Peroxisome Proliferator-Activated Receptor-γ Coactivator 1-α Overexpression Prevents Endothelial Apoptosis by Increasing ATP/ADP Translocase Activity
Objective— Fatty acids increase reactive oxygen species generation and cell apoptosis in endothelial cells. The peroxisome proliferator-activated receptor-γ coactivator 1-α (PGC-1α) is a transcriptional coactivator that increases mitochondrial biogenesis and fatty acid oxidation in various cells. This study was undertaken to investigate the possible preventive effect of PGC-1α on endothelial apoptosis and its molecular mechanism.
Methods and Results— Treatment with linoleic acid in cultured human aortic endothelial cells increased reactive oxygen species generation and cell apoptosis. These effects appeared to be mediated by increases in cytosolic fat metabolites, ie, fatty acyl CoA, diacylglycerol, and ceramide, and consequent decreases in ATP/ADP translocase activity of adenine nucleotide translocator. Adenoviral overexpression of PGC-1α prevented linoleic acid-induced increases in reactive oxygen species generation and cell apoptosis in human aortic endothelial cells by increasing fatty acid oxidation, decreasing diacylglycerol and ceramide, and increasing ATP/ADP translocase activity. In isolated aorta, PGC-1α overexpression prevented linoleic acid-induced decrease in endothelium-dependent vasorelaxation, and this effect was abolished by adenine nucleotide translocator1 shRNA.
Conclusions— PGC-1α regulates reactive oxygen species generation and apoptosis in endothelial cells by increasing fatty acid oxidation and enhancing ATP/ADP translocase activity. Measures to increase PGC-1α expression or ATP/ADP translocase activity in vascular cells may aid in the prevention or treatment of atherosclerosis.
- adenine nucleotide translocator
- peroxisome proliferator-actived receptor-γ coactivator 1-α
- endothelial apoptosis
- mitochondrial membrane potential
- reactive oxygen species
Central obesity is associated with increased cardiovascular morbidity and mortality.1 Endothelial cell apoptosis and consequent impairment of endothelium-dependent vascular relaxation (endothelial dysfunction) are important early events in the pathogenesis of atherosclerosis.2 Increased levels of plasma free fatty acids in obesity may lead to endothelial cell apoptosis by increasing the accumulation of lipid metabolites and the generation of reactive oxygen species (ROS).3,4
Major sites of intracellular ROS generation are mitochondria and cell membrane NAD(P)H oxidase.4 The mitochondrial respiratory chain generates ROS when the electrochemical gradient between the mitochondrial inner membrane is high and the rate of electron transport is limited. In oxidative phosphorylation, electrons are transferred from electron donors, NADH or FADH, to electron acceptors in the mitochondrial electron transport chain. This transfer releases energy, and most of the energy is captured by proton pumps that build a proton gradient across the mitochondrial inner membrane (Δψm), which is the driving force for the phosphorylation of ADP to ATP by ATP synthase.5,6 However, sustained increase in Δψm impairs the flow of electrons through the ETC and increases the accidental transfer of electrons to oxygen to form superoxide.6,7
Δψm is determined by the balance between the export of protons from the mitochondrial matrix into the intermembranous space and the import of protons into the matrix through ATP synthase and uncoupling proteins (UCP).6 The ATP synthesized in the mitochondria is exchanged for cytosolic ADP by adenine nucleotide translocator (ANT) to provide a continuous supply of ADP. Intramitochondrial ADP deficiency resulting from reduced ATP/ADP translocase activity of ANT slows the rate of ATP synthesis and increases Δψm.8 Thus, ATP/ADP exchange by ANT is essential for the maintenance of ATP synthase activity and normal levels of Δψm.8 In addition, ANT is responsible for a significant portion of basal uncoupling or proton leak5,9 independent of its function in ATP/ADP translocase.10 The proton conductance of mitochondria depends on ANT content,11 and ANT and UCP constitute 2 major molecules that determine mitochondrial uncoupling.5
The peroxisome proliferator-activated receptor-γ coactivator 1-α (PGC-1α) is a transcriptional coactivator of nuclear receptors involved in cellular energy metabolism.12 PGC-1α increases mitochondrial biogenesis and fatty acid oxidation (FAO) in various cells, including endothelial cells.13 In addition, recent studies have indicated that PGC-1α is a major regulator of intracellular ROS generation; PGC-1α was shown to increase the expression of ROS-detoxifying enzymes.14,15 Here, we show an additional novel mechanism by which PGC-1α reduces cell apoptosis and ROS generation in endothelial cells; PGC-1α overexpression normalizes fatty acid-induced increases in mitochondrial membrane potential (Δψm) and ROS generation by increasing ATP/ADP translocase activity of ANT.
Materials and Methods
Expanded methods are available in the online data supplement.
Human aortic endothelial cells (HAEC; BioWhittaker) were cultured in endothelial growth medium-2 (BioWhittaker) supplemented with specific growth factors and 2% fetal bovine serum.
Transfection of ANT siRNA, Ad-PGC-1α, and Linoleic Acid Treatment
Fifty percent to 60% confluent HAEC were infected with adenoviruses carrying β-gal (Ad-β-gal) or PGC-1α (Ad-PGC-1α) at a titer of 5×106 pfu/mL for 1 hour at 37°C in DMEM without serum. Infected HAEC were then incubated in endothelial growth medium-2 with growth factors and 2% fetal bovine serum for 48 hours. After that, the media were replaced with M199 (BioWhittaker) without growth factors and with 1% fetal bovine serum for 1 hour, and then the cells were treated with 60 μmol/L of linoleic acid (LA) or same amounts of vehicle for indicated times. For the experiments using ANT1 siRNA, 50 nM of ANT1 siRNA or control siRNA was transfected into HAEC using LipofectAMINE 2000 (Invitrogen) 24 hours before Ad-PGC-1α infection.
Analysis of Apoptosis
Apoptosis was measured by various methods, including a cell death enzyme-linked immunosorbent assay kit examining cytoplasmic histone-associated DNA fragmentation (Roche Diagnostics), ApoAlert caspase 3 fluorescence assay kit (Clontech), and Western blots for cleaved caspase 3 and poly (ADP-ribose) polymerase.16,17
Measurement of ROS Levels
Intracellular ROS levels and mitochondrial-specific ROS generation was measured by flow cytometry using DCFH2-DA (Molecular Probes) and MitoSOX Red fluorescent dye (Molecular Probes), respectively (see online Methods).
Measurement of Δψm
The Δψmwas measured by flow cytometry using JC-1 (Molecular Probes; see online Methods).
Measurement of ATP/ADP Translocase Activity of ANT
ATP/ADP translocase activity was measured as previously described18 (see online Methods).
Western Blot Analysis
Protein expression in cells was measured by Western blot analysis as previously described.19
Real-Time Polymerase Chain Reaction Analysis
The mRNA expression in cells was measured by real-time polymerase chain reaction analysis (see online Methods).
Eight-week-old male Sprague-Dawley rats (Orient, Sungnam, Korea) weighing 250 to 300 g were housed in cages containing 4 rats per cage and allowed ad libitum access to water and food. All experimental procedures were approved by the Institutional Animal Care and Use Committee of the Asan Institute for Life Sciences.
Ex Vivo Measurement of Endothelium-Dependent Vasorelaxation
Endothelium-dependent vascular relaxation was measured as previously described.20 The thoracic aorta was excised from Sprague-Dawley rats after euthanization with pentobarbital, and then cleaned to remove fat and adhering tissue. The vessel was cut into several individual ring segments 2 to 3 mm in width. Adenoviral overexpression of PGC-1α or knockdown of ANT1 were achieved by infection of aortic rings with Ad-PGC-1α or Ad-ANT1 shRNA (6×106 pfu/mL each) for 30 minutes at 37°C in DMEM without serum. The tissues were subsequently incubated in medium containing 5% bovine serum albumin for 24 hours. After that, the aortic rings were pre-exposed to 60 μmol/L of LA for 2 hours, and contraction was induced by treatment with 300 μmol/L phenylephrine. Acetylcholine (from 10−9–10−5 mol/L) was then added serially to the bath to induce endothelium-dependent vasorelaxation. The tension was measured by an isotonic force displacement transducer (Hugo Sachs Elektronik KG D-7806) and recorded using a polygraph (Graphtec Linerecorder mark 8 WR3500).
All data are shown as means±SEM. Comparisons between 2 groups were performed using unpaired Student t tests and comparisons between multiple groups were performed by ANOVA. Time-dependent changes of ROS and Δψm were assessed by 2-way ANOVA. The significance of differences in vascular relaxation between the 4 experimental groups was assessed by ANOVA with repeated measures. Differences were classified as significant at P<0.05.
Expanded results are available in the online data supplement at http://atvb.ahajournals.org.
PGC-1α Overexpression Prevents LA-Induced ROS Generation and Apoptosis
Previous studies established that fatty acids increase ROS generation and cell apoptosis in endothelial cells.21,22 In accordance with these studies, incubation of HAEC with LA significantly increased intracellular and mitochondrial ROS generation and apoptosis (Figure 1, supplemental Figures I, II, available online at http://atvb.ahajournals.org). Adenoviral overexpression of PGC-1α (Ad-PGC-1α) prevented LA-induced increases in ROS generation and apoptosis (Figure 1, Figure I).
PGC-1α Overexpression Increases the Expression of Antioxidant Genes
LA had a slight but significant effect of increasing endogenous PGC-1α protein (Figure 2A). LA or glucose oxidase23 increased the mRNA and protein expression of antioxidant enzymes, including manganese superoxide dismutase, copper-zinc superoxide dismutase, catalase, glutathione peroxidase, and uncoupling protein 2 (UCP-2; Figure 2B, Figure III). This finding is consistent with the notion that oxidative stress can induce cellular antioxidant responses.24,25 The mRNA and protein expressions of the same antioxidant enzymes and UCP-2 were also significantly increased by Ad-PGC-1α (Figure 2B, supplemental Figure III)14,15 and reduced by PGC-1α–specific siRNA (supplemental Figure IV). This result suggests that endogenous PGC-1α plays a role in regulating the expression of antioxidant enzymes and UCP-2.14,15 However, in the presence of LA or glucose oxidase, Ad-PGC-1α did not further increase the expression of antioxidant enzymes or UCP-2 (Figure 2B, supplemental Figure III). This result suggests that the ability of Ad-PGC-1α to normalize LA-induced increases in intracellular ROS production and cell apoptosis cannot be explained by differences in the expression of antioxidant enzymes or UCP-2.
LA Decreases and PGC-1α Overexpression Increases ATP/ADP Translocase Activity by Affecting Intracellular Lipid Metabolites
Recent studies have shown that a significant increase in Δψm (hyperpolarization), which induces ROS production,26 is an earlier prerequisite for apoptosis.27,28 In our study, LA treatment significantly increased Δψm during the initial 4 hours, which was followed by depolarization after 20 hours (Figure 3A). Because intramitochondrial ADP deficiency, resulting from reduced ATP/ADP exchange, is known to slow the rate of ATP synthesis and to increase Δψm,8 we reasoned that decreased ATP/ADP translocase activity might be responsible for the hyperpolarization. In accordance with a previous study,29 ATP/ADP translocase activity, measured by 14C-ADP import, was significantly decreased after 4 hours of LA treatment (Figure 3B).
We next examined the mechanism underlying the LA-induced inhibition of ATP/ADP translocase activity. LA decreased FAO and increased intracellular levels of triglycerides, ceramide, and DAG (Figure 3C, D). In the basal state (ie, without LA), etomoxir, a carnitine palmitoyl transferase inhibitor, significantly decreased ATP/ADP translocase activity (data not shown). However, triasin and myriocin, inhibitors of fatty acyl CoA synthase and ceramide synthase, respectively, significantly reduced LA-induced effects on ATP/ADP translocase activity (Figure 3E). Collectively, these results suggest that increased levels of lipid metabolites, ie, fatty acyl CoA, DAG, and ceramide, may be involved in LA-induced decreases in ATP/ADP translocase activity.
PGC-1α overexpression significantly increased FAO and decreased intracellular triglycerides, ceramide, and DAG levels (Figure 3C, D). PGC-1α overexpression completely reversed LA-dependent decreases in ATP/ADP translocase activity (Figure 3B), which was associated with a complete prevention of LA-induced time-dependent changes in Δψm (Figure 3A).
Role of ANT1 in Cell Apoptosis and ROS Generation
Among the known ANT isoforms, ANT1 is predominantly expressed in the heart, skeletal muscle, and brain.30 We examined the effects of LA and PGC-1α on ANT1 expression in endothelial cells. LA significantly increased ANT1 expression (Figure 4A), and this effect was prevented by the antioxidant N-acetylcysteine (Figure 4B). However, N-acetylcysteine treatment did not affect ATP/ADP translocase activity with or without LA treatment (Figure 4C), suggesting that increased ROS generation with LA may be responsible for the increase in ANT1 expression and, more importantly, that changes in ANT1 expression cannot account for the decrease in ATP/ADP translocase activity with LA. PGC-1α also increased ANT1 expression but did not further increase ANT1 expression in the presence of LA (Figure 4A).
We next examined the effects of ANT1 siRNA (Figure 4D) on ROS generation and apoptosis. ANT1 siRNA significantly increased cell apoptosis and ROS generation (Figure 4E, F). This was associated with increased expression of antioxidant enzymes, ie, manganese superoxide dismutase, copper-zinc superoxide dismutase, catalase, glutathione peroxidase, and UCP-2 (supplemental Figure V), and changes in mitochondrial morphology (supplemental Figure VI). Taken together, these data suggest that reduced ANT1 expression/activity can cause increased cell apoptosis, ROS generation, and antioxidant enzyme expression, as seen with LA, and support the notion that LA increases cell apoptosis and ROS generation by reducing ATP/ADP translocase activity of ANT. Increased ANT1 expression observed with LA may be a compensatory response to increased intracellular ROS.
Increased ATP/ADP Translocase Activity Is Required for Effects of PGC-1α to Prevent LA-Induced Cell Apoptosis and ROS Generation
As already described, PGC-1α overexpression completely prevented LA-induced decreases in ATP/ADP translocase activity (Figure 3B). We tested whether this effect is critically required for the beneficial effects of PGC-1α to prevent LA-induced increases in cellular apoptosis and ROS generation. For this, we examined the effect of ANT1 siRNA on PGC-1α–dependent changes in ATP/ADP translocase activity, ROS generation, and apoptosis in the presence of LA. ANT1 siRNA almost completely abolished PGC-1α–induced changes in ATP/ADP translocase activity, cell apoptosis, ROS generation, and Δψm with LA (Figure 5 A–E). These findings indicate that PGC-1α prevents LA-induced ROS production and apoptosis in HAEC by improving ATP/ADP translocase activity of ANT.
ANT Mediates the Effects of PGC-1α on LA-Induced Endothelial Dysfunction
To investigate whether PGC-1α overexpression in aortic tissue improves endothelial dysfunction, we infected the aortic ring of Sprague-Dawley rat with Ad-PGC-1α or Ad-ANT1 shRNA ex vivo (Figure 6A). As reported previously,31 LA treatment significantly decreased endothelium-dependent vascular relaxation in the adenoviruses carrying β-gal–infected aortic rings (Figure 6B). Administration of nitric oxide synthase inhibitor l-NAME nearly completely inhibited acetylcholine-induced vasorelaxation, both in LA-treated and untreated aorta. Because LA treatment was associated with reduced vasorelaxation in response to acetylcholine, the net difference between l-NAME–treated and l-NAME–untreated vessels was less in the LA-treated than in LA-untreated vessels. These results suggest that vascular dysfunction by LA treatment is attributable to reduced NO bioavailability (supplemental Figure VIII). Ad-PGC-1α significantly inhibited LA-induced decreases in endothelium-dependent vasorelaxation compared to adenoviruses carrying β-gal (Figure 6B, C). This effect was significantly reduced by Ad-ANT1 siRNA (Figure 6C).
In this study, we confirmed that fatty acids (ie, LA) increase ROS production and apoptosis in cultured HAEC. These effects were accompanied by decreases in FAO and ATP/ADP translocase activity. In addition, we found that PGC-1α overexpression prevented fatty acid-induced increases in ROS production and apoptosis. These PGC-1α effects were accompanied by normalization of FAO and ATP/ADP translocase activity. Furthermore, a knockdown of ATP/ADP translocase activity (via ANT1 siRNA) led to an increase in ROS production and apoptosis, similar to the changes with LA, and abolished the effects of PGC-1α to prevent LA-induced increases in ROS production and apoptosis. Taken together, these data indicate ATP/ADP translocase activity of ANT as a major regulatory site of ROS production and apoptosis in endothelial cells affected by LA and PGC-1α overexpression.
Treatment of HAEC with LA caused significant increases in intracellular triglycerides, ceramide, and DAG concentrations, which were all normalized by PGC-1α overexpression. This is consistent with the well-known effect of PGC-1α to increase mitochondrial biogenesis and FAO, which would increase the clearance of cytosolic fat moieties. These data increase the possibility that fat metabolites may be responsible for the changes in ATP/ADP translocase activity induced by LA or PGC-1α overexpression. To support this idea, triascin, a fatty acyl-CoA synthase inhibitor, prevented LA-induced decreases in ATP/ADP translocase activity. In addition, myriocin, which inhibits ceramide synthesis, also prevented LA-induced decreases in ATP/ADP translocase activity.
ANT is considered a dual-edged sword. The prime function of ANT is the exchange of ATP and ADP across the inner mitochondrial membrane, which is rate-limiting for oxidative phosphorylation in the resting state. However, in states promoting apoptosis, ANT plays an important regulatory role in mitochondrial permeability transition pore opening,32 even though recent studies cast doubt on the central role of the ANT as a leading contender for the membrane component that forms the transmembrane channel of the mitochondrial permeability transition pore.33 In addition, ANT is responsible for a significant portion of basal uncoupling or proton leak5,9 independent of its function in ATP/ADP exchange. In fact, ANT and UCP are considered 2 major molecules that determine mitochondrial uncoupling.5
Whereas LA decreased ATP/ADP translocase activity, LA increased ANT1 expression. Fatty acids have long been known to increase mitochondrial uncoupling,34 and LA-induced increases in ANT1 and UCP expression may be a compensatory response to an increase in intracellular ROS. To support this, the present data show that LA-induced increases in ANT1 expression were completely prevented by N-acetylcysteine, a well-established thiol antioxidant. However, N-acetylcysteine treatment did not affect ATP/ADP translocase activity with or without LA treatment, suggesting that LA-induced decrease in ATP/ADP translocase activity is ROS-independent. A previous study suggested that PGC-1α may be involved in cold-induced upregulation of UCP and ANT in skeletal muscle.35 In agreement with this study, forced expression of PGC-1α increased the expression of ANT1 in endothelial cells (without LA treatment). However, this effect did not appear to be the major mechanism by which PGC-1α improves ROS generation and apoptosis in cells treated with LA because PGC-1α did not increase ANT1 or UCP-2 expression above the levels induced by LA. Despite a lack of effect on ANT1 expression, PGC-1α normalized the ATP/ADP translocase activity of ANT in cells treated with LA, and this was associated with the reversal of all of the changes induced by LA, including time-dependent changes in Δψm, ROS production, and apoptosis. In addition, knockdown of ATP/ADP translocase activity of ANT via ANT1 siRNA abolished all of these beneficial effects of PGC-1α. Taken together, these results suggest that PGC-1α–dependent enhancement of ATP/ADP translocase activity of ANT, which may be mediated by cytosolic fat metabolites as discussed, is critically required for the beneficial effects of PGC-1α on endothelial function.
Impaired endothelium-dependent vascular relaxation is generally considered a prerequisite for atherosclerosis.36 In this study, we examined the effect of ex vivo adenoviral overexpression of PGC-1α on endothelium-dependent vasorelaxation. Consistent with in vitro study, LA significantly impaired endothelium-dependent vasorelaxation. This inhibitory effect of LA treatment on vasorelaxation may be through reduction in the activity of endothelial nitric oxide synthase (supplemental Figure VIII). Ad-PGC-1α significantly inhibited LA-induced decreases in endothelium-dependent vasorelaxation compared to adenoviruses carrying β-gal, and this effect was significantly reduced by Ad-ANT1 shRNA. Collectively, these ex vivo results confirm antiatherogenic effects of PGC-1α in vascular endothelial cells.
In conclusion, we propose that fatty acids increase ROS generation and apoptosis in endothelial cells by reducing ATP/ADP translocase activity of ANT, which may be mediated by increased cytosolic fat metabolites. PGC-1α overexpression prevented fatty acid-induced changes in apoptosis, ROS generation, FAO, and Δψm, by enabling endothelial cells to cope better with a high lipid load, normalizing ATP/ADP translocase activity. These findings indicate ATP/ADP exchange activity as a crucial regulatory site for the effects of fatty acids or PGC-1α on fuel metabolism and cell apoptosis in endothelial cells. Measures to increase PGC-1α expression or ATP/ADP translocase activity in vascular endothelial cells may be useful in the prevention or treatment of atherosclerosis.
The authors thank Professor Shey-Shing Sheu (University of Rochester Medical Center) for insightful comments and discussion.
Sources of Funding
This work was supported by the Korea Science and Engineering Foundation (KOSEF) grants funded by the Ministry of Science and Technology (M10642140004-06N4214-00410 to K.U.L. and NRL M1040000000804J000000810 to J.Y.P.), and a grant (2008-122) from Asan Institute for Life Sciences, Seoul, Korea.
J.C.W. and J.-Y.P. contributed equally contributed to this work.
Received January 9, 2009; revision accepted November 23, 2009.
Artwohl M, Roden M, Waldhausl W, Freudenthaler A, Baumgartner-Parzer SM. Free fatty acids trigger apoptosis and inhibit cell cycle progression in human vascular endothelial cells. FASEB J. 2004; 18: 146–148.
Boudina S, Abel ED. Mitochondrial uncoupling: a key contributor to reduced cardiac efficiency in diabetes. Physiology. 2006; 21: 250–258.
Echtay KS, Esteves TC, Pakay JL, Jekabsons MB, Lambert AJ, Portero-Otin M, Pamplona R, Vidal-Puig AJ, Wang S, Roebuck SJ, Brand MD. A signalling role for 4-hydroxy-2-nonenal in regulation of mitochondrial uncoupling. Embo J. 2003; 22: 4103–4110.
Azzu V, Parker N, Brand MD. High membrane potential promotes alkenal-induced mitochondrial uncoupling and influences adenine nucleotide translocase conformation. Biochem J. 2008; 413: 323–332.
Valle I, Alvarez-Barrientos A, Arza E, Lamas S, Monsalve M. PGC-1alpha regulates the mitochondrial antioxidant defense system in vascular endothelial cells. Cardiovasc Res. 2005; 66: 562–573.
Khaled AR, Reynolds DA, Young HA, Thompson CB, Muegge K, Durum SK. Interleukin-3 withdrawal induces an early increase in mitochondrial membrane potential unrelated to the Bcl-2 family. Roles of intracellular pH, ADP transport, and F(0)F(1)-ATPase. J Biol Chem. 2001; 276: 6453–6462.
Lee KU, Lee IK, Han J, Song DK, Kim YM, Song HS, Kim HS, Lee WJ, Koh EH, Song KH, Han SM, Kim MS, Park IS, Park JY. Effects of recombinant adenovirus-mediated uncoupling protein 2 overexpression on endothelial function and apoptosis. Circ Res. 2005; 96: 1200–1207.
Russell JW, Golovoy D, Vincent AM, Mahendru P, Olzmann JA, Mentzer A, Feldman EL. High glucose-induced oxidative stress and mitochondrial dysfunction in neurons. FASEB J. 2002; 16: 1738–1748.
Park JY, Park KG, Kim HJ, Kang HG, Ahn JD, Kim HS, Kim YM, Son SM, Kim IJ, Kim YK, Kim CD, Lee KU, Lee IK. The effects of the overexpression of recombinant uncoupling protein 2 on proliferation, migration and plasminogen activator inhibitor 1 expression in human vascular smooth muscle cells. Diabetologia. 2005; 48: 1022–1028.
Ciapaite J, Bakker SJ, Diamant M, van Eikenhorst G, Heine RJ, Westerhoff HV, Krab K. Metabolic control of mitochondrial properties by adenine nucleotide translocator determines palmitoyl-CoA effects. Implications for a mechanism linking obesity and type 2 diabetes. FEBS J. 2006; 273: 5288–5302.
Esposito LA, Melov S, Panov A, Cottrell BA, Wallace DC. Mitochondrial disease in mouse results in increased oxidative stress. Proc Nat Acad Sci U S A. 1999; 96: 4820–4825.
Ray J, Noll F, Daut J, Hanley PJ. Long-chain fatty acids increase basal metabolism and depolarize mitochondria in cardiac muscle cells. Am J Physiol Heart Circ Physiol. 2002; 282: H1495–H501.