A sensitive assay for ABCA1-mediated cholesterol efflux using BODIPY-cholesterol.

Studies have shown a negative association between cellular cholesterol efflux and coronary artery disease (CAD). Standard protocol for quantitating cholesterol efflux involves labeling cells with [(3)H]cholesterol and measuring release of the labeled sterol. Using [(3)H]cholesterol is not ideal for the development of a high-throughput assay to screen large numbers of serum as would be required in studying the link between efflux and CAD. We compared efflux using a fluorescent sterol (boron dipyrromethene difluoride linked to sterol carbon-24, BODIPY-cholesterol) with that of [(3)H]cholesterol in J774 macrophages. Fractional efflux of BODIPY-cholesterol was significantly higher than that of [(3)H]cholesterol when apo A-I, HDL(3), or 2% apoB-depleted human serum were used as acceptors. BODIPY-cholesterol efflux correlated significantly with [(3)H]cholesterol efflux (p < 0.0001) when apoB-depleted sera were used. The BODIPY-cholesterol efflux correlated significantly with preβ-1 (r(2) = 0.6) but not with total HDL-cholesterol. Reproducibility of the BODIPY-cholesterol efflux assay was excellent between weeks (r(2) = 0.98, inter-assay CV = 3.31%). These studies demonstrate that BODIPY-cholesterol provides an efficient measurement of efflux compared with [(3)H]cholesterol and is a sensitive probe for ABCA1-mediated efflux. The increased sensitivity of BODIPY-cholesterol assay coupled with the simplicity of measuring fluorescence results in a sensitive, high-throughput assay that can screen large numbers of sera, and thus establish the relationship between cholesterol efflux and atherosclerosis.

The effl ux of cholesterol from macrophages in the arterial wall is thought to be an early step in the process termed reverse cholesterol transport (RCT; [1][2]. Effl ux can be mediated by a number of pathways involving 1 ) aqueous diffusion, 2 ) scavenger receptor BI (SR-B1), 3 ) ABCG1, and 4 ) ABCA1 (3)(4)(5). Although all of these pathways can contribute to the removal of free cholesterol (FC) from cells, it is ABCA1 that has been shown to play a major role in the maintenance of normal cholesterol levels in tissues, as demonstrated by the accumulation of large amounts of cholesterol in macrophages in individuals with genetic mutations in ABCA1 (Tangier disease) or in mouse models in which this protein is genetically eliminated (6)(7)(8).
A link between the in vitro effl ux of cholesterol from macrophages and atherosclerosis has recently been established by studies demonstrating a negative correlation between cholesterol effl ux from J774 mouse cells and coronary artery disease (CAD) as measured by either carotid intima-media thickness (cIMT) or angiographic measurements ( 9 ). Although not as yet defi nitely established, preliminary data indicate that the primary effl ux mechanism linking cholesterol effl ux to arterial cholesterol deposition is ABCA1 ( 10 ). In contrast to the other effl ux pathways, ABCA1 uses as a ligand lipid-free/lipidpoor apolipoproteins, primarily apo A-I, whereas other pathways employ mature, fully lipidated HDL as cholesterol acceptors ( 3,4,11 ). Regardless of the pathway, the standard protocol for quantitating cell cholesterol effl ux employs cells labeled with radioactive cholesterol and determination of the fractional release of the labeled sterol to a variety of acceptors ranging from whole serum to isolated apoproteins ( 12 ). Although this approach to measuring serum. The sera used in the study had undergone one cycle of freezing and thawing. The effects of freezing on the effl ux assay were tested by measuring both BODIPY-cholesterol and [ 3 H]cholesterol effl ux on fresh and frozen sera (n = 4). The BODIPY-and [ 3 H]cholesterol effl ux values of fresh sera did not differ significantly from those of frozen sera (paired t -test). To isolate the serum total HDL fraction (apoB-depleted serum), the pools of sera or individual serum samples were depleted of apoB-containing lipoproteins by precipitating them with PEG (MW = 8000) solution and diluted to either 1% or 2% (equivalent to 0.7% or 1.4% whole serum, respectively) as described previously ( 15,16 ), and as indicated in the fi gure legends. The presence of PEG does not alter the effl ux capacity of the extracellular acceptors. The concentrations of pre ␤ -1 protein in selected sera were determined by two-dimensional gel chromatography as previously described ( 16 ).

Preparation of labeling media
Labeling medium containing BODIPY-cholesterol was prepared by complexing the sterols (unlabeled cholesterol and BODIPY-cholesterol) with CD at a molar ratio of 1:40 (cholesterol/CD). The BODIPY-cholesterol represented 20% of the total cholesterol in the labeling medium. Unlabeled cholesterol and BODIPY-cholesterol were dried under nitrogen in the dark to form a thin fi lm in a round-bottom 100-ml glass bottle. The cholesterol mixture was solubilized by adding 20 mM CD in MEM-HEPES buffer. The suspension was sonicated in a water bath (37°C) for 30 min, placed in a shaking water bath for 3 h (37°C), and kept at 4°C until use. Before use, the suspension was sonicated for 30 min, fi ltered using a 0.45-m syringe fi lter, and diluted with an equal volume of MEM-HEPES buffer containing 2 µg/ml ACAT inhibitor. There was no loss of fl uorescent sterol upon fi ltration. The fi nal concentrations of BODIPY-cholesterol, unlabeled cholesterol, and CD in the labeling medium were 0.025 mM, 0.1 mM, and 10 mM, respectively.
The [ 3 H]cholesterol-labeling medium was prepared by drying [ 3 H]cholesterol (4 µCi/ml) on the walls of a conical glass tube and then dissolving the residue in 50 µl of ethanol. The solution was then incubated with 5% FBS overnight and diluted with the desired volume of RPMI medium containing 2 µg/ml ACAT inhibitor; the fi nal concentration of ethanol was <0.5%. The detailed protocol for preparing the [ 3 H]cholesterol-labeling medium has been described previously ( 17 ).

H]cholesterol
J774 cells were plated in RPMI supplemented with 10% FBS and gentamicin in 48-well plates at a density of 75,000 cells/well for 24 h. Cells were labeled with BODIPY-cholesterol by incubating the monolayers with 0.25 ml of labeling medium containing CD/BODIPY-cholesterol/unlabeled cholesterol for 1 h, followed by washing with MEM-HEPES. J774 cells were then equilibrated with RPMI containing 0.2% BSA and the ACAT inhibitor, with or without cAMP (0.3 mmol/L), for 18 h. After this equilibration period, the cells were washed with MEM-HEPES buffer and incubated with MEM-HEPES media containing the cholesterol acceptors as indicated in the fi gures. Incubation times were 4 h or as indicated in the fi gures. At the end of the incubation time, the effl ux media were removed, fi ltered through a 0.45-m fi lter, and the fl uorescence intensity was recorded using a Molecular Devices M2 plate reader (excitation 482 nm, emission 515 nm). After equilibration, the time-zero (t 0 ) cell monolayers (cells not incubated with extracellular acceptors) were solubilized with 1% cholic acid and mixed well by shaking on a plate shaker for 4 h at room temperature; then the fl uorescence intensity was recorded. The fl uorescence intensity values obtained at this time were used as the t 0 measurements. Fractional effl ux of BODIPY-cholesterol cholesterol effl ux has provided large amounts of data on both the effi ciency of various extracellular acceptors and on different effl ux pathways, a protocol using radiolabeled cholesterol does not lend itself to the development of a high-throughput assay that can effi ciently screen large numbers of specimens such as serum or serum fractions.
In the present study we have substituted a fl uorescent sterol, BODIPY-cholesterol, for the generally used radiolabeled cholesterol. BODIPY-cholesterol is a fl uorescent analog of free cholesterol in which carbon-24 of the sterol side chain is linked directly to the dipyrromethene boron difl uoride ("BODIPY") moiety ( 13 ). In a series of experiments using J774 macrophages together with a variety of cholesterol acceptors, we have compared effl ux of BODIPY-cholesterol and [ 3 H]cholesterol. These studies have demonstrated that the use of the fl uorescent-labeled sterol provides an effi cient measurement of effl ux when compared with radiolabeled cholesterol. We found that BODIPY-cholesterol is released from cells primarily via the ABCA1 pathway. Thus, the greater effi ciency for effl ux of the BODIPY-cholesterol compared with labeled cholesterol, together with the greater simplicity of the fl uorescent assay, allows the development of effl ux protocols suitable for rapid, high-throughput screening for effl ux capacity of large numbers of sera or serum fractions.

Preparation of lipoproteins and human sera
Human HDL 3 was isolated by sequential ultracentrifugation from plasma of healthy, normolipemic individuals, as previously described (HDL 3 , d = 1.12-1.21 g/ml) ( 14 ). Apo A-I was purifi ed from delipidated HDL using ethanol/diethyl ether followed by anion-exchange chromatography on a Q-Sepharose column ( 14 ). The blood samples used in these studies were collected, with informed consent, from normolipemic individuals (both males and females) ranging in age from 23 to 70 years. The blood samples were obtained in vacutainer tubes and allowed to clot at room temperature for 45 min. The sera were spun at 2400 rpm for 15 min after clotting and the samples were aliquoted and stored at Ϫ 80°C for up to 2 years before use. The sera had an average level of HDL-cholesterol of 58 ± 15 mg/dl (35-97 mg/dl), LDL-cholesterol of 121 ± 27 mg/dl (66-180 mg/dl), and triglycerides of 115 ± 60 mg/dl (42-230 mg/dl). The human serum samples were used individually or combined to obtain a pool of which 20 replicate wells of cAMP-treated J774 cells were exposed to 20 µg/ml of lipid-free apo A-I or to 2% apoBdepleted pooled human serum for 4 h. BODIPY-cholesterol effl ux based on media values yielded a CV of 8.2% and 9.8% for 2% apoB-depleted serum and apoA-I, respectively, whereas the effl ux based on the residual fl uorescence in the cells had a CV of 21.9% (2% apoB-depleted serum) and 23.7% (20 µg/ml of lipid-free apo A-I). Thus, accumulation of BODIPY-cholesterol in the media gave more reproducible values than assaying for the loss of the compound from the cells. These are similar to the values for [ 3 H]cholesterol effl ux to 2% apoB-depleted serum (CV = 7.2%) and apo A-I (CV = 13.2%). In the present experiments, percent effl ux for both [ 3 H]cholesterol and BODIPY-cholesterol was determined using ((media / time zero values) × 100).
BODIPY-cholesterol effl ux from control and cAMP-treated cells. J774 cells were used throughout this study because the effl ux pathways in these cells, particularly ABCA1mediated effl ux, are easily upregulated by exposure of the cells to cAMP ( 3,18 ). Our initial experiment compared the effl ux of BODIPY-cholesterol and [ 3 H]cholesterol from J774 macrophage cells treated with or without cAMP to 10 µg/ml of lipid-free apo A-I, 25 µg/ml human HDL 3, and 2% apoB-depleted pooled human serum. As shown in Fig. 1A and B , treatment of J774 cells with cAMP stimulated release of both fl uorescent cholesterol and radiolabeled cholesterol with all extracellular acceptors. With each of the acceptors, the effl ux patterns were similar; however, the fractional release of BODIPY-cholesterol was substantially greater than that of [ 3 H]cholesterol. The difference in fractional effl ux between the cAMP-treated and control cells is generally accepted to represent the was calculated based on the fl uorescence intensity of the media divided by the t 0 monolayer fl uorescence values measured after cholic acid solubilization. BODIPY-cholesterol effl ux to media containing no acceptors ("background effl ux") was 6.3 ± 0.6% and was subtracted from BODIPY-cholesterol effl uxes of all the acceptors. ACAT inhibitor was present at all times during the experiment.
The [ 3 H]cholesterol labeling of the cells was accomplished as previously described ( 17 ). Briefl y, J774 cells were incubated for 24 h in 0.25 ml of RPMI media supplemented with 5% FBS and 4 µCi/ml of [ To determine if hemolysis would affect the BODIPY-cholesterol effl ux assay, J774 cells treated with cAMP were incubated with 1% plasma or hemolysed plasma (10-100% lysed red blood cells) for 4 h. BODIPY-cholesterol effl ux to plasma with moderate hemolysis (<10% hemolysis) did not differ from that of plasma with no hemolysis. However, BODIPY-cholesterol effl ux to plasma with 100% lysed red blood cells had a signifi cantly lower effl ux compared with that of plasma with no hemolysis. We also examined if the presence of bilirubin in the serum affected the BODIPY-cholesterol effl ux assay. BODIPY-cholesterol effl ux to 1% serum spiked with bilirubin added at different concentrations (0.25-2.5 mg/dl) was not signifi cantly different from that of control serum (without bilirubin).

Statistical analysis
All statistical analyses were performed using GraphPad Prism (San Diego, CA) software. Data are presented as mean ± SD. Statistical signifi cance was determined by unpaired t -tests unless otherwise indicated. Coeffi cient of variation (CV) was calculated by dividing the SD with the mean of that variable. Deming regression was used to assess the relationship between the [ 3 H]cholesterol and BODIPY-cholesterol effl ux. Signifi cance was assessed at p р 0.05. BODIPY-cholesterol effl ux via the ABCA1 pathway. The data illustrated in Figs. 1-3 suggest that the primary effl ux pathway by which the fl uorescent sterol is released from cells involves ABCA1. Thus the effl ux of BODIPY-cholesterol is considerably faster than that of radiolabeled cholesterol, and the effl ux is enhanced by exposure of J774 cells to cAMP. To further establish the role of ABCA1 in BODIPY-cholesterol effl ux, we compared fractional effl ux values for both BODIPY-cholesterol and [ 3 H]cholesterol with and without probucol pretreatment, as probucol has been shown to inhibit ABCA1-mediated effl ux of cholesterol ( 21 ). In this study, cAMP-treated J774 cells were exposed to media containing 50 µg/ml of apo A-I or 2% apoB-depleted human serum. The effl ux of BODIPY-cholesterol to apoA-I was almost totally inhibited by probucol and effl ux to apoB-depleted serum was greatly reduced ( Table 1 ). Probucol treatment also reduced the effl ux of [ 3 H]cholesterol; however, the magnitude of the inhibition was not as great as that observed with the fl uorescent sterol ( Table 1 ). The enhanced stimulation of effl ux by cAMP, contribution of ABCA1 to effl ux ( 3 ). In the J774 cell system, treatment with cAMP or cholesterol enrichment has also been shown to upregulate the expression of ABCG1 in addition to ABCA1 ( 3 ). However, ABCG1 does not release cholesterol to lipid-free apolipoproteins whereas apo A-I serves as a primary ligand for the ABCA1 effl ux pathway. The observation that cAMP treatment enhanced effl ux of BODIPY-cholesterol to apo A-I is consistent with a substantial amount of effl ux of the fl uorescent sterol occurring via the ABCA1 pathway. In addition, cAMP-treated J774 cells were used in a recent study that demonstrated a negative association between cholesterol effl ux and plaque burden ( 9 ). This recent study used apoB-depleted human serum as the acceptor; this total HDL preparation has been used in several studies because the presence of apoB-containing lipoproteins can confound the measurement of cellular cholesterol fl ux ( 19 ).

BODIPY-cholesterol effl ux to lipid-free apo A-I and phospholipid vesicles.
It has been repeatedly demonstrated that lipid-free or lipid-poor apolipoproteins, particularly apo A-I, are particularly effi cient acceptors of cholesterol released from cells via the ABCA1 pathway ( 11,20 ). Using J774 cells that are upregulated for ABCA1 by treatment with cAMP, we have quantitated and compared BODIPYcholesterol and [ 3 H]cholesterol effl ux to increasing concentrations of apo A-I ( Fig. 2 ). The V max (16.1 ± 0.3%/4 h) for effl ux of BODIPY-cholesterol is greater than that for radiolabeled cholesterol ( V max = 10.2 ± 0.5%/4 h).   and fl uorescent cholesterol to 23 apoB-depleted human sera. The reference range for BODIPY-cholesterol effl ux assay using 23 healthy subjects is 38.9-48.1% (95% CI). There are a number of different HDL subfractions that contribute to the effl ux of cell cholesterol. Spherical, mature HDL has been shown to promote cholesterol effl ux via SR-BI, ABCG1, and aqueous transfer, whereas lipidfree/poor apoproteins function as the acceptor of cholesterol released via the ABCA1 pathway, resulting in the formation of nascent HDL particles ( 22,23 ). The data presented in Figs. 1 and 2 demonstrate that lipid-free apo A-I promotes the release of BODIPY-cholesterol via the ABCA1 pathway. Thus, it was of interest to determine the relationships between both serum HDL cholesterol and pre ␤ -HDL concentrations and the effl ux of BODIPY-cholesterol and [ 3 H]cholesterol. As illustrated in Fig. 5A , effl ux of BO-DIPY-cholesterol correlated with the concentration of pre ␤ -1 (r 2 = 0.6) and had no signifi cant correlation to total HDL-C (n = 23 individual sera, combined results of four independent experiments in triplicate; Fig. 5B ). In contrast, the effl ux of [ 3 H]cholesterol had a signifi cant correlation to HDL-C levels (r 2 = 0.6, Fig. 5D ) and an association that exhibited a trend between [ 3 H]cholesterol effl ux and pre ␤ -1 that did not reach statistical signifi cance (r 2 = 0.2, n = 23 individual sera, combined results of four independent experiments in triplicate; Fig. 5C ). These associations are consistent with BODIPY-cholesterol release occurring, to a large extent, by the ABCA1 pathway and responding more specifi cally to serum pre ␤ -HDL levels than does the release of [ 3 H]cholesterol.
In the experiments described above, apoB-depleted sera were used at a concentration of 2% to act as the acceptors for both fl uorescent and radiolabeled cholesterol. This concentration is similar (2.8%) to that which has been used for a number of effl ux studies in the past ( 9,10,19 ). To establish how changing the concentration of apoB-depleted serum infl uences BODIPY-cholesterol effl ux, effl ux values were obtained using apoB-depleted sera at three different concentrations (0.5%, 1%, and 2%). Fluorescent sterol effl ux obtained using these three different concentrations had the same rank order. However, BODIPY-cholesterol values the release of the fl uorescent sterol to lipid-free apoA-I, and the inhibition by probucol are consistent with the primary pathway for effl ux being via ABCA1. 3 [H]cholesterol effl ux to human sera. The goal of developing a BODIPY-cholesterol effl ux system is to provide a high-throughput assay for comparing the effl ux effi ciency of individual serum donors, particularly because our recent studies have demonstrated a signifi cant negative association between effl ux and atherosclerosis ( 9 ). Many effl ux studies have used apoB-depleted sera as cholesterol acceptors ( 16 ). The removal of apoBcontaining lipoproteins has been employed to obtain a reliable measure of HDL effl ux capacity in the absence of any potential effl ux to apoB-containing lipoproteins. The precipitation of apoB-lipoproteins with PEG yields a total HDL preparation that avoids the potential changes in HDL that can occur during isolation of HDL particles by centrifugation. To determine whether the precipitation protocol was necessary in studies using BODIPY-cholesterol, we conducted a preliminary study in which effl ux from whole serum was compared with effl ux obtained with a comparable concentration of apoB-depleted serum from the same individual. We found that BODIPY-cholesterol effl ux obtained using whole serum was reduced by 25 ± 6% (n = 11) upon removal of apoB lipoproteins, whereas the contribution of apoB-lipoproteins to [ 3 H]cholesterol effl ux was greater (42 ± 8%, n = 11). The contribution of apoB-containing lipoproteins to BODIPY-cholesterol effl ux to whole serum was less than that to radiolabeled cholesterol, which is consistent with effl ux of BODIPY-cholesterol being a more specifi c probe for ABCA1-mediated effl ux.

BODIPY-cholesterol and
The intent of this investigation was to develop a cholesterol effl ux assay in which the use of radiolabeled cholesterol could be eliminated by substituting a fl uorescent-labeled cholesterol analog for the radiolabeled cholesterol. Thus, it was necessary to demonstrate a correlation between fractional effl ux of BODIPY-cholesterol and [ 3 H]cholesterol. Figure 4 demonstrates a signifi cant relationship ( P < 0.0001, Deming regression) between the measurement of effl ux from J774 cells based on the release of radiolabel 3 H]cholesterol for 24 h. The cells were then equilibrated with or without cAMP for 16 h followed by pretreatment with or without probucol (10 µM) for 2 h. Effl ux was measured after incubating the cells with either apo A-I (50 µg/ml), or 2% apoB-depleted pooled human serum for 4 h. Probucol was not present during effl ux phase. % of inhibition by probucol (ABCA1 contribution) was calculated by subtracting the effl ux values obtained from cells treated with cAMP and probucol from effl ux values of cells treated with cAMP but not probucol. between control cells and cells pretreated with cAMP because exposure to cAMP upregulates ABCA1 and ABCG1. Another approach to determining the contribution of each pathway to effl ux is to use inhibitors such as probucol, which has been shown to inhibit ABCA1 effl ux ( 21 ), and BLT-1, which inhibits SR-BI effl ux ( 3 ). All of these approaches utilize cells that contain radiolabeled cholesterol, and the protocols require a number of manipulations, including 1 ) removal of the acceptor medium, 2 ) fi ltration or centrifugation of the media to remove fl oating cells, 3 ) determination of the total amount of labeled cholesterol in the cells at the time of exposure (t 0 ), a process that requires lipid extraction of the cells, and 4 ) liquid-scintillation obtained using 1% sera had the best spread among individual serum samples. In addition, 1% apoB-depleted serum will require smaller amounts of serum to be collected in future screening studies and is less likely to be toxic to the cells ( 24 ). In subsequent experiments, we used 1% apoB-depleted sera becausee this concentration yielded the most consistent values when multiple serum samples were tested. The reproducibility of the fl uorescent assay is illustrated in Fig. 6 . In this study, six individual serum samples were tested for effl ux capacity at three different effl ux times and on subsequent weeks. The agreement between fractional effl ux values determined 1 week apart was excellent and all of the fractional effl ux values obtained at 2 h, 4 h, and 6 h fi t a straight line (r 2 = 0.98). The inter-assay CV of the two experiments was 3.3% (1 week apart). In a larger study using 18 apoB-depleted sera at 1% concentration, the agreement between two assays conducted a week apart had an r 2 = 0.7 (effl ux time, 4 h).

DISCUSSION
Experiments to study cell cholesterol effl ux generally use cells prelabeled with radioactive cholesterol as donors, and the release of isotope provides values from which fractional effl ux (% effl ux) can be determined. This approach has been productive in supplying information on both the effl ux pathways available to the cell being studied and the serum apolipoproteins and lipoproteins that serve as the most effi cient acceptors. A variety of donor cells have been used, with recent emphasis on macrophages, including the established cell lines such as J774 mouse macrophages. In J774 macrophages, the extent to which an effl ux pathway contributes to effl ux to either serum or HDL can be measured by determining the difference in fractional effl ux  of cells for 2 h with probucol resulted in a reduction of both fl uorescent and radiolabeled cholesterol effl ux, effectively eliminating BODIPY-cholesterol effl ux to apo A-I and greatly reducing effl ux to apoB-depleted human serum. Effl ux to the apoB-depleted serum that remained following probucol treatment can be attributed to effl ux mediated by the non-ABCA1 pathways that are known to be present in J774 cells ( 3,34 ) and represents a lower fraction of total effl ux in BODIPY-cholesterol labeled cells.
The fact that the release of BODIPY-cholesterol from cells is mediated primarily by ABCA1 suggests that this fl uorescently-labeled sterol partitions into the pool of membrane lipid that is incorporated into the nascent HDL particles created by the activity of the transporter. The cellular cholesterol released via ABCA1 originates from the plasma membrane ( 35,36 ) as well as from late endosomes ( 37 ). The preference of BODIPY-cholesterol to partition into the liquid-ordered phase of membranes ( 29 ) is consistent with BODIPY-cholesterol originating from plasma membrane raft domains being incorporated effi ciently into nascent HDL particles ( 36 ). Because [ 3 H]cholesterol is released from cells by ABCA1 much more slowly than BODIPY-cholesterol, it seems that incorporation of the BODIPY moiety into the cholesterol molecule enhances partitioning into the substrate pool for nascent HDL formation by ABCA1. The enhanced sensitivity of BODIPYcholesterol to ABCA1-mediated effl ux accounts for the relative effl uxes of BODIPY-cholesterol and [ 3 H]cholesterol to the plasma HDL fraction (apoB-depleted serum) observed with ABCA1-expressing cells ( Fig. 3 ). However, as has been observed with other cell types ( 30 ), BODIPYcholesterol also undergoes effl ux faster than [ 3 H]cholesterol from J774 cells in which ABCA1 is not upregulated with cAMP. In this situation, cholesterol effl ux occurs primarily by diffusion-mediated pathways ( 11 ), and thus, it follows that BODIPY-cholesterol desorbs more readily from the plasma membrane than does [ 3 H]cholesterol. This effect is presumably a consequence of the presence of the BODIPY group in the cholesterol molecule, which weakens interactions between cholesterol and the fatty acyl chains of neighboring phospholipid molecules in the membrane. Such an effect has been observed in mixed phospholipid/BODIPY-cholesterol monolayer experiments ( 13 ) and in calculations of acyl chain order in model membranes containing BODIPY-cholesterol ( 30 ).
The demonstration that the effl ux of radiolabeled cholesterol from J774 cells has an inverse association with both cIMT and angiographic CAD ( 9 ) suggests that the quantitation of cholesterol effl ux, as a measure of HDL functionality, may be useful in predicting cardiovascular risk. The protocol for determining serum effl ux capacity has until now relied on the release of radiolabeled cholesterol from cell monolayers ( 12 ). For effl ux to be used as a probe for the presence of atherosclerosis, or for the effectiveness of pharmacological agents in enhancing RCT, it will be necessary to test large numbers of sera. Because the use of radiolabeled cholesterol for this kind of testing has obvious limitations, the availability of a fl uorescent effl ux assay generally will be preferable. It should be noted that the counting of both the cell extracts and the effl ux medium. The present study was designed to determine whether a cholesterol effl ux assay could be developed that avoided the need to use donor cells that contained radiolabeled cholesterol and eliminated some of the steps needed to establish the effl ux capacity of serum, HDL, or apolipoproteins. To accomplish this, we labeled the cells with a fl uorescent analog of cholesterol, BODIPY-cholesterol, and used the release of this fl uorescent sterol as a surrogate measure of cell cholesterol effl ux.
BODIPY-cholesterol is a fl uorescent analog of free cholesterol in which carbon-24 of the sterol side chain is linked directly to the dipyrromethene boron difl uoride ("BODIPY") moiety ( 13 ). BODIPY-cholesterol has many attractive photophysical properties, such as a high fl uorescence quantum yield, high photostability, and insensitivity to pH and polarity ( 13,(25)(26)(27). It partitions into the liquid-disordered and liquid-ordered phases of model bilayer membranes and cell membranes ( 28 ), with a preference for the liquid-ordered phase in giant unilamellar vesicles ( 29 ). BODIPY-cholesterol has been used for live-cell monitoring of cholesterol traffi cking ( 30 ) and for analysis of the kinetics of transfer of self-quenched BODIPY-cholesterol from donor to acceptor vesicles in the presence of StAR protein ( 31 ).
Recently, another protocol designed to measure cell cholesterol effl ux using a fl uorescent mimic of cholesterol has been published. This assay uses the Pennsylvania Green fl uorophore attached by a linker containing a glutamic acid residue to a derivative of N-alkyl-3 ␤ -cholesterylamine ( 32 ). The fractional release of this compound is less than that obtained with [ 3 H]cholesterol, whereas the fractional effl ux value obtained with BODIPY-cholesterol is greater than that of [ 3 H]cholesterol. Although this fl uorescent assay has the potential of being used for high-throughput screening for effl ux, it was designed to be used as a screen for bioactive compounds that modulate cell cholesterol effl ux. In contrast, the primary application of the BODIPYcholesterol effl ux assay will be to screen large numbers of sera or serum HDL fractions for effl ux capacity, because effl ux capacity has recently been shown to have a negative association with atherosclerosis ( 9 ).
The fact that BODIPY-cholesterol effl ux is saturatable ( Fig. 3 ) suggests that the release of the fl uorescent sterol may occur preferentially via the ABCA1 pathway. Such a specifi city is consistent with the enhanced ABCA1-mediated effl ux of the BODIPY-cholesterol when compared with [ 3 H]cholesterol, as determined by comparing effl ux from cAMP ± cells ( Figs. 1A, B ). Studies have demonstrated that probucol is an effective inhibitor of ABCA1mediated lipid effl ux ( 21 ). The inhibition of effl ux of BODIPY-cholesterol by probucol further documents the importance of ABCA1 in mediating BODIPY-cholesterol effl ux ( Table 1 ). Because of the proven importance of ABCA1 in promoting cholesterol effl ux ( 20,33 ) and the recent results linking effl ux to plaque burden in humans ( 9 ), we used probucol inhibition of effl ux to examine if ABCA1 is the major cell protein promoting effl ux of BODIPY-cholesterol. As presented in Table 1 , pretreatment use of the BODIPY-cholesterol effl ux assay, as described in this study, has the potential of streamlining studies linking cell cholesterol effl ux to clinical endpoints such as IMT and angiography. However, an association of BODIPYcholesterol effl ux and atherosclerosis has not yet been established. Thus, the "gold standard" for determining the association of cholesterol effl ux to clinical endpoints remains the measurement of the effl ux of [ 3 H]cholesterol from cAMP treated J774 macrophages. The association between effl ux and atherosclerosis appears to be linked to the level of ABCA1 in the cholesterol donor cells and the level of pre ␤ -HDL in the serum. The present study indicates that BODIPY-cholesterol effl ux is a more sensitive probe for ABCA1-mediated effl ux than radiolabeled cholesterol. If so, then the increased sensitivity, coupled with the ability to quantitate effl ux by fl uorescent measurements, will result in a sensitive, high-throughput assay that will establish the relationships between ABCA1-mediated effl ux and atherosclerosis.