Apolipoprotein A-I is necessary for the in vivo formation of high density lipoprotein competent for scavenger receptor BI-mediated cholesteryl ester-selective uptake.

The severe depletion of cholesteryl ester (CE) in steroidogenic cells of apoA-I(-/-) mice suggests that apolipoprotein (apo) A-I plays a specific role in the high density lipoprotein (HDL) CE-selective uptake process mediated by scavenger receptor BI (SR-BI) in vivo. The nature of this role, however, is unclear because a variety of apolipoproteins bind to SR-BI expressed in transfected cells. In this study the role of apoA-I in SR-BI-mediated HDL CE-selective uptake was tested via analyses of the biochemical properties of apoA-I(-/-) HDL and its interaction with SR-BI on adrenocortical cells, hepatoma cells, and cells expressing a transfected SR-BI. apoA-I(-/-) HDL are large heterogeneous particles with a core consisting predominantly of CE and a surface enriched in phospholipid, free cholesterol, apoA-II, and apoE. Functional analysis showed apoA-I(-/-) HDL to bind to SR-BI with the same or higher affinity as compared with apoA-I(+/+) HDL, but apoA-I(-/-) HDL showed a 2-3-fold decrease in the V(max) for CE transfer from the HDL particle to adrenal cells. These results indicate that the absence of apoA-I results in HDL particles with a reduced capacity for SR-BI-mediated CE-selective uptake. The reduced V(max) illustrates that HDL properties necessary for binding to SR-BI are distinct from those properties necessary for the transfer of HDL CE from the core of the HDL particle to the plasma membrane. The reduced V(max) for HDL CE-selective uptake likely contributes to the severe reduction in CE accumulation in steroidogenic cells of apoA-I(-/-) mice.

It is well established that the risk of developing coronary heart disease is inversely proportional to plasma HDL 1 cholesterol and apoA-I levels (1,2), suggesting that HDL is protective against the development of atherosclerosis. The premise that HDL and apoA-I are protective against atherosclerosis is supported by studies showing that expression of human apoA-I in transgenic mice protects against the development of aortic lesions (3) and decreases atherosclerosis in apoE-deficient (4,5) and apo(a)-transgenic mice (6). The most widely accepted model explaining the anti-atherogenic properties of apoA-I is reverse cholesterol transport (7). In this process, poorly lipidated apoA-I first removes excess free cholesterol from peripheral cells through a mechanism dependent on ABCA1 (8).
ApoA-I then acts as a co-factor for LCAT, which catalyzes formation of cholesteryl ester (CE), resulting in the conversion of the poorly lipidated apoA-I to a spherical HDL particle. After being remodeled by cholesteryl ester transfer protein, hepatic lipase, and phospholipid transfer protein, HDL delivers its CE either to the liver, where it can be excreted or repackaged into new lipoproteins, or to ovaries, testes, and adrenal glands, where it can be used in the production of steroid hormones.
The importance of apoA-I in reverse cholesterol transport has been confirmed by the creation of mice deficient in this apolipoprotein. Compared with apoA-I ϩ/ϩ mice, apoA-I Ϫ/Ϫ mice have an ϳ70% reduction in total plasma cholesterol and HDL cholesterol (9 -11). In addition, LCAT activity in the plasma of the apoA-I Ϫ/Ϫ mice is 75% lower than in apoA-I ϩ/ϩ mice. Reduced LCAT activity is directly linked to the absence of apoA-I because addition of apoA-I to apoA-I Ϫ/Ϫ plasma restores LCAT activity to near normal levels (12). In addition to the decrease in plasma HDL levels, apoA-I Ϫ/Ϫ mice have a dramatic decrease in CE accumulation in the cortical cells of the adrenal gland, luteal and interstitial cells of the ovary, and Leydig cells of the testis (13). This reduction appears to reflect the absence of apoA-I and not simply the reduced HDL levels because apoA-II Ϫ/Ϫ mice have near normal CE levels in steroidogenic cells despite having plasma HDL and HDL CE levels equivalent to those in the apoA-I Ϫ/Ϫ mice (13). This result suggests a specific effect of apoA-I on CE accumulation in steroidogenic cells.
HDL provides cholesterol to cells via a selective uptake pathway in which HDL CE and free cholesterol (FC) are taken into the cell without the uptake and lysosomal degradation of the HDL particle (14 -19). The selective delivery of HDL cholesterol to cells differs from the classical LDL receptor pathway in which lipoprotein particles are degraded in lysosomes to release the constituent lipids for use by the cell (20). The selective uptake pathway is the major route for the delivery of HDL CE to the liver and steroidogenic cells of rodents in vivo and in vitro (14,17). In steroidogenic cells this pathway accounts for up to 90% of the cholesterol destined for steroid production or CE storage.
Recent studies indicate that scavenger receptor BI (SR-BI) is the cell surface receptor responsible for HDL CE-selective uptake (21)(22)(23). SR-BI is a 57-kDa, 509-amino acid protein that is highly conserved among human, mouse, rat, hamster, and cow (21, 24 -28). Evidence that SR-BI is a selective uptake receptor was provided by studies showing that SR-BI is expressed in tissues exhibiting high rates of HDL CE-selective uptake and is regulated in parallel to steroid production (21, 29 -32). Direct evidence for SR-BI function was provided by studies in which antibody to the extracellular domain of SR-BI inhibited delivery of HDL CE to the steroidogenic pathway in murine adrenocortical cells (33). Inactivation of the SR-BI gene in mice increased plasma HDL cholesterol levels and reduced neutral lipid stores in the adrenal gland and ovary (34,35). Mice carrying an induced mutation that reduced hepatic SR-BI expression levels by 50% also showed a reduction in HDL CEselective uptake (36). These studies are complemented by studies in which hepatic SR-BI was overexpressed by either an adenovirus vector (37) or via a transgene (38,39). In these studies SR-BI overexpression decreased HDL cholesterol and apoA-I levels, increased the plasma clearance of apoA-I, and increased hepatic HDL CE-selective uptake (38). Taken together, these data indicate that SR-BI plays a key role in mediating HDL CE-selective uptake by cells of the liver and steroidogenic organs and in determining the plasma levels of HDL cholesterol in mice.
The severe depletion of CE in steroidogenic cells of apoA-I Ϫ/Ϫ mice suggests that apoA-I plays a specific role in the HDL CE-selective uptake process mediated by SR-BI in vivo. The nature of this role, however, is unclear because a variety of lipoproteins and apolipoproteins can bind to SR-BI expressed in transfected cells (24,40,41). In addition, reconstituted HDL containing apoA-I, apoE, or apoA-II ϩ apoC proteins are all active in HDL CE-selective uptake into cultured adrenal cells (42). In the present study we tested the role of apoA-I in SR-BI-mediated HDL CE-selective uptake by analyses of the biochemical properties of apoA-I Ϫ/Ϫ HDL and the functional interaction of these particles with SR-BI on adrenocortical cells, hepatoma cells, and cells expressing a transfected SR-BI. The results show that large CE-rich apoA-I Ϫ/Ϫ HDL bind to SR-BI with the same affinity as apoA-I ϩ/ϩ HDL. In contrast, apoA-I Ϫ/Ϫ HDL show a reduced V max for CE transfer from the HDL particle to the cell. These results indicate that the absence of apoA-I results in HDL particles with a reduced capacity for SR-BI-mediated CE-selective uptake. The reduced V max for HDL CE transfer likely contributes to the severe reduction in CE accumulation in steroidogenic cells of apoA-I Ϫ/Ϫ mice.
Animals-C57BL/6J-Apoa1 tm1Unc (10) and "wild type" apoA-I ϩ/ϩ C57BL/6J were purchased from the Jackson Laboratory, and colonies were established at the Scripps Research Institute and, subsequently, at the La Jolla Institute for Molecular Medicine. At the State University of New York (Stony Brook, NY), the C57BL/6J-Apoa1 tm1Unc mice were cross-bred with apoA-I ϩ/ϩ FVB/N mice to overcome the reproductive deficiency of the C57BL/6J-Apoa1 tm1Unc mice (small litter size and low survival rate). This was necessary to obtain sufficient HDL for these studies. Progeny heterozygous for the mutated apoA-I allele were then mated with apoA-I ϩ/ϩ FVB/N mice for 3 generations. The genotypes of all offspring were determined by PCR analysis of genomic DNA as described (10). Heterozygotes having a 8:1 FVB/N:C57BL/6 background were then mated to generate mice carrying either two normal (apoA-I ϩ/ϩ ) or two mutated copies of the apoA-I allele (apoA-I Ϫ/Ϫ ). apoA-I ϩ/ϩ and apoA-I Ϫ/Ϫ animals were mated within their respective genotype to achieve the numbers of animals necessary for the experiments listed below. Additional FVB/N apoA-I ϩ/ϩ mice were obtained from Taconic and maintained as a separate colony. All mice were kept on a 12-h light/12-h dark cycle with standard rodent chow and water ad libitum. Housing and experimental procedures were approved by the State University of New York at Stony Brook Committee on Laboratory Animal Resources and the Scripps Research Institute Institutional Animal Care and Use Committee.
Isolation of apoA-I ϩ/ϩ and apoA-I Ϫ/Ϫ Lipoproteins-After an overnight fast, mice were anesthetized and exsanguinated by heart puncture. Cells were removed by centrifugation at 2,000 ϫ g for 30 min at 4°C. If not used immediately, plasma was frozen at Ϫ80°C after adding sucrose to a final concentration of 10%. Prior to isolation of lipoproteins, 0.05% NaN 3 , 5 g/ml aprotinin, 5 g/ml leupeptin, 1 g/ml pepstatin, and 1 mM EDTA (final concentrations) were added. Lipoproteins in the 1.02-1.21 g/ml density range were isolated by sequential ultracentrifugation using KBr at 1.02 and 1.21 g/ml, respectively. Following dialysis against PBS-E (150 mM NaCl, 10 mM K x H x PO 4 , 1 mM EDTA, pH 7.4), an equal amount of apoA-I ϩ/ϩ or apoA-I Ϫ/Ϫ lipoprotein total cholesterol was fractionated by HPLC using a Superose 6 HR 10/30 size exclusion column (Amersham Biosciences) run at a flow rate of 0.4 ml/min. Using the Cholesterol CII enzymatic assay (Wako), the total cholesterol concentration was determined for 40 0.4-ml fractions collected from 10 to 50 min after sample injection.
For structural and compositional analysis of the HDL, apoA-I ϩ/ϩ fractions 24 -30 and apoA-I Ϫ/Ϫ fractions 22-30 were pooled, concentrated using a Centricon 50 (Millipore), and dialyzed against PBS-E. Total cholesterol, free cholesterol, phospholipid, and triglyceride concentrations were measured using commercially available enzymatic kits (Wako). Cholesteryl ester concentration was calculated by multiplying the difference between total cholesterol and free cholesterol by 1.68 to account for the mass of the esterified acyl chain. A modified Lowry assay using horse IgG as a standard (Pierce) was used for protein measurement (43).
Density Gradient Ultracentrifugation of HDL-HDL radiolabeled with 125 I was overlaid on 1 ml of 1.30 g/ml KBr. A step gradient was created by sequentially adding the following KBr solutions: 3 ml of 1.21 g/ml, 4 ml of 1.12 g/ml, 3 ml of 1.063 g/ml, 1 ml of 1.02 g/ml. The gradient was centrifuged in a SW41 rotor at 39,000 rpm for 24 h at 15°C. After collecting 17 0.7-ml fractions from the top to the bottom of the gradient, the samples were counted for 125 I activity and weighed to determine density.
Dynamic Laser Light Scattering Analysis of HDL-For each HDL sample, particle diameters were determined optically by dynamic light scattering analysis with a Microtrac Series 150 Ultrafine particle analyzer fitted with a flexible conduit-sheathed probe tip (UPA-150, Microtrac, Clearwater, FL) (44). Particles were assumed to be transparent and non-spherical with a density of 1.16 g/ml and a refractive index of 1.46. Raw particle-size distributions were converted to population percentiles, which were used to calculate median particle diameter for each decile of lipoprotein size distribution.
Competitive Binding of apoA-I ϩ/ϩ and apoA-I Ϫ/Ϫ HDL-Y1-BS1 cells were washed twice with 1 ml of serum-free Ham's F-10-HEPES medium (4°C). HDL was added such that the final concentration was 5 g/ml protein for 125 I-apoA-I Ϫ/Ϫ HDL and 5, 10, 25, or 75 g/ml protein for unlabeled human HDL 3 , apoA-I ϩ/ϩ HDL, and apoA-I Ϫ/Ϫ HDL. After 2 h at 4°C, cells were washed three times with 0.1% bovine serum albumin/PBS, pH 7.4 and one time with PBS, pH 7.4 (4°C). Cells were lysed with 0.1 N NaOH, and the lysate was passed three times through a 28-gauge needle. Protein was determined (43), and 125 I was measured in a Packard Cobra-II ␥ counter.
Determination of Cell Association, Degradation, and Selective Uptake of HDL CE-Y1-BS1, Fu5AH, and ldlA[mSR-BI] cells in six-well plates were washed, the medium was replaced with serum-free Ham's F-10 medium, EMEM, or DMEM/F-12, respectively, and [ 3 H]COE-[ 125 I]DLT apoA-I ϩ/ϩ or [ 3 H]COE-[ 125 I]DLT apoA-I Ϫ/Ϫ HDL was added to the indicated concentration. Following a 4-h incubation at 37°C, cells were washed three times with PBS plus 0.1% bovine serum albumin, pH 7.4, one time with PBS, pH 7.4, lysed with 0.1 N NaOH, and passed five times through a 28-gauge needle. The lysate was then processed to determine trichloroacetic acid-soluble and -insoluble 125 I radioactivity and organic solvent-extractable 3 H-radioactivity as described (45,47). Trichloroacetic acid-insoluble 125 I radioactivity represents cell-associated HDL apolipoprotein, which is the sum of cell surface-bound apolipoprotein and endocytosed apolipoprotein that is not yet degraded. Trichloroacetic acid-soluble 125 I radioactivity represents endocytosed and degraded apolipoprotein that is trapped in lysosomes because of the dilactitol tyramine label (45,48). The sum of the 125 I-degraded and 125 I-cell-associated undegraded apolipoprotein expressed as CE equivalents was subtracted from the CE measured as extractable 3 H radioactivity to give the selective uptake of HDL CE (45,47). Values for these parameters are expressed as nanograms of HDL CE/mg of cell protein.
Binding and kinetic parameters were calculated by fitting the data to a one-site model using GraphPad Prism 3.00 (GraphPad Software Inc.).

RESULTS
Compositional Analysis of FVB/N apoA-I ϩ/ϩ and apoA-I Ϫ/Ϫ HDL-Prior to functional analysis of apoA-I ϩ/ϩ and apoA-I Ϫ/Ϫ HDL, the structural and compositional properties of the particles were defined because previous studies were unclear as to the effect of apoA-I deficiency on particle size and the content of CE and TG (9,11). Fractionation of lipoproteins within the density range of 1.02-1.21 g/ml by gel filtration chromatography showed that apoA-I Ϫ/Ϫ HDL was larger than apoA-I ϩ/ϩ HDL (Fig. 1A) as noted previously (9). Similar results were seen by nondenaturing gradient gel electrophoresis (Fig. 1A, inset) and by dynamic laser light scattering (Fig. 1B). apoA-I ϩ/ϩ HDL was mainly composed of particles approximately 11 nm in diameter. In contrast, apoA-I Ϫ/Ϫ HDL had a mean size of 13.1 nm with a broader distribution of particles, ranging from 10 to 17 nm in diameter, and with no evident lipoprotein subclasses. Consistent with these properties, density gradient ultracentrifugation showed apoA-I Ϫ/Ϫ HDL to have a broader distribution with a peak density of 1.075 g/ml compared with 1.10 g/ml for apoA-I ϩ/ϩ HDL (data not shown).
Compositional analysis of the FVB/N HDL showed that apoA-I Ϫ/Ϫ HDL had significantly more FC and phospholipid (PL) compared with apoA-I ϩ/ϩ HDL but very similar percent compositions of the core lipids, primarily CE. Because a previous analysis with C57BL/6 mice indicated that apoA-I Ϫ/Ϫ HDL had a lipid core of mainly TG (11), HDL from mice with a C57BL/6 genetic background was also examined (Table I). Like FVB/N apoA-I Ϫ/Ϫ HDL, the C57BL/6 apoA-I Ϫ/Ϫ HDL was enriched in FC and PL compared with apoA-I ϩ/ϩ HDL with a core that was primarily CE and not TG. The reason for the discrepancy in CE and TG content with the previous report is unclear (11).
The apolipoprotein distribution within HDL was analyzed by separating fractions 20 -30 from the gel filtration profile (Fig.  1) on 8 -20% SDS-polyacrylamide gradient gels. In apoA-I ϩ/ϩ HDL, apoA-I and apoA-II/apoC proteins were similarly distributed across the HDL peak ( Fig. 2A). ApoA-IV was localized to the front side of the HDL peak (fractions 24 -27), and a trace amount of apoE was present but restricted mainly to fractions coincident with apoB (fractions 20 -23) and not apoA-I (fractions 24 -30). In contrast to apoA-I ϩ/ϩ HDL, significant levels of apoE were found throughout the apoA-I Ϫ/Ϫ HDL profile (fractions 21-30). ApoA-II was also present throughout the apoA-I Ϫ/Ϫ HDL profile, but the quantitative distribution of apoA-II was shifted to smaller HDL, in contrast to that of apoE, which was primarily on larger HDL. This result, as well as the lack of discrete size fractions upon nondenaturing gradient gel electrophoresis (Fig. 1, inset), may indicate that apoA-II and apoE reside on the same apoA-I Ϫ/Ϫ HDL particles, but that has not been tested directly. ApoA-IV was also found on the apoA-I Ϫ/Ϫ HDL. However, unlike the other apolipoproteins, apoA-IV was found on particles of similar size in both apoA-I ϩ/ϩ and apoA-I Ϫ/Ϫ HDL (fractions 24 -28). This may indicate that apoA-IV forms a separate population of HDL that is not dependent on apoA-I. It is also important to note that, for apoA-I ϩ/ϩ fractions 25-30 and apoA-I Ϫ/Ϫ fractions 23-27, the high molecular protein mass in the 200 -500-kDa range most likely represents protein aggregates and not apoB. This is evident from additional analyses in which the extent of aggregation was proportional to the amount of sample loaded (data not shown).
Functional Properties of FVB/N apoA-I ϩ/ϩ and apoA-I Ϫ/Ϫ HDL-Functional properties of apoA-I ϩ/ϩ and apoA-I Ϫ/Ϫ HDL were tested using the Y1-BS1 mouse adrenocortical cell line (49). Like adrenocortical cells in vivo, Y1-BS1 cells respond to ACTH treatment by up-regulating SR-BI expression and SR-BI-dependent selective HDL CE uptake (31). SR-BI is the major route for the selective uptake and the provision of HDL CE to the steroidogenic pathway in Y1-BS1 cells (33). Similarly, most of the cell surface high affinity HDL binding by Y1-BS1 cells is the result of SR-BI, although HDL may bind to low affinity sites as well (33). Before evaluating the HDL CE-selective uptake properties of apoA-I Ϫ/Ϫ HDL, a competitive binding assay was conducted using 125 I-apoA-I Ϫ/Ϫ HDL with unlabeled human HDL 3 , apoA-I ϩ/ϩ HDL, or apoA-I Ϫ/Ϫ HDL as competitors (Fig. 3A, inset). The results showed that all three HDL particles competed similarly for the binding of the radiolabeled apoA-I Ϫ/Ϫ HDL. Hence, the majority of the high affinity binding of apoA-I Ϫ/Ϫ and apoA-I ϩ/ϩ is to the same cell surface binding site that was shown previously with human HDL 3 to be SR-BI (33).
The ability of Y1-BS1 cells to internalize CE from the apoA-I ϩ/ϩ and apoA-I Ϫ/Ϫ HDL was tested using HDL doubly labeled with [ 3
The ability of the Y1-BS1 cells to internalize HDL CE by either degradative or selective uptake was then analyzed (Fig.  3B). Over an HDL concentration range of 2.5-50 g/ml protein, 1.5-3-fold more CE was internalized by selective uptake from apoA-I ϩ/ϩ HDL compared with apoA-I Ϫ/Ϫ HDL. Fitting these data to a one-site model showed that the difference in selective HDL CE uptake from the two types of HDL was mainly caused by a difference in V max as opposed to the K m (Table II). The difference in the V max most likely represents the inability of SR-BI to efficiently remove the CE core from the apoA-I Ϫ/Ϫ HDL. The selective uptake efficiencies, which are the ratios of HDL CE-selective uptake to HDL CE cell association, were 1.8 -2.8-fold higher for apoA-I ϩ/ϩ HDL compared with apoA-I Ϫ/Ϫ HDL throughout the concentration range. Therefore, even when similar amounts of HDL CE were bound to SR-BI, significantly less CE was internalized from the apoA-I Ϫ/Ϫ HDL. Interestingly, the quantities of HDL taken up and degraded via the endocytic pathway were the same for both apoA-I Ϫ/Ϫ and apoA-I ϩ/ϩ HDL (Fig. 3B, Table II). Thus, the difference in HDL CE delivery to the cell from apoA-I Ϫ/Ϫ HDL is specific to the SR-BI-mediated selective uptake pathway. It is unclear whether the low level of HDL degradation (Fig. 3B) is the result of SR-BI or of other mechanisms.
To determine whether the reduction in selective CE uptake from apoA-I Ϫ/Ϫ HDL occurred with other cell types, studies were done with the Fu5AH and ldlA[mSR-BI] cell lines. The Fu5AH rat hepatoma cell line was chosen for two reasons. First, this cell line was derived from rat hepatocytes, which in vivo internalize the majority of plasma HDL CE by selective uptake (14, 17). Second, the Fu5AH cells have been shown to  naturally express a high level of SR-BI (50). At 25 g/ml HDL protein, Fu5AH cells showed similar amounts of HDL cell association (Fig. 4, panel A) and HDL degradation (panel B) with apoA-I Ϫ/Ϫ and apoA-I ϩ/ϩ HDL, but showed 3-fold more selective CE uptake from apoA-I ϩ/ϩ compared with apoA-I Ϫ/Ϫ HDL (panel C). Interestingly, the HDL cell association and degradation values for Fu5AH cells were reduced approximately 2-fold compared with Y1-BS1 cells, although the HDL CE-selective uptake values were nearly identical in both cell types.
The selective uptake efficiencies, which are the ratios of HDL CE-selective uptake to HDL CE cell association, were 1.5-1.7fold higher for apoA-I ϩ/ϩ HDL compared with apoA-I Ϫ/Ϫ HDL throughout the concentration range. The kinetic parameters for HDL CE-selective uptake in Table II indicate that, as for the Y1-BS1 cells, the difference between the apoA-I Ϫ/Ϫ and apoA-I ϩ/ϩ HDL lies in the V max. Interestingly, the K m values for selective uptake were lower for apoA-I Ϫ/Ϫ HDL as compared with apoA-I ϩ/ϩ HDL when expressed on a protein basis or on a particle molarity basis (Table II). A reduction in K m with apoA-I Ϫ/Ϫ HDL was also seen with Y1-BS1 cells, although the difference compared with apoA-I ϩ/ϩ HDL was greater with the Y1-BS1 cells compared with the ldlA[SR-BI] cells. These data show that apoA-I Ϫ/Ϫ HDL binds as efficiently or with a higher affinity compared with apoA-I ϩ/ϩ HDL, but shows a reduced selective uptake of CE into adrenocortical cells, hepatoma cells, and a cell line expressing a transfected SR-BI.

DISCUSSION
The present results demonstrate structural and functional differences between apoA-I ϩ/ϩ and apoA-I Ϫ/Ϫ mouse HDL. apoA-I Ϫ/Ϫ HDL consisted of particles that were generally larger and more heterogeneous than apoA-I ϩ/ϩ HDL. Although apoA-I Ϫ/Ϫ HDL are CE-rich and bind to SR-BI with a similar affinity compared with apoA-I ϩ/ϩ HDL, apoA-I Ϫ/Ϫ HDL show a 2-3-fold decrease in the V max for CE transfer to Y1-BS1 adrenocortical cells. A similar reduction in the efficiency of HDL CE-selective uptake was seen with Fu5AH hepatoma cells and a lesser reduction in an SR-BI-expressing CHO cell line. It is unclear why the quantitative difference in the selective uptake of apoA-I Ϫ/Ϫ and apoA-I ϩ/ϩ HDL CE is less in the transfected CHO cell, but this may be because of cell typespecific factors that occur in cells naturally expressing SR-BI. These results indicate that the absence of apoA-I results in HDL particles with a reduced capacity for SR-BI-mediated CE-selective uptake.
The apoA-I Ϫ/Ϫ mouse has a dramatic reduction in CE accumulation in steroidogenic cells; adrenocortical cells contain only 5% of the CE found in apoA-I ϩ/ϩ adrenal cells (13). The reduced V max of the apoA-I Ϫ/Ϫ HDL for HDL CE transfer likely contributes to, but cannot completely explain, this severe reduction in adrenal CE accumulation. Additionally, electron microscopic studies suggest that apoA-I Ϫ/Ϫ HDL are not well retained in the microvillar channels that are believed to be the cell surface compartment in which SR-BI-mediated HDL CEselective uptake occurs (13,51,52). The failure to retain HDL in microvillar channels may also contribute to adrenal CE depletion in the apoA-I Ϫ/Ϫ mouse. Whether this failure is a result of the absence of a specific interaction between apoA-I and microvillar channel components or is the result of other properties (e.g. size, lipid composition, apolipoprotein content) of apoA-I Ϫ/Ϫ HDL is unknown.
The observation that apoA-I Ϫ/Ϫ HDL binds to SR-BI with a similar K D (as measured by HDL protein concentration) or a lower K D (on a molar basis) compared with apoA-I ϩ/ϩ HDL is consistent with previous observations that SR-BI can bind a variety of HDL apolipoproteins with high affinity (41) and appears to do so by recognition of the multiple amphipathic ␣-helical repeat units present in the apolipoproteins (53). The reduced efficiency of CE uptake from apoA-I Ϫ/Ϫ HDL, however, clearly shows that lipoprotein binding to SR-BI does not translate to equivalent CE-selective uptake. Lipoprotein properties other than those needed for docking to SR-BI are important for SR-BI-mediated CE transfer to the cell. This conclusion is also consistent with the observation that SR-BI-mediated CE uptake from LDL is 6 -7-fold less efficient than from HDL despite the fact that SR-BI binds LDL with the same or a higher affinity (24, 54). An interesting observation from these studies is that the K m for selective uptake was considerably lower (2-5-fold) than the K D for HDL cell association. This was seen with both apoA-I ϩ/ϩ and apoA-I Ϫ/Ϫ HDL in both Y1-BS1 and ldlA[SR-BI] cells. This was noted previously for human HDL 3 interaction with Y1-BS1 cells (55). The reason for the lower K m is not known. One speculation is that subsets of SR-BI exist in active (perhaps occurring as a multimeric complex) and inactive (monomeric?) forms for HDL CE uptake. If the active form had a higher affinity for HDL, the K D for binding and the K m for CE uptake may be more similar in the subset of active receptors engaged in CE uptake.
Because apoA-I ϩ/ϩ and apoA-I Ϫ/Ϫ HDL differ significantly in their structural and compositional properties, many factors may contribute to the reduced selective uptake efficiency of the apoA-I Ϫ/Ϫ HDL. Included among these are potential effects of HDL lipid composition, particle size, and the increased apoA-II and apoE content, e.g. based on the chemical compositions and mean value for HDL size and density, the molecular compositions on a per particle basis indicate that the FC to PL ratio is greater for the apoA-I Ϫ/Ϫ HDL (0.264 Ϯ 0.024) compared with apoA-I ϩ/ϩ HDL (0.183 Ϯ 0.021). This difference (p ϭ 0.02) may reduce the fluidity of the PL monolayer and disrupt SR-BImediated transfer of CE from the particle core despite the fact that the apoA-I Ϫ/Ϫ HDL contains 1.6-fold more CE. However, this calculation is based on mean compositional values and may not account for the heterogeneity of properties that may occur in the broad size range of apoA-I Ϫ/Ϫ HDL.
The impact of HDL size on selective CE uptake was first studied by Pittman et al. (42), who showed that Y1-BS1 cells display more HDL CE-selective uptake from denser compared with more buoyant rat HDL or reconstituted apoA-I-containing HDL. In contrast, a recent report showed that SR-BI-expressing CHO cells display higher binding affinity (56,57) and selective CE uptake from human HDL with low compared with high density (57). The results of the latter study suggested that the major difference in CE uptake was caused by the altered binding affinity of the more buoyant HDL rather than a difference in the efficiency of CE transfer from the HDL. Our results with apoA-I Ϫ/Ϫ HDL coincide with the former findings because we observed less selective CE uptake from the larger, more buoyant apoA-I Ϫ/Ϫ HDL.
Another factor that could influence selective CE uptake from the apoA-I Ϫ/Ϫ HDL is the apolipoprotein complement of the particles. Compared with apoA-I ϩ/ϩ HDL, apoA-I Ϫ/Ϫ HDL lack apoA-I and are enriched in apoE and apoA-II. An inhibitory effect of apoE was suggested by the observation that the in vivo clearance of CE from apoE-rich HDL 1 in the rat is slower compared with the clearance of CE from apoE-free HDL (58). In contrast, Arai et al. (59) observed that selective uptake of HDL CE is enhanced by the cellular expression of apoE although it was not determined whether this effect was caused by apoE altering HDL binding or CE transfer from the HDL. With regard to apoA-I, Pittman et al. (42) showed that Y1-BS1 cells displayed the highest level of selective CE uptake from recon-  stituted HDL containing apoA-I as compared with apoA-II/Cs or apoE. Similarly, Pilon et al. (60) reported that displacement of apoA-I on human HDL by apoA-II increased HDL binding but decreased selective HDL CE uptake by a human adrenocortical cell line. Studies with human hepatocytes and hepatoma cells also showed greater CE-selective uptake from HDL containing only apoA-I compared with HDL containing both apoA-I and apoA-II (61). These results suggest that apoA-I may enhance whereas apoA-II may inhibit SR-BI-mediated HDL CE-selective uptake. Both of these possibilities could contribute to the reduced efficiency of HDL CE-selective uptake with apoA-I Ϫ/Ϫ HDL. However, de Beer et al. (62) observed that CHO cells expressing SR-BI displayed less binding and more selective CE uptake from recombinant discoidal HDL prepared with apoA-I and apoA-II compared apoA-I alone. All of these studies, including the present results, are consistent with the idea that specific HDL apolipoproteins may influence the actual CE transfer process mediated by SR-BI, although the multiple variables and different experimental models in these studies preclude a clear interpretation at this time.
The increased size and heterogeneity of the apoA-I Ϫ/Ϫ HDL suggest that apoA-I plays a key role in limiting the enlargement of HDL particles during circulation in the blood. The formation and enlargement of spherical HDL depend on CE synthesis by LCAT (7). Although plasma LCAT activity is substantially reduced in the apoA-I Ϫ/Ϫ mouse (12), there is clearly sufficient activity to generate large CE-rich HDL. Two scenarios may be suggested for the generation of the large, CE rich apoA-I Ϫ/Ϫ HDL in vivo. The first is an increased circulation time of the apoA-I Ϫ/Ϫ HDL, which would permit LCAT to slowly enlarge the HDL CE core. The data of Plump et al. (11) would argue against this idea because they observed no difference in the fractional catabolic rates of apoA-I ϩ/ϩ and apoA-I Ϫ/Ϫ HDL CE.
The second scenario is that, in the absence of apoA-I, HDL particles fuse to form larger HDL. There is some precedent for this possibility in that apoA-I will prevent the in vitro fusion of LDL with cholesteryl ester microemulsion particles (63). Perhaps the absence of apoA-I leads to HDL particle fusion in the apoA-I Ϫ/Ϫ mouse, a process that would be expected to generate large particles with excess surface components. Consistent with this possibility is that the surface (% protein ϩ FC ϩ PL) to core (% CE ϩ TG) ratio for the apoA-I Ϫ/Ϫ HDL (2.2) is nearly the same as this ratio for apoA-I ϩ/ϩ HDL (2.1) despite the fact that the apoA-I Ϫ/Ϫ HDL are larger and would be expected to have a reduced surface to core ratio. Based on the average particle diameters of 11 and 13.1 nm for apoA-I ϩ/ϩ and apoA-I Ϫ/Ϫ HDL ( Fig. 2A), respectively, and a surface monolayer thickness of 2.15 nm (64), the predicted surface to core ratio for apoA-I Ϫ/Ϫ HDL can be calculated to be 70% of the value for apoA-I ϩ/ϩ HDL. The observation that the surface to core ratios for the two particles are nearly identical suggests that the apoA-I Ϫ/Ϫ HDL has an excess of surface components, a result consistent with the formation of large apoA-I Ϫ/Ϫ HDL via fusion of smaller particles. Irrespective of whether apoA-I limits HDL size by preventing particle fusion or by defining maximum particle size, or both, this appears to be a unique property of apoA-I that is not shared by apoA-II or the other apolipoproteins that accumulate on the large apoA-I Ϫ/Ϫ HDL.
In summary, these results demonstrate major structural differences between apoA-I ϩ/ϩ and apoA-I Ϫ/Ϫ HDL that translate to differences in SR-BI-mediated HDL CE-selective uptake but have little effect on HDL CE uptake by the endocytic pathway. The absence of apoA-I leads to the formation of HDL particles that bind to SR-BI with the same or greater affinity compared with apoA-I ϩ/ϩ HDL but have a reduced V max for SR-BI-mediated CE uptake. This reduced V max illustrates that HDL properties necessary for binding to SR-BI are distinct from those properties necessary for the transfer of HDL CE to the cell. The reduced V max of apoA-I Ϫ/Ϫ HDL likely contributes to the severe reduction of CE accumulation seen in adrenal cells of the apoA-I Ϫ/Ϫ mouse.