Glycosphingolipid accumulation inhibits cholesterol efflux via the ABCA1/apolipoprotein A-I pathway: 1-phenyl-2-decanoylamino-3-morpholino-1-propanol is a novel cholesterol efflux accelerator.

Cellular glycosphingolipid (GSL) storage is known to promote cholesterol accumulation. Although physical interactions between GSLs and cholesterol are thought to cause intracellular cholesterol "trapping," it is not known whether cholesterol homeostatic mechanisms are also impaired under these conditions. ApoA-I-mediated cholesterol efflux via ABCA1 (ATP-binding cassette transporter A1) is a key regulator of cellular cholesterol balance. Here, we show that apoA-I-mediated cholesterol efflux was inhibited (by up to 53% over 8 h) when fibroblasts were treated with lactosylceramide or the glucocerebrosidase inhibitor conduritol B epoxide. Furthermore, apoA-I-mediated cholesterol efflux from fibroblasts derived from patients with genetic GSL storage diseases (Fabry disease, Sandhoff disease, and GM1 gangliosidosis) was impaired compared with control cells. Conversely, apoA-I-mediated cholesterol efflux from fibroblasts and cholesterol-loaded macrophage foam cells was dose-dependently stimulated (by up to 6-fold over 8 h) by the GSL synthesis inhibitor 1-phenyl-2-decanoylamino-3-morpholino-1-propanol (PDMP). Unexpectedly, a structurally unrelated GSL synthesis inhibitor, N-butyldeoxynojirimycin, was unable to stimulate apoA-I-mediated cholesterol efflux despite achieving similar GSL depletion. PDMP was found to up-regulate ABCA1 mRNA and protein expression, thereby identifying a contributing mechanism for the observed acceleration of cholesterol efflux to apoA-I. This study reveals a novel defect in cellular cholesterol homeostasis induced by GSL storage and identifies PDMP as a new agent for enhancing cholesterol efflux via the ABCA1/apoA-I pathway.

tabolism (1)(2)(3)(4). A clear example of this relationship is the demonstration that the cellular GSL storage detected in fibroblasts derived from patients with a variety of sphingolipid storage diseases (SLSDs) is accompanied by accumulation of endosomal/lysosomal cholesterol (1). Subsequent studies in human macrophage foam cells and fibroblasts indicated that GSL accumulation induced either by exogenously added lactosylceramide (LacCer) or by the glucocerebrosidase inhibitor conduritol B epoxide (CBE) results in concomitant accumulation of cholesterol (2,4).
The factors that control the accumulation of cholesterol with GSLs remain to be identified. Puri et al. (1,4) provided evidence supporting a "molecular trap" hypothesis, whereby increased GSL levels lead to increased cholesterol sequestration in late endosomes/lysosomes. Although the endosomal/lysosomal location of GSL and cholesterol in SLSDs is well known, it is becoming increasingly clear that cellular membrane composition is also altered when lysosomal storage of GSL/cholesterol occurs. Recent detailed studies by te Vruchte et al. (5) revealed that the accumulation of both GSL and cholesterol in ordered lipid microdomains commonly referred to as lipid rafts or, because of their relative resistance to detergent solubilization, detergent-resistant microdomains (DRMs) may contribute directly to lysosomal accumulation by altering endosomal transport pathways. These observations are consistent with previous proposals that alterations to membrane DRM structure could contribute to sphingolipidosis (6 -8).
Although the co-localization of cholesterol with GSLs (4) and the accumulation of cholesterol and GSLs within DRMs (5) under GSL storage conditions support the GSL/cholesterol molecular trap concept (4), it is not known to what extent other pathways that regulate cellular cholesterol homeostasis are affected by GSL storage. Previous work has suggested that cholesterol that accumulates as a result of GSL storage may not be available to the sensing pathways that normally control intracellular cholesterol balance. Specifically, we have shown that macrophages induced to store cholesterol by treatment with LacCer or CBE down-regulate their apoE secretion (2). This is in contrast to the predicted up-regulation of apoE secretion in response to cholesterol loading (9). Similarly, Puri et al. (4) showed that low density lipoprotein (LDL) receptor expression is up-regulated when fibroblasts accumulate cholesterol as a consequence of LacCer treatment, again in contrast to the response predicted by Brown and Goldstein (10). The pathways that contribute to GSL-induced cellular cholesterol accumulation clearly require further study.
A key pathway regulating cellular cholesterol homeostasis relies on the apoA-I-mediated efflux of cholesterol from cells via ABCA1 (ATP-binding cassette transporter A1) (11). Cells from patients with genetic defects in ABCA1 leading to Tangier disease exhibit impaired cholesterol efflux to apoA-I and increased cellular cholesterol accumulation (12). Although it is clear that cellular GSL storage promotes cholesterol accumulation, it is not known whether alterations in the apoA-I-mediated cholesterol efflux pathway are affected by GSL status. In this study, we therefore examined the impact of cellular GSL accumulation or depletion on apoA-I-mediated cholesterol efflux.
Cell Culture-Human foreskin fibroblasts from a "control" subject and patients with GM1 gangliosidosis, Sandhoff disease, and Fabry disease (cell line identification numbers AG01518, GM10919, GM00203A, and GM004390, respectively) were obtained at passages 4 -8 from the Coriell Institute (Camden, NJ) and grown under standard culture conditions at 37°C in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (FCS), 2 mM glutamine, 50 IU/ml penicillin G, and 50 mg/ml streptomycin in a humidified atmosphere containing 5% CO 2 . The cells were subcultivated by trypsinization at a 1:4 ratio and transferred to 35-mm plastic Petri dishes (Costar) for use in experiments. Fibroblasts were routinely used from passages 14 to 20. Human monocytes were isolated from donor buffy coats using Ficoll-Paque as described previously (14). Monocytes were allowed to differentiate into macrophages for 1 week in RPMI 1640 medium containing 10% (v/v) human serum supplemented with L-glutamine, penicillin, and streptomycin as above before further treatments. Cell viability was assessed by trypan blue exclusion.
Cell Treatments-Cells were treated with 40 M LacCer for 48 h as described previously (2,4). Using HPLC methods, we have shown that this protocol results in an approximate doubling of cellular LacCer levels (2). Cells were also treated with PDMP, NB-DNJ, or CBE as indicated for up to 72 h. These compounds were added to cells in complete growth medium containing 10% (v/v) serum except where MDMs were also loaded with acLDL, in which case 10% (v/v) lipoprotein-depleted serum was used.
Cholesterol and Phospholipid Efflux Assays-Cells were labeled with 2 Ci/ml [ 3 H]cholesterol for 72 h as described (13,17). Where indicated, MDMs were labeled with [ 3 H]cholesterol-labeled acLDL (13). [ 3 H]Cholesterol labeling was initiated when fibroblasts were 60% confluent or after 7 days differentiation for MDMs. Cells were subsequently washed with phosphate-buffered saline (PBS) and incubated overnight in RPMI 1640 medium containing 0.1% (w/v) bovine serum albumin (BSA) to allow equilibration of [ 3 H]cholesterol in all cellular pools. Equilibrated [ 3 H]cholesterol-labeled cells were washed with PBS and incubated in 2 ml of efflux medium containing either DMEM (fibroblasts) or RPMI 1640 medium (MDMs) and 0.1% BSA with or without 25 g/ml apoA-I. Where indicated, efflux was also assessed using phospholipid vesicles (PLVs; 200 mg/ml total phospholipid), prepared using 1-palmitoyl-2oleoylphosphatidylcholine and repeated sonication and extrusion as described previously (18), or methyl-␤-cyclodextrin (MBCD; 250 mM) (19) as acceptor. At the times specified (up to 8 h), 50-ml aliquots of the media were removed and analyzed by scintillation counting. A similar approach was used to measure 3 H-phospholipid efflux after labeling with [ 3 H]choline as described previously (13). Directly after equilibration, replicate wells were harvested in 800 l of 0.2 M NaOH, and samples were analyzed by scintillation counting either directly (for cholesterol) or after removal of choline (for phospholipids) (13). These cell lysates served as a check for the effect of the various treatments on [ 3 H]cholesterol and 3 H-phospholipid loading/labeling and were used to calculate lipid efflux to the medium as a percentage of total counts present at t ϭ 0 h. Unless stated otherwise, experiments were performed in triplicate and repeated three times.
Analysis of ABCA1 Expression by Western Blotting and Real-time PCR-Fibroblast ABCA1 expression was measured by Western blotting and real-time PCR (20,21). Cell lysates were mixed with Laemmli sample buffer and incubated for 30 min at 37°C. Samples were separated on SDS-7% polyacrylamide gels using a Mini-PROTEAN II system (Bio-Rad) and subsequently transferred to nitrocellulose membranes (Bio-Rad). Loading equivalence and transfer efficiency were monitored by Ponceau S staining. The membranes were then incubated for 16 h at 4°C with rabbit anti-human ABCA1 polyclonal antibody (1:1000 dilution; Novus Biologicals), followed by incubation for 1 h with horseradish peroxidase-conjugated goat anti-rabbit IgG antibody (catalog no. P0448, Dako Corp.). Where indicated, the blots were stripped and reprobed using goat anti-human ␤-actin polyclonal antibody (1: 1000 dilution; Santa Cruz Biotechnology, Inc.). Blots were developed by enhanced chemiluminescence (Amersham Biosciences); the membranes were exposed to Fuji x-ray film, developed, and scanned; and signal intensity was quantified using NIH Image software. Cells were also harvested for total RNA using TriReagent (Sigma) according to the manufacturer's instructions with the exception that 1-bromo-3-chloropropane was substituted for chloroform. Quantitative real-time PCR was performed using an ABI 7700 sequence detector and analyzed using ABI PRISM sequence detector software (Version 1.6.3, PE Biosystems). IQ SYBR Green Supermix (Bio-Rad) was used according to the manufacturer's instructions using gene-specific primers for ABCA1 (forward, 5Ј-GCACTGAGGAAGATGCTGAAA-3Ј; and reverse, 5Ј-AGT-TCCTGGAAGGTCTTGTTCAC-3Ј), with annealing at 60°C normalized to that of a housekeeping gene, porphobilinogen deaminase (forward, 5Ј-GAGTGATTCGCGTGGGTACC-3Ј; and reverse, 5Ј-GGCTCCGATG-GTGAAGCC-3Ј), with annealing at 55-60°C (21). Melting curve analysis was performed to confirm production of a single product in these reactions.
Cell-surface Biotinylation-Cell-surface biotinylation was performed as described previously (22) with slight modifications. Briefly, fibroblast cells were grown in 10-cm diameter Petri dishes to confluency. They were washed with ice-cold PBS and treated with 1 mg/ml sulfosuccinimidyl 2-(biotinamido)ethyl-1,3Ј-dithiopropionate (Pierce) for 30 min at 4°C on a platform mixer. The biotinylation treatment was stopped with two washes of ice-cold quench buffer (50 mM Tris, 0.1 mM EDTA, and 150 mM NaCl), followed by a wash with ice-cold PBS. Cells were collected by scraping, pelleted by centrifugation, resuspended in lysis buffer (50 mM Tris-Cl (pH 8.0), 150 mM NaCl, 0.2% NaN 3 , 0.1% SDS, 100 g/ml phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, 1% Triton X-100, and 0.5% sodium deoxycholate), and left on ice for 30 min. Following centrifugation, the supernatant was transferred to a new tube, mixed with 100 l of UltraLink-immobilized NeutrAvidin gel (Pierce), and incubated overnight at 4°C on a platform mixer. The gel was pelleted by centrifugation and washed five times with lysis buffer. 55 l of loading dye (60 mM Tris-Cl (pH 6.8), 25% glycerol, 2% SDS, 0.1% bromphenol blue, and 350 mM mercaptoethanol) was added to the pellet, followed by incubation at 37°C for 30 min. The resulting protein sample was analyzed by Western blotting for ABCA1 (as described above) or for scavenger receptor class B, type I (SR-BI) using a rabbit anti-mouse polyclonal antibody (1:1000 dilution; Novus Biologicals) that cross-reacts with human SR-BI.
Cellular GSL Expression-The GSL profiles of fibroblasts were analyzed as described previously (2,23). Briefly, cells grown to confluency in 75-cm 2 flasks were rinsed three times with PBS and extracted in 2:1 (v/v) chloroform/methanol. The GSL fractions were isolated by silicic acid chromatography, and the glycan moiety was released by ceramide glycanase addition (23). The GSL glycans were then fluorescently labeled with 2-aminobenzamide and analyzed by normal phase HPLC as described previously (24). Glycan peaks was identified by calculation of glucose units, derived from a partially hydrolyzed dextran standard, and comparison with previously published glucose units (2,23). Cell-surface GSL expression levels were assessed by measuring the binding of fluorescently labeled cholera toxin B (CTxB) to cell-surface GM1 (10 min at 4°C). Alexa Fluor 448-conjugated CTxB (Molecular Probes, Inc., Eugene, OR) was used according to the manufacturer's instructions. The labeled cells were rinsed five times with PBS and lysed in 50 mM Tris (pH 7.5), 0.15 M NaCl, 1% (v/v) Triton X-100, 1% (v/v) deoxycholate, 10 mM EDTA, 0.1% (w/v) SDS, and Complete protease inhibitors (Roche Applied Science), and a 200-ml aliquot was analyzed at Ex 485 nm /Em 530 nm in a CytoFluor multiwell plate reader (Per-Septive Biosystems Inc., Framingham, MA).
Other Methods-The cellular content of detergent-insoluble cholesterol was analyzed as described previously (3,25). In brief, cells were labeled with [ 3 H]cholesterol and equilibrated as described above. The cells were then cooled on ice and extracted in 10 mM Tris (pH 7.4), 0.15 M NaCl, and 1 mM EDTA containing 1% (v/v) Triton X-100 at 4°C. The lysates were centrifuged at 15,000 ϫ g at 4°C, and the supernatants were carefully removed from the insoluble pellets. The pellets were then dissolved in 0.2 M NaOH, and aliquots of the supernatants and pellets were analyzed by scintillation counting to calculate the proportion of cholesterol in the detergent-insoluble fraction. Estimation of cellular acyl-CoA:cholesterol acyltransferase activity was achieved by analyzing the proportion of esterified [ 3 H]cholesterol present as a percentage of the total cellular [ 3 H]cholesterol by thin-layer chromatography to separate cholesterol from cholesterol esters in 1:1 (v/v) heptane/ethyl acetate. Standards of cholesterol and cholesteryl esters were identified by charring at 100°C in 10% CuSO 4 and 8.5% H 3 PO 4 . Protein concentrations were calculated using the bicinchoninic acid assay and BSA as a standard (Sigma).
Statistical Analysis-Statistical significance was determined using the two-tailed t test for unpaired data. A p value Ͻ0.05 was considered significant.

LacCer, CBE, and SLSDs Inhibit Cholesterol Efflux to
ApoA-I-Previous studies have shown that fibroblasts derived from patients with SLSDs accumulate GSL and cholesterol (1). This accumulation of cholesterol with GSL storage can be reproduced in vitro by treating cells with LacCer or the glucocerebrosidase inhibitor CBE (2)(3)(4). Here, we investigated the impact of such treatments on the ability of fibroblasts to efflux cholesterol to apoA-I, as this is a key regulator of cholesterol homeostasis (26). Treatment of human fibroblasts with 40 mM LacCer inhibited apoA-I-mediated cholesterol efflux while having no effect on the low levels of basal efflux observed in the presence of albumin (Fig. 1A). At the 8-h time point, apoA-Imediated cholesterol efflux was inhibited by 40% (8.9 Ϯ 0.3% versus 5.3 Ϯ 0.5%, mean Ϯ S.E., n ϭ 3; p Ͻ 0.05). Increasing the LacCer concentration to 70 M resulted in a further 13% reduction in apoA-I-mediated cholesterol efflux at 8 h. 2 Similar to the effect of LacCer, treatment of fibroblasts with CBE under conditions shown previously to promote accumulation of both GSLs and cholesterol (2, 4) significantly inhibited apoA-I-dependent cholesterol efflux (Fig. 1B). These data indicate that the accumulation of cellular cholesterol that accompanies GSL storage is associated with an impaired capacity to efflux cholesterol to apoA-I. We also examined cholesterol efflux from SLSD fibroblasts, and this was found to be significantly inhibited compared with healthy controls, although the magnitude of the differences was variable depending on the SLSD (Fig. 1C).
To examine whether LacCer inhibits fibroblast cholesterol efflux specifically, we analyzed apoA-I-mediated phospholipid efflux. LacCer significantly inhibited phospholipid efflux to apoA-I, although the magnitude of this inhibition was not as marked as that seen for cholesterol ( Fig. 2A). ABCA1 is an important plasma membrane transporter that regulates cholesterol and phospholipid efflux to apoA-I, but not to other acceptors such as PLVs and MBCD (27,28). LacCer did not inhibit cholesterol efflux to either PLVs or MBCD, indicating that the action of LacCer may be linked to ABCA1 expression or activity.
Does Cellular GSL Accumulation Correlate with ABCA1 Expression?-Transcriptional regulation of ABCA1 expression is a key modulator of cholesterol efflux to apoA-I (29). To understand the factors that contribute to the impaired efflux of cholesterol from LacCer-treated cells, we used quantitative real-time PCR and Western blotting to analyze for expression of ABCA1. Treatment of fibroblasts with 40 M LacCer did not significantly reduce ABCA1 mRNA levels (n ϭ 4) overall. 2 Similarly, total ABCA1 protein levels were only moderately reduced with LacCer treatment (Fig. 3A). The level of cellsurface ABCA1 expression was reduced, however, by as much as 50% after LacCer treatment (Fig. 3B). These data show that, although LacCer does not regulate ABCA1 expression directly, intracellular pathways, presumably related to ABCA1 trafficking or degradation, are significantly perturbed.
When the SLSD and control fibroblasts were compared, we found no correlation between the levels of ABCA1 mRNA expression and the rate of cholesterol efflux ( Fig. 4A; cf. Fig. 1C). Although, the GM1 gangliosidosis cells were inefficient in cholesterol efflux activity and contained only one-third the level of ABCA1 mRNA compared with control fibroblasts, the Fabry disease fibroblasts expressed increased ABCA1 mRNA levels while still exhibiting a suppressed capacity to efflux cholesterol to apoA-I ( Fig. 4A; cf. Fig. 1C). The suppression of cholesterol efflux from SLSD fibroblasts is apparently not simply a result of dysregulated ABCA1 expression. This implies that multiple factors are likely to contribute to the impaired cholesterol efflux in different lysosomal storage disorders.
Correlates of Impaired Cholesterol Efflux in SLSD Fibroblasts-Acid sphingomyelinase deficiency results in sphingomyelin (SM) accumulation and suppressed cellular cholesterol efflux (30). This impaired cholesterol efflux correlates with decreased trafficking of cholesterol to the acyl-CoA:cholesterol acyltransferase-accessible pool (30). The acyl-CoA:cholesterol acyltransferase-accessible cholesterol pool (predominantly endoplasmic reticulum-associated) is selectively utilized in the ABCA1/apoA-I efflux pathway in fibroblasts (31). As an index of cholesterol trafficking to the acyl-CoA:cholesterol acyltransferase-accessible pool (31), we analyzed the extent of [ 3 H]cholesterol esterification in control and SLSD fibroblasts and in fibroblasts treated with exogenous LacCer. The level of choles-terol esterification was significantly decreased under all conditions resulting in GSL accumulation (Fig. 4B). The relative levels of cholesterol esterification did not, however, strictly correlate with cholesterol efflux efficiency in the SLSD cells ( Fig. 4B; cf. Fig. 1C). Therefore, the potential decrease in acyl-CoA:cholesterol acyltransferase-accessible cholesterol, which contributes directly to the pool that is effluxed to apoA-I (31), appears to contribute to varying degrees to the impaired cholesterol efflux associated with the different SLSDs.
We next investigated whether the impaired cholesterol efflux induced by GSL accumulation in the SLSDs is associated more closely with a specific glycolipid species or total GSL load. HPLC analysis of control fibroblast GSLs revealed the major species as LacCer, ceramide trihexoside, and gangliosides GM3 and GM2, with a minor contribution made by GM1 (Fig. 5). In the SLSD cells, the GSL profiles were altered in concordance with the known genetic defects (i.e. Fabry disease, ␣-galactosidase defect resulting in elevated ceramide trihexoside levels; Sandhoff disease, ␤-hexosaminidase A/B defect resulting in elevated GM2/globoside levels; and GM1 gangliosidosis, ␤-galactosidase defect resulting in elevated GM1 levels). In the SLSD cells, additional GSLs were also markedly increased (e.g. LacCer and ceramide trihexoside in GM1 gangliosidosis and GM2 in Fabry disease) compared with control cells (Fig. 5A). Although cellular LacCer levels were increased to a variable extent in all of the SLSD cells examined, only the total cellular concentration of GSLs correlated (inversely) with apoA-I-mediated cholesterol efflux (Fig. 5B).
Modulation of ApoA-I-dependent Cholesterol Efflux by PDMP-Having established that increased cellular GSL levels suppressed cholesterol efflux, we wanted to know whether the reverse was also true, i.e. could a reduction in cellular GSL levels result in accelerated cholesterol efflux to apoA-I? To test this, we treated control fibroblasts with the GSL synthesis inhibitor PDMP. This compound inhibits glucosylceramide transferase activity, the initial step in GSL synthesis, thereby decreasing cellular levels of a range of GSLs, including glucosylceramide, LacCer, and GM1 (32)(33)(34)(35). Confirming previous studies, treatment of fibroblasts with PDMP for 72 h inhibited cellular GSL levels as determined by binding of fluorescently labeled CTxB to cell-surface GM1 (Fig. 6). PDMP was initially added to subconfluent (ϳ60%) fibroblasts so that, after 72 h of incubation, the cells were confluent. At PDMP concentrations Ն40 M, cell growth was inhibited as assessed by total protein measurements. 2 At concentrations of 2.5-10 M, there was no significant effect of the drug on cell protein, whereas GSL levels were reduced by 11 and 23%, respectively (Fig. 6). We therefore used PDMP at low concentrations to minimize potential growth inhibitory effects in subsequent analyses. PDMP significantly stimulated cholesterol efflux to apoA-I (Fig. 7). At 10 M, PDMP increased cholesterol efflux to apoA-I by up to 6-fold (2.8 Ϯ 0.8-fold increase above apoA-I alone, mean Ϯ S.E., n ϭ 7) as assessed at the 8-h time point while having no effect on basal efflux (Fig. 7). At 2.5 M, PDMP increased efflux to apoA-I by 1.6 Ϯ 0.1-fold (mean Ϯ S.E., n ϭ 3) at t ϭ 8 h. The lack of effect on basal efflux implied that PDMP selectively promotes efflux via ABCA1. To confirm this, we assessed the effect of PDMP on cholesterol efflux to PLVs and MBCD, acceptors that do not require ABCA1 (27,28). At 10 M, PDMP did not affect cholesterol efflux to either PLVs or MBCD (Fig.  8). These data indicate that PDMP promotes cholesterol efflux through an ABCA1-dependent pathway.
Effect of PDMP on ABCA1 Expression and Cholesterol Solubility-At least two mechanisms could account for the effect of PDMP on apoA-I-dependent cholesterol efflux. First, the drug could up-regulate ABCA1 expression or activity either directly or via altered concentrations of an endogenous lipid mediator. Second, if molecular trapping of cholesterol within GSL-en- riched lipid microdomains (also known as DRMs or rafts) occurs, one would predict that depletion of cellular GSL would lead to increased mobility of membrane cholesterol, which may therefore be more efficiently transported through ABCA1 (which is not raft-associated in fibroblasts (36,37)). Analogous to the second mechanism, depletion of plasma membrane SM is thought to "release" cholesterol from ordered lipid microdomains, resulting in accelerated efflux to apoA-I (38), whereas accumulation of cellular SM (as occurs in sphingomyelinasedeficient cells) "traps" cholesterol, thereby inhibiting efflux to apoA-I (30). We therefore conducted experiments to assess these possibilities. PDMP dose-dependently increased fibroblast ABCA1 mRNA levels (Fig. 9A). At 10 M, PDMP increased ABCA1 mRNA levels by up to 2.7-fold (2.1 Ϯ 0.3-fold, mean Ϯ S.E., n ϭ 3; p Ͻ 0.05). There was an overall trend toward increased ABCA1 mRNA expression at 2.5 M, but this was not statistically significant in all experiments (1.3 Ϯ 0.3fold, mean Ϯ S.E., n ϭ 3). We also detected moderate increases in ABCA1 protein levels by Western blotting, but only at the higher PDMP dose (Fig. 9B). The levels of cell-surface ABCA1 expression were increased to a similar degree by PDMP, indicating that PDMP acts primarily by regulating ABCA1 expression and not by modulating trafficking to the cell surface (Fig.  9C). Overall, these data suggest that at least part of the effect of PDMP on cholesterol efflux is mediated through effects on ABCA1 expression.
To assess the potential for PDMP to alter cholesterol distribution in DRMs, we measured cholesterol partitioning in Triton-soluble and Triton-insoluble fractions using an established method (3,25). Using cold 1.0% (w/v) Triton X-100 extraction, we found that ϳ25% of the [ 3 H]cholesterol was insoluble (Fig.  10). There was dose-dependent trend for PDMP to reduce the amount of cholesterol present in the Triton-insoluble fraction; and at a concentration of 10 M, ϳ21% of the cholesterol was insoluble (Fig. 10). At 40 M PDMP, only 13% of the cholesterol remained insoluble, and this difference was statistically signif-icant (Fig. 10B). Although the trend for 10 M PDMP to reduce cholesterol insolubility appears modest, a potential contribution to accelerated efflux cannot be ruled out. These data show that, under the conditions in which PDMP accelerated apoA-Imediated cholesterol efflux, a trend toward increased cholesterol solubility was also observed. This indicates that PDMP may promote the release of cholesterol from within DRMs, consistent with previous work showing that PDMP treatment increases the mobile fraction of synthetic lipid probes in the plasma membrane (39).
Lack of Effect of NB-DNJ on ApoA-I-mediated Cholesterol Efflux-The above data suggest that PDMP may exert its action both through alterations in ABCA1 expression (or activity) and via reducing the physical trapping of cholesterol by GSLs within DRMs. Because compounds that promote cellular cholesterol efflux may be clinically useful as accelerators of reverse cholesterol transport (40), we conducted additional experiments in an attempt to further separate the effect of PDMP on ABCA1 expression from its moderate impact on the implied physical interaction of GSL with cholesterol in DRMs. To do this, we employed a structurally distinct GSL synthesis inhibitor, NB-DNJ, which also inhibits glucosylceramide transferase activity (41). We conducted dose-response experiments in the range of 10 -500 mM NB-DNJ to achieve a level of GSL synthesis inhibition comparable to that observed with 10 M PDMP without adversely affecting cell growth or morphology. 2 Based on the CTxB binding assay, 50 M NB-DNJ reduced GSL levels by 43% (Fig. 11A). This level of GSL depletion was more effective than that achieved using PDMP at subcytostatic concentrations ( Fig. 11A; cf. Fig. 6). Similar to our experiments with PDMP, 50 M NB-DNJ reduced the amount of Triton-insoluble cholesterol from 27 to 16% (Fig. 11A). Although NB-DNJ reduced cellular GSL levels and also increased cholesterol solubility to an extent even greater than observed with PDMP treatment, it did not significantly promote apoA-I-dependent cholesterol or phospholipid efflux (Fig. 11). This indicates that, under normal physiological conditions (in the absence of GSL storage), a reduction in cellular GSL levels per se does not appear to promote cholesterol efflux to apoA-I.
PDMP Promotes ApoA-I-dependent Cholesterol Efflux from Human Monocyte-derived Macrophages-Our studies using NB-DNJ suggest that PDMP may promote cholesterol efflux to apoA-I via a pathway that is not reliant on its ability to inhibit GSL synthesis. Although GSL synthesis inhibition per se may not be sufficient to induce cholesterol efflux, the fact that PDMP selectively promoted apoA-I-dependent cholesterol efflux could offer a new approach to promote reverse cholesterol transport in vivo. This would be of therapeutic interest for the treatment of atherosclerosis, where large amounts of cholesterol accumulate within macrophage foam cells in the arterial intima (42). We therefore assessed the impact of PDMP on apoA-I-mediated cholesterol efflux from human MDMs and acLDL-loaded MDM foam cells. At 10 M, PDMP increased cholesterol efflux from MDMs and MDM foam cells to apoA-I by ϳ2-fold as assessed at the 8-h time point (Table I). Interestingly, although PDMP did not significantly promote basal cholesterol efflux from fibroblasts, it did increase basal efflux from both MDMs and acLDL-loaded MDMs (Fig. 7; cf. Table I). The effect of PDMP on basal cholesterol efflux is potentially related to the high levels of apoE secreted by MDMs, which could offer an additional cholesterol efflux vehicle (43). We have shown previously that CBE inhibits macrophage foam cell apoE secretion while promoting cholesterol accumulation (2). Here, we also assessed the impact of CBE treatment on apoA-I-mediated cholesterol efflux. Similar to the effect of CBE on fibroblasts, treatment of MDMs with 50 M CBE inhibited subsequent cholesterol efflux to apoA-I (7.1 Ϯ 0.82% versus 3.7 Ϯ 0.38%, mean Ϯ S.E., n ϭ 3; p Ͻ 0.05) as assessed at the 8-h time point. CBE treatment also tended to increase total cellular cholesterol counts at t ϭ 0 h by ϳ20%, but this was not statistically significant ((244 Ϯ 10) ϫ 10 3 cpm/well versus (298 Ϯ 21) ϫ 10 3 cpm/well, mean Ϯ S.E., n ϭ 3; p ϭ 0.09). DISCUSSION Our studies show for the first time that cellular LacCer accumulation or agents/genetic diseases that cause GSL storage inhibit apoA-I-mediated cholesterol efflux. These findings indicate that the cholesterol accumulation that accompanies GSL storage in SLSD is compounded by a reduced capacity of these cells to divest themselves of excess cholesterol. Interestingly, certain SLSDs are accompanied by low plasma HDL cholesterol levels. For example, Gaucher disease (patients have a ␤-glucocerebrosidase deficiency, and the condition is therefore mimicked by CBE in vitro) is accompanied by low plasma HDL cholesterol levels, which are increased toward normal levels with enzyme replacement therapy (44). Our experiments using CBE indicate that this may due to alterations in the ABCA1/apoA-I cholesterol efflux pathway. Patients with Niemann-Pick disease (types A and B) are deficient in acid sphingomyelinase activity and also have low plasma HDL cholesterol levels (45). Furthermore, cells derived from sphingomyelinase knockout mice are defective in their ability to efflux cholesterol to apoA-I, illustrating that sphingolipid storage can also inhibit cholesterol efflux (30).
Patients with Niemann-Pick disease type C (deficiency in NPC1 or HE1/NPC2, proteins that regulate intracellular cholesterol trafficking) also have low plasma HDL cholesterol lev- els (46). A plausible explanation for this was revealed when cells derived from an NPC1-deficient subject were found to have low levels of ABCA1 mRNA and protein expression and a reduced capacity to efflux cholesterol to apoA-I (46). It was suggested that NPC1 plays a role in the regulation of oxysterol ligands responsible for the liver X receptor-␣ (LXR␣)-mediated regulation of ABCA1 (46). Our data indicated that, in certain GSL storage cells (notably GM1 gangliosidosis), ABCA1 mRNA levels were lower than in control cells. It is possible that the cholesterol transport pathways that generate oxysterol ligands for LXR␣ may be perturbed in specific GSL storage diseases as they are in NPC1; however, this will clearly require further investigation.
Alternatively, the molecular trapping of cholesterol induced by GSL storage could inhibit the rate of ABCA1-dependent efflux by limiting the rate of transport of internal cholesterol pools to the plasma membrane (1). Our experiments with normal fibroblasts showed that phospholipid efflux to apoA-I was also suppressed by LacCer. Because phospholipids are not predicted to be trapped within DRMs (47), our data argue against such a mechanism limiting cholesterol efflux. It seems more likely that dysregulated intracellular lipid trafficking contributes to impaired efflux to apoA-I. The decrease in the acyl-CoA: cholesterol acyltransferase-accessible cholesterol pool under all conditions of GSL accumulation is consistent with this proposal.
Regardless of the mechanisms operating in the different SLSDs, our data indicate that the impaired capacity of SLSD cells to release cholesterol to apoA-I may contribute to the reduced HDL levels observed in specific GSL storage conditions in vivo. This could ultimately promote additional GSL storage, as HDL also stimulates cellular GSL efflux (2). Our data, in addition to the studies cited above, indicate that the association of impaired ABCA1-mediated cholesterol efflux with SLSDs likely contributes to the storage phenotype.
Previous studies indicated that cellular cholesterol efflux is inhibited by SM accumulation and accelerated when SM is depleted by the addition of sphingomyelinase to cells in vitro (30,38,48). Although these observations appear to be analogous to our findings with cellular GSL accumulation and depletion (via PDMP treatment), important differences exist. Depletion of plasma membrane SM appears to release cholesterol from ordered lipid microdomains, resulting in accelerated efflux to apoA-I (38), whereas accumulation of cellular SM (as occurs in sphingomyelinase-deficient cells) traps cholesterol, thereby inhibiting efflux to apoA-I (30). Although there may be mechanistic similarities when either SM or GSL levels are increased, leading to cholesterol trapping (as suggested previously (1, 7)), depletion of cellular GSL levels was not sufficient to promote cholesterol efflux, as demonstrated by our experiments with NB-DNJ (Fig. 11).
The use of cold detergent extraction provides a facile approach to detect gross changes in membrane lipid order (25,47). Although we detected an increase in Triton-soluble cholesterol when GSL depletion was induced with NB-DNJ, this alone was insufficient to promote apoA-I-mediated cholesterol efflux. It is not clear why plasma membrane SM depletion stimulates apoA-I-mediated cholesterol efflux (38), whereas GSL depletion induced by NB-DNJ does not, as both sphingolipids form complexes with cholesterol (Ref. 49 and references cited therein). The magnitude of change in cholesterol solubility we detected may be insufficient to have an impact on apoA-I-mediated cholesterol efflux, i.e. in the studies of Ito et al. (38), sphingomyelinase treatment decreased 0.1% Triton X-100-insoluble cholesterol levels by ϳ50%. In addition, where GSL levels were chronically reduced in our studies, compensatory pathways (such as increased saturation of SM and glycerophospholipid acyl chains) may be induced to maintain membrane lipid order (50) and thereby minimize alterations in cholesterol solubility.
Despite the fact that GSL synthesis inhibition per se did not directly promote cholesterol efflux, our studies using PDMP did reveal a novel means to promote apoA-I-specific cholesterol efflux. In contrast to previous studies in which sphingomyelinase treatment of cells nonspecifically promoted cholesterol efflux to cyclodextrin (48,51), PDMP appeared to promote efflux through an ABCA1-specific pathway. This activity was associated with a moderate up-regulation of ABCA1 mRNA and protein levels. The mechanisms responsible for this increase in ABCA1 expression are unknown. One possibility is that PDMP treatment induces LXR␣-dependent transcription either directly or via increased synthesis of an oxysterol ligand. Consistent with this, in fibroblasts treated with 10 M PDMP, we also detected a 2-fold increase in apoE mRNA levels and an 8-fold increase in ABCG1 mRNA levels by real-time PCR. 2 Both of these genes are transcriptionally regulated by LXR␣ (9,21,52). Although these changes are not likely to contribute to the increased cholesterol efflux to apoA-I observed (as ABCG1 does not promote efflux to apoA-I (53), and we did not detect secretion of apoE by fibroblasts), it does indicate that other LXR␣-regulated genes are induced.
We have focused on the regulation of ABCA1 expression, as this is a well documented modulator of apoA-I-dependent cholesterol efflux; however, other regulators of ABCA1 activity might also contribute to the effects of PDMP, and these include regulation of ATP binding and ABCA1 phosphorylation and calpain-dependent proteolysis (54 -57). An additional possibility is that changes in ceramide levels induced by PDMP could increase the cell-surface expression of ABCA1 (22); however, our data indicated that there was no selective increase in fibroblast plasma membrane ABCA1 after treatment with 10 M PDMP. Furthermore, if PDMP were used at concentrations high enough to significantly increase intracellular ceramide levels, we would expect cell growth to be inhibited, and this was not the case. Although our data indicate that PDMP promotes cholesterol efflux via LXR␣-mediated transcriptional up-regulation of ABCA1, additional details of this pathway, including the identification of an LXR␣ ligand, remain to be elucidated.
Our finding that cellular GSL accumulation inhibits cholesterol efflux may have implications for the development of atherosclerosis, where GSLs, including LacCer, accumulate within  10 M PDMP Human monocytes were isolated from donor buffy coats and differentiated to MDMs for 1 week in 10% (v/v) human serum. MDMs from Donor 1 were then labeled with ͓ 3 H͔cholesterol in growth medium, whereas MDMs from Donor 2 were loaded with ͓ 3 H͔cholesterol-labeled acLDL (50 g/ml protein). Under both labeling conditions, cells were also incubated in the presence or absence of 10 M PDMP. Cholesterol efflux was assessed after a 8-h incubation with 25 g/ml apoA-I, and counts in cell supernatants are expressed as a percentage of the total. The presence of PDMP significantly reduced ͓ 3 H͔cholesterol counts in MDMs at t ϭ 0 h by 32 and 30% in the experiments for Donors 1 and 2, respectively. Cell proteins (ϳ600 mg/well) were not significantly affected by PDMP. All values are the means of triplicate determinations with S.E. in parentheses. arterial lesions in humans and in animal models of the disease (58 -60). Previous studies have shown that LacCer is proatherogenic, e.g. by acting as a smooth muscle cell mitogen (61,62); however, its impact on apoA-I-mediated cholesterol efflux was not recognized. Numerous studies have focused on the mechanisms resulting in cholesterol accumulation in atherosclerotic lesions, principally within MDM foam cells (42). Our discovery that PDMP promotes cellular cholesterol efflux to apoA-I and reduces GSL levels (processes that appear to be independent of each other) suggests that PDMP or structurally related compounds may represent a novel approach to treat atherosclerosis.
In conclusion, our studies show for the first time that cellular GSL accumulation inhibits apoA-I-mediated cholesterol efflux. Our studies also reveal PDMP as a novel cholesterol efflux accelerator. This may provide a useful approach to promote apoA-I-mediated efflux and reverse cholesterol transport in vivo.