Impairment of Macrophage Cholesterol Efflux by Cholesterol Hydroperoxide Trafficking: Implications for Atherogenesis Under Oxidative Stress

Abstract

Objective

Oxidative stress associated with cardiovascular disease can produce various oxidized lipids, including cholesterol oxides such as 7-hydroperoxide (7-OOH), 7-hydroxide (7-OH), and 7-ketone (7=O). Unlike 7=O and 7-OH, 7-OOH is redox-active, giving rise to the others via potentially toxic free radical reactions. We tested the novel hypothesis that under oxidative stress conditions, steroidogenic acute regulatory (StAR) family proteins not only deliver cholesterol to/into mitochondria of vascular macrophages, but also 7-OOH, which induces peroxidative damage that impairs early stage reverse cholesterol transport.

Approach and Results

Stimulation of human monocyte-derived THP-1 macrophages with dibutyryl-cAMP resulted in substantial upregulation of StarD1 and ABCA1. siRNA-induced StarD1 knockdown (kd) prior to stimulation had no effect on StarD4, but reduced ABCA1 upregulation, linking the latter to StarD1 functionality. Mitochondria in stimulated StarD1-kd cells internalized 7-OOH slower than non-stimulated controls and underwent less 7-OOH-induced lipid peroxidation and membrane depolarization, as probed with C11-BODIPY and JC-1, respectively. Major functional consequences of 7-OOH exposure were (i) loss of mitochondrial CYP27A1 activity, (ii) reduced 27-hydroxycholesterol (27-OH) output, and (iii) down-regulation of cholesterol-exporting ABCA1 and ABCG1. Consistently, 7-OOH-challenged macrophages exported less cholesterol to apoA-I or HDL than did non-challenged controls. StarD1-mediated 7-OOH transport was also found to be highly cytotoxic, whereas 7=O and 7-OH were minimally toxic.

Conclusions

This study describes a previously unrecognized mechanism by which macrophage cholesterol efflux can be incapacitated under oxidative stress-linked disorders such as chronic obesity and hypertension. Our findings provide new insights into the role of macrophage redox damage/dysfunction in atherogenesis.

Introduction

Vascular macrophages express scavenger receptors such as CD36 and SR-BI, which can bind and internalize oxidatively modified low density lipoprotein (oxLDL) arising under oxidative stress conditions associated with chronic obesity, hypertension, and atherosclerosis.^1–4 Unlike expression of normal LDL receptors, that of scavenger counterparts is not regulated by sterol negative feedback. Consequently, macrophages may accumulate large amounts of cholesteryl esters, non-esterified cholesterol, and other lipids from oxLDL, potentially leading to atherogenic “foam cell” formation.^3–5 Cholesterol efflux is a key early step in reverse cholesterol transport (RCT) whereby macrophages attempt to maintain cholesterol homeostasis by exporting the sterol when it reaches excessive levels.^6–8 These cells express a number of proteins besides CD36 and SR-BI that play important roles in inward and outward cholesterol trafficking. Of special interest are proteins of the steroidogenic acute regulatory (StAR) family, which bind and transport preexisting or incoming cholesterol, and deliver it to/into mitochondria for conversion to 27-hydroxycholesterol (27-OH) by 27-hydroxylase (CYP27A1) on the inner membrane.^9–12 27-OH is a prominent agonist of the liver-X receptor transcription factors (LXRα/β), which control expression of the ATP-binding cassette (ABC) transporters ABCA1 and ABCG1.^13–16 These plasma membrane proteins are known to play key roles in excess cholesterol export to extracellular acceptors, most notably apolipoprotein A-I (apoA-I) and high density lipoprotein (HDL), the former mainly recognizing ABCA1 and the latter ABCG1.^15,16 It is important to point out that a backup mode of cholesterol export exists whereby 27-OH itself and/or an acid derivative is released to an acceptor such as HDL when the intracellular sterol load exceeds the capacity of ABCA1/G1.^12,17 Direct export of 27-OH might become more important if ABCA1/G1-mediated cholesterol efflux is incapacitated due to genetic mutations or protein damage/dysfunction.

In addition to oxidized phospholipids and a modified apoB-100 protein component, oxLDL contains several esterified and non-esterified oxysterols, including 7α- and 7β-hydroperoxycholesterol (7α/7β-OOH), 7-ketocholesterol (7=O), 7α- and 7β-hydroxycholesterol (7α/7β-OH), and 5-6-epoxycholesterol (5-O-6).^18–23 Some of these oxysterols can be generated enzymatically.²¹ However, most of them derive mainly from free radical reactions triggered by reactive oxidants in the LDL lipid milieu, 7α/7β-OOH being generated at early stages and the others as secondary products.^18–21 Unlike the other cholesterol oxides mentioned, 7α- and 7β-OOH are redox-active and can undergo iron- or copper-catalyzed one-electron reduction to free radical species which propagate damaging chain peroxidation reactions in lipid compartments.^21,24 In the process, 7α/7β-OOH gives rise to 7α/7β-OH, 7=O, and 5-O-6.^20,21,24 Despite their relative instability, significant levels of 7α/7β-OOH (particularly 7β-OOH) have been detected in atherosclerotic plaques^18,20 and in plasma and tissue samples from animals or humans chronically exposed to increased oxidative pressure.^25,26 The substantially greater amounts of 7=O and 7α/7β-OH found in plaques^21–23 are attributed mainly to one-electron reductive turnover of 7α/7β-OOH,^20,21,24 although this has not been specifically assessed for the sub-endothelial space, where most of the LDL oxidation is known to occur.³ Based on this information and our recent evidence that the StAR homologue, StarD4, transports 7α-OOH to isolated mitochondria with deleterious effects,²⁷ we present the following hypothesis: Under pathophysiological conditions associated with oxidative stress, oxLDL-derived 7-OOHs will be caught up in StAR protein-mediated cholesterol trafficking to/into mitochondria in vascular macrophages, thereby causing site-specific structural/functional damage which impairs RCT at its early stages. In a brief recent report,²⁸ we provided the first supporting evidence for this hypothesis by showing that 7α-OOH uptake by mitochondria of cAMP-activated mouse RAW264.7 macrophages was StarD1-dependent and induced lipid peroxidation, membrane depolarization, and cell killing. For greater cardiovascular relevance, we have extended these findings to human THP-1 monocyte-derived macrophages in the present study and have shown that 7α-OOH inactivation of CYP27A1, with reduced 27-OH synthesis and ABCA1/G1 expression, significantly retards cholesterol efflux to extracellular apoA-I or HDL.

Materials and Methods

Material and Methods are available in the online-only Data Supplement.

Results

Protein expression in Bu₂cAMP-stimulated THP-1 macrophages: effects of StarD1 knockdown

In an initial experiment, fully differentiated THP-1 macrophages were analyzed for StarD1, StarD4, CYP27A1, and ABCA1 protein levels during incubation with Bu₂cAMP. As shown by the Western blots in Figure 1A, all four proteins were detected in cells after 4 h incubation in medium lacking Bu₂cAMP; the level in each case was not significantly different from that observed before incubation (not shown). However, when Bu₂cAMP (200 μM) was present during incubation, the StarD1 and ABCA1 levels at 4 h were ~2.5-fold and ~2-fold greater, respectively, than their control levels (Figure 1A). In contrast, there was no significant change in StarD4 level compared with that in the 4 h control and only a small (probably insignificant) increase in CYP27A1.

Figure 1.

Effect of Bu₂cAMP stimulation and StarD1 knockdown on protein expression in THP-1 macrophages. (A) THP-1 monocytes were differentiated to macrophages by incubating with 150 nM PMA in 10% FBS/RPMI medium for 3 days. The macrophages at ~70% confluency were switched to 1% FBS/RPMI medium and incubated in the absence (not-stimulated, NS) or presence (stimulated, S) of 0.2 mM Bu₂cAMP for 4 h. Cells were then recovered, lysed, and subjected to Western blot analysis for StarD1, StarD4, CYP27A1, ABCA1, and β-actin, the latter serving as a loading standard. (B) StarD1-knockdown cells (kd) and scrambled controls (Scr) were either stimulated (S) or not (NS), then Western analyzed for the indicated proteins. Protein load per lane in (A) and (B): 50 μg. Numbers below lanes represent densitometrically determined band intensities normalized to β-actin and relative to values in non-stimulated wild type cells. Also shown is effect of StarD1 knockdown on 7α-OOH uptake by THP-1 macrophages. StarD1-kd cells and Scr controls in serum-free RPMI medium were either stimulated with 0.2 mM Bu₂cAMP for 4 h or not stimulated (Scr-only) and then treated with POPC/Ch/[¹⁴C]7α-OOH (2:1:1 by mol) SUVs. After incubation at 37 °C for the indicated times, radioactivity in whole cells (C) and in mitochondrial fractions (D) isolated by differential centrifugation of cell homogenates was measured by scintillation counting. Plotted values in (C) and (D) are means ± SE (n=3 in each case).

An siRNA approach was used to deplete StarD1 and assess the effects of this on expression of the other selected proteins. As shown in Figure 1B (columns 3 and 4), StarD1 knockdown by ~60% relative to a scrambled vector control not only blunted StarD1 upregulation by Bu₂cAMP (~70% reduction), but also ABCA1 upregulation (~25% reduction). There was no change in StarD4 and a small increase (~15%) in CYP27A1, which is regarded as insignificant (Figure 1B).

StarD1-dependent 7α-OOH uptake by THP-1 macrophages: whole cell vs. mitochondrial fraction

We found that cholesterol uptake by whole macrophages as well as the mitochondrial fraction was substantially increased after Bu₂cAMP stimulation (results not shown) and similar results were obtained for 7α-OOH. As shown in Figure 1C, stimulated macrophages that had been transfected with a scrambled siRNA construct (S/Scr) took up SUV-borne [¹⁴C]7α-OOH more rapidly than non-stimulated controls (NS/Scr), resulting in ~70% more radioactivity in the former after 5 h of incubation. A more rapid uptake of [¹⁴C]7α-OOH after cell stimulation was also observed when mitochondrial fractions were examined, the S/Scr level being ~80% greater than the NS/Scr level after 5 h of incubation (Figure 1D). However, in stimulated cells with StarD1 knockdown (S/kd), [¹⁴C]7α-OOH uptake was substantially reduced in the total cellular compartment (Figure 1C) as well as the mitochondrial fraction (Figure 1D). The S/kd uptake values approached those of the NS/Scr samples in each case. These results clearly established that StarD1 was required for Bu₂cAMP-enhanced 7α-OOH uptake and delivery to mitochondria, just as it is for cholesterol.^9–12 Of added importance is our finding that the uptake specific radioactivity (CPM/μg protein) for the mitochondrial fraction (Figure 1D) was much greater than that for whole cells, e.g. nearly 3-times greater in S/Scr samples after 5 h (Figure 1C). Since mitochondria comprise on average only 20–25% of cellular mass,²⁹ this result highlights the ability of these organelles to selectively import 7α-OOH (as they do cholesterol) via StarD1.

7α-OOH-induced lipid peroxidation and ΔΨ_m disruption in macrophage mitochondria: role of StarD1

C11-BODIPY is a membrane-localizing ratiometric probe that can be used to detect free radical-mediated lipid peroxidation occurring in its membrane surroundings.³⁰ We used this fluorophore to assess the extent to which 7α-OOH redox activity would induce damaging lipid peroxidation in THP-1 mitochondria. As shown in Figure 2A, Bu₂cAMP-stimulated StarD1-kd macrophages exhibited substantially less green than red fluorescence after SUV-7α-OOH exposure than scrambled controls, indicating more extensive probe oxidation in the latter. The more extensive peroxidation in control cells, which was virtually identical to that in wild type cells (not shown), was localized mainly in punctate perinuclear zones (Figure 2A). These “hot spots” were most likely mitochondria, in keeping with the data shown in Figure 2.

Figure 2.

Mitochondrial lipid peroxidation and loss of membrane potential in 7α-OOH-challenged THP-1 macrophages. (A) Bu₂cAMP-stimulated StarD1-kd cells and scrambled construct controls (Scr) in serum-free RPMI medium were incubated for 4 h with SUVs lacking or containing 7α-OOH, the initial concentration of which was 50 μM in bulk phase. Following incubation, the cells were washed, treated with 2 μM C11-BODIPY for 30 min, and examined by confocal fluorescence microscopy. Integrated green/red ratios were as follows: -7α-OOH: 0.07±0.01 (kd), 0.07±0.02 (Scr); +7α-OOH: 0.28±0.03 (kd), 0.36±0.05 (Scr). Bar: 25 μm. (B) Stimulated (S) and non-stimulated (NS) wild type cells in a 96-well plate were incubated with SUV-7α-OOH in increasing concentrations up to 100 μM for 7 h, Cells were then washed, treated for 30 min with JC-1 (5 μg/ml), washed again, and examined using a plate reader with fluorescence detector. RFI denotes fluorescence intensity ratio: 595 nm (red)/535 nm (green). Data notations: S (Δ); NS (○). The inset shows RFI of stimulated StarD1-kd cells and Scr controls after incubation with 100 μM SUV-7α-OOH for 4 h. Plotted values in (B) are means ± SE (n=3). *P<0.01

We used the ratiometric fluorophore JC-1 to probe for mitochondrial membrane potential (ΔΨ_m) in stimulated vs. non-stimulated THP-1 macrophages and how this might change under a 7α-OOH challenge. As anticipated from the 7α-OOH-induced membrane damage observed in Figure 2A, JC-1-assessed ΔΨ_m (represented as red/green fluorescence intensity ratio (RFI)), decreased more rapidly with increasing 7α-OOH concentration in stimulated cells than in non-stimulated controls (Figure 2B).

For example, the RFI value at 30 μM 7α-OOH was ~45% lower than that of control cells. Moreover, stimulated StarD1-kd cells exhibited a substantially greater ΔΨ_m than scrambled controls after 4 h incubation in the presence of 100 μM SUV-7α-OOH (Figure 2B inset). These results add support to our hypothesis that StarD1 plays a crucial role not only in cholesterol uptake by mitochondria, but also 7α-OOH uptake, the latter triggering free radical-mediated peroxidative damage resulting in membrane depolarization and metabolic dysfunction.

Cytotoxic effects of 7α-OOH compared with other cholesterol-7-oxide species

Various studies have shown that oxysterols such as 7=O and 7-OH, which exist in oxLDL and atherosclerotic lesions,^20–22 are toxic to vascular endothelial cells and macrophages.^19,23 Redox-inert 7=O and 7-OH are produced by redox-turnover of 7-OOH.^20,24 However, these three different cholesterol-7-oxides have not been compared for toxicity in vascular macrophages. To examine this toxicity, we used ethanol as a delivery vehicle instead of liposomes to ensure rapid and consistent cellular uptake of the three 7-oxides. Using stimulated cells, we found that 7α-OOH was much more cytotoxic than 7=O (LD₅₀ ~15 μM vs. ~120 μM), whereas 7α-OH was non-toxic, even at the highest concentration tested (Figure 3A). 7α-OOH’s high lethality is attributed to its robust redox activity (Figure 3A), while relatively weak 7=O may operate via some other toxic mechanism(s).^22,23 The results in Figure 3A clearly demonstrate for the first time that 7α-OOH is exceedingly deleterious to macrophages compared with its redox products (7-OH, 7=O), which have been more widely studied with regard to cytotoxicity.²³

Figure 3.

Cytotoxic effects of 7α-OOH compared with other cholesterol-7-oxides. Separate stock solutions of 7α-OOH, 7=O, and 7α-OH in ethanol (each at ~10 mM) were prepared immediately before experimental use. Bu₂cAMP-stimulated macrophages were incubated with each 7-oxide in increasing concentrations up to 100 μM for 12 h, after which cell viability was determined by MTT assay (A). Immediately after addition to the cell medium, each 7-oxide was rapidly dispersed by swirling to ensure uniform and reproducible contact with cells. The concentration of ethanol at the highest concentration of 7-oxide applied to cells was <1% (v/v). Data notations: (○) 7α-OOH; (△) 7=O; (□) 7α-OH. Plotted values are means ± SD (n=4). Also represented is StarD1 involvement in 7α-OOH cytotoxicity. StarD1-kd macrophages (○) and scrambled construct controls (△) were stimulated with 0.2 mM Bu₂cAMP for 4 h, then exposed to SUV-7α-OOH in increasing concentrations up to 100 μM (B) or t-BuOOH in increasing concentrations up to 0.4 mM (C). After 24 h of incubation at 37 °C, cell viability was determined by MTT assay. Plotted values in (B) and (C) are means ± SE (n=3).

Role of StarD1 in 7α-OOH cytotoxicity

As anticipated from the results in Figures 2 and 3, exposing Bu₂cAMP-stimulated macrophages to SUV-7α-OOH in increasing concentrations up to 100 μM for 24 h resulted in a progressive loss of viability (Figure S1A). In contrast, non-stimulated cells, which expressed far less StarD1 than stimulated counterparts (Figure 1), were less sensitive to 7α-OOH toxicity, the LD₃₀ values being ~50 μM and ~18 μM, respectively. We then asked how cell stimulation would affect sensitivity toward another type of redox-active hydroperoxide, viz. tert-butyl hydroperoxide (t-BuOOH). As shown in Figure S1B, t-BuOOH, which did not require a liposome vehicle, also killed cells in a concentration-dependent fashion; however, there was no difference in sensitivity between stimulated and non-stimulated cells, the LD₅₀ for both being ~170 μM for the indicated incubation time. StarD1 involvement in 7α-OOH-induced cell killing was confirmed by showing (Figure 3B) that stimulated StarD1-kd cells were substantially more resistant to the hydroperoxide than scrambled construct controls (LD₂₅ ~90 μM vs. 40 μM), the latter responding identically to wild type cells (not shown). On the other hand, there was no difference between StarD1-kd and control cells in sensitivity to t-BuOOH (Figure 3C), which agrees with the stimulated vs. non-stimulated results in Figure S1B. Because StAR proteins have a high binding specificity for sterols, t-BuOOH would not have been well recognized or transported by StarD1, thus accounting for the non-effect of StarD1-kd in this case.

Effects of 7α-OOH on 27-hydroxycholesterol biosynthesis

In subsequent experiments, we asked whether mitochondrial peroxidative damage due to StarD1-mediated 7α-OOH delivery would impair 27-OH biosynthesis, 27-OH being an agonist of LXR, which regulates ABCA1 and ABCG1 expression.^11, We measured both mitochondrial CYP27A1 activity and total 27-OH output for stimulated THP-1 macrophages and how each of these was affected by 7α-OOH exposure. As shown in Figure 4A, CYP27A1 activity was reduced by ~40% after cells were incubated with 20 μM SUV-7α-OOH for 10 h. The viable cell count at this point was at least 90%. Overall cellular output of 27-OH was likewise reduced, and in a starting 7α-OOH concentration-dependent fashion. Thus, at 10 μM SUV-7α-OOH, 27-OH yield dropped by ~20%, and at twice this concentration, it dropped by ~40% relative to the control yield (Figure 4B). The decreases in CYP27A1 activity and 27-OH yield are attributed to peroxide-induced mitochondrial free radical damage (cf. Figure 2), resulting in severe metabolic dysfunction.

Figure 4.

Impact of 7α-OOH uptake on CYP27A1 activity and 27-hydroxycholesterol yield. (A) Bu₂cAMP-stimulated THP-1 macrophages were incubated with 0 μM or 20 μM SUV-7α-OOH for 10 h, then washed, recovered and homogenized, after which mitochondria were isolated by differential centrifugation. After protein determination, CYP27A1 activity was measured by incubating the mitochondria with [¹⁴C]cholesterol in Me-β-CD vehicle for 90 min, followed by lipid extraction and HPTLC-PI analysis of [¹⁴C]27-OH. (B) Stimulated macrophages were incubated with 0 μM, 10 μM, or 20 μM SUV-7α-OOH for 10 h, followed by washing and exposure to [¹⁴C]cholesterol in Me-β-CD for 12 h. Lipid extracts from samples of the medium were analyzed for [¹⁴C]27-OH by HPTLC-PI. Each of the sample bands shown represents 10⁵ cells. Plotted values in panels (A) and (B) are means ± SE (n=3); *P < 0.005 compared with 0 μM 7α-OOH.

Diminished StarD1, ABCA1 and ABCG1 expression in cells exposed to 7α-OOH

As a direct follow-up to the results in Figures 2–4, we asked how 7α-OOH exposure might affect StarD1 and ABCA1 expression in THP-1 macrophages. As shown by the Western blots in Figure 5A, the level of StarD1 protein in non-stimulated cells did not change during 6 h of incubation with 100 μM SUV-7α-OOH, whereas in stimulated cells, it decreased by ~48% relative to the non-challenged control. As anticipated from the results in Figure 4, ABCA1 protein also underwent a steady decline in stimulated cells, reaching ~50% of its control level after 6 h. Parallel RT-PCR analysis revealed that the steady state level of ABCA1 mRNA dropped by ~80% after 6 h of cell exposure to 100 μM SUV-7α-OOH (Figure 5B). The observed decrease in ABCA1 protein as well as mRNA implies that transcription was impaired by 7α-OOH, which is consistent with the observed 27-OH shortfall (Figure 4B). Some oxidative modification of ABCA1 protein with possible loss of function and/or antibody recognition may also have occurred, and cannot be ruled out at present. Western analysis indicated that the level of ABCG1 protein in stimulated cells also fell during SUV-7α-OOH treatment, reaching ~30% of the control level after 6 h (Figure 5A). As in the case of ABCA1, the ABCG1 decline is ascribed primarily to the reduced 27-OH output.

Figure 5.

Effect of 7α-OOH on StarD1, ABCA1, and ABCG1 expression. (A) Wild type THP-1 macrophages at ~60% confluency in serum-free RPMI medium were either not stimulated or stimulated with 0.2 mM Bu₂cAMP for 10 h, then washed, overlaid with fresh medium, and either analyzed before (0 h) or after incubating with POPC/Ch/7α-OOH (2:1:1 by mol) SUVs for 2, 4, and 6 h, the starting overall concentration of 7α-OOH being 100 μM in each case. At the indicated times, cells were washed, lysed, and samples of pre-determined total protein were subjected to Western analysis for StarD1, ABCA1, and ABCG1. Numbers below protein bands in the NS and S groups indicate band intensities relative to 0-time NS and normalized to β-actin. (B) RT-PCR-assessed ABCA1 mRNA levels in stimulated cells that had been incubated with 100 μM SUV-7α-OOH for 0, 2, 4, and 6 h; SUV composition was as described in panel A. Numbers below bands indicate band intensities relative to 0-time and normalized to GAPDH. Data in (A) and (B) are from one experiment in each case, which is representative of at least three with similar results.

Negative effect of 7α-OOH on cholesterol export

Of special interest at this point was how early stage RCT in macrophages would be affected by 7α-OOH trafficking. Such interest would address the functional significance of the findings in Figures 2–6. THP-1 cells pre-loaded with [³H]cholesterol were stimulated and exposed to SUVs either lacking or containing 7α-OOH, after which ABCA1-mediated [³H]cholesterol export to apoA-I in the medium was assessed. As shown in Figure 6A, [³H]cholesterol efflux at 6 h was significantly reduced in stimulated cells that had been exposed to 7α-OOH, whereas no difference was observed in non-stimulated controls. The diminished efflux reflects mitochondrial injury inflicted by 7α-OOH that was imported by upregulated StarD1. A small decrease in [³H]cholesterol efflux was also observed after 3 h of exposure to 7α-OOH, but it proved to be insignificant (results not shown). We also determined how 7α-OOH import would affect ABCG1-mediated [³H]cholesterol efflux to HDL in the medium. As shown in Figure 6B, [³H]cholesterol efflux from stimulated cells to HDL after a SUV-7α-OOH challenge was significantly reduced relative to a SUV-cholesterol control under identical reaction conditions. A much smaller effect of 7α-OOH was observed for non-stimulated cells, which proved to be statistically insignificant. These findings provide even more compelling support for our hypothesis that 7α/7β-OOH, an oxysterol often overlooked up to now because of its low content in atherosclerotic lesions, could be a primary driving force in RCT impairment associated with oxidative stress disorders.

Figure 6.

Effect of 7α-OOH on cholesterol export to apoA-I or HDL. THP-1 monocytes were differentiated to macrophages by treating with 30 nM PMA for 3 days. After an overnight incubation in 10% FBS-containing RPMI medium, cells were washed and incubated with 20 μM [³H]Ch / 1.5 mM Me-β-CD in serum-free medium for 12 h. After washing, cells were either not stimulated (NS) or stimulated (S) with Bu₂cAMP, then washed again and exposed to POPC liposomes containing cholesterol only or cholesterol in combination with 7α-OOH for 6 h. The initial concentration of 7α-OOH in bulk phase was 20 μM; cholesterol concentration in the control was also 20 μM. Immediately after these incubations, cells were washed, overlaid with (A) apoA-I (20 μg/ml in 0.5% BSA/RPMI) or (B) HDL (50 μg/ml in 0.5% BSA/RPMI), and extent of [³H]cholesterol efflux over a 6 h period was determined by scintillation counting. Plotted data in (A) and (B) are means ± SE (n=3 in each case), and represent percentages of total cellular cholesterol. *P < 0.01.

Discussion

Scavenger receptor-expressing macrophages normally resist harmful buildup of cholesterol from endogenous or exogenous sources by activating transport processes whereby cholesterol is delivered to mitochondria for CYP27A1-mediated conversion to 27-OH.^8–12 The latter stimulates LXR-mediated transcription of the plasma membrane proteins ABCA1 and ABCG1.^13–16 These proteins play a key role in the export of excess cholesterol, apoA-I serving as its principal extracellular acceptor in the case of ABCA1, and HDL in the case of ABCG1.^15,16 In addition to scavenger receptors, macrophages express a number of other proteins involved in cholesterol homeostasis. These include StAR family proteins, one of which, StarD1 on the outer mitochondrial membrane, has been implicated in cholesterol translocation to the CYP27A1 system on the inner membrane.^9–11 Although less evidence is available for macrophages, the overall mechanism of cholesterol delivery to/into mitochondria appears to be similar to that of steroidogenic cells.^11,29 It is well established that Leydig MA-10 testicular cells, for example, express a network of proteins dedicated to steroid synthesis upon stimulation by chorionic gonadotropin or its downstream effector, cAMP.³¹ These proteins include the cholesterol side-chain cleavage enzyme (CYP11A1) on the inner membrane, StarD1 on the outer membrane, and at least one cytosolic homologue such as StarD4 or StarD5. It is becoming increasingly clear that the StAR transporters act cooperatively in delivering cholesterol to the inner membrane for CYP11A1-catalyzed formation of pregnenolone, the first step in steroid hormone biosynthesis.^31–33 In a recent study,³⁴ we showed that Bu₂cAMP-stimulated MA-10 cells expressed higher levels of StarD1 and StarD4 proteins than non-stimulated controls, and also (i) channeled more liposomal 7α-OOH as well as cholesterol to/into mitochondria, (ii) underwent a greater loss of ΔΨ_m and progesterone output during 7α-OOH exposure, and (iii) underwent more extensive intrinsic apoptosis. These findings strongly support the notion that under oxidative stress conditions, steroidogenic cells may deliver not only cholesterol to mitochondrial compartments, but also ChOOHs such as 7α/7β-OOH, thereby setting the stage for free radical damage, metabolic dysfunction, and even apoptotic cell death.

An analogous mechanism of mitochondria-targeted oxidative damage/dysfunction is described for vascular macrophages in the present study. As also demonstrated for murine RAW 264.7 macrophages recently,²⁸ human THP-1 macrophages robustly overexpressed StarD1 and ABCA1 upon stimulation with Bu₂cAMP. Mitochondria in stimulated cells imported liposomal 7α-OOH and cholesterol at significantly higher rates than in non-stimulated controls and StarD1 knockdown blunted this, thus establishing this protein’s involvement in the uptake. Although no significant upregulation of StarD4 was observed after Bu₂cAMP treatment, this protein was expressed at fairly high constitutive levels in THP-1 cells and this may have been sufficient for a trafficking role along with StarD1. Although StarD4 involvement in 7α-OOH delivery was not specifically assessed by a knockdown approach, we showed previously that recombinant StarD4 was fully capable of transporting 7α-OOH from liposomes to isolated mitochondria, which caused peroxidative damage and membrane depolarization.²⁷ No such effect was observed with a phospholipid hydroperoxide, consistent with the known sterol specificity of StarD4 and most other StAR proteins.²⁷ In the present study, we observed a StarD1-dependent uptake of 7α-OOH by mitochondria of stimulated THP-1 macrophages. This uptake resulted in more rapid induction of free radical-mediated lipid peroxidation and membrane depolarization than observed in non-stimulated controls. These deleterious effects were significantly diminished by StarD1 knockdown, again consistent with its involvement in selective 7α-OOH targeting to mitochondria. t-BuOOH also caused macrophage damage/dysfunction, but this occurred independently of Bu₂cAMP stimulation or StarD1 knockdown. Since t-BuOOH lacks the structural requirements of a StAR ligand,³⁵ this result confirmed StarD1 binding/trafficking specificity for 7α-OOH.

One of the most significant of our observed functional consequences of StarD1-mediated 7α-OOH transfer to mitochondria of stimulated THP-1 cells was the large decrease in 27-OH production. This decrease was dependent on both hydroperoxide dose and contact time. We chose to measure 27-OH in the medium because these cells can release considerable amounts of it as biosynthesis progresses.¹² The decline in 27-OH yield was accompanied by reduced expression of ABCA1 and ABCG1. The latter finding was not unexpected, given that 27-OH is a strong natural agonist of nuclear LXR, which stimulates ABCA1 and ABCG1 transcription.^16,36 The 27-OH decline would have resulted from the observed loss of mitochondrial CYP27A1 activity (Figure 4A). The latter is attributed to 7α-OOH-induced redox damage - either directly to the enzyme itself or indirectly to its inner membrane surroundings. Diminished 27-OH output and ABCA1 expression after 7α-OOH exposure, most likely explains the impaired early stage RCT of these cells, i.e. reduced ability to export cholesterol to apoA-I (Figure 6A). An analogous explanation would apply to their reduced cholesterol export to HDL (Figure 6B), which is mainly attributed to down-regulation of ABCG1 (Figure 5A). We found that prolonged cell exposure to 7α-OOH also resulted in a significant down-regulation of StarD1 protein (Figure 5A). StarD1 is induced during differentiation of THP-1 macrophages and further elevated by cAMP stimulation,^9,10 as confirmed in the present study. However, treatment with acetylated LDL or oxLDL, resulting in cholesterol overloading and incipient foam cell formation, has been shown to down-regulate StarD1.¹¹ The latter may have occurred to prevent cytotoxicity from further cholesterol accumulation as cholesterol export became overwhelmed and/or incapacitated. A similar explanation may apply for the gradual 7α-OOH-induced decline in StarD1 that we observed.

Most prior studies on oxysterol involvement in atherogenesis have concentrated on major secondary products of free radical-mediated cholesterol oxidation in LDL before and after it enters macrophages in vascular sub-endothelial spaces. Considerable attention has been directed to 7=O and 27-OH, which are the most abundant oxysterols in atherosclerotic plaques and at sufficiently high levels can induce apoptosis in fibroblasts, endothelial cells, and macrophages.^{19,21–23,37} Though not considered in this light recently, primary 7-OOHs, which arise from free radical reactions,^21,24 are the major sources of 7=O and 7-OH via redox turnover. This can account for the very low levels of 7-OOH detected in oxLDL and arterial plaques.^18,20,21 In this study, we compared 7α-OH and 7=O with 7α-OOH for cytotoxic potency and found that the hydroperoxide was much more toxic to stimulated THP-1 macrophages than the others, e.g. at 50 μM it produced a near complete loss of viability, whereas 7α-OH and 7=O were essentially non-toxic at this concentration. We used ethanol as a 7-oxide vehicle in this experiment (Figure 3A) in attempt to standardize delivery rate of the three 7-oxides to cells. When SUVs were used as delivery vehicles, LD₅₀ for 7α-OOH was ~100 μM (Figure 3B), whereas that for 7=O, which is much less hydrophilic than 7α/7β-OOH,¹⁹ could not be determined, evidently because it remained tightly associated with the liposomes, even after 24 h incubation with cells. It is reasonable to assume that if oxLDL were the 7-oxide vehicle into macrophages, results similar to those in Figure 3A would be obtained, highlighting the toxic importance of 7α-OOH (and/or its 7β-OOH epimer) relative to the other 7-oxides. This is the first time that these three 7-oxides have been compared in this fashion using vascular macrophages as target cells. Most prior studies used higher concentrations of oxysterols (7=O, 7-OH, 27-OH) delivered in organic solvents (ethanol, isopropanol) to other cell types, e.g. fibroblasts, endothelial, and smooth muscle cells.^19,38,39 Only one of these studies¹⁹ included a 7-hydroperoxide, viz. 7β-OOH, which was detected at relatively high levels in LDL oxidized in vitro. 7β-OOH was found to be far more toxic to dermal fibroblasts than 7=O, 7β-OH, 5-O-6, or 25-OH.¹⁹ It was concluded that 7β-OOH was the 7-oxide most likely to be responsible for oxLDL’s well-known toxicity toward a variety of cells other than macrophages.

By extending 7-OOH-induced damage/dysfunction/cytotoxicity to vascular macrophages in the present study, we provide a new mechanistic model for how early stage oxidative injury to these cells can impair RCT and promote atherogenesis. This model depicts a type of “stealthy” delivery of 7-OOH via a natural trafficking pathway, which normally delivers cholesterol to mitochondria to initiate export signaling when its levels become excessive. Because 7-OOHs are structurally similar to cholesterol, they can be recognized by StAR proteins and trafficked to mitochondria alongside cholesterol. We postulate that significant levels of 7-OOHs and other 7-oxides exist in oxLDL as it enters macrophages in the vascular wall. Most of the 7-OOHs would have arisen during free radical-mediated oxLDL formation in sub-endothelial spaces. At the early stages of this process, the steady levels of 7-OOH would be relatively high, but then decline as chain lipid peroxidation progresses; meanwhile, there would be a build-up of redox-inert products of 7-OOHs, namely 7-OHs and 7=O. All of these 7-oxides, including those derived from hydrolysis of oxidized cholesteryl esters, would be recognized by the StAR trafficking network for distribution to intracellular sites, including mitochondria, but only 7α/7β-OOH would be capable of the redox damage/dysfunction observed in this study, viz. chain lipid peroxidation, membrane depolarization, reduced ABCA1 expression, and RCT impairment. Such effects would be more serious at early stages of LDL oxidation, when 7-OOH levels are relatively high, than at late stages. These are new mechanistic insights into oxidative stress-linked atherogenesis which warrant more extensive study at the in vitro as well as in vivo level.

It is worthwhile to consider how vascular macrophages might cope with stress-inducing 7-OOH by employing some natural antioxidant modality. The only antioxidant enzyme known to be capable of catalyzing direct inactivation of 7-OOHs and other ChOOHs is glutathione peroxidase type-4 (GPx4), a selenoenzyme that converts these peroxides to redox-inactive alcohols at the expense of reduced glutathione.⁴⁰ GPx4 is known to be expressed in mitochondria as well as other compartments of mammalian cells. ChOOHs as a group are detoxified much more slowly by GPx4 than phospholipid counterparts (PLOOHs), making ChOOHs potentially more damaging than PLOOHs.^41,42 Release of incoming ChOOHs with subsequent induction of free radical peroxidation is assumed to occur in the vicinity of the mitochondrial inner membrane, since StarD1 delivers cholesterol at/near this site for processing by CYP27A1.^29,43 It is unlikely that GPx4 would have access to StarD4/D1-bound 7-OOH during transit because of tight constraints of the sterol binding pocket.³⁵ Certain chemical antioxidants might prove beneficial as complements to GPx4 and/or any other natural antioxidants. Mito-Q is of special interest along these lines because it specifically targets mitochondria of healthy cells and can site-specifically quench free radical reactions in mitochondrial membranes.^44,45 It is reasonable to expect that Mito-Q will suppress damage/dysfunction caused by 7-OOH redox in macrophage mitochondria, making it a promising new anti-atherogenic pharmacologic agent.

In summary, we have demonstrated for the first time how a natural transport pathway for cholesterol in macrophages can be co-opted by a redox-active ChOOH, leading to mitochondrial damage and impairment of RCT at its earliest stages. This work provides new insights into how oxLDL may initiate atherogenesis and sets the stage for more comprehensive studies involving 7-OOHs borne by LDL itself.

Significance.

An elevated level of oxidized low density lipoprotein (oxLDL) in the circulation (associated with disorders such as chronic inflammation and diabetes) is a known risk factor in atherogenesis. In the artery wall, macrophages bind and internalize oxLDL as a source of cholesterol and other lipids. Macrophages can also release cholesterol by reverse cholesterol transport (RCT). If cholesterol import exceeds export, macrophages become overloaded with cholesterol, and accumulate in zones called atherosclerotic plaques, which obstruct blood flow and can eventually lead to heart attacks. In this study, we describe a previously unrecognized mechanism by which RCT can be incapacitated in oxidative disorders, namely through delivery of redox-active cholesterol hydroperoxides to mitochondria via a natural cholesterol trafficking pathway. Our findings are highly significant because they provide new insights into the role of macrophage redox damage in atherogenesis. Moreover, our findings suggest that mitochondria-targeted antioxidant drugs can be used to prevent this damage.

Non-standard Abbreviations

Bu₂cAMP

dibutyryl cyclic AMP

Ch

cholesterol

ChOOH

cholesterol hydroperoxide

C11-BODIPY

4,4-difluoro-5-(4-phenyl-1,3-butadienyl)-4-bora-3a,4a-diaza-s-inda-cene-3-undecanoic acid

JC-1

5,5’,6,6’-tetrachloro-1,1’,3,3’-tetraethyl-benzimidazolylcarbocyanine iodide

HPTLC-PI

high performance thin layer chromatography with phosphor-imaging detection

kd

knock down

Me-β-CD

methyl-β-cyclodextrin

MTT

3-(4,5-dimethylthiazolyl-2-yl)-2,5-diphenyltetrazolium bromide

PMA

phorbol 12-myristate 13-acetate

POPC

1-palmitoyl-2-sn-glycero-3-phosphocholine

RCT

reverse cholesterol transport

RFI

fluorescence intensity ratio

SUV

small unilamellar vesicle

StAR

steroidogenic acute regulatory

t-BuOOH

tert-butyl hydroperoxide

7-OH

cholest-5-ene-3β,7α/β-diol

7

O, cholest-5-ene-3β,7-one

7-OOH

3β-hydroxycholest-5-ene-7α/β-hydroperoxide

27-OH

27-hydroxycholesterol