Oxidative Stress Is Markedly Elevated in Lecithin:Cholesterol Acyltransferase-deficient Mice and Is Paradoxically Reversed in the Apolipoprotein E Knockout Background in Association with a Reduction in Atherosclerosis*

Complete lecithin:cholesterol acyltransferase (LCAT) deficiency is a rare cause of severe hypoalphalipoproteinemia, but the affected subjects are surprisingly not particularly prone to premature coronary heart disease. We studied oxidative stress in lcat−/− mice and their cross-breed with apolipoprotein-E knockout mice (apoE−/−xlcat−/−) by measuring vascular ring superoxide production and plasma phospholipid (PL)-bound F2-isoprostane levels and their relationship with aortic atherosclerosis. Compared with wild type control (lcat+/+), lcat−/− and lcat+/− mice showed a 4.9- (p = 0.003) and a 2.1-fold (p = 0.04) increase in plasma PL-F2-isoprostane levels, respectively. There was also a 3.6- (p < 0.0001) and 2.9-fold (p = 0.003) increase in the area under the curve for the aortic ring superoxide excursion by lucigenin-derived chemiluminescence. A comparison of apoE−/−xlcat+/+ mice with wild type control mice showed a more modest 2.1- (p = 0.04) and 2.2-fold (p < 0.00001) increase in these respective markers. Surprisingly, the apoE−/−xlcat−/− mice showed a paradoxical normalization in both oxidation markers. Furthermore, by fast protein liquid chromatography separation, we observed an associated retention and redistribution of serum paraoxonase activities to the non-high density lipoprotein fractions in both the apoE−/−xlcat−/− and apoE−/−xlcat+/− mice. Aortic atherosclerotic lesions in male apoE−/−xlcat−/− and apoE−/−xlcat+/− mice were reduced by 52 (p = 0.02) and 24% (p = 0.46), respectively. Our data suggest that LCAT-deficient mice are associated with an increased oxidative stress that is paradoxically reversed in a hyperlipidemic background, possibly due to the redistribution of paraoxonase. This modulation of oxidative stress may in part contribute to the reduced atherosclerosis seen in the apoE−/− xlcat−/− mice.

Lecithin:cholesterol acyltransferase (LCAT) 1 plays a central role in the reverse cholesterol transport process by mediating the esterification of tissue-derived free cholesterol (FC) and is responsible for the majority of esterified cholesterol (CE) in the circulation (1). Subjects with LCAT deficiency as a result of mutations of the LCAT gene invariably develop severe HDL deficiency, but surprisingly, these subjects do not seem to be particularly prone to premature coronary heart disease (2). The role of LCAT in atherosclerosis remains controversial.
Several lines of experimental evidence suggest that HDL may partially confer its anti-atherogenic action as an antioxidant through activities of its associated enzyme, paraoxonase (PON1) (3). Recent studies on PON1 ko mice (4,5) suggest that PON1 plays a major role as antioxidant in the prevention of atherosclerosis. The PON1 ko mice were found to develop accelerated atherosclerosis with and without the apoE deficiency background. We reported recently (6) that LCAT-deficient mice have significantly lower levels of plasma PON1 arylesterase activities. A number of in vitro studies have also suggested that the LCAT enzyme itself may have intrinsic anti-oxidant properties (7,8). It is therefore of considerable interest to understand better the in vivo effect of LCAT deficiency on oxidative stress and atherosclerosis in this mouse model. Vascular oxidant stress due to superoxide anion (O 2 . ) and other reactive oxygen species has been implicated in the development of atherosclerosis (9,10), possibly through being a major source of free radicals in the oxidative modification of low density lipoproteins (LDL) in the arterial wall (11). Excessive production of vascular O 2 . also attenuates the bioavailability of endothelial derived nitric oxide (NO), which may contribute to endothelial dysfunction and atherosclerosis (12). Increasing evidence, both in vivo and in vitro, suggests that NAD(P)H oxidase is a major contributor to the generation of O 2 . anions in the vascular wall (9,13). In addition to oxidized lipids, the NAD(P)H oxidase activity has also been shown to be modulated by a number of vasoactive peptides, cytokines, and growth factors (9). F2-isoprostanes are chemically stable prostaglandin isomers * This work was supported in part by Grants-in-aid N4124 (to D. S. N.) and T4027 (to P. W. C.) from the Heart and Stroke Foundation of Ontario, the Canada Foundation for Innovation New Opportunities fund, and St. Michael's Hospital Research Center research grant (to D. S. N.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  that result from non-enzymatic, free radical-mediated oxidative modification of arachidonic acid. These compounds can be found in tissues and many other body fluids including plasma and urine. These analytes have been shown to be excellent in vivo markers of oxidative stress both in human and in animal models (14).
In this study, by using the LCAT ko mouse model, we tested the hypothesis that LCAT deficiency is associated with increased oxidative stress by measuring plasma levels of glycerophosphocholine-bound F2-isoprostanes (PL-F2-isoP) and vascular O 2 . production using lucigenin-derived chemiluminescence (LDCL) in LCAT ko mice. We further studied the effect of LCAT deficiency on oxidative stress and atherosclerosis in apoE ko mice by crossbreeding the two strains.
Animals-LcatϪ/Ϫ mice were created in Dr. Rubin's laboratory as reported previously (6,15). ApoEϪ/ϪxlcatϪ/Ϫ mice were generated by first cross-breeding apoEϪ/Ϫ mice in C57BL/6 background (The Jackson Laboratory) with lcatϪ/Ϫ mice to yield apoEϩ/Ϫxlcatϩ/Ϫ. Brothersister matings of the F1 double heterozygous siblings were carried out. ApoEϪ/Ϫxlcatϩ/Ϫ were selected by PCR screening for subsequent serial breeding for 5 generations. All studies were carried out with the mice being fed a chow diet, and littermates were used as controls. All mouse protocols were approved by the Animal Care Committee at St. Michael's Hospital, Toronto, Canada.
Plasma Lipid Analyses-Plasma lipid analyses were performed on mice 7-10 weeks of age. Plasma was obtained as described previously (6). The 1.019 -1.063 g/ml ("LDL") fraction was obtained through ultracentrifugation (16). Fast protein liquid chromatography (FPLC) fractionation on total plasma and the LDL fraction was performed on a Superose 6HR column (10 mm ϫ 30 cm) (Amersham Biosciences) (17). Plasma and Superose fractions were analyzed on an RA-1000 (Bayer Diagnostics) using enzymatic assays for total cholesterol, triglycerides, glycerol blank, FC, and PL. HDL cholesterol was obtained from plasma after dextran sulfate precipitation (18).
Paraoxonase (PON1) activity in heparinized plasma and Superose fractions was measured as arylesterase activity in 7-10-week-old mice using phenylacetate as substrate as described previously (19).
Plasma Phospholipid-bound F2-Isoprostane Determination-Fasting blood from mice 8 to 12 weeks of age was obtained by tail bleed, and butylated hydroxytoluene was added to the plasma immediately after separation to a final concentration of 100 M and frozen at Ϫ80°C. The level of PL-F2-isoprostanes was determined by on-line normal phase high pressure liquid chromatography with on-line electrospray/mass spectrometry as described previously (20). Vascular Ring Superoxide Production by Lucigenin-derived Chemiluminescence (LDCL)-The LDCL method was modified from that of Pagano et al. (21). Briefly, mice 6 -10 months of age were anesthetized with Avertin before isolation of the aortic segments, starting from the aortic root and typically ending just distal to the renal artery (wet weight 11-16 mg). Each segment was cut into 5-mm rings before placing in an opaque 96-well microtiter plate in phosphate-buffered saline at pH 7.5 with NADH at 100 M followed by incubation at 37°C under 95% O 2 , 5% CO 2 for 30 min before counting with a Wallac Victor-II (PerkinElmer Life Sciences) in the luminometry mode. Lucigenin at a final concentration of 250 M was added and luminescence count recorded at 1-min intervals for 30 min. Chemiluminescence is reported as count/mg relative to a reference sample of 14 mg of wet weight. To validate these data, coelenterazine at a final concentration of 10 M was used in place of lucigenin.
Oxidizability of Mouse Plasma 1.019 -1.063 g/ml (LDL) Fraction-The 1.019 -1.063 g/ml fraction was isolated by ultracentrifugation of the pooled plasma from mice 6 -10 weeks of age for the genotypes apoEϪ/ Ϫxlcatϩ/ϩ and apoEϪ/ϪxlcatϪ/Ϫ and pooled plasma from normolipidemic human subjects. This fraction was then dialyzed against phosphate-buffered saline, pH 7.5, for overnight. To induce in vitro oxidation, 300 l of the LDL fraction was incubated with 5 M of CuSO 4 in phosphate-buffered saline at pH 7.5 for 30 h, and the degree of oxidation was monitored by absorbance at 234 nm.
Aortic Atherosclerosis Quantification-Mearsurement of aortic atherosclerotic lesions in mice was carried out by the en face method (22). Male mice in apoE ko background fed a chow diet were sacrificed at 8 -9 months of age. Aortae were dissected out intact from the aortic root to the femoral bifurcation, fixed overnight with formalin, followed by staining with Oil Red-O. Digital images were obtained with a Nikon Coolpix 880 camera (Nikon, Ontario, Canada). The Scion imaging software (Scion Inc.) was used to compute the lesion area, and all results were means of triplicate measurements. The severity is expressed as percent surface occupied by Oil Red-O-stained lesions.
Statistical Analyses-Comparison of group mean and S.D. was by Student's t test. Pearson statistics were used to evaluate correlation and two-way ANOVA was used for evaluating gene-gene interaction using the GraphPad Prism software (GraphPad Software Inc., San Diego) and a two-tailed p value of less than 0.05 was considered statistically significant.
Lipid Analyses-Lipoprotein analyses of the LCAT-deficient mice are summarized in Table I. The lipoprotein profile of the lcatϩ/ϩ, lcatϩ/Ϫ, and lcatϪ/Ϫ mice agreed with those reported previously (15). In the apoEϪ/Ϫ background, LCAT deficiency preserved the severely elevated plasma cholesterol. We also observed an LCAT gene dose-dependent decrease in the FC/CE ratio with and without the apoEϪ/Ϫ background. Lipoprotein analyses on the FPLC fractions showed a preservation of the accumulation of "VLDL" fractions in the apoEϪ/ϪxlcatϪ/Ϫ and apoEϪ/Ϫxlcatϩ/Ϫ mice in comparison with the apoEϪ/Ϫ xlcatϩ/ϩ control mice (Fig. 1). HDL-C was also severely reduced in apoEϪ/ϪxlcatϪ/Ϫ mouse plasma, but unlike the lcatϩ/Ϫ mice, its level was also markedly decreased in the apoEϪ/Ϫxlcatϩ/Ϫ mice (Table I and Fig. 1). On the other hand, the IDL/LDL shoulder (fraction 8) showed a 23% reduction in the apoEϪ/ϪxlcatϪ/Ϫ mice compared with apoEϪ/Ϫxlcatϩ/ϩ and apoEϪ/Ϫxlcatϩ/Ϫ littermates (Fig. 1), consistent with the significant up-regulation of the LDL receptor found in LCAT ko mice (23).
PON1 Arylesterase Activities-The total plasma PON1 arylesterase activities in different genotype groups are shown in Fig. 2a. A 46.5% reduction in PON1 activities was noted in the lcatϪ/Ϫ mice as compared with those of lcatϩ/ϩ mice, in agreement with our previous observation (6). Its reduction in the lcatϩ/Ϫ mice was not significant, suggestive of an autosomal recessive pattern in relation to the lcat mutant allele, which mirrors that of the plasma HDL-C levels. On the contrary, despite significantly lower HDL-C levels in both apoEϪ/Ϫ xlcatϩ/Ϫ and apoEϪ/ϪxlcatϪ/Ϫ mice, their total plasma PON1 activities were not significantly different from the apoEϪ/Ϫ xlcatϩ/ϩ control. Analysis of the PON1 activities in FPLC fractions showed that nearly half of the total plasma PON1 activity eluted in the non-HDL fractions of the apoEϪ/Ϫxl-catϪ/Ϫ mice. A lesser extent of redistribution was also observed in the plasma of apoEϪ/Ϫxlcatϩ/Ϫ mice (Fig. 2b).
Plasma PL-F2-isoP Levels-Because nearly 90% of the plasma isoprostanes exist in esterified form bound to phospholipids (14), the plasma level of F2-isoprostanes was determined by measuring the content of PL-F2-isoP for all six genotypic groups (Fig. 3). Compared with the wild type mice, plasma from lcatϪ/Ϫ mice showed a 4.9-fold increase (p ϭ 0.003) in plasma PL-F2-isoP. The lcatϩ/Ϫ mice showed intermediate levels but were not statistically different from those of the lcatϪ/Ϫ mice. In contrast, apoEϪ/Ϫxlcatϩ/ϩ mice showed a 2.1-fold increase (p ϭ 0.04) in plasma PL-F2-isoP as compared with the wild type mice, in excellent agreement with the 2-fold increase in the plasma level of iPF2␣-VI as reported previously (24). Sur-prisingly, the apoEϪ/ϪxlcatϪ/Ϫ mouse plasma PL-F2-isoP level was normalized. It is also of interest to note that the PL-F2-isoP level of apoEϪ/Ϫxlcatϩ/Ϫ mice was comparable with that of the apoEϪ/ϪxlcatϪ/Ϫ mice, suggestive of an autosomal dominant effect.
CuSO 4 -induced Oxidizability of LDL-We observed a comparable lag time between the LDL fraction from apoEϪ/Ϫ xlcatϩ/ϩ mice and that of a pooled normal human LDL. However apoEϪ/ϪxlcatϪ/Ϫ LDL showed a prolongation of the lag time by 60.0 min. The rate of rise of the absorbance and the level of inflection point were similar between the two mouse strains.
Aortic Ring Superoxide Production Using LDCL-Vascular ring O 2 . production was determined using LDCL on all six genotypic groups of mice (Fig. 4). Compared with lcatϩ/ϩ mice, the AUC for lcatϪ/Ϫ mice was found to be increased 3.6-fold (p ϭ 0.00006) and that for lcatϩ/Ϫ mice 2.9-fold (p ϭ 0.003). Likewise, we also observed a 2.2-fold increase (p ϭ 0.000001) in LDCL-AUC of the apoEϪ/Ϫxlcatϩ/ϩ mice when compared with the lcatϩ/ϩ mice. Again, the LDCL-AUCs for the apoEϪ/ ϪxlcatϪ/Ϫ mice were paradoxically normalized (1.26-fold change from the lcatϩ/ϩ mice, p ϭ 0.1), which paralleled the normalization of the plasma PL-F2-isoP in the same strain. The LDCL-AUC for apoEϪ/Ϫxlcatϩ/Ϫ mice was found to be subnormal (0.6-fold, p ϭ 0.01). This also correlates well with the normal PL-F2-isoprostane level observed in this same genotypic group as stated above. The validity of using lucigenin at high concentrations raised concern because of the potential for superoxide recycling, and this potential artifact is absent when using coelenterazine as a marker (25). All but one genotype group of mice were evaluated using coelenterazine as a superoxide sensor with n Ն 3 in each group. We observed a near-unity correlation (r 2 ϭ 0.9265, p ϭ 0.00086) between the LDCL-AUC and coelenterazine-derived chemiluminescence-AUC, thus validating the use of lucigenin.

DISCUSSION
In this paper, we report the observation of marked enhancement in oxidative stress in chow-fed lcatϪ/Ϫ mice using two independent measurements, namely plasma PL-F2-isoP level and aortic ring O 2 . production. When bred into the apoEϪ/Ϫ mouse background, we observed a paradoxical normalization of the same oxidative markers. This paradoxical change was found to be associated with a retention and redistribution of PON1 arylesterase activities into the non-HDL fractions based on FPLC separation. Furthermore, the normalization of the oxidative markers in the apoEϪ/Ϫxlcat Ϫ/Ϫ mice was also found to be associated with a significant reduction in atherosclerotic lesions in the male apoEϪ/ϪxlcatϪ/Ϫ mice as compared with its age-and sex-matched apoEϪ/Ϫxlcatϩ/ϩ control. We present the first in vivo evidence showing a marked increase in plasma PL-F2-isoP level and aortic O 2 . production (LDCL-AUC) in the lcatϪ/Ϫ mice. The increase is related to the lcat mutant allele in an autosomal dominant fashion, in association with an inverse trend in both HDL-C and PON1 activities, although the latter two are affected in an autosomal recessive fashion. Despite being severely hypolipidemic, the increases in both oxidation markers in the lcatϪ/Ϫ mice are substantially higher than those in the extremely hyperlipidemic apoEϪ/Ϫxlcatϩ/ϩ mice. These data suggest that the enhanced oxidative stress in the lcatϪ/Ϫ mice is modulated by factors beyond plasma level of lipoproteins. On the one hand, the inverse relationship between these oxidative markers and plasma PON1 activities suggest that this circulating antioxidant likely plays an important role, consistent with the recent findings by Shih et al. (4,5). The diverging effect of the lcat mutant allele in the heterozygotes underscores the importance of additional modulating factors. In addition to PON1, several HDL-associated proteins have been found to have antioxidant properties, and it is conceivable that the marked reduction of HDL levels in the lcatϪ/Ϫ mice may also lower the activities of these proteins and contribute to the overall increase in the oxidative stress observed. Paraoxonase 3 (PON3) has recently been cloned in a number of species including human, rabbit, and mouse (26). This protein is expressed primarily in the liver and the kidney and circulates in association with HDL. In vitro studies suggest that PON3 is capable of not only inactivating the preformed oxidized LDL but also preventing their formation (27). Although its antioxidant effects have not been established in vivo, it is conceivable that the low HDL in the lcatϪ/Ϫ mice may be associated with a significant reduction in serum PON3 levels and activities, contributing to the marked increase in oxidative stress seen in the lcatϪ/Ϫ mice. Platelet-activating factor acetylhydrolase (PAF-AH) is a 45-kDa protein secreted primarily by macrophages and circulates in both LDL and HDL in humans but exclusively in HDL in mice. This enzyme has the ability to hydrolyze not only PAF but also oxidized phospholipids (28). Recent studies (29) demonstrated that overexpression of circulating PAF-AH in apoE knockout mice results in a reduction in oxidized ␤-VLDL and atherosclerosis. We reported previously (6) that PAF-AH activity was significantly reduced in the lcatϪ/Ϫ mice. Although the in vivo effect of PAF-AH deficiency in mice has not yet been reported, it is conceivable that a reduced plasma level of PAF-AH activity may contribute to the overall oxidative stress.
The lipid profiles of the apoEϪ/Ϫxlcatϩ/Ϫ and apoEϪ/Ϫ xlcatϪ/Ϫ mice were both characterized by a preservation of the severe hyperlipidemia of the apoEϪ/Ϫxlcatϩ/ϩ control. In the apoE ko background, the effect of lcat mutant allele on HDL-C, PON1 activities, and the oxidation markers are even more divergent. The lcat mutant gene dose leads to a lowering of HDL-C in an autosomal dominant fashion but has no effect on PON1 activity (Table I and Fig. 2). The surprising finding in the double ko mice is the paradoxical normalization of plasma PL-F2-isoP and aortic O 2 . production in both apoEϪ/Ϫxlcatϩ/Ϫ and apoEϪ/ϪxlcatϪ/Ϫ mice. Despite such dramatically discordant findings, the correlation between mean PL-F2-isoP and LDCL-AUC among the six genotypical groups is remarkably linear (r 2 ϭ 0.94, p ϭ 0.0014). Another novel finding is the significant redistribution of PON1 activities into the non-HDL fractions in the apoEϪ/ ϪxlcatϪ/Ϫ mice, sustaining the total PON1 activities. Previous studies (30,31) in human subjects with a variety of HDL deficiency states, excluding LCAT-deficient subjects, failed to document redistribution of PON1 activities into the non-HDL fractions. A redistribution of PON1 arylesterase activity to the non-HDL particles may be a unique feature of complete LCAT deficiency. A recent study by Sorenson et al. (32) demonstrated that PON1 can bind to PL vesicles via its hydrophobic Nterminal signal sequence peptide and remains active enzymatically without apoAI. The authors further demonstrated that HDL-associated PON1 can be transferred to these PL vesicles. The physical characteristics of such synthetic vesicles are not dissimilar to that of lipoprotein X (LpX). In our apoEϪ/Ϫ xlcatϪ/Ϫ mice, we isolated LpX on the basis of an identification of an FC-, PL-rich, and CE-poor VLDL-like Superose peak in the 1.019 -1.063 g/ml plasma fraction (33). In this plasma fraction, a small but definite quantity of PON1 activity co-eluted with this peak. However, inconsistent recovery of PON1 activity after ultracentrifugation precluded a more quantitative measurement. For the same reason, current data do not exclude association of PON1 with other non-HDL lipoprotein classes. Nonetheless, the redistribution of PON1 observed should be considered as one of the most biologically plausible explanations for the paradoxical normalization of oxidative stress in the double knockout mice. Similarly, redistribution of PON3 and/or PAF-AH in LCAT-deficient mice should also be considered. In the case of PON3, a redistribution of the protein to non-HDL particles has not been reported. Due to the lack of a specific assay for murine plasma PON3 (26), this possibility cannot be determined with certainty in the apoEϪ/ϪxlcatϪ/Ϫ mice. In humans, it is the PAF-AH that is associated with circulating LDL that is biologically active. It has been shown that human PAF-AH associates with the C terminus portion of apoB, and the sequences mediating this binding are altered in murine PAF-AH, consistent with its absence from murine apoB lipoproteins (34). It is therefore unlikely that PAF-AH would associate with non-HDL in the apoEϪ/ϪxlcatϪ/Ϫ mice.
The mechanism by which the non-HDL-associated PON1 may impart antioxidative action is unclear. The protective role of PON1 against Cu 2ϩ -induced oxidation of LDL has been well established (26,35). We observed a resistance of the 1.019 -1.063 g/ml plasma fraction from the apoEϪ/ϪxlcatϪ/Ϫ mice to Cu 2ϩ -induced oxidation in conjunction with the detection of PON1 activities co-eluting with LpX in the same fraction. This is consistent with the notion that the vesicle (LpX)-associated PON1 retains not only its arylesterase activity but also its antioxidative properties. PON1 activity has been detected in interstitial fluid (36), a site where HDL-associated PON1 is most likely to impart its anti-oxidative actions. It is conceivable that the non-HDL-associated PON1 may also enter the interstitial space, reducing the degree of oxidative modification in the vessel wall. Because oxidized lipids have been shown to stimulate the NADPH oxidase system (9), this may explain the observed paradoxical normalization of O 2 . in the apoEϪ/Ϫ xlcatϪ/Ϫ mouse aortae. The two oxidative markers, mean LDCL-AUC and PL-F2-isoP, are both inversely related to mean plasma PON1 activities among the six groups. However, based on the opposite effects of the two apoE alleles on the lcat heterozygotes, the correlations between them are poor as expected. This is largely the result of the diverge influence of apoE genotype on the lcat heterozygotes as reflected by the strong gene-gene interactions based on the two-way ANOVA analyses. Exclusion of these two heterozygous groups would have unmasked a remarkably linear inverse relationship (r 2 Ͼ 0.85 and p Ͻ 0.05) in both cases, underscoring the complexity of gene-gene interactions on oxidative stress in heterozygous LCAT deficiency. In addition, the PON1 in the HDL and non-HDL fractions are bound to very different lipoproteins, and their antioxidant activities may be differentially modulated, further contributing to the poor correlation.
Recent studies suggest that the NAD(P)H oxidase system is a major source of arterial O 2 . (13), and oxidized LDL has been shown to be one of the potent stimulators of this enzyme (9). The strong correlation between PL-F2-isoP and aortic O 2 . production in the LCAT-deficient mice therefore suggests a direct mechanistic link. It is conceivable that aortic vessel wall NAD(P)H oxidase activity and circulating oxidized lipids may be fueling a self-perpetuating cycle, establishing a unique steady state for each genotype. The resistance of the apoEϪ/ ϪxlcatϪ/Ϫ mouse LDL to Cu 2ϩ -induced oxidation may have contributed to the attenuation of the fueling of this cycle and the normalization of the oxidative markers in these mice. Oxidative stress has been shown to be important in mouse models of atherosclerosis. In apoE ko mice, oral supplementation with vitamin E resulted in concomitant reductions in plasma, urinary, and arterial isoprostane levels in association with a significant reduction in aortic atherosclerosis, without altering the plasma cholesterol level (24). In the present report, our finding of a significant reduction in spontaneous aortic atherosclerosis in male apoEϪ/ϪxlcatϪ/Ϫ mice, when compared with that of age-and gender-matched apoEϪ/Ϫxlcatϩ/ϩ mice, is in agreement with Lambert et al. (23) despite a difference in genetic backgrounds. Although we found a 23% reduction in plasma IDL and LDL levels in our apoEϪ/ϪxlcatϪ/Ϫ mice, the extent is unlikely to account completely for the normalization of the aortic O 2 . production. Our data therefore suggest that the increased resistance of the LDL to oxidative modification in the apoEϪ/ϪxlcatϪ/Ϫ mice is likely important for the observed reduction in atherosclerotic lesions, and this may in turn be a result of the retention and redistribution of PON1 in the hyperlipidemic LCAT-deficient mice. Furthermore, the non-linearity in the association between plasma PON1 activities and the oxidative stress markers in the heterozygous LCAT-deficient mice suggests that the role of the antioxidants, including PON1, in atherosclerosis is complex in LCAT deficiency.