Overexpression of long noncoding RNA ANRIL inhibits phenotypic switching of vascular smooth muscle cells to prevent atherosclerotic plaque development in vivo

Background: Phenotypic switching of vascular smooth muscle cells (VSMCs) plays a key role in atherosclerosis. Long noncoding RNA ANRIL (lncRNA-ANRIL) is critical in vascular homeostasis. Metformin produces multiple beneficial effects in atherosclerosis. However, the underlying mechanisms need to be elucidated. Methods and Results: Metformin increased lncRNA-ANRIL expression and AMPK activity in cultured VSMCs, and inhibited the phenotypic switching of VSMCs to the synthetic phenotype induced by platelet-derived growth factor (PDGF). Overexpression of lncRNA-ANRIL inhibited phenotypic switching and reversed the reduction of AMPK activity in PDGF-treated VSMCs. While, gene knockdown of lncRNA-ANRIL by adenovirus or silence of AMPKγ through siRNA abolished AMPK activation induced by metformin in VSMCs. RNA-immunoprecipitation analysis indicated that the affinity of lncRNA-ANRIL to AMPKγ subunit was increased by metformin. In vivo, administration of metformin increased the levels of lncRNA-ANRIL, suppressed VSMC phenotypic switching, and prevented the development of atherosclerotic plaque in Apoe mice fed with western diet. These protective effects of metformin were abolished by infecting Apoe mice with adenovirus expressing lncRNA-ANRIL shRNA. The levels of AMPK phosphorylation, AMPK activity, and lncRNA-ANRIL expression were decreased in human atherosclerotic lesions. Conclusion: Metformin activates AMPK to suppress the formation of atherosclerotic plaque through upregulation of lncRNA-ANRIL.


INTRODUCTION
It is known that atherosclerosis is a chronic inflammation in arterial wall [1,2]. Vascular smooth muscle cells (VSMCs) can change the phenotype from contractile to synthetic as a response to multiple stimuli in atherosclerosis [3,4]. Further characterization and better understanding of the underlying signaling could provide a therapeutic target in the prevention of atherosclerosis.
Based on the aforementioned studies, we speculated that, in VSMCs, lncRNA-ANRIL is a regulator of AMPK to inhibit phenotypic switching. The present study was aimed to establish the link between lncRNA-ANRIL and AMPK, and to test whether metformin prevents atherosclerosis through lncRNA-ANRIL/AMPK signaling.

PDGF reduces lncRNA-ANRIL level and increases phenotypic switching in VSMCs
VSMC phenotypic switching is vital in the formation of atherosclerotic plaque [22]. Platelet-derived growth factor (PDGF) can promote VSMCs switching from the contractile phenotype to synthetic phenotype, which contributes to atherosclerosis [23]. We determined the effects of PDGF on lncRNA-ANRIL DQ485454 transcript expression, which the expression level of the DQ485454 transcript was >6-fold higher in atherosclerosis plaque than the levels of full-length NR_003529 transcript or EU741058 transcript as reported by Hyosuk Cho et al [13]. Phenotypic switching was detected by using IFC staining of α-SMA. LncRNA-ANRIL expression was assayed by real-time PCR. As indicated in Figure 1A, 1B, PDGF decreased α-SMA, KLF4, and myocardin protein levels, compared to vehicle-treated cells. The mRNA levels of contractile phenotypic markers, such as calponin and smoothelin, were also reduced in PDGF-treated cells ( Figure 1C). Gene expressions of synthetic phenotypic markers, including osteopontin, collagen I, and collagen III, were upregulated by PDGF. Besides, the mRNA levels of lncRNA-ANRIL were reduced by PDGF in VSMCs ( Figure 1D). These data indicate that PDGF induced phenotypic switching of VSMCs, consistent with other reports [24,25].

Metformin upregulates lncRNA-ANRIL expression and inhibits PDGF-induced phenotypic switching in VSMCs
Metformin has been reported to activate AMPK and is prescribed worldwide as anti-diabetic drug to manage type 2 diabetes [26]. Epidemiological evidence has demonstrated metformin is more likely to reduce cardiovascular event risk [27]. Thus, we determined the effects of metformin on PDGF-induced phenotypic switching in VSMCs. As shown in Figure 1B, 1E, metformin at 1 mM activated AMPK by increasing AMPK phosphorylation at Thr172, which is essential for AMPK activity [28]. As expected, metformin treatment abolished PDGF-induced the reductions of α-SMA, KLF4, and myocardin ( Figure 1B). Metformin increased other contractile phenotypic markers (calponin and smoothelin) but decreased synthetic phenotypic markers (osteopontin, collagen I, and collagen III) in PDGF-treated VSMCs ( Figure 1C). These data demonstrate that metformin is able to inhibit phenotypic switching of VSMCs. Further, metformin abrogated PDGF-induced reduction of lncRNA-ANRIL expression in VSMCs ( Figure 1D).

Gain-function of lncRNA-ANRIL increases AMPK activity in PDGF-treated VSMCs
To test whether AMPK is involved in PDGF-induced VSMC phenotypic switching via lncRNA-ANRIL dysfunction, we next examined AMPK phosphorylation and activity in lncRNA-ANRIL-overexpressed cells. As presented in Figure 2H, 2I, PDGF significantly inactivated AMPK by decreasing AMPK phosphorylation and its activity in cells infected with adenovirus vector alone, but had no effects on AMPK activity and phosphorylation if cells were overexpressed with lncRNA-ANRIL.

Knockdown of lncRNA-ANRIL ablates pharmacological activations of AMPK in cultured VSMCs
To further confirm whether lncRNA-ANRIL is an upstream modulator of AMPK in VSMCs, cells were infected with shRNA to downregulate lncRNA-ANRIL expression and then treated with metformin. Adenovirus-mediated shRNA expression inhibited lncRNA-ANRIL expression ( Figure 3A). As indicated in Figure 3B, 3C, metformin increased both AMPK activity and AMPK phosphorylation in VSMCs infected with adenovirus vector alone. However, metformin failed to activate AMPK if lncRNA-ANRIL was deficient in cells, demonstrating that metformin via lncRNA-ANRIL activates AMPK.

Inhibition of lncRNA-ANRIL reduces the sensitivity of AMPK to AMP in cells
AICAR is an adenosine analog to generate AMPmimetic action to activate AMPK [16]. We next infected cells with adenovirus containing lncRNA-ANRIL shRNA and then treated with metformin or AICAR, which did not affect the effects of lncRNA-ANRIL shRNA ( Figure 3A). As illustrated in Figure  3B, 3C, AICAR or metformin alone increased AMPK phosphorylation at Thr172 and AMPK activity in VSMCs infected with adenovirus vector alone. This is consistent with previous reports [29,30].

Antisense oligonucleotide targeting lncRNA-ANRIL inhibits the effects of metformin on AMPK activation
To further implicate whether metformin via upregulating lncRNA-ANRIL activates AMPK, we used antisense oligonucleotide of lncRNA-ANRIL to inhibit the function of lncRNA-ANRIL. As shown in Figure 3D, antisense oligonucleotide of lncRNA-ANRIL reversed the effects on metformin in AMPK phosphorylation. We also observed that metformin increased AMPK phosphorylation in cells with overexpressed lncRNA-ANRIL ( Figure  3E), demonstrating that metformin activates AMPK through upregulating lncRNA-ANRIL. The roles of lncRNA-ANRIL in PDGF-induced VSMC phenotypic switching were further examined by detecting the combination of PDGF and the lncRNA-ANRIL shRNA treatment on smooth muscle markers. As shown in Figure 3F-3I, PDGF not only decreased AMPK phosphorylation, but also altered the expression of α-SMA, calponin, and osteopontin in cells with overexpressed lncRNA-ANRIL shRNA.

Metformin increases the binding of AMPKγ with lncRNA-ANRIL in VSMCs
LncRNAs are associated to a plethora of cellular functions, most of which require the interaction with one or more RNA-binding proteins [31]. We next performed RIP analysis to determine the binding between AMPKγ and lncRNA-ANRIL. In Figure 4A, lncRNA-ANRIL, but not lncRNA-MALAT1, was positively detected in samples from cells following RIP with AMPKγ primary antibody, implying that lncRNA-ANRIL is specific to AMPKγ.

LncRNA-ANRIL increases the formation of AMPKαβγ complex
To address what happens to AMPKγ subunit when it's bound to lncRNA-ANRIL, we detected the binding among AMPKα subunit, β subunit, and γ subunit. As shown in Figure 4B, the interaction among AMPKα, β and γ were much more solid in cells with lncRNA-ANRIL overexpression, compared with adenovirusinfected cells, indicating that lncRNA-ANRIL may increase the formation of AMPKαβγ complex.

Metformin-induced AMPK activation is AMPKγdependent in cells
To further verify this concept that metformin activates AMPK through increasing the interaction between lncRNA-ANRIL and AMPKγ, we downregulated AMPKγ subunit via siRNA. As shown in Supplementary Figure 1A, 1B, metformin increased both AMPK phosphorylation and AMPK activity in VSMCs transfected with scramble siRNA, but not in cells transfected with AMPKγ siRNA. Collectively, it indicates that AMPKγ is essential for metformin-induced AMPK activation.

Knockdown of lncRNA-ANRIL ablates the suppressive effects of AMPK activation by metformin on atherosclerotic plaque growth in Apoe -/mice
We further investigated if lncRNA-ANRIL is involved in the growth of atherosclerotic plaque suppressed by metformin in vivo. Thus, we infected Apoe -/mice with adenovirus to knockdown lncRNA-ANRIL followed by feeding mice with western diet ( Figure 5A). As indicated in Figure 5B, 5D, the size of atherosclerotic plaque was reduced in metformin-treated Apoe -/mice expressing negative control shRNA, compared with vehicle-treated Apoe -/mice expressing scramble shRNA. However, metformin did not decrease atherosclerotic plaque size in Apoe -/mice infected with adenovirus containing lncRNA-ANRIL shRNA. According, metformin increased AMPK activity in atherosclerotic plaque in Apoe -/mice expressing scramble shRNA, but not in Apoe -/mice expressing lncRNA-ANRIL shRNA ( Figure 5C). While, the plasms levels of TC, TG, glucose, and plasma cytokine levels were comparable among four groups (Supplementary Tables 2, 3). In sum, it indicates that lncRNA-ANRIL is involved in the effects of metformin in AMPK activation and atherosclerotic plaque growth in vivo. We next determined whether knockdown of lncRNA-ANRIL could induce VSMC phenotypic switching in atherosclerosis. As shown in Figure 6A, 6B, administration of metformin increased α-SMA content in atherosclerotic plaque in Apoe -/mice fed with western diet infected with adenovirus expressing scramble shRNA. Meanwhile, metformin increased the mRNA levels of other contractile phenotypic markers (calponin and smoothelin) in these Apoe -/mice. Conversely, the expressions of synthetic markers (osteopontin, collagen I, and collagen III) were markedly decreased by metformin in the plaque of Apoe -/mice expressing scramble shRNA. However, metformin treatment had no any effects on the expression of markers including α-SMA, calponin, smoothelin, osteopontin, collagen I, and collagen III in the plaque of Apoe -/mice expressing lncRNA-ANRIL shRNA. Taken together, these results suggest that metformin via lncRNA-ANRIL upregulation suppresses VSMC phenotypic switching in atherosclerosis.

Reduced AMPK phosphorylation, AMPK activity, and lncRNA-ANRIL expression are associated with atherosclerotic lesions in human patients
To establish the clinical association of this study, we determined the level of AMPK activity and AMPK phosphorylation in patients with atherosclerosis. The clinical characters were presented in Supplementary  Table 4. As shown in Supplementary Figure 2A, the plaques were seen in atherosclerotic samples (stenosis > 50%). The atherosclerotic tissues exhibited lower levels of AMPK phosphorylation and activity than control samples (Supplementary Figures 2B, 2C). Further, the expression of lncRNA-ANRIL was also decreased in patient with atherosclerosis (Supplementary Figure 2D). www.aging-us.com 8 AGING

DISCUSSION
In this study, we not only identified lncRNA-ANRIL as an AMPK regulator, but also demonstrated that metformin via lncRNA-ANRIL-dependent AMPK pathway prevents VSMC phenotype switches in the development of atherosclerotic plaque. In cultured cells, metformin upregulates lncRNA-ANRIL expression to increase the interaction between lncRNA-ANRIL and AMPKγ subunit. In mice, AMPK activation by metformin suppressed the growth of atherosclerotic lesion. Importantly, all effects of AMPK activation by metformin were abolished by knockdown of lncRNA-ANRIL via genetic approaches. Thus, we thought that lncRNA-ANRIL upregulation is essential for metformin-suppressed atherosclerosis.
The major discovery of this study is that we identified lncRNA-ANRIL as a novel modulator of AMPK. This is supported by following evidence: (1) The specific binding of AMPKγ to lncRNA-ANRIL was detectable using RIP analysis; (2) LncRNA-ANRIL knockdown decreased metformin-induced AMPK phosphorylation and AMPK activation; (3) The effects of metformin, as a well-recognized AMPK activator, were abolished by lncRNA-ANRIL deficiency in vitro and in vivo. Although at least two upstream kinases including Calmodulin-dependent kinase kinase (CaMKK) and liver kinase B1 (LKB1) have been identified as AMPK kinases [32,33], we firstly showed that lncRNA-ANRIL regulates AMPK activity by direct interaction. Linking many functions of AMPK, for example, glucose and lipid metabolisms [34], the finding that lncRNA-ANRIL is an AMPK regulator may expand the applications of lncRNA-ANRIL.
Another important finding is that metformin via lncRNA-ANRIL inhibits VSMC phenotypic switching and the growth of atherosclerotic plaque. Hyosuk Cho et al have identified and validated ten target genes downstream of ANRIL [13]. In this study, we not only identified AMPK as a new target of lncRNA-ANRIL, but also demonstrated that lncRNA-ANRIL/AMPK pathway is critical in the progression of atherosclerosis. Our observations demonstrate that dysregulation of lncRNA-ANRIL in VSMCs affects critical cellular functions by alternating the expressional levels of some key genes of VSMC phenotypic switching.
An issue remained in this study is the signaling pathway by how lncRNA-ANRIL increases APMK activity. Based on our observations, we thought this binding between lncRNA-ANRIL and AMPKγ may increase the stability of AMPKαβγ complex. This wound increase the sensitivity of AMP to AMPK complex and preserve the phosphorylation of AMPKα at Thr172 because we found the formation of AMPKαβγ was increased by lncRNA-ANRIL overexpression and the sensitivity of AICAR to AMPK was also enhanced.
In summary, we identify a novel mechanism by which metformin via lncRNA-ANRIL pathway limits VSMC phenotypic switching, resulting in the suppression of atherosclerosis lesion (Supplementary Figure 3). Because loss of lncRNA-ANRIL can effectively attenuate the effects of AMPK activation, both AMPK and lncRNA-ANRIL might be a therapeutic target on the prevention of atherosclerosis-associated vascular diseases, such as stroke.

MATERIALS AND METHODS
Expanded Materials and Methods are available in Supplementary Material.

Cell culture
As described previously [29], primary murine VSMCs were grown in Smooth Muscle Cell Medium (Sciencell, USA). Cells between passages 4 and 8 were used.

Western blot analysis
Total 20 μg protein were subjected to western blot analysis as described previously [35].

RNA-immunoprecipitation (IP) assay
As described previously [36], lysates were incubated in cold room overnight with magnetic protein A/G beads pretreated with antibody. RNA was extracted from IP complex. The PCR primers for lncRNA-ANRIL are shown in Supplementary Table 1.

Animal experimental protocol
Male Apoe -/mice were obtained from Beijing Huafukang Animal Experimental Center. Mice received tail vein injection of adenovirus. Two weeks after injection, mice were fed with western diet plus metformin (150 mg/kg/day) in drinking water for 4 weeks. At the end of experiments, mice were sacrificed under anesthesia by intraperitoneal injection of 0.8% pentobarbital sodium (60 mg/kg).

Atherosclerotic lesion analysis
As described previously [37], Oil Red staining was used for neutral lipids in atherosclerotic lesion. AGING

AMPK activity assay
The SAMS peptide was used to assay AMPK activity as previously described [38,39].

Statistical analysis
All data are expressed as mean ± S.E.M. Multiple comparisons were performed using a one-way ANOVA followed by Bonferroni corrections. Two-sided P < 0.05 was considered significant.

Reagents and animals
Polyclonal or monoclonal antibodies against pAMPKα-Thr172, AMPKα, AMPKγ1, and α-SMA were obtained from Santa Cruz Biotechnology or Cell Signaling Company. All secondary antibodies were from Cell Signaling Technology. AMPKγ1 siRNA was from Santa Cruz. All PCR primers were purchased from Invitrogen. Other chemicals, if not indicated, were from Sigma-Aldrich (St. Louis, MO). Male Apoe -/mice were obtained from Beijing Huafukang Animal Experimental Center. Mice were housed in temperature-controlled cages with a 12-hrs light-dark cycle. The animal protocol was reviewed and approved by the University of Central South Animal Care and Use Committee.

Cell culture
Primary murine vascular smooth muscle cells (VSMCs) isolated from mice were grown in Smooth Muscle Cell Medium (Sciencell, USA) supplemented with 2% fetal bovine serum, penicillin (100U/ml) and streptomycin (10mg/ml) as described previously [1]. In all experiments, cells were used between passages 4 and 8. All cells were incubated at 37° C in a humidified atmosphere of 5% CO2 and 95% air. Cells were grown to 80% confluency before being treated with different agents.

Transfection of siRNA into cells
Transient transfection of siRNA was carried out according to Santa Cruz's protocol [2]. Briefly, the siRNAs were dissolved in siRNA buffer (20 mM KCl; 6 mM HEPES, pH 7.5; 0.2 mM MgCl2) to prepare a 10 µM stock solution. Cells grown in 6 well plates were transfected with siRNA in transfection medium (Gibcol) containing liposomal transfection reagent (Lipofectamine RNAimax, Invitrogen). For each transfection, 100 µl transfection medium containing 4 µl siRNA stock solution was gently mixed with 100 µl transfection medium containing 4 µl transfection reagent. After 30-min incubation at room temperature, siRNA-lipid complexes were added to the cells in 1.0 ml transfection medium, and cells were incubated with this mixture for 6 h at 37° C. The transfection medium was then replaced with normal medium, and cells were cultured for 48 hours.

Generations of DNA constructs
LncRNA-ANRIL cDNA or shRNA was purchased from Origene Company. The adenovirus construction compassing lncRNA-ANRIL cDNA or shRNA was generated using the AdMax (Microbix) and pSilencer™ adeno 1.0-CMV (Ambion) systems according to the manufacturers' recommendations. Viruses were packaged and amplified in HEK293A cells and purified using CsCl2 banding followed by dialysis against 10 mM Tris-buffered saline with 10% glycerol. Titering was performed on HEK293 cells using the Adeno-X Rapid Titer kit (BD Biosciences Clontech, Palo Alto, CA, USA) according to the manufacturer's instructions.

Adenovirus infections to cells or animals
Cells were infected with adenovirus overnight in antibiotics-free medium supplemented with 2% FBS. The cells were then washed and incubated in fresh medium for an additional 12-hour before experiments. For infecting mice, adenovirus was injected via tail vein under pressure in 100μl of PBS with 7.6 X 10 7 IFUs of loaded virus. The concentration of DNA was 10 mg/kg.

Western blot analysis
Cell lysates or tissue homogenates were subjected to western blot analysis, as described previously [3]. The protein content was assayed by BCA protein assay reagent (Pierce, USA). Protein of 20 µg was loaded to SDS-PAGE and then transferred to membrane. Membrane was incubated with a 1:1000 dilution of primary antibody, followed by a 1:5000 dilution of horseradish peroxidase-conjugated secondary antibody. Protein bands were visualized by ECL (GE Healthcare). The intensity (area X density) of the individual bands on Western blots was measured by densitometry (model GS-700, Imaging Densitometer; Bio-Rad).

RNA-immunoprecipitation (IP) assay
The Magna RIP kit was used for RNA IP assay as described previously [4]. Briefly, whole-cell lysates were incubated at 4° C overnight with magnetic protein A/G beads pretreated with 5 μg rabbit IgG or AMPK gamma antibody. Beads were washed and incubated with proteinase K buffer (30 min at 55° C), then RNA was isolated from immunoprecipitates, and cDNA was synthesized. lncRNA-ANRIL cDNA was amplified by PCR.

RNA quantifications by RT-qPCR
Total RNA was isolated using a TRIzol-based (Invitrogen) RNA isolation protocol. For mRNA detections, the iScript cDNA Synthesis Kit (Bio-Rad) was used to synthesize first-strand cDNA according to the manufacturer's protocol [5]. Reactions were run for 30 cycles at conditions as follows: denaturation for 30 seconds at 94° C, annealing for 30 seconds at 57° C, and extension for 30 seconds at 72° C. Constitutively expressed GADPH mRNA was amplified as control. All primers were presented in supplementary Table 1.

Animal experimental protocols
Mice received tail vein injection of adenovirus containing scramble shRNA or lncRNA-ANRIL shRNA. Two weeks later after injection, mice were received western diet with or without metformin (150 mg/kg/day) in drinking water for 4 weeks. At the end of experiments, all mice were sacrificed under anaesthesia by intraperitoneal injection of 0.8% pentobarbital sodium (60 mg/kg).

Atherosclerotic lesion analysis
As described previously [6], after being fed the Western diet for 4 weeks, the mice were fasted for 14 h and then were anesthetized and euthanized. The heart and aortic tissue were removed from the ascending aorta to the ileal bifurcation and placed in 4% paraformaldehyde for 16 h. After fixation, the adventitia was thoroughly cleaned under a dissecting microscope. 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 consecutive sections were collected from each mouse and stained with Oil Red O for neutral lipids, and counterstained with hematoxylin to visualize the nuclei. Plaques were captured under the 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. Digital images of the aorta were captured under a stereomicroscopy, 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).

Immunofluorescence (IFC)
As described previously [7], after treatment, cells on sterile glass cover slips were rinsed by cold PBS and then fixed by incubation with 10% formalin in PBS for 10 minutes. Block cells by 5% BSA for 30 minutes. Incubate cells with primary antibody for 1 hour at room temperature or overnight at 4° C. After washing, incubate with fluoresence-conjugated secondary antibody for 45 minutes. Take picture in fluorescence microscope.

Detection of lncRNA-ANRIL by fluorescence in situ hybridization (FISH)
As described previously with minor modifications [5], FISH of lncRNA-ANRIL was performed on 5-μm paraffin embedded arterial sections. Briefly, paraffin embedded tissue sections were cut using a microtome (Leica, RM2235) and mounted on polylysine microscope slides (Fisher Scientific) and stored at room temperature (RT) until FISH. Paraffin wax was removed in xylene, sections rehydrated in a series of decreasing ethanol solutions and washed with PBS before fixing in 4% paraformaldehyde. To block endogenous peroxide activity, tissue sections were treated with 0.3% H2O2 and washed in PBS before acetylating in acetic anhydride/triethanolamine. Sections were then washed in 2 X SSC and PBS before permeabilisation with proteinase K (5 μg/ml) and washes with PBS. Probes (5' and 3'-DIG labelled LNA miRCURY probes; Exiqon) were denatured at 90° C before dilution inhybridisation buffer (50% formamide, 0.3 M NaCl, 20 mM Tris-HCL, 5 mM EDTA, 10 mM NaPO4, 10% Dextran sulphate, 1 X Denhardt's solution, 0.5 mg/ml yeast tRNA). Tissue sections were hybridised with lncRNA-ANRIL (200 nM) overnight at 21° C below the predicted Tm value of the probe. After post-hybridization washes in 5 X SSC at RT; 50% formamide/1 X SSC/0.1% Tween 20 at the hybridisation temperature; 0.2 X SSC at RT, FISH signals were detected using an Anti-Digoxigenin antibody and the tyramide signal amplification system (PerkinElmer) according to the manufacturer's instructions. Tissue sections were mounted in Vectashield (Vector laboratories). All fluorescence images were analyzed with a Nikon TE2000-U inverted microscope.

AMPK activity assay
AMPK activity was assayed by using the SAMS peptide as previously described [8]. Briefly, duplicate tubes with 200 μg of protein from each sample were prepared and mixed with 500 μl of IP buffer (lysis buffer plus 1 mM dithiothreitol). AMPK was then immunoprecipitated by adding 10 μg of polyclonal antibody against AMPK and 25 μl of Protein A/Gagarose and incubated at 4° C. After centrifugation (14,000 g, 1 min), the beads were washed with IP buffer and then twice with 10X reaction buffer (400 mM HEPES, pH 7.4, 800 mM NaCl, 50 mM MgCl2, 1 mM dithiothreitol). The AMPK activity was assayed by adding 50 μl of reaction mixtures, consisting of 5 μl of reaction buffer, 10 μl of SAMS peptide (1 mg/ml), 10 μl of ATP working stock consisting of 0.1 μl of 100 mM ATP, 1 μl of 32 P-ATP, and 8.9 μl of H2O, 25 μl H2O, or 25 μl of 400 μM AMP, and incubated at 37° C for 10 min. The beads were quickly pelleted, and 25 μl of supernatant was spotted onto P81 Whatman paper. The filter papers were then washed four or five times with 1% phosphoric acid. After the final wash, the filters were quickly dried and counted in a scintillation counter. The difference between the presence and absence of AMPK is calculated as the AMPK activity.

Measurements of blood glucose, cholesterol, triglyceride, and plasma cytokine levels
The determinations of blood glucose, cholesterol, triglyceride, and plasma cytokine levels in serum were assayed using commercial kits, which were accordance with the protocols as described previously [7].

Patients and sample processing
Left anterior descending coronary arteries were obtained from human subjects after sudden death. The demographic data were presented in Supplementary  Table 3. The histology of left anterior descending coronary artery was determined by HE staining. Atherosclerosis was diagnosed as the amount of stenosis of lumen was over 100%. The study was conducted in accordance with the principles of Good Clinical Practice and the Declaration of Helsinki. The study protocol was approved by the Ethical Committee of Central South University Xiangya Hospital, and informed consent was obtained from each human subject.

Statistical analysis
All quantitative results are expressed as mean ± SEM. The normal distribution of data was tested by the Kolmogorov-Smirnov test before statistical comparisons, and the normality/equal variance was tested to determine whether ANOVA was appropriate. All data were analyzed with a 1-way ANOVA followed by Bonferroni post-hoc analyses. Comparisons between two groups were analyzed by unpaired Student's t test between two groups. Statistical analysis was conducted using IBM SPSS statistics 20.0 (IBM Corp., Armonk, NY, USA), and a two-sided P-value < 0.05 was considered significant.