Inflammatory stimuli induce acyl-CoA thioesterase 7 and remodeling of phospholipids containing unsaturated long (≥C20)-acyl chains in macrophages[S]

Acyl-CoA thioesterase 7 (ACOT7) is an intracellular enzyme that converts acyl-CoAs to FFAs. ACOT7 is induced by lipopolysaccharide (LPS); thus, we investigated downstream effects of LPS-induced induction of ACOT7 and its role in inflammatory settings in myeloid cells. Enzymatic thioesterase activity assays in WT and ACOT7-deficient macrophage lysates indicated that endogenous ACOT7 contributes a significant fraction of total acyl-CoA thioesterase activity toward C20:4-, C20:5-, and C22:6-CoA, but contributes little activity toward shorter acyl-CoA species. Lipidomic analyses revealed that LPS causes a dramatic increase, primarily in bis(monoacylglycero)phosphate species containing long (≥C20) polyunsaturated acyl-chains in macrophages, and that the limited effect observed by ACOT7 deficiency is restricted to glycerophospholipids containing 20-carbon unsaturated acyl-chains. Furthermore, ACOT7 deficiency did not detectably alter the ability of LPS to induce cytokines or prostaglandin E2 production in macrophages. Consistently, although ACOT7 was induced in macrophages from diabetic mice, hematopoietic ACOT7 deficiency did not alter the stimulatory effect of diabetes on systemic inflammation or atherosclerosis in LDL receptor-deficient mice. Thus, inflammatory stimuli induce ACOT7 and remodeling of phospholipids containing unsaturated long (≥C20)-acyl chains in macrophages, and, although ACOT7 has preferential thioesterase activity toward these lipid species, loss of ACOT7 has no major detrimental effect on macrophage inflammatory phenotypes.≥

a concomitant increase in PPAR- activation (1). The antiinflammatory effects associated with loss of FABP4 in macrophages have subsequently been shown to be due to increased expression of uncoupling protein 2 (4) and sirtuin 3 (6), likely because of increased levels of free intracellular MUFAs. Consistently, inhibition of FABP4 prevents atherosclerosis in mice (2). Similar results have been found by KO of FABP5 (3). Furthermore, long-chain acyl-CoA synthetase 1 (ACSL1), an enzyme that converts FFAs into their acyl-CoA derivatives, is induced by Toll-like receptor 4 (TLR4) activation in macrophages, and its deletion in myeloid cells results in protection against early atherosclerosis in diabetic mice (5). More recently, elegant studies have demonstrated that unsaturated FAs (including omega-6 and -3 FAs) are increased in the late phase (12-24 h) of TLR4 activation in macrophages, concomitant with induction of enzymes involved in MUFA and PUFA metabolism, and that SREBP1 regulates this late resolution phase of the inflammatory response (6).
We therefore investigated the role of unsaturated FA handling in regulation of inflammation and atherosclerosis by studying one of the acyl-CoA thioesterases (ACOTs) induced by TLR4 activation by lipopolysaccharide (LPS) in macrophages, ACOT7. ACOTs hydrolyze the thioester bond on acyl-CoAs, but their cellular functions have just recently begun to be elucidated. There are 15 known ACOTs in mice, each with differential tissue and intracellular localization, transcriptional regulation, and acyl-CoA substrate preference (7)(8)(9)(10). Only a few studies to date have investigated the function of ACOTs in vivo (11)(12)(13)(14)(15)(16)(17)(18). These studies have revealed that ACOTs are important mediators of FA and acyl-CoA channeling into specific fates, and that the processes mediated by ACOTs appear to require rapid and dynamic interconversion between FFAs and acyl-CoAs. Thus, ACOT2 promotes hepatic FA oxidation, which requires acyl-CoA formation (13). ACOT13 deficiency protects against hepatic triglyceride accumulation in fat-fed mice and increases energy expenditure after cold exposure (14,15), and likewise, ACOT11 deficiency protects against obesity by increasing energy expenditure (16,18). ACOT15 is involved in cardiolipin remodeling (11).
ACOT7 is highly expressed in neurons (19,20), and a neuron-specific ACOT7-deficient mouse exhibits elevated levels of lysophospholipid and sphingosine in the brain (12). ACOT7 has been shown to have preferential activity toward arachidonoyl-CoA (20:4-CoA) over saturated acyl-CoAs in macrophages (21), but also to act on other long-chain acyl-CoAs (21,22). Forced expression of an active ACOT7 mutant has been found to suppress prostaglandin D 2 (PGD 2 ) and PGE 2 production under basal conditions in a macrophage cell line (21). Overexpression of ACOT7 in pancreatic  cells lowered acyl-CoA levels, consistent with what one would predict based on the enzymatic activity of ACOT7 (17). On the contrary, ACOT7 deficiency results in markedly increased levels of free unsaturated FAs in neurons (12), suggesting that compensation may exist for loss of ACOT7 in some cell types.
We investigated the role of ACOT7 in two models of inflammation in which altered FA handling in macrophages is thought to contribute to the inflammatory phenotype: LPS stimulation of macrophages in vitro and an in vivo mouse model of diabetes. Our results demonstrate that ACOT7 is induced by TLR4 through its adaptor protein MyD88 and by diabetes in macrophages and that classical Ly6C high monocytes exhibit elevated ACOT7 gene expression, as compared with Ly6C low monocytes. We demonstrate that LPS causes a marked shift in acyl-chain composition of bis(monoacylglycero)phosphate (BMP), a phospholipid present in late endosomes and lysosomes (23), from shorter saturated and monounsaturated BMPs to BMPs containing long-chain (C20 and C22) unsaturated acyl-chains in the resolution phase of the inflammatory response. Furthermore, ACOT7 contributes a significant fraction of total ACOT activity toward arachidonoyl-CoA (C20:4-CoA), docosahexaenoyl-CoA (C22:6-CoA), and eicosapentaenoyl-CoA (C20:5-CoA) in macrophages, and the limited effect of ACOT7 deficiency on glycerophospholipids is restricted to species containing 20-carbon unsaturated acyl-chains. However, loss of ACOT7 is not sufficient to counter the striking effect of LPS on BMP acylchain composition or to alter the inflammatory response in macrophages.

Mice
All mice were on the C57BL/6 background. Acot7 / mice were generated by crossing Acot7 fl/fl mice (12) to mice expressing Cre recombinase from the germline (Jax 006054) (24). Mice carrying a deleted allele of Acot7 were then bred to homozygosity in the absence of Cre recombinase. For additional lipidomics, Acot7 fl/fl mice were crossed with Lys2 Cre/Cre mice to generate bone marrowderived macrophages (BMDMs) lacking ACOT7 with Acot7 WT/WT ; Lys2 Cre/Cre littermates used as controls. Female LDL receptordeficient (Ldlr / ) mice for atherosclerosis studies were obtained from Jackson Labs (Bar Harbor, ME). Bones from Tlr4 / mice were a generous gift from Dr. Linda Curtiss, the Scripps Research Institute, and bones from Myd88 / mice and TRIF-deficient (Ticam1 / ) mice were generously provided by Dr. Alan Aderem, Seattle Biomedical Research Institute, and Dr. Kelly D. Smith, University of Washington, Seattle. Bones from WT littermates were used as controls. All studies were approved by the Animal Care and Use Committee of the University of Washington. Mice were housed (maximum five per cage) in a specific pathogen-free barrier facility with a 12 h light/dark cycle with free access to food and water.

Isolation and maintenance of BMDMs
BMDMs were harvested from freshly obtained bones or from bones shipped overnight on wet ice by syringe flushing with DMEM supplemented with 1% fungizone and 1% penicillin/ streptomycin. Cells were purified of erythrocytes and maintained in DMEM (450 mg/dl glucose) containing 1% fungizone, 1% penicillin/streptomycin, 7% FBS, and 30% L-cell conditioned medium, as a source of macrophage colony-stimulating factor. For measurements of inflammatory responses, cells were stimulated with ultrapure LPS (Escherichia. coli 0111:B4, 10 or 0.1 ng/ml; List Biological Laboratories, Campbell, CA). Alternative activation was achieved by 24 h treatment with IL-4 (10 ng/ml; eBioscience; catalog no. 14-8041).

Knockdown of ACOT7 in thioglycollate elicited macrophages
For in vitro ACOT7 knockdown experiments, macrophages were harvested from the peritoneal cavity of C57BL/6 mice 5 days after injection of thioglycollate (5). Before plating the macrophages, Acot7 siRNA (Life Technologies, Carlsbad, CA; catalog no. 4390771; i.d. s88465 and s211904) or negative control #2 siRNA (Life Technologies; catalog no. 4390846) was introduced by electroporation (Amaxa mouse macrophage nucleofector kit; Lonza, Cologne, Germany). The samples were electroporated in the Amaxa Nucleofector I device by using program Y-01, as described previously (25). A total of 3 × 10 6 cells/cuvette were electroporated and immediately plated in medium supplemented with 10% FBS, and medium and floating cells were removed 1 h later for adherence purification of macrophages. All experiments were performed by using RPMI 1640 medium (11.2 mM glucose) supplemented with 1% penicillin/streptomycin. For measurements of inflammatory responses, cells were stimulated with ultrapure LPS (E. coli 0111:B4, 10 ng/ml; List Biological Laboratories) for 48 h after electroporation, as 90% knockdown of ACOT7 protein was achieved at this time point.

Determination of thioesterase activity and acyl-CoA species
Thioesterase enzymatic activity was determined in lysates of BMDMs from mice with hematopoietic ACOT7 deficiency or BMDMs overexpressing ACOT7 (by transduction with an Acot7 ecotropic retrovirus). Cells were allowed to differentiate in L-conditioned medium for 7 days before the experiment and then harvested in Western lysis buffer containing protease (Santa Cruz Biotechnology, Dallas, TX, catalog no. sc29130) and phosphatase inhibitors. Samples were frozen before the enzymatic assay at 80°C without degradation. For the assays, 12.5-25 g of protein lysate and DTNB (Cayman Chemical) were incubated with 150 nmol of various acyl-CoAs [Sigma-Aldrich (16:0-CoA, 18:1-CoA, 20:4-CoA) and Avanti Polar Lipids, Alabaster, AL (20:5-CoA 22:6-CoA)] in buffer containing: 50 mM KCl, 10 mM HEPES (pH 7.2), 0.025% Triton X-100, and 0.2 mg/ml FA-free BSA. The assays were performed under V max conditions. Free CoA reacted with DTNB and was detected by measuring absorbance at 412 nm. Quantification was performed by using a standard curve of known CoA concentration (Sigma-Aldrich). In some experiments, ACOT7 was immunoprecipitated before the thioesterase activity assay. ACOT7 was immunoprecipitated by using a rabbit polyclonal anti-ACOT7 (ab8515; 1:1,000, Abcam). A nonspecific rabbit IgG was used for control immunoprecipitations. Briefly, cell lysates at a concentration of 1 mg of protein in 0.5 ml were incubated with 3 g of ACOT7 antibody or IgG overnight. Target proteins bound by the antibodies were immunoprecipitated by the addition of protein G-agarose beads (Protein G-agarose Immunoprecipitation Kit, Roche Diagnostics, Basel, Switzerland) and incubation for 3 h. Antibody-conjugated beads were washed several times according to the manufacturer's protocol before the thioesterase assay. Each sample absorbance was normalized to its no substrate control before quantification of CoA released. All reactions were performed within the linear range. Measurements of acyl-CoA molecular species were performed by LC/ESI-MS/MS, as described previously (26).

Analysis of phospholipids by LC-MS/MS
WT and ACOT7-deficient BMDMs were stimulated with LPS (10 ng/ml) for 24 h. The cells were then quickly harvested in PBS and frozen at 80°C. Phospholipids were extracted from the cell pellets and analyzed by the Mass Spectrometry Lipidomics Core facility at the University of Colorado essentially as described (27) Table S1 contains a complete list of the m/z transitions used. Results were analyzed by using MultiQuant software (SCIEX) and are reported as the ratio between the integrated area of each analyte and the integrated area of the corresponding internal standard for each class.

Generation of nondiabetic and diabetic mice lacking ACOT7 in bone marrow cells for atherosclerosis study
Bone marrow was harvested from ACOT7-deficient (ACOT7 KO) mice or WT controls (12), purified of erythrocytes, and transplanted into lethally irradiated (10 Gy) female Ldlr / recipients (5 × 10 6 cells injected retroorbitally). The mice were maintained on antibiotic (2 mg/ml neomycin) water for 2 weeks after the transplant and allowed to recover for a total of 7 weeks before streptozotocin (STZ) injections. STZ (mixed anomers no. S0130, 50 mg/kg; Sigma-Aldrich) was dissolved in freshly made citrate buffer (0.1 M, pH 4.5) and injected intraperitoneally for five consecutive days. The mice were monitored for development of diabetes (defined as blood glucose >250 mg/dl). If mice did not respond to the STZ treatment 14 days after the first injection, they were reinjected for an additional round of 5 days. In the present study, 37/40 mice exhibited blood glucose values >250 mg/dl. Blood glucose was monitored throughout the study by using test strips (One Touch Ultra, LifeScan, Milpitas, CA), and diabetic mice received insulin (Lantus, Sanofi-Aventis, Bridgewater, NJ) as needed to prevent excessive weight loss and ketonuria. After induction of diabetes, the mice were fed a semipurified low-fat diet for the duration of the 12-week study. This diet has been described previously (28) and is used because diabetic mice do not exhibit hypercholesterolemia as compared with nondiabetic controls when fed this diet. Blood cholesterol was measured by using test strips (Cardiochek, Indianapolis, IN). Plasma lipids were determined by colorimetric assays according to manufacturers' instructions: triglycerides (Sigma-Aldrich), NEFAs (Wako, Richmond, VA), and cholesterol (Wako). For mRNA measurements in leukocytes, mRNA was extracted from blood leukocytes according to manufacturer's instructions (Macherey Nagel; Nucleospin RNA kit). Aortas were dissected longitudinally and stained en face with Sudan IV for visualization of aortic lesions (28). Lesions were quantified in a blinded manner as percent area of the entire aortic area.

Statistical analysis
Statistical analyses for all studies were performed by using GraphPad Prism 5 software (GraphPad, La Jolla, CA). Unpaired two-tailed Student's t-test was used to compare two conditions, whereas multiple groups were compared by one-way ANOVA with a Tukey post hoc test or two-way ANOVA. Enzymatic activity assays were assessed by repeated-measures two-way ANOVA. For lipidomic data, volcano plots were generated to identify patterns of changes and followed up by one-way ANOVA for selected lipid species meeting nominal statistical significance and at least a 2-fold difference in the volcano plots. Error bars indicate SEM. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

ACOT7 is induced in macrophages through TLR4/MyD88 and is elevated in Ly6C high monocytes as compared with Ly6C low monocytes
Macrophages can become classically activated and promote inflammation or can become alternatively activated with wound-healing functions and antiinflammatory properties. The bacterial component LPS induces classical activation of macrophages, and ACOT7 has been previously shown to be induced in these cells by LPS (21). We first verified that in mouse thioglycollate-elicited macrophages and BMDMs Acot7 mRNA and ACOT7 protein are increased after treatment with LPS ( Fig. 1A-D). Acot7 mRNA levels peaked at 6 h after stimulation and then declined (Fig. 1A). For comparison, Acsl1 was more robustly induced by LPS (Fig. 1A). Furthermore, a significant increase in ACOT7 protein levels was observed up to 48 h after LPS stimulation (Fig. 1B). Similar results were obtained in BMDMs stimulated by LPS/IFN (data not shown). Moreover, ACOT7-specific CoA thioesterase activity, measured in ACOT7 immunoprecipitates, was increased by 3.1 ± 0.4-fold in BMDMs stimulated with LPS/IFN as compared with IL-4 for 48 h (P = 0.01; n = 4), reflective of the increase in ACOT7 protein after LPS/IFN polarization. Thus, ACOT7 mRNA and protein are increased after LPS stimulation in macrophages, leading to greater ACOT7-specific thioesterase activity in these cells.
LPS is a ligand of TLR4, which, when activated, signals through two adaptor proteins: Toll/Il-1 receptor domain containing adaptor inducing IFN- (TRIF; gene name Ticam1) and MyD88. We investigated the adaptor protein responsible for LPS-induced Acot7 by taking advantage of BMDMs from Tlr4 / and Ticam1 / mice (Fig. 1C). BMDMs isolated from Tlr4 / mice exhibited no increase in Acot7 mRNA levels after LPS stimulation, as expected, whereas BMDMs from Ticam1 / mice had normal, or even elevated, induction of Acot7 mRNA after LPS stimulation. Therefore, regulation of ACOT7 after LPS treatment in macrophages is entirely dependent on TLR4 signaling, but does not require the TRIF adaptor protein. In an additional experiment using BMDMs from Myd88 / mice, we demonstrated that induction of Acot7 mRNA requires MyD88 (Fig. 1D).
Ly6C high monocytes acutely traffic to regions of inflammation via CCR2 signaling and differentiate into macrophages (29). Flow cytometry was used to sort blood leukocytes into Ly6C high monocytes, Ly6C low monocytes, and neutrophils, and Acot7 mRNA levels were determined in each of these populations. Acot7 mRNA was elevated in the Ly6C high monocyte population, as compared with the Ly6C low monocyte population, suggesting a greater role for the enzyme in the Ly6C high population (Fig. 1E). Because Acot7 mRNA expression is downstream of TLR4 activation, Tlr4 mRNA was assessed as well (Fig. 1F). Tlr4 mRNA levels did not follow the same expression pattern as Acot7 mRNA, suggesting that ACOT7 regulation in myeloid cell populations may be influenced by multiple factors in vivo.

LPS stimulation results in a marked shift in BMP composition in the resolution phase of the inflammatory response
To investigate the effects of LPS stimulation at a time point at which ACOT7 is induced, we performed a lipidomic analysis of phospholipid molecular species in BMDMs under basal conditions and after a 24 h stimulation with LPS. A large majority of phospholipids altered in response to LPS were identified as BMPs. Interestingly, there was a dramatic shift in acyl-chain composition of BMPs in response to LPS, such that BMPs containing saturated and monounsaturated C14-C18 acyl-chains were reduced, and BMPs containing long C20-C22 polyunsaturated acylchains were enriched by LPS (Fig. 3A, D-H).

ACOT7 deficiency has a marginal effect on glycerophospholipids containing 20 carbon unsaturated acyl-chains
TLR4 activation profoundly remodels the lipidome in macrophages (6,30), consistent with the results of the present study. Our lipidomic screen was conducted to assess changes in phospholipid species in macrophages with ACOT7 deficiency. Whereas LPS caused marked alterations in both WT and ACOT7-deficient macrophages, primarily in BMPs (Fig. 3A, B), ACOT7 deficiency did not have a striking  Tables S2-S6). Interestingly, ACOT7 deficiency did not significantly alter arachidonic acid-containing phospholipid species, despite its known action on 20:4-CoA (21). Less than a handful of phospholipids were increased by ACOT7 deficiency under basal conditions (Fig. 3C), and no effect of ACOT7 deficiency was detected in LPS-stimulated cells (supplemental Tables S2-S6). The few phospholipids nominally increased by ACOT7 deficiency all contained unsaturated 20-carbon acyl-chains (Fig. 3C, I).

Loss of ACOT7 does not reduce LPS-induced cytokine production, nor does ACOT7 expression influence PGE 2 secretion
We next investigated whether ACOT7 could modulate the overall inflammatory status of macrophages stimulated with LPS. BMDMs from ACOT7-deficient macrophages had similar levels of LPS-induced Tnfa, Il6, or Ccl2 mRNA at both 4 h (Fig. 4A-C) and 24 h (Fig. 4D-F) after stimulation. Secretion of IL-6 or TNF-, as measured 18-24 h after LPS stimulation, was also unchanged by loss of ACOT7 (Fig. 4G, H). PGE 2 release was measured to assess the effect of ACOT7 loss on eicosanoid mediators of inflammation. Surprisingly, based on previous results on forced expression of active ACOT7 (21), and despite the substrate preference of ACOT7 for 20:4-CoA, PGE 2 release was unchanged by ACOT7 deficiency (Fig. 4I).
We also investigated cytokine expression in LPSstimulated J774 macrophages overexpressing ACOT7 at high levels. Overexpression of ACOT7 resulted in increased secretion of IL-1, TNF-, and IL-6 after LPS stimulation (supplemental Fig. S2A-C). The mechanism for this increase in inflammation is unknown, but likely does not involve eicosanoid signaling because PGE 2 was not affected by ACOT7 overexpression, either under basal conditions or after LPS stimulation (supplemental Fig. S2D), nor was the effect ACOT7 on LPS-stimulated IL-6 release blocked by cyclooxygenase 2 and 5-lipoxygenase inhibitors (CAY 10404 and CJ-13,610, respectively) (supplemental Fig. S2E). It is likely that high levels of forced ACOT7 Fig. 4. ACOT7 deficiency does not affect LPS-induced inflammatory activation. ACOT7-deficient BMDMs were used to evaluate the role of ACOT7 on cytokine production, whereas PGE 2 secretion was analyzed in BMDMs in which ACOT7 had been knocked down by siRNA (and BMDMs treated with an siRNA control). The BMDMs were stimulated with 10 ng/ml LPS for 4 or 24 h or left unstimulated. Tnfa (A, D), Il6 (B, E) ,and Ccl2 (C, F) mRNA levels were measured by real-time PCR. Secretion of TNF- (G), IL-6 (H), and PGE 2 (I) was measured by ELISA 18-24 h after LPS stimulation. Results are expressed as mean ± SEM (n = 3-6). Statistical analysis was performed by ANOVA. ** P < 0.01; *** P < 0.001; **** P < 0.0001. ns, not significant. overexpression results in cellular changes in lipid pools that do not reflect physiological conditions, which is supported by our ACOT activity assay results.
Alternative activation of ACOT7-deficient BMDMs was assessed after 24 h of IL-4 treatment and compared with that of classical activation with low dose (0.1 ng/ml) LPS. Low-dose LPS elicited an increase in Acot7 mRNA (supplemental Fig. S3A). IL-4 treatment increased expression of markers of alternative activation, but did not alter Acot7 mRNA, nor did ACOT7 deficiency influence the response to IL-4 (supplemental Fig. S3A-D).

ACOT7 is modestly increased in macrophages from diabetic mice, but hematopoietic ACOT7 deficiency does not prevent inflammation or early atherosclerosis in diabetic mice
TLR4 stimulation is thought to contribute to the proinflammatory phenotype of macrophages in diabetes (31,32). We therefore investigated whether ACOT7 is upregulated in macrophages in a mouse model of diabetes and whether ACOT7 could modulate inflammation and atherosclerosis in vivo. Thioglycollate-elicited macrophages from diabetic mice expressed modestly more Acot7 mRNA and ACOT7 protein than did their nondiabetic littermates (Fig. 5A, B). To test the effect of ACOT7 deficiency in bone marrow-derived cells in diabetic mice, bone marrow from Acot7 / mice or age-matched WT control mice was transplanted into irradiated Ldlr / recipient mice. After recovery, a subset of mice received STZ to induce diabetes, and the mice were then fed a semipurified low fat diet for the duration of the 12 week study (Fig. 5C). A complete KO of macrophage ACOT7 protein was achieved and sustained for the duration of the experiment in mice transplanted with bone marrow from Acot7 / mice (Fig. 5C). Diabetic mice exhibited increased plasma triglyceride levels, increased blood glucose, and decreased body weight, but had no changes in plasma NEFAs or cholesterol. Importantly, hematopoietic ACOT7 deficiency did not alter any of these parameters (Fig. 5D-H). At the end of the study, leukocytes from mice transplanted with Acot7 / bone marrow maintained significant knockdown of Acot7 mRNA, as compared with those transplanted with WT bone marrow (Fig. 5I). Levels of Il1b mRNA in blood leukocytes were used to evaluate low-grade inflammation associated with diabetes. Il1b was elevated in leukocytes in mice with diabetes, but loss of hematopoietic ACOT7 did not alter this marker of inflammation in diabetic mice (Fig. 5J). Another ACOT induced by LPS in macrophages, Acot9, was not significantly affected by ACOT7 deficiency or diabetes (Fig. 5K). Although diabetes has been noted to increase TLR4 expression in monocytes (31)(32)(33) and ACOT7 expression in macrophages, presence of diabetes did not result in upregulation of Acot7 or Tlr4 mRNA in whole-blood leukocyte preparations in this study (Fig. 5I, L). Most mice developed early macrophage-rich atherosclerotic lesions after 12 weeks of diabetes, but hematopoietic ACOT7 deficiency did not alter the effect of diabetes on atherosclerosis (Fig. 5M). Nondiabetic mice had no lesions or very small lesions at this time point (Fig. 5M).
Together, these results suggest that subtle increases in ACOT7 expression, such as those seen in macrophages from diabetic mice, do not mediate macrophage accumulation in the artery wall or inflammation associated with diabetes. Failure of ACOT7 deficiency to alleviate the inflammatory response in LPS-stimulated macrophages supports this finding.

DISCUSSION
Modulation of the immune response by lipids and enzymes that control their metabolism and subcellular localization within the cell is an area of research that has attracted significant attention in recent years. There are now several examples of proteins involved in intracellular FA handling that are induced by inflammatory stimuli in macrophages (1,(3)(4)(5)(6)25). Our study is the first to investigate the effect of ACOT7 deficiency in macrophages. The results demonstrate that loss of ACOT7 in mouse macrophages is suffi- had not yet been examined. Furthermore, activity toward acyl-CoA substrates was assessed in whole-cell lysates; thus, although some thioesterase activity toward substrates C20 remained, other enzymes with ACOT activity were not able to compensate for loss of ACOT7. Conversely, although ACOT7 can act on 16:0-CoA and 18:1-CoA in our overexpression model, endogenous ACOT7 is not required for complete cellular thioesterase activity toward these shorter acyl-CoA species in macrophage lysates. Our expanded cellular thioesterase activity assays suggest that a broader range of substrates for ACOT7 exists, which likely includes more unsaturated acyl-CoAs >20 carbons in length. Although it is possible that the effects of ACOT7 deficiency on some lipids are compensated for by other enzymes, it is also possible that ACOT7 may have a greater regulatory role of thioesterase activity in specific acyl-CoA pools or membrane compartments in vivo, which could explain the modest changes by ACOT7 deficiency on phospholipids containing 20-carbon unsaturated acyl-chains in the lipidomic data set. ACOT7 has been found to exhibit primarily a cytoplasmic localization in macrophages (7) (S.B., V.Z.W., and K.E.B., unpublished observations) and in other tissues (19,34).
Forced expression of an active ACOT7 mutant in a macrophage cell line has been shown previously to reduce basal levels of PGD 2 and PGE 2 (21). Perturbations that affect FA incorporation into phospholipids can have a profound effect on increasing eicosanoid synthesis (35). It has been hypothesized that the balance of 20:4-CoA/free 20:4 can be influenced by long-chain acyl-CoA hydrolases (36), such as ACOT7. However, our results clearly demonstrate that a large increase in overall 20:4-CoA thioesterase activity in macrophages overexpressing ACOT7 is not sufficient to reduce the steady-state pool of 20:4-CoA, suggesting that 20:4-CoA turnover is rapid, consistent with studies on other ACOTs. Accordingly, neither ACOT7 overexpression nor knockdown affected the ability of macrophages to produce PGE 2 . These findings mimic other work in neurons lacking Loss of ACOT7 in bone marrow-derived cells is not sufficient to prevent inflammation or atherosclerosis stimulated by diabetes. Acot7 mRNA (A) and ACOT7 protein (B) were measured in thioglycollate-elicited macrophages from diabetic and nondiabetic Ldlr / mice by real-time PCR and Western blot analysis, respectively. Lethally irradiated female Ldlr / mice received bone marrow transplants (BMT) from Acot7 / mice or WT controls, were allowed to recover for 7 weeks, and were then injected with STZ to induce diabetes. Some mice were injected with citrate and were used as nondiabetic controls. All mice were then fed a low-fat semipurified diet for 12 weeks (C). Isolated BMDMs harvested at the end of the study from mice that received Acot7 / bone marrow had no detectable ACOT7 protein, showing a near complete chimerism (C). The bands shown are from the same blot and exposure times. Plasma triglycerides were measured at the end of the study (D). Body weights (E) and blood glucose (F) were measured every 4 weeks. Plasma cholesterol and NEFA were measured at the end of the study (G, H). At the end of the study, blood leukocytes were used for real-time PCR measurements of Acot7 (I), Il1b (J), Acot9 (K), and Tlr4 (L). Atherosclerosis was evaluated by Sudan IV staining of the aorta using en face preparations and is expressed as percent lesion area of total aortic area (M). Representative en face preparations of the aortic arch are shown. Results are expressed as mean ± SEM (n = 10 nondiabetic; n = 15-17 diabetic). Statistical analysis was performed by Student's t-test (A, B) or two-way ANOVA. * P < 0.05; ** P < 0.01; *** P < 0.001. D, diabetic mice; ND, nondiabetic mice. n.s., not significant.
ACOT7. In this tissue, a near complete loss of thioesterase activity did not alter levels of acyl-CoAs, PGE 2 , or PGD 2 (12). Furthermore, the lipidomics suggests that ACOT7 deletion does not influence levels of 20:4-containing lipid species in basal conditions or their reorganization after LPS stimulation. This is highly suggestive of compensation by other enzymes in maintaining composition of 20:4-containing glycerophospholipids, even in the presence of significant reorganization. Because ACOT7 deficiency reduced thioesterase activity toward exogenously added 20:4-CoA in whole-cell lysates, it suggests that preservation of whole-cell glycerophospholipids containing 20:4 is accomplished by compensatory mechanisms beyond a simple rescue of thioesterase activity toward 20:4-CoA. However, it is possible that ACOT7 activity on 20:4-CoA is biologically relevant, but is limited to a distinct cellular pool that was not detectable in our whole-cell lipidome. Importantly, ACOT7 deficiency caused a greater reduction in activity toward 20:5-CoA and 22:6-CoA, suggesting that endogenous ACOT7 is a major thioesterase for these omega-3 acyl-CoAs in macrophages.
In neurons, where ACOT7 levels are high, expression of ACOT7 is positively regulated by binding of the transcription factor SREBP2 to the sterol regulatory element (SRE2) of the Acot7 promoter region (37). In macrophages, SREBP1 has recently been shown to be responsible for inducing enzymes involved in MUFA and PUFA synthesis, which, in turn, drives the resolution phase of the inflammatory response to TLR4 activation in these cells (6). However, after LPS stimulation, Acot7 mRNA is induced in the early phase of the inflammatory response, and its expression is decreased to near basal levels by 24 h, suggesting that SREBP1 may not be a major contributor to LPS-mediated induction of Acot7 mRNA, although ACOT7 protein levels are elevated up to 48 h after LPS stimulation. We determined that MyD88, and not TRIF, is the adaptor protein required for TLR4-mediated ACOT7 induction in these cells. MyD88 signaling activates the transcription factors AP-1 and NF-B (38). Thus, Acot7 mRNA transcription in response to LPS in macrophages is likely downstream of one or more of these transcription factors. Similarly, ACSL1 is induced by PPARs in insulin target tissues, but not in macrophages, in which ACSL1 is induced by LPS and other inflammatory mediators (25). This suggests that ACOT7 in macrophages is part of the group of FA handling enzymes involved in the inflammatory response.
The lipidomics demonstrated that LPS has a marked effect on acyl-chain composition of BMPs, a phospholipid present in late endosomes (23,39,40), in macrophages. BMP (also known as lysobisphosphatidic acid) has been shown to be involved in transport of proteins and lipids, such as cholesterol, through the late endosome (41), and 22:6/22:6-BMP has been suggested to protect cholesterol from oxidation (42). Although our finding that LPS regulates BMP acyl-chain composition in macrophages will require further study, the data suggest that LPS modulates endosomal trafficking as part of the inflammatory and resolution response in macrophages. The finding that ACOT7 deficiency did not alter the acyl-chain composition of these BMPs in LPS-stimulated cells, although these BMPs are enriched in omega-3 and omega-6 acyl-chains, suggests that the role of ACOT7 is not to alter this aspect of cellular trafficking.
TLR4 is upregulated in monocytes from humans with type 1 diabetes (31,33) and in a mouse model of the disease (32). Given that ACOT7 is induced by TLR4 activation, we hypothesized that expression of ACOT7 in macrophages would be altered in an STZ mouse model of diabetes and that it might influence the increased inflammatory response and atherosclerosis observed in diabetic mice. ACOT7 was indeed upregulated in macrophages from diabetic mice, although the extent of increase was modest. The increase in ACOT7 expression in the diabetic state may be specific to macrophages or myeloid cells, because diabetes did not significantly increase ACOT7 expression in the whole-blood leukocyte population. Hematopoietic ACOT7 deficiency did not inhibit diabetes-accelerated formation of early macrophage-rich lesions of atherosclerosis. Thus, although forced overexpression of ACOT7 at high levels has proinflammatory effects after LPS stimulation, it does not explain the increased inflammation and atherosclerosis associated with diabetes. Although our work describes the mechanism of Acot7 mRNA induction in response to LPS, it is likely that Acot7 transcription in macrophages is induced by additional stimuli in vivo and could use alternate signaling molecules downstream of these stimuli. Future studies will reveal whether ACOT7 plays a role in other immune cells and other inflammatory conditions, such as sepsis or various types of bacterial or viral infection.
In summary, ACOT7 is induced by TLR4-MyD88 activation and diabetes in macrophages. ACOT7 contributes a significant fraction of thioesterase activity toward C20:4-CoA, C20:5-CoA, and C22:6-CoA in macrophages, but is not required for LPS-induced reorganization of glycerophospholipids, particularly BMPs, containing unsaturated acyl-chains 20 carbon or longer. The functional outcome that ACOT7 may have on cellular processes in macrophages requires further investigation.