at SciVerse ScienceDirect

Atherosclerosis 229 (2013) 62e70

Contents lists available

Atherosclerosis

journal homepage: www.elsevier .com/locate/atherosclerosis

Adiponectin increases macrophages cholesterol efflux and suppressesfoam cell formation in patients with type 2 diabetes mellitus

Min Wang 1, Duan Wang 1, Yuhua Zhang, Xiaoming Wang, Yan Liu, Min Xia*

Guangdong Provincial Key Laboratory of Food, Nutrition and Health, Department of Nutrition, School of Public Health,Sun Yat-sen University (Northern Campus), Guangzhou, Guangdong Province 510080, China

a r t i c l e i n f o

Article history:Received 19 November 2012Received in revised form4 January 2013Accepted 15 January 2013Available online 6 February 2013

Keywords:AdiponectinMacrophagesCholesterol effluxDiabetes

* Corresponding author. Tel.: þ86 20 87332433; faE-mail address: xiamin@mail.sysu.edu.cn (M. Xia)

1 M.W. and D.W. contributed equally to this work.

0021-9150/$ e see front matter � 2013 Elsevier Irelahttp://dx.doi.org/10.1016/j.atherosclerosis.2013.01.017

a b s t r a c t

Objectives: Low levels of blood adiponectin contribute to an increased risk of cardiovascular disease(CVD) in patients with type 2 diabetes mellitus (T2DM). To determine the mechanism through whichadiponectin deficiency mediates accelerated cardiovascular disease in patients with diabetes, weinvestigated the effects of adiponectin on macrophage cholesterol deposition.Methods and results: 35 diabetic patients and 35 nondiabetic healthy subjects were recruited in thisstudy. Macrophages from patients with diabetes mellitus were cultured in adiponectin-free oradiponectin-supplemented media and exposed to oxidized low-density lipoprotein cholesterol (OxLDL).Adiponectin treatment markedly suppressed foam cell formation in OxLDL-treated macrophages fromdiabetic subjects only, which was mainly attributed to an increase in cholesterol efflux. Adiponectintreatment significantly increased ATP-binding cassette transporter (ABC) ABCG1 mRNA and proteinlevels but not ABCA1, without affecting protein expression of scavenger receptors, including scavengerreceptor-A (SR-A) and CD36 in diabetics. Pharmacological or genetic inhibition of liver X receptor a(LXRa) blocks the adiponectin-mediated ABCG1 expression, suggesting that LXRa activation is necessaryfor the attenuation of lipid accumulation of macrophages by adiponectin. In addition, deletion of theadiponectin receptor (adipoR1) in macrophages from diabetic patients accelerated foam cell formationinduced by OxLDL. Finally, a strong positive correlation was noted between decreased serum adiponectinlevels and impaired cholesterol efflux capacity both before and after adjustment for HDL-C and ApoAI indiabetic patients (both P < 0.001).Conclusions: The present study identifies reduced adiopoR signaling as a critical mechanism underlyingincreased foam cell formation and accelerated cardiovascular disease in diabetic subjects.

� 2013 Elsevier Ireland Ltd. All rights reserved.

1. Introduction

Type 2 diabetes mellitus (T2DM) is a well-known risk factor forin the initiation and development of atherosclerotic cardiovasculardisease, accounting for a high proportion of disability and deathsamong diabetics [1,2]. Several studies indicate that in poorly con-trolled diabetes mellitus, altered insulin signaling and/or hyper-glycemia promote unbalanced cholesterol metabolism, whichfavors oxidized low-density lipoprotein (OxLDL) cholesterol reten-tion andmacrophage-derived foam cell formation, a hallmark of theinitiation and development of atherosclerosis [3e5].

x: þ86 20 87330446..

nd Ltd. All rights reserved.

Cellular cholesterol levels reflect a balance between uptake,efflux, and endogenous synthesis. Under hyperglycemia and/or aninsulin-resistant state, macrophages upregulate the expression ofscavenger receptors (SR-A, and CD36), which have the ability totake up modified lipoproteins [6,7]. On the contrary, members ofthe cholesterol reverse transporter family, such as ATP-bindingcassette (ABC) mainly ABCG1, are also downregulated in responseto high glucose [8,9]. Thus, increased scavenger receptor expression[10] and decreased ABCG1 expression [11] promote macrophagefoam cell formation and are considered as a link between diabetesmellitus and atherosclerosis. Therefore, strategies to modulatemacrophage cholesterol deposition could have therapeutic poten-tial for limiting the accelerated vascular disease observed inpatients with T2DM.

Adiponectin (encoded by Adipoq) is an insulin-sensitizingplasma protein expressed in adipose tissue, and it plays animportant role in insulin-sensitizing, anti-inflammatory and anti-

M. Wang et al. / Atherosclerosis 229 (2013) 62e70 63

atherogenic properties [12,13]. Plasma adiponectin levels arereduced not only among obese patients [14] but also in diseasestates frequently associated with insulin resistance and T2DM [14],dyslipidemia [15], hypertension [16] and coronary artery disease(CAD) [17]. The anti-atherogenic effects of adiponectin include thesuppression of adhesion molecule expression on vascular endo-thelial cells [18] and the inhibition of vascular smooth muscle cellproliferation and migration [19]. Adiponectin also stimulates theproduction of nitric oxide (NO) in endothelial cells [20] and reducesatherosclerosis by suppressing the endothelial inflammatory reac-tion and macrophage-to-foam cell transformation [21]. Therefore,understanding the molecular mechanism of the accelerated athe-rosclerosis induced by hypoadiponectinemia may be crucial fortreating the epidemic of CVD in diabetics.

This study was designed to explore whether hypo-adiponectinemia levels contribute to the increase in macrophage-mediated cholesterol deposition observed in patients with diabetesand investigate the effects of adiponectin onmacrophage cholesteroldeposition in diabetics and nondiabetic matched controls.

2. Materials and methods

2.1. Materials

The recombinant human and mouse full-length adiponectinwere obtained from Alexis (San Diego, CA). CP113818, an acyl-coenzyme A:cholesterol acyltransferase (ACAT) inhibitor, was pur-chased from Shanghai Ennopharm Co., Ltd (Shanghai, China).22-(N-(7-Nitrobenz-2-Oxa-1,3-Diazol-4-yl)Amino)-23,24-Bisnor-5-Cholen- 3b-Ol (NBD-cholesterol, Catlog. N-1148), an environment-sensitive probe for investigating lipid transport processes as welllipideprotein interactions, were purchased fromMolecular Probes,Inc. Eugene, OR). Human recombinant lipid-free apolipoprotein A-I(ApoAI, Catlog. SRP4693), high density lipoprotein (HDL, L8039)from human plasma, human recombinant macrophage colony-stimulating factor (M-CSF, SRP4237), 8-(4-chlorophenylthio)-cyclic AMP (8-CPT-cAMP, C3912) were obtained from SigmaeAldrich (St. Louis, MO). 1,10-dioctadecyl-3,3,30,30-tetramethylindocarbocyanine percholate (Dil)-labeled OxLDL (Invitrogen,Grand Island, NY). The primary rabbit polyclonal antibodies anti-LXRa, CD36, goat anti-SR-A and horseradish peroxidase (HRP)-conjugated anti-rabbit or goat secondary antibody were allobtained from Santa Cruz Biotechnology Inc (Santa Cruz, CA).Rabbit monoclonal antibody to ABCG1 and mouse monoclonalantibody to ABCA1 were provided by Abcam.

2.2. Study population

Our study population included 35 adult subjects with type 2diabetes mellitus on medication, with an age of 62 � 8 years, bodymass index of 25.50 � 2.92 kg/m2, diabetes duration of 4 � 1.4years, and hemoglobin A1c level of 7.94 � 0.58%. We excludedrecently diagnosed diabetes mellitus, pregnancy, known coronaryartery disease, and normal adiponectin. This population was com-pared with 35 normal weight control subjects with no history ofdiabetes mellitus or hypertension. Subjects were recruited from theoutpatient clinic at Guangzhou Military General Hospital inGuangzhou. The study was approved by the ethics committee ofSun Yat-sen University and was conducted in accordance withthe Declaration of Helsinki. Participation was voluntary, and eachparticipant provided written informed consent. The ethyl-enediaminetetraacetic acid samples were immediately centrifuged,aliquoted, and stored at �80 �C until batch analysis. Peripheralblood samples were collected from healthy subjects and patientsafter an overnight fast between 8:00 AM and 10:00 AM.

2.3. Biochemical measurements

Serum levels of total cholesterol, HDL cholesterol, triglyceridesand glucose were measured through an enzymatic method(Wako Pure Chemical Industries) using an automatic analyzer(Hitachi 747 autoanalyzer, Hitachi Ltd, Tokyo). LDL cholesterolwas calculated according to the Friedewald formula: LDLcholesterol¼ total�(triglycerides/5þHDLcholesterol). Serum levelsof apoAI and apo Bweremeasured by immunonephelometry using aBN Prospect analyzer (Dade Behring). Plasma insulin levels weremeasured with a chemiluminescent enzyme immunoassay (Immu-lite 1000 Analyzer). The intra-assay and inter-assay coefficients ofvariation of all measured biochemical parameters were <5%.

2.4. Human serum cholesterol efflux capacity assay

We measured the cholesterol efflux capacity of human serumaccording to previously established methods [22]. J774 murinemacrophages were first radiolabeled with 2 mCi/ml 3H-cholesterol.ABCA1 was upregulated by 6 h incubation with 0.3 mM 8-CPT-cAMP. Subsequently, efflux media from control healthy subjects ordiabetic patients containing 2.8% apolipoprotein B (ApoB)-depletedserumwere added for another 4 h. All steps were performed in thepresence of 2 mg/ml CP113818. Liquid scintillation counting wasused to quantify the efflux of radioactive cholesterol from the cells.The quantity of radioactive cholesterol incorporated into cellularlipids was calculated through the isopropanol extraction of controlwells not exposed to patient serum. The percentage efflux wascalculated by the following formula: [(microcuries of 3H-choles-terol in media containing 2.8% ApoB-depleted serum�microcuriesof 3H-cholesterol in serum-free media) O microcuries of 3H-cho-lesterol in cells extracted before the efflux] � 100. All assays wereperformed in duplicate.

2.5. Human serum adiponectin concentrations measurement

Adiponectin levels in serumwere determined by a commerciallyavailable competitive ELISA assay kit (Cat No. AG-45A-0002, Adip-oGen, Seoul) according to the manufacturer’s instruction. The assayused standards in the range of 0.001e1 mg/ml. The intra-assay andinter-assay coefficients of variation were 4% and 3%, respectively.The normal blood levels of adiponectin range from 8.3 to 13.9 mg/ml.

2.6. Human monocyte isolation and macrophage differentiation

Human monocyte isolation and macrophage differentiationwere performed as described previously [23]. Peripheral monocyteswere isolated by standard Ficoll isolation techniques according tothe manufacturer’s protocol and selected by CD14 marker purity(>90% as assessed by flow cytometry). Monocytes were then platedin tissue culture medium in DMEM containing 1.5 mg/ml glucose,100 ng/ml human macrophage colony-stimulating factor (M-CSF),10% charcoal/dextran-treated fetal bovine serum (FBS, Hyclone),and 1% antibiotic/antimycotic mixture. Monocytes were allowed todifferentiate into macrophages for 3 d in adiponectin-free mediabefore being used in further experiments.

2.7. Mouse peritoneal macrophages

Mouse peritoneal macrophages from db/db mice and wild-typeC57BL/6J mice (n ¼ 8 per each group) were isolated 3 d after theintraperitoneal injection of 4% thioglycollate solution. Cholesteroluptake was evaluated after 6 h of stimulation with Dil-labeledOxLDL cholesterol in mouse macrophages cultured for 24 h in theabsence or presence of adiponectin treatment.

M. Wang et al. / Atherosclerosis 229 (2013) 62e7064

2.8. Real-time quantitative PCR

Total mRNAwas collected from human andmouse macrophageswith TRIZOL Reagent (Invitrogen) according to the manufacturer’sprotocol. Real-time PCR was performed with the QuantiTect SYBRGreen PCR Kit (Qiagen, Valencia, CA) on the Applied Biosystems7500 DNA Sequence Detection System (Applied Biosystems). Thereverse transcription of RNAwas carried out using random primersand SuperScript II Reverse Transcriptase (Invitrogen). Real-timePCR was performed using TaqMan Universal PCR Master Mix orSYBR Green PCR Master Mix (Applied Biosystems) according to themanufacturer’s instructions. Primers for human ABCA1, ABCG1 andGAPDH were ordered from SABiosciences Co., Ltd. (Shanghai,China). Samples were normalized to GAPDH using the DCt (cyclethreshold) method [24].

2.9. Western blot

Human monocyte-derived macrophages and mouse peritonealmacrophages were washed twice with PBS and harvested in lysisbuffer (10 mM HEPES, pH 7.9, 10 mM KCl, 1.5 mM MgCl2, 0.5%Nonidet P-40, 1 mg/ml leupeptin, 10 mg/ml aprotinin, and 1 mMphenylmethylsulfonyl fluoride) for immunoblotting. Nuclei werepelleted at 5000 g for 5 min at 4 �C, and the resulting super-natants were used as the cytosolic fraction. Nuclei were resus-pended in lysis buffer, sheared for 15 s with a microprobesonicator and incubated on ice for 5 min. After centrifugation at12,000 g for 5 min at 4 �C, the supernatants were collected asnuclear extracts. Cellular lysates or nuclear protein extracts(40 mg) were separated with 10% SDS-PAGE and transferred ontoan Immobilon-P membrane (Millipore). The membranes wereblocked in 5% skim milk in TBS-Tween buffer (0.1% Tween 20, pH7.4) and incubated with primary antibodies against ABCG1,ABCA1, CD36, SR-AI and LXRa, followed by secondary antibodies.The b-actin was used as the loading control. Antigen detectionwas performed with an enhanced chemiluminescence detectionsystem (Pierce) [24].

2.10. Cholesterol loading and efflux

Human and mouse macrophages were plated in 12-well platesand loaded with 1 mg/ml NBD-cholesterol in serum-free mediumcontaining 0.2% fatty acid-free BSA for 24 h to equilibrate cellularcholesterol pools. Then, cell layers were rinsed and incubated in theabsence or presence of adiponectin for an additional 6 h. Choles-terol efflux proceeded for 6 h at 37 �C in medium containing 0.2%BSA, 0.2% BSA plus 15 mg/ml lipid-free human apoAI, or 0.2% BSAplus 50 mg/ml of human HDL. At the end of this incubation, thesupernatant was collected and centrifuged at 13,000 rpm for10 min to remove debris. Cells were lysed with 0.5 ml of 0.1 MNaOH. The fluorescence-labeled cholesterol released from the cellsinto the medium was measured with a multilabel counter (Perki-nElmer). Cholesterol efflux was expressed as the percentage offluorescence in the medium relative to the total amount of fluo-rescence (cells and medium). The specific efflux of apoAI or HDLwas calculated by subtracting non-specific efflux in the presence of0.2% BSA only [25].

2.11. Cholesterol uptake assay

Cholesterol uptake was evaluated as previously described [26].Human and mouse macrophages (0.5 � 106 cells/well) in 12-wellplates were cultured in the absence or presence of 20 mg/ml adi-ponectin for 24 h. After then, the cells were washed and incubatedwith 10 mg/ml Dil-labeled OxLDL for another 6 h. Cholesterol uptake

was measured by confocal microscopy and the results were nor-malized to total cell protein concentrations.

2.12. Cholesteryl ester determination

Human monocytes were isolated and allowed to differentiatefor 3 d in 100 mm dishes. Adiponectin was added at the con-centration of 20 mg/ml during the final 24 h. At the end of thisincubation, the lipids in the cells and media were separatelyextracted in chloroform and methanol, the samples were driedunder liquid nitrogen, and free cholesterol and total cholesterolwere measured by gaseliquid chromatography and normalized tocellular protein as described previously [26]. Esterified cholesterolwas calculated as the difference between total and free choles-terol times 1.67.

2.13. Oil red O staining

Macrophages derived from the healthy and diabetic patients’peripheral bloodmononuclear cells were cultured in the absence orpresence of 20 mg/ml adiponectin for 24 h. To assess foam cellformation, macrophage slides were fixed with 4% paraformalde-hyde for 15 min and stained with 0.5% Oil red O and hematoxylin.Images were acquired with the �20 or �40 objective of a micro-scope (Leica SP5II, Leica) equipped with a digital camera using theLAS (Leica Application Suite) software program [26].

2.14. Small interfering RNA (siRNA)

Macrophages obtained from diabetic subjects were infectedwith lentivirus containing shRNA lentiviral particles against humanLXRa (sc-38828-V), AdipoR1 (sc-60123-V), AdipoR2 (sc-60125-V),or control shRNA lentiviral particles (sc-108080) for 48 h. Afterthen, the cells were washed and cultured for 24 h in adiponectin-free or -supplemented media. The efficiency of transfectionwas >70e80% as determined by the quantitative analysis of greenfluorescent protein (GFP) transfection. The effectiveness of theshRNA treatment was assessed by measuring LXRa, AdipoR1 andAdipoR2 protein levels by immunoblotting.

2.15. Mouse peritoneal macrophages

Mouse peritoneal macrophages from db/db mice and wild-typeC57BL/6J mice (n ¼ 8 per each group) were isolated 3 d after theintraperitoneal injection of 4% thioglycollate solution. Cholesteroluptake was evaluated after 6 h of stimulation with Dil-labeledOxLDL cholesterol in mouse macrophages cultured for 24 h in theabsence or presence of adiponectin (at 20 mg/ml).

2.16. Statistical analysis

Categorical variables are presented as frequencies and per-centages, and continuous variables as mean � SD. The sig-nificance of the differences in mean values among two groupswas evaluated by two-tailed unpaired Student’s t tests. More thanthree groups were evaluated by analysis of variance (ANOVA)followed by post-hoc analysis using Bonferroni’s Multiple Com-parison Test. Logistic regression was used to estimate the asso-ciation between cholesterol efflux capacity and plasmaadiponectin levels after adjustment for age, sex, smoking status,presence or absence of diabetes, and LDL-C, HDL-C and apoAIlevels were added in subsequent models. Adjusted odds ratiosare reported for a 1-SD increase in efflux capacity. P < 0.05 wasconsidered significant.

M. Wang et al. / Atherosclerosis 229 (2013) 62e70 65

3. Results

3.1. Adiponectin prevents macrophage foam cell formation fromdiabetic patients

We first examined the impact of adiponectin on monocyte-derived macrophage foam cell formation. Macrophages derivedfrom diabetic patients cultured in media exhibited a significantincrease in foam cell formation induced by OxLDL compared withmacrophages cultured under adiponectin-supplemented con-ditions (Fig. 1A). Direct lipid analysis showed that adiponectin-treated macrophages exposed to OxLDL had almost 50% less totalcholesterol (Fig. 1B) content and 55% less cholesteryl ester levels(Fig. 1C) than macrophages maintained in adiponectin-free media(both P < 0.01). However, in macrophages obtained from non-diabetic controls, OxLDL-induced cholesteryl ester formation wasdemonstrated to be reduced in the presence of adiponectin, but to aless extent (33%, P < 0.05) compared with diabetic macrophagescultured with adiponectin (Fig. 1D). These data suggest that adi-ponectin confers a protective role in the formation of macrophagefoam cells from monocytes isolated from diabetic patients.

3.2. Adiponectin enhances cholesterol efflux from macrophagesfrom diabetic patients

To investigate the mechanism underlying the decrease in lipid-laden macrophage foam cells induced by adiponectin in diabetics,we assessed cholesterol uptake and efflux in macrophages culturedin either adiponectin-free or -supplemented media. Confocalmicroscopy analysis showed that diabetes-derived macrophagescultured in adiponectin-supplemented media did not demonstratethe different levels of Dil-OxLDL cholesterol uptake compared with

Fig. 1. Adiponectin prevents macrophage foam cell formation. (A) Oil red O staining of choAdiponectin-free cells; bottom, adiponectin-treated cells. Arrowheads indicate foam cells.incubated in adiponectin-free or -supplemented media (n ¼ 12 per group) (*P < 0.05 vs adipnondiabetics incubated in adiponectin-free or -supplemented media (n ¼ 12 per group).

macrophages cultured in adiponectin-free media (SupplementalFig. 1A and B).

We next performed cholesterol efflux studies in macrophagesisolated from diabetic patients. Treatment with adiponectinmarkedly increased the cholesterol efflux to mature HDL in mac-rophages from diabetic patients compared with control subjects(Fig. 2A). However, cholesterol efflux to lipid-poor ApoAI wasslightly and insignificantly augmented in response to adiponectintreatment (Fig. 2B). These findings indicate clear differencesbetween control subjects and diabetic subjects in adiponectinregulation of macrophage cholesterol metabolism.

3.3. Adiponectin did not alter the protein expression of lipidsynthesis-related genes

We then investigated the effect of adiponectin on endogenouslipid synthesis-related gene expression. Our data showed that nei-ther the protein expression of SREBP1-targeted genes (acetyl-CoAcarboxylase and fatty acid synthase) (Supplemental Fig. 2A) nor thatof SREBP2etargeted genes (3-hydroxy-3-methylglutaryl coenzymeA reductase [HMGCR] and LDL receptor) (Supplemental Fig. 2B) wasinfluenced by adiponectin treatment, excluding the possibility thatthe adiponectin-induced suppression of intracellular lipid accu-mulation is not likely attribute to the inhibition of endogenous lipidsynthesis.

3.4. Adiponectin upregulates ABCG1 expression but not SR-A, CD36,and SR-BI in macrophages

ABCA1 and ABCG1 play critical roles in cholesterol reversetransport and foam cell formation [27,28]. In diabetes-derivedmacrophages cultured in high and normal glucose, macrophages

lesterol accumulation in monocyte-derived macrophages from diabetic patients. Top,(B) Cholesterol and (C) Cholesterol ester and contents in macrophages from diabeticsonectin free OxLDL-treated cells). (D) Cholesteryl ester formation in macrophages from

Fig. 2. Cholesterol efflux from macrophages of diabetic patients. Peripheral blood monocytes were isolated from control subjects (n ¼ 12) and patients with T2DM (n ¼ 12) anddifferentiated into macrophages with M-CSF for 3 d. Macrophages were then incubated in NBD-cholesterol for 24 h, and cholesterol efflux to lipid-free apoAI or HDL was measuredas indicated. (A) Cholesterol efflux efficiency to HDL. (B) Cholesterol efflux efficiency to lipid-free apoAI. **, P < 0.01 vs. untreated diabetic macrophages (Student’s t test).

M. Wang et al. / Atherosclerosis 229 (2013) 62e7066

treated with adiponectin for 24 h showed 5-fold higher ABCG1mRNA levels (Fig. 3A) after OxLDL stimulation under both glucoseconditions comparedwithmacrophages in adiponectin-free media.In contrast, adiponectin exposure did not exert significant effectson ABCA1 mRNA levels (Fig. 3B). Western blot analysis of ABCG1protein expression was consistent the changes of mRNA levels. Inaddition, adiponectin-treated macrophages showed a 3-fold higherABCG1 protein expression than untreatedmacrophages, but did notsignificantly influence ABCA1 protein expression (Fig. 3C). Theeffects of adiponectin on ABCG1 mRNA and protein were inde-pendent of glucose concentrations. Additionally, in macrophagesfrom nondiabetic controls, adiponectin did not significantly induceABCG1 expression (Supplemental Fig. 3A and B). Furthermore,adiponectin did not evidently influence SR-AI (Supplemental

Fig. 3. Adiponectin induces ABCG1 expression in diabetics. Macrophages from diabetic subjeor normal glucose (100 mg/dl) were used to prepare total RNA and protein. After stimuldetermined by quantitative real-time PCR (n ¼ 10). *, P < 0.05 vs. adiponectin-free macro(Normal glucose, Student’s t test). C, Western blot analysis of ABCG1 and ABCA1 expression

Fig. 4A) or CD36 (Supplemental Fig. 4B) mRNA levels and bothprotein expression (Supplemental Fig. 4C) in diabetes-derivedmacrophages cultured under high or normal glucose.

3.5. Increase of macrophage cholesterol efflux induced byadiponectin is ABCG1 dependent

To further determine the influence of adiponectin on cholesterolaccumulation in macrophages under diabetic conditions, we meas-ured cholesterol uptake and cholesterol efflux after OxLDL stim-ulation inperitonealmacrophages fromdiabeticdb/dbmice culturedin the absence or presence of adiponectin. Cholesterol uptakeanalysis did not show significant difference between adiponectin-treated and untreated diabetic macrophages (Supplemental

cts cultured in adiponectin-free or -supplemented media with high glucose (450 mg/dl)ation with oxLDL for another 6 h. (A) ABCG1 and (B) ABCA1 mRNA expression werephages (High glucose, Student’s t test). #, P < 0.05 vs. adiponectin-free macrophagesnormalized to b-actin (n ¼ 4).

M. Wang et al. / Atherosclerosis 229 (2013) 62e70 67

Fig. 5A). On the contrary, adiponectin exposure enhanced choles-terol efflux to HDL from OxLDL-pretreated macrophages in a dose-dependent fashion compared with adiponectin-free macrophages(Supplemental Fig. 5B). To clarify the role of ABCG1 andABCA1 in theregulation of cholesterol efflux after OxLDL stimulation by adipo-nectin, we transfected macrophages isolated from db/db mice withABCG1 siRNA or ABCA1 siRNA and then cultured in the absence orpresence of adiponectin. Notably, the inductive effect of adiponectinon cholesterol efflux was largely blunted by ABCG1 siRNA but notABCA1 siRNA in macrophages (Supplemental Fig. 5C). Collectively,these data suggest that the adiponectin-induced suppression ofintracellular lipid accumulation is likely due to the augment ofcholesterol efflux and at least partially mediated by ABCG1.

3.6. LXRa activation mediates the anti-foam cell formation effect ofadiponectin

Because ABCG1 and ABCA1 are regulated by the nuclear receptorLXRa [29,30], we further examined the nuclear protein level of LXRain adiponectin-treated macrophages. In diabetic macrophages cul-tured in adiponectin-supplemented media, LXRa expression wasincreased before and after OxLDL stimulation (Fig. 4A). Fur-thermore, in this population, to explore the transcriptional

Fig. 4. LXRa activation is required for adiponectin-mediated ABCG1 expression and attenuatT2DM (n ¼ 12) and differentiated into macrophages with M-CSF for 3 d. (A) Macrophages weOxLDL (100 mg/ml) for 6 h. Western blot analysis of LXRa protein expression before and afterand incubated with adiponectin (20 mg/ml) for 12 h. Then, cells were lysed for Luc activity aOxLDL); ##P < 0.01 vs. control (After OxLDL). (CeE) Macrophages were preincubated witreatment for another 24 h in the presence of 100 mg/ml OxLDL. (C) Cellular lysates wereCholesterol efflux and (E) the intracellular lipid accumulation were measured as indicated.

regulation of LXRa in adiponectin-treated macrophages, LXRaactivation assays were performed by transfecting cells with3 � LXRE-Luc followed by adiponectin treatment in the absence orpresence of OxLDL. Compared with the untreated group, adipo-nectin markedly induced LXRE-mediated luciferase activity by 7.2fold. The increase in LXRE-mediated luciferase activity confirmedthat adiponectinmarkedly induced LXRa activation before and afterOxLDL stimulation compared with macrophages cultured inadiponectin-free media (Fig. 4B). No significant change in LXRaactivation was observed in nondiabetic controls (SupplementalFig. 6A and B). Moreover, coincubation with an LXRa inhibitor(GGPP) abolished the induction of ABCG1 by adiponectin (Fig. 4C),thus abrogating the inhibitory effect of adiponectin on cholesterolefflux (Fig. 4D) and lipid accumulation (Fig. 4E) in macrophagesfrom diabetic patients. These results were further confirmed bygenetic inhibition of LXRaby specific siRNA (Supplement Fig. 7AeC).

Since ABCG1 are also regulated by PPARg, we also evaluated therole of PPARg in the regulation of ABCG1 expression by adiponectin.However, GW9662 did not exert significant influence on ABCG1protein expression (Supplement Fig. 8A) and macrophages cho-lesterol efflux (Supplement Fig. 8B) by adiponectin from diabeticpatients, excluding the participation of PPARg in the effects ofadiponectin. These data suggest that the adiponectin-mediated

ion of lipid accumulation. Peripheral blood monocytes were isolated from patients withre incubated with adiponectin (20 mg/ml) for 24 h before and after the pretreatment ofstimulation with oxLDL. (B) Macrophages were transfected with plasmid 3 � LXRE-Lucssays, and renilla activity was used as an internal control. **P < 0.01 vs. control (Beforeth vehicle (methanol) or GGPP (20 mM) for 2 h, followed by adiponectin (20 mg/ml)subjected to Western blot to determine the protein level of ABCG1 and b-actin. (D)**P < 0.01 vs. adiponectin-treated cells.

M. Wang et al. / Atherosclerosis 229 (2013) 62e7068

upregulation of LXRa is a unifying signaling pathway that inducescholesterol efflux in diabetic patients.

3.7. Activation of adipoR signaling prevents foam cell formation

AdipoR1 and AdipoR2 serve as the major receptors for adipo-nectin and key physiological regulators of glucose and lipidmetabolism in vitro and in vivo [31]. To identify whether theadiponectin-induced effects on cholesterol efflux were AdipoRdependent, we analyzed diabetic monocyte-derived macrophagescultured in adiponectin-supplemented media in the presence ofAdipoR1 siRNA, AdipoR2 siRNA or control siRNA. AdipoR siRNA-transfected macrophages showed an 80% reduction in AdipoRprotein levels compared with control siRNA-transfected macro-phages (Supplemental Fig. 9). Oil Red O staining (Fig. 5A) andcholesterol efflux quantification (Fig. 5B) confirmed that adipo-nectin increased macrophage cholesterol efflux and limited cho-lesterol accumulation from macrophage foam cells only inmacrophages with intact AdipoR signaling pathways; this responsewas blunted in macrophages lacking an AdipoR1 signaling pathwaybut not AdipoR2. In addition, adiponectin upregulated ABCG1expression (Fig. 5C) and LXRa protein expression (Fig. 5D) in thepresence of intact AdipoR signaling, but these effects were reducedin AdipoR1 siRNA-infected macrophages. These data confirm theimportance of the activation of AdipoR signaling in the regulationof both cholesterol transporters and cell signaling pathwaysinvolved in macrophage foam cell formation.

3.8. Serum adiponectin levels are strongly associated withcholesterol efflux capacity in diabetic subjects

Finally, we studied 35 adult subjects with T2DM and 35 adulthealthy volunteers as controls. Sixty percent of patients were ondiabetic oral medications. Both groups were similar with respect toage, gender, total cholesterol, lipid medications, and tobacco use.

Fig. 5. Adiponectin activation of AdipoR signaling increases macrophage cholesterol effludifferentiated into macrophages using M-CSF for 3 d. Then the cells were transfected with esupplemented media (20 mg/ml) for 24 h. (A) Cholesterol accumulationwas assessed by Oil re(C and D) Western blot analysis was performed to determine the ABCG1 (C) and LXRa (D)

Patients with diabetes had laboratory profiles that were typical ofmetabolic syndrome and T2DM, with high levels of triglycerides,low HDL cholesterol, and elevated fasting glucose and hemoglobinA1c. The control population was within normal parametric varia-bles for each category. All of the parametric variables tested werenormally distributed (Supplemental Table 1).

Compared with control healthy subjects, patients with diabeteshad significantly lower levels of not only HDL-C and ApoAI (P< 0.01for each comparison) but also cholesterol efflux capacity (meanvalue for diabetic patients, 0.80; mean value for control subjects,0.96; P ¼ 0.006) (Supplemental Table 1). The HDL-C level was thestrongest predictor of efflux capacity (P < 0.001) but accounted foronly 21% of the observed variation. Male sex and current smokingwere associated with decreased efflux capacity (P ¼ 0.025 andP ¼ 0.048, respectively). Subsequent adjustment for HDL-C largelyattenuated the relationship between sex and cholesterol effluxcapacity (P ¼ 0.114), although smoking remained a significantinverse predictor of efflux capacity (P ¼ 0.002). In the logisticregression analysis adjusted for age, sex, and traditional car-diovascular risk factors, we found that lower adiponectin levelswere strongly correlated with a decrease in cholesterol effluxcapacity. This relationship remained robust after the addition ofHDL-C as a covariate. The results were similar when the ApoAI levelwas substituted for the HDL-C level (Supplemental Table 2). Thus,lower levels serum adiponectin are closely correlated withimpaired cholesterol efflux capacity in diabetic subjects.

4. Discussion

In this study, we first showed that lower plasma adiponectinlevels were strongly correlated with the impairment of cholesterolefflux capacity in patients with diabetes mellitus. Then, we dem-onstrated that activation of adiponectin receptor (AdipoR) signalingprevents foam cell formation by inducing cholesterol efflux inhuman monocyte-derived macrophages from diabetics.

x in diabetics. Peripheral blood monocytes were isolated from diabetic patients andither adipoR1 siRNA, adipoR2 siRNA or control siRNA cultured in adiponectin-free or -d O staining. (B) HDL-mediated cholesterol efflux from oxLDL-pretreated macrophages.protein expression.

M. Wang et al. / Atherosclerosis 229 (2013) 62e70 69

Furthermore, adiponectin upregulates the expression of ABCG1, acritical transporter involved in macrophage cholesterol deposition,via the induction of LXRa activation. The impairment of AdipoR1signaling confirmed the increase in foam cell formation in dia-betics. Taken together, these results suggest that the modulation ofAdipoR signaling is a potential therapeutic target to prevent vas-cular disease progression in diabetes.

T2DM significantly increases the risk for the development ofatherosclerosis, primarily due to the imbalance of cholesterol influx[7] and efflux [11] in macrophages, leading to increased intra-cellular cholesteryl ester accumulation. We first examined thepotential role of adiponectin on lipid deposition in monocyte-derived macrophages from diabetic subjects. Our data showed thatadiponectin indeed ameliorated the OxLDL-induced intracellularlipid accumulation in human macrophages from diabetic patientsbut not normal healthy subjects.

The intracellular lipid homeostasis of foam cells is dynamicallyregulated by OxLDL internalization and cholesterol efflux. Macro-phages from mice lacking SR-A and/or CD36 show reduced OxLDLinternalization and are less prone to foam cell formation in vitro [6].In addition to the scavenger receptors that regulate the influx ofcholesterol in macrophages, several equally important ABC trans-porters are involved in the efflux of cholesterol from cells: ABCA1,ABCG1, and ABCG4. Vaughan and Oram [32] have shown that theseABC transporters act sequentially to eliminate excess cholesterolfrom the cell, with ABCA1 acting first by effluxing cholesterol tolipid-poor ApoAI. ABCG1 can then efflux cholesterol to these newlymade HDL moieties before ABCG4 adds additional cholesterol.However, macrophages do not express ABCG4, leaving ABCA1 andABCG1 as the known primary regulators of cholesterol efflux inmacrophages. Previous studies using macrophages from patientswith T2DM [11] and diabetic mice [33] showed that diabetic mac-rophages have an increased activation of proatherogenic pathwaysbecause of the significant reductions in ABCG1-specific cholesterolefflux. Therefore, we examined the regulatory role of adiponectinand showed that adiponectin augmented both mRNA and proteinexpression of ABCG1 but not ABCA1, SR-A and CD36 in diabeticmacrophages. We also demonstrated that adiponectin treatmentmay affect HDL-mediated cholesterol efflux but not ApoAI-mediated cholesterol efflux or OxLDL uptake during the trans-formation of foam cells. Although adiponectin had been shown toprotect against atherosclerosis by increasing apoAI-mediated cho-lesterol efflux from macrophages through ABCA1-dependentpathway [34], ABCG1 is the primary protein in macrophagesaffected by adiponectin on the setting of diabetes that regulatescholesterol efflux to HDL. Thus, upregulation of ABCG1 in T2DM byadiponectin most likely will be important for the prevention ofatherosclerosis and diabetic vascular complications.

The liver X receptors (LXRs) are nuclear receptors that playcentral roles in the transcriptional control of lipid metabolism [35].Activation of LXRs in lipid-loadedmacrophages leads to induction ofgenes involved in the cholesterol efflux pathway in an attempt toreduce the intracellular cholesterol burden [35]. The ABC trans-porters such as ABCA1 and ABCG1 are downstream target genes ofLXRa and critical for the ability LXRs to enhance efflux to cholesterolacceptors [36]. More important, we also showed that the upregu-lation of ABCG1 by adiponectin was accompanied by increasednuclear LXRa levels and enhanced LXRa transcriptional activity.Moreover, the pharmacological and genetic inhibition of LXRaactivation by GGPP or siRNA abolished the adiponectin-mediatedupregulation of ABCG1. These results suggest that LXRa-mediatedtranscriptional regulation is required for the induction of ABCG1 byadiponectin. Our findings are in agreement with those of Tian [37],who found that adiponectin upregulated ABCA1 and ABCG1 andinhibited the development of atherosclerosis in mice. Despite the

unique pathway discovered in this study, the detailed mechanismsof how adiponectin affects cholesterol efflux merit further study.

We finally examined the correlation between adiponectin andcholesterol efflux capacity in diabetic patients in a cross-sectionalstudy. We used a previously established assay that integrates thepathways known to mediate cholesterol efflux from macrophages(i.e., ABCA1, ABCG1, scavenger receptor B1, and aqueous diffusion)[22]. Our findings demonstrated that only a small part of theobserved relationship between diabetes and cholesterol effluxcapacity was explained by variations in plasma HDL-C and ApoAIlevels. Indeed, adiponectin levels served as a stronger predictor ofcholesterol efflux capacity. The association between serum adipo-nectin levels and cholesterol efflux capacity remained significantafter adjustment for the HDL cholesterol level in diabetic patients inthe regression models. Although cholesterol efflux from macro-phages represents only a small fraction of overall flux through thereverse cholesterol transport pathway [38], a substantial body ofevidence suggests that cholesterol efflux capacity, an integratedmeasure of HDL quantity and quality, is most likely the most rele-vant component of atheroprotection [38e40]. Nevertheless, ourfindings indicate that the impaired cholesterol efflux capacity inpatients with T2DM is largely due to the lower levels ofadiponectin.

In conclusion, the present study reveals a direct mechanistic linkbetween hypoadiponectinemia and the pathophysiology of accel-erated atherosclerosis in diabetes. Our findings suggest that adi-ponectin may potentially be of therapeutic value in inhibiting theprocess of atherosclerosis. Interventional studies are needed toassess the effects of adiponectin status on CVD in diabetic subjects.

Conflicts of interests

The authors have no conflicts of interests.

Sources of funding

This work was supported by grants from the National NaturalScience Foundation of China (81172663), “Thousand, Hundred and,Ten” project of Guangdong Province, and the Foundation for theAuthor of National Excellent Doctoral Dissertation of PR China(200978).

Appendix A. Supplementary data

Supplementary data related to this article can be found athttp://dx.doi.org/10.1016/j.atherosclerosis.2013.01.017.

References

[1] Lee CD, Folsom AR, Pankow JS, Brancati FL. Cardiovascular events in diabeticand nondiabetic adults with or without history of myocardial infarction. Cir-culation 2004;109:855e60.

[2] Mokdad AH, Ford ES, Bowman BA, Dietz WH, Vinicor F, Bales VS. Marks JSPrevalence of obesity, diabetes, and obesity-related health risk factors, 2001.JAMA 2003;289:76e9.

[3] Liang CP, Han S, Senokuchi T, Tall AR. The macrophage at the crossroads ofinsulin resistance and atherosclerosis. Circ Res 2007;100:1546e55.

[4] Lusis AJ. Atherosclerosis. Nature 2000;407:233e41.[5] Glass CK, Witztum JL. Atherosclerosis: the road ahead. Cell 2001;104:503e16.[6] Kunjathoor VV, Febbraio M, Podrez EA, et al. Scavenger receptors class A-I/II

and CD36 are the principal receptors responsible for the uptake of modifiedlow density lipoprotein leading to lipid loading in macrophages. J Biol Chem2002;277:49982e8.

[7] Rahaman SO, Lennon DJ, Febbraio M, Podrez EA, Hazen SL, Silverstein RL.A CD36-dependent signaling cascade is necessary for macrophage foam cellformation. Cell Metab 2006;4:211e21.

[8] Kennedy MA, Barrera GC, Nakamura K, et al. ABCG1 has a critical role inmediating cholesterol efflux to HDL and preventing cellular lipid accumu-lation. Cell Metab 2005;1:121e31.

M. Wang et al. / Atherosclerosis 229 (2013) 62e7070

[9] Wang N, Silver DL, Thiele C, Tall AR. ATP-binding cassette transporter A1(ABCA1) functions as a cholesterol efflux regulatory protein. J Biol Chem 2001;276:23742e7.

[10] Xue JH, Yuan Z, Wu Y, et al. High glucose promotes intracellular lipid accu-mulation in vascular smooth muscle cells by impairing cholesterol influx andefflux balance. Cardiovasc Res 2010;86:141e50.

[11] Mauldin JP, Nagelin MH, Wojcik AJ, et al. Reduced expression of ATP-bindingcassette transporter G1 increases cholesterol accumulation in macrophages ofpatients with type 2 diabetes mellitus. Circulation 2008;117:2785e92.

[12] Stefan N, Vozarova B, Funahashi T, et al. Plasma adiponectin concentration isassociated with skeletal muscle insulin receptor tyrosine phosphorylation,and low plasma concentration precedes a decrease in whole-body insulinsensitivity in humans. Diabetes 2002;51:1884e8.

[13] Ouchi N, Parker JL, Lugus JJ, Walsh K. Adipokines in inflammation and met-abolic disease. Nat Rev Immunol 2011;11:85e97.

[14] Weyer C, Funahashi T, Tanaka S, et al. Hypoadiponectinemia in obesity andtype 2 diabetes: close association with insulin resistance and hyper-insulinemia. J Clin Endocrinol Metab 2001;86:1930e5.

[15] Matsubara M, Maruoka S, Katayose S. Decreased plasma adiponectin con-centrations in women with dyslipidemia. J Clin Endocrinol Metab 2002;87:2764e9.

[16] Ouchi N, Ohishi M, Kihara S, et al. Association of hypoadiponectinemia withimpaired vasoreactivity. Hypertension 2006;42:231e4.

[17] Kumada M, Kihara S, Sumitsuji S, , et alOsaka CAD Study Group 2003 Coronaryartery disease. Association of hypoadiponectinemia with coronary arterydisease in men. Arterioscler Thromb Vasc Biol 2003;23:85e9.

[18] Ouchi N, Kihara S, Arita Y, et al. Novel modulator for endothelial adhesionmolecules: adipocyte-derived plasma protein adiponectin. Circulation 1999;100:2473e6.

[19] Arita Y, Kihara S, Ouchi N, et al. Adipocyte-derived plasma protein adiponectinacts as a platelet-derived growth factor-BB-binding protein and regulatesgrowth factor-induced common postreceptor signal in vascular smoothmuscle cell. Circulation 2002;105:2893e8.

[20] Chen H, Montagnani M, Funahashi T, Shimomura I, Quon MJ. Adiponectinstimulates production of nitric oxide in vascular endothelial cells. J Biol Chem2003;278:45021e6.

[21] Ouchi N, Kihara S, Arita Y, et al. Adipocyte-derived plasma protein, adipo-nectin, suppresses lipid accumulation and class A scavenger receptorexpression in human monocyte-derived macrophages. Circulation 2001;103:1057e63.

[22] Khera AV, Cuchel M, de la Llera-Moya M, et al. Cholesterol efflux capacity,high-density lipoprotein function, and atherosclerosis. N Engl J Med 2011;364:127e35.

[23] McGillicuddy FC, de la Llera Moya M, Hinkle CC, et al. Inflammation impairsreverse cholesterol transport in vivo. Circulation 2009;119:1135e45.

[24] Wang Y, Zhang Y, Wang X, Liu Y, Xia M. Cyanidin-3-O-b-glucoside inducesoxysterol efflux from endothelial cells: role of liver X receptor alpha. Athe-rosclerosis 2012;223:299e305.

[25] Lu KY, Ching LC, Su KH, et al. Erythropoietin suppresses the formation ofmacrophage foam cells: role of liver X receptor alpha. Circulation 2010;121:1828e37.

[26] Oh J, Weng S, Felton SK, et al. 1,25(OH)2 vitamin D inhibits foam cell for-mation and suppresses macrophage cholesterol uptake in patients with type 2diabetes mellitus. Circulation 2009;120:687e98.

[27] Out R, Hoekstra M, Habets K, et al. Combined deletion of macrophage ABCA1and ABCG1 leads to massive lipid accumulation in tissue macrophages anddistinct atherosclerosis at relatively low plasma cholesterol levels. Arte-rioscler Thromb Vasc Biol 2008;28:258e64.

[28] Yvan-Charvet L, Wang N, Tall AR. Role of HDL, ABCA1, and ABCG1 transportersin cholesterol efflux and immune responses. Arterioscler Thromb Vasc Biol2010;30:139e43.

[29] Wang Y, Kurdi-Haidar B, Oram JF. LXR-mediated activation of macrophagestearoyl-CoA desaturase generates unsaturated fatty acids that destabilizeABCA1. J Lipid Res 2004;45:972e80.

[30] Sabol SL, Brewer Jr HB, Santamarina-Fojo S. The human ABCG1 gene: identi-fication of LXR response elements that modulate expression in macrophagesand liver. J Lipid Res 2005;46:2151e67.

[31] Yamauchi T, Nio Y, Maki T, et al. Targeted disruption of AdipoR1 and AdipoR2causes abrogation of adiponectin binding and metabolic actions. Nat Med2007;13:332e9.

[32] Vaughan AM, Oram JF. ABCA1 and ABCG1 or ABCG4 act sequentially toremove cellular cholesterol and generate cholesterol-rich HDL. J Lipid Res2006;47:2433e43.

[33] Mauldin JP, Srinivasan S, Mulya A, et al. Reduction in ABCG1 in Type 2 diabeticmice increases macrophage foam cell formation. J Biol Chem 2006;281:21216e24.

[34] Tsubakio-Yamamoto K, Matsuura F, Koseki M, et al. Adiponectin preventsatherosclerosis by increasing cholesterol efflux from macrophages. BiochemBiophys Res Commun 2008;375:390e4.

[35] Tontonoz P. Transcriptional and posttranscriptional control of cholesterolhomeostasis by liver X receptors. Cold Spring Harb Symp Quant Biol 2011;76:129e37.

[36] Venkateswaran A, Laffitte BA, Joseph SB, et al. Control of cellular cholesterolefflux by the nuclear oxysterol receptor LXR alpha. Proc Natl Acad Sci U S A2000;97:12097e102.

[37] Tian L, Luo N, Klein RL, Chung BH, Garvey WT, Fu Y. Adiponectin reduces lipidaccumulation in macrophage foam cells. Atherosclerosis 2009;202:152e61.

[38] Cuchel M, Rader DJ. Macrophage reverse cholesterol transport: key to theregression of atherosclerosis? Circulation 2006;113:2548e55.

[39] Terasaka N, Wang N, Yvan-Charvet L, Tall AR. High-density lipoprotein pro-tects macrophages from oxidized low-density lipoprotein-induced apoptosisby promoting efflux of 7-ketocholesterol via ABCG1. Proc Natl Acad Sci U S A2007;104:15093e8.

[40] Wang X, Collins HL, Ranalletta M, et al. Macrophage ABCA1 and ABCG1, butnot SR-BI, promote macrophage reverse cholesterol transport in vivo. J ClinInvest 2007;117:2216e24.