Apolipoprotein A-I–Mediated Efflux of Sterols From Oxidized LDL–Loaded Macrophages
Abstract Although oxidized low-density lipoprotein (OxLDL) can accumulate in macrophages in vitro, generating cholesterol-loaded cells, little attention has been paid to the capacity of such macrophages loaded with OxLDL to export cholesterol and oxidized sterol moieties. In vitro lipid-loaded cells were generated by incubating primary cultures of mouse peritoneal macrophages with acetylated LDL (AcLDL) or OxLDL for 24 hours. The cellular content of native cholesterol, individual cholesteryl esters, and 7-ketocholesterol was determined by high-performance liquid chromatography. These cells were then incubated with medium containing apolipoprotein (apo) A-I and albumin or albumin alone for up to 24 hours; cholesterol and oxidized sterol efflux were measured both in terms of intracellular depletion and extracellular accumulation. Macrophages loaded with AcLDL accumulated cholesterol and large quantities of cholesteryl esters, whereas OxLDL-loaded cells accumulated cholesterol, a number of oxidized compounds (predominantly 7-ketocholesterol), and a relatively small quantity of cholesteryl esters. AcLDL-derived cells released approximately 50% of their total cholesterol (unesterified and esterified) to apo A-I–containing medium over 24 hours in the form of unesterified cholesterol, whereas OxLDL-derived cells released approximately 30% of their total cholesterol and 7% of their total content of 7-ketocholesterol over the same period. There was minimal efflux of any sterol in the absence of apo A-I. The proportions of cholesterol and 7-ketocholesterol released by either AcLDL- or OxLDL-loaded cells were not reduced by inhibiting cellular acyl-CoA:cholesterol acyl transferase using Sandoz 58-035, despite substantial alterations in the proportions of both free cholesterol and (in OxLDL-loaded cells) free 7-ketocholesterol in these cells. Furthermore, the subcellular distributions of both cholesterol and 7-ketocholesterol in individual subcellular organelle fractions were identical to that of free cholesterol in nonloaded cells, indicating that these sterols in OxLDL-loaded cells are not selectively sequestered in lysosomes. 7-Ketocholesterol is released much less efficiently than cholesterol from OxLDL-loaded cells. In addition, OxLDL-loaded cells release cholesterol less efficiently than do cells derived from AcLDL. It is possible that this impairment of efflux from OxLDL-loaded cells influences the generation and persistence of the foam cell phenotype in vivo and may therefore contribute to the atherogenicity of OxLDL.
Presented in part at the 1992 Annual Scientific Meeting of the European Lipoprotein Club, Tutzing, Germany.
- Received July 13, 1993.
- Accepted October 10, 1994.
It has been proposed that one of the mechanisms by which high-density lipoprotein (HDL) might exert its apparent protective effect against cardiovascular mortality is that of “reverse cholesterol transport.”1 By promoting the removal of cholesterol from intimal macrophages, HDL may permit the transfer of this cellular cholesterol to the liver, either directly or indirectly via transfer to other lipoprotein particles, and thus facilitate its excretion in the bile.2 3
The initiation of reverse cholesterol transport in vivo is thought to occur in the subintimal space where foam cells accumulate.4 Because interstitial fluid contains a greater concentration than plasma of lipoprotein-unassociated apolipoprotein (apo) A-I with pre-β mobility5 and because cholesterol-laden fibroblasts release cholesterol to an apo A-I–only–containing particle with pre-β mobility,6 apo A-I may be involved in the initial steps of cholesterol efflux. Moreover, purified apolipoproteins including apo A-I7 and other particles containing apolipoproteins8 9 10 have been shown to promote cholesterol efflux in vitro.
While much research has concerned possible mechanisms of reverse cholesterol transport, relatively little attention has been directed to the comparative release from lipid-loaded macrophages of oxidized and unoxidized lipids derived from oxidized low-density lipoprotein (OxLDL) and the unoxidized lipids derived from acetylated LDL (AcLDL). This may be of major importance, as there is now substantial evidence suggesting a role for LDL oxidation in the development of atherosclerosis. This includes the identification of lipid oxidation products (including 7-ketocholesterol) in human atheroma,11 12 in macrophage foam cells isolated from rabbit and human atheroma,13 and in plasma14 ; the suspected localization of adducts to apo B of OxLDL in rabbit atherosclerotic plaque15 ; and the observation that various cells of the arterial intima can oxidize LDL.16
Whereas AcLDL-derived in vitro foam cell models accumulate cholesterol and cholesteryl esters,17 OxLDL-derived foam cells contain much lower quantities of cholesteryl esters and contain a number of oxidized cholesterol compounds.18 19 We hypothesized that there may be a major difference in the ability of cells loaded with unoxidized cholesterol compounds (AcLDL-loaded) or oxidized cholesterol compounds (OxLDL-loaded) to release their respective lipid contents. This difference could in turn contribute to the persistence of the foam cell phenotype.
We have developed a method to measure some of the cholesterol and cholesteryl ester oxidation products of OxLDL20 and have used this method to directly compare the accumulation and release of cholesterol and 7-ketocholesterol from macrophages loaded with OxLDL or AcLDL. Our studies have identified (1) impaired release of 7-ketocholesterol to apo A-I from OxLDL-loaded cells, as compared with the release of cholesterol from these cells, and (2) impaired release of cholesterol from OxLDL-loaded cells as compared with that from AcLDL-loaded cells.
Preparation of LDL
LDL (1.05>d>1.02 g/mL) was isolated from plasma of normolipidemic, healthy subjects using two sequential centrifugation steps at 10°C using a Beckman L8-M centrifuge and Ti 70 rotor at 50 000 rpm (242 000g) for 24 hours. The LDL was dialyzed against 4×1 L deoxygenated phosphate-buffered saline (PBS, calcium and magnesium free; Flow Laboratories) containing 0.1 mg/mL chloramphenicol (Boehringer Mannheim) and 1.0 mg/mL ethylenediaminetetraacetic acid (EDTA, British Drug Houses). The LDL was stored in the dark at 4°C under nitrogen until use (within 7 days).
Modification of LDL
LDL was acetylated using a modification of a previously described method21 22 using 6 μL acetic anhydride per milligram LDL protein. After acetylation, LDL was dialyzed against 4×1 L PBS (LDL:PBS, 1:200 vol/vol) containing chloramphenicol (0.1 g/L) and Chelex-100 (1 g/L, BioRad Laboratories) over 16 hours to remove excess saturated sodium acetate and acetic anhydride.
LDL was dialyzed against 3×1 L PBS (LDL:PBS, 1:200 vol/vol) containing chloramphenicol (0.1 g/L) and Chelex-100 (1 g/L), and then 1×1 L PBS containing chloramphenicol only over 16 hours before oxidation. Oxidation was achieved as previously described20 by incubating LDL at a concentration of 400 μg LDL protein/mL in a sterile solution of PBS containing 10 μmol/L cupric chloride (British Drug Houses) for 24 hours at 37°C.
Nondenaturing Agarose Gel Electrophoresis
Two- to four-microliter samples of AcLDL and OxLDL were loaded directly onto 1% Universal Agarose gels (Ciba-Corning) and subjected to electrophoresis in Tris-barbitone buffer (pH 8.6) at 90 V for 45 minutes. LDL was visualized by staining with fat red 7B. LDL that had not undergone oxidation or acetylation was used as a reference. The relative electrophoretic mobility of modified LDL was calculated by dividing the distance traveled during electrophoresis by the distance traveled by the reference LDL, and satisfactory acetylation or oxidation was confirmed if the relative electrophoretic mobility was greater than or equal to 3.
Purified, lecithin:cholesterol acyltransferase–free, human apo A-I was generously provided by Prof P. Barter and Dr K.-A. Rye (University of Adelaide, Australia) after isolation and lyophilization, as previously described.23 The apo A-I was reconstituted in 3 mol/L guanidine HCl solution and dialyzed against 5×1 L 0.01 mol/L Tris-HCl buffer (pH 8.2) over 5 days. Purity of the reconstituted apo A-I was confirmed using SDS PAGE, as described below, by detection of a single band of molecular weight 28 000.
Samples of the reconstituted apo A-I were mixed with sample buffer containing 0.125 mol/L Tris-HCl (pH. 6.8), 2% (wt/vol) SDS, 10% (vol/vol) glycerol, 0.1% (wt/vol) bromophenol blue, and 2% (vol/vol) mercaptoethanol heated to 100°C for 2 minutes. Samples containing 2.4 μg of apo A-I were applied to wells on either 20% or 4% to 15% precast (BioRad) gradient SDS–polyacrylamide gels (Laemmli). The bands were visualized by staining for protein with Coomassie brilliant blue. Nonradioactive “rainbow” molecular weight markers (range, 14 300 to 200 000; Amersham) were run concurrently for calibration. Gels were run in 10 mmol/L Tris-glycine buffer (pH 8.3) for 50 minutes at 150 V and were fixed in ethanol/acetic acid/water (38:10:52, vol/vol/vol).
Isolation and Lipid Loading of Macrophages and Conditions for Sterol Efflux
Procedures followed were in accordance with ethical guidelines of the National Health and Medical Research Council of Australia. Resident macrophages were isolated from Quackenbush-Swiss strain (QS) mice, after asphyxiation using carbon dioxide gas by peritoneal lavage with ice-cold Dulbecco’s minimum essential medium (DMEM; GIBCO cat 320-1885 AJ) containing 0.38% (wt/vol) sodium citrate, penicillin G (50 U/mL), and streptomycin (50 μg/mL). The isolated cells were immediately plated in 35-mm–diameter tissue culture wells (Costar) at 5.0 to 6.0×106 cells per well, incubated at 37°C for 1 to 2 hours, and then washed four times with prewarmed PBS to remove nonadherent cells. After washing, cells were incubated in DMEM containing 10% (vol/vol) human lipoprotein-deficient serum (LPDS; d>1.25), 50 U/mL penicillin, 50 μg/mL streptomycin, and either AcLDL (50 or 100 μg/mL) or OxLDL (25 or 50 μg/mL) for 24 hours. Representative cultures were then taken for analysis. The remaining cultures then were washed with warm PBS three times to remove LDL-containing medium and incubated for a further 24 hours in DMEM (with penicillin and streptomycin as specified above) containing 100 μg/mL bovine serum albumin (BSA, essentially fatty acid free; Sigma) with or without apo A-I (25 μg/mL, efflux medium). Cells and media were then separately extracted and analyzed for oxidized and unoxidized cholesterol and cholesteryl esters (as described below).
In some experiments, an overnight (18-hour) or 4-hour equilibration of cells loaded with AcLDL or OxLDL was performed before their being incubated with efflux medium. In this case, cells were washed three times with warm PBS to remove LDL-containing medium, incubated with DMEM containing 100 μg/mL albumin overnight (or for 4 hours), and then washed again with warm PBS before the addition of efflux medium.
Macrophage acyl-CoA:cholesterol acyltransferase (ACAT) was inhibited in some experiments using Sandoz 58-035 (generously provided by Drs Nordmann and Nadelson, Sandoz Pharmaceuticals). This compound was dissolved in ethanol at a concentration of 10 mg/mL, diluted in DMEM-containing medium, and filter sterilized immediately before its addition to cells. Preliminary experiments confirmed that concentrations of 2, 5, and 10 μg/mL of inhibitor in DMEM were not toxic to cells. As maximal inhibition of intracellular cholesteryl ester accumulation was seen at both 5 and 10 μg/mL, 5 μg/mL was used in all subsequent experiments. When used, the inhibitor was present in both phases of experiments, that is, during loading (in DMEM with LPDS and AcLDL or OxLDL) and during efflux (in DMEM with albumin with or without apo A-I).
Extraction of Medium
Efflux medium (1.5 mL) was removed and spun in an Eppendorf centrifuge at 4°C for 2 minutes at 14 000 rpm (16 000g) to remove any cellular debris. One milliliter of the supernatant was mixed with 10 μL trifluoroacetic acid (TFA) and was extracted into methanol (2.5 mL) and then hexane (10 mL).20 Preliminary experiments had shown that standards such as cholesteryl linoleate were poorly extracted from DMEM containing albumin, but that this could be overcome (with 100% extraction efficiency) by the addition of TFA. Samples were stored at −80°C until analysis.
Preparation of Cell Extracts
After washing three times with ice-cold PBS, cells were lysed by incubating in 0.6 mL of cold 0.2 mol/L NaOH for 15 minutes at 4°C. Of this, 0.1 mL was removed and stored at −20°C for later protein assay, and 0.4 mL was removed, added to 0.6 mL ice-cold PBS, and immediately extracted into methanol and hexane (as above) in the presence of 20 μmol/L butylated hydroxytoluene (Sigma) and 2 mmol/L EDTA. In some earlier experiments, a slightly different protocol was used. The only significant difference between the protocols was the addition of 0.6 mL of PBS before extraction in later experiments (as stated above), which resulted in a 30% improvement in extraction of 7-ketocholesterol from cell extracts (total extraction efficiency of 7-ketocholesterol increasing from 70% without PBS to 100% with PBS). Experiments in which this addition had not been made have been corrected accordingly and have been repeated using the improved extraction and the corrected values have been confirmed. Samples were stored after extraction at −80°C until analysis, which was usually performed within 7 days.
Resident macrophage cultures (2×108 per 150-cm2 tissue culture flask) were prepared as described above and incubated for 24 hours in DMEM containing 10% (vol/vol) LPDS with (OxLDL-loaded cells) or without (nonloaded cells) 25 μg/mL OxLDL. The cultures then were washed with warm PBS and incubated overnight in DMEM containing 100 μg/mL albumin. The cells then were removed by scraping and ruptured in homogenization buffer (0.25 mol/L sucrose, 20 mmol/L HEPES, 0.5 mmol/L EDTA, pH 7.0) using a shear force device comprising an 8020-μm precision bore in a stainless steel block that contained an 8004-μm stainless steel ball (Industrial Tectonics) as previously described.24 Cell rupture was assessed using ethidium bromide25 and was routinely 80% to 95% efficient. The homogenate was centrifuged at 800g for 15 minutes to remove cellular debris, unbroken cells, and nuclei; the postnuclear supernatant was fractionated on a 1% to 22% Ficoll gradient underlaid with a 45% Nycodenz cushion in a Beckman VTi 65.2 rotor at 240 000g and 4°C for 90 minutes.24 Fractions of 26×200 μL were collected from the bottom of the tube, and aliquots were assayed for density (by refractometry), aryl sulfatase activity,24 and protein. The fractions then were combined into five major fractions representing the density ranges d<1.02, 1.02 to 1.035, 1.035 to 1.05, 1.05 to 1.095, and 1.095 to 1.25, areas in which free lipids, plasma membrane, endosomes, lysosomes, and cell proteins partition, respectively.24 The lipid compositions of these individual fractions were determined after extraction into methanol/hexane, as described below.
Analysis of Cholesterol and Cholesteryl Esters and Their Oxidation Products
This method has been published recently in detail.20 In brief, unoxidized cholesterol and cholesteryl esters were separated using reverse-phase high-performance liquid chromatography (HPLC) at room temperature on a C-18 column (Supelco) using an eluent of acetonitrile/isopropanol (30:70, vol/vol) and detected by their 210-nm absorbance using an Activon UV-200 absorbance detector. Oxidation products, in particular 7-ketocholesterol, were also analyzed using HPLC on a C-18 column but were separated with a solvent system of acetonitrile/ isopropanol/water (44:54:2, vol/vol/vol) and detected by their absorbance at 234 nm.
Cholesterol, cholesteryl esters (cholesteryl arachidonate, cholesteryl linoleate, cholesteryl oleate, cholesteryl palmitate, and cholesteryl stearate), 7-ketocholesterol, and cholesteryl linoleate hydroxide (CL-OH) were quantified by deriving standard curves using commercially available standards (Sigma for cholesterol, cholesteryl esters, and 7-ketocholesterol; Cayman Chemicals for CL-OH). The curves expressed a linear relation between the chromatographic peak areas and the mass of the standard. Chromatograms of cell and medium extracts were integrated, and the masses of individual cholesterol compounds were quantified by reference to these standard curves. Results are expressed as nanomoles per milligram of cell protein and are presented as mean±SD of triplicate determinations or as mean±range of duplicate determinations. Duplicate points are always indicated.
In each experiment, samples of AcLDL and OxLDL were extracted and analyzed by HPLC for comparison with cell extracts. AcLDL contained identical quantities of cholesterol and its esters to unmodified LDL, whereas OxLDL usually contained no detectable cholesteryl arachidonate or cholesteryl linoleate, small quantities of cholesteryl oleate and palmitate, and reduced quantities of unesterified cholesterol. These changes have been described in detail previously.20
The choice of an appropriate internal standard to correct for preparative losses from a mixture of several different extracted lipids is not simple, since the varying solubilities of the individual lipids can be reflected in the accuracies and efficiencies with which each is recovered during extraction and analysis. Since it was necessary to monitor the recovery of up to nine different lipids in a single extraction, the use of a cocktail of these individual radiolabeled standards was not practicable. Unlabeled cholesteryl heptadecanoate standard was chosen because it could be resolved from other naturally occurring esters on the HPLC chromatogram at 210 nm. The extraction yield of cholesteryl heptadecanoate was 97±5% (mean±SD, n=40), and extraction efficiencies of cholesterol, cholesteryl linoleate, 7-ketocholesterol, and oxidized ester standards, all tested under the extraction conditions described, were all approximately 100%.20
In practice, we noted that when added to a standard mixture of cholesteryl esters, cholesteryl heptadecanoate was extracted less consistently than all other esters. Thus, the use of this internal standard was not ideal, since internal standard correction of measured values generally increased the magnitude of the standard deviations without any significant impact on the mean values. For example, in one experiment the percent cholesterol efflux was calculated as 48.9±2.0% without correction and 47.2±8.0% after use of internal standard correction. All data presented are without internal standard correction; however, all of the studies described here have contained at least one replicate experiment in which internal standard recoveries have been applied, with the same results.
The protein content of LDL samples and cell lysates was measured using the bicinchoninic acid method (Sigma) using BSA as standard. For LDL, standards were prepared in water, and for cell lysates, standards were prepared in 0.2 mol/L NaOH, which corresponds to the solutions in which the two analytes were prepared. Samples were incubated for 60 minutes at 60°C before measurement of absorbance at 562 nm.
Cell viability was established by trypan blue staining at the end of the efflux incubation. Counting was performed twice in representative wells, and the result was averaged in each case. Both AcLDL-loaded cells and OxLDL-loaded cells showed a viability of 90%.
Oil Red O Staining
Macrophages were grown on sterile coverslips under the experimental conditions specified. They were then fixed in 6% paraformaldehyde and stained in sequence with osmium tetroxide, oil red O (in isopropanol), and hematoxylin. Cells were photographed under oil immersion with an Olympus BH-2 microscope at magnification ×125.
Cholesterol Efflux From AcLDL-Loaded Cells
Cells Incubated With AcLDL Accumulate Cholesterol and Cholesteryl Esters and Release Cholesterol in Response to Incubation With Apo A-I
Macrophages that had been incubated for 24 hours in the absence of LDL did not contain detectable quantities of cholesteryl esters (Fig 1A⇓), whereas cells incubated with AcLDL for 24 hours accumulated both cholesterol and cholesteryl esters (Fig 1B⇓). During subsequent incubations with efflux media, AcLDL-loaded cells released large quantities of free cholesterol to medium containing apo A-I and albumin (Fig 2A⇓), but only very small quantities to control medium containing albumin alone (Fig 2B⇓). No extracellular esterified cholesterol was detected under either condition.
The concentration dependence of apo A-I–mediated cholesterol efflux from AcLDL-loaded cells was tested. Maximal efflux was obtained with concentrations as low as 20 μg/mL (Fig 3⇓), in agreement with previous data.7 The same concentration dependence was seen with OxLDL-loaded cells (data not shown). Consequently, 25 μg/mL apo A-I was used in all subsequent experiments. Although reductions in both unesterified and esterified cholesterol were seen in the presence of apo A-I, there was a more marked loss of intracellular cholesteryl esters than unesterified cholesterol (Fig 3⇓). This is despite the appearance of only unesterified cholesterol in the efflux medium (Fig 2A⇑) and is consistent with previous studies.26
Kinetics of Apo A-I–Mediated Efflux of Cholesterol
The rate of apo A-I–mediated cholesterol efflux from AcLDL-loaded macrophages was measured. There was a progressive, time-dependent appearance of cholesterol in medium containing apo A-I over 24 hours (Fig 4A⇓). This displayed a nonlinear kinetic but did not appear to be complete by 24 hours. The release of cholesterol into the medium was paralleled by its loss from cells (Fig 4B⇓). The proportion of the intracellular cholesterol that was released to the apo A-I–containing medium after 24 hours varied minimally between experiments, approximating 50% (n=4, each performed in triplicate), despite wide variation in the amount of cholesterol accumulated before the incubation with efflux medium. Two of these experiments are shown in detail in Table 1⇓⇓ (chosen to highlight extremes of intracellular total cholesterol loading). This constant relation between the amount of sterol exported from the cells and the total intracellular cholesterol pool remained when susequently studied over an even wider range of loadings (cellular total cholesterol contents of 160 to 650 nmol/L cholesterol per milligram of cell protein). Preliminary experiments demonstrated similar decreases in intracellular total cholesterol after incubation with medium containing human HDL (data not shown). The proportion released in the presence of albumin alone was 5.3±2.7% (mean±SD).
Sterol Efflux From OxLDL-Loaded Cells
OxLDL-Loaded Cells Contain a Mixture of Oxidized and Unoxidized Lipids
The accumulation of lipids in OxLDL-loaded cells was established both quantitatively, by HPLC analysis of intracellular oxidized and unoxidized lipids, and visually, by oil red O staining. Cells loaded with OxLDL were found to contain unoxidized cholesterol, 7-ketocholesterol, cholesteryl linoleate hydroxide (this is a tentative identification as described in detail previously20 ), other oxidation products (yet to be characterized), and only small quantities of unoxidized cholesteryl esters (Fig 1C⇑ and 1D⇑). 7-Ketocholesterol has been identified previously in OxLDL-loaded macrophages,18 27 in OxLDL,27 28 in human atherosclerotic plaque,11 12 and recently, in rabbit and human foam cells,13 while cholesteryl linoleate hydroxide has been identified in LDL and in aortas from atherosclerotic rabbits.29 This composition is distinct from that of AcLDL-loaded cells, which contained none of the detectable oxidation products (data not shown) and much larger quantities of cholesteryl esters (Fig 1B⇑). The accumulation of the oxidized lipids during exposure of macrophages to OxLDL was progressive over 24 hours (Fig 5⇓).
In accordance with the clearly differing lipid profiles between AcLDL- and OxLDL-loaded cells, there was a difference in the morphology of cells stained with oil red O (Fig 6⇓). Whereas cells loaded with AcLDL were larger, more extensively spread, and contained large lipid droplets dispersed to the periphery of the cell, cells loaded with OxLDL were more rounded and contained smaller lipid droplets that were present throughout the cytoplasm and did not show peripheral distribution. The rounded morphology persisted for the duration of the incubation with OxLDL and partially resolved upon incubation with fresh medium without OxLDL. This resolution was equally evident when loaded cells were incubated in efflux medium containing albumin only as with medium containing apo A-I, suggesting that sterol efflux was not a necessary component of this change. As far as we are aware, this reversible change in morphology has not been described previously.
Release of Oxidized Cholesterol Compounds From OxLDL-Loaded Cells
7-Ketocholesterol was the only detectable oxidized lipid released from OxLDL-loaded cells into medium containing apo A-I (Fig 2D⇑). Cholesteryl linoleate hydroxide is neither an unoxidized cholesteryl ester nor a 7-ketocholesterol–containing compound. Furthermore, as it is quantitatively minor as an intracellular lipid and because its identification is not yet conclusive, it was not included in further calculations. The time-dependent release of 7-ketocholesterol into apo A-I–containing medium was much greater than into control media containing albumin alone (Fig 7⇓). The appearance of 7-ketocholesterol in the medium was compared with its intracellular loss, expressed as a proportion of the sum of intracellular and extracellular pools after 24 hours. Between 4.7% and 11.5% (mean of four experiments, each performed in triplicate, was 7.5±2.9%; mean±SD) of the total 7-ketocholesterol pool had redistributed to the extracellular medium in the presence of apo A-I by 24 hours. This compared with an average release of 7-ketocholesterol into medium containing albumin alone of 0.5±0.5% (range, 0% to 1.2%). Two representative experiments are described in detail in Table 1⇑⇓. The amount of 7-ketocholesterol released was much smaller in absolute and proportional terms than the release of cholesterol from AcLDL-loaded cells.
Release of Unoxidized Cholesterol From OxLDL-Loaded Cells
OxLDL-loaded cells released cholesterol into medium containing apo A-I simultaneously with the release of 7-ketocholesterol (Fig 7⇑). Whereas the release of 7-ketocholesterol was linear during the 24-hour period, the release of cholesterol was nonlinear, resembling the kinetics of cholesterol release from AcLDL-loaded cells (Fig 4A⇑). In addition, the absolute mass of cholesterol released was approximately 10 times that of 7-ketocholesterol released over the same period from the same cells (Fig 7⇑).
The cholesterol released from OxLDL-loaded cells incubated with apo A-I comprised 26.6±4.2% of the total (sum of cholesterol and cholesteryl ester) pool at 24 hours compared with a release of 1.7±0.7% with control (apo A-I–free) medium (Table 1⇑⇓). Because the cholesteryl ester pool in OxLDL-loaded cells is small, even if cholesteryl esters are removed from calculations of total cholesterol pool, the proportional efflux is insignificantly affected. There was therefore a clear impairment in the ability of OxLDL-loaded cells to release 7-ketocholesterol as compared with cholesterol (7.5±2.9% of 7-ketocholesterol pool versus 26.6±4.2% of total cholesterol pool).
OxLDL-loaded cells released less cholesterol than did AcLDL-loaded cells (26.6±4.2% versus 47.3±2.9% of total cholesterol pool). When the relation between cholesterol loading and efflux was examined further using cells loaded by incubation with 5 to 50 μg/mL OxLDL, leading to intracellular free cholesterol contents of 50 to 160 nmol/mg cell protein and total cholesterol ester levels of 0 to 45 nmol/mg cell protein, the amount of cholesterol exported from OxLDL-loaded cells remained directly proportional to the total intracellular cholesterol content over this entire range (data not shown).
OxLDL-loaded cells often contained at least as much free cholesterol as AcLDL-loaded cells but usually contained a smaller quantity of cholesteryl esters (Table 1⇑⇓). Thus, cholesterol efflux from OxLDL-loaded cells was less, both in absolute and relative terms, than that from AcLDL-loaded cells.
A direct comparison of cells containing similar free cholesterol contents (Ac2 versus Ox1 from Table 1⇑⇓) indicates that while the unesterified cholesterol content of the controls incubated in albumin-containing medium is almost identical in these two experiments, OxLDL-loaded cells released fivefold less mass of cholesterol to apo A-I–containing medium than AcLDL-loaded cells after 24 hours. As the intracellular unesterified cholesterol content after efflux was also similar in these two experiments, the difference in the mass of cholesterol released is probably due to the release of free cholesterol derived from cholesteryl esters in the AcLDL-loaded cells.
Proportion of Cholesterol Released to Apo A-I–Containing Medium and the Mass of Cholesterol in AcLDL-Loaded Cells
In each experiment, ratios of the molar mass of cholesterol or 7-ketocholesterol released into the medium containing apo A-I to the known molar mass of apo A-I in the medium were calculated. For these calculations it was presumed that all released cholesterol or 7-ketocholesterol was associated with apo A-I particles (Table 1⇑⇓). The ratio of cholesterol to apo A-I particles in the medium from cells loaded with AcLDL varied fourfold between experiments. In contrast, the proportion of the total cholesterol pool released to the extracellular space in these same experiments varied very little: from 44% to 50%, respectively. Thus, there is a notable consistency in the proportion of the total cholesterol pool released from cells loaded with AcLDL despite multifold variations in the mass of sterol accumulated before, and consequently released during, incubation with apo A-I.
The proportion of cholesterol released from OxLDL-loaded cells also varied minimally over a fourfold range of total cellular cholesterol loading (50 to 205 nmol/mg cell protein; see above), although data comparable to the greater amounts of cholesterol loading described for AcLDL-loaded cells are still required to fully test the independence of the proportion released from these cells and the degree of loading. In contrast, the proportion of the 7-ketocholesterol pool released to apo A-I varied approximately threefold (4.7% to 11.5%) between experiments.
Intracellular Sterol Distribution and Its Relation to Sterol Efflux
It was possible that a significant proportion of the accumulated intracellular cholesterol after AcLDL or OxLDL loading for 24 hours comprised undegraded lipoprotein still present in lysosomes.24 To allow time for more complete lysosomal hydrolysis of endocytosed lipids before the addition of efflux medium, macrophages to which efflux medium had been added immediately after a 24-hour loading period were directly compared with cells that had been loaded and chased by an overnight incubation (18 hours) in DMEM-albumin.
For AcLDL-loaded cells, at the time of addition of efflux medium there were only small differences in the quantity of cholesteryl esters between the two conditions (Table 2⇓, AcLDL versus AcLDL+O/N). Overnight equilibration slightly increased the proportion of total cell cholesterol that was esterified (67.2% versus 76.5%), indicating that during the chase period a small net increment in esterification occurred. This might be derived from residual lysosomal cholesterol that had not been exported by the end of the loading period or free cytoplasmic cholesterol that had not yet undergone esterification. However, the same proportion of total cellular cholesterol was released to apo A-I–containing medium by cells incubated with efflux medium immediately after loading and by cells equilibrated overnight before efflux (Fig 8A⇓).
The effect of overnight equilibration was also studied with OxLDL-loaded cells (Table 2⇑). As seen with AcLDL-loaded cells, there was no significant difference in the cholesteryl ester content of cells extracted immediately after loading and cells extracted after equilibration. In this case also, there were only very minor differences between the degree of cholesterol efflux (Fig 8B⇑) and also 7-ketocholesterol efflux (Fig 8C⇑) from cells incubated with efflux medium immediately after incubation with OxLDL or after overnight equilibration. Thus, it is unlikely that the difference in relative cholesterol efflux between AcLDL- and OxLDL-loaded cells can be attributed to insufficient time for equilibration of intracellular lipids during the 24-hour loading period.
To further eliminate the possibility that sterol deposits in OxLDL-loaded cells remained physically trapped in the lysosomes (as has been shown for oxidized apo B,24 ) macrophages were loaded with OxLDL, equilibrated overnight, and then fractionated and the subcellular distribution of cholesterol determined (Fig 9A⇓). The majority of the cholesterol content of nonloaded cells was distributed between a low-density fraction (d=1.031), which we have shown previously contains plasma membrane,24 an intermediate-density fraction (d=1.045) containing endosomes, and a lysosomal fraction (d=1.091). This is consistent with the predicted partitioning of cholesterol between all membrane-containing regions of the cell. There was no detectable shift in the density distribution of these organelles in fractions isolated from cells loaded with oxidized LDL compared with unloaded cells (data not shown; also see Reference 2424 ). The small amount of lysosomal enzyme activity (22%) present in the plasma membrane fractions of both nonloaded and OxLDL-loaded cells (Fig 9B⇓) may originate from lysosomes damaged during homogenization, since the latency of this fraction is low (15% compared with 90% for the lysosomal fraction), and we have shown previously that treatment of homogenates with detergent causes redistribution of lysosomal arylsulfatase to this region of the gradient. We cannot exclude the possibility that a subpopulation of low-density lysosomes exists; however, the size of this population is unaffected by OxLDL loading (Fig 9B⇓).
While the total cholesterol content of homogenates from cells loaded with OxLDL was higher than that of nonloaded cells, the relative distribution of cholesterol between the subcellular fractions followed an identical profile (Fig 9A⇑). Furthermore, 7-ketocholesterol subcellular distribution followed the same pattern. This is not compatible with a selective lysosomal trapping of OxLDL-derived cholesterol and 7-ketocholesterol suggested by others.30 Similarly, only a small proportion of the cholesteryl esters present in OxLDL-loaded cells were found in the lysosomes (Fig 9C⇑). The majority partitioned to the lowest density region of the gradient, consistent with a free cytosolic location similar to that seen in AcLDL-loaded cells (E. Mander, unpublished observation), implying that these were derived from ACAT-mediated esterification (see below). Taken together, these subcellular distributions of free and esterified sterols in OxLDL-loaded cells indicate that (1) free sterols are able to redistribute from the lysosomes into other cellular membranes, and (2) cholesterol esters are present in a membrane-free fraction, probably via ACAT (see also below). These observations demonstrate that sterols derived from OxLDL are not physically trapped within the lysosomes of primary macrophages.
Influence of Inhibition of ACAT on Cholesterol Efflux
Macrophages loaded with either AcLDL or OxLDL in the presence of an ACAT inhibitor (Sandoz 58-035) were directly compared with cells loaded in the absence of ACAT inhibitor. Cells incubated with AcLDL and Sandoz 58-035 for 24 hours contained similar total cholesterol contents but a much greater quantity of free cholesterol and a lesser quantity of esterified cholesterol than cells incubated with AcLDL alone. The total cholesterol content of the cells was not affected by the presence of ACAT inhibitor (Table 2⇑). Cells incubated with AcLDL in the presence of Sandoz 58-035 and harvested immediately after a 24-hour loading period contained only 18.4% of the ester content of cells incubated without the ACAT inhibitor. In a separate experiment using AcLDL under identical loading conditions (total cellular cholesterol esters, 239±10 nmol/mg cell protein; free cholesterol, 146±13 nmol/mg cell protein), the cholesteryl ester content of cells incubated continuously with Sandoz 58-035 was similarly reduced by 84% at the end of the AcLDL incubation. However, when these cells were washed and incubated for a further period without AcLDL in DMEM containing 100 μg/mL albumin plus inhibitor, the cholesteryl ester content was further reduced to 5% of uninhibited controls after 4 hours and to 0.5% after 28 hours. These data indicate that, as found in previous studies, this agent is a highly efficient inhibitor of ACAT.31 The incomplete loss of cholesteryl esters detected immediately after the loading incubation is likely to represent a residual pool of endosomal and lysosomal AcLDL whose degradation is substantially complete before 4 hours, consistent with previous literature.22
Despite the efficacy of Sandoz 58-035, the release of cholesterol from AcLDL-loaded cells was not decreased by the marked relative decrease in cellular content of esterified cholesterol (Fig 10A⇓). Indeed, there was a slightly greater release of cholesterol from AcLDL-loaded cells in the presence of inhibitor (48.9% compared with 41.5% without Sandoz 58-035; P<.01; Fig 10A⇓). In separate experiments, a 4-hour chase period in the continued presence of Sandoz 58-035 was included before the addition of efflux medium to allow time for more complete lysosomal hydrolysis of endocytosed AcLDL. Under such conditions (95% reduction in cholesteryl ester content), the efflux of cholesterol was also unaffected (50.8±1.8% compared with 46.7±6.5% efflux for control and inhibited cells, respectively). These data confirm that mobilization of cholesteryl esters is not rate limiting in cholesterol efflux from mouse macrophages.
Incubation of cells with OxLDL and Sandoz 58-035 reduced by 37% the already low relative content of intracellular cholesteryl esters, whereas free cholesterol increased by 16.5% (Table 2⇑). However, these esters declined to undetectable levels during a 4-hour chase in the continued presence of Sandoz 58-035. In contrast, use of the ACAT inhibitor had a major immediate impact on cellular free 7-ketocholesterol content, causing a 45% increase compared with uninhibited cells (Table 2⇑). Thus, a substantial proportion of intracellular 7-ketocholesterol is normally a substrate for esterification, consistent with its extensive extralysosomal distribution (Fig 9A⇑). However, as for AcLDL-loaded macrophages, the release of cholesterol (Fig 10B⇑) and 7-ketocholesterol (Fig 10C⇑) from OxLDL-loaded cells were unaffected by the presence of ACAT inhibitor.
Thus, reduction in the relative size of the cholesteryl ester pool in AcLDL-loaded cells toward that seen in OxLDL-loaded cells did not decrease the degree of cholesterol efflux from AcLDL-loaded cells to that seen from OxLDL-loaded cells. The difference in cholesterol release from AcLDL- and OxLDL-loaded cells cannot therefore be simply explained by the relative amounts of free cholesterol and cholesteryl ester in these cells. Similarly, the relatively smaller release of 7-ketocholesterol compared with cholesterol in OxLDL-loaded cells cannot be explained by the sizes of their free and esterified sterol pools. Moreover, 7-ketocholesterol appeared to be at least as accessible to esterification by ACAT as cholesterol in these cells.
Unexplained Losses of Sterol
In both cell models, comparisons were made between efflux in the presence of albumin and apo A-I with albumin alone to allow for changes in the cellular and medium contents of sterol with respect to time. The relative intracellular content of cholesterol and cholesteryl esters and oxidation products were very similar at the end of the initial loading phase and after incubation with DMEM-albumin (controls), as has been indicated in the 18-hour overnight incubations described in Table 2⇑. In both systems, there was a small loss and variation of total sterol per milligram of cell protein over time that was not accounted for even after total medium and cellular sterols were summed. This remained true when internal standard corrections were applied to the data. The proportional drop in total sterol varied between experiments but was always less than 20% of the total. As the reproducibility of our results for cellular and media analyses is generally satisfactory (SD for each experiment usually less than 10% of the mean), this is unlikely to be due to analytical error.
Two explanations are proposed. First, cholesterol and 7-ketocholesterol may be processed by the cells into other compounds not detected by our system. In support of this, we have preliminary evidence (A. Brown and I. Gelissen, personal communication; see also Table 2⇑) for the formation of 7-ketocholesteryl esters in OxLDL-loaded cells. Second, as the cells were noted to enlarge after changing medium after loading, it is possible that the protein per cell increased, thus reducing the cholesterol to cell protein ratio. This second explanation is unlikely, as we have documented a minor progressive reduction of total cell protein per well over 24 hours in this system (data not shown), and, in the absence of cell loss, an increase in protein per cell would cause a rise in protein per well.
These losses should not interfere with the internal consistency of our results because albumin controls at identical time points were used for all calculations and comparisons. With few exceptions (such as the cholesterol pool in Ox1 in Table 1⇑⇓), the sums of intracellular and extracellular sterol pools in albumin-incubated and apo A-I and albumin-only incubated cultures were the same. Where minor discrepancies did exist, they did not materially affect the interpretation of the data.
The relatively consistent proportions of cholesterol released from lipoprotein-loaded macrophages in this closed system appear to be controlled by cellular factors because the association of apo A-I with sterol is not saturated. OxLDL-loaded macrophages release 7-ketocholesterol to apo A-I much less efficiently than they release cholesterol. In addition, OxLDL-loaded cells release proportionally less cholesterol than do AcLDL-loaded cells. The difference in cholesterol release between the types of lipid-loaded cells is not explained simply by their relative contents of cholesteryl esters, since it persisted irrespective of whether (1) cellular free sterol content was manipulated through use of an ACAT inhibitor or (2) the comparison was made between cells on the basis of their free or total cholesterol contents. The independence of the rate of cholesterol export from intracellular ester hydrolysis is not surprising but rather consistent with previous reports demonstrating that increasing cellular free cholesterol content through ACAT inhibition was not inhibitory to efflux.32 33 34
Relevance of Model Systems Used
Acetylation and 24-hour copper oxidation are well-described LDL modifications that are known to result in scavenger receptor recognition of LDL and foam cell formation.16 Copper-mediated oxidation of LDL results in very extensive cholesteryl ester oxidation.20 Although LDL extracted from human atheroma in one study was only mildly oxidized,12 7-ketocholesterol was identified in the same LDL, yet 7-ketocholesterol is a product of quite extensive, not mild, LDL oxidation.20 27 28 It is thus difficult to know a priori which degree of in vitro LDL oxidation is most physiologically relevant. Indeed, substantial oxidation is required for protein modification of degree equal to AcLDL, as defined by its relative electrophoretic mobility (as in “Methods”).20 Thus, to compare unoxidized and oxidized LDL particles endocytosed by the same route, which is a primary aim of this work, 24-hour copper modification or other oxidation to a similar degree is appropriate for comparison with AcLDL.
We have demonstrated that macrophages incubated with AcLDL or OxLDL clearly become lipid loaded on the basis of demonstrated lipid accumulation and ability to release cholesterol to a suitable acceptor. The two foam cell models described in this study differ markedly in their cholesteryl ester content. It is clear from our lipid analyses of these cells that in the absence of external sources of fatty acyl chains, the lipid profile of LDL incubated with the cells dictates the nature of accumulated lipids. Thus, the minimal residual cholesteryl esters in the OxLDL at the time of incubation with cells correlates with the very small quantities of these esters in OxLDL-loaded cells, whereas AcLDL and AcLDL-loaded cells both contained large quantities of cholesteryl esters. Our analysis by HPLC demonstrated that AcLDL-loaded cells accumulate large quantities of all cholesteryl esters, not only cholesteryl oleate, consistent with reports by others.17 35 The relative proportion of cholesteryl esters to cholesterol increased during overnight equilibration without LDL and decreased profoundly with ACAT inhibition. Therefore, these cellular esters were principally derived from cellular reesterification rather than from accumulated, undegraded LDL lipid.
Apolipoproteins facilitate removal of cell cholesterol by phospholipids.36 37 38 39 Moreover, the interaction of HDL with cells may be mediated by apolipoprotein binding to specific HDL-binding proteins,40 41 42 and, as free apo A-I may be present in interstitial fluid,4 5 apo A-I may be a physiologically relevant cholesterol acceptor. For these reasons, we used free apo A-I as acceptor.
The release of cellular cholesterol to apolipoproteins may be simultaneous with release of phospholipid,7 43 whereas in phophatidylcholine monolayers, cholesterol desorption may be subsequent to phospholipid.44 Thus, it is probable that extracellular apo A-I cholesterol–containing complexes generated in these experiments contained cellular phospholipid. That clear mass transfer to apoA-I was demonstrable and sustainable in a closed system without LCAT is of interest, and was one reason for using purified apo A-I rather than HDL. This observation will permit selective later delineation of the role of LCAT and other agents in the promotion of sterol efflux from these lipid-loaded model cells.
Impaired Release of Sterols From OxLDL-Loaded Cells
The accumulation of cholesteryl ester lipid droplets in cells loaded with AcLDL involves seven stages: (1) lysosomal hydrolysis of cholesteryl esters and (2) transfer of cholesterol to the extralysosomal cytosol for reesterification by ACAT to form cholesteryl ester lipid droplets. Normal cholesterol efflux then requires (3) hydrolysis of these cholesteryl esters by nCEH (nonlysosomal neutral cholesteryl ester hydrolase) to release free cholesterol, (4) synthesis of phosphatidylcholine and sphingomyelin for the packaging of cholesterol, (5) delivery of cholesterol to the plasma membrane, (6) desorption from the plasma membrane across the unstirred water layer, and (7) the presence of an acceptor in the medium that will bind cholesterol.2 26 43 45 46
It is possible that oxysterols (such as 7-ketocholesterol) or oxidized protein moieties from the oxidized apo B of OxLDL may interfere with stages 1 through 5 described above. Such interference would be expected to affect the removal of both cholesterol and 7-ketocholesterol. Oxysterols can affect intracellular enzymes that may be involved in this sequence. For example, 7-ketocholesterol can promote activation of ACAT, whereas 7α- dihydroxycholesterol can inhibit ACAT.18 7-Ketocho- lesterol also has been demonstrated to inhibit hydroxy- methyl glutaryl–CoA reductase.47
Studies using radiolabeled cholesteryl oleate indicated that a maximum of 50 minutes is required to transport cholesterol from lysosomes to plasma membrane.48 However, there could be substantial differences in the rate of intracellular transport of LDL cholesterol in AcLDL- and OxLDL-loaded cells. OxLDL protein is known to be less readily hydrolyzed than AcLDL protein.49 50 In addition, we have shown using cell fractionation techniques24 that some OxLDL protein accumulates in the lysosome, whereas AcLDL protein does not. Thus, if the handling of protein and lipid components of modified LDL is similar, OxLDL lipids might be less readily hydrolyzed in the lysosome than those of AcLDL and may therefore experience delayed intracellular transport. While we have not directly demonstrated that the rates of lipid hydrolysis and export from the lysosome are identical in AcLDL- and OxLDL-loaded cells, we have shown that there is no major difference in these lipid-loaded cells in their subcellular distribution of sterols (and in particular no evidence of lysosomal enrichment with sterols) or their availability for esterification by ACAT (which is consistent with an extralysosomal location) within the time scale of the experiments described here.
We have additionally noted that the proportion of cholesterol that is subject to cellular esterification after uptake of OxLDL by macrophages is much lower than that seen in AcLDL-loaded cells. We believe that this is largely a consequence of the restricted availability of fatty acyl cosubstrates for ACAT in such cells, since supplementation of cells with fatty acyl albumin leads to an increased content of cholesteryl esters at the expense of free cholesterol (W. Jessup and E. Mander, unpublished observations). It should be remembered that the mass of cholesterol delivered to macrophages by AcLDL or OxLDL uptake greatly exceeds that delivered using the radiolabeled tracer amounts of free cholesterol often applied to study cellular esterification and therefore that it is perhaps not surprising that the ability of the cell to synthesize sterol esters under circumstances of relatively large sterol influx may be limited by cosubstrate supply.
Stages 6 and 7 could also be disturbed either by changes to the plasma membrane environment or by differential affinities for acceptor particles of different sterols (discussed below). The proportionately different efflux of cholesterol from OxLDL-loaded and AcLDL-loaded cells may in part be due to a less compressible plasma membrane with the former, which would render phospholipid displacement by apo A-I more difficult. For example, increasing the cholesterol content decreased apo A-I adsorption to phosphatidylcholine monolayers,51 and the enrichment of Fu5AH hepatoma cells with cholesterol reduced whole-cell fractional efflux.48 As seen in “Results,” the total cell unesterified cholesterol was in some experiments as great or greater in cells loaded with OxLDL than in cells loaded with AcLDL despite much lower quantities of esterified cholesterol. If the pool of 7-ketocholesterol is added to the total unesterified sterol pool, these cells could sometimes contain greater amounts of sterol in the cell membrane than AcLDL-loaded cells. Such an explanation would be consistent with a recently described “domain hypothesis” of efflux52 but presumes similar proportions of unesterified sterols are available to the plasma membrane in the two cell systems.
It is possible that esterified cholesterol is more easily mobilized for efflux than unesterified cholesterol either because of differing location or because of differing sensitivity to second-messenger systems triggered by extracellular acceptors.53 However our data with AcLDL-loaded cells and previous literature33 34 indicate that ACAT inhibition can promote efflux. This suggests that the difference in the release of cholesterol between AcLDL-loaded and OxLDL-loaded cells cannot be explained by an intrinsic difference between efflux of cholesterol derived from cellular unesterified cholesterol and that derived from cellular cholesteryl esters. It may be explained by a different, extralysosomal, intracellular location (perhaps cytosolic) of much cholesterol, esterified or unesterified, in AcLDL-loaded cells, as compared with OxLDL-loaded cells. However, the subcellular distribution of sterols in OxLDL-loaded cells does not suggest that selective sequestration of cholesterol or 7-ketocholesterol within the lysosomes occurs in these cells. This is in disagreement with a recent report in which unesterified cholesterol derived from OxLDL uptake by the macrophage cell line J774 A.1 was reported to be selectively trapped in a lysosome-rich fraction.30 The contrast between this and our observations may be based on differences in the analytical methods used to fractionate the cells and detect lipids and/or variations in the cholesterol metabolism of the primary macrophages (used here) and the J774 A.1 cell line. For example, in direct comparisons between uptake of AcLDL by these two cell types, we have found a major difference in the proportion of intracellular cholesterol that undergoes esterification. While in primary macrophages we routinely measure 60% to 75% of the total cellular cholesterol pool present as esters, for J774 A.1 cells loaded under identical conditions only 25% of the total pool was esterified, leaving 75% of the total cellular cholesterol as free sterol. J774 cells also selectively retain free cholesterol in a lysosomal fraction after uptake of cholesteryl ester droplets,54 suggesting that this phenomenon is peculiar to the type of cells used rather than to simultaneous presence of lipid oxidation products.
If HDL-binding proteins40 41 42 are important to the interaction of apo A-I with foam cells, disturbed expression of such proteins in OxLDL-loaded cells could explain decreased sterol efflux from these cells. However, as other data suggest apolipoprotein-mediated efflux involves amphipathic α-helical repeats in general rather than specific apolipoproteins and their receptors,52 55 56 57 the sterol and phospholipid content of the plasma membrane may be more important explanatory variables.
Selective Impairment of the Release of 7-Ketocholesterol
Although all of the above mechanisms may apply equally to 7-ketocholesterol and cholesterol, additional restrictions on the release of 7-ketocholesterol are evident. This is suggested by the 2- to 4-fold greater proportional release of cholesterol (10-fold greater by mass) compared with 7-ketocholesterol into apo A-I–containing medium from OxLDL-loaded cells.
This observation contrasts with a more rapid exchange of 7-ketocholesterol than cholesterol from lipid vesicles58 but is consistent with the slow release of 7β-hydroxycholesterol from Mycoplasma cells.59 The former vesicle system differed from our cell system in several important respects: (1) pure lipid vesicles differ from cell membranes that contain protein and a complex phospholipid mixture; (2) incorporation of individual oxysterols was used in the vesicle study, whereas 7-ketocholesterol was a relatively small contributor to the total sterol available in OxLDL foam cells; (3) it is possible that intracellular sterol transport plays a significant role in our more complex system. Although diffusion across the unstirred water layer is considered rate limiting in cholesterol efflux,2 45 60 it would appear difficult to attribute the difference between cholesterol and 7-ketocholesterol in our system to differences in water solubility and therefore ability to traverse the unstirred water layer, given these previous observations.
Two possible contributing factors to the differences between 7-ketocholesterol and cholesterol release from oxidized LDL-loaded cells are proposed: (1) apo A-I may be less able to bind 7-ketocholesterol than cholesterol, either directly or in the postulated apo A-I–phospholipid complex; (2) 7-ketocholesterol may experience strong phospholipid, cholesterol, or protein interactions in the plasma membrane, requiring a higher activation energy for release into the water layer than cholesterol.
These data suggest a novel mechanism for the atherogenicity of OxLDL, namely, that its accumulation in macrophages is associated with an impairment of the reverse cholesterol transport pathway. It appears that normal pathways used by cells to remove cholesterol are much less efficient at removing 7-ketocholesterol (and probably other oxysterols not detected by our HPLC system) and, to a lesser extent, cholesterol in cells that accumulate OxLDL. It is therefore possible that small quantities of OxLDL in atherosclerotic plaque may be very difficult for normal clearance mechanisms to remove (hence will tend to accumulate) and that these small quantities of accumulated oxysterols may exert major effects on the release of unoxidized cholesterol in vivo. Recent studies13 that indicate enrichment of several oxysterols (of which 7-ketocholesterol is a major component) in rabbit and human arterial macrophage foam cells support this proposal.
Dr Kritharides is supported by a postgraduate medical research scholarship from the National Health and Medical Research Council of Australia. E. Mander is supported by a postgraduate research scholarship from The Queen’s Trust. We gratefully acknowledge Prof P. Barter and Dr K. Rye (University of Adelaide) for providing us with purified apo A-I and for helpful discussions and Dr Andrew Brown and Ingrid Gelissen for helpful discussions and making data available. In addition, we are very grateful to Sandoz Pharmaceuticals for providing ACAT inhibitor S58-035 for use in this work.
Goldberg DI, Beltz WF, Pittman RC. Evaluation of pathways for the cellular uptake of high density lipoprotein cholesterol esters in rabbits. J Clin Invest. 1991;87:331-346.
Roheim PS, Dory L, Lefevre M, Sloop CH. Lipoproteins in interstitial fluid of dogs: implications for a role in reverse cholesterol transport. Eur Heart J. 1990;11E:225-229.
Lefevre M, Sloop CH, Roheim PS. Characterisation of dog prenodal peripheral lymph lipoproteins: evidence for the peripheral formation of lipoprotein-unassociated apo A-I with slow pre-β electrophoretic mobility. J Lipid Res. 1988;29:1139-1148.
Hara H, Yokoyama S. Interaction of free apolipoproteins with macrophages. J Biol Chem. 1991;266:3080-3086.
Steinbrecher UP, Lougheed M. Scavenger receptor-independent stimulation of cholesterol esterification in macrophages by low density lipoprotein extracted from human aortic intima. Arterioscler Thromb. 1992;12:608-625.
Mattsson L. Studies of Arterial Derived Foam Cells in Atherogenesis. PhD thesis, 1994, University of Goteborg, Sweden.
Rosenfeld ME, Khoo JC, Miller E, Parthasarathy S, Palinsky W, Witzum JL. Macrophage-derived foam cells, freshly isolated from rabbit atherosclerotic lesion degrade modified lipoproteins, promote oxidation of low density lipoproteins, and contain oxidation specific lipid-protein adducts. J Clin Invest. 1991;87:90-99.
Zhang H, Basra HJK, Steinbrecher UP. Effects of oxidatively modified LDL on cholesterol esterification in cultured macrophages. J Lipid Res. 1990;31:1361-1369.
Jialal I, Chait AJ. Differences in the metabolism of oxidatively modified low density lipoprotein and acetylated low density lipoprotein by human endothelial cells: inhibition of cholesterol esterification by oxidatively modified low density lipoprotein. J Lipid Res. 1989;30:1561-1568.
Goldstein JL, Ho YK, Basu SK, Brown MS. Binding site on macrophages that mediates uptake and degradation of acetylated low density lipoprotein producing massive cholesterol deposition. Proc Natl Acad Sci U S A. 1979;76:333-337.
Brown MS, Goldstein JL, Krieger M, Ho YK, Anderson RGW. Reversible accumulation of cholesteryl esters in macrophages incubated with acetylated lipoproteins. J Cell Biol. 1979;82:597-613.
Ho YK, Brown MS, Goldstein JL. Hydrolysis and excretion of cytoplasmic cholesteryl esters by macrophages: stimulation by high density lipoprotein and other agents. J Lipid Res. 1980;21:391-398.
Jialal I, Freeman DA, Grundy SM. Varying susceptibilities of different low density lipoproteins to oxidative modification. Arterioscler Thromb. 1991;11:482-488.
Bhadra S, Arshad MAQ, Rymaszewski Z, Norman E, Wherley R, Subbiah MTR. Oxidation of cholesterol moiety of low density lipoprotein in the presence of human endothelial cells or Cu ions: identification of major products and their effects. Biochem Biophys Res Commun. 1991;176:431-440.
Maor I, Aviram M. Oxidized low-density lipoprotein leads to macrophage accumulation of unesterified cholesterol as a result of lysosomal trapping of lipoprotein hydrolysed cholesterol ester. J Lipid Res. 1994;35:803-819.
Ross AC, Go KJ, Heider JG, Rothblat GH. Selective inhibition of acyl coenzyme A:cholesterol acyltransferase by compound 58-035. J Biol Chem. 1984;259:815-819.
Schmitz G, Robenek H, Beuck M, Krause R, Schurek A, Niemann R. Ca antagonists and ACAT inhibitors promote cholesterol efflux from macrophages by different mechanisms, I: characterization of cellular lipid metabolism. Arteriosclerosis. 1988;8:46-56.
Bernard DW, Rodriguez A, Rothblat GH, Glick JM. Influence of high density lipoprotein on esterified cholesterol stores in macrophages and hepatoma cells. Arteriosclerosis. 1990;10:135-144.
Bierman EL, Oram J, Mendez A. HDL receptor-mediated cholesterol efflux from cells and its regulation. In: Malmendier CL, ed. Hypercholesterolemia, Hypocholesterolemia, Hypertriglyceridemia. New York, NY: Plenum Press; 1990:81-83.
Mahlberg FH, Glick JM, Lund-Katz S, Rothblat GH. Influence of apolipoproteins A-I, A-II and C on the metabolism of membrane and lysosomal cholesterol in macrophages. J Biol Chem. 1991;266:19930-19937.
Mahlberg FH, Rothblat GH. Cellular cholesterol efflux: role of cell membrane kinetic pools and interaction with apolipoproteins A-I, A-II, and Cs. J Biol Chem. 1992;267:4541-4550.
Graham DL, Oram JF. Identification and characterisation of a high density lipoprotein-binding protein in cell membranes by ligand blotting. J Biol Chem. 1987;262:7439-7442.
Fidge NH, Nestel PJ. Identification of apolipoproteins involved in the interaction of human high density lipoprotein 3 with receptors on cultured cells. J Biol Chem. 1985;260:3570-3575.
Oram JF, Brinton EA, Bierman EL. Regulation of high density lipoprotein receptor activity in cultured human skin fibroblasts and human arterial smooth muscle cells. J Clin Invest. 1983;72:1611-1621.
Bielicki JK, Johnson WJ, Weinberg RB, Glick JM, Rothblat GH. Efflux of lipid from fibroblasts to apolipoproteins: dependence on elevated levels of cellular unesterified cholesterol. J Lipid Res. 1992;33:1699-1709.
Schmitz G, Beuck M, Fischer H, Nowicka G, Robeneck H. Regulation of phospholipid biosynthesis during cholesterol influx and high density lipoprotein-mediated cholesterol efflux in macrophages. J Lipid Res. 1990;31:1741-1752.
Kandutsch AA, Chen HW. Inhibition of sterol synthesis in cultured mouse cells by 7α-hydroxycholesterol, 7β-hydroxycholesterol and 7-ketocholesterol. J Biol Chem. 1973;248:8408-8417.
Johnson WJ, Chacko GK, Phillips MC, Rothblat GH. The efflux of lysosomal cholesterol from cells. J Biol Chem. 1990;265:5546-5553.
Lougheed M, Zhang H, Steinbrecher UP. Oxidized low density lipoprotein is resistant to cathepsins and accumulates within macrophages. J Biol Chem. 1991;266:14519-14525.
Rothblat GH, Mahlberg FH, Johnson WJ, Phillips MC. Apolipoproteins, membrane cholesterol domains, and the regulation of cholesterol efflux. J Lipid Res. 1992;33:1091-1097.
Mendez AJ, Oram JF, Bierman EL. Protein kinase C as a mediator of high density lipoprotein receptor-dependent efflux of intracellular cholesterol. J Biol Chem. 1991;266:10104-10111.
Tangirala RK, Mahlberg FH, Glick JM, Jerome WG, Rothblat GH. Lysosomal accumulation of unesterified cholesterol in model macrophage foam cells. J Biol Chem. 1993;268:9653-9660.
Leblond L, Marcel Y. The amphipathic α-helical repeats of apolipoprotein A-I are responsible for binding of high density lipoproteins to HepG2 cells. J Biol Chem. 1991;266:6058-6067.
Jonas A. Lipid-binding properties of apolipoproteins. In: Rosseneu M, ed. Structure and Function of Apolipoproteins. Ann Arbor, Mich: CRC Press; 1992:218-250.
Phillips MC. Interactions of apolipoproteins at interfaces. In: Rosseneu M, ed. Structure and Function of Apolipoproteins. Ann Arbor, Mich: CRC Press; 1992:186-217.