Myeloid cell-specific ABCA1 deletion does not worsen insulin resistance in HF diet-induced or genetically obese mouse models.

Obesity-associated low-grade chronic inflammation plays an important role in the development of insulin resistance. The membrane lipid transporter ATP-binding cassette transporter A1 (ABCA1) promotes formation of nascent HDL particles. ABCA1 also dampens macrophage inflammation by reducing cellular membrane cholesterol and lipid raft content. We tested the hypothesis that myeloid-specific ABCA1 deletion may exacerbate insulin resistance by increasing the obesity-associated chronic low-grade inflammation. Myeloid cell-specific ABCA1 knockout (MSKO) and wild-type (WT) mice developed obesity, insulin resistance, mild hypercholesterolemia, and hepatic steatosis to a similar extent with a 45% high-fat (HF) diet feeding or after crossing into the ob/ob background. Resident peritoneal macrophages and stromal vascular cells from obese MSKO mice accumulated significantly more cholesterol. Relative to chow, HF diet markedly induced macrophage infiltration and inflammatory cytokine expression to a similar extent in adipose tissue of WT and MSKO mice. Among pro-inflammatory cytokines examined, only IL-6 was highly upregulated in MSKO-ob/ob versus ob/ob mouse peritoneal macrophages, indicating a nonsignificant effect of myeloid ABCA1 deficiency on obesity-associated chronic inflammation. In conclusion, myeloid-specific ABCA1 deficiency does not exacerbate obesity-associated low-grade chronic inflammation and has minimal impact on the pathogenesis of insulin resistance in both HF diet-induced and genetically obese mouse models.

HF diet containing 45% of energy as lard and 0.015% cholesterol for 16-24 weeks. The HF diet was made by our institutional diet kitchen (see detailed diet composition in supplementary Table I). Body weight (BW) was measured biweekly. To generate MSKO-ob/ ob mice, MSKO mice were fi rst bred with heterozygous ob/ob (ob/+) mice. The resulting double heterozygous MSKO-ob/+ mice were intercrossed to generate ob/+ or MSKO-ob/+ mice, which were then intercrossed to generate ob/ob or MSKO-ob/ob mice, respectively. Ob/ob and MSKO-ob/ob male mice were fed chow for 14-15 weeks for the described studies.

Cell culture
Peritoneal macrophages were harvested from mice by fl ushing the peritoneal cavity with cold PBS. The peritoneal cells were plated in RPMI media containing 100 U/ml penicillin, 100 g/ml streptomycin, and 1% Nutridoma SP media (Roche Applied Science). After a 2 h incubation, fl oating cells were removed by washing with PBS and adherent macrophages were used for experiments.

Plasma lipid, glucose, and insulin analysis
Blood samples were taken from tail veins following a 4 h fast during the light cycle (chow phase) before, and at 2-4 week intervals after initiation of HF diet feeding. Plasma cholesterol (Wako) and TGs (Roche) were determined by enzymatic analysis according to the manufacturer's instructions. Plasma glucose levels were measured using a glucometer (Ascensia Contour, Bayer). Plasma insulin levels were measured by ELISA (Crystal Chem, Inc., Downers Grove, IL). For fed and overnight fasting plasma samples, blood was collected by tail vein at 9:00 AM in ad libitum fed mice (fed) or in mice fasted during the dark cycle for 15 h (5:00 PM-8:00 AM).

Hepatic lipid analysis
Liver lipids were extracted with chloroform:methanol (2:1) and the extract was used for enzymatic assays (cholesterol, Wako; TG, Roche). Data were normalized to liver protein mass, measured by the Lowry protein assay.

Glucose homeostasis analysis
Glucose tolerance tests (GTTs) and insulin tolerance tests (ITTs) were done after 12-20 weeks of HF diet consumption or with ob/ob mice at 13-14 weeks of age. Briefl y, for GTTs, mice were fasted overnight before intraperitoneal (ip) injection of 1 g glucose/kg BW. Blood was collected before and after injection (0, 15, 30, 60, and 120 min) to measure glucose concentrations using a commercial glucose monitor. One week later, the same groups of mice were used for ip injection of 1.5 U of regular human insulin/kg BW (HF diet-fed mice) or 3 U/kg BW (ob/ob mice) after a 5 h fast. Blood glucose concentrations were measured at 0, 15, 30, 60, and 120 min after injection.

Real-time PCR
Total RNA in peritoneal macrophages and white adipose tissue was extracted using Trizol (Invitrogen) and RNeasy lipid tissue kits (Qiagen), respectively, according to the manufacturer's protocols. cDNA preparation and real-time PCR were conducted as described previously ( 24 ).

Immunohistochemistry staining
Epididymal fat tissue sections were incubated with the primary antibodies to CD68 (abD Serotec) or cleaved caspase-3 (Cell Signaling), followed by the biotinylated secondary antibody. The staining was visualized using ABC reagent (ABC vector kit; Vector) and DAB substrate chromogen (Dako). The area of fat sections disorder characterized by near absence of HDL, elevated plasma triglycerides (TGs), reduced low density lipoprotein concentrations, and increased storage of cholesteryl esters (CEs) in macrophages (13)(14)(15). Insulin resistance has not been documented in people with Tangier disease, perhaps because the disease is so rare ( 16 ). However, ABCA1 single nucleotide polymorphisms are associated with altered glucose metabolism and insulin resistance in humans (17)(18)(19)(20). Of interest, leukocyte ABCA1 gene expression is associated with fasting glucose concentrations in normoglycemic men, and reduced leukocyte ABCA1 gene expression is associated with type 2 diabetes ( 21,22 ). To date, only one study has directly linked ABCA1 expression to hyperglycemia, reporting that pancreatic ␤ cell-specifi c ABCA1 deletion results in defective insulin release from islet cells ( 23 ). Whether deletion of ABCA1 expression in other cell types, such as macrophages, alters glucose metabolism is unknown.
Using myeloid cell (macrophage and neutrophil)-specifi c ABCA1 knockout (MSKO) mice, we demonstrated that macrophages from MSKO mice have a signifi cant increase in FC and are more responsive to pro-infl ammatory stimuli [e.g., lipopolysaccharide (LPS)] in vivo and in vitro compared with wild-type (WT) mice ( 24 ). This response was mediated through the TLR and myeloid differentiation primary-response protein 88 (MyD88)-dependent pathway and was independent of alterations in plasma lipid concentrations ( 24 ). Hypersensitivity to LPS is most likely due to increased lipid raft content and increased traffi cking of TLR4 into plasma membrane lipid rafts in MSKO mouse macrophages ( 25 ). Furthermore, MSKO mice challenged with the Listeria monocytogenes cleared the bacterium better than WT mice ( 26 ). These studies clearly demonstrated a regulatory role of myeloid ABCA1 in macrophage infl ammation and innate immunity.
Based on the documented relationship between macrophage infl ammation and insulin resistance, we hypothesized that myeloid-specifi c ABCA1 defi ciency may exacerbate obesity-induced chronic infl ammation and insulin resistance in mice fed a HF diet or crossed into the leptindefi cient ob/ob background. However, MSKO and WT mice developed obesity, adipose infl ammation, and insulin resistance to a similar extent, suggesting that myeloid ABCA1 expression does not signifi cantly worsen obesity-induced chronic infl ammation and insulin resistance.

Animals
WT (ABCA1 +/+ ) and MSKO (ABCA1 Ϫ M/ Ϫ M ) mice were generated as described previously ( 24,25 ). Mice were backcrossed into the C57BL/6 background for six generations before use in these studies and housed in a specifi c pathogen-free facility with a 12 h light/dark cycle. Experiments were conducted in conformity with the Public Health Service Policy on Humane Care and Use of Laboratory Animals, and the experimental protocol was approved by the Wake Forest University Animal Care and Use Committee. At 8 weeks of age, male mice were switched from chow to a only IL-6 was upregulated in macrophages from chow-fed MSKO-ob/ob versus ob/ob mice ( Fig. 1F ). In addition, the expression of TLR4 and its lipid-binding coreceptor CD36 ( 28 ) did not differ between genotypes ( Fig. 1G, H ). These data suggest that despite signifi cant increases in macrophage cholesterol accumulation resulting from ABCA1 defi ciency, HF diet feeding was not suffi cient to stimulate pro-infl ammatory activation of macrophages in vivo.
Myeloid-specifi c ABCA1 defi ciency does not worsen insulin resistance in HF diet-fed mice or chow-fed ob/ob mice Obesity-associated low-grade chronic infl ammation underlies the pathogenesis of type 2 diabetes and insulin resistance ( 29,30 ). Macrophage ABCA1 dampens proinfl ammation via downregulating MyD88-dependent TLR signaling ( 24,25 ). We hypothesized that myeloid-specifi c ABCA1 defi ciency may exacerbate chronic infl ammation and insulin resistance associated with diet-induced or genetically obese mice. To test this hypothesis, we fi rst fed the WT and MSKO male mice a HF diet containing 45% of energy as lard and 0.015% cholesterol. When challenged with the HF diet, WT and MSKO mice gained similar BW over the 24 week diet feeding period ( Fig. 2A ). Liver and fat mass were comparable between genotypes after 24 weeks of diet consumption (supplementary Fig. IA, B).
We then assessed several key parameters of glucose homeostasis in WT and MSO mice. Blood glucose concentrations after a 4 h fast were comparable between both groups of mice ( Fig. 2B ). MSKO mice also exhibited similar plasma glucose clearance (GTT) compared with WT mice ( Fig.  2C ). Consistent with this fi nding, the ability of insulin to lower blood glucose concentrations during ITTs was equivalent in both groups of mice ( Fig. 2D ). We next measured plasma insulin concentrations in overnight-fasted and fed mice; again, no differences were observed between genotypes ( Fig. 2E, F ). Comparable glucose homeostasis between the two genotypes also occurred with prolonged diet feeding (e.g., 12-24 weeks) using different sets of mice. Adding cholesterol (0.2%) to the HF diet did not signifi cantly alter the phenotype (data not shown). Similar outcomes were also observed for HF diet-fed female mice (data not shown). Together, our data suggest that myeloid cell-specifi c ABCA1 deletion does not affect the development of systemic insulin resistance in the presence of HF diet-induced obesity.
Next, we tested our hypothesis using an early-onset genetic obesity mouse model by crossing MSKO mice into the ob/ob background. At 8 and 14 weeks of age, ob/ob and MSKO-ob/ob mice had comparable BWs ( Fig. 3A, B ). No differences in liver or fat mass were observed between genotypes (supplementary Fig. IC, D). A GTT was performed when mice were 13 weeks old, followed by an ITT the next week. Due to the low yield of the MSKO-ob/ob mice, we combined the data from three sets of mice by normalizing data to baseline values. Similar to the HF diet study, we saw no differences in the GTTs and ITTs for ob/ob and MSKO-ob/ob mice ( Fig. 3C, D ), indicating comparable systemic insulin resistance. Similar plasma insulin covered by CD68 + or cleaved caspase-3 + cells was measured using Image-Pro software to quantify the percentage of positive staining cells in adipose tissue.

Stromal vascular cell isolation
Stromal vascular cells were fractioned according to the procedure described by Brown et al. ( 27 ). Briefl y, epididymal fat was minced and enzymatically digested for 45 min with 5 ml/g tissue of collagenase type I (1 mg/ml, Worthington Biochemical Corporation) media. The digestion mixture was fi ltered through a 100 m cell strainer and stromal vascular cells were pelleted by centrifugation (3,000 rpm for 10 min). The stromal vascular fraction (SVF) was then used for fl ow cytometry.

Cholesterol content of macrophages and stromal vascular cells
Macrophages or stromal vascular cells isolated from adipose tissue were extracted with isopropanol (including 5 ␣ -cholestane as internal standard) at room temperature overnight and analyzed for cholesterol content by gas-liquid chromatography ( 27 ).

Statistical analysis
Data are presented as the mean ± SEM unless indicated otherwise. Differences were compared with two-tailed Student's t -test or one-way ANOVA using GraphPad Prism software. P < 0.05 was considered statistically signifi cant.

Macrophages from MSKO versus WT obese mice had signifi cantly more cholesterol accumulation, but comparable infl ammatory cytokine expression
MSKO macrophages have increased plasma membrane FC and lipid rafts, resulting in increased response to TLR4 stimulation (LPS) compared with WT cells ( 24 ). Saturated fatty acids also activate TLR4 and induce chronic infl ammation in the setting of obesity ( 5-7 ). We hypothesized that increased dietary saturated fatty acids (i.e., HF diet), functioning as TLR4 agonists, might increase infl ammation during diet-induced obesity, worsening development of insulin resistance in MSKO versus WT mice. To test this possibility, we fi rst measured cholesterol content in the resident peritoneal macrophages from obese mice. Consistent with our previous fi ndings, macrophages from HF diet-fed MSKO mice ( Fig. 1A, C ) or chow-fed MSKO-ob/ob mice ( Fig. 1B, D ) had signifi cantly more FC and CE accumulation compared with their WT counterparts. However, similar infl ammatory cytokine gene expression was observed for macrophages from 24 week HF diet-fed WT and MSKO mice ( Fig. 1E ). Among the examined cytokines, WT fat had signifi cantly lower F4/80 expression and a trend toward less CD68 expression. Furthermore, immunohistochemistry staining of macrophages using CD68 antibody revealed that CD68 + cells in MSKO adipose tissues were greatly reduced compared with WT ( Fig. 4B ), indicating signifi cantly less macrophage infi ltration into fat. The comparable level of cleaved caspase-3 in adipose tissues (supplementary Fig. IIA, B) rules out a major role for apoptosis in the differential accumulation of macrophages between genotypes. Interestingly, this difference was not observed with a longer period of HF diet feeding (24 weeks, supplementary Fig. IIIA). concentrations ( Fig. 3E, F ) further confi rmed that insulin resistance did not differ between genotypes.

Myeloid-specifi c ABCA1 defi ciency does not worsen adipose tissue infl ammation
We next examined the infl ammatory status in fat, an important insulin target tissue. We fi rst measured expression of F4/80 and CD68 (macrophage markers) in epididymal fat from HF diet-fed mice (17 weeks). Compared with chow, the HF diet signifi cantly increased F4/80 and CD68 expression in adipose tissue, indicating macro phage infi ltration into fat tissue ( Fig. 3A ). Unexpectedly, MSKO versus MSKO-ob/ob versus ob/ob mice had similar levels of IL-1 ␤ , TNF-␣ , and MCP-1 mRNA expression and a trend toward decrease in IL-6 and IL-12p40 mRNA expression in fat tissue ( Fig. 5B ). Note that similar to peritoneal macrophages, stromal vascular cells from MSKO-ob/ob fat had signifi cantly higher FC (23% increase) and CE (254% increase) accumulation relative to their WT counterparts ( Fig. 5C ). Thus, our data suggest that myeloid cell-specifi c ABCA1 deletion enhances cholesterol accumulation in adipose macrophages; however, it does not enhance obesity-induced macrophage infi ltration or chronic infl ammation in adipose tissue.

Myeloid cell-specifi c ABCA1 defi ciency does not alter plasma and liver lipid profi les in HF diet-induced or genetically-induced obese mice
To determine whether ABCA1 defi ciency in myeloid cells affects lipid homeostasis in HF diet-fed or chow-fed ob/ob obese mice, we examined the plasma and liver lipid concentrations. There was no genotypic difference in plasma total cholesterol (TC) and TG concentrations during the 24 weeks of HF diet feeding ( Fig. 6A, C ) or after mice were crossed into the ob/ob background ( Fig. 6B, D ). We also saw comparable levels of hepatic TC ( Fig. 6E, F ), FC, Obesity induces a phenotypic switch from M2-polarized state (alternatively activated) to an M1 (classically activated) pro-infl ammatory state in adipose macrophages ( 31 ). We examined expression of the M1 type of infl ammatory cytokine/chemokine and M2 macrophage markers in fat. Feeding the HF diet for 17 weeks signifi cantly induced pro-infl ammatory cytokine (TNF-␣ ) or chemokine (MCP-1) expression in adipose tissues compared with chow ( Fig. 4C ). However, no genotypic differences were observed between the chow-fed or HF diet-fed mice ( Fig.  4C ). Not surprisingly, the expression of M2 macrophage markers did not show any difference between genotypes ( Fig. 4D ). Again, no signifi cant difference was observed in TLR4 or CD36 expression between genotypes (supplementary Fig. IIIB ) . Increasing the duration of HF diet feeding from 17 to 24 weeks did not signifi cantly increase adipose infl ammation in MSKO versus WT mice (supplementary Fig. IIIC).
Macrophages play a central role in HF diet-induced obesity and subsequent development of systemic insulin resistance through secretion of pro-infl ammatory cytokines that act on local tissues or circulate to distal tissues. The infl ammatory cytokines then block insulin receptor signaling by serine phosphorylation of insulin-like substrate, interfering with its ability to engage in insulin receptor signaling ( 32 ). During development of diet-induced obesity, saturated fatty acids can activate TLR4, triggering a lowgrade infl ammatory state in macrophages and other cell types, such as adipose tissue and muscle ( 7,8 ). Deletion of TLR4 in bone marrow-derived cells blunts the macro phage and adipose tissue response to dietary fatty acids and protects mice from HF diet-induced insulin resistance ( 8 ). Deletion of hematopoietic cell JNK1 (a key component of the TLR4 signaling pathway) attenuates infl ammation in metabolic tissues and improves insulin sensitivity in HF diet-fed mice ( 9 ). Moreover, myeloid-specifi c IKK ␤ deficiency improves systemic insulin sensitivity in mice on a HF diet ( 10 ), suggesting a causative link between increased macrophage infl ammation and insulin resistance.
Our previous studies reported that myeloid-specifi c ABCA1 defi ciency exacerbates macrophage infl ammatory response to a moderate LPS dose in vivo (3 mg/kg) and in vitro (100 ng/ml), and this response was mediated through a MyD88-dependent TLR4 pathway ( 24 ). Unstimulated resting macrophages were indistinguishable between WT and CE (data not shown) accumulation between HF diet-fed WT and MSKO mice. WT and MSKO mice developed hepatic steatosis to a similar extent when fed a HF diet for 24 weeks ( Fig. 6G ) or crossed into the ob/ob background ( Fig. 6H ), respectively. Together, these data suggest that myeloid cell-specifi c ABCA1 deletion does not signifi cantly affect plasma and liver lipid metabolism in the setting of obesity-induced insulin resistance.

DISCUSSION
Obesity and insulin resistance are chronic low-grade infl ammatory diseases associated with macrophage accumulation and activation in adipose tissue ( 29,30 ). We previously demonstrated that myeloid cell-specifi c deletion of ABCA1 results in increased macrophage infl ammatory response to TLR agonists ( 24 ), and human studies suggest an association between ABCA1 single nucleotide polymorphisms, leukocyte ABCA1 expression, and insulin resistance (17)(18)(19)(20)(21)(22). However, whether there is a link between macrophage ABCA1-regulated innate immunity and insulin resistance is unknown. In the present study, HF diet-fed MSKO mice and chow-fed MSKO-ob/ob mice did not exhibit exacerbated obesity-associated chronic infl ammation or insulin resistance compared with their WT counterparts, despite a signifi cant increase in cholesterol content in both resident peritoneal macrophages and adipose stromal vascular cells that accompanies ABCA1 deletion. Our results suggest that increased dietary saturated fatty acid intake (i.e., HF Fig. 3. Myeloid-specifi c ABCA1 defi ciency does not worsen insulin resistance in male ob/ob mice. MSKO mice were crossed with ob/ob mice to induce obesity. A, B: BW was measured at 8 and 14 weeks of age. GTTs (C) and ITTs (D) were performed at 13 and 14 weeks of age, respectively. Plasma insulin levels in fed (E) or overnight-fasted (F) mice were measured at 13 weeks of age.
would exacerbate obesity-associated chronic infl ammation, worsening insulin resistance. However, MSKO and WT mice developed obesity, insulin resistance, and mild hepatic steatosis to a similar extent with HF diet feeding. Relative to chow, HF diet feeding markedly induced macrophage infi ltration into adipose tissue and infl ammatory cytokine expression. However, we did not observe enhanced macrophage infi ltration and adipose infl ammation in MSKO versus WT mice after a HF diet feeding. Similar data were obtained with chow-fed MSKO mice crossed into the ob/ob background. Consistent with our and MSKO mice, except for a small ( ‫ف‬ 10%) but signifi cant increase in cellular FC. Furthermore, MSKO mice cleared infection with L. monocytogenes more effi ciently than WT mice ( 26 ). Collectively, these studies clearly establish a regulatory role for myeloid ABCA1 in acute macrophage infl ammation and innate immunity. In addition, leukocyte ABCA1 expression protects against atherosclerosis, a chronic disease with well-characterized infl ammatory components (33)(34)(35). Based on the causative association between macrophage infl ammation and insulin resistance, we hypothesized that myeloid-specifi c ABCA1 defi ciency  ( 7 ). Second, previous studies demonstrating TLR4-specifi c activation of infl ammation by HF diets used mice lacking TLR4 globally ( 7 ) or in bone marrow cells ( 8 ), leaving open the possibility that cells other than macrophages could have contributed to the exacerbated infl ammation and systemic insulin resistance. In fact, to our knowledge, there are only two examples in which myeloid-specifi c deletion of genes (IKK-␤ or JNK) involved in canonical infl ammatory signaling pathways resulted in HF-induced systemic insulin resistance ( 10,36 ). However, IKK-␤ and JNK are directly involved in infl ammatory signaling, whereas ABCA1 indirectly affects TLR signaling by increasing membrane lipid raft content, resulting in more TLR4 in lipid rafts ( 25 ). In addition, the concept that saturated fatty acids activate TLR4 is not universally accepted. previous fi ndings, macrophages from HF diet-fed MSKO versus WT mice accumulated signifi cantly more cholesterol, but cytokine and chemokine gene expression was similar to WT macrophages. Therefore, myeloid-specifi c ABCA1 defi ciency impairs cholesterol effl ux, leading to intracellular cholesterol accumulation, but does not exacerbate obesityassociated low-grade chronic infl ammation and does not lead to insulin resistance.
Our study shows that absence of myeloid cell ABCA1 expression is not sufficient to mount an exaggerated infl ammatory response in the setting of obesity and does not result in insulin resistance. In our previous study, an increased macrophage infl ammatory response in ABCA1-deficient versus WT macrophages was triggered by a moderate dose of LPS in vivo or in vitro, and not observed in nonstimulated macrophages ( 24 ). There may be several explanations for the lack of an exaggerated infl ammatory response with HF diet feeding in our current study. First, the amount of saturated fatty acid necessary to differentially activate an infl ammatory response in macrophages with and without ABCA1 expression may have been   6. Myeloid-specifi c ABCA1 defi ciency does not alter plasma and liver lipid homeostasis in the setting of obesity. A, C: Blood samples from HF diet-fed mice were collected periodically to assay total cholesterol (TC) and TGs by enzymatic assays. B, D: Blood samples were collected from 14-week-old MSKO-ob/ob and ob/ob mice to assay TC and TGs by enzymatic assays. E-H: After 24 weeks HF diet consumption or at the age of 14 weeks for ob/ob mice, mice were euthanized and liver lipids were quantifi ed by enzymatic assay and normalized to liver protein content. +/+, WT; Ϫ M/ Ϫ M, MSKO.
For example, functional TLR4 was necessary to attenuate trans fatty acid-induced obesity, hyperglycemia, and hyperinsulinemia ( 37 ). Similarly, mice lacking MyD88, a TLR4 signaling adaptor protein, were more susceptible to dietinduced insulin resistance, suggesting a protective role for MyD88 signaling in obesity-associated insulin resistance ( 38 ). Finally, in vitro studies suggest that TLR4 activation by fatty acids can potentially be confounded by LPS and lipopeptide contamination ( 39 ).
In summary, myeloid-specifi c ABCA1 defi ciency impairs cholesterol effl ux, but does not signifi cantly exacerbate macrophage infi ltration and low-grade infl ammation in adipose tissue, or insulin resistance in both HF diet-induced and genetically obese mouse models. However, because we examined advanced obesity resulting from 17 weeks of HF diet feeding, we cannot rule out the possibility that myeloid-specifi c ABCA1 defi ciency might worsen adipose infl ammation or metabolic abnormalities in the context of less advanced obesity.