Activation of AMP-Activated Protein Kinase Inhibits Oxidized LDL-Triggered Endoplasmic Reticulum Stress In Vivo

Antibodies against phospho-AMPK, phospho-acetyl-CoA carboxylase (ACC), phospho-eukaryotic translation initiation factor 2α (eIF2α), anti-endothelial nitric oxide synthase, phospho–Jun NH₂-terminal kinase (JNK), and 3-nitrotryosine were obtained from Cell Signaling Biotechnology (Danvers, MA). The antibodies against phospho-PKR (protein kinase R)-like ER kinase (PERK), X-box binding protein 1 (XBP-1), and SERCA, scrambled small interfering RNA (siRNA), and the specific siRNA for calcium/calmodulin-dependent protein kinase kinase 2 (CaMKK2), SERCA2, and p67^phox were obtained from Santa Cruz Biotechnology Inc. (Santa Cruz, CA). Antibodies against glucose-regulated protein 78 (GRP78) and oxidized LDL (ox-LDL) were from Abcam (Cambridge, MA). Antibody against activating transcription factor 6 (ATF6) was obtained from Imgenex (San Diego, CA). 3-(4-morpholinyl)sydnone imine hydrochloride (SIN-1) was from Dojindo Laboratories USA (Rockville, MD). 5′-aminoimidazole-4-carboxymide-1-β-d-ribofuranoside (AICAR) was from Toronto Research Chemicals Inc. (North York, ON, Canada). Fluo-4 NW kits were obtained from Invitrogen Inc. (Carlsbad, CA). All other chemicals, if not indicated, were purchased from Sigma-Aldrich (St. Louis, MO).

All mice were housed in temperature-controlled cages under a 12-h light-dark cycle and given free access to water and normal chow. AMPKα2^−/− mice that had been backcrossed onto a C57BL/6 background were crossed with ApoE^−/− mice (The Jackson Laboratory), also on the C57BL/6 background, to generate ApoE^−/−/AMPKα2^−/− mice. Accelerated atherosclerosis was induced by feeding the mice a Western diet containing 0.21% cholesterol and 21% fat (D12079B; Research Diets Inc.). This diet was administered beginning at 5 weeks of age and continued for 8 consecutive weeks. At 5 weeks of age, Tempol (Sigma-Aldrich) was also added to the drinking water (1 mmol/l) for 8 weeks. ApoE^−/− mice fed with normal chow were used as controls. The animal protocol was reviewed and approved by the University of Oklahoma Institutional Animal Care and Use Committee.

Native LDL (N-LDL) and HOG-LDL were prepared as described previously (27). Final preparations were dialyzed extensively against PBS buffer. Characterization of native and HOG-LDL was done by gel electrophoresis (fluorescence at 360 nm excitation/430 nm emission, and absorbance at 234 nm, as described previously [27,28]). The study protocol was approved by the Institutional Review Board at the University of Oklahoma Health Sciences Center.

Both bovine aortic endothelial cells (BAECs) and human umbilical vein endothelial cells (HUVECs) were cultured in endothelial basal medium with 2% FBS. After confluence, cells were incubated with HOG-LDL at the concentrations and times indicated. Cells exposed to the same concentrations of N-LDL were used as controls. If needed, pharmacological reagents, including 1,2-B is BAPTA-AM (10 μmol/l), AICAR (1 mmol/l), or Tempol (10 μmol/l), were added 30 min prior to the addition of N-LDL or HOG-LDL.

Confluent HUVECs were transfected with control or gene-specific siRNA according to manufacturer's recommendations. To improve suppression of the target gene, two rounds of consecutive siRNA gene silencing were performed. The efficiency of siRNA-silenced genes was evaluated by Western blotting of the targeted protein with specific antibodies.

Adenoviral vector AMPK-CA (a constitutively active AMPK mutant) was generated as previously reported (25). Adenoviral vector AMPK-DN (a dominant-negative mutant of AMPK) was generated by subcloning the cDNA encoding AMPKα1-DN-(D159A) into adenoviral vector pAdEasy-1.

Confluent cells were infected with adenoviruses at multiplicity of infection of 50 for 36 h, as described previously (25). Adenovirus encoding green fluorescent protein (GFP) was used as control. Under this condition, the expression efficiency was >80% as seen by GFP expression.

Cellular cytosolic and membrane fractions were prepared as described previously (25). Translocation of p47^phox subunit of NAD(P)H oxidase was determined in Western blot.

The endothelium-dependent (assays of bioactive nitric oxide [NO]) and -independent (assays of vascular smooth muscle responsiveness to nitric oxide donor sodium nitroprusside) vasorelaxation was measured as described previously (27).

Intracellular calcium concentration ([Ca²⁺]_i) was measured using a Fluo-4 NW kits in accordance with manufacturer's recommendations with the fluorophore excited at 485 nm and detection at 520 nm (27). [Ca²⁺]_i was expressed as a percentage of control. SERCA activity was measured as described previously (29). Biotinylated-iodoacetamide (b-IAM) labeling of SERCA cysteine 674 was performed according to the methods described by Tong et al. (30). Proteins were separated by SDS-PAGE, and b-IAM labeled Cys674-SERCA was detected using SERCA-specific antibody. Ratiometric Ca²⁺ measurement, which can assay SERCA function dynamically, was conducted as described previously (31).

Mice underwent a 14- to 15-h fast before blood samples were collected. Blood glucose levels were determined by applying tail blood to an OneTouch Ultra Blood Glucose Monitoring System (LifeScan). Serum cholesterol and triglyceride levels were measured enzymatically, using Infinity reagents from Thermo DMA according to the manufacturer's instructions.

After being fed the Western diet for 8 weeks, mice were fasted for 14 h and then were anesthetized and killed. The heart and aortic tissue were removed from the ascending aorta to the ileal bifurcation and placed in 4% paraformaldehyde for 16 h. For analyzing the lesion area in the aortic root, the heart was dissected from the aorta, embedded in Polyfreeze tissue freezing medium (Polysciences, Inc.), and sectioned (5-μm thickness). Four sections representing every tenth serial section over a distance of 200 μm were collected from each mouse and stained with Oil Red O for neutral lipids, and counterstained with hematoxylin to visualize the nuclei. Plaques were imaged with an Olympus microscope connected to a QImaging Retiga CCD camera. The aortic lesion size of each animal was obtained by the averaging of lesion areas in four sections from the same mouse. For analyzing the lesion area in the aortic arch, the intimal surface was exposed by a longitudinal cut from the ascending arch to 5 mm distal of the left subclavian artery to allow the lumen of the aortic arch to be laid flat. The aorta was rinsed for 5 min in 75% ethanol, stained with 0.5% Sudan IV in 35% ethanol and 50% acetone for 15 min, destained in 75% ethanol for 5 min, and then rinsed with distilled water. Digital images of the aorta were captured under a stereomicroscope, and the lesion area was quantified from the aortic arch to 5 mm distal of the left subclavian artery using Alpha Ease FC software (Version 4.0; Alpha Innotech).

Immunocytochemistry stainings for oxidized LDL and ER stress markers were performed as described previously (32).

Results were analyzed with a one-way ANOVA with appropriate post hoc analysis of results as well as Student t test. Values were expressed as mean ± SD. P < 0.05 was considered statistically significant.

We had previously reported (27) that exposure of BAECs to HOG-LDL impairs endothelial function by inducing calpain-mediated endothelial NO synthase degradation in a reactive oxygen species (ROS)– and Ca²⁺-dependent manner. Whether HOG-LDL alters ER stress was not known. To determine the effect of HOG-LDL on ER stress, confluent BAECs were exposed to HOG-LDL at a concentration of 100 μg/ml, which is considered to be pathologically relevant to type 2 diabetes (28), for 30 min to 6 h. BAECs exposed to N-LDL (100 μg/ml) were used as controls. As depicted in Fig. 1,A, HOG-LDL but not N-LDL, markedly increased the detection of p-PERK and p-eIF2α, two established ER stress markers, as early as 30 min after HOG-LDL exposure. Concomitantly, HOG-LDL also increased the expression of GRP78 and the cleavage of ATF6 (Fig. 1 A).

The effects of HOG-LDL appeared to be dose dependent. Low concentration of HOG-LDL (50 μg/ml) markedly increased the detection of p-PERK, p-eIF2α, and p-JNK in BAECs at 6 h (Fig. 1 B and C). ER stress markers (p-PERK, p-eIF2α, and p-JNK) increased as the doses of LDL increased (25–200 μg/ml) within 6-h exposure.

Because exposure of BAECs to HOG-LDL elevates intracellular calcium [Ca²⁺]_i in BAECs (27), we next determined whether the rise of [Ca²⁺]_i contributed to increased detection of ER stress. We first determined whether calcium mobilizing reagents triggered ER stress in BAECs. As expected, exposure of BAECs to calcimycin, a well-known Ca²⁺ mobilizer, markedly increased the detection of ER stress markers (Fig. 1 D). Consistently, tunicamycin (16), meflouquine, or SIN-1, all of which are known to increase intracellular [Ca²⁺]_i, also increased ER stress in BAECs (supplementary Fig. 1, available in an online appendix at http://diabetes.diabetesjournals.org/content/early/2010/03/10/db09-1637/suppl/DC1).

It was important to determine whether chelating [Ca²⁺]_i reversed the effects of HOG-LDL on ER stress. Pretreatment of BAECs with BAPTA-AM (10 μmol/l) significantly inhibited HOG-LDL–induced ER stress (Fig. 1 D and supplementary Fig. 1C), implying that a rise in [Ca²⁺]_i was critical for HOG-LDL–enhanced ER stress.

As peroxynitrite (ONOO⁻) is known to increase both intracellular calcium and ER stress (33), we next determined whether ONOO⁻ mediated the effects of HOG-LDL. 3-nitrotyrosine (3-NT)–positive proteins are considered a footprint of ONOO⁻ in cultured cells. We first assayed whether HOG-LDL increased the levels of 3-NT. As depicted in Fig. 2 A, exposure of BAECs to HOG-LDL, but not N-LDL, dose-dependently increased the levels of 3-NT–positive proteins, implying that HOG-LDL increased ONOO⁻ in endothelial cells.

ONOO⁻ is a potent oxidant formed by the combination of superoxide and NO. Our previously published data (27) demonstrate that HOG-LDL enhances both superoxide and ONOO⁻ upon its reaction with NO by activating NAD(P)H oxidase. Therefore, it was interesting to determine whether inhibition of NAD(P)H oxidase or scavenging superoxide altered HOG-LDL–triggered ER stress. To this end, BAECs were pretreated with apocynin, a potent inhibitor for NAD(P)H oxidase, or Tempol, a superoxide scavenger, prior to the addition of HOG-LDL. As shown in Fig. 2 B, both apocynin and Tempol significantly attenuated HOG-LDL–induced ER stress.

Next, we determined whether genetic inhibition of NAD(P)H oxidase attenuated ER stress caused by HOG-LDL. As depicted in Fig. 2,C, transfection of p67^phox-specific siRNA significantly reduced the levels of p67^phox, an essential component of NAD(P)H oxidases. Further, transfection of p67^phox-specific siRNA, but not control siRNA, abrogated HOG-LDL–induced ER stress (Fig. 2 C and D). Taken together, our results suggest that NAD(P)H oxidase–derived ROSs were required for HOG-LDL–triggered ER stress.

Because SIN-1, an ONOO⁻ donor, increased both intracellular Ca²⁺ and ER stress (supplementary Fig. 1B and C), we next investigated how ONOO⁻ increased intracellular Ca²⁺ in BAECs. Because SERCA is a key enzyme that controls Ca²⁺ transfer from the cytosol to the ER lumen and its inhibition is known to cause aberrant ER stress in various cell types (34), we hypothesized that ONOO⁻ generated during HOG-LDL exposure might alter SERCA function resulting in increased levels of [Ca²⁺]_i. Indeed, exposure of BAECs to thapsigargin, a specific SERCA inhibitor, triggered ER stress in both a time- and concentration-dependent manner in BAECs (supplementary Fig. 2A and B).

To avoid potential off-target effects of thapsigargin, we next determined whether genetic inhibition of SERCA2 altered ER stress in endothelial cells. Because bovine-specific SERCA siRNA is not available, we performed siRNA experiments in HUVECs, in which HOG-LDL caused a Tempol-sensitive aberrant ER stress and SERCA inhibition (supplementary Fig. 3). As expected, gene silencing of SERCA2 markedly increased HOG-LDL–induced ER stress in endothelial cells (supplementary Fig. 2C and D), suggesting SERCA inhibition triggered ER stress in endothelial cells.

To explore how HOG-LDL exposure could trigger ER stress in endothelial cells, we first determined whether HOG-LDL suppressed SERCA activity. As depicted in Fig. 3,A, HOG-LDL significantly inhibited SERCA activity in BAECs. However, HOG-LDL had no effect on the expression of SERCA(s) in endothelial cells (Fig. 3 B), suggesting that HOG-LDL–induced suppression of SERCA activity was independent of transcriptional and translational regulation of SERCA.

Because Cys674 of SERCA is reported to be oxidized and suppressed by ROSs (30), we next determined whether HOG-LDL increased the SERCA oxidation at Cys674. We first determined whether inhibition of NAD(P)H oxidase with apocynin or scavenging superoxide with Tempol, both of which suppressed HOG-LDL–induced ER stress (Fig. 2,B), also alleviated the effects of HOG-LDL on SERCA activity and SERCA oxidation on Cys674. As depicted in Fig. 3,A, Tempol restored SERCA activity in HOG-LDL–treated cells. Similarly, apocynin ablated the reduction in SERCA activity induced by HOG-LDL (Fig. 3,C). Consistently, HOG-LDL significantly increased oxidation of Cys674 (Fig. 3,D). Furthermore, Tempol prevented HOG-LDL–enhanced SERCA oxidation (Fig. 3 D). Cumulatively, these findings suggest that HOG-LDL attenuates SERCA activity via ROS-mediated oxidation of the Cys674 residue of these ATPases.

We had reported that AMPK activation by AICAR reduces oxidative stress in diabetes (35). It was therefore interesting to investigate whether AMPK activation alleviated SERCA oxidation and consequent ER stress caused by HOG-LDL. Short exposure (<3 h) of BAECs to HOG-LDL (100 μg/ml) markedly increased the detection of AMPK phosphorylation at Thr172 (Fig. 4,A), an index for AMPK activation. AICAR significantly enhanced AMPK Thr172 phosphorylation caused by HOG-LDL alone (Fig. 4 A).

We next determined whether the reduction of ER stress was via a reduction of [Ca²⁺]_i. As depicted in Fig. 4,B, AICAR blocked HOG-LDL–induced [Ca²⁺]_i rise and reduced the markers of ER stress (Fig. 4 C).

As apocynin ablated HOG-LDL–induced ER stress and SERCA oxidation, it was logical to determine whether AMPK activation by AICAR lowered HOG-LDL–induced ER stress by suppressing the activation of NAD(P)H oxidase. As depicted in Fig. 4,D, HOG-LDL exposure increased the membrane translocation of p47^phox, a key step for NAD(P)H oxidase assembly and activation. Importantly, AICAR suppressed the effects of HOG-LDL on p47^phox translocation (Fig. 4 D).

We next assayed whether AICAR attenuated SERCA inhibition caused by HOG-LDL. AMPK activation by AICAR inhibited HOG-LDL–induced reduction in SERCA activity in BAECs (Fig. 4,E). To further confirm the effect of AICAR on SERCA function, [Ca²⁺]_i was monitored using ratiometric measurements dynamically (31). As shown in Fig. 4,F, compared with N-LDL–treated cells, HOG-LDL–treated BAECs exhibited impaired ability to transfer [Ca²⁺]_i to the ER lumen, which is in line with an earlier report (28). Importantly, AICAR pretreatment abolished the effects of HOG-LDL (Fig. 4 F). These data further support that AICAR alleviates the reduction of SERCA function caused by HOG-LDL.

To exclude potential off-target effects of AICAR, we tested whether genetic activation or inhibition of AMPK modulates HOG-LDL–induced ER stress. Overexpression of AMPK-CA, a constitutively active AMPK mutant, suppressed HOG-LDL– and calcimycin-induced ER stress (Fig. 5,A and B). Conversely, AMPK inhibition by overexpression of AMPK-DN, an AMPK dominant negative mutant, significantly accentuated ER stress induced by either HOG-LDL or calcimycin (Fig. 5,C and D). Concomitantly, overexpression of AMPK-CA maintained SERCA activity when challenged with HOG-LDL, whereas AMPK-DN failed to maintain SERCA activity (Fig. 5 E). Taken together, our results suggest that AMPK activation suppresses ER stress caused by HOG-LDL through maintaining SERCA activity.

To further test the role of AMPK in response to HOG-LDL–induced ER stress and consequent vascular dysfunction, we assayed the effects of metformin and statins, two clinically used AMPK activators, in isolated aortas ex vivo. Metformin or simvastatin was added 30 min prior to HOG-LDL exposure. As shown in Fig. 6,A, both metformin and statin treatment activated AMPK when aortic rings were incubated with or without HOG-LDL. HOG-LDL induced ER stress, as indicated by stress markers p-PERK and p-eIF2α (Fig. 6,B). Pretreatment of isolated aortae with metformin or statins ablated the effects of HOG-LDL on ER stress (Fig. 6 B and C).

Next, we assessed the effects of AMPK activation on acetylcholine (ACh)–induced endothelium-dependent relaxation under ex vivo conditions. As indicated in Fig. 6 D, HOG-LDL impaired ACh-induced relaxation. Pretreatment with metformin or simvastatin, which did not alter basal ACh-induced relaxation, reversed the reduction of Ach-induced vasorelaxation caused by HOG-LDL. In addition, the endothelium-independent vasorelaxation, as assayed by monitoring vasorelaxation to sodium nitroprusside, a NO donor, was unchanged among these groups (data not shown), suggesting unchanged function of vascular smooth muscle cells in terms of NO. Overall, these results suggest that AMPK activation is sufficient to preserve endothelium-dependent relaxation impaired by HOG-LDL.

It was important to determine whether AMPK is essential in maintaining SERCA activity and inhibiting ER stress in vivo. Compared with ApoE^−/− mice fed normal diet, ApoE^−/− mice fed high-fat diet (HFD) had a reduced SERCA activity and exhibited greater degree of ER stress in the aorta (Fig. 7,A–C). In addition, AMPKα2 depletion caused further reduction of SERCA activity with higher degree of ER stress in the aorta than ApoE counterparts (Fig. 7 A–C).

Next, we determined whether chronic administration of Tempol, a potent antioxidant, prevented SERCA oxidation and attenuated ER stress in vivo. As shown in Fig. 7,C, Tempol treatment protected SERCA activity from reduction, suggesting that oxidation played a critical role in the reduction of SERCA activity. Importantly, Tempol markedly suppressed ER stress in both ApoE^−/− and ApoE^−/−/AMPKα2 ^−/− mice in vivo (Fig. 7 A and B). Taken together, our data imply that inhibition of SERCA oxidation is able to suppress ER stress in vivo.

Compared with C57BL6 mice, the endothelium-dependent vasorelaxation in response to acetylcholine was normal in mice fed normal mouse chow (data not shown). The endothelium-dependent relaxation in ApoE^−/− mice fed HFD was significantly less than those of ApoE^−/− mice fed normal mouse chow (data not shown). In addition, the endothelium-dependent relaxation in HFD-fed ApoE^−/−/AMPKα2^−/− mice was further reduced compared with ApoE^−/− mice with HFD (Fig. 7,D). Importantly, Tempol administration markedly attenuated impaired endothelium-dependent vasorelaxation in ApoE^−/−/AMPKα2^−/− mice (Fig. 7 D) but had no effect on the endothelium-independent vasorelaxation (data not shown).

Next, we evaluated whether Tempol altered the levels of oxidized LDL, ER stress, and aortic lesions in ApoE^−/− mice. Compared with normal diet, HFD significantly increased ox-LDL, ER stress, and aortic lesion in ApoE^−/− mice (Fig. 7,E and F). Further, Tempol reduced the levels of oxidized LDL, ER stress markers, and aortic lesion areas (Fig. 7 E and F).

Next, we compared metabolic parameters among ApoE^−/− and ApoE^−/−/AMPKα2^−/− mice. No differences in blood glucose, serum cholesterol, and triglyceride were observed between ApoE^−/−/AMPKα2^−/− and ApoE^−/− mice (Table 1). Tempol treatment did not affect blood glucose and triglyceride levels in both groups but increased cholesterol levels in ApoE^−/−/AMPKα2^−/− mice (Table 1).

In this study, we have demonstrated that LDL oxidation and glycation cause aberrant ER stress via the oxidation of SERCA in endothelial cells. We have also provided evidence that AMPK activation is effective in inhibiting HOG-LDL–enhanced SERCA oxidation and consequent ER stress in vivo. Finally, we found chronic administration of Tempol restored SERCA activity, ER stress, endothelium function, and atherosclerosis caused by HFD in vivo.

One of the most important findings in the present study is that HOG-LDL increases ER stress by ROS-dependent oxidation and inhibition of SERCA activity, a key enzyme that controls intracellular [Ca²⁺] (34,36,,–39). We have further found HOG-LDL increases the oxidation of Cys674, a known site for SERCA glutathiolation and oxidation, likely via NAD(P)H oxidase–derived ROSs. Our observations are in line with several published studies that have demonstrated that SERCA is prone to oxidation at one of its thiol groups under certain conditions (30,40,41) and that its oxidation is significantly increased in atherosclerotic lesions (41). Importantly, we found that chronic administration of Tempol significantly inhibited ER stress, endothelial dysfunction, and aortic lesions in vivo. Our results are also consistent with a recent report in which antioxidant metallothionein is reported to inhibit ER stress caused by diabetes and angiotensin II (42). Thus, ROS might be a common ER stress inducer in different cell types including endothelial cells, vascular smooth muscle cells, and myocytes.

Another important finding of the present study is that AMPK activation suppresses ER stress by inhibiting NAD(P)H oxidase–derived ROSs, a well-known ER stress initiator that has been implicated in the progression of atherosclerosis in many studies (43,,–46). Our unpublished data suggest that AMPKα2 deletion increases the expression of NAD(P)H oxidase subunits and NAD(P)H oxidase–derived ROSs in endothelial cells. How AMPK suppresses NAD(P)H oxidase remains undefined but warrants further investigation.

Circulating levels of oxidized LDL have been proposed to be a predictor of secondary cardiovascular events (47,48). Plasma levels of oxidized LDL correlate with endothelial dysfunction and are reduced after lipid-lowering therapy using apheresis or statins (49). The concentration of oxidized and oxidized glycated LDL in the circulation in healthy individuals is thought to be low (1–5 μg protein/ml). Because conventional lipid profiles are often abnormal in type 2 diabetes including increased triglycerides, normal or elevated levels of LDL, and reduced HDL, it is expected that glycation and oxidation of LDL are enhanced by diabetes. Further, the levels of HOG-LDL are likely to be much higher in the extravascular environment such as aortic lesions, when LDLs are retained in a less antioxidant-rich milieu and exposed to pro-oxidant stresses. For example, Nishi et al. demonstrated that in the general population, levels of oxidized LDL are ∼70-fold higher in arterial lesions than in the circulation (50). For these reasons, we consider the concentrations of LDL proposed in this work to be a reasonable estimate of those present in vivo. Thus, our findings might be important for understanding the etiology of atherosclerosis.

In summary, the present study has demonstrated that activation of AMPK suppresses HOG-LDL–induced ER stress by inhibiting NAD(P)H oxidase–derived ROS and SERCA oxidation. Therefore, AMPK may be a physiological regulator that maintains ER homeostasis and endothelial function. The proposed mechanism is outlined in supplementary Fig. 4.

This study was supported by National Institutes of Health (NIH) grants (HL079584, HL080499, and HL096032), the Juvenile Diabetes Research Foundation, and the American Diabetes Association to M.-H.Z., and by grants from the American Diabetes Association (1-05-RA-74) and the Oklahoma Center for the Advancement of Science and Technology (HR08-67; NIH P20RR024215) to T.J.L. M.-H.Z. is a recipient of the National Established Investigator Award from the American Heart Association.

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

We thank Drs. Najeeb Shirwany for his assistance in article preparation and Leonidas Tsiokas for technical support in ratiometric analysis of intracellular calcium.