Sterol Carrier Protein-2 Alters High Density Lipoprotein-mediated Cholesterol Efflux*

Although sterol carrier protein-2 (SCP-2) participates in the uptake and intracellular trafficking of cholesterol, its effect on “reverse cholesterol transport” has not been explored. As shown herein, SCP-2 expression inhibited high density lipoprotein (HDL)-mediated efflux of [3H]cholesterol and fluorescent 22-(N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino)-23,24-bisnor-5-cholen-3b-ol (NBD-cholesterol) up to 61 and 157%, respectively. Confocal microscopy of living cells allowed kinetic analysis of two intracellular pools of HDL-mediated NBD-cholesterol efflux: the highly fluorescent lipid droplet pool and the less fluorescent pool outside the lipid droplets, designated the cytoplasmic compartment. Both the whole cell and the cytoplasmic compartment exhibited two similar kinetic pools, the half-times of which were consistent with protein (t b 1 2 near 1 min) and vesicular (t d 1 2 = 10–20 min) mediated sterol transfer. Although SCP-2 expression did not alter cytoplasmic sterol pool sizes, the rapid t b 1 2 decreased 36%, while the slower t d 1 2 increased 113%. Lipid droplets also exhibited two kinetic pools of NBD-cholesterol efflux but with half-times over 200% shorter than those of the cytoplasmic compartment. The lipid droplet slower effluxing pool size and t d 1 2 were increased 48% and 115%, respectively, in SCP-2-expressing cells. Concomitantly, the level of the lipid droplet-specific adipose differentiation-related protein decreased 70%. Overall, HDL-mediated sterol efflux from L-cell fibroblasts reflected that of the cytoplasmic rather than lipid droplet compartment. SCP-2 differentially modulated sterol efflux from the two cytoplasmic pools. However, net efflux was determined primarily by inhibition of the slowly effluxing pool rather than by acceleration of the rapid protein-mediated pool. Finally, SCP-2 expression also inhibited sterol efflux from lipid droplets, an effect related to decreased adipose differentiation-related protein, a lipid droplet surface protein that binds cholesterol with high affinity.

Although the HDL-mediated 1 steps of cholesterol transfer from the cell surface membrane and subsequent fate of choles-terol in the vasculature have been extensively studied, much less is known about intracellular components of cholesterol efflux (reviewed in Refs. [1][2][3][4][5]. Plasma membrane cholesterol is distributed into multiple pools or domains (reviewed in Refs. 6 and 7). It is now recognized that there may be a connection between such domains and HDL receptor-mediated reverse cholesterol transport (reviewed in Refs. [7][8][9]. The transbilayer distribution of cholesterol in plasma membranes is asymmetric, with the cholesterol enriched 400% in the cytofacial leaflet versus exofacial leaflet (reviewed in Refs. 10 -14). Transbilayer movement of cholesterol across the plasma membrane appears fast (t1 ⁄2 ϭ 1-6 min; reviewed in Ref. 10). Plasma membrane cholesterol is also distributed into lateral cholesterol-rich and -poor membrane domains (reviewed in Refs. 6 and 10). Most of the cholesterol in the plasma membrane is localized in lateral domains that are, for the most part, relatively inert in terms of transfer kinetics (i.e. t1 ⁄2 ϭ hours to days), and movement between such domains is also slow. However, a small pool of plasma membrane cholesterol appears highly dynamic (reviewed in Ref. 6) and is associated with cholesterol-rich, HDL receptor containing microdomains called caveolae (reviewed in Refs. 8, 9, 15, and 16). Molecular details of cholesterol entry/ exit, cholesterol organization, and mechanism(s) of cholesterol transbilayer movement in caveolae remain to be determined. Likewise, the relationships between caveolae, "rafts," and other cholesterol-rich plasma membrane microdomains is not yet clear (reviewed in Ref. 9).
The intracellular steps preceding cellular cholesterol efflux include transfer of cholesterol from the Golgi, endoplasmic reticulum, and lipid droplets to the plasma membrane (reviewed in Refs. 1, 9, and 17). The time frame of bidirectional vesicular transfer of cholesterol between plasma membranes and Golgi has a t1 ⁄2 of 10 -20 min (reviewed in Refs. 8, 18, and 19). Alternately, molecular cholesterol transfer, mediated by cholesterol-binding proteins in the cytoplasm, occurs much faster (t1 ⁄2 near 1-2 min) from the lysosome (exogenous cholesterol) to the plasma membrane (20, 21) and from the endoplasmic reticulum (newly synthesized cholesterol) to the plasma membrane (reviewed in Refs. 7, 8, and 22). Which one (or both) of the processes participate in cholesterol trafficking from lipid droplets, a primary site of cholesterol storage, is presently unknown.
The purpose of this investigation was to visualize three aspects of cholesterol efflux in intact cells. (i) To determine the role of SCP-2 in HDL-mediated cholesterol efflux. Previously it was shown that SCP-2 stimulates biliary cholesterol secretion (23)(24)(25) but inhibits lipoprotein secretion (22). (ii) To address the lack of information regarding the lipid droplet as a source for cholesterol efflux via the HDL pathway. Lipid droplets contain a surface coat of cholesterol and phospholipid surrounding a cholesterol ester/triglyceride core (26,27). Further, the lipid droplet surface monolayer contains unique proteins including perilipin and adipose differentiation-related protein (ADRP), neither of whose function is well understood (28). Because ADRP binds cholesterol, this suggests a potential role of this protein in cholesterol trafficking from the lipid droplet (29). (iii) To examine the effect of SCP-2 expression on HDLmediated cholesterol efflux from the lipid droplet. This possibility was suggested by the fact that SCP-2 stimulates cholesterol transfer from lipid droplets to mitochondria in adrenal cells (reviewed in Ref. 27). The results presented herein provide fresh insights into the efflux process and for the first time establish new information demonstrating that SCP-2 expression: (i) inhibits HDL-mediated cholesterol efflux and (ii) inhibits cholesterol efflux from a little understood, subcellular compartment, i.e. lipid droplets.

Materials-[ 3 H]Cholesterol was from PerkinElmer
Life Sciences. Bovine serum albumin (essentially fatty acid free) and fetal bovine serum (FBS) were from Sigma. HDL was from Calbiochem (San Diego, CA). Lab-Tek Coverglass slides were from Fisher. NBD-cholesterol and Nile red were from Molecular Probes (Eugene, OR). Sandoz 58-035 was a gift from T. Y. Chang (Dartmouth Medical School, Hanover, NH). Rabbit polyclonal antisera to ADRP were prepared (30) using ADRP generously provided by Dr. G. Serrero (University of Maryland, Baltimore, MD). Monoclonal anti-caveolin-1 was from Transduction Laboratories (Lexington, KY). Rabbit polyclonal anti-scavenger receptor B1 antisera were from Novus Biologicals (Littleton, CO). All reagents and solvents used were of the highest grade available and were cell culture tested.
L-cell Culture-Murine L-cells (L arpt Ϫ tk Ϫ ) were grown to confluency in Higuchi medium (31) supplemented with 10% fetal bovine serum (Hyclone, Logan, UT) as described (30). Development of mocktransfected cells (designated as control) and cells transfected with cDNA corresponding to the 15-kDa pro-SCP-2 protein was as described earlier (30,32). In high expression cells, SCP-2 comprised 0.036 ϩ 0.002% of the total cytosolic proteins, whereas levels in the control and mock transfected cells were below the level of detectability. Cellular cholesterol and cholesterol ester mass were increased 20 and 100%, respectively, in the SCP-2 expression clones. For radiolabeled experiments, control and SCP-2 expressors were plated to 24-well plates at approximately 5 ϫ 10 5 cells/well. For fluorescence imaging experiments, cells were seeded at a density of 50,000 cells/chamber onto Lab-Tek Chamber Coverglass slides (Nunc, Naperville, IL). To ensure a monolayer, samples analyzed by imaging were examined within 20 h of seeding. Cells for radiolabeled experiments were treated as described below.

Measurement of Cellular [ 3 H]Cholesterol
Efflux-Cholesterol efflux was done as described earlier (33). Briefly, mock-transfected and SCP-2 expressors were plated in medium containing 1 Ci/ml of [ 3 H]cholesterol and 2.5% fetal bovine serum (1 ml/well) overnight. The medium was then replaced with medium containing 0.3% bovine serum albumin to allow equilibration between different cholesterol pools. After 24 h, cells were washed twice with phosphate-buffered saline. HDL at indicated concentrations was then added, and radiolabeled aliquots (50 l) were removed at indicated time points and counted. At the end of the time course, the monolayers were washed three times with phosphatebuffered saline, cells were dissolved in 0.1 N NaOH (500 l) overnight, and protein was determined (34 mM NaCl, and 10 mM HEPES) and placed in serum-free medium. A medial section passing through 5-10 cells was selected, and the section was scanned for initial fluorescence intensity of NBD-cholesterol. HDL was added to start the efflux experiment. The fluorescence in the total area of each cell was measured over time and was used to gauge the extent of NBD-cholesterol efflux from the cell.
NBD-cholesterol Co-localization with Lipid Droplets-Three types of co-localization studies were performed to determine the identity of high intensity regions of NBD-cholesterol visible within the cell. First, NBDcholesterol was co-localized with Nile red, a lipid stain. Second, Nile red was co-localized with ADRP, a protein closely associated with lipid droplets. Third, NBD-cholesterol was co-localized with ADRP. Cells overexpressing SCP-2 were incubated with both NBD-cholesterol (0.35 M) and Nile red (0.4 M) for 30 min in a 37°C CO 2 incubator. Although Nile red is known to stain other nonlipidic structures, care was taken to titer both Nile red and NBD-cholesterol to low levels to give a low cytoplasmic background with emphasis on lipid droplet staining in the co-localization experiments. However, it should be noted that Nile red and NBD-cholesterol are also located in sites outside lipid droplets where they also co-localize to a large degree. After incubation, cells were washed with Puck's buffer and then simultaneously imaged for NBDcholesterol (excitation, 488 nm; emission, 540/30; green channel) and Nile red (excitation, 568 nm; emission, HQ598/40; red channel). The confocal images from the green and red channels were merged and appeared yellow where superimposition occurred (red and green are additive and yield yellow to orange in RGB color space). Pixel fluorograms were constructed, and correlation coefficients generated from the fluorograms were derived from the following equations as described earlier (35).
where R i, coloc is the sum of intensities of all red pixels which also have a green component; R i is the sum of intensities of all red pixels in the image; G i, coloc is the sum of intensities of all green pixels, which also have a red component; and G i is sum of intensities of all the green pixels in the image. To co-localize ADRP with Nile red or with NBD-cholesterol, SCP-2 expressors were plated on 8-well Lab-Tek chamber slides (Nunc, Naperville, IL) at subconfluency. The cells were washed with serum-free medium and incubated with 0.005% of either Nile red or NBD-cholesterol in serum-free medium for 30 min, at 37°C. Cells were washed with Hanks' solution, fixed with 70:30 (v/v) acetone:ethanol for 10 min at 4°C, permeabilized with 0.05% saponin in Hanks' for 5 min at room tempera-ture, and then blocked with 10% FBS in Hanks' for 1 h at room temperature. Anti-ADRP (rabbit polyclonal anti-mouse ADRP, affinity purified, concentrated to 0.5 mg/ml; dilution, 1:20) was incubated in 5% FBS in Hanks' for 1 h at room temperature, followed by extensive washing with 2% FBS in Hanks' buffer. Next, the cells were incubated with either goat anti-rabbit IgG conjugated to fluorescein (for the ADRP/Nile red co-localization) or goat anti-rabbit IgG conjugated to Texas red (for the ADRP/ NBD-cholesterol co-localization) for 1 h at room temperature. After a final wash with Hanks', cells were mounted in SlowFade (Molecular Probes) according to the manufacturer's procedure and were visualized with the MRC1024 laser scanning system as described above.
NBD-cholesterol Efflux from Subcellular Compartments: Cytoplasm and Lipid Droplets-NBD-cholesterol efflux from the whole cell was compared with that arising from two compartments designated as lipid droplets (based on the above co-localization data) or cytoplasmic (defined as the nonlipid droplet, lower intensity, diffuse pattern of the remaining NBD-cholesterol fluorescence in the cell). Cells were imaged and analyzed by Metamorph software to graphically determine the separate contributions of the efflux process. Data points were fitted to either a biexponential (y ϭ Ae Ϫbt ϩ Ce Ϫdt ) or multiparameter (y ϭ (Ae ϪbT0 ϩ Ce ϪdT0 )(e Ϫh(tϪT0) ) exponential decay equation where y was the fraction of initial cellular NBD-cholesterol remaining in the cells at time t; A and C were the size of the cholesterol pools available for efflux; b, d, and h were the apparent rate constants; and T 0 was the time at which depletion of the pools increased to completion. The multiparameter equation was employed to adjust for an inflection point observed in the time course data obtained from both control and SCP-2 overexpressors. It appeared that at some time T 0 , the rate of efflux was altered, which effected a decrease in the half-life. To compare the data, a model was developed that allowed for the initial decay curve (Ae Ϫbt ϩ Ce Ϫdt ) to be modified at some T 0 to become (Ae ϪbT0 ϩ Ce ϪdT0 ) e Ϫh(tϪT0) . Using a nonlinear least squares routine in Sigma Plot 4.0, all parameters were obtained simultaneously from the fitted data. Unique fittings were obtained individually for each time course involving individual cells. This produced excellent results for the cytoplasmic compartment contribution from the cell. Statistical methods involving the paired t test were used to compare the control and SCP-2 expressors. However, the lipid droplet contribution was quite "noisy," and an inflection point could not be determined for these data, which were subsequently fit to the equation Ae Ϫ bt ϩ Ce Ϫdt for all times. This also contributed some complication in determining the inflection point of the whole cell because it is the combination of lipid droplet plus cytoplasm. To overcome this difficulty, the time courses of n individual cells for the specified concentration and cell type were averaged to produce a single time course, which was then fitted to the multiparameter equation.
Isolation of Mouse Lipid Droplets from L-cell Fibroblasts-Lipid droplets from control cells and SCP-2 expressors were isolated as described by Chanderbhan et al. (36). Briefly, cells scraped from 10 confluent trays (245 ϫ 245 ϫ 25 mm each; Nunc, Naperville, IL) were homogenized in 50 mM NaH 2 P0 4 buffer, pH. 7.4, containing 154 mM NaCl and 5 mM MgCl 2 and centrifuged at 800 ϫ g for 10 min. The supernatant was then centrifuged at 5,000 ϫ g for 20 min to sediment mitochondria. This was followed by centrifuging at 35,000 rpm in a SW 40.1 rotor for 2 h at 4°C. The lipid droplet fraction, forming a distinct, white band on the surface of the preparation, was removed, and protein was quantified (34).
Quantitative Western Blotting for ADRP, Caveolin, and SRB1-Lipid droplets and cell homogenates were analyzed by Western blot analysis to determine ADRP content in control and SCP-2 expressors. Cell homogenates were analyzed for presence of caveolin and SRB1. For all quantitative Western blots, protein samples (17-20 g) were run on Tricine gels (12%) and transferred to nitrocellulose membranes. The blots were blocked in 3% gelatin in 10 mM Tris-HCl, pH 8, 100 mM NaCl, 0.055 Tween-20 before incubation with the polyclonal rabbit antibodies against either ADRP, caveolin, or SRB1. Alkaline-phosphatase conjugates of goat anti-rabbit IgG and Sigma Fast 5-bromo-4chloro-3-indolyl phosphate/nitro blue tetrazolim tablets (Sigma) were used to visualize and quantitate bands of interest (37).
Statistics-All values were expressed as the means Ϯ S.E. with n and p indicated under "Results." Statistical analyses were performed using Student's t test (GraphPad Prism, San Diego, CA). Values with p Ͻ 0.05 were considered statistically significant. protein/ml (Fig. 2B, open bars). In SCP-2-expressing cells, the initial rate of HDL-mediated [ 3 H]cholesterol efflux similarly increased with medium HDL, except that the slope of the increase was 40% higher (Fig. 2B, filled bars). Thus, SCP-2 expression inhibited the initial rate of HDL-mediated [ 3 H]cholesterol efflux at low, but not high, HDL concentration. However, even at 250 g HDL/ml the initial rate of HDL-mediated [ 3 H]cholesterol efflux in SCP-2-expressing cells was not increased over that in mock transfected controls.

Dependence of the Extent of HDL-mediated Efflux of [ 3 H]Cho
Further analysis of the HDL-mediated [ 3 H]cholesterol efflux curves in Fig. 1 revealed that the efflux curve fit single exponential kinetics that allowed resolution of two parameters: the time at which half the cholesterol pool was depleted (t1 ⁄2 ) and the pool size of HDL-mediated [ 3  H]cholesterol efflux was inhibited by SCP-2 expression, radiolabeled studies did not provide direct visualization of this process. Therefore, cells were preloaded with NBD-cholesterol and washed, and NBDcholesterol was imaged by laser scanning confocal microscopy as described under "Experimental Procedures." The total fluorescence intensity/cell was then determined individually for a large number of cells. The time-dependent disappearance of NBD-cholesterol fluorescence of each cell was then determined in the presence and absence of HDL (see "Experimental Procedures"). As with [ 3 H]cholesterol efflux, NBD-cholesterol fluorescence intensity/cell (corrected for a small amount of photobleaching over the time period studied) was essentially zero in the absence of HDL for both mock transfected control and SCP-2-expressing cells (not shown). The addition of HDL dramatically stimulated NBD-cholesterol efflux from both the SCP-2 expressor and mock transfected control cells (Fig. 3). However, the extent of this effect was highly dependent on the expression of SCP-2 and the HDL concentration in the medium. HDL-mediated NBD-cholesterol efflux from mock transfected control cells was greater than 90% by 40 min (Fig. 3, filled triangles). In contrast, SCP-2 expression decreased the extent of HDL-mediated NBD-cholesterol efflux over the same time period studied (Fig. 3, open triangles). Examination of a large number of cells showed that the extent of HDL-mediated NBDcholesterol efflux was 40% less than that from mock transfected control cells (p Ͻ 0.006, n ϭ 11-14).
The extent of NBD-cholesterol efflux was dependent on the HDL concentration. The extent of efflux increased proportional to medium HDL concentration in near linear fashion for both mock transfected control (slope, 0.74; Fig. 4A, open bars) and for SCP-2-expressing cells (slope, 1.04; Fig. 4A, solid bars). SCP-2 expression inhibited HDL-dependent NBD-cholesterol 157% (p Ͻ 0.006, n ϭ 6 -9) at low HDL concentration (Fig. 4A) but not at high HDL concentration. However, even when the SCP-2-mediated inhibition was abolished at high HDL concentration, the extent HDL-mediated NBD-cholesterol efflux in SCP-2-expressing cells was not increased above that exhibited by mock transfected control cells (Fig. 4A).  Fig. 2A), SCP-2 expression inhibited HDL-mediated sterol efflux at low HDL levels to the same degree (65-75%), and this inhibition was abolished at similar high ratios of HDL/number of cells.

Comparison of the Effect of SCP-2 Expression on Extent of HDL-mediated NBD-cholesterol and [ 3 H]Cholesterol Ef
Effect of SCP-2 Expression on Kinetics of HDL-mediated NBD-cholesterol Efflux in Living Cells-SCP-2 expression decreased the initial rate of HDL-mediated NBD-cholesterol efflux as much as 220% at low HDL concentration in the medium (Fig. 4B). Increasing HDL concentration abolished the inhibitory effect of SCP-2 expression on HDL-mediated NBD-cholesterol efflux. Visual examination of the curves in Fig. 3 indicated that the kinetics of HDL-mediated NBD-cholesterol efflux were not simple exponentials. Attempts at fitting these curves to simple exponential kinetics for either mock transfected controls or SCP-2-expressing cells were unsuccessful (see "Experimental Procedures"). In contrast, the efflux curves in Fig. 3 fit very well (R 2 ϭ 0.999) to a multiparameter exponential decay equation, y ϭ (Ae ϪbT0 ϩ Ce ϪdT0 ) (e Ϫh(tϪT0) ) (see "Experimental Procedures"). This equation describes two pools of cholesterol (a large, slow pool and a smaller, faster pool) showing increased efflux after some time point T 0 . In mock transfected control cells, the respective half-times of these cholesterol pools (t b 1 ⁄2 and t d 1 ⁄2 ) were 1.2 and 13.1 min, respectively (Table I). The longer half-time was increased 100% (p Ͻ 0.0025) in SCP-2expressing cells. In mock transfected control cells, the pool sizes (A and C) corresponding to these half-times were 0.23 and 0.77, respectively (Table I). Although the slowly effluxing NBDcholesterol pool was 230% larger than the rapidly effluxing pool, SCP-2 expression did not affect the respective pool sizes. Overexpression of SCP-2 resulted in a 59% decrease in the half-life associated with h. T 0 was not affected by SCP-2 expression. In summary, these data suggested that SCP-2 expression inhibited NBD-cholesterol efflux from the cells primarily by inhibiting the half-time rather than altering the pool size of NBD-cholesterol.
Intracellular Localization of the NBD-cholesterol: Confocal Laser Scanning Microscopy of L-cells Expressing SCP-2-Although the NBD-cholesterol fluorescence in cells preloaded with NBD-cholesterol appeared localized throughout the cell, distinct highly fluorescent regions resembling lipid droplets in size and shape were also prominent (Fig. 6). To establish whether the intensely fluorescent regions were indeed lipid droplets, three co-localization experiments were performed.
First, cells preloaded with NBD-cholesterol were stained with Nile red, a selective lipid droplet stain, and emission of the two fluorophores was simultaneously determined through separate photomultipliers: NBD-cholesterol (green) and Nile red (red). Upon superposition of such images (Fig. 7A), co-localization appeared as yellow/orange areas (red plus green ϭ yellow/ orange), whereas lack of co-localization appeared as separate green and red areas. A graphical representation of the superposition is shown in the form of a pixel fluorogram (Fig. 7B). High co-localization in pixel fluorograms is indicated both by the localization of many pixels along the diagonal, the yellow/

TABLE I Kinetic analysis of HDL-mediated NBD-cholesterol efflux
from intact cells Parameters were derived from the following multi-parameter exponential decay equation. When t Ͻ T 0 , y ϭ (Ae Ϫbt ϩ Ce Ϫdt ); when t Ն T 0 , y ϭ (Ae ϪbT0 ϩ Ce ϪdT0 ) ϫ eϪ h(tϪT0) where A and C represent the fraction of available NBD-cholesterol in each rate differentiated pool; b, d, and h are the apparent rate constants; t is time in minutes; and T 0 (derived from the fitted curve) is the time at which depletion of the pools increased to completion (18.2 Ϯ 0.6 and 16.8 Ϯ 0.9 min for the control and SCP-2 expressor, respectively).  Fig. 7B) is ratio of all the red (Nile red) intensities showing a green (NBD-cholesterol) component divided by the sum of all the red intensities. It should be noted, however, that although a significant population of data points fell along the diagonal line, NBD-cholesterol also appeared throughout the cell (green in the superimposed image in Fig. 7A). Morphometric analysis on intensity measurements using Metamorph software (see "Experimental Procedures") revealed the percentage of NBD-cholesterol outside the lipid droplets (i.e. designated as the cytoplasmic compartment of the cell) ranged from 80 to 96% for both control and SCP-2 expressors. Although the correlation coefficients given in Fig. 7B determined the degree of co-localization of Nile red in NBDcholesterol (and vice versa), they did not measure the amount of NBD-cholesterol in the lipid droplets versus that in the rest of the cell.
Second, the lipid droplet specific stain, Nile red, was colocalized with ADRP, a protein closely associated with lipid droplets. Cells were simultaneously labeled with Nile red (red) and anti-ADRP antibodies (green). Confocal fluorescence images for ADRP (Fig. 8B) highly resembled those obtained for Nile red (Fig. 8A), the lipid droplet specific stain. Magnification of a representative lipid droplet showed uniform staining with Nile red throughout (Fig. 8A, inset). In contrast, magnification of the same lipid droplet stained with anti-ADRP showed much more intense staining at the lipid droplet surface (Fig. 8B,  inset). Superposition of the two images (Fig. 8, A and B) yielded a yellow-orange color where co-localization occurred. Although both stains co-localized to lipid droplets, the magnified image of the lipid droplet revealed that ADRP (green) was present on the surface of the lipid droplet, whereas the Nile red stain was seen both at the surface (orange/yellow) and throughout (red) the lipid droplet (Fig. 8C).
Third, NBD-cholesterol was co-localized with ADRP by simultaneous labeling with anti-ADRP antibodies (red) and NBD-cholesterol (green). Simultaneous acquisition of confocal images for ADRP (Fig. 8D) and NBD-cholesterol (Fig. 8E) showed that ADRP and NBD-cholesterol were both present in intense staining areas, i.e. lipid droplets shown in superposed images (Fig. 8F). Magnified images of representative lipid droplets (Fig. 8F, inset) showed that ADRP again appeared on the lipid droplet surface (distinct red region on left side of lipid droplet), whereas NBD-cholesterol was present at the surface co-localized with a portion of ADRP (yellow) as well as throughout the interior of the lipid droplet (green).
Effect of SCP-2 Expression on ADRP Levels in Transfected L-cell Fibroblasts-Because ADRP binds cholesterol and NBD-cholesterol (29) and is significantly co-localized with ADRP, the inhibition of HDL-mediated cholesterol efflux by SCP-2 may occur in part by altered level of ADRP. Because the mocktransfected and untransfected cell lines had similar levels of ADRP (not shown), Western blots of cell homogenates and lipid droplets, isolated as described under "Experimental Procedures," were run for control and SCP-2-expressing cells samples (Fig. 9). Although SCP-2 was not detected in lipid droplets, ADRP was enriched 140% in lipid droplets as compared with cell homogenates for both control (Fig. 9, lane 2 versus lane 1) and SCP-2-expressing cells (Fig. 9, lane 4 versus lane 3). SCP-2 expression reduced the level of ADRP by 60% (Ϯ0.36, n ϭ 11) in the cell homogenate and by 70% (Ϯ0.08, n ϭ 11) in the isolated lipid droplets. These data suggest an important role for ADRP as well as SCP-2 in regulating cellular cholesterol levels.
Effect of SCP-2 Expression on Scavenger Receptor B1 and Caveolin Levels in Transfected L-cells-It is possible that SCP-2 inhibition of HDL-mediated cholesterol efflux was due to alterations in plasma membrane components of the cholesterol efflux pathway, caveolin and the HDL receptor (SRB1). However, Western blotting of cell homogenates; Fig. 9, B and C) from control and SCP-2 expression clones showed no significant change in the levels of caveolin (panel B) or SRB1 (panel C).
Relative Contributions of the Intracellular Compartments to HDL-mediated Cholesterol Efflux Kinetics-The relative contributions of the low intensity, diffuse pattern typical of distribution throughout the cytoplasm and the high intensity pattern localized to lipid droplets to HDL-mediated cholesterol efflux kinetics were examined over time: 2 min (Fig. 6A), 28 min (Fig. 6B), and 58 min (Fig. 6C). After 58 min, the lipid droplet NBD-cholesterol intensity (indicated by arrows) was still visible, whereas that in cytoplasmic areas of the cell was almost depleted. Cholesterol was retained in the lipid droplets; this suggested either that the rate-limiting step of NBD-cholesterol efflux from the cell was efflux from lipid droplets or that the concentration of NBD-cholesterol in the lipid droplets was higher than in the cytoplasm. To resolve these possibilities, cells were imaged and graphically partitioned (see "Experimental Procedures") to separate the lipid droplet contribution from the cytoplasmic component of the cell. Surprisingly, NBDcholesterol efflux from lipid droplets (Fig. 10B, filled triangles) was initially faster than from cytoplasm ( Fig. 10A) but quickly reached an equilibrium level. Multicomponent kinetic analysis of NBD-cholesterol fluorescence decrease confirmed this observation. The data points derived from the lipid droplet compartment (Fig. 10B, filled triangles) best fit (R 2 ϭ 0.99) to a biexponential decay equation, whereas the cytoplasmic compartment (Fig. 10A, filled triangles) best fit to a multi-parameter decay equation (R 2 ϭ 0.99), suggesting that multiple pools of NBD-cholesterol were available for efflux from each compartment. In control cells, the half-times of NBD-cholesterol efflux from the lipid droplets were t b 1 ⁄2 ϭ 0.66 and t d 1 ⁄2 ϭ 8.4 min, significantly smaller (p Ͻ 0.005) than the respective halftime from the cytoplasm t b 1 ⁄2 ϭ 1.9 and t d 1 ⁄2 ϭ 14.7 min (Table  II). Because Pool A, the rapid efflux component, was nearly 200% larger than that of cytoplasm (0.56 versus 0.18), the larger pool size further contributed to the more rapid efflux of NBD-cholesterol from the lipid droplet than from the cytoplasm. Thus, the efflux of NBD-cholesterol from the lipid droplet was not the rate-limiting step in HDL-mediated cholesterol efflux from the cell, and efflux kinetics from the two components in the whole cells (Table I) basically reflected those from the two cytoplasmic components rather than the lipid droplets (Table II). Lower apparent loss of NBD-cholesterol from the lipid droplets versus cytoplasm was apparently due to a higher concentration of NBD-cholesterol in the lipid droplets than in FIG. 6. NBD-cholesterol efflux by HDL in SCP-2 overexpressing cells. Transfected L-cells expressing the SCP-2 protein were labeled with NBD-cholesterol (0.35 M) and incubated with HDL (10 g protein/ml). Cells at 2 min (A), 28 min (B), and 58 min (C) during the efflux process were examined with the Bio-Rad MRC-1024 confocal system as described under "Experimental Procedures." Arrows indicate high retention lipid droplets during the time course. Objective, 63ϫ. the cytoplasm rather than a slower rate of NBD-cholesterol efflux from the lipid droplet.
Effect of SCP-2 Expression on HDL-mediated Cholesterol Efflux Kinetics from Cytoplasm and Lipid Droplets-In SCP-2expressing cells, efflux of NBD-cholesterol from lipid droplets (Fig. 10B, open triangles) was initially faster than from cytoplasm but did not reach an equilibrium level in either cytoplasm or lipid droplets as quickly as observed with control cells. Multicomponent kinetic analysis of NBD-cholesterol fluorescence decrease showed that efflux from the lipid droplet compartment of SCP-2-expressing cells (Fig. 10B, open triangles) again best fit (R 2 ϭ 0.99) to a biexponential decay equation, whereas from the cytoplasmic compartment (Fig. 10A, open triangles) best fit to a multi-parameter decay equation (R 2 ϭ 0.99). Examination of the relative half-times and pool sizes of NBD-cholesterol efflux from cytoplasm and lipid droplets of SCP-2 expression showed: (i) The half-times of NBD-cholesterol efflux from the lipid droplets of SCP-2-expressing cells were shorter (faster) than those from cytoplasm (Table II) Co-localization patterns of NBD-cholesterol and Nile red were shown using pseudo-coloring derived from confocal image acquisition from red-and green-specific PMT channels. A 24-bit RGB image was created from the red plus green plus blue (null) channels. Red and green are additive in RGB color space yielding yellow-orange. L-cells expressing the SCP-2 protein stained with NBD-cholesterol and Nile red were combined to yield yellow-orange (A) where co-localization occurred. Superimposition of the probes was graphically demonstrated in a pixel fluorogram (B) where co-localization of NBD-cholesterol (green) and Nile red (red) resulted in yellow to orange points falling along the diagonal line in the fluorogram. The correlation coefficients corresponding to red and green were proportional to the degree of fluorescence probe co-localizing in each component of the image relative to the total fluorescence and indicated the extent of overlap between the probes. The cells were examined using the Bio-Rad MRC-1024 confocal system. Objective, 63ϫ.

FIG. 8. Double-label immunofluorescence with ADRP and Nile red or NBD-cholesterol in transfected L-cells expressing SCP-2.
Cells were simultaneously labeled for Nile red (A) and ADRP with fluorescein isothiocyanate-conjugated goat anti-rabbit IgG (B) or ADRP with Texas Red-conjugated goat anti-rabbit IgG (D) and NBD-cholesterol (E). Co-localization (C and F) were shown using pseudo-coloring resulting in yellow to orange where superimposition occurred. Insets were lipid droplets located within the larger frame. expressing cells the half-times and pool sizes of NBDcholesterol efflux from the whole cells (Table I) reflected those of the cytoplasmic compartment (Table II) rather than those observed for the lipid droplets (Table II).
SCP-2 expression differentially affected the two half-times of NBD-cholesterol efflux from the cytoplasm as compared with those in the mock transfected control cells. The half-time t b 1 ⁄2 for rapid NBD-cholesterol efflux from the cytoplasmic compartment was significantly shorter (faster) in SCP-2-expressing cells than that observed for control cells, 1.4 Ϯ 0.1 versus 1.9 Ϯ 0.1 min, p Ͻ 0.005 (Table II). In contrast, the slower half-time t d 1 ⁄2 was 113% longer in SCP-2-expressing cells as compared with control cells (31.4 Ϯ 2.0 versus 14.7 Ϯ 0.8 min, p Ͻ 0.005) (Table II). Although SCP-2 expression altered the half-times of NBD-cholesterol efflux from the cytoplasmic compartment, there was little effect of SCP-2 expression on the respective pool sizes. The pool size C associated with the slower half-time t d 1 ⁄2 was 460% greater than that of pool size A associated with the rapid half-time t b 1 ⁄2 (Table II). However, neither pool size differed significantly from those in mock transfected control cells. In summary, SCP-2 expression enhanced efflux from the rapidly transferring NBD-cholesterol pool in the cytoplasm. Concomitantly, SCP-2 expression inhibited the efflux from the slowly transferring, but much larger, NBD-cholesterol pool in the cytoplasm.
c Parameters were derived from the biexponential decay equation. y ϭ Ae Ϫbt ϩ Ce Ϫdt .
First, overexpression of SCP-2 in transfected L-cell fibroblasts inhibited the extent and rate of HDL-mediated [ 3 H]cholesterol and NBD-cholesterol efflux. SCP-2 expression inhibited the maximal extent of HDL-mediated [ 3 H]cholesterol and NBD-cholesterol efflux to the same degree at equivalent, nonsaturating ratios of HDL/cell in the culture medium. SCP-2 expression also increased half-times (longer) and decreased cholesterol pool size available for HDL-cholesterol efflux.
Third, confocal imaging of living cells allowed graphic delineation of HDL-mediated cholesterol efflux into at least two components: lipid droplets and outside lipid droplets. Because of its diffuse distribution, the latter was defined as cytoplasmic. Efflux curves of NBD-cholesterol were best fit to bi-exponential and multi-parameter exponential decay equations, indicating the presence of a small, rapid pool with half-time near 1 min and a larger, slow effluxing pool with half-time near 13 min. These data were consistent with data reported elsewhere wherein two pools of intracellular cholesterol transfer with similar half-times were observed and attributed to proteinmediated and vesicular cholesterol transfer, respectively (18,20,21,50,50,53). Furthermore, the efflux kinetics from the cytoplasmic compartment largely reflected that of the whole cell. This suggested that transfer of cholesterol through the cytoplasm, rather than desorption from lipid droplets, was the rate-limiting step in HDL-mediated cholesterol efflux.
Fourth, SCP-2 expression differentially affected cholesterol efflux from the cytoplasmic compartment: (i) SCP-2 stimulated the smaller, fast pool of protein-mediated cholesterol through the plasma membrane to the HDL acceptor. In the absence of an extracellular cholesterol acceptor, SCP-2 expression in transfected L-cells was shown to stimulate the rapid transfer of plasma membrane derived cholesterol to the endoplasmic reticulum for esterification by acyl-CoA:cholesterol O-acyltransferase (49). This was consistent with data from transfected hepatoma cells overexpressing SCP-2 wherein rapid cholesterol cycling of plasma membrane and intracellular cholesterol was observed (22). Also pertinent were observations that the fast pool of protein-mediated cholesterol transfer was abolished by treatment of human fibroblasts with antisense DNA to SCP-2 (50). We speculate that this fast pool represents, at least in part, SCP-2-bound NBD-cholesterol because SCP-2 has been shown to bind NBD-cholesterol with high affinity (7,54,55). Parallel studies on the intracellular diffusion of NBD-stearic acid, a ligand that SCP-2 also binds with high affinity, also showed that SCP-2 expression increased the rapid diffusion coefficient of this ligand in the cytoplasm (30,56,57). Lipidbinding proteins enhance the intracellular diffusion of poorly water soluble molecules (i.e. lipids) by increasing the desorption membranes and increasing their aqueous solubility (reviewed in Refs. 58 and 59). (ii) SCP-2 expression increased the half-time of the slow vesicular pool of cholesterol transfer. In contrast, treatment of human fibroblasts with antisense DNA to SCP-2 stimulated the slower, vesicular component of choles-terol transfer through the cytoplasm (50). Likewise, HDL cholesterol secretion (mediated by vesicular transport) was decreased in transfected rat hepatoma cells overexpressing SCP-2 (22).
Fifth, cholesterol efflux from the lipid droplet was characterized by both fast and slow components that were distinct from those observed in the cytoplasm. The data presented herein and earlier (29) demonstrated that lipid droplet cholesterol is present in at least three forms: esterified and localized primarily in the core of the lipid droplet; unesterified and localized in the surface lipid of the lipid droplet; and unesterified and bound to ADRP, a cholesterol-binding protein present in the surface of the lipid droplet. The effect of SCP-2 expression on lipid droplet cholesterol pool size, taken together with the observation that ADRP binds cholesterol, suggests that the slowly effluxing component was the unesterified lipid. Pool sizes in lipid droplets were dramatically affected by SCP-2 expression where the larger cholesterol pool (C) was increased, whereas that of the smaller rapidly transferable cholesterol pool (A) was decreased. The 60% decrease in the rapidly effluxing pool size in the lipid droplet of SCP-2-expressing cells correlated with a 70% decrease in ADRP level in lipid droplets from SCP-2-expressing cells as compared with the control cells. It should be noted that the two pools were not due to ADRPbound versus unbound NBD-cholesterol that was not esterified. Although NBD-cholesterol is esterified at least as well as cholesterol in L-cells, CaCo2 cells, and hamster intestinal enterocytes (29,40), after 24 h less than 8% of the NBD-cholesterol was observed to be esterified in L-cells (29). Because the time frame of the efflux experiments using NBD-cholesterol was less than an hour, little to no esterification of the probe was expected. Thus, the fact that ADRP binds cholesterol with nM K d (29) taken together with data showing NBD-cholesterol colocalizes with ADRP suggests a role for ADRP in determining the lipid droplet cholesterol pool size available for efflux. Because this observation might be correlative, work is in progress to further delineate the role of ADRP in efflux and other intracellular processes in cells overexpressing ADRP.
In summary, NBD-cholesterol proved a useful probe for monitoring the relative effects of SCP-2 expression on HDL-mediated cholesterol efflux. However, this does not imply that the behavior of NBD-cholesterol is identical to that of cholesterol. One important distinction is that NBD-cholesterol efflux achieved equilibrium in 20 -40 min, significantly faster than [ 3 H]cholesterol (20 -24 h). Conversely, HDL-mediated uptake of NBD-cholesterol is also faster than that of [ 3 H]cholesterol in L-cells, hamster intestinal enterocytes, and CaCo2 cells (29,60). Although the mechanism by which HDL facilitates the removal of cholesterol from the cell is not yet fully resolved, the more rapid dynamics of NBD-cholesterol efflux/uptake may simply reflect the higher aqueous solubility of NBD-cholesterol versus cholesterol (55). For example, in the sterol efflux pathway HDL interacts with the scavenger receptor (SRB1) at the plasma membrane surface. The existing evidence favors a mechanism whereby plasma membrane cholesterol then desorbs and diffuses through the aqueous medium to be taken up by the acceptor HDL (reviewed in Ref. 61). Thus, the higher aqueous solubility of NBD-cholesterol favors an increased rate of HDL-mediated efflux or, in the reverse direction, uptake. Substantial evidence suggests that NBD-cholesterol undergoes metabolism and utilizes the same intracellular trafficking pathways as cholesterol. NBD-cholesterol distributes in L-cells similarly as dehydroergosterol, a naturally occurring fluorescent sterol (29). NBD-cholesterol is esterified similarly as cholesterol and dehydroergosterol in L-cells, intestinal enterocytes, and CaCo2 cells (29,60). Finally, NBD-cholesterol trafficks by similar uptake, intracellular, and secretory pathways as was shown herein and previously by others in studies with L-cell fibroblasts and CaCo2 cells (29,60). However, there are limitations of the use of NBD-cholesterol that include the inability to incorporate to as high a degree as cholesterol in some membranes (such as the plasma membrane) (62). Notwithstanding, although NBD-cholesterol does not accurately reflect all aspects of cholesterol, the above studies taken together with those presented herein indicate that NBD-cholesterol is an acceptable probe for examining relative effects of factors such as HDL concentration and SCP-2 expression on cholesterol uptake, efflux, and metabolism.
The physiological importance of the findings reported herein is that SCP-2 may participate in maintaining high intracellular pools of cholesterol. This is especially significant to tissues such as liver, intestine, and steroidogenic tissues wherein cholesterol utilization is high and SCP-2 expression greatest (reviewed in Refs. 7 and 27). The observation that SCP-2 expression stimulated cholesterol uptake in transfected cells is also consistent with SCP-2 affecting the balance of influx/efflux in favor of influx/retention of cholesterol (reviewed in Refs. 7 and 32). The ability of SCP-2 cells to retain cholesterol, along with longer half-times and changing pool size contributed to the overall effect of inhibition of HDL-mediated cholesterol efflux. Interestingly, the data suggest that the SCP-2 protein may have an opposite effect on cellular cholesterol balance as compared with caveolin, a cholesterol-binding protein also enriched in tissues where cholesterol utilization is high, i.e. liver and steroidogenic tissues (reviewed in Ref. 8). Thus, the opposite effects of caveolin and SCP-2 on HDL-mediated cholesterol efflux suggest that these two proteins may participate in fine tuning cellular cholesterol efflux versus uptake.