibitory effect of palmitate on AMPK phosphorylation (Fig. 5, E and F). In addition, fumonisin B1 also increased AMPK phosphorylation in non-stimulated cells by 40% (p Ͻ 0.05, Fig. 5, E and F). Taken together, these findings implied that palmitate attenuated AMPK via de novo ceramide synthesis.
Palmitate Induces de Novo Ceramide Synthesis in BAECs-Prior studies had implicated that palmitoyl-CoA, the activated form of palmitate, is the substrate of de novo formation of ceramide (32). We next determined if exposure of BAECs to palmitate increased de novo synthesis of ceramide. As shown in Fig. 6 (A and B), compared with the controls, palmitate instead of palmitoleate and oleate increased the formation of ceramide in BAECs.
Activation of PP2A by Ceramide in BAECs-We next determined if palmitate activated PP2A in BAECs. To directly assess PP2A activity in response to different factors, immunoprecipitation of PP2A was performed with a peptide antibody directed against the C subunit of PP2A. The immunoprecipitated PP2A activity was further assayed by using the specific phosphopeptide as its substrate, as described under "Experimental Procedures." As indicated in Fig. 7A, pretreatment with palmitate increased PP2A activity by 50% compared with the control, whereas oleate did not change PP2A activity. Direct exposure of cells with C 2 -ceramide mim-  icked the effect of palmitate and caused a 2-fold elevation in PP2A activity. In contrast, fumonisin B1, which suppresses ceramide formation, caused a 60% reduction in PP2A activity. Furthermore, fumonisin B1 abolished palmitate-induced PP2A activation. Palmitate did not change the abundance of the PP2A catalytic subunit (Fig. 7B), suggesting that palmitate activated AMPK without altering its expression.
PP2C Is Not Involved in Palmitate-induced Decrease in AMPK Phosphorylation-There is evidence that another type of phosphatase, protein phosphatase 2C (PP2C) negatively regulates AMPK in vitro and in skeletal muscle (33). To determine whether or not PP2C was involved in AMPK inhibition induced by palmitate, we next determined the effects of palmitate on PP2C expression and activity. As shown in Fig. 7 (C and D), palmitate altered neither PP2A expression nor the PP2C activity. PP2C was known to be OA-insensitive but Mg 2ϩ -sensitive (34), and no specific inhibitors for PP2C were available. High concentrations (10 mM) of EDTA were reported to inhibit PP2C by chelating Mg 2ϩ , which is required for PP2C activity. High concentrations of EDTA had no effect on palmitate-in-duced AMPK inhibition (Fig. 7, E and F). These results suggest that Mg 2ϩ -sensitive phosphatases, like PP2C, might not be involved in palmitate-induced AMPK inhibition.
Palmitate-induced Inhibition of AMPK Is Associated with PP2A Activation in Mice-We next determined the effects of palmitate on AMPK and the roles of PP2A. Palmitate but not oleate significantly inhibited AMPK and ACC phosphorylation in aorta from C57BL/6J mice (Fig. 8, A and B) in parallel with increased aortic PP2A activity (Fig. 8, E and F). The PP2A activity was inversely related to the levels of phosphorylated AMPK suggesting that PP2A played an important role in palmitateinduced AMPK inhibition. Previous studies from us and others (20,35) suggested that AMPK activation might improve vascular endothelial functions by increasing endothelial nitric-oxide synthase (eNOS) phosphorylation on Ser 1177 . Therefore, effects of palmitate or oleate on eNOS Ser 1177 phosphorylation on mouse aortas were also assayed in these mice. As indicated in Fig. 8 (A and B), palmitate significantly decreased eNOS Ser 1177 phosphorylation by 65%, and, to a less extent, oleate decreased it by 25% (p Ͻ 0.05) compared with control groups. The endogenous PP2A was immunoprecipitated for the determination of PP2A activity using a synthetic phosphopeptide substrate as described under "Experimental Procedures." The relative PP2A activities were normalized by arbitrarily setting the A 650 nm absorbance of control cells to 100% (means Ϯ S.E., n ϭ 5; *, p Ͻ 0.05; **, p Ͻ 0.01 versus control; #, p Ͻ 0.05 versus palmitate). B, palmitate increases PP2A activity but not PP2A expression. BAECs were lysed and proteins were extracted, as described under "Experimental Procedures." Proteins were separated in SDS-PAGE and detected by using the specific antibodies in Western blots. The blot is a representative of four blots obtained from four separate experiments. C and D, palmitate (0.4 mM, 40 h) altered neither the levels nor the activity of PP2C. The activity of PP2C in BAECs treated with vehicle, palmitate, or oleate was assayed using p-nitrophenyl phosphate as substrate. The initial rate of liberation of p-nitrophenol was determined spectrophotometrically at 405 nm. E, inhibition of PP2C by 10 mM EDTA could not ameliorate palmitate-induced decrease in AMPK phosphorylation. The blot is a representative of three blots obtained from three independent experiments. F, bar graph showing the levels of phosphorylated AMPK determined by quantitative Western blotting (n ϭ 3; *, p Ͻ 0.05 versus non-EDTA-treated control groups; #, p Ͻ 0.05 versus non-EDTA-treated AICAR groups).
In addition, we explored the effects of these fatty acids on another important cell type of aorta, VSMCs. Interestingly, palmitate inhibited AMPK phosphorylation (by 28%, p Ͻ 0.05) in cultured VSMC but had no effects in endothelium-denuded aortas from palmitate-fed mice (Fig. 8, C and D), suggesting that endothelium-derived AMPK might be a primary targets of HFD-induced PP2A activation.
Inhibition of PP2A with PP2A-specific siRNA Reverses Palmitate-induced AMPK Inhibition in Mice-We next determined if inhibition of PP2A reversed the inhibitory effect of palmitate on AMPK as well as eNOS phosphorylation. Normal chow or palmitate-HFD fed mice were injected retro-orbitally with NPspecific PP2A siRNA in complexes with jetPEI TM . As controls, mice were injected retro-orbitally with scrambled siRNA in complexes with jetPEI TM . Forty-eight hours after infection, PP2A protein content in aortas was measured by Western blot. Administration of PP2A-specific siRNA diminished aortic PP2A protein expression by 50%. As depicted in Fig. 9 (A and B), PP2A siRNA not only increased AMPK phosphorylation in mice fed with normal chow fed mice (ϳ2.7-fold, p Ͻ 0.05) but also abolished palmitate-induced inhibition of AMPK phosphorylation in mouse aortic tissues (Fig. 9, A and B). In parallel with the change of AMPK phosphorylation, eNOS Ser 1177 phosphorylation was decreased in response to HFD rich in palmitate, whereas inhibition of PP2A with PP2A-specific siRNA but not scrambled siRNA reversed HFD-triggered reduction of eNOS-Ser 1177 phosphorylation in mice fed with HFD (Fig. 9, A and B). These data indicated that HFD inhibited eNOS phosphorylation, likely via a PP2A-dependent inhibition on AMPK in vivo.
Palmitate Causes Insulin Resistance in Vivo-Plasma glucose, insulin, food intake, body weight, and AUC glucose during intraperitoneal glucose tolerance test measured after 3 months of normal chow, diet rich in palmitic acid, or oleic acid are shown in Table 1. No differences in food intake were observed among the three groups. Both palmitate-HFD and oleate-HFD increased body mass relative to normal chow (28 Ϯ 2 g and 22.8 Ϯ 1.5 g versus 18.9 Ϯ 0.8 g, p Ͻ 0.01). However, body weights were significantly higher in the palmitate-HFD than in the oleate-HFD group after 3 months of the diet (p Ͻ 0.05). The changes in net weight gain were similar to those in body mass among these groups. These data indicated that the mice were more prone to obesity when fed with palmitate rather than oleate. After 3 months, levels of plasma glucose and insulin FIGURE 8. Palmitate-induced inhibition of AMPK is associated with activation of PP2A in C57BL/6J mice. A, high fat diet rich in palmitate instead of oleate decreased AMPK and ACC phosphorylation. Male C57BL/6J mice were fed with normal chow, high fat diet rich in palmitate (palmitate-HFD, 16:0 ϭ 50% of total fatty acids), or oleate (oleate-HFD, 18:1 ϭ 80% of total fatty acids). The mice were sacrificed, and thoracic aortas were isolated for determination of AMPK, ACC, and eNOS phosphorylation 12 weeks after starting the diets. The blot is representative of three blots obtained from three independent experiments. B, bar graph showing the levels of phosphorylated AMPK, ACC, and eNOS determined by quantitative Western blotting (n ϭ 3; *, p Ͻ 0.05 versus control groups; #, p Ͻ 0.05 versus palmitate-HFD groups). C, palmitate inhibited AMPK phosphorylation in cultured VSMC instead of endothelium-removed aortas. The adventitia and endothelium of aortas from diet rich in palmitate or oleate fed mice were removed as described under "Experimental Procedures." Isolated aortas and fatty acids-treated VSMC were homogenized, and proteins were separated by SDS-PAGE and Western blotted with the antibodies against phosphor-AMPK or AMPK. The blot is a representative of three independent experiments (n ϭ 3). D, bar graph showing the levels of phosphorylated AMPK determined by quantitative Western blotting (n ϭ 3; *, p Ͻ 0.05 versus control groups; #, p Ͻ 0.05 versus palmitate-HFD groups). E, palmitate-HFD instead of oleate-HFD increased PP2A activity in mice. After 12 weeks of each diet, the mice were sacrificed, and thoracic aortas were isolated for detection of PP2A activity as described under "Experimental Procedures." The data are means Ϯ S.E. from three independent experiments, n ϭ 10 in each dietary group. F, PP2A protein levels remained unchanged in aortas after each diet. The blot is a representative of three blots obtained from three independent experiments.
were significantly higher in the palmitate-HFD group than in the control and oleate-HFD group (p Ͻ 0.01 and p Ͻ 0.05, respectively). At 3 months, AUC glucose was ϳ1.5-fold higher in the palmitate-HFD group than in the control group (p Ͻ 0.01), but there was no difference in AUC glucose between the oleate-HFD and control groups. In palmitic acid HFD-fed mice, a marked rise in both plasma glucose and insulin levels plus impaired glucose tolerance implied that mice fed with palmitic acid developed the phenotypes of insulin resistance. These results are consistent with the hypothesis that a decreased AMPK activity induced by palmitate contributes to excess lipid accumulation and insulin resistance in vivo.

DISCUSSION
The data presented in the study have demonstrated that prolonged exposure of endothelial cells to palmitate significantly suppresses the phosphorylation of AMPK and its downstream enzyme, ACC. In addition, we have for the first time provided compelling evidence obtained in vitro and in vivo that palmitate inhibits both AMPK and eNOSphosphorylationbyceramidedependent PP2A activation. Thus, our results might provide a novel mechanism by which palmitate inhibits AMPK in endothelial cells.
One metabolic fate of the palmitate taken up by cells is that, following its conversion into palmitoyl-CoA, it can be condensed with serine in a reaction catalyzed by serine palmitoyl transferase (36) to generate 3-ketodihydrosphingosine. This latter molecule can be reduced to sphinganine whose acylation produces dihydroceramide, which can be desaturated to yield ceramide (37). The present study suggests that palmitate, instead of palmitoleate and oleate, increase de novo synthesis of ceramide in BAECs (Fig. 6). Ceramide is a second messenger in the sphingomyelin signaling pathway and has been implicated in the regulation of diverse cellular responses, including cell death, differentiation, and in the pathogenesis of insulin resistance (38) and lipotoxicity (39). Our data suggest that ceramide may be the intermediate molecule mediating palmitate-induced PP2A activation and subsequent AMPK inhibition (Fig. 8). This notion is supported by several lines of evidence. First, incubation of BAECs with palmitate promotes intracellular accumulation of ceramide and inhibits AMPK phosphorylation (Figs. 5 and 6), and palmitoleate or oleate, which cannot produce ceramide, have no effect. Second, a cellpermeable analog of ceramide, C 2 -ceramide, was capable of mimicking the effect of palmitate on AMPK phosphorylation. Conversely, fumonisin B1, a ceramide synthase inhibitor, prevented the inhibitory effect of palmitate on AMPK phosphorylation (Fig. 5). In addition, we also found palmitate-induced PP2A activation is ceramide-dependent in cultured BAECs (Fig. 7). The latter observation provides further, albeit indirect, support for the concept that PP2A is an important component of dephosphorylating and inactivating AMPK. Third, the inhibition of PP2A, by either a specific inhibitor, OA, or by PP2A siRNA, increased AMPK phosphorylation and normalized palmitate-induced increase in AMPK Thr 172 dephosphorylation (Fig. 4), suggesting that PP2A is important for AMPK phosphorylation in vivo. Fourth, high fat diet rich in palmitate FIGURE 9. Inhibition of PP2A with selective siRNA for PP2A reverses palmitate-induced AMPK and eNOS inhibition in C57BL/6J mice. After 12 weeks of feed, normal chow, or palmitate-HFD, mice were injected retro-orbitally with NP-specific PP2A siRNA in complexes with jetPEI TM . Control mice were injected with scrambled siRNA through the same route. Forty-eight hours after injection, PP2A protein content and AMPK or eNOS phosphorylation (A) in aortas were measured by Western blot. The blot is a representative of three blots obtained from three independent experiments. B, bar graph showing the levels of phosphorylated AMPK and eNOS determined by quantitative Western blotting (n ϭ 3; *, p Ͻ 0.05 versus normal diet control groups; #, p Ͻ 0.05 versus palmitate-HFD control groups). C, proposed scheme for palmitate-induced AMPK inhibition in endothelial cells. Palmitate is converted into palmitoyl-CoA. It can be condensed with serine in a reaction by serine palmitoyl transferase to generate 3-ketodihydrosphingosine. This latter molecule can be reduced to sphinganine whose acylation produces dihydroceramide, which can be desaturated to yield ceramide. Increased formation of ceramide activates PP2A, which dephosphorylates AMPK. Decreased AMPK activation leads to metabolic and cardiovascular diseases by impairing glucose uptake and fatty acid oxidation. ACS, acyl-CoA synthetase; SPT, serine palmitoyl transferase.

TABLE 1
Plasma glucose and insulin, food intake, body mass, net weight gain, and AUC glucose during intraperitoneal glucose tolerance test of C57BL/6J mice after 12 weeks of normal chow, diet rich in palmitic acid, or oleic acid rather than oleate lead to activation of PP2A and inhibition of AMPK in C57BL/6J mice (Fig. 8). Moreover, introduction of PP2A siRNA to the mice reversed the palmitate-induced decrease in AMPK phosphorylation (Fig. 9). The important finding of the present study is that PP2A may directly modulate AMPK function. In support of this idea, it has been reported that the PP2A complex is involved in regulating the interaction between AMPK ␣2 and ␥1 (40) and inactivating of AMPK in pancreatic ␤-cells (41), and the active phosphorylated form of AMPK can be inactivated in cell-free assays by PP2A (42). Thus, we conclude that PP2A might be a physiologically relevant AMPK phosphatase. Our data suggest that palmitate inhibits AMPK phosphorylation by activating the serine/threonine phosphatase, PP2A, instead of inhibiting the AMPK kinase, LKB1 (Fig. 2), or enhancing ATP levels after its metabolism and ␤-oxidation (Fig. 3).
PP2C is an Mn 2ϩ -or Mg 2ϩ -dependent protein serine/threonine phosphatase. Its exact physiological role is still unclear. It has been shown that PP2C is responsible for dephosphorylation of AMPK (33). Wang et al. (43) have demonstrated that AMPK phosphorylation was significantly reduced in both ZDF fa/fa rats and ob/ob mouse hearts compared with lean, wild-type controls, and the reduction in active P-AMPK␣ is associated with an increase in PP2C. In the present study, our data indicate that a decrease in AMPK phosphorylation after chronic exposure of endothelial cells to high levels of palmitate is correlative with PP2A activation instead of PP2C. The discrepancy is probably due to the differences in species, organs, and nutrition types. In addition, it is likely that PP2C might negatively regulate AMPK in physiological conditions, and PP2A might be applicable for palmitate-induced AMPK inhibition, because it is ceramide-dependent, as we demonstrate in the present study.
The physiological relevance of these findings is obvious and important. On the one hand, high circulating plasma fatty acid levels are a hallmark of obesity and poorly controlled Type II diabetes. Fatty acid oversupply is also associated with fatty acid metabolite accumulation in non-adipose tissues, which causes insulin resistance (44). There is increasing acceptance of the idea that excessive exposure of non-adipose tissues to lipids in excess of their oxidative or storage capacities causes cell dysfunction or death. These FFA-induced disturbances are referred to as lipotoxicity. The lipotoxicity is the main causation of endothelial dysfunction and development of cardiovascular diseases (45). However, the concrete mechanisms mentioned above are not elucidated clearly. On the other hand, AMPK pathways play an important role in development of insulin resistance and endothelial damage. Activation of AMPK induced by exercise, hypoxia, and physiological or pharmacological inducers can improve insulin sensitivity by increasing glucose transport and oxidation, fatty acid oxidation, and subsequent decrease in lipid accumulation in non-adipose tissues. More importantly, AMPK can inhibit the mTOR-S6 kinase 1 pathway, which phosphorylates insulin receptor substrate proteins on serine residues resulting in decreasing tyrosine phosphorylation and insulin signaling (8). The exact role of AMPK in endothelial cells is not very clear, but it may have some potential benefits. The most important aspect is that activated AMPK increases fatty acid oxidation by phosphorylating and inhibiting ACC leading to a decrease in the concentration of malonyl-CoA (46). In addition, it decreases fatty acid incorporation into glycerolipids, either secondary to its effect on fatty acid oxidation or by virtue of the fact that in some tissues it phosphorylates and inhibits sn-glycerophosphate acyltransferase, the first committed enzyme in diacylglycerol and triglyceride synthesis (47). An additional benefit of endothelial AMPK activity is that it may inhibit glycerol-3-phosphate acyltransferase, required for de novo synthesis of diacylglycerol (47). In other words, AMPK may lessen endothelial diacylglycerol production (and thus protein kinase C activation) both by diminishing availability of the FFA substrate for this synthesis, and by directly inhibiting the enzyme that catalyzes it. Endothelial AMPK has the further advantage of activating nitric-oxide synthase via phosphorylation on Ser 1177 (35), thus opposing the adverse impact of protein kinase C on this enzyme. Consistent with these findings, our data indicated that inhibition of AMPK activation by palmitate blocked eNOS phosphorylation, and increasing AMPK phosphorylation by PP2A siRNA administration could increase eNOS phosphorylation (Fig. 9, A and B) suggesting that eNOS phosphorylation is associated with the status of AMPK activity. However, there are other possibilities that PP2A affects eNOS phosphorylation directly (48) or through altering protein kinase B/Akt activity (49), an important upstream kinase of eNOS phosphorylation. Therefore, it remains possible that activation of PP2A by palmitate can impair eNOS phosphorylation by a direct dephosphorylation of eNOS-Ser 1177 or an inhibition of phosphatidylinositol 3-kinase/Akt pathway. Therefore, whether or not AMPK reduction plays a causal effect in FFA-induced reduction of eNOS phosphorylation or endothelium dysfunction deserves further investigation. Taken together, our studies might reveal a novel mechanism by which long term high levels of FFA trigger a PP2A-dependent inhibition of both AMPK and eNOS, which might contribute to vascular endothelial injury and dysfunction (Fig. 9C).