ABCA1 Overexpression in the Liver of LDLr-KO Mice Leads to Accumulation of Pro-atherogenic Lipoproteins and Enhanced Atherosclerosis*

The identification of ABCA1 as a key transporter responsible for cellular lipid efflux has led to considerable interest in defining its role in cholesterol metabolism and atherosclerosis. In this study, the effect of overexpressing ABCA1 in the liver of LDLr-KO mice was investigated. Compared with LDLr-KO mice, ABCA1-Tg × LDLr-KO (ABCA1-Tg) mice had significantly increased plasma cholesterol levels, mostly because of a 2.8-fold increase in cholesterol associated with a large pool of apoB-lipoproteins. ApoB synthesis was unchanged but the catabolism of 125I-apoB-VLDL and -LDL were significantly delayed, accounting for the 1.35-fold increase in plasma apoB levels in ABCA1-Tg mice. We also found rapid in vivo transfer of free cholesterol from HDL to apoB-lipoproteins in ABCA1-Tg mice, associated with a significant 2.7-fold increase in the LCAT-derived cholesteryl linoleate content found primarily in apoB-lipoproteins. ABCA1-Tg mice had 1.4-fold increased hepatic cholesterol concentrations, leading to a compensatory 71% decrease in de novo hepatic cholesterol synthesis, as well as enhanced biliary cholesterol, and bile acid secretion. CAV-1, CYP2b10, and ABCG1 were significantly induced in ABCA1-overexpressing livers; however, no differences were observed in the hepatic expression of CYP7α1, CYP27α1, or ABCG5/G8 between ABCA1-Tg and control mice. As expected from the pro-atherogenic plasma lipid profile, aortic atherosclerosis was increased 10-fold in ABCA1-Tg mice. In summary, hepatic overexpression of ABCA1 in LDLr-KO mice leads to: 1) expansion of the pro-atherogenic apoB-lipoprotein cholesterol pool size via enhanced transfer of HDL-cholesterol to apoB-lipoproteins and delayed catabolism of cholesterol-enriched apoB-lipoproteins; 2) increased cholesterol concentration in the liver, resulting in up-regulated hepatobiliary sterol secretion; and 3) significantly enhanced aortic atherosclerotic lesions.

The inverse relationship between plasma high density lipoprotein-cholesterol (HDL-C) 3 concentrations and the incidence of coronary artery disease is well established in humans (1)(2)(3)(4)(5). The identification of the ATP-binding cassette transporter A1 (ABCA1) as the molecular defect in Tangier disease (6 -12), a disease characterized by plasma HDL deficiency and increased susceptibility to atherosclerosis, has led to major advances in our understanding of the formation of plasma HDL and its role in reverse cholesterol transport (RCT) (13). RCT is considered an anti-atherogenic process, in which excess cholesterol is transported in the form of HDL from peripheral tissues to the liver, where it can be excreted from the body directly or indirectly after conversion into bile acids (13,14). In vitro studies have shown that ABCA1 facilitates the transfer of excess cellular cholesterol and phospholipids to apolipoprotein (apo) acceptors, such as apoA-I and apoE, resulting in the formation of nascent or pre-␤ HDL (15)(16)(17). Consistent with these findings, overexpression of ABCA1 in C57Bl/6 mice led to increased plasma-and HDL-cholesterol, as well as increased apoA-I and apoA-II levels compared with wild-type mice (18 -20). In agreement with the proposed function of ABCA1 in HDL formation and RCT, ABCA1-deficient mice (21-23) had a striking decrease in total plasma cholesterol and phospholipids, with virtually undetectable levels of HDL. Thus, ABCA1-mediated formation of nascent HDL represents a critical initiating step in the atheroprotective mechanism of RCT, followed by HDL maturation via LCAT, which converts cholesterol to the more hydrophobic cholesterol esters (CE) present in the HDL core.
Because ABCA1 enhances cellular cholesterol efflux and facilitates RCT, therapeutic up-regulation of ABCA1 gene expression has been anticipated to promote the prevention and regression of atherosclerosis. Unexpectedly, studies in various mouse models have led to divergent results regarding the antiatherogenic role of ABCA1 expression. Complete ABCA1 deficiency in wild type, LDLr-KO, and apoE-KO mice did not alter atherosclerosis (22,24), but repopulation of ABCA1-KO mice with wild-type macrophages (24) or macrophages from ABCA1-overexpressing mice (25) led to significantly decreased atherosclerosis. In contrast, selective inactivation of macrophage ABCA1 in mice results in markedly increased atherosclerosis (24,26). Transgenic mice overexpressing a human ABCA1 cDNA under the control of the apoE promoter in both liver and macrophages developed less atherosclerosis in C57Bl/6 mice when fed a high fat-high cholesterol diet containing cholate (19), but showed increased atherosclerosis when crossed into an apoE-KO background (19). In contrast, Singaraja et al. (27) reported decreased atherosclerosis in apoE-KO mice, which overexpressed human ABCA1 from a human ABCA1-BAC transgene. The ABCA1-BAC transgene contained endogenous regulatory promoter elements and led to a different expression profile than the transgene construct used by Joyce et al. (19).
Although the underlying mechanisms for these contrary results remain unclear, the different promoters (28) regulating the tissue expression levels of the human ABCA1 transgenes may provide a partial explanation for the opposing results in atherosclerosis. However, an increase in the pro-atherogenic plasma apoB-containing lipoproteins (apoB-Lps) was observed in these as well as other studies, including C57Bl/6 as well as apoE-KO mice transgenic for ABCA1 (18 -20), and C57Bl/6 mice with low level (29) or high level (30) adenoviral ABCA1 expression in the liver. Conversely, ABCA1 deficiency in apoE-KO or LDLr-KO mice led to decreased levels of the plasma apoB-Lps (24).
Another important finding that resulted from studies using adenoviral liver-specific ABCA1 overexpression (29,30) or down-regulation (31), was the demonstration that hepatic ABCA1 is primarily responsible for regulating the level of plasma HDL. Recent findings in mice with selective inactivation of ABCA1 in either liver (32) or intestine (33) have confirmed the role of hepatic ABCA1 as the primary source of plasma HDL-cholesterol as well as a critical factor for the HDL plasma residence time, accounting for ϳ70 -80% of plasma HDL-cholesterol in mice on a chow diet, with the remaining fraction of HDL coming largely from the intestine (33).
Taken together, these studies revealed that macrophage ABCA1 expression does not contribute to plasma HDL levels but does control the development of atherosclerosis in apoE-KO and LDLr-KO mice, while the role of hepatic ABCA1 expression in the development of atherosclerosis and the regulation of non-HDL lipoprotein formation has not yet been fully established.
In the present study, LDLr-KO mice overexpressing human ABCA1 in the liver were used to investigate the effect of hepatic ABCA1 on hepatic cholesterol balance, apoB-Lp metabolism and the development of atherosclerosis. We report that hepatic overexpression of ABCA1 in LDLr-KO mice consuming a chow diet leads to significantly increased hepatic cholesterol levels, resulting in the induction of compensatory sterol secretion and reduced cholesterol synthesis. The plasma lipoprotein profile in ABCA1-transgenic LDLr-KO mice was found to be highly pro-atherogenic with accumulation of cholesterol-enriched apoB-Lps because of delayed apoB-Lp catabolism and rapid transfer of free cholesterol from HDL to apoB-Lps. In accordance with the plasma lipoprotein profile changes, aortic atherosclerosis was significantly increased in ABCA1-Tg mice compared with LDLr-KO mice. These results establish that hepatic up-regulation of ABCA1 in LDLr-KO mice, a mouse model with genetic abnormality in apoB-clearance, is pro-atherogenic and further support the concept that genetic background as well as tissue-specific regulation of ABCA1 are critical factors that modulate the atheroprotective properties of ABCA1.

EXPERIMENTAL PROCEDURES
Animals-Human ABCA1-Tg mice were produced with a construct containing the human ABCA1 cDNA driven by the murine apoE promoter, as previously described (18). These mice were crossed to homozygosity with C57Bl/6 LDLr-KO mice. Expression of human ABCA1 was determined by dot blot hybridization, as previously described (19). LDLr genotype status was determined by PCR screening, using mouse LDLr-specific PCR primers that generate a 383-bp PCR amplicon (5Јaccccaagacgtgctcccaggatga-3Ј and 5Ј-cgcagtgctcctcatctgacttgt-3Ј), and neo-specific PCR primers that generate a 200-bp amplicon (5Ј-aggatctcgtcgtgacccatggcga-3Ј and 5Ј-gagcggcgataccgtaaagcacgagg-3Ј). LDLr-KO and human ABCA1-Tg mice were maintained on a chow diet containing 0.02% cholesterol and 4% fat, or were placed on a Western Diet (TD88137; Harlan Teklad; Madison, WI), containing 0.2% cholesterol and 21.2% fat for either 4, 9, or 12 weeks prior to sacrifice. The mice were housed under protocols approved by the Animal Care and Use Committee of the National Heart, Lung, and Blood Institute.
Real-time PCR-Total RNA from liver, small intestine, kidney, heart, adrenals, testis, lung, brain, and spleen from agematched male mice was isolated using TRIzol (Invitrogen, Carlsbad, CA) and further purified using RNeasy RNA-binding columns with on-column DNase treatment (Qiagen, Valencia, CA). Total RNA from peritoneal macrophages was extracted from a pool of macrophages (n ϭ 6 male mice per genotype) after allowing the cells to attach for 3 h on Primaria cell culture dishes (BD Biosciences, Franklin Lakes, NJ) in DMEM containing 10% fetal bovine serum, 4.5 g of glucose/liter and penicillin/ streptomycin/L-glutamine (Sigma). Cells were washed, and RNA was isolated with RNeasy RNA-binding columns followed by an on-column DNase treatment (Qiagen).
Western Blot Analysis-Protein analyses of m, hABCA1, mABCG1, mCYP27␣1, mLRP, mSR-BI, and mCAV-1 were performed by homogenizing ϳ100 mg of snap-frozen liver samples in cell lysis buffer containing 25 mM Tris, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, and 1ϫ protease inhibitor (Roche Applied Science, Indianapolis, IN) as well as 50 M calpain inhibitor II (Sigma). Macrophage protein was isolated from pools of macrophages of six male mice from each group, after allowing the cells to attach to Primaria cell culture dishes (BD Biosciences) for 3 h in DMEM containing 10% fetal bovine serum, 4.5g glucose/liter, and penicillin/streptomycin/L-glutamine (Sigma). Cells were washed, and protein was isolated as described above.
Protein concentrations were determined by BCA assay (Pierce), and 5-20 g of protein were separated under reducing conditions on a Novex 3-8% Tris-acetate gel or 4 -12% Bis-Tris gel (Invitrogen, Carlsbad, CA). After transfer to polyvinylidene difluoride membranes (Invitrogen), blots were incubated with the following antibodies: rabbit anti-mouse LRP (Resarch Diagnostics Inc.), rabbit anti-human/mouse ABCA1 and SR-BI (Novus, Littleton, CO), rabbit anti-human/mouse ABCG1 (E-20; Santa Cruz Biotechnology) and rabbit anti-human/ mouse caveolin-1 (Abcam Inc., Cambridge, MA) according to the manufacturer's protocol. Rabbit anti-human/mouse CYP27␣1 was provided by Dr. David Russell, University of Texas Southwestern Medical Center, Dallas, TX and was diluted 1:2000 before use (1 h at room temperature). Corresponding horseradish peroxidase-linked secondary antibodies were purchased from Amersham Biosciences and from Santa Cruz Biotechnology for all other primary antibodies. Visualization of chemiluminescence was performed with Western Lightning Reagent (PerkinElmer Life Sciences). Anti-Human/mouse ␤-actin (BioLegend, San Diego, CA) was used to control for equal loading and transfer. If not otherwise indicated, equal amounts of protein were analyzed. Band intensities were quantified using the Image Quant 5.2 software (Molecular Dynamics, Sunnyvale, CA).
Efflux Studies-Mouse peritoneal macrophages were seeded at a density of ϳ400,000 cells per well, using a 24-well Primaria cell culture plate (BD Biosciences), and maintained in DMEM medium containing 10% LPDS, 0.3 m 8-Br-cAMP, and penicillin/streptomycin/L-glutamine (Sigma). Macrophages were loaded with medium containing 10 g/ml [ 3 H]CE-labeled acetylated-LDL for 24 h. Efflux of [ 3 H]cholesterol was performed as previously described (34), using either HDL (50 g/ml), apoA-I (10 g/ml), or bovine serum albumin alone (0.2%) as acceptors. Data were normalized to cell protein and expressed as percent of total counts. Preparation of hepatocytes was performed as previously described (29).
Native gel electrophoresis of pooled plasma was performed using the Titan Gel Lipoprotein Electrophoresis System (Helena Laboratories, Beaumont, TN). Plasma levels of apoB were quantitated by sandwich ELISA (36), using polyclonal antibodies raised against purified mouse apoB. In addition, mouse apolipoproteins A-I, A-II, B, and E were also identified by immunostaining plasma lipoproteins of pooled plasma after separation on 4 -20% Novex Tris-glycine or 3-8% Tris acetate gels (Invitrogen) and transfer to polyvinylidene difluoride membranes (Millipore Corporation, Bedford, MA). Polyclonal rabbit anti-mouse IgGs purchased from Biodesign International (Kennebunkport, ME) were used as primary antibodies. Supersignal Chemiluminescent Reagent (Pierce) and biotinylated secondary antibody, included in the Vectastain Elite ABC Kit (Vector Laboratories, Burlingame, CA) were used for visualization.
Lipoprotein Isolation-VLDL, LDL, and HDL were isolated from 2.5 ml of fasted mouse plasma by sequential ultracentrifugation (4°C, 100,000 rpm for 5 h) after density adjustment with KBr (VLDL, d ϭ 1.0006 g/ml; LDL, d ϭ 1.019 -1.063 g/ml; and HDL, d ϭ 1.063-1.21 g/ml), using a TLA-100.2 rotor on a Beckman tabletop ultracentrifuge (Beckman Instruments, Palo Alto, CA). After extensive dialysis against phosphate-buffered saline containing EDTA and sodium azide, particle integrity was confirmed by FPLC profiling and native agarose gel as described above.
Fatty Acid Profile of Plasma Cholesterol Esters-Plasma cholesterol esters were isolated by thin-layer chromatography, following Bligh-Dyer extraction of plasma total lipids (37). The cholesterol ester fatty acid moieties were converted to volatile cholesterol ester fatty acid methyl esters (FAME) by the method of Metcalfe et al. (38). FAME were separated by GC (39).
In Vivo HDL-C Transfer-Autologous mouse plasma (900 l) was incubated for 1 h with dried [ 3 H]cholesterol (12.5 Ci) on sterile filter paper at 4°C on an orbital shaker. HDL was isolated as described above and homogeneous labeling of HDL was ascertained by FPLC profiling and quantification of 3 H (Cytoscint scintillation mixture, Fisher Scientific International Inc.), using a Tri-Carb 2500 TR liquid scintillation counter (Packard, Downers Grove, IL). Three male mice per genotype were injected 3-6 ϫ 10 5 dpm into the saphenous vein. Blood was collected by retro-orbital bleeding at 3-and 10-min postinjection, EDTA plasma was isolated, and 100 l of plasma were subjected to FPLC analysis, as described above. 3 H dpm were counted in each fraction and plotted against the fraction elution volume.
ApoB Synthesis Studies-ApoB production was measured as previously described (18). Briefly, fasted (for hepatic apoB production) or non-fasted (for intestinal apoB production) male LDLr-KO and ABCA1-Tg mice (n ϭ 5 each group) were injected with Triton X-100 (Sigma) to block lipolysis, and VLDL-and chylomicron-apoB secretion into plasma was followed within 2 h, using [ 35 S]methionine as radioactive tracer.
ApoB Catabolic Studies-Purified mouse VLDL and LDL fractions were labeled with 125 I using the iodine monochloride method (41) and analyzed by FPLC to ensure particle integrity. Briefly, ϳ1.5 ϫ 10 6 dpm of the purified labeled particles were injected, and the remaining plasma counts at 30 min, 90 min, 3 h, 6 h, 12 h, and 24 h were measured and expressed as a percentage of radioactivity remaining in plasma 3 min after saphenous vein injection.
Northern Blot Analysis-Total liver RNA was isolated using the Ultraspec RNA Isolation System (Biotecx, Houston, TX). Ten micrograms of total liver RNA were separated on a 1% glyoxal gel and transferred to a 0.45-m Nytran membrane (Schleicher and Schuell). Hybridizations were performed in PerfectHyb Plus Hybridization Buffer (Sigma) using cDNA probes generated by RT-PCR with the indicated forward and reverse PCR primer pairs: 3-hydroxy-3-methylglutaryl-coenzyme A reductase (Hmgcr, 5Ј-tggcaccatgtcaggcgtcc-3Ј and 5Ј-tgtgtccccagggcggatgg-3Ј) and ␤-actin cDNA (Ambion) was used for standardization.
Hepatic and Fecal Sterol Content-Liver cholesterol was quantified following Folch extraction (ϳ200 mg of wet tissue) by a fluorescent enzymatic assay (42). Fecal neutral sterol determination was performed by GC analysis, as well as by a fluorescent enzymatic assay (43).
In Vivo Cholesterol Synthesis-The rates of in vivo cholesterol synthesis in the liver and intestine during mid-dark cycle were measured after intraperitoneal injection of 10 mCi of 3 H 2 O (PerkinElmer Life Sciences) as described (44,45).
Bile Cannulation-The common bile duct of studied mice was cannulated and bile was collected as previously described (43). Biliary cholesterol, phospholipids, and bile acid concentrations were determined with colorimetric kits (cholesterol E, phospholipids B, total bile acid kit, Wako).
Fractional Intestinal Cholesterol Absorption-Cholesterol absorption studies were performed in male mice (4 -8 months of age), using the dual isotope plasma ratio method (46,47).
Microarray Analysis-Liver samples taken at mid-dark cycle (n ϭ 4 male mice each group) were subjected to RNA expression analysis. Total RNA was isolated with TRIzol (Invitrogen) and further purified, using the RNeasy RNA-binding columns (Qiagen, Valencia, CA) followed by an on-column DNase treatment (RNase-free DNase, Qiagen). Purified RNA was reversetranscribed and amplified for microarray analysis, using the protocol and reagents from the RiboAmp OA kit (Arcturus, Mountain View, CA). Amplified cDNA samples were hybridized to Affymetrix 430A GeneChips according to the manufacturer's protocol (Affymetrix). Expression array data were normalized using Affymetrix GCOS 1.2/MAS 5.1 and scaled to a target value of 1000. Expression data and present/absent flags were imported into GeneSpring 7.3 (Agilent, Redwood City, CA) for subsequent analysis. Robust Multichip Average, RMA, was used to estimate the expression level for all genes. The total set of genes was reduced to those with a present call in all arrays of either group. Gene lists were constructed containing 2-fold or greater differences between the two groups and with Benjamini-Hochberg false discovery rate of Ն 50%. Subsets were analyzed and imported into Ingenuity (Ingenuity, San Diego, CA) and MetaCore (GeneGo, Minneapolis, MN) pathway analysis tools. The total set of data is accessible through the GEO repository as series ID GSE5496.
Analysis of Aortic Lesions-Atherosclerotic lesion formation was quantitated from the heart and the attached section of ascending aorta as previously described (48).
Statistics-Data are expressed as mean Ϯ S.E. Statistically significant differences between ABCA1-Tg and control mice were assessed by a two-tailed Student's t test. Non-parametric data were analyzed by the Mann-Whitney test (Instat Software, Graphpad Inc, San Diego, CA).

Hepatic Overexpression of ABCA1 in LDLr-KO Mice-To
further elucidate the role of hepatic ABCA1 in lipid metabolism and atherosclerosis, we studied the effect of ABCA1 overexpression in mice lacking the LDL-receptor (LDLr) consuming a rodent chow diet. Human ABCA1-Tg ϫ LDLr-KO mice, which will be referred to as ABCA1-Tg mice, were produced by crossing the previously established human ABCA1-Tg (18) with C57Bl/6 LDLr-KO mice. The original human ABCA1 transgene had been generated by using a construct containing the human ABCA1 cDNA driven by the murine apoE promoter (18). Fig. 1A depicts mRNA levels of mouse Abca1 in liver, small intestine, macrophages, kidney, heart, adrenal, testis, lung, brain, and spleen of both mouse study groups as well as the mRNA levels of the human ABCA1 transgene in ABCA1-Tg mice using real-time PCR analysis. Human ABCA1 mRNA was found to be overexpressed in livers of ABCA1-Tg mice, leading to a 2-fold increase of ABCA1 (mouse and human) mRNA levels (Fig. 1A), while human ABCA1 expression was very low in all other investigated tissues, including macrophages (Fig. 1A). As anticipated, the human ABCA1 mRNA was not detected in LDLr-KO mouse tissues (data not shown). Furthermore, overexpression of human ABCA1 did not significantly alter the mRNA expression levels of endogenous mouse ABCA1 in LDLr-KO mice (Fig. 1A) compared with control mice (Fig. 1A).
This striking result of very low level hABCA1 transgene expression in macrophages of ABCA1-Tg mice led us to use different primer and probe sets, which detected the 3Ј-and 5Ј-end, as well as exon 30 -31 of the human ABCA1 mRNA (data not shown). However, in contrast to the parental transgenic mouse line (18), our ABCA1-Tg model again revealed only low levels (20%) of hABCA1 transcripts in macrophages (Fig. 1A). Western blot analysis (Fig. 1B) confirmed these findings of a liver specific ABCA1-overexpressing mouse model. By using an antibody that detects both mouse and human ABCA1, ABCA1 protein levels were increased only in livers (2.3-fold; To assess the functionality of the ABCA1 expression, efflux of radiolabeled [ 3 H]cholesterol from mouse hepatocytes as well as macrophages to apoA-I and HDL was examined (Fig. 1C). Cholesterol efflux to apoA-I, but not to HDL, was increased by 3-fold ( p Ͻ 0.05) in ABCA1-Tg compared with LDLr-KO hepatocytes (Fig. 1C). In contrast, apoA-I-and HDL-mediated cholesterol efflux were similar in peritoneal macrophages of ABCA1-Tg versus LDLr-KO mice (Fig. 1C). As a result of this unique phenotype we were able to evaluate the effects of ABCA1 overexpression exclusively in the liver.
Native agarose gel electrophoresis confirmed increased ␤-migrating apoB-Lps (Fig. 2B), and FPLC analysis demonstrated that the increase in plasma TC in ABCA1-Tg mice was mostly caused by the accumulation of cholesterol in all non-HDL fractions, including VLDL, IDL, and LDL (fractions 1-5, Fig. 2C). Representative FPLC fractions were analyzed for apolipoprotein expression by Western blot analysis (Fig. 2C, inset). Increased apoB-48 protein levels were found in the LDL (fractions 3-4, Fig. 2C), and increased apoA-I was found in virtually all analyzed FPLC (fractions 1-7, Fig. 2C inset) of ABCA1-Tg mice compared with nontransgenic LDLr-KO mice. Our finding that apoA-I is present not only in the HDL but also in the apoB-Lp fractions has been previously reported in mice (49,50).
Similar increases were observed in plasma lipids of female ABCA1-Tg mice (supplemental Fig. S1A), including TC (1.8fold), TG (1.2-fold), PL (1.7-fold), FC (2.2-fold), CE (1.7fold), non-HDL-C (2-fold), and HDL-C (1.5-fold) in female ABCA1-Tg versus LDLr-KO mice ( p Ͻ 0.05 all). However, in contrast to males, female ABCA1-Tg mice did not exhibit a significant cholesterol increase in the VLDL fraction (supplemental Fig. S1B). These data demonstrate that in the absence of the LDLr, hepatic ABCA1 overexpression leads to increased plasma cholesterol levels mostly in apoB-Lps in both male and female ABCA1-Tg mice on a regular chow diet. The 2.8-fold increase in apoB-Lp-cholesterol together with a comparably only modest 1.4-fold increase in plasma apoB levels ( Fig. 2A) suggests that hepatic ABCA1 overexpression in LDLr-KO mice results in a modest expansion of the plasma pool of apoB-Lp particles that are markedly cholesterol-enriched. Conversion of our cholesterol and apoB data to moles revealed that on average, each apoB-Lp particle in ABCA1-Tg mice contained 1.6fold more cholesterol (2.4-fold CE; 0.9-fold FC (mol/mol apoB)) than the apoB-Lp particles in LDLr-KO control mice.
ABCA1 Increases the Cholesterol Content and Alters the CE Composition of the ApoB-containing Lipoproteins-Analysis of plasma lipids in the VLDL, IDL/LDL, and HDL fractions obtained by FPLC separation (Table 1)   Livers from male mice (n ϭ 2 each group) were taken at mid-dark cycle. Primary peritoneal macrophages were pooled from six male mice per genotype and allowed to attach for 3 h before protein extraction. Protein extracts were prepared in cell lysis buffer as described under "Experimental Procedures," and a commercial antibody was used to detect both endogenous mouse ABCA1 and human ABCA1 protein. Arrow indicates size marker band at 188 kDa (SeeBlue Plus2, Invitrogen). C, cholesterol efflux was measured from primary hepatocytes and peritoneal macrophages from the studied mouse lines. After loading with 3 H-labeled cholesterol or cholesterol ester, cells were incubated with medium containing either bovine serum albumin, human apoA-I (10 g/ml) or human HDL (50 g/ml) as cholesterol acceptor. After normalization of counts to cellular protein, [ 3 H]cholesterol efflux was calculated as % counts (dpm).
LDLr-KO mice were observed (Table 1, p Ͻ 0.05 all). Of interest, HDL particles from ABCA1-Tg mice revealed a 2.3-fold ( p Ͻ 0.05) increase in FC but no significant difference in CE content ( Table 1).
Characterization of the fatty acid moiety of CE in control LDLr-KO versus ABCA1-Tg mice provided further insight into the origin of the accumulated plasma CE found primarily in the non-HDL fraction of ABCA1-Tg mice. Table 2 reveals the changes in the CE fatty acid content of the two mouse study groups as measured by GC. Overexpression of ABCA1 in LDLr-KO mice led to an increase of 100 mg/dl (158.9 Ϯ 4.3 versus 59.3 Ϯ 2.5 mg/dl; p Ͻ 0.001) of LCAT-derived cholesteryl-linoleate (18:2), which accounted for ϳ40% of total plasma CE in this mouse model ( Table 2). In contrast, the increase in acyl-CoA:cholesterol acyltransferase (ACAT)-derived cholesteryl oleate (18:1) was ϳ50 mg/dl (63.3 Ϯ 3.5 versus 17.2 Ϯ 1.6 mg/dl; p Ͻ 0.001) in ABCA1-Tg mice compared with control mice, and accounted for only 14% of total plasma CE in this mouse line (Table 2).
These results demonstrate that although some of the CE increase observed in the plasma apoB-Lps of ABCA1-Tg mice is ACAT-derived, the major portion is derived from the esterification by LCAT. Consistently, plasma LCAT activity, determined by a proteoliposome assay, was significantly 1.7-fold increased in the ABCA1-Tg mice (32.7 Ϯ 0.79 nmol of fatty acids/ml/h)) compared with the control mice ( Fig. 1C), we injected mice intravenously with [ 3 H]FC-labeled HDL and followed the tracer after separation of the major plasma lipoprotein fractions by FPLC (Fig. 3). Within 3 min after injection, [ 3 H]cholesterol could be detected in apoB-Lps in ABCA1-Tg mice (Fig. 3), as well as in LDLr-KO controls (data not shown). These data suggest that were determined by immunoblot analysis (inset). C, cholesterol.

TABLE 1 Lipid particle composition in plasma of LDLr-KO and ABCA1-Tg mice
Plasma from fasted male mice was subjected to FPLC analysis followed by lipid quantification of all fractions. Lipid content was calculated for the VLDL, IDL/LDL, and HDL fraction. n ϭ 2 each group, with each sample representing pooled plasma from 2-3 mice. cholesterol is indeed transferred from HDL to apoB-Lps, where it can be esterified by plasma LCAT.

VLDL IDL/LDL HDL
Plasma Metabolism of ApoB-containing Lipoproteins in ABCA1-Tg and LDLr-KO Mice-Another potential mechanism by which ABCA1 overexpression might raise non-HDLcholesterol and plasma apoB levels is by altering the catabolism of the apoB-Lps. Therefore, we measured the synthesis and fractional catabolic rates (FCR) of apoB-Lps in the plasma of ABCA1-Tg and control mice. First, plasma apoB production was examined by measuring the incorporation of [ 35 S]methionine in the apoB-Lps of mice injected with triton to inhibit lipase-mediated lipolysis of chylomicrons and VLDL. No differences in VLDL and chylomicron apoB production were found in fasted or non-fasted ABCA1-Tg versus LDLr-KO control mice (data not shown).
Major alterations in the composition and size of apoB-Lps have been shown to alter their catabolism (53,54). To evaluate whether the catabolism of the apoB-Lps was altered in ABCA1-Tg mice, we isolated endogenous mouse VLDL and LDL, labeled the protein moiety with 125 I and monitored the clearance of injected, autologous 125 I-apoB using plasma gels over a time period of 24 h. The catabolism of 125 I-apoB-labeled VLDL (Fig. 4A) (Fig. 4B). Consistent with our findings in the Triton study, the calculated production rates of apoB in VLDL and LDL, using the FCR data, were similar in both mouse models (data not shown). These combined data indicate that delayed catabolism of apoB-Lps accounts for the higher plasma apoB levels and provide a further explanation for the expanded pool of apoB-Lp cholesterol present in the plasma of ABCA1-Tg mice.

Hepatic Overexpression of ABCA1 in LDLr-KO Mice Leads to Increased Hepatic Cholesterol Concentrations and the Induction of Biliary Sterol Secretion
Pathways-Despite the increased hepatic cholesterol efflux and the 14 and 26% reduction in plasma clearance of VLDL and LDL, hepatic cholesterol content was 1.35-fold increased (4.20 Ϯ 0.22 versus 3.12 Ϯ 0.10 g chol/mg liver, respectively; p Ͻ 0.001) in ABCA1-Tg versus LDLr-KO controls (Fig. 5A). These findings are consistent with the hepatic uptake of markedly cholesterol-enriched apoB-Lps (containing 2.4-fold more mol of cholesterol ester per mol of apoB) by the unregulated LDLr-related protein (LRP) in ABCA1-Tg mice (Fig. 5D). Consistently, compensatory changes aimed at lowering hepatic cholesterol content were observed in ABCA1-Tg mice. De novo cholesterol synthesis from 3 H-water was decreased by 71% in ABCA1-Tg versus LDLr-KO mice (177 Ϯ 19 versus 610 Ϯ 45 nmol/h/g, respec-

TABLE 2 Fatty acid content in the plasma cholesterol ester moiety of LDLr-KO and ABCA1-Tg mice
Total lipids were extracted from plasma of fasted male mice, and after separation by thin-layer chromatography the CE moiety was quantitated by GC. n ϭ 5 for each group.   NOVEMBER 3, 2006 • VOLUME 281 • NUMBER 44 tively; p Ͻ 0.001) (Fig. 5B) and accordingly, HMGCoA-Reductase (Hmgcr) mRNA expression was down-regulated by 55% ( p Ͻ 0.05) (Fig. 5C). ABCA1-Tg mice also showed reduced SR-BI expression in the liver when compared with LDLr-KO mice (Fig. 5D).

Hepatic ABCA1 Overexpression in LDLr-deficient Mice
To determine whether hepatic ABCA1 overexpression also alters the hepatobiliary cholesterol balance of ABCA1-Tg mice (Table 3), we measured daily dietary cholesterol intake, biliary cholesterol secretion, intestinal cholesterol absorption, and fecal cholesterol excretion. Biliary cholesterol and bile acid (BA) secretion rates into the intestine were significantly increased by 1.9-fold (3.98 Ϯ 0.53 versus 2.06 Ϯ 0.14 mol cholesterol/d/mouse; p Ͻ 0.05) and 1.8-fold (57.6 Ϯ 5.6 versus 33.0 Ϯ 4.4 mol BA/d/mouse; p Ͻ 0.05), respectively, in ABCA1-Tg versus control LDLr-KO mice (Table 3), represent-ing another compensatory mechanism to reduce liver cholesterol content in ABCA1-Tg mice. However, more than 60% of the total intestinal cholesterol, consisting of similar dietary but increased biliary cholesterol contributions in ABCA1-Tg versus LDLR-KO mice, was readily absorbed in both mouse study groups (3.82 Ϯ 0.02 mol/d/mouse versus 2.71 Ϯ 0.01 mol/d/mouse; p Ͻ 0.05) ( Table 3). Although fecal cholesterol excretion was also increased (by 1.4-fold) in ABCA1-Tg compared with LDLr-KO mice (Table 3), the greater amount of intestinally absorbed cholesterol, together with the significantly higher hepatobiliary cycling BA pool in ABCA1-Tg mice, led to a net increase in sterol absorption from the intestine and transport back to the liver via the enterohepatic circulation in ABCA1-Tg versus control LDLr-KO mice.
Surprisingly, the increases in hepatobiliary sterol secretion were not associated with induced hepatic expression of genes involved in sterol transport, such as Abcg5, Abcg8, Abcb4, or Abcb11, as measured by real time-PCR (data not shown). Furthermore, hepatic expression levels of the two major regulated enzymes in BA synthesis, Cyp7␣ and CYP27a, remained unchanged (real time-PCR and Western blot analysis, respectively; data not shown). These findings prompted us to perform an mRNA expression analysis in the livers of both mouse lines, using Affymetrix cDNA microarrays. The total set of data is accessible in the Gene Expression Omnibus (GEO) data base, series GSE5496. Data have been reduced to genelists of significantly regulated genes (Ն2-fold changes) as described under "Experimental Procedures," and have been screened for genes that are potentially involved in sterol modification and transport.

TABLE 3 Biliary, intestinal, and fecal sterol homeostasis in LDLr-Ko and liver ABCA1-overexpressing mice
Data are calculated per mouse if not specified otherwise; average mouse weights were 28.7 Ϯ 3.6 g for LDLr-KO and 30.9 Ϯ 1.1 g for ABCA1-Tg mice (p Ͻ 0.069). n ϭ 5-6 male mice each group at the age of 4 -6 months. nificantly increased RNA expression in ABCA1-Tg livers (4.4fold; Table 4), as validated by real-time PCR (12-fold; p Ͻ 0.01, Fig. 6). However, mRNA levels of caveolin-1 (cav-1), which has recently been implicated in hepatic cholesterol transport and secretion (55), were found to be 2.8-fold up-regulated (Table 4). Hepatic caveolin-1 up-regulation was further validated by Real-Time PCR (4-fold; p Ͻ 0.01, Fig. 6) as well as by analysis of protein levels (Fig. 5D). In addition, we analyzed expression levels of hepatic ABCG1, which is thought to complement ABCA1 in its role in cholesterol transport. Abcg1 mRNA expression levels were indeed 1.4-fold increased (Table 4) as confirmed by real-time PCR (2-fold; p Ͻ 0.01, Fig. 6) and resulted in an ϳ2-fold induction in ABCG1 protein levels in livers of ABCA1-Tg versus LDLr-KO mice. In summary, hepatic ABCA1 overexpression in the absence of the LDLr significantly alters liver cholesterol homeostasis, down-regulating endogenous cholesterol synthesis and up-regulating hepatobiliary sterol secretion. The absorption of an increased pool of intestinal sterols in ABCA1-Tg mice further contributes to the maintenance of higher liver and plasma cholesterol levels in ABCA1-Tg compared with LDLr-KO mice.

Biliary cholesterol concentration
Hepatic Overexpression of ABCA1 Increases Aortic Atherosclerosis in LDLr-KO Mice-To investigate whether the proatherogenic plasma lipid profile in ABCA1-Tg mice altered atherosclerosis, proximal aortas of 4-month-old male and female ABCA1-Tg and control LDLr-KO mice consuming a chow diet were examined for atherosclerosis (Fig. 7A). The mean lesion area was significantly increased by 10-fold in aortas of male ABCA1-Tg versus LDLr-KO mice (23674 Ϯ 4350 m 2 mice versus 2213 Ϯ 863 m 2 ; p Ͻ 0.001) and by 6-fold in female ABCA1-Tg versus LDLr-KO mice (15439 Ϯ 2402 m 2 versus 2141 Ϯ 703 m 2 ; p Ͻ 0.001). Lesion sizes were dramatically enhanced when the mice were placed on a Western diet, but the difference in lesion size between control and ABCA1 transgenic mice was less pronounced (1.6-fold increase in male ABCA1-Tg compared with control LDLr-KO mice after 12 weeks on Western diet; Fig. 7B).

DISCUSSION
The role of ABCA1 in facilitating efflux of excess cellular cholesterol to acceptor lipoproteins and raising HDL levels in plasma is well established. Several mouse models that overexpress ABCA1 have been developed with the expectation that increased ABCA1 would reduce aortic lesion development.  Table 4) were further validated using real-time PCR as an independent method. Total RNA from livers of mice used for microarray experiments was utilized as decribed under "Experimental Procedures." Relative quantities were normalized to the expression levels of ␤-actin mRNA per sample. Expression in LDLr-KO livers were arbitrarily set to one. (Ⅺ), LDLr-KO (males n ϭ 4 mice); (f), ABCA1-Tg (males n ϭ 4).

TABLE 4 cDNA microarray study: significantly regulated genes potentially implicated in altered sterol homeostasis and bile acid detoxification in the livers of ABCA1-Tg ؋ LDLr-KO mice
Liver samples of ABCA1-Tg and LDLr-KO controls (n ϭ 4 each group) were subjected to Affymetrix cDNA microarray expression analysis as reported under "Experimental Procedures." Genes with mRNA expression changes Ն 2-fold between the two genotypes, using a Benjamini-Hochberg false discovery rate of Յ 50%, were selected and screened for possible sterol modifying or bile acid transporting enzymes. The total set of data is accessible through the GEO repository as series ID GSE5496. ABCG1 was added as a gene of high interest with regard to its complementary role as lipid transporter besides ABCA1. Uncorrected p values for t-tests of individual genes are reported for comparison. However, the various mouse models generated to date have yielded conflicting results. Whereas expression of ABCA1 in macrophages of otherwise ABCA1-deficient mice has clearly been shown to be atheroprotective (24,25), the overexpression of ABCA1 in other tissues of different transgenic mouse models resulted in atheroprotective as well as atherogenic effects, depending on the genetic background and tissue-specificity of ABCA1 expression and the promoter used to drive expression of the transgene (19,27). Increased non-HDL-C levels, as observed in ABCA1-transgenic (18 -20) but also in liver-specific adenoviral ABCA1-overexpressing mice (29 -31) could account for the pro-atherogenic effect. The importance of liver ABCA1 contributing to the formation and maintenance of HDL-C as well as non-HDL-C levels has recently been confirmed in liver specific ABCA1 knock-out and siRNA knockdown mice (31,32), but the mechanisms underlying the mixed findings of ABCA1 overexpression and its effect on atherosclerosis have not yet been fully elucidated.

Gene name
In the present study ABCA1-Tg mice overexpressing ABCA1 protein only in liver were generated by crossing ABCA1-Tg (18) with LDLr-KO mice, both on a C57Bl/ 6 background. Although human ABCA1 mRNA was expressed in both liver and macrophages, ABCA1 protein levels and cholesterol efflux were increased only in livers of ABCA1-Tg mice. Thus, in contrast to the parental strain overexpressing ABCA1 in wild-type mice (18), macrophage ABCA1 protein levels and activity did not differ between ABCA1-Tg ϫ LDLr-KO (ABCA1-Tg) mice and their nontransgenic LDLr-KO littermates.
Recently, intestinal ABCA1 has been shown to contribute to the plasma levels of intestinally derived apoB-Lps (31,32). However, our human ABCA1 transgene under the control of the apoE promoter was not detected in the small intestine of our ABCA1-Tg mice. This unique phenotype enabled us to investigate the effects of stable, liver-specific overexpression of ABCA1 on lipoprotein metabolism and atherosclerosis in mice lacking the LDLr, a model that mimics human familial hypercholesterolemia.
Hepatic ABCA1 overexpression in LDLr-KO mice caused striking alterations in plasma lipids and lipoproteins. Plasma cholesterol, PL, TG, HDL-C, and non-HDL-C levels were increased in ABCA1-Tg versus control LDLr-KO mice. The increased plasma cholesterol in ABCA1-Tg mice (2.8-fold) was predominantly because of accumulation of FC and CE in the apoB-Lps, although plasma HDL-C levels were also higher (1.5fold), mainly because of increased FC. In contrast to the 2.8-fold apoB-Lp-cholesterol increase, the plasma apoB levels were increased by only 1.4-fold in ABCA1-Tg versus LDLr-KO mice. Thus, ABCA1 overexpression in LDLr-KO mice resulted in the accumulation of cholesterol-enriched apoB-Lps. These findings are consistent with the increase in apoB-Lps previously reported in transgenic mice overexpressing ABCA1 on a chow diet (19) and in control mice overexpressing ABCA1 after injection of adenovirus vectors (29,30).
We evaluated the possibility that excess cholesterol in HDL, resulting from hepatic ABCA1 overexpression, could be rapidly transferred to apoB-Lps in the form of FC. Transfer of FC from HDL to apoB-Lps has been reported in vivo in humans by Schwartz et al. (57). Our in vivo FC-transfer studies similarly demonstrated a transfer of [ 3 H]FC from HDL to apoB-Lps in mice within minutes of injection. According to previous findings, LCAT-mediated esterification contributes significantly to the formation of CE not only in HDL but also in non-HDL fractions (51,52); therefore the excess FC in apoB-Lps in ABCA1-Tg mice could serve as a substrate for LCAT. In our ABCA1-Tg mouse model we found a significant 1.7-fold increase in plasma LCAT activity of ABCA1-Tg mice versus control mice; furthermore, the fatty acid composition of the plasma CE moiety provided direct evidence that most of the CE enrichment, which was found primarily in apoB-Lps of ABCA1-Tg mice, was LCAT-derived cholesteroyl linoleate ( Table 2). These data are consistent with a model in which liver overexpression of ABCA1 in LDLr-KO mice increases the efflux of hepatic FC to HDL, followed by its rapid transfer to apoB-Lps where it can be esterified by LCAT.
Changes in LDL particle composition and size can alter the exposure of apolipoprotein epitopes resulting in altered affinity to proteoglycans and cell surface receptors that are responsible for their removal from the circulation (53,54). In ABCA1-Tg mice, generation of excess core lipids increases the particle size and hence alters epitope exposure of apoB-Lps. Consistently, the catabolism of the cholesterol-enriched apoB-Lps in ABCA1-Tg mice was delayed by ϳ20% compared with LDLr-KO mice, accounting for the increase in plasma apoB levels in the transgenic mice. These combined findings provide an explanation for the accumulation of apoB and apoB-Lp-cholesterol in plasma of transgenic mice that overexpress hepatic ABCA1 in the absence of the LDLr. Thus, hepatic overexpression of ABCA1 increases efflux of hepatic FC to HDL, followed by a rapid transfer of FC to apoB-Lps where it can be esterified by LCAT as well as ACAT. Consequently, the plasma pool of cholesterol-rich, pro-atherogenic apoB-Lps is expanded in ABCA1-Tg versus control LDLr-KO mice.
The marked changes in lipoprotein metabolism in ABCA1-Tg mice prompted us to evaluate the effects of hepatic ABCA1 overexpression on liver and intestinal sterol transport in LDLr-KO mice. Previous studies using ABCA1-deficient and transgenic mice have not conclusively defined the role of ABCA1 in the hepatobiliary and enterohepatic circulation, reporting either no changes in hepatobiliary secretion and increased cholesterol absorption (22,58), decreased cholesterol absorption (59), or increased biliary cholesterol secretion (18). Despite the increased ABCA1-mediated cholesterol efflux from hepatocytes and the decrease in apoB-Lp catabolism, hepatic overexpression of ABCA1 led to significantly higher cholesterol concentrations in the liver of ABCA1-Tg mice. This accumulation is likely due, in part, to the expanded pool of cholesterolenriched apoB-Lps and the uptake of these apoB-Lps by LRP, which is not regulated by changes in liver cholesterol in ABCA1-Tg mice.
The hepatic cholesterol enrichment was associated with a host of compensatory mechanisms aimed at lowering hepatic liver cholesterol, including down-regulation of endogenous hepatic cholesterol synthesis, decreased expression of HMG-CoA-reductase and of SR-BI, a compensatory pathway that reduces the uptake of HDL-derived CE (60). In addition, biliary cholesterol and bile acid secretion were significantly enhanced in ABCA1-Tg mice compared with their nontransgenic LDLr-KO littermates. As a result, the amount of intestinal cholesterol available for absorption and fecal excretion was increased in ABCA1-Tg mice. However, since hepatic ABCA1 overexpression did not alter intestinal cholesterol absorption, the net cholesterol absorption in ABCA1-Tg mice was greater than in LDLr-KO control mice. Together with the marked increase in the hepatobiliary circulation of bile acids, these changes further contributed to intestinal sterol absorption and the accumulation of plasma cholesterol in ABCA1-Tg mice. Brunham et al. (33) recently reported that intestinal ABCA1 can mediate cholesterol efflux from enterocytes directly into plasma HDL fractions as well as contribute to the plasma pool of intestinally derived chylomicrons. Taken together, our data indicate that although ABCA1-Tg mice have increased cholesterol efflux from the liver to apoA-I and enhanced biliary cholesterol secretion, the livers of ABCA1-Tg ϫ LDLr-KO mice accumulate cholesterol from taking up cholesterol-rich apoB-Lps.
A possible compensatory pathway accounting for the increased hepatobiliary sterol and bile acid secretion in ABCA1-Tg mice was found in the up-regulation of caveolin-1, a lipid transport protein involved in cellular cholesterol uptake, and potentially also Cyp2b10, an enzyme involved in lipid drug metabolism. Liver-specific adenoviral overexpression of caveolin-1 in the liver of C57Bl/6 mice has recently been shown to increase the flow of all biliary lipids (55). We have found a significant induction of caveolin-1 expression in whole liver extracts of ABCA1-Tg mice, whereas sterol transporters such as ABCG5/G8, ABCB4, and ABCB11 showed no differences in expression.
Hepatic sterol modification involves also cytochrome P450 enzymes, which convert cholesterol to bile acids by hydroxylation reactions. Cyp7␣1 or Cyp27␣1 are often up-regulated when the biliary sterol flow is increased. However, in livers of our ABCA1-Tg ϫ LDLr-KO mice, Cyp2b10 was identified by cDNA array analysis as the only cytochrome P450 enzyme induced. Cyp2b10 has been shown to be a phenobarbital-as well as diet-inducible enzyme involved in drug metabolism (61,62), which is regulated by the pregnane X receptor, PXR, the constitutive androstane receptor, CAR, and the Bile Acid/Farnesoid X receptor, BAR/FXR (63), all of which have demonstrated roles in lipid metabolism. Furthermore, Cyp2b10 has been implicated in the hepatic hydroxylation and detoxification process that leads to the formation of polyhydroxylated bile acids secreted via the basolateral hepatic membrane for transport to the kidney and excretion via the urine (64). Thus, Cyp2b10 may constitute an alternative pathway to the normal apical secretion of bile acids into the gall bladder, aimed at decreasing the hepatic cholesterol accumulation observed in these animals.
Given the increased hepatic cholesterol content, as well as the enhanced biliary bile acid secretion rate in ABCA1-Tg ϫ LDLr-KO mice, it is unclear why the expression of the apical bile acid transporters was not also up-regulated in ABCA1-Tg ϫ LDLr-KO mice. Interestingly, we found ABCG1 to be induced in ABCA1-overexpressing livers. ABCG1 is involved in lipid efflux by providing cholesterol pools for the removal by mature HDL (65), thereby complementing ABCA1-mediated cholesterol efflux to nascent HDL particles. Further detailed investigation will be necessary to elucidate the pathways that are implicated in the bile acid homeostasis of ABCA1-Tg ϫ LDLr-KO mouse livers.
The increase of cholesterol in apoB-Lps in the plasma of ABCA1-Tg ϫ LDLr-KO mice thus reflects the increased contribution of excess HDL-C and its conversion to CE, reduced catabolism because of the significant particle enlargement, as well as the continuous recycling of the increased pool of cholesterol that is secreted by the sterol-loaded ABCA1-overexpressing liver. One would anticipate that the accumulation of cholesterol-rich pro-atherogenic lipoproteins results in increased atherosclerosis. As expected, we found increased atherosclerotic lesion development in male and female ABCA1-Tg ϫ LDLr-KO mice on a chow diet. The mean proximal aortic lesion size was markedly increased by feeding a Western type diet for 12 weeks. However, the difference between ABCA1-Tg and LDLr-KO mice decreased from 10-fold on a chow diet to 1.6-fold on a Western diet, possibly reflecting a saturation effect of the diet-induced effect of ABCA1 expression. Our studies demonstrate that increases in the plasma levels of pro-atherogenic apoB-Lps can occur as a result of the selective overexpression of ABCA1 in the absence of the LDLr, which represents a major pathway for the removal of apoB-Lps.
In summary, hepatic overexpression of ABCA1 leads to a further expansion of the pro-atherogenic plasma pool of LDLr-KO mice by (i) enhancing the efflux of liver cholesterol to HDL-C, which then transfers to the apoB-Lps, (ii) delaying the catabolism of the lipid-enriched apoB-Lps, and (iii) increasing the absorption of hepatobiliary sterols from the intestine, ultimately leading to markedly enhanced proximal aortic atherosclerosis. Thus, hepatic ABCA1 modulates the plasma cholesterol levels of both HDL as well as non-HDL. These combined findings indicate that while ABCA1 overexpression in macrophages exerts a clear atheroprotective effect, overexpression of ABCA1 in the liver may be less desirable and the overall effect on cardiovascular disease will likely depend on the ability of the organism to metabolize the increased plasma pool of cholesterol-enriched apoB-Lps.