Macrophage 12/15 lipoxygenase expression increases plasma and hepatic lipid levels and exacerbates atherosclerosis.

12/15 lipoxygenase (12/15LO) oxidizes polyunsaturated fatty acids (PUFAs) to form bioactive lipid mediators. The role of 12/15LO in atherosclerosis development remains controversial. We evaluated atherosclerosis development and lipid metabolism in 12/15LO-LDL receptor (LDLr) double knockout (DK) vs. LDLr knockout (SK) mice fed a PUFA-enriched diet to enhance production of 12/15LO products. Compared with SK controls, DK mice fed a PUFA-enriched diet had decreased plasma and liver lipid levels, hepatic lipogenic gene expression, VLDL secretion, and aortic atherosclerosis and increased VLDL turnover. Bone marrow transplantation and Kupffer cell ablation studies suggested both circulating leukocytes and Kupffer cells contributed to the lipid phenotype in 12/15LO-deficient mice. Conditioned medium from in vitro incubation of DK vs. SK macrophages reduced triglyceride secretion in McArdle 7777 hepatoma cells. Our results suggest that, in the context of dietary PUFA enrichment, macrophage 12/15LO expression adversely affects plasma and hepatic lipid metabolism, resulting in exacerbated atherosclerosis.

separated by size using fast protein liquid chromatography, and cholesterol distribution among lipoprotein fractions was determined by enzymatic cholesterol assays. After 16 weeks of atherogenic diet consumption, mice were euthanized, and liver was harvested for measurement of lipid content as previously described ( 23 ). Liver samples were also fi xed in 10% formalin and embedded in paraffi n for hematoxylin and eosin staining.

Quantifi cation of aortic lesion area and lipid content
Whole aorta (from sinotubular junction to iliac bifurcate) was fi xed in 10% formalin, and the adventitia was cleaned. Aortas were opened along the longitudinal axis, pinned onto black silicon elastomer, and photographed with a Sony DXC-S500 digital camera. The digital images were used to quantify the percentage of total aortic surface covered with lesion using Image Pro 6.2 software. After surface lesion quantifi cation, aortas were transferred to a glass tube containing 10 µg of 5 ␣ -cholestane as internal standard for gas-liquid chromatographic analysis of cholesterol content as previously described ( 24 ).

Real-time PCR and Western blotting
Real-time PCR and Western blot analysis were performed as described previously ( 25 ). Primer sequences used for real-time PCR were summarized in Supplementary Table IV.

In vivo quantifi cation of TG and apoB secretion rate
Detergent block to inhibit TG lipolysis in vivo was used to quantify the rate of TG and apoB secretion into plasma of SK and DK mice after 12 weeks of PUFA diet consumption ( 25,26 ). Another detergent block study was performed with BM transplanted mice after 15 weeks of PUFA diet consumption.

VLDL turnover study
After feeding mice the PUFA diet for 16 weeks, plasma was collected from SK and DK mice after a 4 h fast. VLDL was separated from plasma by ultracentrifugation (d = 1.006, 100,000 rpm for 4 h with a Beckman Coulter TLA100.2 rotor). VLDL chemical composition was determined using enzymatic assays, and particle size was quantifi ed with a Zetasizer Nano S dynamic light scatter instrument (Malvern Instruments Ltd., Worcestershire, UK). VLDL from SK and DK mice were radioiodinated with 125 I and 131 I, respectively, as described ( 27 ). Ninety-eight percent of the radiolabel was trichloroacetic acid precipitable (i.e., protein bound), and 80% was associated with apoB, as determined by isopropanol extraction ( 28 ). VLDL tracers ( 125 I and 131 I; 350,000 cpm each) were mixed, diluted to a volume of 200 µl with saline, and injected into PUFA-diet fed SK-and DK-recipient mice (fasted 4 h) through a jugular vein catheter ( 29 ). Plasma samples were taken periodically after injection (5 and 30 min and 1, 2, 3, 5, 8, and 24 h), and 125 I and 131 I radioactivity in isopropanol-extracted plasma was quantifi ed with a Perkin Elmer automatic ␥ counter. The fractional catabolic rate (FCR) of VLDL in plasma was calculated using SAAM II software (SAAM Institute, Seattle, WA) as described previously ( 21 ).

BM transplantation
BM was harvested from cleaned femurs and tibias of female SK and DK mice and resuspended in RPMI 1640 medium for injection. SK and DK male recipient mice were fasted overnight and received a lethal dose of radiation (900 rads) 4 h before BM injection. Mice were anesthetized with isofl urane, and 5 × 10 6 BM cells were injected into the retro-orbital venous plexus. Three days before and 2 weeks after the BM transplantation, recipient mice received autoclaved acidifi ed (pH 2.7) water supplemented with 100 mg/l neomycin and 10 mg/l polymyxin B sulfate. After is atherogenic ( 3,5,(9)(10)(11)(12)(13)(14)(15), but other studies, using similar experimental approaches, have concluded that 12/15LO expression is atheroprotective ( 16,17 ). One possible explanation for the inconsistent results is that diets used in these studies varied in fatty acid composition, resulting in formation of different oxidative lipid mediators by macrophages in vivo, which had different biological effects on infl ammation, cytokine production, and endothelial activation (18)(19)(20).
To date, no atherosclerosis study has addressed the role of 12/15LO expression in an experimental setting in which dietary substrate fatty acids are enriched to drive production of 12/15LO-catalyzed lipid mediators. To address this gap in knowledge, we compared atherosclerosis development in 12/15LO KO mice in the LDLrKO background [double knockout (DK)] to LDLrKO (SK) mice using an atherogenic diet enriched in PUFAs. For comparison, we also fed a control diet enriched in saturated and monounsaturated fatty acids (Sat/mono), similar to that used in previously published studies ( 11 ). Our results suggest a dietary fatty acid-gene interaction for 12/15LO that affects plasma and hepatic lipid metabolism as well as atherosclerosis.

Mice and diets
LDLrKO and 12/15LOKO mice in the C57BL/6 background were obtained from The Jackson Laboratory (Bar Harbor, ME). The mice were crossbred to generate heterozygotes at both targeted alleles, and heterozygous matings were used to generate 12/15LO-LDLr DK mice. Genotypes were ascertained by PCR analysis using tail DNA as described previously ( 21 ). For LDLr genotyping, the following primers were used: forward primer, ACCCCAAGACGTGCTCCCAGGATGA; reverse primer for wildtype allele, CGCAGTGCTCCTCATCTGACTTGT; and reverse primer for knockout allele, CTTGTTCAATGGCCGATCCC. For 12/15LO genotyping, the following primers were used: forward primer, CGAAATCGCTGGTCTACAGG; reverse primer for wildtype allele, CGTGGTTGAAGACTCTCAAGG; and reverse primer for knockout allele, AGGTGAGATGACAGGAGATC. All mice used in this study were male, except for BM transplantation studies, in which female mice served as BM donors. Mice were housed in a specifi c pathogen-free facility with a 12 h light/dark cycle and, unless otherwise specifi ed, were fasted 4 h before blood samples were collected. All protocols and procedures were approved by the animal care and use committee of Wake Forest School of Medicine. The atherogenic diets used in this study were prepared by the diet kitchen in the Department of Pathology at Wake Forest School of Medicine and were similar in composition to those previously described ( 22 ). For the initial studies, two experimental diets were fed: one containing 10% of calories as palm oil diet (i.e., Sat/mono) and the other containing 10% of calories as saffl ower oil (PUFA). Both diets contained 0.2% cholesterol. Fatty acid composition of the diets is given in Supplementary Table I.

Plasma and liver lipid analysis
Plasma cholesterol and triglyceride (TG) concentrations were measured using enzymatic assays with commercially available reagents (cholesterol: Wako; TG: Roche). Plasma lipoproteins were

Statistical analyses
Data are presented as mean ± SEM and were analyzed for statistical signifi cance using a two-tailed unpaired Student's t -test (i.e., SK vs. DK comparisons). FCR values for different radioiodinated VLDL tracers in the same recipient mice were compared using a two-tailed paired-t -test. All the analyses were performed with GraphPad Prism 5.01 (GraphPad Software Inc., San Diego, CA).

12/15LO expression is pro-atherogenic in SK mice
To determine whether 12/15 LO expression is proatherogenic, we fed DK and SK mice atherogenic diets that were enriched in Sat/mono (i.e., palm oil) or PUFA (i.e., saffl ower oil) for 16 weeks (Supplementary Table I). Atherosclerosis was evaluated by two measurements: the percentage of aortic surface covered with lesion and aortic cholesterol content. DK mice fed either diet developed less atherosclerotic surface lesion area compared with SK mice ( Fig. 1A, B ). The attenuation of atherosclerotic surface lesion area in DK vs. SK mice was more robust in the PUFA diet group compared with the Sat/mono diet group (68.9% vs. 35.7%). Aortic cholesterol content was lower in PUFA-fed DK vs. SK mice but was similar for Sat/mono-fed mice ( Fig. 1C ). To determine the extent to which aortic surface lesion area and cholesterol content were associated, that, mice were given acidifi ed water until the end of the experiment. The replacement of hematopoietic cells by donor BM was evaluated by quantifying the remaining relative copy number of male DNA in hematopoietic genomic DNA 4 weeks after BM transplantation. Briefl y, genomic DNA was extracted from whole blood with a Wizard Genomic DNA Purifi cation kit (Promega). The y-chromosome associated sex-determining region Y gene (Sry) was amplifi ed by PCR to a linear amplifi cation phase with the following conditions: 95°C for 3 min followed by 30 cycles of 94°C for 30 s, 61°C for 1 min, and 72°C for 1 min. The amplifi ed 380 bp PCR product was separated on 0.8% agarose gels and visualized with ethidium bromide. Acyl-CoA:cholesterol acetyltransferase 2 (ACAT2) was amplifi ed by PCR as an internal control. A series of male and female genomic DNA mixtures (100, 50, 25, and 0% male DNA) were made and used to construct a standard curve of male:female DNA ratio vs. Sry:ACAT2 ratio, determined by quantifi cation of density of PCR bands. Male genomic DNA in the whole blood samples of transplanted mice was estimated using the standard curve.

GdCl 3 -induced ablation of Kupffer cells
GdCl 3 ·6H 2 O (Sigma Aldrich) powder was reconstituted to 6 mg/ml with saline and sterilized by passing through a 0.22 µm fi lter. After 16 weeks of PUFA diet consumption, mice were anesthetized with isofl urane, and the GdCl 3 solution (25 µg/g body weight) was injected into mice through the retro-orbital venous plexus. Injections were repeated every 4 days for a total of three injections.

Hepatocyte and Kupffer cell isolation
Hepatocytes and Kupffer cells were harvested from collagenaseperfused livers ( 30 ). Nonparenchymal cells were separated from hepatocytes by spinning at 50 × g for 5 min and loaded onto the top of a Percoll gradient. Kupffer cells were harvested at a density of 1.037-1.066 g/ml. Cells were immediately treated with Trizol to isolate RNA. Gene expression was quantifi ed by real-time quantitative PCR.

Macrophage-conditioned medium preparation and McArdle RH7777 TG secretion studies
Resident peritoneal macrophages were collected from mice fed the PUFA diet for 12-16 weeks by fl ushing the peritoneal cavity with PBS. The macrophages were pelleted by low-speed centrifugation; resuspended in RPMI 1640 medium containing 100 U/ml penicillin, 100 g/ml streptomycin, and 10% of FBS; and seeded onto cell culture plates. After a 2 h incubation, fl oating cells were removed by washing with PBS and adherent macrophages were stimulated with ±2 µM Ca 2+ ionophore A23187 for 20 min. Then, cells were changed to fresh medium (RPMI 1640, DMEM equal volume) containing 100 U/ml penicillin, 100 g/ml streptomycin, and 1% Nutridoma-SP (Roche Applied Science) and cultured overnight. Macrophage-conditioned medium was transferred to McArdle RH7777 cells along with 5 µCi 3 H-oleate, 0.8 mM oleate, and 5% BSA and incubated for 4 h. After incubation, media lipids were extracted ( 31 ), TG was separated by TLC, and radiolabel in TG was quantifi ed by liquid scintillation counting and expressed as dpm/mg cell protein.

Macrophage lipid mediator analysis
Macrophages were stimulated with Ca 2+ ionophore (A23187) for 20 min. Then, cells and medium were collected together to separate and quantify lipid mediator concentration by LC/MS/MS as previously described ( 32 ). Sat/mono-fed DK mice also had lower plasma TG levels ( ‫ف‬ 16%) compared with SK mice. There was a signifi cant association between aortic surface lesion area and TPC ( Fig. 2E ), suggesting that the additional reduction in aortic surface lesion area in PUFA vs. Sat/mono-fed DK mice was due, in part, to the reduction in TPC. After mice were fed the atherogenic diets for 16 weeks, plasma lipoproteins were fractionated by fast protein liquid chromatography, and cholesterol distribution was measured. Although cholesterol concentration was signifi cantly lower in all lipoprotein fractions in DK mice fed the PUFA diet ( Fig. 2F ), the most striking differences were among the apoB-containing lipoprotein particles (VLDL and LDL).

Hepatic steatosis is improved in 12/15LO-defi cient mice
Given the critical role of liver in lipoprotein and lipid metabolism and the dramatic reduction in plasma VLDL and LDL cholesterol in PUFA-fed DK mice, we examined hepatic lipid content of the mice. Liver weight was similar among the four groups (data not shown). Liver cholesteryl ester (CE) content was signifi cantly lower in PUFA-fed DK compared with SK mice ( Fig. 3B ). Interestingly, liver TG content was 50% to 60% lower in DK vs. SK mice in both diet groups ( Fig. 3D ), suggesting a major effect of 12/15LO expression on liver TG storage. Hepatic free cholesterol and phospholipid were similar among the groups ( Fig. 3A, C ). Finally, hematoxylin and eosin staining showed fewer and smaller neutral lipid droplets in livers of DK mice in both diet groups compared with SK mice ( Fig. 3E ), indicating improved hepatic steatosis in the absence of 12/15LO expression.

Liver lipogenic gene expression is reduced in 12/15LO-defi cient mice
Because PUFA diet feeding resulted in reduced plasma ( Fig. 2 ) and liver ( Fig. 3 ) cholesterol in DK mice, whereas the Sat/mono diet did not, the remainder of the studies was focused on PUFA-fed DK and SK mice. To investigate the underlying mechanisms contributing to the protective role of 12/15LO defi ciency on lipid metabolism and atherosclerosis, expression of key hepatic lipid regulatory genes was analyzed by real-time PCR after the mice were fed the PUFA diet for 16 weeks. Hepatic expression of fatty acid and TG synthesis-related genes, such as sterol regulatory element binding protein-1c (SREBP-1c), fatty acid synthase (FAS), acetyl-CoA carboxylase-1 (ACC-1), and stearoyl-CoA desaturase-1, were decreased in DK mice ( Fig. 4A ). mRNA level of cholesterol synthetic regulatory genes, such as SREBP-2 and 3-hydroxy-3-methylglutaryl-CoA (HMG-CoA) synthetase and reductase, also trended lower in DK mice. Acyl-CoA oxidase 1 and carnitine palmitoyltransferase 1, two genes involved in regulation of fatty acid ␤ oxidation, had similar expression in both genotypes of mice. Consistent with the decreased hepatic mRNA expression of lipogenic genes, FAS and stearoyl-CoA desaturase-1 protein expression was 56% and 33% lower in DK compared with SK mice ( Fig. 4B, C ). Expression of proteins involved in removal of cholesterol from the liver, such as ATP-binding cassette transporter A1, liver X receptor we performed linear regression analysis and found a highly signifi cant association between the two measurements of atherosclerosis ( Fig. 1D ).

PUFA-fed DK mice have reduced plasma lipid concentrations
Previous studies in different mouse genetic backgrounds (i.e., apoE-KO and LDLrKO) have not observed an effect of 12/15LO expression on plasma lipid concentrations ( 9,11 ). However, none of those studies used an atherogenic diet enriched in PUFA or substrate fatty acids for 12/15LO enzymatic activity. To determine whether the additional reduction in aortic surface lesion area observed for PUFAfed vs. Sat/mono-fed DK mice ( Fig. 1B ) was due to a reduction in plasma lipids, we measured total plasma cholesterol (TPC) and TG concentrations and lipoprotein cholesterol distribution. Unexpectedly, DK mice fed the PUFA diet had signifi cantly lower TPC and TG concentrations (40%) compared with SK mice ( Fig. 2A-D ). However, Sat/mono-fed SK and DK mice had similar TPC values, in agreement with previously published studies ( 9, 11 ). after mice were fed the PUFA diet for 16 weeks. Plasma VLDL from both genotypes of mice were similar in size ( ‫ف‬ 49 nm diameter), but VLDL from DK mice were relatively enriched in TG (45% vs. 36% of total mass) and poor in CE (23% vs. 33%) compared with those isolated from SK mice (Supplementary Table II). Because the core lipid compositions were different, VLDLs from SK and DK mice were used as tracers for the turnover studies. VLDLs from DK and SK mice were radiolabeled with 131 I and 125 I, respectively, mixed together, and injected into PUFA-fed recipient mice intravenously. 131 I-VLDL tracer from DK mice was cleared signifi cantly faster ( P < 0.05) from plasma in both genotypes of recipient mice compared with 125 I-VLDL tracer from SK mice (Supplementary Fig. I ; FCR, 2.0 ± 0.3 vs. 1.0 ± 0.2 pools/day in SK recipients and 2.81 ± 0.45 vs. 1.85 ± 0.18 pools/day in DK recipients, n = 5). There was also a trend toward faster plasma clearance of both VLDL tracers in DK recipient mice that reached statistical signifi cance for the 125 I-VLDL tracer from SK mice. These data support the conclusion that 12/15LO defi ciency in PUFA-fed mice results in hypercatabolism of plasma VLDL particles, which partially explains the reduced plasma apoB lipoprotein concentrations in DK mice.

VLDL secretion is reduced in PUFA-fed DK mice
Decreased liver lipid synthetic gene and protein expression in PUFA-fed DK mice suggested that lower plasma lipid levels might result from decreased hepatic lipid synthesis and secretion. To investigate this possibility, we measured the VLDL secretion rate in vivo after intravenous injection of detergent to inhibit TG lipolysis in mice fed the PUFA diet for 12 weeks. TG secretion rates were measured as an increase in plasma TG concentration ( Fig. 5A, C ) or as 3 H-oleate incorporation into plasma TG ( Fig. 5B and D ); both measures of TG production were signifi cantly reduced in DK mice. There was a signifi cant reduction in apoB 48 secretion rate ( Fig. 5G ) and a trend toward decreased apoB100 secretion rate ( Fig. 5F ) in PUFA-fed DK vs. SK mice.

VLDL clearance is increased in PUFA-fed 12/15LOdefi cient mice
To evaluate plasma VLDL clearance rate as a possible mechanism for reduced plasma VLDL and LDL concentrations in DK mice, VLDL turnover studies were performed  BM transplanted into DK recipients (DK → DK) resulted in signifi cantly reduced TPC and TG concentrations compared with SK recipients transplanted with SK BM (SK → SK) ( Fig. 6A-D ), similar to results observed for nontransplanted SK and DK mice ( Fig. 2A-D ). However, when SK or DK mice received BM from the opposite genotype (i.e., SK → DK or DK → SK), plasma lipid levels were intermediate between those of SK → SK and DK → DK recipients. This was also the case when we evaluated liver TG secretion using the detergent block method ( Fig. 6E-F ). These data suggested that transplanted BM cells could only partially explain the hepatic and plasma lipid lowering observed in PUFA-fed DK mice.
Atherosclerosis, measured as a percentage of aortic surface lesion area, was reversed by transplantation of DK BM

Transplantation of DK mouse BM into SK mice reversed atherosclerosis but only partially restored plasma lipid phenotype
Our data indicated that the lower TPC and TG in PUFAfed DK mice was the result of altered hepatic lipid metabolism. However, 12/15LO is mainly expressed in BM-derived cells, such as macrophages ( 3,4 ). We also could fi nd no detectable 12/15LO mRNA expression in isolated hepatocytes from SK mice, but we found abundant expression in Kupffer cells of SK vs. DK mice (Supplementary Fig. II ). We hypothesized that macrophages were affecting hepatic lipid metabolism, presumably by secreting a 12/15LO product that directly or indirectly affected hepatic lipid metabolism. To investigate this possibility, we transplanted BM from SK and DK donor mice into lethally irradiated SK and DK recipient mice. More than 98% of the hematopoietic cells in the circulation were replaced by the donor BM 4 weeks after transplantation ( Supplementary  Fig. III ). Eight weeks after transplantation, mice were fed the PUFA diet for 16 weeks to induce atherosclerosis. DK  between Kupffer cell 12/15LO expression and lipid metabolism in vivo, a GdCl 3 solution was injected into 16-week PUFA-diet fed DK and SK mice three times at 4-day intervals to ablate hepatic Kupffer cells. Plasma cholesterol and TG concentrations were measured periodically for up to 32 days after the fi rst GdCl 3 injection. Plasma cholesterol and TG levels in SK mice began to decrease 6 days after the fi rst GdCl 3 injection ( Fig. 8 ). The decrease in plasma lipid values continued for 12-14 days and then began to return to preinjection values, coincident with the reported time course of Kupffer cell repopulation by BM ( 34 ). Thirty-two days after the fi rst injection, plasma lipid levels had recovered to preinjection values. Plasma lipid concentrations of DK mice did not change after GdCl 3 injection. These results support our hypothesis that Kupffer cells expressing 12/15LO regulate hepatic lipid metabolism, raising plasma cholesterol and TG concentrations, whereas ablation of Kupffer cells not expressing 12/15LO (i.e., DK) had no effect on plasma lipids.

Profi le of lipid mediators produced by activated macrophages
Because 12/15LO enzymatic activity generates bioactive lipid mediators, we profi led the lipid mediator content of macrophages (cells + conditioned media) 20 min after stimulation with the Ca 2+ ionophore A23187. 12-HETE, the direct product of 12/15LO, was nearly absent in 12/15LO-defi cient macrophages (Supplementary Table  III). By comparison, nonenzymatic generation of 15-HETE was low and similar for both genotypes of macrophages. However, 5-HETE, the product of 5LO, was doubled in medium from 12/15LO-defi cient macrophages compared with SK-conditioned medium, suggesting that PUFA substrate shunts to 5LO in the absence of 12/15LO. PGE 2 and ( Fig. 7A ), such that SK and DK mice receiving BM from DK mice had >50% reduction in aortic surface lesion area compared with recipient mice receiving SK BM. A similar but less robust trend was observed for aortic cholesterol content ( Fig. 7B ), and there was a signifi cant association between aortic cholesterol content and surface lesion area ( Fig. 7D ), as previously observed for nontransplanted mice ( Fig. 1D ). However, the signifi cant association between plasma cholesterol concentration and aortic lesion area observed in the nontransplanted mice ( Fig. 2E ) was no longer observed in the BM transplantation study ( Fig. 7C ), suggesting that circulating BM cells that lacked 12/15LO expression were still able to protect against atherosclerosis but were less effective in reducing hepatic lipid synthesis and secretion.

Plasma lipid concentrations are decreased with ablation of Kupffer cells in SK mice
The BM transplantation experiment partially supported our hypothesis that 12/15LO expression in macrophages plays a role in regulation of liver lipid metabolism. However, resident macrophages, in general, and Kupffer cells, in particular, require a long time for complete replacement after BM transplantation ( 33 ). Incomplete replacement of Kupffer cells by donor BM in our study might explain why liver and plasma lipid levels were not completely reversed by BM replacement. GdCl 3 is a reagent that is specifi cally taken up by Kupffer cells and induces Kupffer cell apoptosis, resulting in temporary ablation of Kupffer cells ( 34 ). To further study the relationship  and TG concentrations, hepatic cholesterol content, and atherosclerosis compared with SK mice with intact 12/15LO expression. The decrease of TPC in PUFA-fed DK mice that drives the reduction in atherosclerosis is primarily due to decreases in plasma VLDL and LDL cholesterol resulting from decreased hepatic lipogenic gene expression and VLDL TG secretion as well as increased VLDL turnover. BM transplantation and Kupffer cell ablation studies suggested that 12/15LO activity has different pathological outcomes depending on where it is expressed, with circulating leukocytes primarily affecting the atherosclerosis outcome, with minimal impact on plasma lipids, whereas Kupffer cells affected plasma lipid levels. Finally, macrophages from DK relative to SK mice fed the PUFA diet lack one or more components responsible for increased TG secretion in McArdle cells. These results suggest a dietary fatty acid-gene interaction for 12/15LO that affects plasma and hepatic lipid metabolism as well as atherosclerosis.
In this study, we show that 12/15LO expression is proatherogenic, in agreement with other studies ( 3,5,(9)(10)(11)(12)(13)(14)(15). The increased atherosclerosis with 12/15LO expression in previous studies was not accompanied by a change in plasma cholesterol concentrations. We also observed a reduction in atherosclerosis in DK mice fed the Sat/mono diet that was independent of plasma cholesterol changes, in agreement with previous studies. A potential explanation for decreased atherosclerosis in the absence of plasma cholesterol changes is that 12/15LO expression primarily affects oxidation and infl ammation in the artery wall. Cyrus et al. (12) reported that 12/15LO promoted atherosclerosis by increasing in vivo oxidative status. Additional support for this idea came from another study in which vitamin E, an antioxidant, attenuated atherosclerosis in SK mice to the same level as DK mice not treated with vitamin E ( 15 ). 12/15LO was also reported to increase endothelial expression of intercellular adhesion molecule 1, which in turn increased the recruitment of monocytes to atherosclerotic lesions, exacerbating the disease ( 2,14 ). However, Merched et al. ( 16 ) observed that 12/15LO expression systemically or in BM protected against atherosclerosis and proposed that the anti-infl ammatory products of 12/15LO, such as lipoxins, resolvins, and protectins, protect mice from the development of atherosclerosis by dampening infl ammation. Past studies have not used diets enriched in PUFA, which are the substrate fatty acids for 12/15LO generation of bioactive lipid mediators, which can be pro-or anti-infl ammatory (18)(19)(20). In our study, dietary PUFA resulted in additional reduction of atherosclerosis in the absence of 12/15LO expression beyond that observed with a Sat/mono diet, suggesting a dietary fatty acid-gene interaction that resulted in this outcome. The additional reduction in atherosclerosis was directly attributable to a signifi cant reduction in plasma VLDL and LDL cholesterol concentrations because there was a strong positive association between total plasma cholesterol and aortic surface lesion area ( Fig. 2E ). These results suggest another environmental variable, dietary fatty acid composition, to consider in interpreting the role of 12/15LO in PGF 2 ␣ levels were also doubled in 12/15LO-defi cient macrophages, whereas the LTB 4 level was 50% lower. Thus, 12/15LO expression altered multiple bioactive lipid mediators generated from macrophages of mice fed the PUFA diet.

Secretion products from DK macrophages attenuate TG secretion from McArdle cells
Results from the BM transplantation and Kupffer cell ablation studies support our hypothesis that macrophage 12/15LO expression regulates hepatic lipid metabolism. We hypothesized that this likely resulted from a secretion product of macrophages that affected hepatic lipid synthesis/secretion. To test this concept, resident peritoneal macrophages were isolated from SK and DK mice fed the PUFA diet for 12 weeks. Conditioned medium was prepared by culturing the macrophages overnight after an acute (20 min) stimulation ± Ca 2+ ionophore (A23187) to activate the macrophages. The macrophage conditioned medium was then added to McArdle cells, along with 3 Holeate and 0.8 mM oleate to stimulate VLDL TG secretion, and 3 H-TG in the medium was quantifi ed after a 4 h incubation. Macrophage-conditioned medium from both genotypes stimulated McArdle cell TG secretion compared with unconditioned medium ( Fig. 9 ). However, McArdle cells receiving conditioned medium from DK macrophages secreted less TG than those receiving conditioned medium from SK macrophages, suggesting that secretory products from SK macrophages stimulated McArdle cell TG secretion more than those of DK macrophages.

DISCUSSION
The role of 12/15LO expression in atherosclerosis development is controversial, with 12/15LO expression reported as pro-and antiatherogenic ( 3,5,(9)(10)(11)(12)(13)(14)(15)(16)(17). However, no studies have been conducted in the context of 12/15LO substrate fatty acid enrichment. In this study, we demonstrate that DK mice fed a PUFA diet have reduced TPC Fig. 9. Macrophage secretion products attenuate TG secretion from McArdle cells. Peritoneal resident macrophages were isolated from PUFA-diet fed mice and stimulated with Ca 2+ ionophore for 20 min. The medium was replaced with fresh culture medium and incubated overnight. The conditioned medium from SK and DK macrophages was added to McArdle cells along with 5 µCi hepatic TG secretion must await further studies, our data suggest a new pathway for the regulation of hepatic lipid production by macrophages.
Although transplantation of DK BM into SK or DK recipients had little effect on plasma lipids or hepatic TG secretion rate ( Fig. 6 ), it strikingly decreased aortic surface lesion area in both SK and DK recipients ( Fig. 7A ). Furthermore, the strong positive association between total plasma cholesterol and aortic surface lesion area observed in nontransplanted mice ( Fig. 2E ) was not apparent after BM transplantation. These results, along with previous BM transplantation data ( 3 ), suggest that 12/15LO expression in circulating BM-derived cells does not infl uence plasma lipid concentrations but is atherogenic due to production of proinfl ammatory lipid mediators. The combined results of the BM transplantation and Kupffer cell ablation studies suggest a dual role for 12/15LO in the pathogenesis of atherosclerosis, with circulating BM-derived cells directly affecting atherosclerosis through infl ammation pathways in the artery wall and resident Kupffer cells indirectly affecting hepatic lipid metabolism in a paracrine manner that may be related to infl ammatory cytokines.
the pathogenesis of diseases such as atherosclerosis and obesity ( 35 ).
PUFA-fed DK mice had lower hepatic neutral lipid (TG and CE) content and hepatic lipogenic gene expression compared with their SK counterparts. A recent study of chow and high fat diet-fed 12/15LO knockout mice in the apoE knockout background also reported decreased hepatic Oil Red-O staining compared with apoE knockout controls and decreased expression of FAS but not SREBP-1c; however, no signifi cant difference in total serum cholesterol or TG concentrations was observed ( 36 ). Our hepatic TG results are in agreement with this published study as our liver TG content was lower in DK vs. SK mice regardless of diet ( Fig. 3 ), suggesting that dietary PUFA enrichment is not necessary for the reduced hepatosteatosis in the absence of 12/15LO expression, whereas it is for decreased plasma cholesterol and TG concentrations.
The mechanisms by which 12/15LO expression affected plasma lipid levels in PUFA-fed mice in our study involved both production and catabolism. The reduced plasma lipid concentrations in PUFA-fed DK mice were due to reduced VLDL and LDL concentrations ( Fig. 2F ). Detergent block studies demonstrated reduced plasma accumulation of VLDL particles (i.e., apoB) and TG in PUFA-fed DK mice compared with SK mice ( Fig. 5 ). Because the detergent block studies were performed in fasted mice, the reduced secretion was likely due to decreased hepatic, not intestinal, secretion. In addition, there was an increase in particle turnover for DK VLDL tracer in both genotypes of recipient mice, suggesting that the composition of the VLDL particle was affecting its clearance. The increased TG content in DK VLDL (Supplementary Table II) may have resulted in smaller VLDL remnants after intravascular TG lipolysis, with subsequent accelerated particle clearance from plasma.
Because the lipid phenotype of the PUFA-fed DK mice implicated decreased liver lipid production and 12/15LO is mainly expressed in macrophages, not hepatocytes ( 3 ) (Supplementary Fig. II ), we hypothesized that macrophage 12/15LO expression was altering a secretory product(s) that affected hepatic lipid synthesis. Although transplantation of homologous BM (i.e., SK → SK or DK → DK) recapitulated the plasma phenotype of nontransplanted mice, heterologous BM transplantation (SK → DK and DK → SK) resulted in an intermediate phenotype ( Fig. 6 ). Resident liver macrophages (i.e., Kupffer cells) are replaced slowly after BM transplantation ( 33 ), likely resulting in a chimeric (i.e., donor and recipient) Kupffer cell population after transplantation. Thus, failure to recapitulate the nontransplanted mouse lipid phenotype may be due to chimeric Kupffer cells exerting a paracrine effect on hepatic lipid synthesis. In support of this concept, ablation of Kupffer cells resulted in decreased TPC and TG in SK mice but not in DK mice lacking 12/15LO expression ( Fig. 8 ). Subsequent in vitro studies showed that incubation of conditioned medium from DK mouse macrophages resulted in less McArdle cell TG secretion compared with conditioned medium from SK macrophages. Although elucidating the molecular mechanism by which macrophages regulate