The aim of the present study was to establish mitochondrial cholesterol trafficking 18 kDa translocator protein (TSPO) as a potential therapeutic target, capable of increasing macrophage cholesterol efflux to (apo)lipoprotein acceptors. Expression and activity of TSPO in human (THP-1) macrophages were manipulated genetically and by the use of selective TSPO ligands. Cellular responses were analysed by quantitative PCR (Q-PCR), immunoblotting and radiolabelling, including [3H]cholesterol efflux to (apo)lipoprotein A-I (apoA-I), high-density lipoprotein (HDL) and human serum. Induction of macrophage cholesterol deposition by acetylated low-density lipoprotein (AcLDL) increased expression of TSPO mRNA and protein, reflecting findings in human carotid atherosclerosis. Transient overexpression of TSPO enhanced efflux (E%) of [3H]cholesterol to apoA-I, HDL and human serum compared with empty vector (EV) controls, whereas gene knockdown of TSPO achieved the converse. Ligation of TSPO (using PK11195, FGIN-1-27 and flunitrazepam) triggered increases in [3H]cholesterol efflux, an effect that was amplified in TSPO-overexpressing macrophages. Overexpression of TSPO induced the expression of genes [PPARA (peroxisome-proliferator-activated receptor α), NR1H3 (nuclear receptor 1H3/liver X receptor α), ABCA1 (ATP-binding cassette A1), ABCG4 (ATP-binding cassette G4) and APOE (apolipoprotein E)] and proteins (ABCA1 and PPARα) involved in cholesterol efflux, reduced macrophage neutral lipid mass and lipogenesis and limited cholesterol esterification following exposure to AcLDL. Thus, targeting TSPO reduces macrophage lipid content and prevents macrophage foam cell formation, via enhanced cholesterol efflux to (apo)lipoprotein acceptors.
Enhancing the efficiency of mitochondrial cholesterol trafficking can increase CYP27A1-dependent generation of oxysterol ligands for LXRs, which regulates the anti-atherogenic cholesterol efflux pathway.
In the present study, we have found that a modest cholesterol ‘load’ induces the expression of mitochondrial cholesterol TSPO, and that overexpression and/or ligation of TSPO facilitates cholesterol efflux to acceptors, such as apoA-I, HDL and human serum, inducing the expression of NR1H3, PPARA, ABCA1, ABCG4 and APOE and reducing macrophage cholesterol and triacylglycerol mass.
Small molecules targeting mitochondrial TSPO activity or maintaining its function could prove therapeutically useful in regressing atherosclerotic lesions and reducing the incidence of coronary heart disease.
Mitochondrial dysfunction, associated with increased production of reactive oxygen species, accumulation of mitochondrial DNA damage and progressive loss of respiratory chain function, is found in atherosclerotic lesions in both human studies and animal models of atherogenesis . Oxidized low-density lipoprotein (oxLDL), aging, hyperhomocysteinaemia, hyperglycaemia, hypertriglyceridaemia, Type 2 diabetes mellitus and cigarette smoking are all associated with mitochondrial damage and increased production of reactive oxygen species [2–4]. Increased arterial oxidative stress modifies low-density lipoprotein (LDL) to a form recognized by macrophage scavenger receptors, resulting in further mitochondrial damage and the unregulated accumulation of excess cholesterol and cholesteryl esters within macrophage ‘foam cells’, a hallmark of early and developing atheroma [1–5]. Reversal of this process, regressing and stabilizing atherosclerotic lesions, can be effected by efficient removal of cholesterol from ‘foam cells’ to acceptor particles such as (apo)lipoprotein A-I (apoA-I), (apo)lipoprotein E (apoE) or nascent high-density lipoproteins (HDLs), which then enter the athero-protective reverse cholesterol transport pathway responsible for delivery of excess cholesterol to the liver for excretion in bile and bile acids .
Previously, we  and others  have demonstrated that overexpression of the mitochondrial cholesterol trafficking protein steroidogenic acute regulatory protein (StAR) decreases macrophage lipid content [7,8], represses inflammation  and increases macrophage cholesterol efflux [7,8], whereas a viral vector expressing StAR reduces aortic lipids and atheroma in apoE−/− mice . Overexpression of StAR enhances the rate-limiting step in mitochondrial generation of oxysterol ligands for liver X receptors (LXRα/β)–transfer of cholesterol to sterol 27-hydroxylase (CYP27A1) located on the inner mitochondrial membrane [10,11]. CYP27A1 generates oxysterol ligands for LXRα/β, nuclear transcription factors, which regulate gene expression of proteins involved in the cholesterol efflux pathway, including the ATP-binding cassette transporters ABCA1 and ABCG1/ABCG4, which orchestrate transfer of cholesterol and/or phospholipids across the plasma membrane to (apo)lipoproteins, such as apoA-I and apoE [12–14]. Loss of CYP27A1 results in cerebrotendinous xanthomatosis, which can be characterized by increased risk of premature atherosclerosis, despite normal circulating concentrations of plasma LDL . Exploiting the anti-atherogenic functions of StAR may prove problematical, however, due either to induction of lipogenesis  or dearth of small molecules capable of modulating the activity of this protein in vivo.
However, StAR interacts with a protein complex, located at contact sites between outer and inner mitochondrial membranes, which includes 18 kDa translocator protein [TSPO/peripheral benzodiazepine receptor (PBR)], voltage-dependent anion channel (VDAC) and adenine nucleotide transporter (ANT), and associated proteins, including cAMP-dependent protein kinase associated protein [PAP7/acyl-CoA-binding domain-containing 3 protein (ACBD3)], PBR-associated protein (PRAX1) and its proposed endogenous ligand diazepam-binding inhibitor (DBI/ACBD1) [16,17]. TSPO has five transmembrane domains and a high-affinity cholesterol/interaction recognition amino acid consensus (CRAC)-binding C-terminal domain [16,17]; the CRAC peptide was used recently as an inter-chelating agent to reduce cholesterol and aortic lesions in apoE−/− mice . Knockdown of TSPO induces arrest of mitochondrial cholesterol transport, which can be rescued by TSPO cDNA, whereas global deletion of TSPO is embryonic lethal in mice [16,17]. Intriguingly, TSPO ligands were shown recently to activate fasting metabolism, reducing lipogenesis and hepatosteatosis in obese mice , suggesting that additional activities associate with this protein.
The contribution, and therapeutic potential, of TSPO and associated proteins in regulation of the cholesterol efflux pathway in macrophages and macrophage ‘foam cells’ remains uninvestigated, despite the potential importance of this pathway in activating the anti-atherogenic cholesterol efflux process. In the present study, we have studied gene and protein expressions of TSPO during macrophage ‘foam cell’ formation and have demonstrated the positive impact of TSPO overexpression and/or ligation on macrophage cholesterol efflux; notably, TSPO overexpression causes a decline in macrophage total neutral lipid mass, without induction of lipogenesis, and effectively prevents ‘foam cell’ formation following exposure to acetylated low-density lipoprotein (AcLDL). We have postulated that existing ligands for TSPO may possess anti-atherosclerotic properties, readily amenable to testing in clinical studies.
MATERIALS AND METHODS
The human monocytic cell line (THP-1) and murine RAW264.7 macrophages were purchased from the European Cell Culture Collection. Cryopreserved ultrapure human peripheral blood mononuclear cells (107) isolated from fresh blood from a single donor (C-12907) were purchased from PromoCell. The sample of total RNA (50 μg) derived from normal human aortas pooled from four male/female Caucasians (aged 27–45 years), who died of sudden death, was purchased from the Clontech Laboratories. Tissue culture reagents were purchased from Lonza, and other sources include Amaxa Transfection reagent (Lonza), NuPAGE gels and buffers (Life Technologies), antibodies (Abcam) and primers and probes (Eurogentec). ApoA-I, HDL and LDL were purchased from the Athens Research UK; LDL was acetylated according to Brown et al. . Radiochemicals ([3H]cholesterol, [14C]acetic acid and [3H]oleic acid) were provided by the ICN Biologicals, and all other chemicals were from Sigma–Aldrich. Mammalian expression vectors (pCMV6) encoding TSPO, DBI, VDAC and ANT were supplied by OriGene, via Cambridge Biosciences; TriFecta Dicer Substrate RNAi duplexes directed against TSPO and scrambled control sequences were supplied by the Integrated DNA Technologies. We are indebted greatly to Professor D.J. Mangelsdorf (South Western Medical Center, USA) for generously providing the LXR-response element (LXRE) reporter plasmid (pCMX.LXRE).
Human THP-1 monocytes were maintained using a split ratio of 1:10 in RPMI medium supplemented with FBS (10%, v/v), L-glutamine (200 mM) and penicillin/streptomycin (50 μg·ml−1 and 50 units·ml−1 respectively). For experiments, cells were plated on to 12-well tissue culture dishes at a density of 1×106 cells·well−1 in complete RPMI medium (above) supplemented with 100 nM PMA to induce macrophage differentiation. Transfection of monocytes, immediately prior to differentiation, was achieved using the ‘Human Monocyte’ transfection reagent (VPA-1007, Lonza) and the electroporation protocol (Y001, Amaxa), as per manufacturer's instructions and as described previously ; our studies have established a transfection efficiency of 70–80%, without significant (<5%) loss of cell viability, using this system. Mammalian expression vector pCMV6_TSPO (0.5 μg) was used to transiently (48 h) overexpress the mitochondrial TSPO, whereas validated siRNA (10 nM) duplexes directed against TSPO and peroxisome-proliferator-activated receptor α (PPARA) were used for gene knockdown (48 h). Transient transfections were also performed using the same quantities of vector encoding full-length VDAC, ANT and DBI. The empty vector (EV; pCMV6) and scrambled siRNA sequences of the same nucleotides were used as respective controls. Macrophages were treated with TSPO ligands for 48 h prior to assessment of cholesterol efflux, using DMSO (<0.01%) as vehicle. Transient expression of pCMX.LXRE (0.5μg) in RAW 264.7 macrophages was achieved using FuGene6 (6 μl: 1 μg of DNA; 24 h), and luciferase activity (Britelite Plus, Perkin Elmer) was determined following treatment with conditioned medium derived from EV- and TSPO-transfected macrophages, as described previously [7,22,23]. Cell viabilities were determined by conversion of Methyl Thiazolyl Blue Tetrazolium Bromide with formazan, as described previously [7,22,23].
Macrophage lipid analysis and cholesterol efflux
Incorporation of [14C]acetic acid (1 μCi·ml−1; 24 h) into fatty acid, phospholipid, cholesterol, cholesteryl ester and triacylglycerol pools was initiated 48 h post-transfection and as described previously [7,22,23]. Esterification of cholesterol, in the presence of AcLDL (50 μg·ml−1), was monitored by flux of [3H]oleic acid (10 μM, 1 μCi·ml−1) into cholesteryl [3H]oleate, as described previously [7,20,22,23]. Macrophage lipids were extracted using hexane/propan-2-ol (3:2, v/v) and dried under nitrogen, before separation by TLC, using the mobile phases as described previously [7,20,22,23]. Lipids were identified by co-migration with authentic standards, and incorporation of radiolabel determined by scintillation counting, with data expressed as nmol incorporation per mg of cell protein. Production of 14CO2 from [1-14C]oleic acid (3 μM; 0.15 μCi·ml−1) was assessed as described previously . Neutral lipid mass (total cholesterol and triacylglycerol) in macrophage lipid extracts (above) was determined using the colorimetric assays (Infinity™) and are expressed as μg of lipid per μg of cell protein [7,20,22,23]. Macrophage efflux of [3H]cholesterol to apoA-I (20 μg·ml−1), HDL (20 μg·ml−1) and human serum (1%, v/v) was determined as described previously [22,23]. Efflux (24 h) from radiolabelled macrophages was initiated 48 h post-transfection by addition of serum-free RPMI containing each acceptor, and results are expressed as percentage cholesterol efflux, calculated as [d.p.m.medium/d.p.m.(medium + cells)] × 100%.
Analysis of gene and protein expressions
Total RNA was isolated (TriFast, Peqlabs) from macrophages and reverse transcribed to cDNA (BioLine) prior to measurement of cellular levels of mRNA encoding the mitochondrial cholesterol trafficking complex (TSPO, VDAC and ANT), lipid-responsive transcription factors [sterol-regulatory-element-binding transcription factor 1 (SREBF1), SREBF2, PPARA, PPARG, PPARD and nuclear receptor 1H3 (NR1H3; encoding LXRα)] and key elements of the cholesterol efflux pathway (ABCA1, ABCG1, ABCG4 and APOE). Quantitative PCR (Q-PCR), relative to glyceraldehyde-3-phosphate dehydrogenase (GAPDH), was performed using the primers and fluorescent [6-carboxyfluorescein (FAM)/6-carboxytetramethylrhodamine(TAMRA)] probe sequences indicated in Supplementary Table S1 (http://www.clinsci.org/cs/127/cs1270603add.htm); negative reverse transcription controls were performed for all samples. Gene expression of an array of 84 genes implicated in atherosclerosis was analysed in control and TSPO-overexpressing macrophages using the RT2 Profiler Human Atherosclerosis PCR array (PAHS-038A, Qiagen). Assessment of CPT-1A (carnitine palmitoyltransferase 1A) gene expression was performed using forward (5′-ACAGTCGGTGAGGCCTCTTATGAA-3′) and reverse (5′-TCTTGCTGCCTGAATGTGAGTTGG-3′) primers, generating a 252 bp product, semi-quantified by densitometry using GelDoc XR (Bio-Rad Laboratories).
Macrophage cell lysates were prepared in RIPA buffer [25 mM Tris HCl pH 7.6, 150 mM NaCl, 1% (v/v) Nonidet P40, 1% (w/v) sodium deoxycholate and 0.1% SDS] supplemented with Complete™ protease inhibitor cocktail (Roche), and proteins (20–50 μg·lane−1) were separated using NuPAGE (10%, w/v) gels. Transfer to a PDVF membrane was followed by immuno-blotting using rabbit polyclonal antibodies (1:1000 dilution) against ABCA1 TSPO, PPARα, acyl-CoA: cholesterol acyltransferase (ACAT1) and GAPDH (1:2000 dilution), and detection was achieved for the majority of the immunoblob, using secondary antibodies coupled to the ECL detection system, as described previously [7,22,23]. Densitometry was performed using a GelDoc XR using Image Lab Software for Western Blot Analysis (Bio-Rad Laboratories). However, the images presented for siRNA knockdown of TSPO in Figure 2(A) and for PPARα following TSPO overexpression in Figure 5(A) were detected using a donkey anti-rabbit secondary antibody (1:12 500 dilution) labelled with IRDye® 800CW (LI-COR) and the LI-COR Odyssey Fc Imaging System with LI-COR Image Studio version 3.1.4 software.
All values are means±S.E.M., with numbers of independent experiments denoted by n. Significant (P<0.05) differences were determined using a Student's t test, or by using ANOVA and post-hoc Bonferroni tests, as indicated in the legends to the Figures.
An indication of the levels of gene expression of TSPO, VDAC and ANT, relative to the housekeeping gene GAPDH, in human mononuclear cells (primarily lymphocytes and monocytes) derived from peripheral blood and in normal heart aorta is shown in Figure 1(A) to provide a physiological context and comparison for the studies carried out using macrophages derived from the human THP-1 monocytic cell line; mRNA levels of these genes in differentiated THP-1 macrophages and THP-1 macrophage ‘foam cells’ are shown in Figure 1(B). Exposure of THP-1 macrophages to two doses of 50 μg·ml−1 AcLDL (24 h) led to a modest but significant increase (27%; P<0.05) in total cholesterol mass (control 58.0±1.7 μg·mg−1 cell protein compared with AcLDL 73.7±7.2 μg·mg−1 cell protein; P<0.05; n=4), a mild cholesterol loading condition that avoids the overt toxicity associated with exposure to higher concentrations of modified LDL . Levels of TSPO mRNA were increased by 1.8±0.3-fold following exposure to AcLDL, reflected in increased levels of TSPO (37%; P<0.05; n=3), relative to VDAC or GAPDH (Figure 1B). Analysis (MatInspector) of the −3 kb upstream promoter region of TSPO revealed a number of putative transcription-factor-binding sites, including sterol-regulatory elements (−2113 to 2097; −1957 to 1922; −1783 to 1769) and PPAR-responsive elements (−2214 to 2191; −366 to 344).
Expression of TSPO in human peripheral blood mononuclear cells, heart aorta and in THP-1 macrophage ‘foam cells’
Overexpression and gene knockdown of TSPO in THP-1 macrophages: impact on cholesterol efflux
Overexpression and knockdown of dimeric TSPO (44 kDa) relative to GAPDH, and compared with the EV controls (pCMV6) or siRNA negative control, are shown in Figures 2(A) and 2(B), measured as detailed in the Materials and methods section. No changes in cell viability were associated with TSPO overexpression [EV 37.8±0.9 μM formazan compared with TSPO 36.2±0.9 μM formazan, n=3; not significant (NS)]. Overexpression of TSPO significantly increased efflux (24 h), of [3H]cholesterol, to 20 μg·ml−1 apoA-I (60±27.4%; P<0.01; n=3), 20 μg·ml−1 HDL (74±13.2%; P<0.05) and 1% (v/v) human serum (51.7±21.2%; P<0.05; n=3) (Figure 2C) when compared with cells transfected with the EV. Conversely, gene knockdown of TSPO subsequently reduced [3H]cholesterol efflux (24 h) to apoA-I (20 μg·ml−1) and HDL (20 μg·ml−1) by 15.7±5.7% (P<0.05) and 14.3±9.2% (P<0.05) respectively when compared with the scrambled siRNA negative control (Figure 2D).
Genetic manipulation of TSPO regulates macrophage cholesterol efflux
Transfection with the same quantity of vector encoding the proposed endogenous ligand for TSPO and DBI induced a small increase in efflux to apoA-I, which did not quite reach significance (EV 5.1±0.78% compared with DBI 6.4±0.3%; P=0.06; n=3) but did significantly (P<0.01) increase efflux of [3H]cholesterol to HDL (EV 19.8±1.8% compared with DBI 21.3±1.9%; n=3). By contrast, equivalent transfections with vector encoding either VDAC or ANT did not significantly alter efflux of [3H]cholesterol to either apoA-I (EV 4.0±0.2% compared with VDAC 4.2±0.3% compared with ANT 4.4±0.4%; n=3; NS) or HDL (EV 8.8±0.7% compared with VDAC 8.8±0.8% compared with ANT 9.7±0.8%; n=3; NS), when tested under the same conditions.
Ligation of TSPO increases macrophage cholesterol efflux to apoA-I and HDL
Wild-type THP-1 macrophages were treated with the established TSPO ligands PK11195, FGIN-1-27 and flunitrazepam, using concentrations commonly used to increase mitochondrial cholesterol trafficking and steroidogenesis [16,17]. Basal efflux was not affected significantly by treatment with ligands at these concentrations (control 4.07±0.63% compared with PK11195 4.40±0.81% compared with FGIN-1-27 4.69±0.97% compared with flunitrazepam 4.09±1.13%; n=3). Significant increases in [3H]cholesterol efflux to apoA-I (20 μg·ml−1; Figure 3) were observed following treatment with PK11195 (30 μM; P<0.05; n=3), flunitrazepam (30 μM; P<0.01; n=3) and FGIN-1-27 (10 μM; P<0.05; n=3) (Figures 3A–3C). By contrast, only PK11195 (P<0.01; n=3) and FGIN-1-27 (P<0.01; n=3) increased efflux of cholesterol to HDL (20 μg·ml−1) (Figures 3A–3C). Notably, in macrophages overexpressing TSPO, treatment with FGIN-1-27 increased efflux to apoA-I by a further 2.2±0.2-fold (P<0.001; n=3) resulting in a 5.2±0.6-fold (P<0.001; n=3) increase in efflux above basal levels (Figure 3D). Equally, knockdown of TSPO abrogated the stimulatory (31.7±8.6%) effect of FGIN-1-27 (10 μM) on cholesterol efflux to apoA-I (scrambled control 8.3±0.7% compared with siRNA to TSPO 6.8±0.6%; P<0.05; n=3).
Ligation of TSPO can increase macrophage cholesterol efflux
Overexpression of TSPO: impact on lipid phenotype
Transient TSPO overexpression (48 h) was associated with a marked loss of macrophage neutral lipid mass, due to reductions in total cholesterol (52.6±8.9%; P<0.05; n=4) and triacylglycerol (34.2±10.9%; P<0.05; n=4) contents (Figure 4A). However, we have reported previously that overexpression of the mitochondrial cholesterol trafficking protein StAR (STARD1), is associated with induction of lipogenesis in macrophages . Accordingly, we have investigated the impact of transient TSPO overexpression (48 h) on the subsequent incorporation of [14C]acetate into macrophage lipid pools (24 h) (Figure 4B). A small reduction in total lipogenesis (10±3.1%; P<0.05; n=6) was observed, predominantly due to a 15±6.2% (P<0.05; n=6) reduction in incorporation of [14C]acetate into the fatty acid pool. Note that it is likely that steady-state levels of incorporation into lipid pools other than cholesterol are measured over the time period used in the present study. Increased oxidation of [1-14C]oleic acid, expressed as a percentage of cellular uptake (EV 0.30±0.03%·h−1 compared with TSPO 0.52±0.05%·h−1; P<0.05; n=3), was associated with a 2-fold induction of gene expression of CPT-1A (arbitrary fluorescence units: EV 61.2±6.7 compared with TSPO 119.8±2.5; P<0.01; n=3).
Overexpression of TSPO and modulation of macrophage lipid phenotype
The ability of TSPO overexpression (48 h) to increase cholesterol efflux and protect macrophages against accumulation of cholesteryl ester was investigated following exposure to AcLDL (50 μg·ml−1) for 24 h (Figures 4C and 4D). Following a cholesterol load, apoA-I stimulated cholesterol efflux by 1.9±0.2-fold (P<0.05) above basal (EV) control. By contrast, in TSPO-overexpressing cells, apoA-I increased cholesterol efflux by 3.8±0.4-fold (P<0.001), an increase of 1.6±0.1-fold (P<0.01) compared with the EV control (Figure 4C). Equally, although addition of AcLDL (50 μg·ml−1) increased incorporation of [3H]oleate into the cholesteryl ester pool by 81±12.0% (P<0.01; n=3) in macrophages transfected with the EV, overexpression of TSPO prevented significant increases in formation of cholesteryl [3H]oleate following treatment with this modified lipoprotein (Figure 4D, upper panel). Equally, TSPO overexpression prevented effectively the increase in cholesterol mass following AcLDL (50 μg·ml−1; 24 h) treatment (Figure 4D, lower panel) compared with the EV control. No change in expression of ACAT1 protein was observed in TSPO overexpressing cells (Figure 5A).
Changes in macrophage gene and protein expression mediated by TSPO overexpression
Overexpression of TSPO: changes in expression of genes and proteins involved in the cholesterol efflux pathway
Gene expression of the sterol-responsive transcription factors SREBF2, PPARD and PPARG was repressed in cells overexpressing TSPO, whereas expression of NR1H3 was increased significantly; this last gene reflected in increased LXRE reporter activity (TSPO 184.8±23.2% of control; P<0.05; n=4). Gene expression of PPARA (Figure 5B) was increased, accompanied by an increase in PPARα protein (32±5.7%; P<0.05; n=3) in macrophages transiently expressing TSPO compared with GAPDH (Figure 5A). Treatment with FGIN-1-27 (10 μM; 24 h) also increased mRNA levels of PPARA (2.0±0.2-fold; n=3) and NR1H3 (1.5±0.5-fold; n=7), replicating observations in TSPO-overexpressing macrophages.
Increases in the expression of genes encoding proteins involved in the cholesterol efflux pathway were observed in macrophages overexpressing TSPO (Figure 5B). Levels of ABCA1 mRNA were increased by more than 6.5±2.3-fold (n=6), and this was reflected in increased levels of ABCA1 protein (39±9.3%; P<0.05; n=3; Figure 5A) compared with GAPDH. Gene expression of ABCG4 and the endogenously produced cholesterol acceptor, APOE, was also increased by 2- to 3-fold (n=6) in macrophages overexpressing TSPO (48 h). Analysis of an array of genes implicated in atherosclerosis, relative to five housekeeping genes, revealed only two more genes up-regulated by >3-fold following TSPO overexpression, namely FN1 (fibronectin 1) and PDGFB (platelet-derived growth factor β).
Overexpression and ligation of TSPO: blockade by PPARα antagonism
Since induction of PPARα is implicated in the pathway by which TSPO overexpression or ligation stimulates the cholesterol efflux, a short series of experiments was performed using PPARα antagonists. Induction of ABCA1 gene expression by TSPO (ratio to GAPDH: control 0.98 compared with TSPO 2.45) was blocked by addition of 10 μM GW6471  (ratio to GAPDH: control 0.85 compared with TSPO 0.98) and partially reduced by treatment with geranylgeranyl pyrophosphate (GGPP)  (10 μM) (ratio to GAPDH: control 1.05 compared with TSPO 1.99). Induction of NR1H3 by TSPO (ratio to GAPDH: control 1.14 compared with TSPO 3.89) was inhibited effectively by both 10 μM GW6471 (ratio to GAPDH: control 0.58 compared with TSPO 0.39) and 10 μM GGPP  (ratio to GAPDH: control 0.10 compared with TSPO 0.31); the values are averages from a single representative experiment, which was performed three times.
Furthermore, the stimulatory effect of TSPO overexpression on [3H]cholesterol efflux to apoA-I (20 μg·ml−1; 24 h) was effectively blocked by the addition of the PPARα antagonist GW6471  (10 μM) (Figure 6A), and by treatment with siRNA (10 nM) directed against PPARα (Figure 6B), which allow us to posit the pathway described in Figure 6(C).
A putative role for PPARα in mediating the effect of TSPO overexpression on macrophage cholesterol efflux
The present study investigated the relationship between components of the mitochondrial cholesterol trafficking and macrophage cholesterol homoeostasis and demonstrated that overexpression, gene silencing and ligation of mitochondrial TSPO modulate cholesterol efflux, with TSPO overexpression channelling cholesterol away from the cholesteryl ester pool and reducing macrophage total neutral lipid mass, without triggering compensatory increases in lipogenesis. These effects appear to be related to the induction and/or activation of LXRα and PPARα caused by TSPO overexpression or ligation–blockade of either pathway using antagonists and/or siRNA directed against these transcription factors prevented the TSPO-dependent increases in cholesterol efflux.
Overexpression of the 18 kDa TSPO is associated with increased efflux of cholesterol to apoA-I and HDL, and induction of the expression of genes was involved in the cholesterol efflux pathway [7,8], including NR1H3, ABCA1 and ABCG4 but not ABCG1 (Figure 5). The reason for this distinctive regulatory pattern is not clear at present, but dissociation of ABCA1 and ABCG1 gene expression has been noted in bovine mammary tissue during the switch from lactating to dry periods . Induction of LXRα is consistent with sequestration of SREBP2 at the endoplasmic reticulum and loss of SREBF2 expression (Figure 5B), and both may contribute to increased induction of ABCA1 mRNA and protein [28–31]. Expression of ABCG4 in human monocyte macrophages lies under the control of oxysterols and retinoids , suggesting that both ABCG4 and ABCA1 might be regulated by LXRα activation or induction (Figure 5B), and one mechanism by which [3H]cholesterol efflux to HDL is increased (Figure 2B).
Moreover, overexpression of TSPO and/or ligation of this protein by FGIN-1-27 induced PPARα gene and/or protein levels, associated previously with increased ABCA1 expression and cholesterol efflux [32–35]. Expression of PPARα can be regulated by stress, hormones and starvation in rodents, whereas human PPARα can mediate its own expression and is induced during macrophage differentiation in response to high glucose levels . Notably, the PPARA promoter can be activated by PPAR agonists in an LXRα-dependent manner and LXRα is a PPARα target , so that these transcription factors can work together in a co-operative manner to control ABCA1 expression and cholesterol efflux . Overexpression of TSPO induces the expression of LXRα and PPARα, whereas the use of either siRNA or an antagonist directed against PPARα (GW6471) (Figure 6) indicates an obligate role for PPARα in mediating TSPO-dependent increases in [3H]cholesterol efflux to apoA-I and possibly other cholesterol acceptors (Figure 2B).
Indeed, the co-ordinated induction of PPARα and LXRα by TSPO may be a key factor in abrogating the adverse effects associated previously with activation of LXRα by agonists such as T091317 . In particular, our previous study  reported that overexpression and activation of the mitochondrial cholesterol trafficking protein StAR in murine macrophages were associated not only with enhanced cholesterol efflux and increased expression of ABCA1, but also with substantive induction of lipogenesis, thereby limiting the potential therapeutic utility of this approach [7,36]. By contrast, overexpression of TSPO was associated not only with increased cholesterol efflux to acceptors (Figure 2B), but also with a marked loss of macrophage neutral lipid content (Figure 4A) and a small reduction in lipogenesis, primarily within the fatty acid pool (Figure 4B). Furthermore, consistent with PPARα induction, levels of CPT-1A gene expression were increased, reflected in enhanced fatty acid oxidation, in TSPO-overexpressing cells. Activation of PPARα is known to induce modest increases in lipogenesis, possibly via participation in the generation of an endogenous LXRα ligand , and enhances fatty acid oxidation, by inducing the expression of long-chain fatty acyl CoA synthetases and CPT-1A which are essential for generating fatty acyl CoA and facilitating the entry of fatty acyl carnitine into mitochondria for β-oxidation [32–34]. These data provide a plausible explanation for the observed reductions in macrophage triacylglycerol content (Figure 4A) and [14C]fatty acid levels (Figure 4B). Overexpression of TSPO was also associated with reduced incorporation of [3H]oleate into the cholesteryl ester pool and with decreased total cholesterol mass, following exposure to modified LDL (Figure 4D), which may reflect the reduced availability of de novo-synthesized fatty acids for esterification (Figure 4B) [32–34]; indeed, activation of PPARα and LXRα has previously been associated with reductions in ACAT activity and cholesterol esterification in human macrophages [32,33]. These results resonate with those of Falchi et al. , who demonstrated that the TSPO ligand PK11195 (40 μM) increased the efflux of radiolabelled cholesterol from murine Swiss 3T3 fibroblasts to FBS (10%). Although FBS is not a specific cholesterol acceptor, in terms of its interactions with ABCA1 or ABCG1/G4, the effect of PK11195 was accompanied by decreases in cholesterol esterification without alteration in cholesterol biosynthesis, in good agreement with the data shown in the present study.
Notably, treatment with AcLDL, which, while not a biological ligand, does achieve effective cholesterol loading in THP-1 macrophages, appears to reflect the induction of TSPO observed in human carotid atherosclerotic plaques, wherein binding of TSPO ligands such as [3H]PK11195 and [3H]DAA1106 has been shown to significantly correlate with macrophage burden . Lipid regulation of TSPO gene expression has not previously been reported during macrophage cholesterol loading, but Tspo can be regulated by protein kinase Cε, by downstream signalling pathways involving mitogen-activated protein kinase [MAPK; Raf/extracellular-signal-regulated kinase 1/2 (ERK1/2)], and by acting on targets such as c-Jun and the signal transducer and activator of transcription 3 (STAT3) [16,17]. It is known that STAT3 gene delivery can reduce atherosclerotic lesions in LDL-receptor-knockout mice fed on a high-cholesterol diet , but whether induction of TSPO forms part of that response is not established. However, it is known that expression of TSPO is modulated by PPARs in a tissue-specific manner, with transrepression of Tspo by PPARα cited in Leydig cells, but the reverse is observed in non-steroidogenic tissues . Given the presence of a number of putative PPAR-response elements within the promoter region of TSPO, it is possible to speculate that induction of PPARα in human macrophages may form part of a feed-forward regulatory loop, sustaining or amplifying TSPO expression (Figure 6).
Finally, the ability of TSPO to moderate gene transcription has been observed in lower species and thus may be evolutionarily conserved in mammalian cells. For example, the bacterial orthologue, TspO outer membrane protein, found in Rhodobacter sphaeroides, which is involved in effluxing intermediates of the porphyrin biosynthetic pathway, acts as a co-activator of transcription of a number of genes, encoding enzymes involved in photopigment biosynthesis [39,40]. Arabidopsis TSPO is also linked to porphyrin metabolism and appears to act as a signal linking water-related stress to transient increases in gene expression of the plant stress hormone, absicisc acid .
We have proposed a model wherein a modest cholesterol load induces macrophage overexpression of TSPO mRNA and protein, facilitating cholesterol efflux to apoA-I, HDL and human serum, inducing the expression of NR1H3, PPARA, ABCA1, ABCG4 and APOE, and reducing macrophage cholesterol and triacylglycerol mass. In turn, loss of this protective TSPO-co-ordinated response, exemplified in the speculative model provided in Figure 6(C), might be part of the pathophysiology of ‘foam cell’ formation; certainly, chronic high fat, high cholesterol supplementation, and associated oxidative stress, decreases the TSPO-binding capacity in rodent hepatic and aortic tissues . Targeting mitochondrial TSPO activity or maintaining its function could prove therapeutically useful in regressing atherosclerotic lesions, particularly since bioavailable TSPO ligands, without obvious side effects, are currently in development for other disease conditions .
- ABC transporter
acyl-CoA: cholesterol acyltransferase
acyl-CoA-binding domain-containing protein
acetylated low-density lipoprotein
adenine nucleotide transporter
carnitine palmitoyltransferase 1A
cholesterol recognition/interaction amino acid consensus
liver X receptor
nuclear receptor 1H3 (gene encoding LXRα)
peripheral benzodiazepine receptor
steroidogenic acute regulatory protein
sterol-regulatory-element-binding transcription factor
sterol-regulatory-element-binding protein 2
signal transducer and activator of transcription 3
voltage-dependent anion channel
The present study was originated and designed by Annette Graham, in consultation with Janice Taylor; Janice Taylor performed the majority of the laboratory work, with additional input from Anne-Marie Allen. All three authors have been involved in preparation of the manuscript.
This work was supported by the British Heart Foundation [project grant number PG/04/0002871].