Botanical oils enriched in n-6 and n-3 FADS2 products are equally effective in preventing atherosclerosis and fatty liver.

Echium oil (EO), which is enriched in 18:4 n-3, the immediate product of fatty acid desaturase 2 (FADS2) desaturation of 18:3 n-3, is as atheroprotective as fish oil (FO). The objective of this study was to determine whether botanical oils enriched in the FADS2 products 18:3 n-6 versus 18:4 n-3 are equally atheroprotective. LDL receptor KO mice were fed one of four atherogenic diets containing 0.2% cholesterol and 10% calories as palm oil (PO) plus 10% calories as: 1) PO; 2) borage oil (BO; 18:3 n-6 enriched); 3) EO (18:4 n-3 enriched); or 4) FO for 16 weeks. Mice fed BO, EO, and FO versus PO had significantly lower plasma total and VLDL cholesterol concentrations; hepatic neutral lipid content and inflammation, aortic CE content, aortic root intimal area and macrophage content; and peritoneal macrophage inflammation, CE content, and ex vivo chemotaxis. Atheromas lacked oxidized CEs despite abundant generation of macrophage 12/15 lipooxygenase-derived metabolites. We conclude that botanical oils enriched in 18:3 n-6 and 18:4 n-3 PUFAs beyond the rate-limiting FADS2 enzyme are equally effective in preventing atherosclerosis and hepatosteatosis compared with saturated/monounsaturated fat due to cellular enrichment of ≥20 PUFAs, reduced plasma VLDL, and attenuated macrophage inflammation.

they ate a chow diet. At 7-8 weeks of age, mice were randomly assigned to one of four atherogenic diet (AD) groups (n = 15 per diet group) containing 10% calories as PO and 0.2% cholesterol, supplemented with an additional 10% of calories as: 1 ) PO, 2 ) BO (18:3 n-6 enriched), 3 ) EO (18:4 n-3 enriched), or 4 ) FO (20:5 n-3 and 22:6 n-3 enriched) for 16 weeks. The synthetic ADs were prepared by the diet kitchen in the Department of Pathology at Wake Forest School of Medicine as previously described ( 18 ). Detailed composition and quality control data for similar ADs have been published ( 16 ).

Body and organ weights
Mice were weighed every 2-4 weeks. After 16 weeks of AD feeding, body weights were recorded after a 4 h fast. Blood was collected via the tail vein at baseline and after 2, 4, 8, and 16 weeks of AD feeding. Mice were then anesthetized using ketamine-xylazine and perfused via the left ventricle using cold PBS at the rate of ‫ف‬ 3 ml/min for 3-4 min before organs were harvested. After perfusion, liver wet weights were measured and normalized to terminal body weight.

Fatty acid analysis
Lipid extraction of lyophilized diets, red blood cells (RBCs), plasma, and liver was performed using the Bligh-Dyer method ( 19 ). The total lipid extracts from plasma and liver were separated into cholesteryl ester (CE), TG, and phospholipid (PL) fractions by TLC. Lipids were re-extracted from CE, TG, and PL fractions and then transmethylated using boron trifl uoride ( 20 ), and percentage fatty acid composition was quantifi ed by gas-LC (GLC) as described previously ( 18 ). To estimate recovery of fatty acids from TLC and the transmethylation procedure, a known amount of tripentadecanoin (C15:0) and cholesteryl nonadecanoate (C19:0) were used as internal standards prior to TLC and triheptadecanoin (C17:0) was used as an internal standard prior to transmethylation. All internal standards were purchased from Nu-Chek-Prep, Inc., (T-145, Ch-818, and T-155). Percentage loss during TLC and transmethylation was <10% and <15%, respectively. Diet and RBC lipid extracts were directly transmethylated for fatty acid analysis.

Plasma lipid and lipoprotein analysis
Plasma was isolated by low-speed centrifugation. Plasma total and free cholesterol (FC) (Wako), and TG (Roche) concentrations were determined using enzymatic assays as described earlier ( 21 ) at baseline and after 2, 4, 8, and 16 weeks of AD feeding. CE content was calculated as [total cholesterol (TC)−FC] × 1.67; this calculation corrects for loss of fatty acid during the cholesterol esterase step of the assay . Data were expressed as area under the curve (AUC), which integrates plasma lipid concentrations over the 16 weeks, representing a time average estimate of hypercholesterolemia during the 16 week experiment. Plasma lipoprotein cholesterol mass distribution was determined using fast protein LC (FPLC) fractionation on a Superose 6 10/300 column (GE Healthcare). Three equal volumes of plasma samples from each time point (5 mice per group) were pooled and subjected to FPLC fractionation and cholesterol quantifi cation. VLDL, LDL, and HDL cholesterol concentrations were determined and then expressed as AUC.
One strategy to circumvent the poor conversion of dietary 18-carbon PUFAs to their у 20 carbon bioactive products is to identify botanical oils enriched in PUFAs beyond the FADS2 rate-limiting step of desaturation and elongation. We previously showed that echium oil (EO), which is relatively enriched in stearidonic acid (SDA, 18:4 n-3), the immediate product of FADS2-mediated desaturation of ALA, effectively enriches plasma and tissue lipids in EPA ( 16 ). In LDL receptor KO (LDLrKO) mice, isocaloric replacement of palm oil (PO) with EO attenuated atherosclerosis severity, splenic monocytosis, monocyte infl ux into aortic intima, and aortic root intimal macrophage content to an equivalent extent as FO, lending proof of principle for this strategy ( 17 ). However, whether a similar strategy will be atheroprotective for the n-6 PUFA pathway is unknown.
To address this gap in knowledge, we identifi ed borage oil (BO) as a potential botanical oil to test this strategy in the n-6 PUFA conversion pathway. BO is enriched ( ‫ف‬ 20%) in ␥ -linolenic acid (GLA, 18:3 n-6), the immediate product of FADS2-mediated desaturation of LA. GLA can be elongated to DGLA, a precursor of the anti-infl ammatory PGE 1 , or arachidonic acid (AA, 20:4 n-6), a precursor of pro-infl ammatory eicosanoid species. The atheroprotective effects of BO have not been explored, and little is known about its effects on plasma lipids and lipoproteins. The purpose of this study was to directly compare the atheroprotective potential of BO with EO and elucidate potential mechanisms for atheroprotection.

Dietary oils
The seed oil of Borago offi cinalis L. (a member of the Boraginaceae family) and the seed oil of Echium plantagineum L. (a member of the Boraginaceae family) were generous gifts from Croda Europe Ltd. (Leek, Staffordshire, UK) and authenticated by the Wake Forest University Center for Botanical Lipids and Infl ammatory Disease Prevention. The seed oil of palm, Elaeis guineensis Jacq (a member of the Arecaceae family) was purchased from Shay and Company (Portland, OR). For these oils, a certifi cate of analysis is on fi le and retention samples are deposited at the Wake Forest School of Medicine. The FO source, Brevoortia tyrannis Latrobe (a member of the Clupeidae family), was manufactured and generously provided by Omega Protein (Reidsville, VA) with a report of analysis on fi le for reference.

Animals and atherogenic diets
Female LDLrKO (C57BL/6 background) mice (5-6 weeks of age) were purchased from The Jackson Laboratory (Bar Harbor, ME). Mice were housed in a specifi c pathogen-free facility on a 12 h light/dark cycle. All protocols and procedures were approved by the Institutional Animal Care and Use Committee. Mice were allowed to acclimate for 1-2 weeks during which time LC/MS/MS analysis. Before analysis, the organic layer was dried under N 2 . The residue was then weighed and diluted to 1 g/ l with isooctane, and stored at Ϫ 20°C until analysis of oxidized CE (ox-CE) using LC/MS (/MS) ( 28 ). The remaining 1 ml of the original aortic extract in 75:25 isooctane:ethyl acetate was used to quantify aortic total cholesterol (TC) and FC content by GLC ( 17 ). Results were normalized to aortic wet weight.

Peritoneal macrophage studies
Thioglycollate-elicited peritoneal macrophages were isolated and cholesterol content was measured as described previously ( 21 ). Macrophage gene expression was measured after 2 h of PBS or lipopolysaccharide (LPS) (200 ng/ml) stimulation ( 21 ) and eicosanoids were identifi ed and quantifi ed using LC/MS/MS in macrophage-conditioned media ( 29,30 ). In vitro macrophage chemotaxis in response to MCP-1 and MIP-1 ␣ was performed in a 48-well microtaxis chamber ( 31 ).

Real-time PCR analysis
Total RNA was isolated from mouse macrophages and liver using TRIzol (Invitrogen) and quantitative real-time PCR was performed to determine gene expression ( 21 ). Primer sequences have been reported previously ( 20,27,32 ). GAPDH was used as the control for normalization of results.

Statistics
Data are reported as mean ± SEM. Statistical analyses were performed using one-way ANOVA and individual paired diet comparisons were made using Tukey's post hoc analysis. All statistical analyses were performed using GraphPad Prism software.

Systemic response to ADs enriched in n-3 versus n-6 fatty acid products of FADS2
Dietary fatty acid compositions are given in Table 1 and show relative enrichment of 18-carbon fatty acids beyond FADS2 in the BO (19.7% 18:3 n-6) and EO (6% 18:4 n-3) diets. AD feeding over 16 weeks resulted in uniform food consumption ( ‫ف‬ 3-4 g/day/mouse), body weight gain, and terminal liver/body weight ratios among all groups ( Fig. 1 ). Upon initiation of AD feeding, signifi cant RBC fatty acid enrichment over baseline (chow diet) was evident 180 min after injection were determined by enzymatic assay. TG secretion rate was calculated using linear regression analysis to determine the slope of the time versus TG concentration plot for each animal ( 20 ).

Liver lipid analysis
Livers were harvested at necropsy, fl ash-frozen in liquid N 2 , and stored at Ϫ 80°C. Liver lipids were quantified using a detergent-based enzymatic assay ( 23 ).

Western blot analysis
Nuclear proteins were prepared from frozen livers using ultracentrifugation as described ( 24,25 ). Equal aliquots (20 g) of nuclear protein from individual livers were subjected to SDS-PAGE on 4-12% gels and transferred to a polyvinylidene difl uoride membrane. Immunoblot analyses were performed using monoclonal anti-mouse SREBP-1 and rabbit polyclonal SREBP-2 antibodies as described ( 26 ). Anti-SREBP-1 and -2 antibodies were generously donated by Dr. Timothy Osborne (Sanford Burnham Medical Research Institute). Anti-YY1 antibody (Abcam #43058) was used as a nuclear loading control. Whole cell lysates were prepared using frozen livers ( 27 ) and equal aliquots (50 g) of protein from individual livers were subjected to SDS-PAGE on 4-12% gels and transferred to polyvinylidene fl uoride membranes. Immunoblot analyses were performed using rabbit monoclonal anti-FAS antibody (Cell Signaling #3180) and goat polyclonal anti stearoyl-CoA desaturase-1 (SCD-1) antibody (Santa Cruz Biotechnology #14719).

Atherosclerotic lesion analysis
A subset of mice (n = 3 per group) was euthanized after 8 weeks of AD feeding to measure aortic CE content. In the remaining mice, aortic CE content, aortic root intimal area, and intimal macrophage content (CD68 + ) were measured after 16 weeks of AD feeding ( 17 ). Aortic root intimal area was measured with Oil Red O staining.

Aortic oxidized CE and aortic cholesterol content analysis
At necropsy, aortas were cleaned of visible adventitial fat and placed into a 15 ml etched glass tissue grinder containing 1 ml of 1:1 methanol:water (v/v) and a known amount of internal standard 5-␣ cholestane. Aortas were homogenized on ice and neutral lipids were extracted twice using 75:25 isooctane:ethyl acetate (v/v). The combined organic layers were brought to a volume of 2 ml with 75:25 isooctane:ethyl acetate. One milliliter of the extract was added to an argon-purged ampule, heat-sealed, and shipped on dry ice to the University of Colorado Denver for  within 1 week and appeared to reach equilibrium by 4 weeks ( Fig. 2 ). RBC fatty acid compositions refl ected efficient in vivo elongation-desaturation of dietary 18:3 n-6 derived from BO to its longer chain (20-carbon) counterparts, 20:3 n-6 and 20:4 n-6. Similarly, 18:4 n-3 derived from EO was elongated-desaturated to 20:5 n-3. These data indicate that dietary enrichment of 18-carbon fatty Fig. 1. Body weight gain and terminal liver/body weight ratios. A: Body weight gain was monitored periodically from baseline (chow diet) to 16 weeks of feeding the indicated ADs. B: Mice were then euthanized and liver wet weights were measured and normalized to terminal body weight. Data are expressed as mean ± SEM; n = 15 per diet group. No signifi cant differences were found by one-way ANOVA. and equilibrated by 4 weeks for PO-fed mice ( Fig. 5A ). VLDL-c concentrations were signifi cantly and equivalently attenuated in the other diet groups ( Fig. 5 A, D ). LDL cholesterol (LDL-c) levels showed a similar pattern; only FOfed mice had signifi cantly lower LDL-c concentrations versus PO-fed mice ( Fig. 5B, E ). HDL cholesterol (HDL-c) concentrations also increased sharply within 4 weeks among BO-, EO-, and FO-fed mice, and were relatively stable thereafter. HDL-c concentrations were signifi cantly lower in PO-fed mice compared with the other groups ( Fig. 5C, F ); however, HDL-c was <10% of the total pool of plasma cholesterol. Hence, most of the TC lowering came from decreases in VLDL-c. BO-induced reduction in VLDL-c was equivalent to that of EO and FO, but BO did not reduce plasma TG concentrations.

EO, but not BO, diet reduces hepatic VLDL TG secretion rate
We investigated whether the difference in plasma TG concentrations between EO-versus BO-fed mice could be explained by hepatic TG secretion. Using detergent inhibition of plasma TG lipolysis, hepatic TG secretion rates were signifi cantly lower for EO-and FO-fed groups compared with BO-and PO-fed mice ( Fig. 6 ). These data suggest that the higher plasma TG concentrations in BO-versus EO-fed mice are likely due, in part, to a higher hepatic VLDL TG secretion rate.
Fatty acid composition was also measured in plasma and liver CE, TG, and PL fractions ( Fig. 3 ). In plasma from BOfed mice, equivalent ( ‫ف‬ 20%) AA enrichment was observed in CE, TG, and PL fractions relative to plasma from PO-fed mice ( Fig. 3 A-C ). Plasma neutral lipids from mice fed BO were also relatively enriched in GLA [CE ( ‫ف‬ 6%) and TG ( ‫ف‬ 15%)], unlike plasma PL. In BO-fed livers, AA was modestly, but signifi cantly, enriched in PL ( ‫ف‬ 20%), but not in neutral lipids (below 5%), relative to the other diet groups ( Fig. 3 D-F ). In general, PUFA enrichment in liver CE and TG was low for all diet groups relative to liver PL.

Plasma lipid and lipoprotein response to dietary BO versus EO
Chow-fed mice had similar baseline measurements, but after 2 weeks of AD feeding, plasma TC, CE, and TG concentrations increased signifi cantly, peaked by week 4, and remained elevated over the 16 weeks of diet feeding for PO-fed mice ( Fig. 4 ). This pattern was signifi cantly attenuated in the other diet groups. BO and EO induced equivalent cholesterol lowering, whereas FO induced even further TC and CE lowering, compared with PO. Plasma TG concentrations increased equivalently in PO-and BO-fed mice after 2 weeks of AD feeding, but stayed at baseline levels in the EO-and FO-fed mice.
A sharp increase in VLDL cholesterol (VLDL-c) concentrations occurred within 2 weeks of AD feeding that peaked To determine the reason for the decreased hepatic neutral lipid content with BO and EO feeding, we examined hepatic lipogenic gene expression using quantitative real-time study, ADs containing BO, EO, or FO for 16 weeks reduced hepatic total neutral lipid content (i.e., TG and CE) relative to PO ( Fig. 7A ), but FC and PL contents were similar among all diet groups. Furthermore, BO and EO were  Aortic root intimal area was signifi cantly lower (300-400 vs. 700 mm 2 ) in the three PUFA-containing ADs compared with PO ( Fig. 8C ), as was aortic root intimal macrophage content (CD68 + ) ( Fig. 8D ). Collectively, these results show that BO, EO, and FO were equally effective in reducing aortic atherosclerosis and macrophage content.

Mouse aortas lack ox-CE species
Human atheromas contain several distinct families of ox-CE species derived from PUFAs (e.g., 18:2 n-6, 20:4 n-6, and 22:6 n-3) that may play a role in atherogenesis ( 28 ). Because three of the four ADs showed substantial PUFA enrichment, we examined ox-CE content in mouse aortas after 16 weeks of AD feeding. Aortic CE fatty acyl molecular species refl ected the fatty acid enrichment of the diet; however, aortas had undetectable levels of ox-CE in all diet groups ( Fig. 9 ). The relative abundance of the nonoxidized CE molecular species presented in Fig. 9 insets do not precisely refl ect absolute differences in molecular species due to different electrospray ionization response factors (mass spectrometric parameters) for the polyunsaturated CEs, as previously noted ( 29 ). Nonetheless, these raw data reveal dietary modifi cation of PUFA-containing CE molecular species in aorta tissues refl ecting the dietary fats.

Thioglycollate-elicited peritoneal macrophage eicosanoid release is similar for BO-and EO-fed mice
We studied the effects of AD feeding on thioglycollateelicited peritoneal macrophage eicosanoid biosynthesis. After 16 weeks of AD feeding, thioglycollate-elicited peritoneal macrophages were incubated with or without LPS for 2 h before media eicosanoids were quantifi ed. This model of stimulated eicosanoid biosynthesis has been recently reported, including details of gene expression related to PUFA metabolism stimulated by LPS ( 30 ). Among the eicosanoids measured, 12/15 lipoxygenase-derived LA metabolites 9-and 13-HODE were most abundant ( Fig. 10 ). Furthermore, the type of dietary fat had little impact on production of thioglycollate-elicited peritoneal macrophage eicosanoid species in the basal state or after LPS stimulation. Only in the FO group were signifi cant reductions observed for generation of TXB 2 , PGE 2 , and 12-hydroxyeicosatetraenoic acids after LPS stimulation ( ‫ف‬ 2-to 4-fold lowering vs. PO, EO, and BO). These data indicate that LPS stimulation revealed little difference in thioglycollate-elicited peritoneal macrophage eicosanoid biosynthesis among BO-, EO-, and PO-fed mice.
PCR. Compared with PO, all three ADs signifi cantly lowered SREBP-2 mRNA and its target gene HMG-CoA reductase, but not HMG-CoA synthase ( Fig. 7B ). Although SREBP-1c mRNA expression levels were similar among all diet groups, expression of its target genes, FAS and SCD-1, was signifi cantly reduced in all three ADs relative to PO ( Fig. 7B ). There were no signifi cant differences among diet groups in LXR ␣ and ABCA1 expression, whereas PPAR ␣ expression was signifi cantly induced in FO-fed mice ( Fig. 7B ). Because mice consuming the BO diet had reduced hepatosteatosis, similar to that of mice fed the EO or FO diets, we determined whether BO feeding attenuated hepatic infl ammation as well. mRNA expression of CD68, TNF-␣ , and MCP-1 were signifi cantly and equivalently reduced in BO-, EO-, and FO-fed mice compared with those fed PO ( Fig. 7B ).
Expression of other pro-infl ammatory cytokines, such as IL-1 ␤ , IL-6, and IL-18, and the alternatively activated macrophage markers, arginase-1 and anti-infl ammatory cytokine IL-10, were similar among all four diet groups (data not shown). To determine whether FADS2 products reduce hepatic lipogenic gene expression via attenuation of the SREBP pathway, we measured accumulation of proteolytically cleaved mature/nuclear SREBP-1 and -2 isoforms in liver nuclear preparations. Content of nuclear SREBP-1 (but not SREBP-2) was signifi cantly reduced in BO-, EO-, and FO-fed mice relative to PO ( Fig. 7C ). Hepatic protein content of the SREBP-1c target genes, FAS and SCD-1, was also signifi cantly lower in those groups relative to PO ( Fig. 7D ).

BO and EO are equally atheroprotective
Aortic FC content was comparable among the groups after 8 weeks; however, CE content was signifi cantly lower ( ‫ف‬ 2-fold lower) in all groups versus PO-fed mice ( Fig. 8A ). After 16 weeks of AD feeding, aortic FC and CE content was signifi cantly lower in all groups ( ‫ف‬ 2-fold reduction in FC; ‫ف‬ 3-to 4-fold reduction in CE) versus PO-fed mice ( Fig. 8B ), indicating that BO, EO, and FO induced equivalent atheroprotection compared with PO. Signifi cant aortic FC accumulation occurred between 8 and 16 weeks of AD feeding among all groups ( ‫ف‬ 3-4 g/mg at 8 weeks vs. 6-14 g/mg at 16 weeks). CE accumulation increased modestly among all groups ( ‫ف‬ 1-3 g/mg at 8 weeks vs. ‫ف‬ 2-4 g/mg at 16 weeks) relative to the PO group ( ‫ف‬ 4 g/mg at 8 weeks to ‫ف‬ 16 g/mg at 16 weeks), indicating a disproportionate increase of CE deposition in PO-fed mice. Fig. 6. Hepatic VLDL-TG secretion rate. A: Plasma TG levels were measured by enzymatic assay before (0 min) and after (60, 120, and 180 min) intravenous Triton ® injection. B: Hepatic TG secretion rate during the 3 h experiment was calculated for each animal as the slope of the regression line. Data are plotted as mean ± SEM, n = 4-5. Groups with different letters are signifi cantly different ( P < 0.05) by oneway ANOVA and Tukey's post hoc analysis. isolated from mice after 16 weeks of AD feeding. Basal ( Ϫ LPS) gene expression of the pro-infl ammatory cytokines IL-6, IL-1 ␤ , TNF-␣ , and the chemokine MCP-1 was comparable among all groups ( Fig. 11A ). LPS-treated macrophages from mice fed BO, EO, and FO versus PO had decreased mRNA expression of IL-6, IL-1 ␤ , and MCP-1, whereas TNF-␣ expression was signifi cantly higher ( Fig. 11A ).

BO and EO attenuate macrophage infl ammatory response to LPS
HODEs are natural ligands for macrophage PPAR ␥ , which, when activated, can result in downregulation of the canonical NF-B pathway ( 36,37 ). Thus, we measured expression of NF-B target genes after LPS-induced activation in thioglycollate-elicited peritoneal macrophages compared with PO-fed mice ( Fig. 11B ). After 16 weeks of AD feeding, macrophage CE content increased considerably ( ‫ف‬ 2-to 8-fold) compared with 8 weeks and was significantly attenuated in BO-, EO-, and FO-fed mice versus PO-fed mice ( Fig. 11B ). We next investigated whether reduced macrophage chemotaxis might partially explain the reduction in aortic root macrophage content in BO-, EO-, and FO-fed mice. All three PUFA-containing ADs were equally effective in reducing chemotaxis to MIP-1 ␣ and MCP-1 compared with PO ( Fig. 11C ).

DISCUSSION
In the current study, we tested the hypothesis that botanical oils enriched in 18:3 n-6 (BO) and 18:4 n-3 (EO) PUFAs beyond the rate-limiting FADS2 enzyme are equally atheroprotective compared with saturated/monounsaturated fat (PO) in LDLrKO mice. Although BO differs from EO and FO in its inability to lower plasma TGs, BO and EO were comparable in their ability to lower plasma cholesterol concentrations, especially VLDL-c. Additionally, BO was as effective as EO and FO in alleviating hepatic steatosis, and in regulating hepatic lipogenic and infl ammatory However, 6 h after LPS stimulation, TNF-␣ mRNA expression was signifi cantly lower in macrophages from BO-, EO-, and FO-fed mice versus PO-fed mice (data not shown), similar to the results for IL-6, IL-1 ␤ , and MCP-1 at 2 h. mRNA abundance for the alternatively activated macrophage markers, arginase 1 and CD206, was not different among groups (data not shown). Collectively, these data indicate that BO attenuates macrophage LPS-induced proinfl ammatory gene expression to the same extent as EO and FO, without affecting the gene expression of alternatively activated macrophage markers.

BO and EO equally attenuate macrophage CE accumulation and chemotaxis
Given the equivalent effectiveness of BO in attenuating aortic CE and macrophage content, we sought to determine whether macrophages from BO-fed mice contributed to reduced aortic atherosclerosis via reduced macrophage foam cell formation. We used thioglycollate-elicited peritoneal macrophages (as a surrogate for aortic macrophages) from mice after 8 and 16 weeks of AD exposure to estimate CE content. At 8 weeks, macrophage CE content ranged from 5 to 20 g/mg among the groups, but only the FOfed mice had signifi cantly lower macrophage CE content Our current results show that this is also true for the n-6 pathway of fatty acid elongation-desaturation. BO, which is enriched in GLA, was as atheroprotective as EO and FO despite its signifi cant enrichment of RBCs and plasma and liver lipid fractions with AA, a fatty acid precursor to pro-infl ammatory leukotrienes and prostaglandins ( 42 ). Concerns that elevated membrane AA may result in increased cellular infl ammation that exacerbates atherosclerosis lack support in human studies ( 13 ). Moreover, LAenriched diets have not enriched AA in plasma and tissue lipid fractions ( 40,43,44 ), likely due to ineffi cient FADS2 conversion of LA to AA ( 45,46 ). In addition, a recent meta-analysis of 13 cohort studies (involving 310,602 individuals and 12,479 coronary heart disease events) revealed an inverse association between dietary LA intake and coronary heart disease risk, such that a 5% increase in energy intake from LA was associated with a 10% and 13% lower risk of coronary heart disease events and deaths, respectively ( 12 ). Collectively, these results suggest that increased consumption of dietary LA and GLA is not harmful and is potentially atheroprotective in the general population. Our results also suggest that assessing cardiovascular risk and infl ammatory potential by dietary n-3/n-6 PUFA ratio may be an oversimplifi cation that does not take into account differences between 18-versus у 20-carbon PUFAs.
A recent study reported a single nucleotide polymorphism (rs174537) in the FADS1/2 gene cluster that affects plasma AA levels ( 47 ). The GG allele, which is nearly twice as frequent in African Americans as in European Americans, was associated with a small ( ‫ف‬ 3%), but statistically significant, enrichment in plasma AA. A subsequent study demonstrated that GG homozygotes for the rs174537 allele had increased plasma AA enrichment and production of LTB 4 gene expression compared with PO. As a result, BO and EO resulted in signifi cant atheroprotection relative to PO at early (8 weeks) and advanced (16 weeks) atherosclerotic stages. We also report that ox-CEs were undetectable in early and advanced atherosclerotic arteries, suggesting that ox-CEs play a minimal role in murine atherosclerosis progression. Additionally, BO and EO signifi cantly and equivalently attenuated macrophage infl ammatory gene expression, CE content, and chemotaxis. BO had these atheroprotective outcomes despite signifi cant enrichment of RBC membranes and plasma and liver lipids with 20:4 n-6, a precursor for several pro-infl ammatory eicosanoids. Our results support the conclusion that botanical oils enriched in 18:3 n-6 and 18:4 n-3 PUFAs beyond the rate-limiting FADS2 enzyme are equally atheroprotective and hepatoprotective compared with saturated/ monounsaturated fat.
Although previous studies in nonhuman primates and mice have demonstrated n-6 PUFA-mediated atheroprotection, these studies used dietary fats primarily enriched in LA (38)(39)(40). Our study focused on the hypothesis that dietary enrichment with n-6 PUFAs beyond FADS2 (i.e., GLA) would be as atheroprotective as EO, which we have previously demonstrated as equal to FO in preventing atherosclerosis progression ( 17,41 ). On the other hand, fl axseed oil, which is enriched in ALA, a substrate for FADS2, was not as atheroprotective as FO; this outcome occurred despite signifi cant enrichment of liver PLs with EPA and was likely due to the lower plasma LDL-c concentrations in the FO versus the fl axseed oil group ( 9 ). The combined results of the fl axseed oil and EO studies support our hypothesis that dietary enrichment in n-3 PUFAs beyond FADS2 is atheroprotective. Fig. 9. Mouse atherosclerotic plaque ox-CE analysis. LDLrKO mice were fed the indicated ADs for 16 weeks before aortas were harvested for ox-CE analysis using normal phase-HPLC-MS/MS. Ion chromatograms of CE (2-4 min elution) and ox-CEs (6-36 min elution) for one aorta from each diet group. Inset presents the raw mass spectrometric data corresponding to CE molecular species labeled with m/z values and their acyl components, denoted by total acyl carbons and double bonds in this qualitative study. No ox-CE species were detectable for any of the four diet groups.
Plasma HDL-c was signifi cantly elevated in all three diet groups, and may have contributed to atheroprotection relative to the PO group; however, only a small fraction ( ‫ف‬ 10%) of plasma cholesterol was distributed in HDL particles, making this a less likely possibility. In other studies investigating the infl uence of dietary n-3 and n-6 PUFAs on atherogenesis, LDL was the predominant plasma atherogenic lipoprotein ( 9,38 ). Differences in plasma apoB lipoprotein response (VLDL vs. LDL) among these studies are likely due to genetic background (LDLrKO vs. apoB100 only-LDLrKO) and diet composition (0.02 vs. 0.2% cholesterol; 10 vs. 20% calories as fat).
A signifi cant driver of atherogenesis in nonhuman primates and LDLrKO mice is hepatic production of saturated and monounsaturated CEs that are core constituents of secreted VLDL particles ( 11 ). Plasma VLDL particles undergo intravascular metabolism to become LDL particles, a major atherogenic lipoprotein particle in plasma. Deletion of the cholesterol esterifi cation enzyme, steroyl O-acyltransferase 2 (SOAT2), strikingly reduces atherosclerosis regardless of dietary fat saturation ( 38 ), supporting a critical role for SOAT2 in the generation of and 5-hydroxyeicosatetraenoic acids in zymosan-stimulated blood, suggesting that genetic polymorphisms may infl uence tissue AA content and infl ammatory response to external pathogens ( 48 ). Thus, individuals harboring rs174537 homozygous GG alleles may be hyperresponsive to diets enriched in LA and GLA. Further studies are required to determine whether this potential hyperresponsiveness affects coronary heart disease risk.
Our results also suggest that BO and EO slow atherosclerosis progression in multiple ways, including reduced plasma VLDL-c concentrations, macrophage cholesterol content, infl ammatory gene expression, and decreased macrophage migration toward a chemokine gradient. Plasma VLDL-c reduction likely had the greatest infl uence on atherosclerosis outcome in this study, because VLDL-c concentrations are the best lipoprotein/lipid predictor of aortic root atherosclerosis in LDLrKO mice ( 49 ), and we observed a strong positive association between plasma VLDL-c and aortic CE content (r 2 = 0.88; P < 0.0001, data not shown). LDL-c concentrations were similar among PO-, BO-, and EO-fed mice, suggesting they had a minimal effect on atheroprotection in BO-and EO-fed mice. AA in the BO group ( Figs. 2,3,9 ). Thus, ox-CEs did not appear to contribute signifi cantly to atherogenesis in our study.
CE-loaded macrophages are a hallmark of atherosclerosis, which result from chemokine-induced chemotaxis of monocytes to atherosclerotic lesions, differentiation of monocytes into macrophages expressing scavenger receptors, and unregulated uptake of modifi ed apoB lipoproteins and apoB lipoprotein-proteoglycan complexes ( 54 ). In addition to reduced aortic cholesterol and aortic root intimal area in mice fed BO and EO versus PO, we also observed decreased aortic root macrophage (CD68 + cells) content and decreased macrophage chemotaxis in vitro. Thioglycollate-elicited peritoneal macrophages isolated from BO-and EO-fed mice also had reduced sterol content and infl ammatory response to LPS compared with their PO-fed counterparts. Macrophage FC accumulation resulting from genetic deletion of macrophage effl ux genes ABCA1 and ABCG1 results in hyper-responsiveness to pro-infl ammatory stimuli and increased chemotaxis in vitro and in vivo ( 21,55 ). However, in this study, the atherogenic CEs in apoB-containing lipoproteins, such as VLDL and LDL. Hepatic CE content was signifi cantly and equivalently reduced in mice fed BO and EO, suggesting the production of atherogenic CEs was blunted in these mice compared with those fed PO. This likely contributed to lower VLDL-c concentrations for the EO and BO groups, and reduced monounsaturated 18:1 n-9 CE species in plasma at the expense of FADS2-derived CEs.
One of the earliest events in atherogenesis is arterial retention of apoB lipoproteins by proteoglycans ( 50 ). A recent study showed that SOAT2 KO versus WT mice or mice fed FO versus monounsaturated fat had plasma LDL enriched in PUFA CE species and bound with less affi nity to proteoglycans, suggesting a mechanism by which apoB lipoproteins from mice fed dietary PUFA may be less atherogenic ( 51 ). Although PUFA lipid species are more easily oxidized than their saturated and monounsaturated counterparts ( 52,53 ) and ox-CE species have been identifi ed in human tissue samples from endarterectomies ( 28 ), no arterial ox-CEs were detected in our study, despite signifi cant enrichment of circulating and tissue lipids with Fig. 11. Macrophage infl ammation, foam cell formation, and chemotaxis. LDLrKO mice were fed the indicated ADs for 8 or 16 weeks before thioglycollateelicited macrophages were isolated. A: Infl ammatory gene expression was measured using RT-PCR after 2 h treatment with LPS (200 ng/ml) or PBS ( Ϫ LPS) in thioglycollate-elicited peritoneal macrophages. B: Peritoneal macrophage CE content was measured using GLC after 8 and 16 weeks of AD feeding. C: Ex vivo chemotaxis of thioglycollate-elicited peritoneal macrophages toward MCP-1 and MIP-1 ␣ was measured after 16 weeks of AD feeding. Data are expressed as mean ± SEM, n = 5. In (B) and (C), data points for individual mice are also shown. Groups with different letters are signifi cantly different ( P < 0.05) by one-way ANOVA and Tukey's post hoc analysis.
large decrease in hepatic TG content does not necessarily result in decreased TG secretion. For example, knockdown of SCD-1 with a targeting anti-sense oligonucleotide resulted in a >90% reduction in hepatic TG content, but did not affect hepatic TG secretion compared with mice treated with a nontargeting anti-sense oligonucleotide ( 66 ). We also observed a 50% decrease in newly synthesized TG secreted from livers of monkeys fed FO versus lard diets, although liver TG synthesis was similar between diet groups ( 34 ). These combined results suggest a unique secretory pool of TG that may be regulated differently than the bulk TG storage pool in hepatocytes.
Collectively, our results support the hypothesis that dietary enrichment with FADS2 fatty acid products, such as SDA and GLA, results in membrane and plasma lipid enrichment in EPA and AA, which in turn is associated with reduced plasma lipids, atherosclerosis, and hepatosteatosis. This hypothesis was based on data in humans and animal models showing that conversion of 18-carbon PUFAs to у 20-carbon PUFAs is ineffi cient, and that bypassing the rate-limiting FADS2 step is possible by feeding botanical oils enriched in FADS2 products. Furthermore, allowing for body surface area differences between mice and humans ( 67 ), achieving this dose of botanical oils in the diet is feasible. For example, our ADs contained BO and EO as 10% energy, which corresponds to a human equivalent dose of 0.81% energy [10% × 3/37 ( 67 )], well within the range of n-3 PUFA doses administered in many randomized clinical trials ( 68 ) and the American Heart Association's recommended n-3 PUFA intake for individuals with documented coronary heart disease ( 69 ). While replacing dietary saturated fat with PUFAs is viewed as atheroprotective in general, our study suggests a more targeted approach of dietary PUFA replacement using known biochemical pathways may enhance the beneficial outcomes of increased dietary PUFA consumption.
Anti-SREBP-1 and -2 antibodies were generously donated by Dr. Timothy Osborne (Sanford Burnham Medical Research Institute). The authors gratefully acknowledge Karen Klein (Biomedical Research Services and Administration, Wake Forest School of Medicine) for editing the manuscript. cholesterol elevation was due to CE, not FC, and there was no increase in macrophage ABCA1 and ABCG1 gene expression among diet groups (data not shown). These results suggest that decreased uptake of cholesterol from apoB lipoproteins likely explained the decreased macrophage atherogenic phenotype for mice fed EO and BO.
Diets enriched in n-3 PUFAs reduce hepatosteatosis ( 9,16,33,56 ), unlike those containing n-6 PUFAs ( 27,56 ). The reduced hepatic neutral lipid content in animals fed n-3 PUFAs is primarily mediated through decreased hepatic lipogenesis and is usually accompanied by reduced hepatic TG secretion ( 34,57 ). In vitro, n-3 and n-6 PUFAs suppress SREBP-1c gene transcription ( 58,59 ), decreasing hepatic lipogenesis, by competing with activators of LXR, a potent inducer of hepatic TG synthesis ( 58,60 ). The n-3 and n-6 PUFAs also increase mRNA degradation ( 61 ), inhibit the proteolytic processing of SREBP-1c in vitro ( 59 ), and accelerate the degradation of nuclear SREBP-1c ( 62 ), reducing lipogenesis. Thus, we hypothesize that BO, but not the other n-6 PUFA-enriched diets ( 27,56 ), reduces hepatosteatosis relative to diets containing saturated/ monounsaturated fatty acids via its ability to enrich liver lipids in AA through elongation and desaturation of GLA. Many n-6 PUFA-enriched botanical oils contain LA as the predominant fatty acyl species, accounting for 85-90% of n-6 PUFA consumption in the US ( 10 ). However, as discussed above, diets enriched in LA do not result in plasma and tissue AA enrichment ( 44,63 ). Genetic deletion of Elovl5, the gene encoding the enzyme that elongates 18:3 n-6 to 20:3 n-6 and 18:4 n-3 to 20:4 n-3, results in diminished hepatic lipid AA and DHA and increased neutral lipid storage ( 64 ). Feeding Elovl5 KO mice AA or DHA rescued the hepatosteatosis phenotype by decreasing nuclear SREBP-1c and de novo lipogenesis with no changes in mRNA or membrane-bound SREBP-1c, supporting a role for AA in blocking SREBP-1c cleavage and activation. In our study, liver membrane (i.e., PL) AA content was elevated, nuclear SREBP-1 content was reduced, and SREBP-1c-targeted genes (FAS, SCD-1) were reduced in BO-versus PO-fed mice, supporting a role for elevated AA in reducing hepatosteatosis in BO-fed mice. In another study, BO reduced ethanol-induced hepatosteatosis ( 65 ), suggesting that BO protects against hepatosteatosis regardless of the method of induction. How elevated liver AA inhibits the proteolytic processing of membrane SREBP-1c is unclear, but may be related to fl uidity of the endoplasmic reticulum membrane. Regardless of the detailed mechanism for decreasing hepatic lipogenesis, our study reveals a distinct advantage of BO in reducing hepatosteatosis and atherosclerosis in contrast to other n-6 PUFA-enriched botanical oils that are atheroprotective, but do not prevent hepatosteatosis.
Although BO, EO, and FO all reduced hepatosteatosis comparably relative to PO, BO did not reduce hepatic TG secretion or plasma TG concentrations, whereas EO and FO did ( Fig. 6 ). FO consistently reduces hepatic TG secretion, likely due to reduced hepatic lipogenesis ( 34,41 ). This paradox is likely due to the fact that only a small fraction of hepatic TG is mobilized for secretion; therefore, a