Myeloid Deletion of α1AMPK Exacerbates Atherosclerosis in LDL Receptor Knockout (LDLRKO) Mice

Macrophage inflammation marks all stages of atherogenesis, and AMPK is a regulator of macrophage inflammation. We therefore generated myeloid α1AMPK knockout (MAKO) mice on the LDL receptor knockout (LDLRKO) background to investigate whether myeloid deletion of α1AMPK exacerbates atherosclerosis. When fed an atherogenic diet, MAKO/LDLRKO mice displayed exacerbated atherosclerosis compared with LDLRKO mice. To determine the underlying pathophysiological pathways, we characterized macrophage inflammation/chemotaxis and lipid/cholesterol metabolism in MAKO/LDLRKO mice. Myeloid deletion of α1AMPK increased macrophage inflammatory gene expression and enhanced macrophage migration and adhesion to endothelial cells. Remarkably, MAKO/LDLRKO mice also displayed higher composition of circulating chemotaxically active Ly-6Chigh monocytes, enhanced atherosclerotic plaque chemokine expression, and monocyte recruitment into plaques, leading to increased atherosclerotic plaque macrophage content and inflammation. MAKO/LDLRKO mice also exhibited higher plasma LDL and VLDL cholesterol content, increased circulating apolipoprotein B (apoB) levels, and higher liver apoB expression. We conclude that macrophage α1AMPK deficiency promotes atherogenesis in LDLRKO mice and is associated with enhanced macrophage inflammation and hypercholesterolemia and that macrophage α1AMPK may serve as a therapeutic target for prevention and treatment of atherosclerosis.

Macrophage inflammation marks all stages of atherogenesis, and AMPK is a regulator of macrophage inflammation. We therefore generated myeloid a1AMPK knockout (MAKO) mice on the LDL receptor knockout (LDLRKO) background to investigate whether myeloid deletion of a1AMPK exacerbates atherosclerosis. When fed an atherogenic diet, MAKO/LDLRKO mice displayed exacerbated atherosclerosis compared with LDLRKO mice. To determine the underlying pathophysiological pathways, we characterized macrophage inflammation/chemotaxis and lipid/cholesterol metabolism in MAKO/LDLRKO mice. Myeloid deletion of a1AMPK increased macrophage inflammatory gene expression and enhanced macrophage migration and adhesion to endothelial cells. Remarkably, MAKO/LDLRKO mice also displayed higher composition of circulating chemotaxically active Ly-6C high monocytes, enhanced atherosclerotic plaque chemokine expression, and monocyte recruitment into plaques, leading to increased atherosclerotic plaque macrophage content and inflammation. MAKO/LDLRKO mice also exhibited higher plasma LDL and VLDL cholesterol content, increased circulating apolipoprotein B (apoB) levels, and higher liver apoB expression. We conclude that macrophage a1AMPK deficiency promotes atherogenesis in LDLRKO mice and is associated with enhanced macrophage inflammation and hypercholesterolemia and that macrophage a1AMPK may serve as a therapeutic target for prevention and treatment of atherosclerosis.
Inflammatory processes mark all stages of atherogenesis, from early endothelial activation to eventual rupture of atherosclerotic plaque (1)(2)(3). Experimental interventions aimed at reducing inflammation are atheroprotective, whereas removing or blocking anti-inflammatory molecules (i.e., interleukin [IL]-10 or transforming growth factor-b) accelerates atherosclerosis (1). Arterial wall macrophage accumulation is a key feature of atherosclerosis development (4,5) that is initiated with monocyte chemotaxis into atheroma. Monocytes expressing the chemokine (C-C motif) receptor 2 (CCR2) migrate to activated endothelial cells that secrete monocyte chemoattractant protein 1 (MCP1) and adhere to the injured endothelial cells by vascular cell adhesion molecule 1 (6)(7)(8)(9)(10). Within the intima, monocytes mature into macrophages, which accelerate atherosclerotic formation by producing proinflammatory cytokines, the accumulation of cholesterol, and foam cell formation (1,2). Selective depletion of monocytes/macrophages by diphtheria treatment in a transgenic mouse model expressing a CD11b-driven diphtheria toxin receptor in monocyte/ macrophages provides resistance to atherosclerosis (11). Accumulating evidence strongly supports the notion that macrophage inflammation is a key component of atherosclerosis. Therefore, understanding how macrophage inflammation/ function is altered in deregulated metabolic conditions with excess nutrients (e.g., cholesterol, fatty acids, and glucose) may provide novel mechanisms and therapeutic targets in the prevention and treatment of atherosclerosis. As such, AMPK, a cellular nutrient and energy sensor, has generated interest due to its anti-inflammatory function.
AMPK is an evolutionarily conserved cellular energy sensor that regulates metabolic pathways in lipid, cholesterol, and glucose metabolism (12). We (13) and others (14,15) have shown that activating macrophage AMPK significantly suppresses inflammation induced by lipopolysaccharide (LPS) and free fatty acid (FFA), which is exerted primarily through the a1AMPK isoform in macrophages. Therefore, AMPK may serve as an energy sensor linking nutrient metabolism to the regulation of inflammatory signaling (13). Other studies also reported that AMPK activation suppresses macrophage proliferation induced by oxidized LDL (16). However, whether the anti-inflammatory function of AMPK protects against atherosclerosis is still not clear. For example, administration of the AMPK activator 5-aminoimidazole-4-carboxymide-1-b-D-ribofuranoside (AICAR) to apolipoprotein (apo)E 2/2 mice prevents atherosclerosis development (17). However, the pleiotropic functions of AMPK in multiple tissues have made it difficult to understand whether and to what extent macrophage AMPK expression protects against atherosclerosis. This study was therefore designed to directly investigate the role of macrophage a1AMPK in the atherosclerosis development through the generation of myeloid-specific a1AMPK knockout (MAKO) mice on the genetic background of LDL receptor knockout (MAKO/LDLRKO). These MAKO/LDLRKO mice and their littermate controls (floxed [fl/fl]/LDLRKO) were fed an atherogenic diet for 16 weeks, followed by quantification of 1) atherosclerosis extent; 2) in vitro macrophage inflammation and chemotaxis; 3) macrophage content, inflammation, and chemotaxis in the atherosclerotic aortic lesions; and 4) plasma lipid/cholesterol profile.

RESEARCH DESIGN AND METHODS
Mice a1AMPK fl/fl mice were provided by Dr. Sean Morrison at The University of Texas Southwestern Medical School (18). Lysozyme Cre +/2 (Lyz-Cre) (19) and LDLRKO mice (20) were purchased from The Jackson Laboratory (Bar Harbor, ME). The fl/fl mice were crossed with Lyz-Cre mice to generate the MAKO mice. MAKO mice were further bred with LDLRKO to generate MAKO/LDLRKO mice with myeloid-specific deletion of a1AMPK on the genetic background of LDLRKO. Eight-week-old MAKO/ LDLRKO and their littermate control fl/fl/LDLRKO mice (eight mice/group) were fed an atherogenic diet containing 42% calories from fat and 0.2% (wt/wt) cholesterol for 16 weeks (Research Diets D12079B). Body weight was monitored weekly. All animal procedures were approved by the Georgia State University Institutional Animal Care and Use Committee.
To study insulin sensitivity, 6-week-old MAKO and their littermate control fl/fl mice (eight mice/group) were fed a high-fat (HF) diet containing 60% of calories from fat (Research Diets D1249) for 16 weeks to establish dietinduced obesity. Blood glucose was measured with an OneTouch Ultra Glucose Meter (LifeScan, Mulpitas, CA), and insulin was measured by a rat insulin ELISA kit (Crystal Chem, Downers Grove, IL). Plasma adiponectin was also measured by an ELISA kit (Crystal Chem). Glucose tolerance tests (GTTs) and insulin tolerance tests (ITTs) were conducted as we previously described (21,22).

Quantification of Atherosclerosis
Atherosclerosis development was evaluated as described previously (23). Briefly, mice were killed after 4 h of food removal. The circulatory system was perfused with PBS before the heart and aorta were removed. The upper onethird of the heart was dissected and embedded in optimal cutting temperature compound (Sakura Tissue-Tek, Torrance, CA), frozen, and stored at 280°C. Blocks were serially cut at 8-mm intervals and stained with hematoxylin and 0.5% Oil Red O (Sigma-Aldrich) to evaluate aortic sinus atherosclerotic intimal area. Atherosclerotic lesion area and Oil Red O-positive area were quantified using Image-Pro Plus software (Media Cybernetics, Bethesda, MD). Three sections of staining were quantified for each mouse, and eight mice per group were used in each cohort we studied.
The whole aorta (from the sinotubular junction to the iliac bifurcate) was dissected and fixed in 10% formalin, and the adventitia was cleaned. Aortas were opened along the longitudinal axis and pinned onto black silicon elastomer (Rubber-Cal, Santa Ana, CA) for the quantification of atherosclerotic lesion area. The percentage of total aortic surface covered with atherosclerotic lesions was quantified by Image-Pro Plus software.

Plasma Lipid Analysis
Mice were fasted for 4 h prior to euthanization. Blood was collected to analyze plasma lipid and cholesterol profiles, as described previously (23). Briefly, plasma total cholesterol (TC) (Pointe Scientific, Canton, MI), free cholesterol (FC) (Wako, Richmond, VA), and triglyceride (TG) (Wako) concentrations were measured using enzymatic assays according to the manufacturer's instructions. Cholesterol ester (CE) was calculated by multiplying the difference between TC and FC by 1.67 (24). Plasma lipoproteins were separated by size using fast protein liquid chromatography, and cholesterol distribution among lipoprotein fractions was determined by enzymatic cholesterol assays.

Cell Culture
To isolate bone marrow-derived macrophages (BMDMs), bone marrow was flushed from the femur and tibia, dispersed, and cultured in DMEM containing 20% FBS and 30% L929 conditional medium for 5-7 days. Peritoneal macrophages were isolated by lavage 5 days after an intraperitoneal injection of 3% thioglycolate solution (Difco; BD Biosciences, San Jose, CA). Cells were stimulated with 10 ng/mL LPS for 4 h, and mRNA expression levels were measured by quantitative real-time RT-PCR.

Cholesterol Efflux to ApoA-I or HDL
Macrophage cholesterol efflux to apoA-I or HDL was measured as described previously (23). Briefly, BMDMs were isolated from MAKO/LDLRKO and their littermate control fl/fl/LDLRKO mice as described above. Cells were incubated with 1 mCi/mL [ 3 H]-cholesterol in the presence of 50 mg/mL acetylated LDL (Kalen Biomedical, Montgomery Village, MD) for 24 h. Cells were washed and equilibrated for 2 h in medium with lipid-free BSA. HDL-or apoA-I-mediated cholesterol efflux was performed by adding HDL or apoA-I for an additional 24 or 4 h, respectively. Radioactivity was measured in medium and cell lysates. Cholesterol efflux was calculated as radioactivity in the medium divided by the sum of radioactivity in medium and cell lysate.

Cholesterol Content in Peritoneal Macrophages
Peritoneal macrophages were isolated from MAKO/LDLRKO and their fl/fl/LDLRKO littermate controls fed an atherogenic diet for 16 weeks, as described above. Cells were plated at 1 3 10 6 /well in six-well plates for 2 h and washed, and lipid was extracted with isopropanol. Cellular cholesterol content was measured as we previously described (23).

Hepatic Lipid Measurement
Liver lipid extraction was conducted as previously described with minor modifications (25). Briefly, ;100 mg of liver was thawed, minced, and weighed in a glass tube, and 3 mL CHCl 3 :MeOH (2:1) was added to the samples to extract lipid at room temperature overnight. The samples were further washed with CHCl 3 :MeOH (2:1) twice, and the combined solvent extracts were dried at 60°C. Three milliliters of CHCl 3 :MeOH (2:1) and 0.6 mL 0.05% H 2 SO 4 were added into tubes, and the phases were separated by 2,000 revolutions per minute centrifugation for 15 min. An aliquot of the bottom phase was dried down. The lipid extraction was dissolved in 2% Triton X-100 and used for TG quantification using a L-Type TG M test (Wako Chemicals). The data were normalized to liver weight.

Immunoblotting
Immunoblotting was performed as described previously (23). SDS-PAGE was run with 2 mL plasma, and goatanti-human apoB antibody (1:1,000; Academy Bio-Medial Co.) was used to detect apoB protein levels.

Immunohistochemistry
Sections of aortic sinus were immunostained with the following primary antibodies: rat monoclonal antibody against CD68 (Clone FA11, 1:75; AbD Serotec, Raleigh, NC), followed by staining with alkaline phosphatase-conjugated mouse anti-rat (for CD68, 1:50) secondary antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA). Control slides contained no primary antibody. The CD68-positive areas were analyzed using Image-Pro Plus software.
Total RNA Isolation From the Whole Aorta, Aortic Macrophages, and Quantitative RT-PCR For chemokine expression in aortas, the whole aortas (from the sinotubular junction to the iliac bifurcation) were dissected and immediately frozen in liquid nitrogen. Total RNA was extracted using the TRI Reagent kit (Molecular Research Center, Cincinnati, OH), and gene expression was assessed by quantitative RT-PCR using ABI Universal PCR Master Mix (Applied Biosystems, Foster City, CA) using a Stratagene Mx3000p thermocycler (Stratagene, La Jolla, CA). Cyclophilin was used to normalize the gene expression data. The primer and probe sets used in the assays were purchased from Applied Biosystems/Life Technologies (Grand Island, NY).
For gene expression in aortic macrophages, the digested aortas were used to isolate macrophages with rat anti-mouse F4/80 antibodies (AbD Serotec), followed by a pull down of F4/80-positive cells with sheep anti-rat microbeads using a magnetic-activated cell sorting system (Miltenyi Biotec, Auburn, CA), as described previously (22,23). The isolated F4/80-positive aortic macrophages were used for RNA extraction with TRI Reagent (Molecular Research Center) and gene expression analysis as described above (22,23).

Blood Monocyte Labeling and Quantification of Monocyte Recruitment Into Intima
Ly-6C high (Ly-6C hi ) monocytes were labeled as described previously (26). Briefly, MAKO/LDLRKO and their littermate control fl/fl/LDLRKO mice were injected with 250 mL clodronate-liposome through the retro-orbital venous sinus to deplete blood monocytes. Eighteen hours later, 250 mL diluted 1.0 mm Fluoresbrite Yellow Green Microspheres (1:4 diluted with PBS) were injected into the bloodstream of mice to label Ly-6C hi monocytes. After 24 h, blood was drawn to determine the percentage of bead-positive monocytes each mouse. The aortic sinus was dissected 3 days after injection to quantify the number of intimal beads, as described previously (26).

Macrophage Migration and Adhesion Assay
Macrophage migration and adhesion assay was performed as described previously (23). Macrophage migration assay was performed using a CytoSelect 96-Well Cell Migration Assay kit (Cell Biolabs, San Diego, CA) according to the manufacturer's instructions. Briefly, BMDMs (1.5 3 10 5 /well) were plated in the membrane chamber, which was placed on top of the feeder tray. Cells were incubated with 10 ng/mL MCP1 at 37°C for 20 h. Migrated cells were dissociated from the membrane, lysed, and quantified using CyQUANT GR fluorescent dye with a Victor 3 plate reader (PerkinElmer, Waltham, MA).
BMDM adhesion assays were conducted with a CytoSelect Leukocyte-endothelium Adhesion Assay kit (Cell Biolabs) according to the manufacturer's instructions. Briefly, human microvascular endothelial cells, adult dermis (HMVECad) (Invitrogen, Carlsbad, CA) were seeded at 6 3 10 4 /well in a 96-well plate overnight. BMDMs were labeled with LeukoTracker dye for 1 h at 37°C and incubated with a monolayer of the HMVECad cells, which were pretreated with 10 ng/mL tumor necrosis factor-a (TNF-a) for 6 h. Wells were washed 1 h after incubation, and adherent macrophages were lysed. The fluorescent signal from each well was measured with a Victor 3 plate reader (PerkinElmer).

Statistics
All data are expressed as mean 6 SEM. Differences between groups were analyzed for statistical significance by t test or one-or two-way ANOVA, as appropriate.

Myeloid Deletion of a1AMPK Exacerbates the Development of Atherosclerosis in LDLRKO Mice
To study the role of macrophage a1AMPK in the development of atherosclerosis, we generated MAKO/LDLRKO mice as described in RESEARCH DESIGN AND METHODS. The a1AMPK mRNA and protein levels in BMDMs isolated from MAKO/LDLRKO mice were decreased by 82% and 60% ( Supplementary Fig. 1A-C), respectively, compared with littermate controls. Similar suppression of a1AMPK mRNA and protein levels was also observed in peritoneal macrophages isolated from MAKO/LDLRKO mice (Supplementary Fig. 1D-F). Moreover, we found that the inhibition of a1AMPK mRNA and protein in BMDMs isolated from MAKO/LDLRKO mice also significantly suppressed the basal phosphorylation of aAMPK and acetyl-CoA carboxylase (ACC), a direct downstream target of AMPK (Supplementary Fig. 2A). Further, the activated phosphorylation of aAMPK and ACC by AICAR, a specific agonist of AMPK, was also substantially blocked in BMDMs with a1AMPK deficiency (Supplementary Fig. 2B). These data suggest that myeloid deletion of a1AMPK downregulates AMPK signaling in macrophages. We next determined whether the myeloid deletion of a1AMPK affects the AMPK expression and signaling in other key metabolic tissues, such as liver, fat, and skeletal muscle, due to potential leakiness of the Cre expression in those tissues. We found, however, that myeloid deletion of a1AMPK did not change a1AMPK expression in liver, fat, and skeletal muscle and that the myeloid deficiency of a1AMPK did not alter the AMPK signaling, including the phosphorylation of aAMPK and ACC in those tissues ( Supplementary  Fig. 3). These data suggested that the a1AMPK deletion is likely restricted to myeloid lineage cells.
Eight-week-old MAKO/LDLRKO and their littermate control fl/fl/LDLRKO mice were fed an atherogenic diet containing 42% calories from fat and 0.2% (wt/wt) cholesterol for 16 weeks. Because previous studies reported that the expression of macrophage a1AMPK was suppressed by inflammatory stimuli such as LPS and FFAs (13) and by obesity (15), we therefore determined whether myeloid deletion of a1AMPK by the genetic approach can further decrease the supposedly lower macrophage a1AMPK by feeding of an atherogenic diet. Indeed, we still observed an 80% reduction of a1AMPK mRNA in peritoneal macrophages isolated from MAKO/LDLRKO mice compared with their littermate control fl/fl/LDLRKO mice fed the atherogenic diet for 16 weeks (Supplementary Fig. 4A). Similarly, a comparable inhibition of a1AMPK mRNA was also observed in macrophages isolated from the aortas of MAKO/LDLRKO mice fed the atherogenic diet ( Supplementary Fig. 4B).
Myeloid deletion of a1AMPK did not alter body weight ( Supplementary Fig. 4C) or liver and fat pad mass (Supplementary Fig. 4D). However, myeloid deletion of a1AMPK significantly increased the atherosclerotic lesions in intimal areas of aortic roots by 38% (Fig. 1A and B) and also enhanced the atherosclerotic surface lesion areas in the whole aorta by 63% (Fig. 1C and D) compared with the control mice.

Myeloid Deletion of a1AMPK Increases Macrophage Inflammation and Infiltration Into Atherosclerotic Plaques
Macrophage inflammation is a key component in atherosclerosis development (2). We (13) and others (14) have reported that a1AMPK is a negative regulator of macrophage inflammation. We therefore examined whether myeloid deletion of a1AMPK worsens atherosclerosis by promoting macrophage inflammation and chemotaxis. Indeed, we found in this study that macrophage a1AMPK deficiency substantially promoted LPS or interferon-gstimulated expression of proinflammatory genes, such as TNF-a, IL-6, IL-1b, MCP1, and inducible nitric oxide synthase (iNOS), in primarily cultured BMDMs isolated from MAKO mice ( Fig. 2A-E). To test whether macrophage a1AMPK deficiency also affects macrophage function, we performed macrophage adhesion and chemotaxis assays with BMDMs isolated from MAKO/LDLRKO mice. We found that a1AMPK deficiency increased macrophage adhesion to endothelial cells and enhanced the ability of macrophages to migrate toward an MCP1 gradient ( Fig. 2F and G).
We then determined whether the enhanced macrophage inflammation and chemotaxis induced by a1AMPK deficiency affected atherosclerotic plaque macrophage content. Using immunohistochemistry, we stained macrophages in aortic sinus sections with anti-CD68 antibodies and found that atherosclerotic plaque macrophage content was higher in MAKO/LDLRKO mice compared with the control mice (Fig. 3). Interestingly, we found that the expression of genes involved in chemotaxis, such as MCP1, chemokine (C-X-C motif) ligand 1 (CXCL1), CXCL2, intercellular adhesion molecule 1, and vascular cell adhesion molecule, was significantly increased in aortas isolated from MAKO/LDLRKO mice compared with that of control mice (Fig. 4A), suggesting that the increased production of chemokines in the atherosclerotic plaques may attract monocyte infiltration. We also determined the content of T cells in the whole aortas. We found that the expression of T-cell markers did not change in the aortas isolated from MAKO/LDLRKO and their littermate control mice ( Supplementary Fig. 5), suggesting that T-cell infiltration was not involved in the development of atherosclerosis in MAKO/LDLRKO mice. To determine the inflammatory status of macrophages infiltrated in the atherosclerotic plaques, we isolated macrophages from the plaques and measured the expression of inflammatory genes. We found that the expression of proinflammatory cytokines, such as TNF-a, IL-1b, MCP1, and the inflammasome marker NLR family pyrin domain containing 3 (NLRP3), was significantly upregulated in macrophages isolated from atherosclerotic plaques of MAKO/LDLRKO mice (Fig. 4B). Ly-6C hi monocytes are a subset of circulating monocytes that infiltrate into inflammatory sites, including atherosclerotic lesions (27,28). Using FACS analysis, we examined this particular monocyte subpopulation and found that the percentage of Ly-6C hi monocytes was increased in overall circulating monocytes of MAKO/LDLRKO mice, whereas the composition of Ly-6C low monocytes was decreased accordingly (Fig. 5A). However, we did not detect any differences in the composition of major immune cell types in blood between MAKO/LDLRKO mice and control mice, including monocytes, CD4 + and CD8 + T cells, and CD19 + B cells ( Supplementary  Fig. 6). To further test the ability of in vivo migration of a1AMPK-deficient macrophages, we performed in vivo monocyte recruitment to intima in MAKO/LDLRKO and control mice. We found that macrophage a1AMPK deficiency markedly increased monocyte migration to the intima of MAKO/LDLRKO mice ( Fig. 5B and C).

Myeloid Deletion of a1AMPK Causes Hypercholesterolemia
Another important risk factor for the development of atherosclerosis is hypercholesterolemia (29). Interestingly, MAKO/LDLRKO mice exhibited hyperlipidemia and hypercholesterolemia evident by increased plasma TG, TC, and CE levels, as well as FFA levels ( Fig. 6A and  B), compared with control mice. In addition, increased plasma cholesterol content in both VLDL and LDL was observed in MAKO/LDLRKO mice, whereas the HDL cholesterol concentration was without change (Fig. 6C  and D). Because apoB100 is a major structural apo in VLDL and LDL particles and is important for VLDL and LDL production, we measured the abundance of this apo in plasma. Immunoblotting analysis showed higher plasma apoB100 and apoB48 protein levels in MAKO/ LDLRKO mice than in control mice fed an atherogenic diet for 10 weeks and 16 weeks (Fig. 6E-G and Supplementary Fig. 7). We also found that hepatic apoB mRNA levels (Fig. 7A) and protein expression (Fig. 7B) were increased in MAKO/LDLRKO mice compared with the control mice. Because inflammation is one of the factors that have been shown to increase apoB expression (30,31), we examined the hepatic inflammatory status of MAKO/LDLRKO mice. Real-time quantitative RT-PCR analysis revealed that the expression of proinflammatory genes, such as TNF-a, IL-1b, MCP1, and iNOS, was upregulated in the liver of MAKO/LDLRKO mice fed an atherogenic diet (Fig. 7C). However, we did not observe changes in hepatic TG, TC, FC, and CE content in MAKO/LDLRKO mice fed an atherogenic diet (Supplementary Fig. 8).
Macrophages play a critical role in the development of atherosclerosis by accumulating lipid and cholesterol to form lipid-laden foam cells (29). However, we found that macrophage a1AMPK deficiency did not change the content of TC, CE, and FC accumulation in peritoneal macrophages isolated from MAKO/LDLRKO mice fed the atherogenic diet (Supplementary Fig. 9A). In addition, MAKO/LDLRKO and control mice had similar expression of macrophage genes involved in cholesterol metabolism and transport, such as liver X receptor a (LXR-a), ATPbinding cassette, subfamily A, member 1 (ABCA1), and ATP-binding cassette, subfamily G, member 1 (ABCG1) (Supplementary Fig. 9B), and genes involved in fatty acid oxidation, such as carnitine palmitoyltransferase 1a (CPT1), cytochrome c oxidase subunit I (COX1), peroxisome proliferative-activated receptor g coactivator 1 a (PGC1-a), PGC1-b, and acyl-CoA oxidase 1 (ACOX1) (Supplementary Fig. 9C). Finally, macrophage a1AMPK deficiency had no effect on cholesterol efflux to apoA-I and to HDL (Supplementary Fig. 9D).

Myeloid Deletion of a1AMPK Does Not Affect Insulin Sensitivity in MAKO/LDLRKO Mice Fed an Atherogenic Diet but Causes Insulin Resistance in MAKO Mice Fed an HF Diet
Since Galic et al. (15) reported that disruption of macrophage AMPK signaling by AMPK-b1 knockout through bone marrow transplantation exacerbates obesity-induced inflammation and insulin resistance, we therefore determined whether myeloid deletion of a1AMPK would affect insulin sensitivity in MAKO/LDLRKO mice. Fed the atherogenic diet containing 42% of fat and 0.2% (wt/wt) of cholesterol for 16 weeks, MAKO/LDLRKO mice had no changes in insulin sensitivity evident by normal GTTs, ITTs, and fed insulin levels compared with their littermate controls ( Supplementary Fig. 10A-C), and they did not show any difference in circulating adiponectin levels (Supplementary Fig. 10D). To more carefully examine the potential effect of macrophage a1AMPK on obesityinduced insulin resistance, we also generated another mouse model with myeloid deletion of a1AMPK on the C57BL/6J background, in which the intact LDL receptor (LDLR) gene was preserved. a1AMPK fl/fl mice were crossed with Lyz-Cre mice to generate MAKO mice. MAKO and their littermate control fl/fl mice were fed a standard HF diet containing 60% of calories from fat, which we have successfully used to induce diet-induced obesity and insulin resistance (21,22). There was no difference in body weight between MAKO and the control fl/fl mice (Supplementary Fig. 11A). However, MAKO mice displayed a phenotype of glucose intolerance assessed by GTTs and insulin resistance assessed by ITTs (Supplementary Fig. 11B and C) with enhanced blood insulin levels ( Supplementary Fig. 11D), whereas there was no difference in blood adiponectin levels ( Supplementary Fig.  11E). This was associated with increased expression of proinflammatory cytokine IL-6 and TNF-a in adipose tissue of MAKO mice fed the HF diet ( Supplementary Fig.  12).

DISCUSSION
The current study was designed to address the hypothesis that macrophage a1AMPK negatively regulates the development of atherosclerosis by dampening inflammation. The plausibility of this hypothesis derives from several prior observations. First, macrophage inflammation plays a key role in the development of atherosclerosis and is involved in all stages of atherogenesis, including atheroma formation, progression, and rupture (1,2,23). Second, we (13) and others (14) (15) have shown that a1AMPK is a negative regulator of macrophage inflammation. Whether the anti-inflammatory function of a1AMPK protects against atherogenesis is not clear, however. To address this question, we generated mice with myeloidspecific deletion of a1AMPK in the LDLRKO background and investigated diet-induced atherosclerosis progression. We found that myeloid deficiency of a1AMPK markedly promotes atherosclerosis development. To examine the underlying pathophysiological changes, we investigated macrophage inflammation/chemotaxis and lipid/ cholesterol metabolism.
Because macrophage inflammation plays a key role in the development of atherosclerosis (1,2), we first determined whether a1AMPK deficiency worsened macrophage inflammation/chemotaxis, contributing to atherosclerosis development in MAKO/LDLRKO mice. AMPK is an evolutionarily conserved serine/threonine kinase and has been viewed as a metabolic sensor or fuel gauge that monitors cellular AMP and ATP levels (12). Activation of AMPK represses ATP-consuming anabolic pathways (e.g., fatty acid and cholesterol synthesis) and induces ATP-producing catabolic pathways (e.g., fatty acid oxidation) (12). Thus, AMPK is a key signaling molecule that integrates hormonal and nutrient signals to regulate metabolic pathways. Recently, we (13) and others (14) have shown that macrophage AMPK activation prevents LPS-and FFA-induced proinflammatory cytokine expression and nuclear factor-kB signaling, indicating that AMPK is an important negative regulator of macrophage inflammation (13). The anti-inflammatory function of macrophage AMPK might be mediated by upregulation of SIRT1 expression and activity (13,21). SIRT1 can subsequently deacetylate the nuclear factor-kB subunit p65, leading to downregulation of macrophage inflammation (13,32). Moreover, Galic et al. (15) demonstrated that inhibiting macrophage AMPK signaling by AMPK-b1 deletion reduces fatty acid oxidation and increases diacylglycerol accumulation, leading to enhanced expression of proinflammatory markers. With this in mind, the inhibition of fatty acid oxidation caused by AMPK deficiency may increase accumulation of fatty acids in macrophages, leading to endoplasmic reticulum stress that may further activate inflammatory response (33,34). Downregulation of macrophage AMPK may also mediate fatty acid activation of NLRP3 inflammasome, resulting in IL-1b production (35,36). In sum, macrophage AMPK controls multiple pathways in regulation of the inflammatory network.
AMPK may therefore be the potential link between nutrient metabolism and inflammation. Indeed, we showed that AICAR, an AMPK activator, reduces obesity-induced insulin resistance through inhibiting macrophage inflammation (21). As such, macrophage AMPK activation becomes a very attractive target for treatment of insulin resistance. A report from the Steinberg group showed that hematopoietic AMPK-b1 deficiency worsens obesity-induced inflammation and insulin resistance (15). Activation of AMPK by AICAR also decreases atherosclerosis in apoE 2/2 mice (17). However, pleiotropic AMPK activation by AICAR in multiple metabolic tissues has made it difficult to determine the specific role of macrophage AMPK in insulin resistance and atherosclerosis. Additional genetic studies involving deletion of AMPK in specific tissues are required to separate the systemic effects on energy metabolism from the direct anti-inflammatory effects and to determine the extent to which the anti-inflammatory effects of AMPK alter atherosclerosis progression. Our current study provides in vitro and in vivo evidence that myeloid deletion of a1AMPK promotes macrophage inflammation and chemotaxis (adhesion and migration) and increases macrophage infiltration into atherosclerotic plaques, which likely contributes to exacerbated atherosclerosis in MAKO/LDLRKO mice. Interestingly, we have previously shown that AMPK signaling is downregulated by inflammatory stimuli and in nutrientrich conditions, such as exposure to LPS and FFAs, and in diet-induced obesity (13). Downregulation of macrophage AMPK in obesity, which is characterized by excess nutrient intake and chronic inflammation (e.g., increased circulating cytokine levels), may result in exaggerated macrophage inflammation and chemotaxis that worsens atherosclerosis development.
We also found that MAKO/LDLRKO mice exhibit increased plasma TG, FFA, TC, and CE concentrations and that the increased cholesterol was confined to the atherogenic lipoproteins, LDL, and VLDL. These results suggest that macrophage a1AMPK-deficient mice have increased hepatic lipoprotein production and/or decreased catabolism, leading to hyperlipidemia and increased atherosclerosis relative to control mice. Although the systemic change of the circulating lipid profile was somewhat unexpected, we think that this is not likely due to leakiness of the Cre gene expression in other tissues such as liver. The Lyz-Cre mouse (19) we purchased from The Jackson Laboratory has been extensively used to generate myeloid deletion of genes of interest and has been reported in many publications from other groups (37)(38)(39)(40) and our group (21). The Cre construct knocked into the endogenous lysozyme gene locus is faithfully expressed in myeloid lineage cells (19). In fact, we also determined whether the myeloid deletion of a1AMPK affects the AMPK expression and signaling in other key metabolic tissues, such as liver, fat, and skeletal muscle, due to potential leakiness of the Cre expression in those tissues. We found that myeloid deletion of a1AMPK did not change a1AMPK expression in liver, fat, and skeletal muscle and that the myeloid deficiency of a1AMPK did not alter the AMPK signaling in those tissues. Our data suggested that the a1AMPK deletion is likely restricted to myeloid lineage cells. However, to exclude potential leakiness of the Cre expression, other approaches, such as bone marrow transplantation, will be warranted to confirm and will also be complementary to our observation on the protective effect of macrophage a1AMPK in atherosclerosis. Alternatively, it seems plausible that increased macrophage inflammation is responsible for altered hepatic lipid metabolism and increased plasma lipid levels. This idea is supported by published data demonstrating that acute inflammation increases the synthesis of apoB (30,31) and hepatic lipids (41), leading to hyperlipidemia. Although MAKO/LDLRKO mice had no increase in hepatic lipid content relative to control mice, we have previously shown that hepatic lipid content does not necessarily reflect hepatic lipid secretion (42)(43)(44). Because macrophage a1AMPK is a negative regulator of macrophage inflammation, hepatic macrophages (e.g., Kupffer cells) without a1AMPK presumably secrete proinflammatory cytokines that contribute to hepatic inflammation, although our results do not rule out the possibility that extrahepatic macrophages may contribute to this phenotype. Nonetheless, additional studies are required to address the causative effect of hepatic inflammation in apoB protein production in our study.
We noted the discrepancy of the phenotype of insulin sensitivity between MAKO/LDLRKO and MAKO mice fed the respective atherogenic diet and the standard HF diet. There are two potential reasons. First, MAKO/LDLRKO and MAKO mice were fed two different diets with different fat contents. MAKO/LDLRKO mice were fed the atherogenic diet containing 42% of calories from fat, and MAKO mice were fed the standard HF diet containing 60% of calories from fat (including lards, a more detrimental type of saturated fat). As a result, the MAKO mice fed the HF diet are more obese than the MAKO/LDLRKO mice, with MAKO mice being ;5 g heavier. In addition to the difference in body weight, the two different diets themselves may cause different metabolic consequences. Second, MAKO/ LDLRKO and MAKO mice have different genetic backgrounds. In addition to both lacking myeloid a1AMPK, MAKO/LDLRKO mice have whole-body deletion of LDLR, whereas MAKO mice have the intact LDLR gene. Whether or how the presence or absence of LDLR gene may affect insulin sensitivity is not clear.
In summary, our data demonstrate that myeloid-specific a1AMPK deletion exacerbates atherosclerosis in LDLRKO mice. Myeloid a1AMPK deficiency upregulates macrophage proinflammatory gene expression and enhances macrophage migration and adhesion to endothelial cells, leading to monocyte infiltration into atherosclerotic plaques. MAKO/ LDLRKO mice fed the atherogenic diet also display hypercholesterolemia and hyperlipidemia that further exacerbate atherosclerosis progression. We conclude that macrophage a1AMPK is atheroprotective in LDLRKO mice and may serve as a therapeutic target for prevention and treatment of atherosclerosis. Author Contributions. Q.C., X.C., R.W., L.Z., and X.W. performed most of the experiments. J.S.P. provided technical support on lipid/cholesterol/monocyte traffic/atherosclerosis analyses. J.S.P. and L.Y. contributed to discussion and reviewed and edited the manuscript. Q.C., H.S., and B.X. conceived the hypothesis, designed the study, analyzed the data, and wrote the manuscript. H.S. and B.X. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.