Molecular and Cellular Physiology of Apolipoprotein A-I Lipidation by the ATP-binding Cassette Transporter A1 (ABCA1)*

The dynamics of ABCA1-mediated apoA-I lipidation were investigated in intact human fibroblasts induced with 22(R)-hydroxycholesterol and 9-cis-retinoic acid (stimulated cells). Specific binding parameters of 125I-apoA-I to ABCA1 at 37 °C were determined: Kd = 0.65 μg/ml, Bmax = 0.10 ng/μg cell protein. Lipid-free apoA-I inhibited the binding of 125I-apoA-I to ABCA1 more efficiently than pre-β1-LpA-I, reconstituted HDL particles r(LpA-I), or HDL3 (IC50 = 0.35 ± 1.14, apoA-I; 1.69 ± 1.07, pre-β1-LpA-I; 17.91 ± 1.39, r(LpA-I); and 48.15 ± 1.72 μg/ml, HDL3). Treatment of intact cells with either phosphatidylcholine-specific phospholipase C or sphingomyelinase affected neither 125I-apoA-I binding nor 125I-apoA-I/ABCA1 cross-linking. We next investigated the dynamics of apoA-I lipidation by monitoring the kinetic of apoA-I dissociation from ABCA1. The dissociation of 125I-apoA-I from normal cells at 37 °C was rapid (t½ = 1.4 ± 0.66 h; n = 3) but almost completely inhibited at either 15 or 4 °C. A time course analysis of apoA-I-containing particles released during the dissociation period showed nascent apoA-I-phospholipid complexes that exhibited α-electrophoretic mobility with a particle size ranging from 9 to 20 nm (designated α-LpA-I-like particles), whereas lipid-free apoA-I incubated with ABCA1 mutant (Q597R) cells was unable to form such particles. These results demonstrate that: 1) the physical interaction of apoA-I with ABCA1 does not depend on membrane phosphatidylcholine or sphingomyelin; 2) the association of apoA-I with lipids reduces its ability to interact with ABCA1; and 3) the lipid translocase activity of ABCA1 generates α-LpA-I-like particles. This process plays in vivo a key role in HDL biogenesis.

step in reverse cholesterol transport (RCT) process, one of the several proposed mechanisms by which HDL may protect against atherosclerotic vascular disease (1)(2)(3).
The molecular interaction of apoA-I with ABCA1 promotes cholesterol efflux from peripheral cells and macrophages and is critical for the initial formation of HDL particles (1). The importance of ABCA1 in the lipidation of apoA-I has been strikingly demonstrated by the identification of mutations at the ABCA1 gene locus as the molecular defect of Tangier Disease (TD) and Familial HDL Deficiency (FHD) (4,5). These patients are characterized by extremely low HDL-cholesterol levels, caused by inadequate transport of cellular cholesterol and phospholipids to the extracellular space, leading to hypercatabolism of lipid-poor nascent HDL particles (6).
ApoA-I has been shown to interact with many proteins including high-density lipoprotein-binding protein (HBP, vigilin), HB2 (7), annexin I, annexin VII (8), fibronectin, collagen I (9,10), and the human ␤-chain of ATP synthase (11). However, the physiological significance of these interactions remains unknown. On the other hand, it is well established that apoA-I binds to the scavenger receptor class B type I (SR-BI) (12), which participates in selective uptake of HDL-derived cholesteryl esters, but so far no role for SR-BI in apoA-I-mediated lipid efflux has been found.
Although several studies have suggested a molecular interaction between apoA-I and ABCA1 at the cell surface (13)(14)(15), the role of ABCA1 as a candidate apoA-I receptor is still a matter of debate. At least two different mechanisms are proposed for this interaction. First, it is reported that a direct protein-protein interaction occurs between apoA-I and ABCA1 on the basis of chemical cross-linking experiments (13). A second hypothesis has been proposed suggesting an interaction between apoA-I and lipid domains in the cell membranes formed by the phospholipid translocase activity of ABCA1 (14). Indeed, studies by Remaley et al. (16,17) have shown that a majority of the plasma apolipoproteins containing lipophilic class A amphipathic helices can also promote lipid efflux and bind to ABCA1. Furthermore, the amphipathic helix was found to be a key structural motif for peptide-mediated lipid efflux from ABCA1.
Without knowledge of specific binding parameters of apoA-I-containing particles to ABCA1, it is not possible to predict whether ABCA1 might function as a significant receptor for apoA-I in the presence of other apolipoproteins, which demonstrate affinity for the same protein. A recent study by Basso et al. (18) demonstrating that the hepatic expression of ABCA1 is an important source of plasma HDL-C has stimulated our interest for apoA-I lipidation in peripheral cells. In the present study, experiments were directed at defining the mechanism by which apoA-I is lipidated by ABCA1 and how the formation of the apoA-I/ABCA1 complex can be affected by apoA-I conformation within discoidal and spherical HDL particles, by specific hydrolysis of plasma membranes phospholipids, or by naturally occurring mutants of ABCA1. In addition, the dynamics of apoA-I lipidation were investigated by determining the kinetic parameters of apoA-I/ABCA1 dissociation and the characterization of apoA-I-containing particles generated during this process.

EXPERIMENTAL PROCEDURES
Patient Selection-For the present study, we selected fibroblasts from 3 normal control subjects and 1 patient with TD (homozygous for Q597R at the ABCA1 gene). The protocol for the study was reviewed and accepted by the Research Ethics Board of the McGill University Health Center. Separate consent forms for blood sampling, DNA isolation, and skin biopsy were provided.
Cell Culture-Human skin fibroblasts were obtained from 3.0-mm punch biopsies of the forearm of patients and healthy control subjects and were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 0.1% nonessential amino acids, penicillin (100 units/ml), streptomycin (100 g/ml), and 10% fetal bovine serum.
ApoA-I Binding Assay-ApoA-I was iodinated with 125 I by IODO-GEN® (Pierce) to a specific activity of 800 -2500 cpm/ng apoA-I. Cells were grown on 24-well plates and were stimulated or not with 2.5 g/ml 22(R)-hydroxycholesterol and 10 M 9-cis-retinoic acid for 20 h. Cells were then incubated at 37°C with 125 I-apoA-I in DMEM/BSA (1 mg/ml) as specified for each experiment in the presence or absence of a 30-fold excess of unlabeled apoA-I, to subtract the nonspecific binding. The cells were then washed rapidly two times with ice-cold PBS/BSA, two times with cold PBS and lysed with 0.1 N NaOH. The amount of bound iodinated ligand was determined by ␥-counting.
Chemical Cross-linking and Immunoprecipitation Analysis-Chemical cross-linking was performed as described by Wang et al. (19) with a minor modification. Fibroblasts were grown to confluence in 100-mm diameter dishes and then stimulated or not with 2.5 g/ml 22(R)hydroxycholesterol and 10 M 9-cis-retinoic acid for 20 h in DMEM/ BSA. Cells were incubated in the presence or absence of either 3 g/ml of unlabeled apoA-I or 10 g/ml of 125 I-apoA-I in DMEM/BSA for 1 h at 37°C. Cells were then placed on ice for 15 min and washed three times with PBS. DSP (cross-linker agent) was dissolved immediately before use in dimethyl sulfoxide (Me 2 SO) and diluted to 500 M with PBS. 8 ml of DSP solution was added in each well. Cells were then incubated at room temperature for 1 h; the medium was removed, and the cells were washed twice with PBS. Cells were lysed at 4°C with IP buffer containing 20 mM Tris (pH 7.5), 0.5 mM EDTA, 0.5 mM EGTA, 1% Triton X-100 (Invitrogen), and the suspension was allowed to stand for 30 min at 4°C in presence of a protease inhibitor mixture (Roche Diagnostics). ApoA-I/ABCA1 complex was immunoprecipitated with an affinity-purified polyclonal anti-ABCA1 antibody (Novus Biologicals) as previously described (20). After SDS-gel electrophoresis either apoA-I or ABCA1 were detected by immunopurified polyclonal anti-apoA-I antibody (Biodesign) or affinity-purified human anti-ABCA1 antibody (Novus). The presence of labeled 125 I-apoA-I/ABCA1 complexes were directly detected by autoradiography using XAR-2 Kodak film.
Quantitative Cross-linking of ApoA-I to ABCA1-Fibroblasts were grown to confluence in 100-mm diameter dishes and then stimulated for 20 h. Cells were incubated at 37°C for 1 h in the presence or absence of 5 units/ml PC-PLC or 0.4 units/ml SM-ase. After washing to remove phospholipases, cells were incubated with 10 g/ml of 125 I-apoA-I (1500 -2500 cpm/ng) in the presence or absence of a 20-fold excess of unlabeled apoA-I. Cells were then placed on ice for 15 min and washed three times with PBS, and then cross-linking with DSP was performed as described above. Samples containing 125 I-apoA-I cross-linked to ABCA1 (200 g of total protein) were incubated with 10 l of affinitypurified human anti-ABCA1 antibody for 20 h at 4°C, followed by the addition of protein A bound to Sepharose (30 l) as we have described previously (21). The amount of bound iodinated apoA-I to ABCA1 in the immunoprecipitates was determined by ␥-counting. ABCA1 mutant (Q597R) was used as a negative control.
Dissociation of Specifically Bound 125 I-ApoA-I from Intact Cells-Fibroblasts were grown to near confluence in 24-well plates and then stimulated with 2.5 g/ml 22(R)-hydroxycholesterol and 10 M 9-cisretinoic acid for 20 h in DMEM/BSA. The cells were incubated for 2 h at 37°C with 10 g/ml of 125 I-apoA-I in the presence of 1 mg/ml BSA. For nonspecific binding determination, cells were incubated with a 30-fold excess of unlabeled apoA-I. After washing to remove unbound 125 I-apoA-I, 0.5 ml of DMEM was added, and the plates were immediately incubated at 37°C, 15°C, or 4°C for the indicated times. The medium was then collected, cells were lysed in 0.1 N NaOH, and the radioactivity in the medium and in the cells was determined by ␥-counting.
Cellular Lipid Efflux and Lipid Labeling-Phospholipid and cholesterol efflux were determined as previously described (3)  were stimulated or not with 2.5 g/ml of 22(R)-hydroxycholesterol and 10 M 9-cis-retinoic acid for 20 h. Phospholipid or cholesterol efflux were determined at either 2 or 24 h with 10 g/ml apoA-I. Cellular lipid efflux was determined as follow: 3 H cpm in medium/( 3 H cpm in medium ϩ 3 H cpm in cells); the results were expressed as percent of total radiolabeled phospholipids or cholesterol. Cell phospholipids were also labeled with [ 32 P]orthophosphate as follows: fibroblasts from control subject were grown to confluence in 100-mm or 150-mm diameter dishes and were incubated for 72 h with 300 -1500 Ci of [ 32 P]orthophosphate mixed with DMEM. The cells were stimulated as described above before incubation with lipid-free apoA-I as specified for each experiment.
Separation of Lipoproteins by Two-dimensional Non-denaturing Gradient Gel Electrophoresis-ApoA-I-containing particles were separated by two-dimensional-PAGGE, as previously described (22,23). Briefly, samples (30 -100 l) were separated in the first dimension (according to their charge) by 0.75% agarose gel electrophoresis (100 V, 3 h, 4°C) and in the second dimension (according to the size) by 5-23% polyacrylamide concave gradient gel electrophoresis (125 V, 24 h, 4°C). Iodinated high molecular weight protein mixture (7.1-17.0 nm, Amersham Biosciences) was run as a standard on each gel. Electrophoretically separated samples were electrotransferred (30 V, 24 h, 4°C) onto nitrocellulose membranes (Hybond ECL, Amersham Biosciences). ApoA-I-containing particles were detected by incubating the membranes with immunopurified polyclonal anti-apoA-I antibody (Biodesign) labeled with 125 I. The presence of labeled 125 I-apoA-I or 32 P-phospholipids were directly detected by autoradiography using XAR-2 Kodak film.
Preparation of Reconstituted HDL Particles (rLpA-I)-Complexes comprising apoA-I, POPC, and cholesterol were prepared using the sodium cholate dialysis method (24). ApoA-I/POPC/cholesterol molar ratio of 1:100:5 was used in this experiment. r(LpA-I) particles were further concentrated by ultrafiltration (spiral ultrafiltration cartridge, MWCO 50,000, Amicon) to discard any lipid-free apoA-I or proteolytic peptides. ApoA-I-lipid complex formation was verified by analysis with two-dimensional-PAGGE.
Pre-␤ 1 -LpA-I Purification from Plasma-Pre-␤ 1 -LpA-I was purified from freshly drawn venous blood under nondenaturing conditions as described by Kunitake et al. (25) with the following modifications. Typically, blood is drawn into a tube containing 1 mM sodium EDTA, 0.02% NaN 3 , 2 mM DTNB, and cooled immediately on ice. Plasma is separated by low speed centrifugation (1,800 ϫ g, 30 min) and aliquots (20 ml) were subjected to human immunopurified anti-apoA-I antibody (12171-21A, Genzyme Corp)-coupled Sepharose column (23,26). ApoA-I-containing fractions were then dialyzed and concentrated. Samples were separated by agarose gel electrophoresis, and the pre-␤-migrating region was excised out. Agarose gel pieces containing the pre-␤-migrating region were placed at the top of 3-26% non-denaturing gradient gels, as previously described (27). An immunoblot of apoA-I-containing lipoproteins separated by two-dimensional-PAGGE gels was used as a template to localize pre-␤ 1 -LpA-I particles, which are recovered from the gels by electroelution. Pre-␤ 1 -LpA-I particles were further concentrated by ultrafiltration (spiral ultrafiltration cartridge, MWCO 50,000, Amicon) to discard any lipid-free apoA-I or proteolytic peptides. The integrity of isolated plasma pre-␤ 1 -LpA-I fraction was verified by two-dimensional-PAGGE.

RESULTS
In the present study, we have examined the binding of 125 I-apoA-I to ABCA1 in normal cultured human fibroblasts. To determine the specific binding of 125 I-apoA-I to ABCA1, binding studies were performed in fibroblasts in which ABCA1 was induced with 22(R)-hydroxycholesterol and 9-cis-retinoic acid (stimulated cells), as well as in unstimulated cells. As shown in Fig. 1A, a marked and consistent increased binding of 125 I-apoA-I to stimulated cells was measured. However, significant binding was also observed in unstimulated cells. This is presumably due to basal level of ABCA1 expression and the presence of other apoA-I binding sites at the cell surface. We have not been able to detect any SR-BI receptor presence in fibroblasts as compared with hepatocytes as examined by gel electrophoresis of cellular membranes fraction followed by immunoblotting with an anti-SR-BI antibody (data not shown). The specific binding curve and the binding parameters K d and B max for apoA-I/ABCA1 interactions were determined by subtracting the binding values for the unstimulated cells from the corresponding values from stimulated cells. In the present binding assay apoA-I binds to ABCA1 (ABCA1 specific) with relatively high affinity (K d ϭ 0.65 Ϯ 0.20 g/ml), and the binding was saturable (B max ϭ 0.10 Ϯ 0.05 ng/g cell protein) (Fig. 1A). Maximum specific binding of apoA-I to ABCA1 was reached in less than 30 min and remained constant for the remaining 2 h of the experiment (data not shown). To ensure that the binding parameters obtained reflect specific increased of 125 I-apoA-I/ ABCA1 association in stimulated cells, the cross-linking of apoA-I to ABCA1 was examined. As shown in Fig. 1B, apoA-I forms a complex with ABCA1. Furthermore, stimulation of cells lead to an increase of both cellular ABCA1 expression and apoA-I/ABCA1 cross-linking compared with unstimulated cells. At the same time, phospholipid and cholesterol efflux were increased in stimulated cells, as shown in Fig. 1C. In order to verify that the specific association of 125 I-apoA-I with ABCA1 was dependent on the temperature, stimulated cells were incubated with 10 g/ml of 125 I-apoA-I for 2 h at either 37°C, 20°C or 4°C, and then specific 125 I-apoA-I cell association was determined as described above. Association with 125 I-apoA-I-stimulated cells showed remarkable temperature dependence (100 Ϯ 2%, 29 Ϯ 4% and 13 Ϯ 2%; 37°C, 20°C and 4°C, respectively). Results are expressed as percent of the incubation at 37°C (100%).
It is well established that the conformation of apoA-I within HDL particles is affected by its association with lipid molecules. It was therefore of interest to determine whether apoA-I conformation/organization within particles would affect its interaction with ABCA1. Competition assays were performed to determine the ability of pre-␤ 1 -LpA-I, as well as discoidal reconstituted HDL r(LpA-I) and spherical HDL particles (HDL 3 ), to compete for the binding of 125 I-apoA-I to ABCA1 in stimulated cells. As shown in Fig. 2A, lipid-free apoA-I inhibited the binding of 125 I-apoA-I to ABCA1 more efficiently than either isolated plasma pre-␤ 1 -LpA-I, reconstituted HDL particles r(LpA-I) (11-12 nm of diameter), or native HDL 3 (IC 50 ϭ 0.35 Ϯ 1.14, apoA-I; 1.69 Ϯ 1.07, pre-␤ 1 -LpA-I; 17.91 Ϯ 1.39, r(LpA-I); and 48.15 Ϯ 1.72 g/ml, HDL 3 ). Control experiments were conducted to examine whether the apparent decrease in cell binding of the labeled apoA-I may be due to the 125 I-apoA-I binding to different competitors particles instead of the cells. An experiment was therefore carried out in which either r(LpA-I) or HDL 3 particles were incubated with 125 I-apoA-I under similar conditions used for apoA-I binding assay, and then the samples were separated by fast protein liquid chromatography (FPLC). No significant amount of 125 I-apoA-I was found associated with r(LpA-I) or HDL 3 (data not shown), supporting our results shown in Fig. 2A. To verify the integrity of competitors particles, either isolated pre-␤ 1 -LpA-I, lipid-free apoA-I, r(LpA-I), or plasma were separated by two-dimensional-PAGGE and apoA-I was detected with immunopurified polyclonal anti-apoA-I antibody labeled with 125 I, as shown in Fig. 2B.
Because the lipid binding characteristics of apoA-I have been proposed to be important in the apoA-I/ABCA1 interaction (14,15,28), the question was raised whether the binding of apoA-I to ABCA1 was dependent on the presence of lipids at the cell surface. Stimulated cells were incubated for 60 min at 37°C in the presence or absence of either 5 units/ml phosphatidylcholine-specific phospholipase C (PC-PLC) or 0.4 units/ml sphingomyelinase (SM-ase). To assess how effectively phospholipids were removed by phospholipases treatment, the cells were labeled with [ 3 H]choline and the lipids separated by TLC and counted. PC-PLC and SM-ase treatment digested greater than 65% of phosphatidyl [ 3 H]choline and 80% of [ 3 H]sphingomyelin, respectively. Cells were then incubated with 10 g/ml of 125 I-apoA-I for 2 h at 37°C and specific 125 I-apoA-I binding was determined as described above. As shown in Fig. 3A, no significant effect of phospholipases treatment on the 125 I-apoA-I binding level was observed. In order to further verify that the interaction of apoA-I with ABCA1 was not dependent on the presence of plasma membrane phosphatidylcholine or sphingomyelin, we determined whether the cross-linking of apoA-I to ABCA1 could be affected by phospholipases treatment. Quantitative cross-linking of apoA-I to ABCA1 was performed as described in "Experimental Procedures." As shown in Fig. 3B (lower panel), treatment of intact cells with phospholipases did not affect significantly 125 I-apoA-I cross-linking to ABCA1. The presence of a 20-fold excess of unlabeled apoA-I (200 g/ml) reduced the cross-linking of 125 I-apoA-I to ABCA1 by 78% of control. In addition, ABCA1 mutant (Q597R) that has been shown previously to not cross-link to apoA-I (13) was used as a negative control for the present experiment and showed no binding or cross-linking to ABCA1 (Fig. 3, A and B). To ensure that the immunoprecipitates contained only 125 I-apoA-I/ ABCA1 complex, immunoprecipitated samples were analyzed by 4 -22.5% SDS-polyacrylamide gel electrophoresis. As shown in Fig. 3B (upper panel), only 125 I-apoA-I/ABCA1 complex was detected in immunoprecipitated samples. Also, to rule out the possibility that treatment with phospholipases might induce membrane aggregation or affect ABCA1 protein structure, which may trap apoA-I and result in nonspecific cross-linking, we examined the effect of phospholipases treatment on the cross-linking of 125 I-apoA-I to ABCA1 mutant (Q597R). As shown in Fig. 3B (upper panel), 125 I-apoA-I did not cross-link to Q597R mutant whether treated with PC-PLC, SM-ase, or left intact.
We initially hypothesized that any specific apoA-I dissociation from ABCA1 would be associated with a significant increase in apoA-I lipidation state. To better understand the mechanism by which apoA-I was lipidated by ABCA1, the kinetics of the dissociation of apoA-I from ABCA1 were investigated in stimulated cells. Fig. 4A depicts the time course of the dissociation of bound 125 I-apoA-I from stimulated normal cells at 37°C. The dissociation of 125 I-apoA-I from ABCA1 at 37°C was rapid (t1 ⁄2 ϭ 1.4 Ϯ 0.66 h; n ϭ 3). In contrast, 125 I-apoA-I dissociation from ABCA1 was almost completely inhibited at either 4 or 15°C (Fig. 4A). Practically all radioactivity that disappeared from the cell surface appeared as intact 125 I-apoA-I in the medium (more than 95% of the radioactivity released to the medium was precipitated by 10% trichloroacetic acid).
To further investigate the relationship between apoA-I dissociation from ABCA1 and apoA-I-mediated cellular cholesterol efflux in our stimulated cell culture system, the kinetics of cholesterol efflux in stimulated cells was determined in the presence of 10 g/ml of apoA-I (saturating binding concentrations). As shown in Fig. 4B, apoA-I-mediated cholesterol efflux reached saturation after a 16-h incubation period. In addition, to ensure that cholesterol efflux was ABCA1-dependent in our cell culture system, apoA-I-mediated cholesterol efflux from ABCA1 mutant (Q597R) cells was also monitored.
In order to investigate the nature of apoA-I-containing particles generated by ABCA1 activity, stimulated cells from either normal or from TD (Q597R) subjects in 100 mm diameter dishes were incubated with 10 g/ml of 125 I-apoA-I in 8 ml of DMEM for 24 h at 37°C. The medium was concentrated and 125 I-apoA-I-containing particles were separated by two-dimensional-PAGGE. As shown in Fig. 5 (panel B), apoA-I-containing particles generated by stimulated normal cells exhibited ␣-electrophoretic mobility with a particle diameter ranging from 9 to 20 nm, however, a significant amount of apoA-I was detected in the pre-␤-region. In contrast, lipid-free apoA-I incubated with stimulated mutant Q597R cells was unable to form such particles (panel C), which had a molecular diameter and charge similar to the lipid-free apoA-I incubated in the same conditions without cells (panel A).
To further characterize apoA-I-containing particles released specifically from ABCA1 during the dissociation period, stimulated cells in 150-mm diameter dishes were incubated with 10 g/ml 125 I-apoA-I for 2 h at 37°C. After washing to remove unbound 125 I-apoA-I, 15 ml of DMEM was added, and the plates were immediately incubated at 37°C for 1.4, 8, and 24 h. The medium was concentrated and 125 I-apoA-I-containing particles were electrophoretically separated by two-dimensional-PAGGE. As shown in Fig. 6, 125 I-apoA-I incubated for 24 h without cells had a pre-␤ electrophoretic mobility with a molecular diameter of 7.1 nm (panel A). However, apoA-I-containing particles dissociated from normal stimulated cells at either 1.4, 8, or 24 h exhibited ␣-electrophoretic mobility with a particle size ranging from 9 to 20 nm (designated ␣-LpA-I-like particles) (panels B-D, respectively). Both the charge and size of these nascent particles are stable over a 24-h dissociation period. We next examined whether the ␣-electrophoretic mobility of LpA-I-like particles may be caused by specific phospholipid composition. Cells were first labeled with [ 32 P]orthophosphate, then stimulated and incubated with 10 g/ml of unlabeled apoA-I for 2 h at 37°C. Dissociated 32 Pphospholipidated apoA-I was analyzed by two-dimensional-PAGGE as described above. As shown in Fig. 6 (panels E-G), 32

DISCUSSION
In view of the importance of ABCA1 in the regulation of plasma HDL cholesterol (1, 3, 18), we investigated the molecular and physiological mechanisms of ABCA1-dependent apoA-I lipidation in fibroblasts as a model for peripheral cells. Consistently with an earlier study by Remaley et al. (16), we show that specific binding parameters of 125 I-apoA-I to ABCA1 could be measured in ABCA1-stimulated cells (Fig. 1A). The specificity of 125 I-apoA-I binding to ABCA1 was supported by experiments showing that apoA-I forms a complex with ABCA1 in unstimulated cells, and this effect was markedly enhanced in stimulated cells (Fig. 1B). Moreover, increased apoA-I/ ABCA1 complex formation was concomitant with increased cellular phospholipid and cholesterol efflux (Fig. 1C). Several groups have reported binding studies with apoA-I conducted at 4°C and the results have been somewhat inconsistent (14,29). Here we demonstrate that the apoA-I cell association showed remarkable temperature dependence, suggesting that apoA-I binding to ABCA1 may be controlled by an energy-dependent process or, alternatively, the change in temperature may alter the lipid conformation in the plasma membrane, which then could affect apoA-I cell association. This result underscores the importance of using physiological temperatures to study apoA-I/ABCA1 interactions. To gain further insight into the relationships between the conformation/organization of apoA-I within lipidated HDL particles and its interaction with ABCA1, we performed competition assays that clearly showed that plasma pre-␤ 1 -LpA-I, reconstituted HDL particles r(LpA-I), and native HDL 3 particles are poor competitors for the binding of 125 I-apoA-I to ABCA1 compared with lipid-free apoA-I ( Fig. 2A). This experiment indicates an important role for the association of apoA-I with lipids in controlling apoA-I/ABCA1 interactions. Surprisingly, pre-␤ 1 -LpA-I, which comprises apoA-I combined with only a small amount of phospholipids (30) had a 4-fold lower efficiency to interact with ABCA1 relative to lipid-free apoA-I ( Fig. 2A). Previous studies established that the lipid composition of pre-␤ 1 -LpA-I species as well as the conformation of apoA-I within these particles differ from those of spherical HDL (25,30). Furthermore, pre-␤ 1 -LpA-I is proposed to be an initial acceptor of cell-derived cholesterol (30). This supports the idea that pre-␤ 1 -LpA-I removes cellular lipid by an aqueous diffusion process rather than an ABCA1-dependent pathway. The physiological relevance of the ABCA1-HDL interaction remains to be determined.
Although evidence have been presented demonstrating molecular interactions between ABCA1 and apoA-I (13,14,16), it remains controversial whether there is a "molecule-to-molecule contact" between apoA-I and ABCA1. Several competing models have been proposed for this interaction: 1) Burgess et al. FIG. 3. Effect of phospholipase treatment on apoA-I/ABCA1 interactions. A, stimulated cells were incubated for 60 min at 37°C in the presence or absence of 5 units/ml PC-PLC or 0.4 units/ml SM-ase. Cells were then incubated with 10 g/ml of 125 I-apoA-I for 2 h at 37°C. Specific 125 I-apoA-I binding was determined as described in Fig. 1A. Control value (100%) represent 24 ng of apoA-I/mg cell protein. ABCA1 mutant (Q597R) was used as a negative control. B, upper panel, intact stimulated normal or Q597R cells in 100-mm diameter dishes were incubated or not with PC-PLC or SM-ase as described above and then incubated with 10 g/ml of 125 I-apoA-I for 1 h at 37°C in the presence or absence of a 20-fold excess of unlabeled apoA-I (200 g/ml). Crosslinking with DSP was performed as described above. Samples containing 125 I-apoA-I cross-linked to ABCA1 (200 g of total protein) were incubated with 10 l of affinity-purified human anti-ABCA1 antibody for 20 h at 4°C, followed by addition of protein A bound to Sepharose (30 l). Immunoprecipitated samples were separated on 4 -22.5% SDS-polyacrylamide gel electrophoresis and 125 I-apoA-I/ABCA1 complexes were directly detected by autoradiography. The ABCA1 protein was detected on the same membrane by anti-ABCA1 antibody. Lower panel, intact normal cells were incubated or not with PC-PLC or SM-ase as described above and then incubated with 10 g/ml of 125 I-apoA-I for 1 h at 37°C in the presence or absence of a 20-fold excess of unlabeled apoA-I. Quantitative crosslinking of 125 I-apoA-I to ABCA1 was performed as described under "Experimental Procedures." The amount of bound iodinated apoA-I to ABCA1 in the immunoprecipitates was determined by ␥-counting. Results shown are representative of two different independent experiments. (31) suggested that phospholipids contained in the extracellular matrix of macrophages act as an initial tether point for apoA-I, bringing it into close proximity to membrane-bound ABCA1; 2) Chambenoit et al. (14) have reported that even though ABCA1 expression increases the amount of membrane bound apoA-I, its association with cellular membranes exhibits diffusional properties that are consistent with apoA-I binding to membrane lipids rather than an integral membrane protein.
In the present study, experiments have been designed to answer this controversy and evidence was in fact obtained demonstrating that both of these models cannot be applied to apoA-I/ABCA1 interactions. Here, we demonstrate that treatment of intact stimulated cells with phospholipases (PC-PLC or SMase) affected neither the specific binding of 125 I-apoA-I nor apoA-I/ABCA1 cross-linking (Fig. 3, A and B). It is likely that apoA-I/ABCA1 interactions are due to a direct protein-protein contact, which is not dependent on the presence of membrane phosphatidylcholine or sphingomyelin. However, it is not excluded that other membrane phospholipids or the phospholipase lipid products may serve to bind the amphiphathic helix of apoA-I to ABCA1. This is consistent with a previous study by Smith et al. (32) showing that although ABCA1 expression is associated with an increase in cell surface phosphatidylserine level, the cellular association of apoA-I is not competed by annexin V, a phosphatidylserine binding protein. Moreover, Mendez et al. (33) documented that cholesterol and sphingomyelin-rich membrane rafts do not provide lipid for efflux promoted by apolipoproteins through the ABCA1-mediated lipid secretory pathway.
It has been suggested that the correct conformation of ABCA1 thought to be maintained by the ATP hydrolysis action of ABCA1 or its lipid flipping activity (14,19) was necessary for apoA-I binding. Furthermore, recent studies from our laboratory and others have shown that ABCA1 phosphorylation by cAMP/PKA-dependent pathway plays an important role in the apoA-I lipidation reaction (20,34,35), suggesting that lipidation of apoA-I by ABCA1 is an active process. We confirmed and extended this observation by showing that 125 I-apoA-I dissociation from ABCA1 was almost completely inhibited at either 4 or 15°C (Fig. 4A).
The structural requirements of apoA-I lipidation by ABCA1 have not yet been determined. However, in an attempt to understand this process in fibroblasts, we examined apoA-I lipidation reaction in a tissue culture model by monitoring the kinetic parameters of apoA-I dissociation from ABCA1. We initially hypothesized that any specific apoA-I dissociation from ABCA1 would be associated with a significant increase in apoA-I lipidation state, consistent with the concept that the transfer of phospholipid and cholesterol from the active site of ABCA1 transporter to apoA-I molecule weakens the interaction of apoA-I/ABCA1 and causes dissociation of the lipidated apoA-I product. Our hypothesis is supported by the finding that: 1) specific apoA-I dissociation from ABCA1 is rapid (Fig. 4A); 2) the association of apoA-I with lipids reduces its ability to interact with ABCA1; 3) the lipid translocase activity of ABCA1 generates ␣-LpA-I-like particles; and 4) ABCA1 did not mediate hydrolysis of apoA-I in fibroblasts. However, chlorpromazine has been shown to block cAMP-mediated cholesterol efflux in macrophages (36), supporting the idea that ABCA1 may be involved in the endocytosis and resecretion of apoA-I in macrophages. More thorough investigations are required to establish definitively a possible role of ABCA1 in the endocytosis of apoA-I in fibroblasts and macrophages.
Of interest, comparison of the dissociation rate constant of apoA-I from ABCA1 and apoA-I-mediated cholesterol efflux showed for the first time that apoA-I dissociation from ABCA1 is rapid (t1 ⁄2 ϭ 1.4 h, Fig. 4A). In contrast, in our stimulated cell culture system, apoA-I-mediated cholesterol efflux reached saturation after a 16-h incubation (Fig. 4B). Previous studies have demonstrated that lipid-free apolipoproteins access both cellular FC and PL during incubations of 4 -24 h (37, 38), however, others such as Gillotte et al. (39) do show saturation in a short time frame. It should be noted that in their study the fibroblasts were not enriched with cholesterol, ABCA1 was not induced and the cells were labeled with a very high amount of strating that apoA-I/ABCA1 interactions result in the dephosphorylation of the ABCA1 PEST sequence and thereby inhibits calpain degradation leading to an increase of both ABCA1 cell surface expression and activity.
Several laboratories have demonstrated that apoA-I incubated with cells including fibroblasts (37), CHO cells (40), and macrophages (38) was able to recruit phospholipid and cholesterol from the cells to form protein-lipid complexes. Our experiment presented in Fig. 5 shows that the apoA-I-lipid complexes thus formed during apoA-I incubation with stimulated normal cells represent a spectrum of particles with distinct molecular diameters. In contrast, lipid-free apoA-I was unable  I-apoA-I was directly detected by autoradiography using XAR-2 Kodak film. Lower panels, [ 32 P]orthophosphate-labeled normal cells were stimulated, and then incubated with 10 g/ml unlabeled apoA-I for 2 h at 37°C. After washing to remove unbound apoA-I, 15 ml DMEM was added, and the plate was immediately incubated at 37°C for either 1.4, 8, or 24 h. The medium was recovered, concentrated, dialyzed, and 32 P-labeled phospholipids associated with apoA-I-containing particles were analyzed by two-dimensional-PAGGE. 32  to form larger particles during its incubation with Q597R mutant cells. The interrelationship between these particles was unclear; however, a time course analysis of apoA-I-containing particles dissociated from ABCA1 (Fig. 6) showed nascent apoA-I-phospholipid complexes that exhibited ␣-electrophoretic mobility with a particle size ranging from 9 to 20 nm (designated ␣-LpA-I-like particles). The stability of the charge, molecular diameter and phospholipid species content of these nascent particles over a 24-h dissociation period did not support the existence of a clear precursor-product relationship between the various particles and provide strong support for their common origin. It is important to note that the newly formed ␣-LpA-I-like particles had distinctly different sizes, suggesting that larger particles contained both phospholipids and cholesterol whereas the smallest particles contained only phospholipids and apoA-I (41). Because of the absence of cholesterol acyltransferase activity in the extracellular medium to convert FC to cholesteryl ester, it is most likely that ␣-LpA-I-like particles are discoidal. Indeed, it was documented that a lipoprotein with a high concentration of phosphatidylinositol could have a high negative charge and consequently an ␣-electrophoretic mobility (42), consistent with our finding that ␣-LpA-I-like particles have a high content in phosphatidylinositol (14 Ϯ 0.2%).
During the preparation of this article a study by Liu et al. (43) reported that incubation of apoA-I with macrophages leads to the formation of more than one type of lipidated apoA-Icontaining particles with a molecular diameter of 6 -16 nm. In addition, this study supports the idea that there is a simultaneous release of PL and FC to apoA-I molecules through a membrane microsolubilization process. It is interesting to contrast our results with those reported in that study, which demonstrated that apoA-I-containing particles have a smaller size and an important amount of apoA-I remaining in its lipidfree form. It is possible that the cell species used in the two studies affect ABCA1-dependent lipidation of apoA-I: we used FIG. 7. A proposed model of free apoA-I lipidation by ABCA1 in peripheral cells. Cholesteryl ester-rich HDL 2 gain triacylglycerols from VLDL under the action of CETP. HDL 2 undergo lipolysis by the hepatic lipase (H-TGL) and a possible involvement of the SR-BI receptor generating lipid-free apoA-I, which can be rapidly lipidated by ABCA1 and form ␣-LpA-I-like particles. Continuous action of LCAT contributes to the maturation of ␣-LpA-I-like particles to form cholesteryl ester-enriched HDL. A model of apoA-I lipidation by ABCA1 can be proposed assuming that 1) initial binding of apoA-I to ABCA1 is irreversible or slowly reversible (k 1 Ͼ Ͼ k Ϫ1 ); 2) lipidated apoA-I (␣-LpA-I-like particles) dissociate rapidly (k 2 ) from ABCA1 without any detectable reassociation; and 3) this system contains no other apolipoprotein that could compete for the binding of lipid-free apoA-I to ABCA1. PL, phospholipids; FC, free cholesterol; TG, triglycerides; CE, cholesteryl esters. human fibroblasts and Liu et al. (43) used J774 macrophages. In addition our cells were loaded directly with FC (20 g/ml) and were stimulated with 22(R)-hydroxycholesterol and 9-cisretinoic acid and their cells were loaded with 25 g/ml acetyl-LDL and induced with cAMP. More importantly, we observed that both the charge and diameter of the newly formed LpA-Ilike particles are markedly different from those of lipid-free apoA-I (Figs. 5 and 6). We therefore suggest that our experimental design based on the analysis of LpA-I particles released during the dissociation period might be critical for the study of LpA-I product generated by a specific lipid translocase activity of ABCA1.
Evidence has been presented here demonstrating that only lipid-free apoA-I is able to interact efficiently with ABCA1 in vitro (Fig. 2). However, it seemingly paradoxical that lipid free-apoA-I molecules, which are not normally present in significant quantities in plasma (44), play a similar role in vivo. We postulate that lipid-free apoA-I generated during apoA-Icontaining particles remodeling cycle (45) are rapidly lipidated by ABCA1 and form ␣-LpA-I-like particles or, alternatively, may be incorporated in preexisting plasma HDL. Our current results support the first hypothesis. We demonstrate that 50% of specifically bound 125 I-apoA-I was rapidly dissociated from ABCA1 at physiological temperatures (t1 ⁄2 ϳ1.4 h) (Fig. 4A). At the same time, the majority of apoA-I-containing particles generated during the dissociation period was shown to be associated with phospholipids having an ␣-electrophoretic mobility (Fig. 6E). This concept is supported by recent study by Kee et al. (46) demonstrating that the electrophoretic mobility of 125 I-apoA-I (lipid-free) changed from pre-␤ to ␣electrophoretic mobility only 2 min after injection into wild-type rabbits. In addition, the same study documented that hepatic lipase has the capacity to decrease the size of ␣-migrating HDL, in agreement with the earlier work of Barrans et al. (45).
Different kinetic models can be proposed to explain the mechanism of apoA-I/ABCA1 interaction. Here, our cell culture system represents a relatively simple model. However, in peripheral tissues and interstitial fluid, many other lipid-free apolipoproteins (e.g. apoE, apoJ, apoA-IV) might compete with apoA-I for ABCA1 binding. Our results show that apoA-I binding to ABCA1 was found to occur in a time-and concentrationdependent manner (Fig. 1A). Thus, apoA-I/ABCA1 association can be described as a receptor-ligand interaction or a proteinprotein interaction in solution under apparent equilibrium condition (Fig. 7). On the other hand, the lipid translocase activity of ABCA1 transforms lipid-free apoA-I to ␣-LpA-I-like particle; here ABCA1 seems to act as an enzyme catalyzing the lipidation of the substrate. Although admittedly speculative, we believe that our data support this hybrid model better than either ligand/receptor or substrate/enzyme model. The important finding that the interaction of lipid-free apoA-I with ABCA1 generates only ␣-LpA-I-like particles might help to explain why lipid-free apoA-I is found in trace amounts in human plasma. Indeed, following the release of lipid-free apoA-I by the action of hepatic lipase on HDL 2 and a possible involvement of SR-BI in this process (47), lipid-free apoA-I molecules might be very rapidly lipidated by ABCA1 and transformed into ␣-LpA-I-like particles (Fig. 7).
The results presented in this study provide a biochemical basis for a cellular apoA-I lipidation pathway that involves ABCA1 protein in peripheral cells. This process plays in vivo a key functional role in the biogenesis of nascent HDL particles.