Headgroup-specific Exposure of Phospholipids in ABCA1-expressing Cells*

, ABCA1 has been established to be required for the efflux of cholesterol and phospholipids to apolipoproteins such as apoA-I. At present, it is unclear whether ABCA1-mediated lipid exposure is specific with regard to lipid headgroups and whether it requires calcium activation and the presence of a lipid acceptor. In the present work, we found exofacial exposure of endogenous phosphatidylserine in the absence of apoA-I to be enhanced in ABCA1-GFP expressing MDCKII and HeLa cells compared with control cells. By using C6- N -(7-ni-trobenz-2-oxa-1,3-diazol-4-yl) (NBD)-labeled phospholipid analogues, we observed elevated redistribution of phosphatidylserine and phosphatidylethanolamine but not of phosphatidylcholine analogues from the cytoplasmic to the exoplasmic leaflet of the plasma membrane of ABCA1-GFP expressing cells. Whereas glyburide affected neither the level of exofacial endogenous PS nor the outward movement of the amino

Tangier disease, an inherited plasma high density lipoprotein deficiency. The abnormal accumulation of cholesterol in macrophages of Tangier disease patients suggested that ABCA1 is involved in lipid efflux (1)(2)(3). Indeed, subsequent studies have shown that ABCA1 is required for cholesterol and phospholipids efflux to apolipoproteins such as apoA-I (4 -7). Upon expression of functional ABCA1 in cultured cells, enhanced binding of apoA-I to the plasma membrane and increased cellular cholesterol and phospholipids efflux to apoA-I have been found. It has been suggested that ABCA1 serves as a receptor for docking of apoA-I to the plasma membrane. A chemical cross-linking approach revealed complex formation between both proteins (4,5). The relevance of ABCA-1 for binding of apoA-I is supported by the observation that Tangier patients typically fail to bind nascent apoA-I (8,9).
An alternative model suggests that ABCA1 acts as a lipid transporter pumping the amino phospholipid phosphatidylserine (PS), which is typically sequestered to the cytoplasmic leaflet of mammalian plasma membranes to the exoplasmic leaflet (6,10). Exposure of PS on the exoplasmic plasma membrane leaflet generates a microenvironment facilitating binding of apoA1 (7,8). Another recent study also provided a clear indication that ABCA1 expression increases cell surface PS (11). Furthermore, the authors (11) found a strong enhancement of apoA-I binding to apoptotic cells with an elevated level of exofacial PS. However, this binding was not sufficient to trigger phospholipid and cholesterol efflux to apoA-I. Thus, apart from PS, other phospholipids such as phosphatidylcholine (PC) might also be transported by ABCA1 to enable lipid removal from the plasma membrane to apoA-I (12). This is supported by Wang et al. (13) who concluded a direct transport of both cholesterol and phospholipids via ABCA1.
The lipid transport activity of ABCA1 may also be involved in other cellular processes as endocytosis. An enhanced endocytotic activity in Tangier disease cells has been associated with an impairment of PS exposure (14).
From these studies, it appears that ABCA1-mediated surface exposure of lipids is of rather low specificity. However, so far no systematic study on ABCA1-dependent lipid exposure has been done. Here we investigated the exposure of endogenous PS and various fluorescent lipid analogues in the plasma membrane of ABCA1-expressing MDCKII cells. In order to correlate directly the expression of ABCA1 and exposure of endogenous PS, ABCA1-GFP, which has been shown to be fully functional with respect to wild type ABCA1 (15), was expressed in MDCKII cells. We observed an ABCA1-mediated redistribution of amino phospholipid analogues of phosphatidylserine and phosphati-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ Supported by a fellowship of the Schering Foundation. dylethanolamine from the cytoplasmic side to the exoplasmic plasma membrane leaflet. The enhanced surface exposure of the PS analogue is in line with the higher level of exofacial endogenous PS in comparison with control cells, which was confirmed for ABCA1-GFP expressing HeLa cells. In contrast to the amino phospholipid analogues, we did not find an enhanced surface exposure for the phosphatidylcholine analogue in ABCA1-GFP expressing cells. In parallel, we investigated the endocytotic activity observing that the down-regulated receptor-mediated and fluid-phase endocytosis in ABCA1-GFP expressing cells cannot solely be linked to the lipid transport activity of ABCA1.
Cells-Cells were cultured in Dulbecco's modified Eagle's medium (DMEM, Invitrogen) supplemented with 10% fetal calf serum (FCS, Invitrogen), penicillin, streptomycin, and sodium pyruvate (Biochrom KG, Berlin, Germany). pBI ABCA1-GFP plasmid (Clontech) was constructed as detailed (6). Transfection was performed according to the manufacturer's procedure. Briefly 10 6 MDCKII or HeLa Tet-Off (Clontech) cells maintained in DMEM with 10% FCS were incubated overnight with the transfection mix (5 g of pBI ABCA1-GFP and 100 ng of pTK-Hyg mixed to 20 l of ExGen 500 in 0.15 M NaCl). After transfection, cells were trypsinized and seeded into 10-cm culture dishes. Transfection efficiency was monitored by flow cytometry (FACScan, BD Biosciences) before seeding. Hygromycin B selection (200 g/ml, Invitrogen) was started 48 h after transfection. Three-week-old individual cell clones were picked with cloning cylinders and expanded subsequently on the basis of GFP-positive cells visualized by fluorescence microscopy (passage 0). Cells were regularly passed before complete confluence to reduce loss of expression. Cells were grown in the absence or presence of cytostatic drugs for 24 or 72 h on 35-mm culture dishes (Nunc, Wiesbaden, Germany) for transport assays, and on two-chamber cover glasses (Nunc) for microscopy.
Expression of HA-GFP-MDCKII cells (50% confluent) were grown in DMEM supplemented with 10% FCS. Influenza virus hemagglutinin (strain X31) tagged with a GFP at the C terminus (HA-GFP) in vector pTM1 was expressed by using vaccinia virus T7 system (16). Cells were washed and infected with modified vaccinia virus (vTF7-3 virus) with a multiplicity of infection of ϳ2.5 and incubated at 37°C with 5% CO 2 for 1 h. The virus inoculum was removed, and the cells were washed once with DMEM without FCS. Subsequently, cells were transfected with a mixture of 6 l of cDNA (1.2 g/l) and 30 l of Lipofectin (1 g/l) according to the manufacturer's instructions (Invitrogen). Incubation was continued at 37°C for another 2 h. Upon removal of medium and substituting with DMEM with 2% FCS medium, cells were incubated at 31°C for 24 h. As controls, MDCKII cells were taken that have undergone the whole transfection procedure but without the HA-GFP cDNA.
Inward Redistribution of C6-NBD-PS and -PE-Cells were washed and labeled with 1 ml of 10 M C6-NBD-PS or -PE in ice-cold mPBS (17,18). After incubation on ice for 20 min, non-inserted analogues were removed by washing two times with cold mPBS. Inward redistribution of analogues was initiated by addition of mPBS prewarmed to 20°C (with or without inhibitors), containing 5 mM DFP (as in all following incubations) to prevent hydrolysis of analogues (19). After 30 min, culture dishes were transferred on ice, and cold mPBS containing BSA (final 2%, w/v) was added to extract NBD lipid analogues from the exoplasmic leaflet. C6-NBD-lipids present in the outer leaflet, not in the inner leaflet, of the plasma membrane have been shown to be selectively extracted by incubation with an excess of BSA (referred to as back exchange) (20 -23). Phospholipids with two long fatty acid chains as endogenous phospholipids, for example, are not extracted by BSA. After 10 min, BSA-containing media were collected, and back exchange was repeated with cold mPBS plus 2% BSA for 10 min, followed by lipid analysis of cells and media.
Measurement of Outward Redistribution of C6-NBD-PS and -PE by Direct Labeling-Cells were labeled with 1 ml of 10 M C6-NBD-PS or -PE on ice and incubated at 20°C for 30 min to allow inward movement of the NBD lipid analogues as described above. Inhibitors (200 M glyburide; 25 M MK 571; 10 M PSC833) and A23187 (0.5 M) in Me 2 SO were added 20 min after the beginning of the incubation at 20°C. NBD lipid analogues remaining on the cell surface were extracted twice by incubation with BSA (2% (w/v)) in mPBS for 10 min on ice. After removing the second BSA-containing medium, cells were washed twice with cold mPBS. For t ϭ 0 min, cold medium containing 2% (w/v) BSA and 5 mM DFP in mPBS with or without inhibitor was added to the cell dish and incubated for 10 min on ice. To measure C6-NBD-PS or -PE outward transport, prewarmed (15 or 37°C) mPBS with 2% (w/v) BSA and 5 mM DFP was added to the dish, and cells were incubated at 15 or 37°C for 30 min. Removal of the BSA-containing media and a second extraction with 2% BSA in mPBS for 10 min on ice terminated incubations, followed by lipid analysis.
Measurement of Outward Redistribution of C6-NBD-PC Biosynthetically Produced from C6-NBD-PA-C6-NBD-PC was biosynthetically produced from C6-NBD-PA, and its transport to the exoplasmic leaflet of the plasma membrane was assessed at 15°C as described by van Helvoort et al. (23). Briefly, cells were incubated at 15°C with 1 ml of 25 M C6-NBD-PA in mPBS. After 180 min, analogues on the exoplasmic leaflet (C6-NBD-PA and respective metabolites) were removed by washing cells two times with 2% (w/v) BSA-containing mPBS for 10 min on ice. Subsequently, cells were incubated at 15°C in 2% (w/v) BSA containing mPBS. After 30 min, the amount of extracted analogues was determined as described below.
Lipid Analysis-After the incubations, the second BSA back exchange medium was pooled with the first, and cells were scraped into mPBS. For C6-NBD-PS analysis, lipids from both scraped cells and media were extracted with isopropyl alcohol (5.5 ml per 1 ml of medium) to prevent substantial loss into the aqueous phase. Samples were centrifuged at 780 ϫ g for 5 min, and the supernatant was transferred into new glass tubes and dried. For C6-NBD-PC and -PE analysis, lipids from both cells and incubation media were extracted by the method of Bligh and Dyer (24) using 20 mM acetic acid in the aqueous phase. After two-dimensional separation (I, chloroform, methanol, 25% ammonium hydroxide (65:25:4, v/v); II, chloroform/acetone/methanol/acetic acid/ water (10:4:2:2:1, v/v)) on TLC plates, fluorescent lipid spots were visualized under ultraviolet light, scraped, and quantified as described before (17,18).
Endocytosis-Cells grown on chambered cover glasses were washed with mPBS, 1 ml of 2 mg of BSA in DMEM containing 20 mM HEPES was added, and cells were incubated for 10 min at 37°C. After washing cells twice in mPBS, cells were labeled with 1 ml of 5 mg/ml TMRdextran ("fluid-phase" endocytosis) or 1 ml of 0.2 mg/ml TR-transferrin (receptor-mediated endocytosis) in DMEM containing 20 mM HEPES, for 30 min at 37°C. After removal of supernatant, cells were washed four times with mPBS and fixed for 10 min on ice in the presence of 3% paraformaldehyde. Cells were washed twice with mPBS, and intracellular uptake of fluorophore was visualized by fluorescence microscopy (inverse standard microscope equipped with a Plan-Neofluar ϫ100 (1.3 NA) objective) and quantified employing the software package Metamorph software (version 6.1, Universal Imaging). The barrier filter sets (Carl Zeiss, Oberkochen, Germany) used were as follows: green fluorescence (NBD, GFP): BP 450 -490 excitation filter, FT 510 dichroic mirror, and LP 520 emission filter; red fluorescence: BP 546 excitation filter, FT 580 dichroic mirror, and LP 590 emission filter. Each cell was defined as a region of interest, and its average fluorescence intensity was determined. Background fluorescence, assessed by quantification of fluorescence intensity in cell free regions, was subtracted.
Statistical Analysis-Results are presented as means Ϯ S.E., if not stated otherwise, and were analyzed statistically using a two-way ANOVA (Jandel SigmaStat 2.0) considering treatment and day of experiment. Differences were considered significant for p Ͻ 0.05.  Table I), whereas the remaining cells showed only a low level or no expression of ABCA1-GFP (GFP-low). Staining of cells with the membrane-impermeable nucleic acid stain propidium iodide (PI) did not correlate with GFP fluorescence intensity ( Fig. 1). Thus, viability of cells was not impaired by expression of ABCA1-GFP.

Expression and Localization of ABCA1-GFP-Expression
Expression of ABCA1-GFP was dependent on number of cell passages. For example, the number of highly ABCA1-GFP expressing MDCKII cells declined by about 50 and 70% after the first and second passage, respectively. A similar effect was observed for HeLa cells (data not shown).
ABCA1-GFP was localized to the plasma membrane in MD-CKII and HeLa cells (Fig. 2). Cross-section images show that ABCA1-GFP is expressed in the plasma membrane with no difference between the side facing the dish and the opposite side. As revealed from co-labeling with TetramethylX-wheat germ agglutinin (not shown), ABCA1-GFP was also found in the Golgi apparatus. When MDCKII cells were allowed to reach confluency and to polarize, ABCA1-GFP distributed preferentially to the basolateral domain of the plasma membrane in line with previous observations (see Refs. 6 and 15 and data not shown). However, for our studies only nonpolarized cells were used. Both cell lines did not express endogenous ABCA1 as shown previously (6) and were verified by immunoblotting (data not shown).
C6-NBD-PS Outward Redistribution Is Enhanced in ABCA1-GFP Expressing MDCKII Cells-To study outward redistribution of C6-NBD-PS from the cytoplasmic to the exoplasmic leaflet of the plasma membrane, the cytoplasmic side was labeled following the approach of Pohl et al. (18). To this end, cells were incubated for 30 min at 20°C upon incorporation of PS analogues into the exoplasmic leaflet (see "Experimental Procedures"). During this period, analogues rapidly redistributed to the cytoplasmic side mediated by an amino phospholipid translocase activity present in the plasma membrane of MDCKII cells (17). Analogues remaining on the exoplasmic leaflet were removed by repeated washing with BSA. About 50% of analogues became internalized in control cells. We observed a slightly reduced amount of intracellularly localized analogues for ABCA1-GFP expressing cells with respect to control cells after no passage or one passage of cells. The ratio of internalized analogues between ABCA1-GFP expressing and control cells was 0.83 Ϯ 0.13 (mean Ϯ S.E.; n ϭ 3). However, the difference between ABCA1-GFP expressing cells and control cells disappeared when cells were used after three culturing passages correlating with the decline of ABCA1-GFP expressing cells (see above).
Subsequently, to follow outward redistribution of analogues, cells were continuously incubated at 37 or 15°C in BSA-containing medium. Any analogue reaching the cell surface was extracted by BSA into the suspension medium. The kinetics of outward redistribution at 37°C are shown in Fig. 3.
For control cells, about 50% of analogues originally located intracellularly became accessible to BSA within 30 min of incubation at 37°C. At any time point measured, exposure of C6-NBD-PS to the exoplasmic leaflet was significantly higher (ANOVA, p Ͻ 0.05) for ABCA1-GFP expressing cells with respect to control cells (Fig. 3A).
When outward redistribution was measured for 30 min at 15°C, only 30% of analogues could be extracted. However, exposure of C6-NBD-PS was also enhanced in ABCA1-GFP expressing cells and was even more pronounced in comparison to 37°C (Fig. 4, passage 0, and Fig. 5). At this temperature (15°C) vesicular transport is abolished (25), suggesting that the enhanced exposure of the PS analogue in ABCA1-GFP expressing cells is related to redistribution from the cytoplasmic leaflet. The differences between ABCA1-GFP expressing cells and control cells cannot be explained by cell-dependent hydrolysis of analogues, which was similar for ABCA1-GFP expressing and control cells (data not shown).
Differences in outward redistribution between ABCA1-GFP expressing and control cells depended on the number of passages. Usually, after two passages the exposure of C6-NBD-PS in ABCA1-GFP expressing cells was reduced to the level of control cells (Fig. 4) mostly due to a reduction in the number of ABCA1-GFP expressing cells (see above). Very likely, ABCA1-GFP expressing cells grow at a slower rate than non-expressing cells.
After labeling of cells with C6-NBD-PS, inward redistribution and outward translocation of analogues was also followed by fluorescence microscopy (images not shown). Upon labeling of cells on ice, only the plasma membrane became fluorescent. NBD fluorescence was much more intense than GFP fluorescence. After incubation of cells at 20°C for 30 min and subsequent BSA extraction of analogues from the exoplasmic leaflet, bright labeling of the cytoplasm was observed. Thus, in all cases the ratio of exposed C6-NBD-PS between ABCA1-GFP expressing and control cells was not affected (Fig.  5; only shown for glyburide).
Upon pretreatment of cells with the calcium ionophore A23187, significantly enhanced (ANOVA, p Ͻ 0.05) exposure of C6-NBD-PS in the presence of 2 mM Ca 2ϩ was found for ABCA1-GFP expressing cells (Fig. 5). The ratio of extracted C6-NBD-PS between ABCA1-GFP expressing and control cells increased from about 1.25 to 1.6, and was only slightly reduced in the presence of 200 M glyburide (Fig. 5). A23187 treatment in medium without Ca 2ϩ did not affect the outward redistribution of C6-NBD-PS in ABCA1-GFP expressing and in control cells (not shown). Similarly, neither addition of EGTA to iono-  phore-treated cells in the absence of Ca 2ϩ nor addition of Ca 2ϩ to cells without ionophore led to increased exposure of the PS analogue (data not shown). Thus, whereas A23187 was dispensable in order to observe an influence of ABCA1 on C6-NBS-PS exposure in our experiments, we found it to stimulate exposure in ABCA1-GFP expressing cells but not in controls. Similar observations have been made for ABCA1-expressing HeLa cells (6,26).
C6-NBD-PE Outward Redistribution Is Enhanced in ABCA1-GFP Expressing MDCKII Cells-Subsequent to incubation of cells labeled with C6-NBD-PS, we observed significant metabolic conversion into C6-NBD-PE. About 25 and 50% of intracellularly localized C6-NBD-PS were converted into C6-NBD-PE at 15 and 37°C, respectively, with no significant difference between ABCA1-GFP expressing and control cells. The extent of outward redistribution of C6-NBD-PE was much lower with respect to C6-NBD-PS. Although about 50% of the PS analogue became exposed after 30 min at 37°C, the amount of accessible C6-NBD-PE was in the order of 10%.
Similar to the PS analogue, we measured enhanced exposure (ANOVA, p Ͻ 0.05) of C6-NBD-PE on the exoplasmic leaflet of In a second approach, cells were labeled with C6-NBD-PE and incubated for 30 min at 20°C following the technique used for C6-NBD-PS. It has been shown previously that the PE analogue is also rapidly transported to the cytoplasmic leaflet of the plasma membrane of MDCKII cells (17). After removal of analogues remaining on the exoplasmic leaflet, cells were incubated in BSA-containing medium for 30 min at 37 or 15°C. The amount of accessible PE analogue was much higher in comparison to the previous situation where exposure of metabolically converted C6-NBD-PE was studied. About 80 -90 and 50 -70% of C6-NBD-PE was accessible to BSA at 37 and 15°C, respectively. For both temperatures, we found an enhanced appearance (ANOVA, p Ͻ 0.05) of the PE analogue on the surface of ABCA1-GFP expressing cells with respect to control cells (Fig. 6B). However, the difference between both cell lines was less pronounced in comparison to that measured by exposure of C6-NBD-PE metabolically converted from the PS analogue. We surmise that the high fraction of accessible PE analogue may partly mask the difference between both cell lines.
Outward Redistribution of C6-NBD-PC Is Not Affected by Expression of ABCA1-GFP-Next, we examined the outward transport of C6-NBD-PC in ABCA1-GFP expressing and control MDCKII cells. To this end, cells were incubated with C6-NBD-PA. This lipid analogue is partially converted to C6-NBDdiacylglycerol, which rapidly crosses the plasma membrane and becomes available for intracellular synthesis of C6-NBD-PC (23). Transport of the newly synthesized analogues to the cell surface was measured at 15°C where vesicular traffic is blocked (25). After incubation of cells (passage 1) with C6-NBD-PA for 120 min at 15°C and removal of NBD lipid analogues from the exoplasmic leaflet (see "Experimental Procedures"), cells were incubated for a further 30 min at 15°C in the presence of BSA. Less than 5% of C6-NBD-PC was extracted into the BSA medium with no significant difference between the ABCA1-GFP expressing cell line (3.62 Ϯ 0.22%

ABCA1-mediated Lipid Exposure
(n ϭ 2)) and control cells (3.38 Ϯ 0.55% (n ϭ 2)). 2 We have also measured the amount of C6-NBD-PC extracted during a 3-h incubation of cells in the presence of BSA at 15°C, following labeling with C6-NBD-PA in the presence of BSA. After this period, the amount of extracted PC analogue was on the order of about 50 -60%. Again, no significant difference was seen between control and ABCA1-GFP expressing cells.
Notably, we detected only conversion to C6-NBD-PA into C6-NBD-PC and C6-NBD-PG. C6-NBD-PA may also be converted metabolically to other lipid analogues, e.g. to C6-NBD-PE. However, under our conditions synthesis of the PE analogue was beyond the level allowing quantitative detection.
C6-NBD-PS Outward Redistribution Is Also Enhanced in ABCA1-GFP Expressing HeLa Cells-The outward redistribution of C6-NBD-PS was also studied in HeLa cells by BSA extraction as described (see above). Similar to MDCKII cells, we found an enhanced exposure of the fluorescent PS analogue in ABCA1-GFP expressing cells compared with control cells (transfected with irrelevant plasmid but the same vector) (Fig. 3B).
No enhanced outward redistribution was found for HeLa cells expressing the mutant form of ABCA1-GFP with impaired ATPase activity because of mutation of the lysine residue in the consensus motif in both ATP binding folds (for details of the mutant, see Ref. 6) (Fig. 3B). The fraction of cells expressing mutant ABCA1-GFP at a high level was similar to that of wild type ABCA1-GFP expressing HeLa cells (see above). Furthermore, mutation did not affect trafficking and localization of the protein as verified by confocal microscopy (data not shown).
ABCA1-expressing Cells Expose More Endogenous PS on the Cell Surface than Controls-Exposure of endogenous PS in MDCKII and HeLa cells (passage 0) was tested by flow cytometry. PS present on the exoplasmic plasma membrane leaflet of cells was detected by labeling with APC-annexin V, a high affinity PS-binding protein (27). To exclude damaged cells, co-labeling with propidium iodide was performed. Regions were set to exclude subcellular particles, and only single cells were counted (10,000 cells/sample). For propidium iodide staining, an additional gate was set. Usually, about 5-10% of cells were excluded because of staining with propidium iodide. No difference was observed between control and ABCA1-GFP express-

ABCA1-mediated Lipid Exposure
ing cells (see above, Fig. 1, E and F). Simultaneously, GFP fluorescence was measured (see above). Low or no ABCA1-GFP expressing MDCKII and HeLa cells bound about 2.5-and 3.1-fold more, whereas high ABCA1-GFP-expressing cells bound 3.5-and 4.1-fold more APC-annexin V than control cells, respectively (Fig.  7, A and B). Thus, binding of APC-annexin V was significantly enhanced for ABCA1-GFP expressing cells with respect to control cells and did correlate with the extent of ABCA1-GFP expression, which was not observed for HeLa cells expressing mutant ABCA1-GFP with impaired ATPase activity (see above, data not shown). There are no indications that glyburide affected APCannexin V binding neither for control nor for ABCA1-GFP expressing cells (data not shown). Consistent with the observation presented in Fig. 4, the difference in annexin V binding between ABCA1-GFP expressing and control cells declined with increasing numbers of cell culture passages (data not shown).
To verify that enhanced exposure of endogenous PS is related to the activity of ABCA1-GFP but not simply because of expression of a GFP-tagged membrane protein, we investigated binding of APC-annexin V to MDCKII cells expressing influenza virus hemagglutinin (HA-GFP; see "Experimental Procedures"). No difference of APC-annexin V binding was observed between HA-GFP expressing cells and control cells that were treated according to the transfection procedure but without the HA-GFP plasmid (Fig. 7C).
Receptor-mediated and "Fluid-phase" Endocytosis Is Reduced in ABCA1-GFP Expressing MDCKII Cells-Recently, the enhanced endocytosis in Tangier fibroblasts has been associated with the activity of ABCA1 (14). Therefore, we investigated whether expression of ABCA1-GFP in MDCKII cells affects endocytosis. Transferrin is known to be taken up by clathrin-coated pits upon binding to surface receptors (28). Cells were incubated with Texas Red-labeled transferrin for 30 min at 37°C. Internalization of transferrin was seen by punctate structures corresponding to endosomes (not shown). Quantification of intracellular fluorescence revealed a significantly lower amount (ANOVA, p Ͻ 0.05) of internalized transferrin in ABCA1-GFP expressing cells in comparison to control cells (Fig. 8A). Reduced endocytosis cannot be explained by a lower binding of fluorescent transferrin to the plasma membrane of ABCA1-GFP expressing cells. The amount of transferrin bound to the plasma membrane after incubation for 30 min on ice was similar to control and ABCA1-GFP expressing cells (not shown). In the presence of glyburide (200 M), uptake of transferrin increased to the level of control cells, which was not affected by the inhibitor.
To assess fluid-phase endocytosis, cells were incubated with a tetramethylrhodamine-labeled high molecular weight dextran for 30 min at 37°C. Similar to transferrin, uptake of dextran was significantly reduced (ANOVA, p Ͻ 0.05) in ABCA1-GFP expressing cells (Fig. 8B). Addition of glyburide raised the amount of internalized dextran to that of control cells in the absence of glyburide. However, in contrast to receptor-mediated endocytosis, glyburide caused a dramatic decrease of dextran uptake by about 50% for control cells in all experiments (n ϭ 3). As a result, internalization of dextran in the presence of glyburide was about 1.5 times higher in ABCA1-GFP expressing cells than in controls (Fig. 8B). control cells at 37°C. Likewise, exofacial C6-NBD-PE originating from intracellularly metabolized C6-NBD-PS was enhanced in ABCA1-GFP expressing cells. No influence of inhibitors of MDR1 Pgp and MRP1 was observed precluding any role of those ABC transporters in exposure of amino phospholipid analogues.

Involvement of ABCA1 in Transport of
An elevated appearance of amino phospholipid analogues on the surface of ABCA1-GFP expressing cells was also observed at 15°C when vesicular transport is inhibited in MDCKII cells (25). This confirms that the increased accessibility of amino phospholipid analogues to BSA in ABCA1-GFP expressing cells is associated with an enhanced outward redistribution of analogues in the plasma membrane. However, these data do not preclude that ABCA1 affects both intracellular vesicle transport (to the plasma membrane) and transbilayer organization of lipids in subcellular membranes (29) which may contribute to/interfere with the increased exposure of amino phospholipid analogues at the cell surface of ABCA1-GFP expressing cells at 37°C and to differences in the temperature dependence of lipid exposure observed between C6-NBD-PE and C6-NBD-PS.
In agreement with our results on the surface exposure of C6-NBD-PS and consistent with a role of ABCA1 in the transport of endogenous PS, binding of APC-annexin V to the cell surface of ABCA1-expressing cells was significantly higher than binding to control cells. The dependence of PS exposure on ABCA1-GFP expression was confirmed by two independent observations. First, the amount of exofacial C6-NBD-PS and endogenous PS correlated with the expression of ABCA1-GFP. With increasing number of cell passages, the level of ABCA1-GFP expression decreased. This was accompanied by a reduction of C6-NBS-PS exposure almost to the level of control cells and a significant decline of annexin V binding. Second, binding of APC-annexin V was higher for high ABCA1-GFP expressing cells in comparison to low ABCA1-GFP expressing cells. These results confirm previous observations that have consistently shown enhanced binding of annexin V to ABCA1-GFP expressing cells (6,11,14). For example, an enhanced binding of fluorescent annexin V was found in nontransfected RAW264.7 cells after induction of ABCA1 expression by cAMP analogues, indicating an increased exposure of endogenous PS on the exoplasmic leaflet of those cells (11).
The exposure neither of C6-NBD-PS nor of endogenous PS was reduced in ABCA1-expressing cells upon treatment of cells with glyburide. Glyburide has been found to inhibit ABCA1mediated efflux of phospholipids and cholesterol to apoA-I even at a lower concentration (100 M) as used in our study (6,11,13,30). However, although several studies have shown that the level of exofacial PS is elevated in ABCA1-GFP expressing cells, it has not been reported to our knowledge that the exposure of PS is inhibited by glyburide. A reasonable explanation to this different influence of glyburide on lipid transfer may be provided by a recent study (26) that has shown that lipid transport activity and docking properties of ABCA1 are governed by topographically distinct domains. Glyburide might affect the docking side of ABCA1 and thereby the lipid efflux to apoA-I, but not its lipid transport activity.
Involvement of ABCA1 in Transport of PC across the Plasma Membrane-Precursor labeling of cells with C6-NBD-PA at 15°C permitted us to investigate the transport of the newly synthesized C6-NBD-PC from the cytoplasmic to the exoplasmic plasma membrane leaflet. By using the same experimental approach, it has been shown that the ABCB1 (MDR1 Pgp) mediates a transport from the cytoplasmic to the exoplasmic leaflet of the C6-NBD-PC analogue (17,20). For ABCA1-GFP expressing cells, we did not observe an enhanced outward redistribution of the PC analogue with respect to control cells.
Thus, it is unlikely that ABCA1 functions as a PC transporter, at least of the PC analogue used. We cannot preclude that endogenous PC is transported by ABCA1. Recently, it has been suggested that ABCB1 and ABCB4 are involved in cell surface translocation of natural PC (31). However, these transporters have been shown to recognize NBD analogues of PC as well.
Furthermore, our results do not preclude an ABCA1-mediated transport of PC upon binding of apoA-I to ABCA1. This could be caused either by switching or by broadening of the lipid specificity of the transport activity upon apoA-I binding. An alternative hypothesis is that ABCA1 facilitates the transfer of PC from the exoplasmic leaflet to apoA-I. A similar function has been proposed for half-size ABC transporters ABCG5 and ABCG8. Small (32) suggested that the heterodimer ABCG5-ABCG8 may facilitate the exposure of membranebound cholesterol to bile salts. Thus, ABCA1 might present PC to apoA-I in an ATP-dependent manner. Once apoA-I complexes have been formed, cholesterol is taken up from the plasma membrane by these complexes. Previously, it has been shown that the transfer of phospholipids and cholesterol from cells to apoA-I represents different pathways. Although the transfer of phospholipids was dependent upon ABCA1 and sensitive to glyburide, cholesterol efflux was independent of functional ABCA1 (30). Furthermore, phospholipid transfer preceded the binding of cholesterol to apoA-I.
Lipid-transporting ABC Transporters-Taken together, our results support the hypothesis that ABCA1 may function as a lipid transporter specifically recognizing amino phospholipids. Other members of the ABC transporter family have been identified in mammalian cells as mediators of phospholipid outward transport, differing in lipid specificity. ABCB4 (MDR3) and its mouse homologue (mdr2 Pgp) transport PC (31,(33)(34)(35)(36)(37). In contrast, ABCB1 (MDR1 Pgp), giving rise to a frequent form of pleiotropic resistance in tumor cells, transports various lipid analogues (18,23,36,38) and probably also the endogenous lipids platelet-activating factor (a short-chain PC), phosphatidylcholine, phosphatidylserine, and glucosylceramide (18,31,39,40) from the cytoplasmic to the exoplasmic plasma membrane leaflet. ABCC1 (MRP1), belonging to the ABCC family, was reported to mediate outward transport of C6-NBD-PC and -PS in murine erythrocytes (41,42) and NBD-sphingolipids in epithelial cells (43). At the present state, ABCA1 seems to belong to a class of ABC lipid transporters of a rather high specificity, recognizing amino phospholipids. However, to provide convincing evidence for lipid transport activity of ABCA1 as well as for its putative lipid specificity, functional studies on reconstituted proteoliposomes are required. ABCA1 reconstituted into giant liposomes with a distinct lipid composition and a marginal degree of membrane curvature similar to that of a cell, avoiding the generation of high lateral pressure that could potentially inhibit lipid transport, might resemble the system of choice to adequately address its lipid transport activity (18,44).
Role of ABCA1 in Endocytosis-Recently, Zha et al. (14) have shown that receptor-mediated endocytosis is up-regulated in homozygous Tangier fibroblasts in comparison to control cells. In addition, the authors (14) also observed enhanced receptormediated endocytosis upon glyburide treatment of control fibroblasts with functional wild type ABCA1 when studying uptake of fluorescent transferrin. These results were rationalized in the frame of a model originally developed by Devaux (45,46) and supported by subsequent experimental studies (47)(48)(49). According to the model, invaginations of biological membranes can originate from transbilayer transport of lipids, leading to a difference of surface area between both membrane monolayers and, because of the very low compressibility of lipid membranes, finally to bending of the membrane. For example, such a difference of surface area could be built up by the rapid and efficient transport of amino phospholipids, in particular of phosphatidylserine, from the exoplasmic to the cytoplasmic leaflet of the plasma membrane of eukaryotic cells by the amino phospholipid translocase (50). Thus, this directed transport could be an essential determinant of endocytosis. Indeed, knocking out of putative lipid transporters of the plasma membrane of yeast cells leads to an inhibition of endocytosis (49). Zha et al. (14) proposed that transport of PS by ABCA1 in the direction opposite to that of the amino phospholipid translocase would reduce the surface area difference between monolayers and counteract invagination of the plasma membrane.
Here we have re-investigated the endocytotic activity of control and ABCA1-GFP expressing cells. Following the protocol of Zha et al. (14), we also observed a significantly reduced receptor-mediated endocytosis of fluorescent transferrin in ABCA1-GFP expressing cells in comparison with controls. Moreover, the level of endocytosis of ABCA1-GFP expressing cells increased to the level of control cells upon treatment with glyburide. Although these data are in agreement with those of Zha et al. (14), they cannot be explained straightforward in the frame of the model outlined above because we did not find an inhibition of PS transport in glyburide-treated ABCA1-GFP expressing cells. Our data do not necessarily contradict that such a model might be applicable to explain the influence of ABCA1 on endocytosis. Apart from its lipid transport activity, ABCA1 might modulate endocytosis by different yet unknown modes, at least one of them being sensitive to glyburide. Even in the fibroblast cell system used by Zha et al. (14), it is not known whether the increase of endocytosis in glyburide-treated wild type ABCA1-GFP expressing cells is due to an inhibition of PS outward transport. This issue has not been studied by the authors (14) (see above also). Thus, further studies are warranted to explore how ABCA1 can influence endocytosis, as well as which ABCA1-dependent sides of the endocytotic machinery are sensitive to glyburide. Indeed, at the present stage it cannot be excluded that glyburide affects endocytosis by various pathways as illustrated by our results on fluid-phase endocytosis. Again in agreement with the work of Zha et al. (14), we found that the reduced fluid-phase endocytosis in ABCA1-expressing cells increased upon treatment with glyburide to the level observed for control cells in the absence of glyburide. However, treatment of glyburide caused a significant decrease of this endocytotic activity in control cells. Unfortunately, Zha et al. (14) did not report on the influence of glyburide on fluid-phase endocytosis in their cell systems. Nevertheless, our results clearly show that glyburide may differently affect the various endocytotic pathways of eukaryotic cells. Finally, it has to be emphasized that endocytosis phenotype and consequences of glyburide treatment may depend on the cell system used.