Targeted inactivation of hepatic Abca1 causes profound hypoalphalipoproteinemia and kidney hypercatabolism of apoA-I

Patients with Tangier disease exhibit extremely low plasma HDL concentrations resulting from mutations in the ATP-binding cassette, sub-family A, member 1 (ABCA1) protein. ABCA1 controls the rate-limiting step in HDL particle assembly by mediating efflux of cholesterol and phospholipid from cells to lipid-free apoA-I, which forms nascent HDL particles. ABCA1 is widely expressed; however, the specific tissues involved in HDL biogenesis are unknown. To determine the role of the liver in HDL biogenesis, we generated mice with targeted deletion of the second nucleotide-binding domain of Abca1 in liver only (Abca1^–L/–L). Abca1^–L/–L mice had total plasma and HDL cholesterol concentrations that were 19% and 17% those of wild-type littermates, respectively. In vivo catabolism of HDL apoA-I from wild-type mice or human lipid-free apoA-I was 2-fold higher in Abca1^–L/–L mice compared with controls due to a 2-fold increase in the catabolism of apoA-I by the kidney, with no change in liver catabolism. We conclude that in chow-fed mice, the liver is the single most important source of plasma HDL. Furthermore, hepatic, but not extrahepatic, Abca1 is critical in maintaining the circulation of mature HDL particles by direct lipidation of hepatic lipid-poor apoA-I, slowing its catabolism by the kidney and prolonging its plasma residence time.

HDL cholesterol (HDL-C) concentration is inversely proportional to cardiovascular disease risk (1). This relationship is thought to be mediated by the ability of HDL to transport excess cholesterol from peripheral tissues back to the liver for excretion in a process known as reverse cholesterol transport (RCT) (2). An understanding of the molecular events in HDL formation is necessary for the development of therapeutic strategies to raise HDL-C levels and protect against atherosclerosis.

HDL biogenesis is poorly understood. Initially, HDL particle formation was thought to occur inside the cell, by a process similar to that for the formation of VLDL and LDL particles (3). However, the assembly of free cholesterol (FC) and phospholipid (PL) with lipid-free apoA-I to form nascent HDL particles is now thought to occur extracellularly (3–5). Fibroblasts from patients with Tangier disease are unable to assemble PL and FC with apoA-I (6), and these patients are characterized by a near absence of plasma HDL and the accumulation of cholesterol esters in tissues enriched with macrophages (7). The discovery that mutations in the (ATP-binding cassette, sub-family A, member 1) ABCA1 gene cause Tangier disease and familial hypoalphalipoproteinemia has clearly established ABCA1 as the key protein responsible for the assembly of FC and PL with lipid-free apoA-I and a critical molecule regulating an initial step in RCT and nascent HDL particle assembly (8–10).

According to the traditional model of RCT, HDL-C originates from peripheral tissues and is subsequently transferred to the liver (2). However, recent studies have challenged this model based on the finding of overexpression of ABCA1 by the liver, which raised plasma HDL concentrations and suggested that significant HDL particle assembly occurred at the hepatocyte surface (11–13). In addition, bone marrow transplantation studies revealed that the macrophage is not a significant source of plasma HDL-C (14). Furthermore, only 2 tissues, the liver and intestine, are quantitatively important in the synthesis and secretion of apoA-I (15). These findings suggest that the liver may be a significant source of HDL particles. However, the contribution of the liver to HDL biogenesis is unknown.

To definitively determine the role of hepatic Abca1 in HDL particle formation and catabolism, we developed liver-specific Abca1-knockout mice (Abca1^–L/–L) by gene targeting. Our results demonstrate that hepatic Abca1 is essential for the lipidation of nascent apoA-I and the maintenance of the majority of the plasma HDL pool.

Creation of liver-specific Abca1-knockout mice. Conditional targeting of the mouse Abca1 gene was achieved by flanking exons 45–46, which encode the second nucleotide-binding fold, with loxP sites, as depicted in Figure 1A (Floxed allele). Tissue-specific expression of Cre recombinase is predicted to eliminate exons 45–46 and result in an inactive Abca1 protein. Eight of 267 embryonic stem cell clones surviving selection with G418 and ganciclovir were found to be correctly targeted by PCR and Southern blot analysis. However, only 2 of the correctly targeted clones were observed to have the loxP site in intron 46 after homologous recombination. This was determined by PCR amplification of a 500-bp fragment of genomic DNA including the HindIII site of intron 46, followed by DNA sequencing of the agarose gel–isolated PCR product. One of these clones was used to develop the mice described in this article.

Figure 1

Targeting strategy and genotypic analysis of liver-specific Abca1-knockout mice. (A) Schematic of 3′ region (exons 44–49) of Abca1 gene showing wild-type (top), floxed (middle), and knockout (bottom) Abca1 alleles. Three loxP sites, 2 flanking the neomycin (Neo) resistance gene and 1 in intron 46, are shown as arrowheads. Arrows below the floxed allele indicate relative position of primers used for PCR screen of alleles. The size of the EcoRV (RV) fragment is shown above each allele, and the relative location of the probe used for Southern blot analysis is shown above the wild-type allele (Probe A). Cre recombinase–mediated elimination of exons 45 and 46 will delete the second ATP-binding cassette, resulting in a knockout allele. Restriction sites: Bg2, Bgl II; E, EcoRI; H, HindIII; S, SacI. (B) Southern blot analysis of liver (L) and kidney (K) genomic DNA from mice that inherited both the Cre and wild-type or floxed Abca1 alleles. DNA was digested with EcoRV and hybridized with probe A. –L denotes a liver-specific knockout allele. (C) Quantitative real-time PCR analysis of RNA isolated from liver and kidney. Relative fold change compared with a wild-type (+/+) liver sample was calculated using the 2^–ØØCT method (42). (D) Western blot analysis of liver membranes isolated from 3 mice of the indicated Abca1 genotypes. (E) Multi-tissue Southern blot of genomic DNA from the indicated tissues from a wild-type and liver-specific knockout (–L/–L) mouse.

The initial genotyping of mice was performed by PCR analysis of tail DNA, and presumptive genotypes were assigned to animals based on coinheritance of the Cre transgene, under control of the albumin promoter for liver-specific expression, and wild-type or floxed Abca1 alleles. To verify the initial genotypic assignment, we performed Southern blot analysis of genomic DNA from liver and kidney when animals were killed. Only the 6-kb wild-type allele was observed in DNA from liver and kidney in wild-type mice (Figure 1B, lanes 1 and 2). In heterozygous mice, both floxed and wild-type alleles were present in kidney DNA; however, liver DNA showed the presence of the predicted knockout allele (4 kb) and some residual floxed allele (7 kb), as well as the wild-type allele (6 kb) (lanes 3 and 4). In homozygous mice, only the floxed allele was observed in the kidney, but the liver contained predominantly the knockout allele (70% based on PhosphorImager analysis; lane 6) and some residual floxed allele (30%). We attribute the residual floxed allele to genomic DNA from nonhepatic cells (Kupffer, endothelial, white blood cells, etc.) in the liver sample, as described in other liver-specific knockout mice (16). Results from quantitative real-time PCR analysis of Abca1 mRNA demonstrated a gene dosage–dependent decrease in mRNA, to 30% of normal for Abca1^–L/–L mice compared with wild-type littermates, in the liver but not in the kidney (Figure 1C). Western blot analysis of liver membranes also showed a gene dosage–dependent decrease in Abca1 protein to near undetectable levels in the homozygous knockout mice (Figure 1D). Multi-tissue Southern blot analysis demonstrated that only the liver DNA from Abca1^–L/–L mice contained the knockout allele, whereas other tissues from this animal had only the floxed allele, and all tissues in the wild-type littermate control had only the wild-type allele (Figure 1E). These data show that Abca1 recombination and targeted deletion were specific for liver.

PC and FC efflux from primary hepatocytes and elicited peritoneal macrophages. To ensure that the recombination of the floxed allele resulted in loss of Abca1 function, we isolated and cultured primary hepatocytes from Abca1^–L/–L and Abca1^+/+ mice and stimulated them with liver X receptor/retinoid X receptor (LXR/RXR) agonists prior to performing PC and FC efflux studies. FC (Figure 2A) and PL (Figure 2B) efflux from hepatocytes in the absence of lipid-free apoA-I was similar for both genotypes. However, in the presence of apoA-I, there was a significant (P < 0.01) stimulation of efflux for both lipids with hepatocytes derived from wild-type mice compared with those from Abca1^–L/–L mice. In addition, no Abca1 protein was detected in hepatocytes from the knockout mice, whereas abundant protein was detected in wild-type controls (Figure 2D, top panel). These data confirm that our liver-specific knockout mice had no residual apoA-I–dependent Abca1 lipid efflux activity in hepatocytes. As a control, lipid efflux was measured in thioglycolate-elicited peritoneal macrophages from mice of the same genotype. Macrophages from wild-type and knockout mice showed the expected stimulation of cholesterol efflux in the presence of apoA-I and with additional treatment with the LXR agonist, T0901317 (Figure 2C). However, regardless of the experimental condition, efflux of cholesterol was similar for macrophages from wild-type and Abca1^–L/–L mice. Abca1 protein expression was also similar for cultured elicited macrophages from mice of both genotypes (Figure 2D, bottom).

Figure 2

Lipid efflux from primary hepatocytes and peritoneal macrophages. Primary mouse hepatocytes were isolated from chow-fed Abca1^+/+ or Abca1^–L/–L mice, stimulated with 9-cis-retanoic acid and 22-OH-cholesterol, and radiolabeled with either [³H]cholesterol or [¹⁴C]choline chloride for 24 hours. After an hour of equilibration, cells were incubated in the presence or absence of 10 μg apoA-I/ml for 24 hours. (A) Hepatocyte cholesterol efflux in the presence or absence of apoA-I. (B) Hepatocyte choline PL efflux in the presence or absence of apoA-I. Data with unlike symbols are significantly different from one another (P < 0.05). (C) Thioglycolate-elicited peritoneal macrophages were isolated from Abca1^+/+ or Abca1^–L/–L mice, radiolabeled with [³H]cholesterol for 24 hours, and then incubated with 10 μM T0901317 or vehicle for an additional 24 hours. Cholesterol efflux was measured after 24-hour incubation of cells, which were stimulated with T0901317 or vehicle in the presence or absence of 20 μg apoA-I/ml. Radiolabel in medium and the cellular isopropanol extract was quantified, and percentage efflux was calculated as the ratio of radioactivity in the medium divided by total radioactivity (cells + media) × 100%. Data are mean ± SD for 3 mice of the indicated genotypes, assayed in triplicate. (D) Western blot of Abca1 or load control protein (GAPDH or β-actin) in cultured hepatocytes (top) or cultured elicited macrophages (bottom).

Birth frequency of genotypes. Previous studies have documented that matings of homozygous Abca1-knockout (Abca1^–/–) are nonproductive and lead to neonatal death as a result of placental malformation and fetal distress (17, 18). Consistent with these observations, we found the frequency of genotypes of our total Abca1-knockout (Abca1^–/–) mice generated from intercross matings of Abca1^+/–EIIa-Cre transgenic mice did not follow Mendelian inheritance (Abca1^+/+, 18%; Abca1^+/–, 69%; and Abca1^–/–, 13%; n = 21). However, in the case of the liver-specific knockout, the frequency of genotypes derived from intercross matings of heterozygous liver-specific Abca1-knockout mice that inherited the Cre recombinase transgene (i.e., Abca^+/–L(flox)Alb-Cre Tg) was close to that expected for Mendelian inheritance (Abca1^+/+, 22%; Abca1^+/–L, 44%; and Abca1^–L/–L, 33%; n = 72), which indicates that neither absence of hepatic Abca1 nor reduced HDL-C levels are responsible for the decreased fertility observed in Abca1^–/– mice (19).

Plasma lipid and lipoprotein analysis. The plasma lipid and lipoprotein phenotype for chow-fed liver-specific Abca1 and Abca1^–/– mice after a 4-hour fast is shown in Table 1. Total and HDL-C concentrations in plasma of Abca1^–L/–L mice were 80% lower than those of wild-type littermates (P < 0.001). Plasma FC, esterified cholesterol, and PL were all markedly lower in concentration (70–90% reduction) in Abca1^–L/–L mice compared with wild-type animals, and plasma triglyceride (TG) concentrations were significantly higher, which is consistent with the phenotype in Tangier disease patients (7). Plasma apoA-I, measured by ELISA, was also 90% lower in Abca1^–L/–L mice. Values for heterozygous mice were intermediate to those of wild-type and homozygous liver-specific knockout mice, which confirms a gene dosage effect of hepatic Abca1 expression on plasma lipid and HDL-C concentrations. Values for Abca1^–/– mice, generated using EIIa-Cre transgenic mice, were slightly lower than those for Abca1^–L/–L mice. Lecithin cholesterol acyltransferase (LCAT) and hepatic lipase (HL) activities in plasma were 50% lower in Abca1^–L/–L compared with wild-type littermates (Table 1). However, PL transfer protein (PLTP) activity was reduced to background levels (i.e., equivalent to that in plasma of PLTP-knockout control mice; courtesy of Xian-Cheng Jiang, SUNY Downstate Medical Center, New York, New York, USA) in the Abca1^–L/–L mice (Table 1).

Table 1

Plasma measurements of liver-specific and Abca1^–/– mice

Lipoprotein particle analysis. Given the striking decrease in plasma HDL-C levels of Abca1^–L/–L mice as determined by the heparin-manganese precipitation method, we investigated the size distribution and apolipoprotein content of the plasma lipoproteins. The distribution of cholesterol among lipoproteins after fractionation of plasma by fast performance lipid chromatography (FPLC) confirmed the profound reduction in HDL-C concentrations in Abca1^–L/–L mice compared with wild-type controls (Figure 3A). The FPLC cholesterol elution profile also suggested a slight shift of the curve to the right in elution time of HDL particles for Abca1^+/–L and Abca1^–L/–L compared with wild-type mice, which suggests that the average size of HDL particles was smaller in these animals. However, nondenaturing gradient PAGE and apoA-I Western blot analysis revealed a similar size distribution for HDL particles in plasma of all 3 genotypes, which suggests that the decrease in plasma HDL concentration was due primarily to a decrease in the number of similar-sized particles (Figure 3B). There was also a decrease in cholesterol concentration in the LDL elution region of the FPLC column for heterozygous and homozygous mice, which is consistent with reported decreases in LDL concentration in patients with Tangier disease and in Abca1^–/– mice (7, 18, 20).

Figure 3

Plasma lipoprotein and apolipoprotein characterization of liver-specific Abca1-knockout mice. Plasma was obtained from chow-fed Abca1^+/+, Abca1^+/–L, and Abca1^–L/–L mice fasted for 4 hours. Equal-volume pools of plasma were made using 5 mice of each genotype for FPLC (A) and apolipoprotein analysis (C). (A) One hundred microliters whole plasma from each pool was fractionated on Superose 6 FPLC columns. Fractions were collected at 1-minute intervals, and total cholesterol was measured using an enzymatic assay. (B) One microliter of whole plasma from 3 individual mice of each genotype was fractionated on a 4–30% nondenaturing gradient gel for 1,400 V/h. The proteins were transferred to nitrocellulose, and the blot was developed with anti-mouse apoA-I antiserum. (C) Pooled plasma from each genotype was subjected to ultracentrifugation at a density of 1.25 g/ml to float plasma lipoproteins. Fifteen micrograms of lipoprotein protein was added to each lane of the gel, and apolipoproteins were separated by 4–16% SDS-PAGE. Gels were stained with Coomassie blue and destained to visualize the apolipoproteins. Standard low-molecular-weight markers (Std) are indicated on the left. Estimated migration position of apoB100, apoB48, albumin, apoE, and apoA-I are indicated on the right.

To determine whether the apolipoprotein distribution of plasma lipoprotein particles was modified in Abca1^–L/–L mice, we isolated by ultracentrifugation and analyzed them by SDS-PAGE (Figure 3C). Overall, the apolipoprotein pattern among genotypes was similar except for an increase in the proportion of apoB48, albumin, and apoA-IV (by Western blot analysis; data not shown) and a relative decrease in apoA-I for Abca1^–L/–L mice. These data, as well as the increase in plasma TG concentration (Table 1), suggest a modest accumulation of remnant lipoproteins of intestinal origin in the plasma of chow-fed liver-specific knockout mice.

Relationship of hepatic Abca1 expression and plasma HDL-C concentrations. To investigate the relationship between hepatic Abca1 protein expression and the steady-state concentration of plasma HDL among genotypes of liver-specific Abca1-knockout mice, we analyzed hepatic membrane Abca1 protein expression by Western blot analysis for a subset of mice and compared the results with the plasma concentration of HDL-C. There was a gene dosage effect on hepatic Abca1 protein expression (Figure 4A) that paralleled the decrease in plasma HDL-C concentration among liver-specific genotypes (Figure 4B), which suggests that the liver is the major source of plasma HDL particles in these chow-fed mice.

Figure 4

Hepatic Abca1 protein expression and plasma HDL cholesterol concentrations in liver-specific Abca1-knockout mice. Liver membranes were isolated from a subset of mice of the indicated genotypes that were allowed to consume chow. Membranes were fractionated by SDS-PAGE, after which proteins were transferred to nitrocellulose membranes and probed with primary antibody to Abca1 or β-actin. Blots were developed using a ¹²⁵I-radiolabeled secondary antibody, and PhosphorImager analysis was then used to quantify the signal intensity ratio of Abca1 to β-actin (A). HDL cholesterol concentrations in plasma were measured by enzymatic assay after precipitation of apoB lipoproteins with heparin and MnCl2 (B). Points represent data from individual mice, and the horizontal lines denote the mean for each genotype.

Plasma turnover and tissue uptake of wild-type HDL particles in liver-specific Abca1-knockout mice. Although ABCA1 is not thought to play a major role in the maturation and metabolism of mature plasma HDL particles (21), intravenously injected radiolabeled HDL from normal donors is rapidly cleared in patients with Tangier disease (20, 22), which suggests that ABCA1 may play a role in HDL catabolism. Similar observations have been made in studies on the Wisconsin hypoalpha mutant (WHAM) chicken, an animal model of Tangier disease in which the Abca1 gene is functionally inactive (23, 24). We isolated HDL particles that were homogeneous in size from wild-type mice and radiolabeled them with [¹²⁵I]tyramine cellobiose ([¹²⁵I]TC), a residualizing reagent (25), to trace plasma decay and tissue uptake of HDL. The tracer HDL particles were 10.8 nm in diameter, and more than 80% of the radiolabeled protein migrated as apoA-I (Figure 5A). The radiolabeled HDL particles were injected into Abca1^+/+, Abca1^+/–L, and Abca1^–L/–L recipient mice, and plasma decay of radiolabeled tracer was followed for 24 hours before the animals were sacrificed and tissues removed for quantification of tracer particle uptake. The plasma decay of radiolabeled tracer was more rapid with progressive inactivation of liver Abca1 alleles (Figure 5B). Abca1^–L/–L mice had a 2-fold greater tracer fractional catabolic rate (FCR) compared with wild-type littermate recipient mice (2.88 ± 0.14 vs. 1.68 ± 0.61 pools/day; P < 0.02), whereas heterozygous recipient mice had an intermediate FCR (2.00 ± 0.12 pools/day) (Figure 5C). Production rate, calculated as the product of plasma apoA-I pool size and plasma FCR, was 10-fold higher in wild-type compared with Abca1^–L/–L mice (3.1 ± 1.7 vs. 0.33 ± 0.09 mg apoA-I/d; P < 0.03). We interpret this as the production rate of HDL particles, since apoA-I synthesis and secretion are normal in Tangier patients, Abca1^–/– mice, and the WHAM chicken (7, 24, 26). Monitoring of HDL tracer size in plasma during the 24-hour turnover by nondenaturing gradient PAGE and PhosphorImager analysis showed no discernible shift in HDL particle size (Figure 5D), which suggests that HDL particles were catabolized from plasma without major remodeling. Previous studies have shown that the major sites for HDL apoA-I uptake are the liver and kidney (27–29). Uptake of tracer HDL by the liver, which included tracer that was recovered in the intestine and intestinal contents (30), was not different among genotypes and was nearly 3-fold greater than that observed for kidney uptake in wild-type mice (19.3% vs. 7.3%; Figure 5E). However, there was a significant increase in tracer HDL uptake by the kidney with progressive inactivation of hepatic Abca1 alleles, which resulted in a 2-fold increase in tracer HDL uptake in Abca1^–L/–L compared with wild-type mice (12.4% vs. 7.3% of injected dose, respectively).

Figure 5

In vivo catabolism of wild-type HDL tracer in Abca1^+/+, Abca1^+/–L, and Abca1^–L/–L recipient mice. HDL particles were isolated from the plasma of chow-fed wild-type mice, radiolabeled with [¹²⁵I]TC, a residualizing reagent, and injected into chow-fed mice of the indicated genotype. Plasma samples were taken over 24 hours, after which animals were sacrificed, and tissues were harvested for quantification of radiolabel uptake. In C and E, genotypes with unlike symbols are significantly different from one another (P < 0.05). (A) Characterization of [¹²⁵I]TC HDL tracer by 4–30% nondenaturing gradient gel electrophoresis and 4–16% SDS-PAGE. Both gels were visualized by PhosphorImager analysis. Standard proteins are shown for reference. (B) Whole plasma die-away of wild-type HDL tracer in Abca1^+/+, Abca1^+/–L, and Abca1^–L/–L mice. Individual data points are mean ± SD (n = 3). (C) Tracer FCR calculated from the plasma die-away curves in B. The horizontal lines denote the mean for each genotype. (D) Size analysis of [¹²⁵I]TC HDL tracer in plasma after injection into Abca1^+/+ and Abca1^–L/–L recipient mice. Plasma samples were collected at the indicated times from recipient mice injected with [¹²⁵I]TC HDL and separated on 4–30% nondenaturing gradient PAGE. [¹²⁵I]TC HDL migration was visualized by PhosphorImager analysis. (E) Liver and kidney uptake of [¹²⁵I]TC HDL tracer 24 hours after injection into Abca1^+/+, Abca1^+/–L, Abca1^–L/–L mice. Liver (including intestine and contents; ref. 30) and kidney tissue was digested overnight in 1 N NaOH at 60°C, and ¹²⁵I radioactivity in the digest was quantified using a γ-ray counter.

Plasma turnover and tissue uptake of lipid-free apoA-I in liver-specific Abca1-knockout mice. To determine the role of hepatic Abca1 on apoA-I catabolism, we performed turnover studies in Abca1-knockout mice using purified human lipid-free [¹²⁵I]TC apoA-I tracer. The tracer migrated as authentic apoA-I on a SDS-PAGE gel, with a size range of less than 7.1 nm on nondenaturing gradient gels, and contained less than 1 molecule of PL per molecule apoA-I (Figure 6A) (28). Plasma turnover of apoA-I (Figure 6B) was more rapid for Abca1^–L/–L recipients compared with wild-type mice, with a 2-fold greater tracer FCR (3.98 ± 0.34 vs. 1.95 ± 0.04 pools/day, respectively; P < 0.001; Figure 6C). Lipid-free apoA-I tracer FCR was significantly higher than that of HDL tracer in Abca1^–L/–L mice (3.98 vs. 2.88 pools/day; P < 0.003), whereas there was no difference in plasma FCR for these tracers in wild-type mice (1.95 vs. 1.68 pools/day, respectively). apoA-I tracer was found to be bound to mouse HDL particles in plasma throughout the 24-hour turnover study (Figure 6D), with no tracer observed in the migration position of lipid-free apoA-I (<7.1 nm; Figure 6A). As observed for the HDL turnover study, while there was no difference in liver uptake of the apoA-I tracer, there was a 2.3-fold increase in kidney uptake in Abca1^–L/–L compared with wild-type recipients (Figure 6E). Radiolabel recovery in the liver and intestine was lower in the apoA-I tracer turnover study compared with that in the HDL study for 2 reasons. First, the more rapid removal of [¹²⁵I]TC apoA-I from the plasma compartment compared with [¹²⁵I]TC HDL likely resulted in a greater loss of radiolabel from the liver and kidney into the feces and urine during the time course of the turnover (27, 30). Second, the liver uptake data in the HDL turnover study included radiolabel that had leaked into the intestine (27, 30), whereas the apoA-I turnover study did not.

Figure 6

In vivo catabolism of human lipid-free apoA-I tracer in Abca1^+/+ and Abca1^–L/–L recipient mice. apoA-I was isolated from human plasma, radiolabeled with [¹²⁵I]TC, and injected into chow-fed mice of the indicated genotype. Details are presented in the Figure 5 legend. (A) PhosphorImager analysis of [¹²⁵I]TC apoA-I tracer after separation by 4–30% nondenaturing gradient gel electrophoresis and 4–16% SDS-PAGE. (B) Whole plasma die-away of lipid-free apoA-I tracer in Abca1^+/+ and Abca1^–L/–L mice. Individual data points are mean ± SD (n = 4). (C) Tracer FCR calculated from the plasma die-away curves in B. The horizontal lines denote the mean for each genotype. (D) Size analysis of [¹²⁵I]TC apoA-I tracer in plasma after injection into Abca1^+/+ and Abca1^–L/–L recipient mice. (E) Liver and kidney uptake of [¹²⁵I]TC apoA-I tracer 24 hours after injection into Abca1^+/+ and Abca1^–L/–L mice. Liver and kidney tissue was digested overnight in 1 N NaOH at 60°C, and ¹²⁵I radioactivity in the digest was quantified using a γ-ray counter. Genotypes with unlike symbols are significantly different from one another (P < 0.001).

Studies in patients with Tangier disease and Abca1-knockout mice have shown that efflux of lipid by ABCA1 is the rate-limiting step in HDL particle formation and is absolutely required for the maintenance of plasma HDL-C concentrations (18, 20, 31). ABCA1 is variably expressed in nearly all tissues of the body, and its major function is to assemble FC and PL with lipid-free apoA-I to form nascent HDL particles (32). However, only 2 tissues, the liver and intestine, synthesize and secrete apoA-I (15). In addition, adenoviral and transgenic overexpression of hepatic ABCA1 increases plasma HDL-C concentrations (11–13), which suggests that the liver may contribute to the plasma HDL pool. Using a liver-specific Abca1-knockout mouse model, we provide what we believe to be the first definitive proof that hepatic Abca1 is essential for approximately 80% of the steady-state pool of plasma HDL in chow-fed mice and that the gene dosage–dependent decrease in plasma HDL-C is matched by a similar decrease in hepatic Abca1 protein expression. These results support an unequivocal and essential role for hepatic Abca1 in HDL production in an experimental setting in which mice were fed a chow diet and Abca1 was not upregulated by cholesterol feeding or LXR activation. The finding that plasma HDL-C and apoA-I levels were reduced by 80% and 90%, respectively, in the absence of hepatic Abca1 but in the presence of functional Abca1 in extrahepatic tissues, also suggests that lipidation of the nascent apoA-I molecule occurs during or soon after secretion by hepatocytes, prior to entry of the particle into the circulation. This finding is in agreement with studies in cell culture models, which indicate that hepatocyte lipidation of apoA-I occurs in an intra- or pericellular fashion (33, 34). Our data also suggest that lipidation by hepatocytes is necessary for maintaining plasma HDL concentrations by prolonging the circulation time of HDL apoA-I in vivo.

HDL particle assembly in vivo is complex and poorly understood. The demonstration that fibroblasts from patients with Tangier disease are unable to mediate the efflux of FC and PL to lipid-free apoAI, and the discovery that mutations in ABCA1 cause Tangier disease, established ABCA1 as the key molecule regulating the formation of nascent HDL. The identification of the sites of in vivo HDL biogenesis is fundamental to our understanding of RCT. This pathway, as originally envisioned by Glomset (2), involves the release of cellular cholesterol from extrahepatic tissues. Assembly of lipid-free apoA-I with PL and FC in peripheral tissues would provide a source of nascent HDL particles, which, upon maturation to spherical HDL in plasma by LCAT and PLTP, would direct the flux of cholesterol to the liver upon subsequent HDL particle catabolism by this organ. However, studies using Abca1-knockout bone marrow transplanted into wild-type mice do not support a major role for macrophage ABCA1 in the maintenance of plasma HDL concentrations (14). Results from our study, involving selective deletion of hepatic Abca1, demonstrate that the liver is the major source of plasma HDL and hepatic Abca1 maintains the pool of plasma HDL by lipidating newly secreted apoA-I to produce nascent HDL particles that undergo maturation in plasma. These results therefore profoundly alter our concept of in vivo HDL particle assembly by establishing the liver as the single most important source of nascent HDL in chow-fed mice.

Since efflux of FC from cells in culture to plasma HDL particles by non-ABCA1 pathways is quantitatively greater than that observed for lipid-free apoA-I by ABCA1 pathways, the main role of liver ABCA1 may be to provide plasma HDL particles to participate in RCT by pathways other than ABCA1, such as those mediated by scavenger receptor class B, type I (SR-BI) and ABCG1 (35, 36). Efflux of FC from lipid-loaded macrophages by ABCA1 is an important antiatherogenic pathway of the RCT pathway, since transplantation of Abca1^–/– macrophages into atherosclerosis-susceptible mice results in increased atherosclerosis; however, this RCT pathway is not quantitatively important in maintaining plasma HDL levels (14, 37, 38). The relative importance of RCT via macrophage ABCA1 versus non-ABCA1 HDL efflux pathways with regard to atherosclerosis development is unknown and will require further investigation.

Normal HDL particles are hypercatabolized in patients with Tangier disease and the WHAM chicken (20, 22, 24). Hypercatabolism of plasma HDL tracer has been shown to occur even after normalization of HDL-C levels in patients with Tangier disease (22, 39). We observed an increased turnover of wild-type plasma HDL apoA-I in mice with progressive inactivation of hepatic Abca1 alleles (Figure 5, B and C). However, increased catabolism of the tracer was only observed in the kidney, even though the liver was responsible for approximately 3-fold more HDL apoA-I catabolism than the kidney in wild-type mice (Figure 5E). These data argue against a nonspecific hypercatabolism of the HDL apoA-I tracer due to a lower HDL pool size in the Abca1^–L/–L mice and are compatible with the observation that repletion of the plasma HDL pool in Tangier patients does not correct the hypercatabolism of HDL. Since HDL particles do not bind to ABCA1 with high affinity and are not efficient acceptors of ABCA1-mediated lipid efflux compared to lipid-free apoA-I (21, 40), why should hypercatabolism of normal HDL apoA-I occur in the absence of Abca1? We suggest that following selective uptake of HDL cholesteryl ester by SR-BI in the liver, lipid-free or lipid-poor apoA-I is released from the remodeled HDL particle and immediately relipidated by Abca1 at the hepatocyte surface before reentry into plasma, which results in a renewable source of nascent HDL particles that can enter the plasma HDL pool after being processed to mature particles. Our data also suggest that functional Abca1 in extrahepatic tissues is not sufficient to relipidate and recycle the apoA-I that is released at the hepatocyte surface during HDL catabolism, because in the absence of hepatic Abca1, lipid-free or lipid-poor apoA-I is not relipidated, enters the circulation, and is rapidly cleared by the kidney. This argument is also supported by the significantly more rapid turnover of lipid-free apoA-I tracer (Figure 6) compared with that of HDL tracer (Figure 5) in Abca1^–L/–L mice, which suggests that extrahepatic Abca1 expression in chow-fed mice is not sufficient to rescue the hypercatabolism of lipid-free apoA-I from plasma. Thus, we believe that the maintenance of plasma HDL concentrations by hepatic Abca1 is achieved by 2 mechanisms, the lipidation of newly secreted hepatic apoA-I and the recycling of some proportion of apoA-I after catabolism of HDL particles at the hepatocyte surface.

This work was supported by NIH grants HL49373 (to J.S. Parks), HL07115 (Cardiovascular Pathology Training grant; to J.M. Timmins), and HL42630 (to N. Maeda) and by grants from the Canadian Institutes of Health Research (to L.R. Brunham and M.R. Hayden), the Michael Smith Foundation for Health Research (to L.R. Brunham), The Saal van Zwanenberg foundation (to J.M. Coutinho), The Netherlands Heart Foundation (to J.M. Coutinho), and the Heart and Stroke Foundation of BC & Yukon (to M.R. Hayden). M.R. Hayden holds a University Killam Professorship and is a Canada Research Chair in Human Genetics.

Nonstandard abbreviations used: ABCA1, ATP-binding cassette, sub-family A, member 1; FC, free cholesterol; FCR, fractional catabolic rate; FPLC, fast performance lipid chromatography; HDL-C, HDL cholesterol; HL, hepatic lipase; LCAT, lecithin cholesterol acyltransferase; LXR, liver X receptor; PL, phospholipid; PLTP, PL transfer protein; RCT, reverse cholesterol transport; SR-BI, scavenger receptor class B, type I; TC, tyramine cellobiose; TG, triglyceride; WHAM, Wisconsin hypoalpha mutant.

Conflict of interest: The authors have declared that no conflict of interest exists.

1. Assmann, G, Gotto (Jr), AM. HDL cholesterol and protective factors in atherosclerosis. Circulation. 2004. 109(Suppl. 1):III8-III14.
   View this article via: PubMed Google Scholar
2. Glomset, JA. The plasma lecithin:cholesterol acyltransferase reaction. J. Lipid Res. 1968. 9:155-167.
   View this article via: PubMed Google Scholar
3. Hamilton, RL, Moorehouse, A, Havel, RJ. Isolation and properties of nascent lipoproteins from highly purified rat hepatocytic Golgi fractions. J. Lipid Res. 1991. 32:529-543.
   View this article via: PubMed Google Scholar
4. Hamilton, RL, Guo, LS, Felker, TE, Chao, YS, Havel, RJ. Nascent high density lipoproteins from liver perfusates of orotic acid-fed rats. J. Lipid Res. 1986. 27:967-978.
   View this article via: PubMed Google Scholar
5. Oram, JF, Yokoyama, S. Apolipoprotein-mediated removal of cellular cholesterol and phospholipids. J. Lipid Res. 1996. 37:2473-2491.
   View this article via: PubMed Google Scholar
6. Francis, GA, Knopp, RH, Oram, JF. Defective removal of cellular cholesterol and phospholipids by apolipoprotein A-I in Tangier disease. J. Clin. Invest. 1995. 96:78-87.
   View this article via: JCI PubMed CrossRef Google Scholar
7. Assman, G., von Eckardstein, A., and Brewer, H.B., Jr. 2001. Familial analphalipoproteinemia: Tangier disease. In The metabolic and molecular bases of inherited disease. C.R. Scriver et al., editors. McGraw-Hill. New York, New York, USA. 2937–2960.
8. Brooks-Wilson, A, et al. Mutations in ABC1 in Tangier disease and familial high-density lipoprotein deficiency. Nat. Genet. 1999. 22:336-345.
   View this article via: PubMed CrossRef Google Scholar
9. Bodzioch, M, et al. The gene encoding ATP-binding cassette transporter 1 is mutated in Tangier disease. Nat. Genet. 1999. 22:347-351.
   View this article via: PubMed CrossRef Google Scholar
10. Rust, S, et al. Tangier disease is caused by mutations in the gene encoding ATP-binding cassette transporter 1. Nat. Genet. 1999. 22:352-355.
    View this article via: PubMed CrossRef Google Scholar
11. Wellington, CL, et al. Alterations of plasma lipids in mice via adenoviral-mediated hepatic overexpression of human ABCA1. J. Lipid Res. 2003. 44:1470-1480.
    View this article via: PubMed CrossRef Google Scholar
12. Basso, F, et al. Role of the hepatic ABCA1 transporter in modulating intrahepatic cholesterol and plasma HDL cholesterol concentrations. J. Lipid Res. 2003. 44:296-302.
    View this article via: PubMed CrossRef Google Scholar
13. Vaisman, BL, et al. ABCA1 overexpression leads to hyperalphalipoproteinemia and increased biliary cholesterol excretion in transgenic mice. J. Clin. Invest. 2001. 108:303-309. doi:10.1172/JCI200112517.
    View this article via: JCI PubMed Google Scholar
14. Haghpassand, M, Bourassa, PA, Francone, OL, Aiello, RJ. Monocyte/macrophage expression of ABCA1 has minimal contribution to plasma HDL levels. J. Clin. Invest. 2001. 108:1315-1320. doi:10.1172/JCI200112810.
    View this article via: JCI PubMed Google Scholar
15. Wu, AL, Windmueller, HG. Relative contributions by liver and intestine to individual plasma apolipoproteins in the rat. J. Biol. Chem. 1979. 254:7316-7322.
    View this article via: PubMed Google Scholar
16. Postic, C, et al. Dual roles for glucokinase in glucose homeostasis as determined by liver and pancreatic beta cell-specific gene knock-outs using Cre recombinase. J. Biol. Chem. 1999. 274:305-315.
    View this article via: PubMed CrossRef Google Scholar
17. Christiansen-Weber, TA, et al. Functional loss of ABCA1 in mice causes severe placental malformation, aberrant lipid distribution, and kidney glomerulonephritis as well as high-density lipoprotein cholesterol deficiency. Am. J. Pathol. 2000. 157:1017-1029.
    View this article via: PubMed Google Scholar
18. McNeish, J, et al. High density lipoprotein deficiency and foam cell accumulation in mice with targeted disruption of ATP-binding cassette transporter-1. Proc. Natl. Acad. Sci. U. S. A. 2000. 97:4245-4250.
    View this article via: PubMed CrossRef Google Scholar
19. Selva, DM, et al. The ATP-binding cassette transporter 1 mediates lipid efflux from Sertoli cells and influences male fertility. J. Lipid Res. 2004. 45:1040-1050.
    View this article via: PubMed CrossRef Google Scholar
20. Schaefer, EJ, et al. Metabolism of high-density lipoprotein apolipoproteins in Tangier disease. N. Engl. J. Med. 1978. 299:905-910.
    View this article via: PubMed Google Scholar
21. Wang, N, Silver, DL, Costet, P, Tall, AR. Specific binding of ApoA-I, enhanced cholesterol efflux, and altered plasma membrane morphology in cells expressing ABC1. J. Biol. Chem. 2000. 275:33053-33058.
    View this article via: PubMed CrossRef Google Scholar
22. Schaefer, EJ, et al. Metabolism of high density lipoprotein subfractions and constituents in Tangier disease following the infusion of high density lipoproteins. J. Lipid Res. 1981. 22:217-228.
    View this article via: PubMed Google Scholar
23. Attie, AD, et al. Identification and functional analysis of a naturally occurring E89K mutation in the ABCA1 gene of the WHAM chicken. J. Lipid Res. 2002. 43:1610-1617.
    View this article via: PubMed CrossRef Google Scholar
24. Schreyer, SA, Hart, LK, Attie, AD. Hypercatabolism of lipoprotein-free apolipoprotein A-I in HDL-deficient mutant chickens. Arterioscler. Thromb. Vasc. Biol. 1994. 14:2053-2059.
25. Pittman, RC, et al. A radioiodinated, intracellularly trapped ligand for determining the sites of plasma protein degradation in vivo. Biochem. J. 1983. 212:791-800.
    View this article via: PubMed Google Scholar
26. Francone, OL, Subbaiah, PV, Van Tol, A, Royer, L, Haghpassand, M. Abnormal phospholipid composition impairs HDL biogenesis and maturation in mice lacking Abca1. Biochemistry. 2003. 42:8569-8578.
    View this article via: PubMed CrossRef Google Scholar
27. Huggins, KW, et al. Determination of the tissue sites responsible for the catabolism of large high density lipoprotein in the African green monkey. J. Lipid Res. 2000. 41:384-394.
    View this article via: PubMed Google Scholar
28. Lee, JY, et al. Prebeta high density lipoprotein has two metabolic fates in human apolipoprotein A-I transgenic mice. J. Lipid Res. 2004. 45:716-728.
    View this article via: PubMed CrossRef Google Scholar
29. Glass, C, Pittman, RC, Civen, M, Steinberg, D. Uptake of high-density lipoprotein-associated apoprotein A-I and cholesterol esters by 16 tissues of the rat in vivo and by adrenal cells and hepatocytes in vitro. J. Biol. Chem. 1985. 260:744-750.
    View this article via: PubMed Google Scholar
30. Glass, CK, Pittman, RC, Keller, GA, Steinberg, D. Tissue sites of degradation of apoprotein A-I in the rat. J. Biol. Chem. 1983. 258:7161-7167.
    View this article via: PubMed Google Scholar
31. Singaraja, RR, Brunham, LR, Visscher, H, Kastelein, JJP, Hayden, MR. Efflux and atherosclerosis: the clinical and biochemical impact of variations in the ABCA1 gene. Arterioscler. Thromb. Vasc. Biol. 2003. 23:1322-1332.
    View this article via: PubMed CrossRef Google Scholar
32. Wellington, CL, et al. ABCA1 mRNA and protein distribution patterns predict multiple different roles and levels of regulation. Lab. Invest. 2002. 82:273-283.
    View this article via: PubMed Google Scholar
33. Chisholm, JW, Burleson, ER, Shelness, GS, Parks, JS. ApoA-I secretion from HepG2 cells: evidence for the secretion of both lipid-poor apoA-I and intracellularly assembled nascent HDL. J. Lipid Res. 2002. 43:36-44.
    View this article via: PubMed Google Scholar
34. Kiss, RS, et al. The lipidation by hepatocytes of human apolipoprotein A-I occurs by both ABCA1-dependent and -independent pathways. J. Biol. Chem. 2003. 278:10119-10127.
    View this article via: PubMed CrossRef Google Scholar
35. Rothblat, GH, et al. Cell cholesterol efflux: integration of old and new observations provides new insights. J. Lipid Res. 1999. 40:781-796.
    View this article via: PubMed Google Scholar
36. Wang, N, Lan, D, Chen, W, Matsuura, F, Tall, AR. ATP-binding cassette transporters G1 and G4 mediate cellular cholesterol efflux to high-density lipoproteins. Proc. Natl. Acad. Sci. U. S. A. 2004. 101:9774-9779.
    View this article via: PubMed CrossRef Google Scholar
37. Van Eck, M, et al. Leukocyte ABCA1 controls susceptibility to atherosclerosis and macrophage recruitment into tissues. Proc. Natl. Acad. Sci. U. S. A. 2002. 99:6298-6303.
    View this article via: PubMed CrossRef Google Scholar
38. Aiello, RJ, et al. Increased atherosclerosis in hyperlipidemic mice with inactivation of ABCA1 in macrophages. Arterioscler. Thromb. Vasc. Biol. 2002. 22:630-637.
    View this article via: PubMed CrossRef Google Scholar
39. Assmann, G, Smootz, E. High density lipoprotein infusion and partial plasma exchange in Tangier disease. Eur. J. Clin. Invest. 1978. 8:131-135.
    View this article via: PubMed CrossRef Google Scholar
40. Denis, M, et al. Molecular and cellular physiology of apolipoprotein A-I lipidation by the ATP-binding cassette transporter A1 (ABCA1). J. Biol. Chem. 2004. 279:7384-7394.
    View this article via: PubMed CrossRef Google Scholar
41. Furbee (Jr), JW, Francone, OL, Parks, JS. Alteration of plasma HDL cholesteryl ester composition with transgenic expression of a point mutation (E149A) of human lecithin:cholesterol acyltransferase (LCAT). J. Lipid Res. 2001. 42:1626-1635.
    View this article via: PubMed Google Scholar
42. Livak, KJ, Schmittgen, TD. Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods. 2001. 25:402-408.
    View this article via: PubMed CrossRef Google Scholar
43. Furbee (Jr), JW, Francone, OL, Parks, JS. In vivo contribution of lecithin:cholesterol acyltransferase (LCAT) to apolipoprotein B lipoprotein cholesteryl esters in low density lipoprotein receptor and apolipoprotein E knockout mice. J. Lipid Res. 2002. 43:428-437.
    View this article via: PubMed Google Scholar
44. Koritnik, DL, Rudel, LL. Measurement of apolipoprotein A-I concentration in nonhuman primate serum by enzyme-linked immunosorbent assay (ELISA). J. Lipid Res. 1983. 24:1639-1645.
    View this article via: PubMed Google Scholar
45. Parks, J.S., Gebre, A.K., and Furbee, J.W., Jr. 1998. Lecithin-cholesterol acyltransferase. Assay of cholesterol esterification and phospholipase A2 activities. In Methods in molecular biology. M. Doolittle and K. Reue, editors. Humana Press. Totowa, New Jersey, USA. 123–131.
46. Wilcox, RW, Thuren, T, Sisson, P, Kucera, GL, Waite, M. Hydrolysis of neutral lipid substrates by rat hepatic lipase. Lipids. 1991. 26:283-288.
    View this article via: PubMed CrossRef Google Scholar
47. Furbee (Jr), JW, Sawyer, JK, Parks, JS. Lecithin:cholesterol acyltransferase deficiency increases atherosclerosis in the low density lipoprotein receptor and apolipoprotein E knockout mice. J. Biol. Chem. 2002. 277:3511-3519.
    View this article via: PubMed CrossRef Google Scholar
48. Twisk, J, et al. The role of the LDL receptor in apolipoprotein B secretion. J. Clin. Invest. 2000. 105:521-532.
    View this article via: JCI PubMed CrossRef Google Scholar
49. Wellington, CL, et al. Truncation mutations in ABCA1 suppress normal upregulation of full-length ABCA1 by 9-cis-retinoic acid and 22-R-hydroxycholesterol. J. Lipid Res. 2002. 43:1939-1949.
    View this article via: PubMed CrossRef Google Scholar
50. See, RH, et al. Protein kinase A site-specific phosphorylation regulates ATP-binding cassette A1 (ABCA1)-mediated phospholipid efflux. J. Biol. Chem. 2002. 277:41835-41842.
    View this article via: PubMed CrossRef Google Scholar