Chai, Hong MD, PhD; Yang, Hui MD, PhD; Yan, Shaoyu PhD; Li, Min PhD; Lin, Peter H MD; Lumsden, Alan B MD; Yao, Qizhi MD, PhD; Chen, Changyi MD, PhD
Highly active antiretroviral therapy (HAART) has greatly improved clinical outcomes for patients with HIV infection.1 Among all the treatments, HIV protease inhibitors (PIs) have been considered one of the most significant advances in the past 2 decades in controlling HIV infection.1,2 With marked improvement in the long-term survival, however, concern is growing about serious cardiovascular complications from the prolonged use of PIs. Some investigations have revealed a possible association between HIV PIs and metabolic abnormality including hyperlipidemia, lipodystrophy, and insulin resistance.3,4
Previous studies have shown that HIV PIs increased CD36-dependent cholesterol ester accumulation in macrophages in both in vitro and in vivo models independent of dyslipidemia.5 In addition, a clinical cross-sectional study of HIV patients revealed a correlation between the use of PIs and endothelial dysfunction independent of lipid levels.6 Furthermore, we have recently demonstrated that the HIV PI ritonavir (RTV) significantly impaired vasomotor activities through the increase of oxidative stress and the decrease of endothelial nitric oxide synthase (eNOS).7-9 Thus, there could be a direct effect of HIV PIs on endothelial cell functions. Direct comparison of 5 HIV PIs for their adverse effects is not available. It is possible that some HIV PIs have more potent effects than others under the same conditions such as drug concentrations, cell types, and culture environments.
The objective of this study was to determine and compare the effects of HIV PIs including RTV, amprenavir (APV), saquinavir (SQV), indinavir (IDV), and nelfinavir (NFV) on vasomotor functions as well as underlying molecular mechanisms in porcine coronary arteries. Specifically, the expression and activity of eNOS and the status of oxidative stress were investigated. This study may provide a better understanding of the mechanisms of HIV PI-associated vascular complications and suggest a new therapeutic strategy to control this clinical problem.
Tissue Harvest and Culture
Fresh porcine hearts were harvested from farm pigs at a local slaughterhouse. Right coronary arteries were dissected and cut into 5-mm rings. From each heart, several rings were allocated into 6 groups (dimethyl sulfoxide [DMSO], RTV, APV, SQV, IDV, and NFV, kindly provided by NIH AIDS Research and Reference Reagent Program, Germantown, MD). The rings were then incubated in Dulbecco modified Eagle medium (DMEM) with 15 μM of each drug or vehicle (DMSO) at 37°C and 5% CO2 for 24 hours. In separate experiments, rings were co-cultured with antioxidant seleno-L-methionine (SeMet, 100 μM; Sigma, St. Louis, MO) plus RTV (15 μM) in DMEM for 24 hours.
The myograph tension system used in our laboratory has been previously described.10,11 Briefly, rings were suspended between the wires of the organ bath chamber (Multi myograph system 700MO; Myo Technology, Aarhus, Denmark). Each ring was precontracted with 20 μL of thromboxane A2 analogue U46619 (10−7 M). After 60 to 90 minutes of contraction, a relaxation dose-response curve was generated by adding 60 μL of 5 cumulative additions of the endothelium-dependent vasodilator bradykinin (10−9, 10−8, 10−7, 10−6, and 10−5 M) every 3 minutes. In addition, 60 μL of sodium nitroprusside (SNP, 10−6 M) was added into the organ bath and endothelium-independent vasorelaxation was recorded.
Real-Time Polymerase Chain Reaction
Endothelial cells were purified from cultured rings by gently scraping with a surgical blade.10 Total RNA from endothelial cells was isolated using Tri-Reagent (Sigma). cDNA was generated by reverse transcription using iScript cDNA Synthesis Kit (Bio-Rad Laboratories, Hercules, CA). GAPDH was used as an internal control for eNOS expression. Porcine eNOS primers were: sense 5′-CCCTACAACGGCTCCCCTC-3′ and antisense 5′-GCTGTCTGTGTTACTGGATTCCTT-3′. Porcine GAPDH primers were: sense 5′-TGTACCACCAACTGCTTGGC-3′ and antisense 5′-GGCATGGACTGTGGTCATGAG-3′. Real-time polymerase chain reaction (PCR) was performed in an iCycler iQ real-time PCR detection system using iQ SYBR Green SuperMix Kit (Bio-Rad). PCR cycling conditions were set as follows: 95°C for 90 seconds, 40 cycles at 95°C for 20 seconds, and 60°C for 1 minute. Sample cycle threshold (Ct) values were determined from plots of relative fluorescence units (RFUs) vs. PCR cycle number during exponential amplification. Relative eNOS mRNA levels were calculated as 2−(40−Ct) and further normalized to GAPDH expression as 2−[Ct(GAPDH)−Ct(eNOS)].
Treated rings were fixed in 10% neutral buffered formalin and embedded in paraffin. Cross-sections were stained with monoclonal antibody against human eNOS (1:1000, BD Transduction Laboratories, Lexington, KY) using avidin-biotin complex immunoperoxidase procedure.
Detection of Superoxide Anion (O2−)
Levels of O2− produced by endothelial cells were measured by using the lucigenin-enhanced chemiluminescence method with Sirius Luminometer and FB12 software from Berthold Detection System GmbH (Pforzheim, Germany). The rings were cut open longitudinally and trimmed into approximately 5 × 5 mm pieces. Assay tube (12 × 75 mm) was filled with 500 μL of buffer and 25 μL of lucigenin (Molecular Probes, Eugene, OR). After gentle vortexing, the vessel segments were placed endothelium side down in the tubes to record signals from the endothelial layer. Time-based reading of the luminometer was recorded by FB12 software. The data in relative light units per second (RLU/s) for each sample were averaged between 5 and 10 minutes. The area of each vessel segment was measured using a caliper and used to normalize the data for each sample. Final data were represented as RLU/s/mm2.
Oxidative Fluorescent Microscopy
Dihydroethidium (DHE) (Molecular Probes, Eugene, OR) was used to observe the in situ levels of O2−. Treated rings were embedded in OCT compound immediately after culturing. Ten-micrometer-thick frozen sections were cut on a cryostat (Leica Microsystems, Bannockburn, IL) and mounted on glass slides. The tissue sections were covered in DHE (2 μM) and incubated at 37°C for 30 minutes, then viewed with an Olympus BX41 fluorescent microscope (Melville, NY).
NO release from vessel rings was determined by measuring the accumulation of its stable degradation products, nitrite and nitrate. Nitrate was detected after reduction to nitrite using nitrate reductase. Artery rings were cultured in DMEM medium (without phenol red; Biosource, Rockville, MD) with each of the 5 PIs, respectively, as well as RTV plus SeMet (100 μM) for 24 hours. Nitrite levels of the supernatant of artery ring cultures were measured by Griess reaction using a Nitric Oxide Assay kit (Calbiochem, San Diego, CA). The amount of nitrite formed was normalized to the protein content of the cultured rings.
Measurement of Nitrotyrosine
The protein (15 μg) from porcine artery endothelial cells was resolved electrophoretically by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (10% polyacrylamide). Next, the proteins were electrophoretically transferred to nitrocellulose filters. Nitrotyrosine was detected using a mouse monoclonal antibody (HM.11; Abcam, Cambridge, MA) diluted 1:1000. Blots were developed using enhanced chemiluminescence (ECL). The membranes were stripped and reprobed with antibody against β-actin. The relative protein level was expressed as the ratio of density of a major nitrotyrosine protein (45-kD band) and β-actin. Immunohistochemistry was also performed using the same antibody on 5-μM paraffin sections.
A total of 39 pig hearts were harvested and used for this study. Data from different groups were compared using a paired Student t-test (2-tailed). Significance was considered at P < 0.05. Data are reported as mean ± SEM.
Effect of HIV PIs on Vasomotor Function
Porcine coronary arteries were cultured for 24 hours with a clinically relevant dose of 15 μM of 5 different PI drugs (Table 1) and subsequently subjected to physiological contraction (U46619), endothelium-dependent (bradykinin), and endothelium-independent (SNP) relaxation assays (n = 6). In response to U46619, the contraction of the vessel rings was significantly reduced by 63% for RTV and 32% for SQV as compared with controls (P < 0.05, Fig. 1A). APV and NFV treatment also showed reduced vessel contractility, but this did not reach statistical difference (P > 0.05, Fig. 1A). IDV did not affect the vessel contractility (Fig. 1A). In response to bradykinin at 10−5 M, the endothelium-dependent relaxation of vessel rings was significantly reduced by 47% for RTV, 26% for APV, and 20% for SQV as compared with controls (P < 0.05, Fig. 1B). Although IDV and NFV also reduced the relaxation by 15% and 19%, respectively, there was no statistical difference as compared with controls (P > 0.05, Fig. 1B). In response to SNP, RTV treatment significantly reduced vasorelaxation by 24% (P < 0.05, Fig. 1C). Although APV, SQV, and NFV also slightly reduced SNP-induced relaxation, the difference was not statistically significant (P > 0.05, Fig. 1C). Interestingly, IDV increased SNP-induced relaxation, but it did not reach statistical significance as compared with controls (P = 0.1, Fig. 1C).
Effect of HIV PIs on eNOS Expression and NO Production
To determine whether eNOS expression was associated with the reduction of endothelium-dependent vasorelaxation in HIV PI treatments, eNOS levels of the endothelial cells of treated vessels were determined by real-time PCR (n = 6) and immunohistochemistry analysis (n = 3). The eNOS mRNA levels were significantly reduced by 34% for RTV, 43% for APV, and 26% for SQV as compared with controls (P < 0.05, Fig. 2A). IDV and NFV did not alter eNOS mRNA levels (P > 0.05, Fig. 2A). Immunoreactivity of eNOS of the endothelial cells in RTV-, APV-, and SQV-treated vessel rings was also substantially reduced, but it was much less affected in IDV- and NFV-treated vessels as compared with controls (Fig. 2B). The NO production in rings after PI treatment was also investigated (n = 8). Total nitrite accumulation in the supernatant of cultured vessel rings is shown in Figure 2C. In the control group, the basal nitrite level was 18.8 μM/mg protein. It was significantly reduced by 33.5% for RTV and 28.7% for APV, as compared with controls (P < 0.01). SQV, IDV, and NFV also decreased nitrite levels to a lesser degree (Fig. 2C).
Effect of HIV PIs on O2− Production
Oxidative stress is one of the important mechanisms of endothelial dysfunction and vascular injury. To determine whether this mechanism is involved in HIV PI-induced vasomotor dysfunction, O2− production was analyzed by a lucigenin-enhanced chemiluminescence assay. The O2− levels of the endothelial layer of vessel rings were significantly increased by 47% for RTV and 52% for APV as compared with control samples (n = 7, P < 0.05, Fig. 3A). SQV, IDV, and NFV did not affect O2− levels (Fig. 3A). Differences in O2− levels were also investigated using oxidative fluorescent dye (DHE) staining of vessel sections. DHE becomes ethidium bromide upon interaction with O2− and produces red fluorescence when excited. There was a marked increase in red fluorescence (DHE staining) in RTV-and APV-treated samples in both endothelial and smooth muscle layers as compared with control samples (n = 3, Fig. 3B), which is consistent with the results from the lucigenin-enhanced chemiluminescence assay.
Effect of Antioxidant SeMet on RTV-Induced O2− Production and Nitrite Levels
SeMet is known as an antioxidant and has been shown to increase the activity of glutathione peroxidase in endothelial cells.12 The effect of antioxidant SeMet on RTV-induced O2− production was investigated. Because RTV was the most potent HIV PI, which caused vasomotor dysfunction and downregulation of eNOS expression, it was chosen as the representative drug of HIV PIs for further mechanistic investigations. RTV significantly increased endothelial O2− production by 47% as compared with controls (12.415 ± 0.902 vs. 8.456 ± 0.810 RLU/s/mm2, n = 6, P < 0.05), whereas SeMet and RTV coculture showed similar levels of O2− to control vessels (8.670 ± 1.129 vs. 8.456 ± 0.810 RLU/s/mm2, P > 0.05). SeMet (100 μM) alone did not affect O2− levels as compared with controls (7.875 ± 0.039 vs. 8.456 ± 0.810 RLU/s/mm2, P > 0.05). In addition, SeMet also significantly blocked the effect of RTV on NO production. Nitrite levels were 18.78 ± 1.45, 12.47 ± 0.71, and 18.00 ± 0.60 μM/mg protein in the supernatants of control, RTV-, and RTV-plus-SeMet-treated pig coronary artery rings, respectively.
Effects of RTV on Nitrotyrosine Protein Levels
Peroxynitrite (ONOO−) is a powerful oxidant and cytotoxic agent formed from the reaction between NO and superoxide. The best-known property of ONOO− is its ability to nitrite tyrosine residues in proteins.13 Nitrotyrosine is generally viewed as a molecular marker for the actions of peroxynitrite (ONOO−). Nitrotyrosine immunoreactivity was substantially increased in RTV-treated vessels, whereas SeMet attenuated this effect of RTV (Fig. 4A). Western blot analysis showed there were 4 nitrotyrosine protein bands identified, with 45, 60, 70, and 80 kD, respectively. The intensity of 45- and 60-kD nitrotyrosine proteins was substantially increased in RTV-treated vessels as compared with controls (Fig. 4B). The signal for 2 other protein species (80 and 70 kD) was too low to make a reliable comparison among groups, although 70 kD showed a slight increase in the group of RTV plus SeMet (Fig. 4B). The 45-kD band was chosen for density measurement based on the fact that it was the strongest band and had the lowest noise/signal ratio. Thus, comparison of band density ratios between the 45-kD band and β-actin could be more sensitive than that between whole blot and β-actin. Specifically, RTV significantly increased 45-kD nitrotyrosine protein by 85% as compared with controls (P < 0.05, n = 3, Fig. 4C). By contrast, SeMet successfully blocked RTV-increased nitrotyrosine levels (Fig. 4C).
Effect of SeMet on RTV-Induced Vascular Injury
To further confirm whether oxidative stress is involved in HIV PI-induced vessel injury, the effect of SeMet on vasomotor alteration was investigated. When vessel rings were cultured with RTV alone, the maximal contraction was decreased by 60%, the bradykinin (10−5 M)-induced relaxation was decreased by 50%, and the SNP (10−6 M)-induced relaxation was also decreased by 40% as compared with controls (P < 0.05, Fig. 5). SeMet completely reversed the effect of RTV on these abnormalities of the vasomotor function (Fig. 5), however.
This study has compared the direct effects of 5 HIV PIs on coronary artery injury as well as eNOS expression and superoxide anion production. Our results demonstrated that in the model of porcine coronary artery cultures, the HIV PIs RTV, APV, and SQV have more detrimental effects than IDV and NFV, resulting in vasomotor dysfunction, eNOS downregulation, and O2− overproduction. Furthermore, antioxidant therapy could completely block these adverse effects of HIV PIs.
The pharmacokinetic properties of RTV, APV, SQV, IDV, and NFV have been evaluated in both healthy volunteers and AIDS patients. The maximal plasma concentrations of clinical doses of RTV, APV, and IDV are about 15 μM, whereas those for SQV and NFV are 3.7 and 6.0 μM, respectively (Table 1).14 In a previous study, we determined a dose-response curve of RTV on endothelial function.9 RTV at 7.5 μM showed no significant effect, whereas RTV at 15 and 30 μM showed a significant decrease in both vasomotor function and eNOS expression. In the current study, we have carefully chosen the drug concentration of 15 μM for all 5 PIs to simulate a clinical situation and make a meaningful comparison. At this identical condition, RTV significantly impairs vessel contractility, endothelium-dependent vasorelaxation, and endothelium-independent vasorelaxation, indicating RTV-induced smooth muscle cell and endothelial cell injury. APV and SQV had similar but less potent effects on vasomotor function as compared with RTV. IDV and NFV had very limited effects on vasomotor function in porcine coronary arteries. These data are consistent with clinical observations showing impaired flow-mediated vasorelaxation of brachial arteries in HIV patients using PIs.6 Direct effects shown in this study and indirect adverse effects of HIV PIs by clinical dyslipidemia could contribute to vascular complications. Our study provides evidence of these direct effects of HIV PIs. Indeed, we previously demonstrated a direct cytotoxic effect of RTV on endothelial cell culture.15 Recently, a direct effect of HIV PIs (ie, RTV and APV) on macrophage cholesterol ester accumulation has been reported, and the most potent effects have been seen with RTV.5
A key process in the early pathogenesis of atherosclerosis is diminished bioavailability of the endothelium-derived signaling molecule NO, which is generated by eNOS and is a potent vasodilator with multiple additional cardiovascular functions.16 Both basic and clinical studies have shown that eNOS plays a very important role in cardiovascular disease.17 In this study, HIV PI-induced vessel injury may be due to the decrease in eNOS expression and NO production. Indeed, the real-time PCR data showed a 34% to 43% decrease in eNOS mRNA expression in RTV- and APV-treated vessel rings, and immunohistochemistry data also showed a decrease in eNOS protein levels. In addition, APV- and RTV-treated groups showed a marked decrease of NO, which can be reversed by co-culturing with an antioxidant in the porcine coronary artery culture model.
Oxidative stress contributes to mechanisms of vascular dysfunction.18 Endothelium can generate a substantial amount of O2−. Nitric oxide reacts with O2− to form peroxynitrite anion (ONOO−), which subsequently decomposes to form the highly reactive hydroxyl radical (OH). This interaction occurs approximately 3 times faster than the rate for O2−with superoxide dismutase, and therefore, a certain portion of NO may always be reacting with O2−and thus become unavailable for other biologic functions.19 Evidence from cell culture and animal studies suggests that overproduction of O2− and subsequent oxidative inactivation of NO may both be important in the pathogenesis of atherosclerosis.20 In our study, about a 50% increase in O2− production was shown in RTV- and APV-treated samples as compared with controls. Likewise, an increase in endothelial cell DHE staining was observed in vessels treated with RTV and APV, indicating an increase in O2− production. Because RTV showed the most potent effect on vasomotor functions as compared with other HIV PIs, it was chosen as a representative of HIV PIs to further investigate protein nitration and the role of the antioxidant SeMet. Western blot and immunohistochemistry showed an elevated level of nitrotyrosine in RTV-treated rings. This increase can be reversed by co-culturing with the antioxidant SeMet. Furthermore, when rings were co-cultured with RTV and antioxidant SeMet, endothelium-dependent vasorelaxation as well as vessel contractility and endothelium-independent vasorelaxation were returned to the normal levels of control vessels. SeMet is a commonly used antioxidant. We have used a single concentration of 100 μM of SeMet based on our previous experience and publications in this porcine artery culture model.21,22 SeMet (100 μM) alone did not affect endothelial function including vasomotor activity and eNOS expression in our studies. Other studies also used this concentration of SeMet.23-25 In a previous study, we have tested other antioxidants including curcumin26 and ginsenosides,8 which also significantly blocked RTV-induced endothelial dysfunction in the porcine coronary arteries. These data are well correlated with O2− production in these vessels, indicating that oxidative stress may be the major mechanism in HIV PI-induced vascular injury.
Five HIV PIs have some effects in common and some differences. For example, RTV and APV showed more potent effects on vasomotor function relative to others. Both drugs significantly reduced endothelium-dependent relaxation, but APV had no effects on contraction or endothelium-independent relaxation. In contrast, both drugs significantly reduced the levels of eNOS mRNA and nitrites as well as enhanced generation of superoxide radicals. The underlying reasons for these differences between RTV and APV are not clear. Besides oxidative stress and NO availability affecting HIV PI-induced vascular injury, other mechanisms may exist. The data regenerated from RTV may be applicable to other HIV PIs including APV. This issue must be directly verified by direct experiments with these specific HIV PIs in future investigations. In addition, cytotoxicity of RTV could play a role in vascular injury. The viability of RTV-treated vessels was not directly measured, however. Instead, the potent mechanism of RTV-induced vessel injury was studied, and it was related to oxidative stress and eNOS downregulation as compared with the untreated vessels as controls. Antioxidant SeMet could effectively block these effects of RTV on porcine coronary arteries.
HIV PIs tested in vitro may have different biologic activities in vivo because of plasma protein-binding properties of these drugs. In fact, all of the HIV PIs tested except IDV are highly bound to serum proteins, and RTV, SQV, and NFV are >97% bound.14 For example, using a reasonable Cmax concentration of 2500 ng/mL for SQV and 97% plasma protein binding,14 the free plasma drug concentration is calculated to be 0.11 μM. This in vitro concentration of 15 μM of SQV represents 136-fold the calculated free plasma Cmax concentration. Similarly, the concentration of NFV tested in vitro represents approximately 160-fold the calculated free plasma concentration. Thus, the 15-μM concentration of tested HIV PIs in vitro may represent the excessive (supratherapeutic) concentrations for some tested HIV PIs after consideration of free plasma concentrations in vivo. This concern could be a limitation of current investigation in vitro. Kinetics between plasma-bound and free PIs in vivo are not clear. The impact of protein-bound PIs on endothelial functions is also not clear. Further investigations including a dose range of HIV PIs and in vivo models as well as human cells or tissues are warranted. Whereas the potency of the PIs on vasomotor function differed, we may also consider the therapeutic concentrations of the PIs to define the “therapeutic window” in regards to potential effects on vasomotor function.
In summary, we have demonstrated that several HIV PIs can significantly cause vasomotor dysfunction, decrease eNOS expression, and increase O2− production in porcine coronary arteries. RTV and APV showed more potent effects than SQV. However, IDV and NFV had very limited effects in the same condition. Antioxidant SeMet can successfully block RTV-induced effects, indicating a major molecular mechanism and a potential therapeutic strategy in HIV patients with antiviral therapy. Further investigations including in vivo experiments are warranted to gain a better understanding of this important issue.
The following reagents were obtained through the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH: ritonavir, amprenavir, saquinavir, indinavir, and nelfinavir.
1. Leen CL. Perspectives on HAART: switch maintenance therapy. Int J STD AIDS
2. Barbaro G. Highly active antiretroviral therapy and the cardiovascular system: the heart of the matter. Pharmacology
3. Rhew DC, Bernal M, Aguilar D, et al. Association between protease inhibitor use and increased cardiovascular risk in patients infected with human immunodeficiency virus: a systematic review. Clin Infect Dis
4. Calza L, Manfredi R, Chiodo F. Hyperlipidaemia in patients with HIV-1 infection receiving highly active antiretroviral therapy: epidemiology, pathogenesis, clinical course and management. Int J Antimicrob Agents
5. Dressman J, Kincer J, Matveev SV, et al. HIV protease inhibitors promote atherosclerotic lesion formation independent of dyslipidemia by increasing CD36-dependent cholesteryl ester accumulation in macrophages. J Clin Invest
6. Stein JH, Klein MA, Bellehumeur JL, et al. Use of human immunodeficiency virus-1 protease inhibitors is associated with atherogenic lipoprotein changes and endothelial dysfunction. Circulation
7. Conklin B, Fu W, Lin P, et al. HIV protease inhibitor ritonavir decreases endothelium-dependent vasorelaxation and increases superoxide in porcine arteries. Cardiovasc Res
8. Chai H, Zhou W, Lin PH, et al. Ginsenosides block HIV protease inhibitor ritonavir-induced vascular dysfunction of porcine coronary arteries. Am J Physiol Heart and Circ Physiol
9. Fu W, Chai H, Yao Q, et al. Effects of HIV protease inhibitor ritonavir on vasomotor function and endothelial nitric oxide synthase expression. J Acquir Immune Defic Syndr
10. Fu W, Conklin BS, Lin PH, et al. Red wine prevents homocysteine-induced endothelial dysfunction in porcine coronary arteries. J Surg Res
11. Surowiec SM, Conklin BS, Li JS, et al. A new perfusion culture system used to study human vein. J Surg Res
12. Jornot L, Junod AF. Differential regulation of glutathione peroxidase by selenomethionine and hyperoxia in endothelial cells. Biochem J
13. Koppenol WH, Moreno JJ, Pryor WA, et al. Peroxynitrite, a cloaked oxidant formed by nitric oxide and superoxide. Chem Res Toxicol
14. Staff ME. Physicians' Desk Reference 2004,
57th ed. Montvale, NJ: Medical Economics; 2004.
15. Zhong DS, Lu XH, Conklin BS, et al. HIV protease inhibitor ritonavir induces cytotoxicity of human endothelial cells. Arterioscler Thromb Vasc Biol
16. Vane JR, Anggard EE, Botting RM. Regulatory functions of the vascular endothelium. N Engl J Med
17. Ganz P, Vita JA. Testing endothelial vasomotor function: nitric oxide, a multipotent molecule. Circulation
18. Nedeljkovic ZS, Gokce N, Loscalzo J. Mechanisms of oxidative stress and vascular dysfunction. Postgrad Med J
19. Beckman JS, Koppenol WH. Nitric oxide, superoxide, and peroxynitrite: the good, the bad, and ugly. Am J Physiol
20. Ballinger SW, Patterson C, Yan CN, et al. Hydrogen peroxide- and peroxynitrite-induced mitochondrial DNA damage and dysfunction in vascular endothelial and smooth muscle cells. Circ Res
21. Kougias P, Chai H, Lin PH, et al. Adipocyte-derived cytokine resistin causes endothelial dysfunction of porcine coronary arteries. J Vasc Surg
22. Zhou W, Chai H, Lin PH, et al. Ginsenoside Rb1 blocks homocysteine-induced endothelial dysfunction in porcine coronary arteries. J Vasc Surg
23. Jornot L, Junod AF. Differential regulation of glutathione peroxidase by selenomethionine and hyperoxia in endothelial cells. Biochem J
24. Ody C, Junod AF. Effect of variable glutathione peroxidase activity on H2
-related cytotoxicity in cultured aortic endothelial cells. Proc Soc Exp Biol Med
25. Housset B, Ody C, Rubin DB, et al. Oxygen toxicity in cultured aortic endothelium: selenium-induced partial protective effect. J Appl Physiol
26. Chai H, Yan S, Lin PH, et al. Curcumin blocks HIV protease inhibitor ritonavir-induced vascular injury in porcine coronary arteries. J Am Coll Surg
© 2005 Lippincott Williams & Wilkins, Inc.