|Home | About | Journals | Submit | Contact Us | Français|
In the vasculature, reactive oxidant species including reactive oxygen, nitrogen, or halogenating species, and thiyl, tyrosyl, or protein radicals, may oxidatively modify lipids and proteins with deleterious consequences for vascular function. These biologically active free radical and non-radical species may be produced by increased activation of oxidant-generating sources and/or decreased cellular antioxidant capacity. Once formed, these species may engage in reactions to yield more potent oxidants that promote transition of the homeostatic vascular phenotype to a pathobiological state that is permissive for atherothrombogenesis. This dysfunctional vasculature is characterized by lipid peroxidation and aberrant lipid deposition, inflammation, immune cell activation, platelet activation, thrombus formation, and disturbed hemodynamic flow. Each of these pathobiological states is associated with an increase in the vascular burden of free radical species-derived oxidation products and, thereby, implicates increased oxidant stress in the pathogenesis of atherothrombotic vascular disease.
Atherosclerosis and its attendant morbidity and mortality remain a global health issue with more than 17.5 million worldwide deaths attributable to the cardiovascular complications of this disease . The common risk factors for developing atherosclerotic cardiovascular disease, hyperlipidemia, hypertension, tobacco use, diabetes mellitus, and family history, alone or in combination, are associated with a pathophysiological vascular phenotype. As such, the dysfunctional vasculature becomes susceptible to lipid deposition, thrombus formation, immune cell infiltration, inflammation, and aberrant hemodynamic profiles. At a mechanistic level, a number of hypotheses have been advanced to unite these features; central to these theories is the tenet that increased oxidant stress perturbs vascular homeostasis to promote atherogenesis.
The importance of oxidant stress in atherothrombotic cardiovascular disease is highlighted by the observation that increased markers of oxidant stress have been shown to predict coronary heart disease. The F2-isoprostanes, formed by the free-radical catalyzed peroxidation of phospholipid-bound arachidonic acid and released into the circulation, are considered reliable markers of oxidant stress [2–4]. Elevated levels of F2-isoprostanes have been detected in patients with coronary heart disease risk factors, including hyperlipidemia, diabetes mellitus, hypertension, tobacco use, and obesity [5–9]. Furthermore, levels of both urinary and plasma F2-isoprostane have been shown to correlate with the number of coronary artery disease risk factors present [10, 11]. In a matched case-control study of 93 coronary heart disease patients and 93 control subjects, urinary F2-isoprostane levels were elevated in patients compared to control subjects (139 vs. 77 pmol/mmol creatinine, p<0.001), and multivariate analysis revealed that individuals with F2-isoprostane levels in the highest tertile had a 30.8-fold increased risk of coronary heart disease . In a similar case-control study of 50 patients with angiographically significant epicardial coronary artery disease compared with 54 patients with insignificant atherosclerosis, plasma levels of F2-isoprostanes were increased significantly in patients with coronary artery disease (9.4 ± 4.6 vs. 6.2 ± 3.4 µmol/mol arachidonate, p<0.001). Multivariable analysis adjusted for Framingham Risk Score revealed that F2-isoprostane levels remained an independent predictor of angiographic coronary artery disease risk (OR = 9.7, 95% CI: 2.6 – 36.9, p=0.016) . Interestingly, the methodology used in this study allowed for the simultaneous measure of specific fatty acid oxidation products, including individual hydroxy-octadecadienoic acids (HODEs) and hydroxy-eicosatetraenoic acids (HETEs). Measures of these markers revealed no significant difference between cases and controls with respect to 9-HODE and 13-HODE levels; however, there was a significant increase in 9-HETE in patients with coronary artery disease (8.7 ± 3.9 vs. 6.8 ± 3.6 µmol/mol arachidonate, p=0.011), and 9-HETE levels were also shown to predict angiographic disease (OR = 4.8, 95% CI: 1.3 – 17.1, p=0.016) .
Nitrated protein tyrosine residues also predict coronary heart disease risk. In a case-control study of patients with coronary artery disease, nitrotyrosine levels were found to be significantly higher in patients with coronary artery disease than in controls (9.1 vs. 5.2 µmol/mol tyrosine, p<0.001), and individuals with the nitrotyrosine levels in the highest quartile were found to have an increased risk of disease (OR = 6.1, 95% CI: 2.6 – 14.0, p<0.001), even after adjustment for Framingham Risk Score (OR = 5.4, 95% CI: 2.0–14.3, p<0.001) . Baseline serum levels of lipid hydroperoxides were also found to be an independent predictor for the development of major adverse cardiovascular events in a study of 634 individuals followed over three years (HR = 2.23, 95% CI for relative risk; 1.44–3.44, p<0.0003) . Thus, atherosclerosis and the risk for developing disease are associated with increases in several indices of systemic oxidant stress; however, it should be recognized that these studies merely define a correlation between indices of oxidant stress and atherothrombotic vascular disease and do not establish oxidant stress as a causative determinant for atherosclerosis.
Oxidant stress broadly defines the redox state achieved when there is an imbalance between antioxidant capacity and reactive oxidant species (ROS), such as reactive oxygen, nitrogen, or halogenating species, and free radical species such as thiyl, tyrosyl, or protein radicals. Within the cell, ambient levels of some ROS are utilized as signaling molecules to maintain basal cellular functions. In contrast, when reactive oxidants and free radicals are generated in the absence of a physiological stimulus, small molecule antioxidants are depleted, or antioxidant enzymatic systems are overwhelmed, there is a net increase in biologically active ROS and oxidant stress ensues. In blood vessels, oxidant stress has deleterious consequences for basal vascular function that have been associated with atherosclerosis. It is, therefore, not surprising that the cellular mechanisms that result in vascular redox imbalance leading to an increase in oxidant stress have been implicated in the pathogenesis of atherothrombosis.
In biological systems, a diverse array of free radical and non-radical reactive species may be generated, which, under pro-atherogenic conditions, may lead to lipid and protein modification to accelerate the disease process. Basal oxidative cellular metabolism, through the activation of enzymes that produce superoxide anion (O2•−) and/or as a byproduct of mitochondrial respiration, generate a number of oxygen-derived free radical species of importance to atherogenesis.
Superoxide anion is a one-electron reduced product of molecular oxygen that may be protonated to form a hydroperoxyl radical (HOO•: pKa = 4.69 for HOO• O2•− + H+) . Hydroperoxyl radicals have been shown to initiate lipid peroxidation in the absence of free transition metal ions as demonstrated in EtOH/H2O-fatty acid dispersions by the abstraction of a bis-allylic H• from polyunsaturated fatty acids [16, 17]. Using a stopped flow technique, the rate constants for the reaction of HOO• and linoleic, linolenic, and arachidonic acids were found to be 1.2 × 103 M−1s−1, 1.8 × 103 M−1s−1, and 3.0 × 103 M−1s−1, respectively . Superoxide may spontaneously dismutate to oxygen and the nonradical oxidant hydrogen peroxide (H2O2) in aqueous environments at a rate of ~2 × 105 M−1s−1 at pH 7.4 or undergo enzymatical dismutation by the superoxide dismutases (SOD), which at physiological pH, increase the reaction rate over the spontaneous dismutation of O2•− ~10,000-fold [19–22].
Superoxide may react with other free radicals such as nitric oxide (•NO) to form peroxynitrite anion (ONOO−), or react with non-radical derivatives by abstracting H• from a −C-H, −O-H, or −S-H moiety that exists in the non-radical species. These reactions typically involve low-molecular-weight antioxidants and enzyme cofactors, nucleic acids, lipids, proteins, and sugars [23–26]. At neutral pH, O2•− may also oxidize [4Fe-4S] clusters in enzymes to release Fe(II) allowing for the reduction of H2O2 to form a hydroxyl radical (•OH) through Fenton chemistry [27–29]. The •OH is a potent oxidant [E0’ (•OH, e−/H2O) = 2.31 V, vs. NHE at pH 7.0] that reacts with organic compounds at a diffusion-limited rate (k = ~108–1010 M−1s−1); samples of the lipid-rich core of atherosclerotic lesions obtained at autopsy confirmed that these specimens were able to generate •OH in the presence of H2O2 and ascorbate [30–32].
Reactions between these oxygen-derived free radicals and non-radical species may also participate in chain reactions as occurs with lipid peroxidation. Here, a free radical such as •OH will abstract H• from a carbon of a fatty acyl side chain leaving the remaining carbon radical accessible to molecular oxygen to form a lipid peroxyl radical (LOO•). This newly generated LOO• is itself highly reactive and propagates the chain reaction by reacting with nearby lipid moieties (L-H) to yield L• and LOOH; the newly generated L•, in turn, reacts with molecular oxygen to yield another LOO• and propagate the chain reaction .
Lipid peroxidation may also be initiated by ONOO−. Studies performed with soybean phosphatyidylcholine liposomes incubated with ONOO− demonstrated malondialdehyde and conjugated diene formation that was not dependent upon Fe2+/Fe3+ to initiate or enhance lipid peroxidation . Further study with human low-density lipoproteins (LDL) exposed to the ONOO− (0.125 – 1 mmol/L) or the ONOO−-generating agent sydnonimine (0.5 or 1 mmol/L) revealed that ONOO− promoted lipid peroxidation, increased the formation of F2-isoprostanes, and converted the lipoproteins to more negatively charged forms [34, 35].
Nitric oxide is a heterodiatomic free radical that is a key determinant of vascular homeostasis. In the vasculature, •NO-derived reactive nitrogen species react with aromatic amino acids, lipids, and thiols resulting in lipid and protein modification [36– 40]. The terminal oxidation product of •NO is nitrite (NO2−), formed in a series of reactions that generates the intermediate N2O3, a mild oxidant [calculated E0’ (N2O3/•NO, NO2−) = 0.8 V, vs. NHE at pH 7.0] which nitrosates thiols and other nucleophiles [41–44] Nitric oxide also reacts with oxygen (autooxidation) to form nitrogen dioxide (•NO2); the kinetics of •NO2 formation are second order with respect to •NO and first order with respect to oxygen (k = 7 × 106 M−2s−1), indicating that oxidation of •NO is strongly dependent upon its concentration [44–47]. In addition, head space gas analysis of helium-swept reaction mixtures has also demonstrated that •NO2 may be formed by leukocyte peroxidases utilizing H2O2 and NO2− as substrates . Nitrogen dioxide is a highly reactive species, and has been shown to abstract hydrogen atoms from unsaturated lipids, as well as participate in addition and abstraction reactions with cyclohexane, methyl oleate, linoleate, and linolenate [49–52].
The reaction of •NO with O2•− occurs in a nearly diffusion-limited manner (k = 6.7 × 109 M−1s−1) to yield ONOO− . Peroxynitrite may be protonated to form peroxynitrous acid (ONOOH), which has a pKa ~ 6.8 and may account for up to approximately 20% of total ONOO− at physiological pH [54, 55]. In the trans-configuration, ONOOH is a highly reactive and strong oxidizing species; however, ONOOH undergoes rapid homolytic cleavage with a half-life of 1.9 sec at pH 7.4 to yield •NO2 and •OH, which participate in oxidation, nitrosation, and nitration reactions [56, 57]. Peroxynitrite also reacts readily with CO2 to generate a nitrosoperoxycarbonate anion adduct (ONOOCO2−) that is unstable and rearranges to form a nitrocarbonate anion or generates •NO2 and carbonate anion as a pair of caged radicals that may escape the solvent cage as free radicals. Nitrocarbonate anion, in turn, may undergo hydrolysis to yield nitrate (NO3−) and carbonate . As the concentration of plasma CO2 is high (~ 1.3 mM), the reaction rate of CO2 with ONOO− is approximately 60-fold faster than that of the reaction of ONOO− with thiols [58, 59]. The reaction between ONOO− and either O2•− or •NO has been suggested to be thermodynamically feasible with a calculated Gibbs free energy [ΔrxnG°’ (kcal/mol)] of −36 and −27, respectively, for the reaction at pH 7.0 and 25°C [54, 60]. Evidence also suggests that O2•− or •NO-mediated consumption of ONOO− may be indirect and a consequence of the reaction of the ONOOH homolytic cleavage products •NO2 and •OH with •NO leading to the formation of N2O3 and HNO2, respectively, and with O2•− to yield O2NOO−, and −OH and O2, respectively . Peroxynitric acid (HO2NOO−/O2NOO−) is unstable and decomposes to HO2• and •NO2 with •NO2 reacting with itself in water to form NO2−, NO3−, and H+ . Peroxynitrite may also react with H2O2 to form NO2− and H2O .
Nitric oxide may also react with other peroxyl radicals react to form lipid peroxynitrite compounds (LOONO) in reactions that are near diffusion-limited (k = 1 – 3 × 109 M−1s−1) [37, 64]. Experimental evidence also indicates that •NO may react with H2O2 to generate •OH; however, in this study it was not determined if •NO reacted directly with H2O2 or if an intermediate yielded •OH by homolysis . In aqueous solutions, •NO may also react with •OH itself at a diffusion-controlled rate (k = 1 × 1010 M−1s−1) to generate HNO2 .
Reactions between •NO-derived radical species and proteins or lipids lead to post-translational modification(s) that influence function and have been associated with atherosclerosis [67–70]. In particular, the presence of 3-nitrotyrosine is evidence of reactive nitrogen species-mediated modifications that occur under conditions of oxidant stress. Tyrosine nitration may occur via the addition of •NO2 to a tyrosyl radical; the reaction between •NO and a tyrosyl radical to form 3-nitrosotyrosine, which is oxidized via the intermediacy of an iminoxyl radical to yield 3-nitrotyrosine; and/or by the addition of a nitrating species similar to NO2+ to the aromatic ring of tyrosyl residues . Early studies using monoclonal and polyclonal antibodies revealed extensive 3- nitrotyrosine immunostaining and protein tyrosine nitration in fatty streaks in coronary arteries of young subjects at autopsy and in human atherosclerotic plaques . Later studies using gas chromatography/mass spectrometry analysis of LDL isolated from human atherosclerotic lesions revealed that 3-nitrotyrosine levels were 90-fold greater than that observed in human plasma LDL . Following these seminal studies, 3-nitrotyrosine formation has been detected in human macrophages and vascular smooth muscle cells of excised carotid plaques ; in atherosclerotic human coronary artery or peripheral arterial vessels ; in high-density lipoproteins (HDL) isolated from human atherosclerotic intima ; in plasma from patients with abdominal aortic aneurysms ; and in plasma from patients with ischemic stroke . While ONOO− was initially believed to be the sole source of protein tyrosine nitration and 3-nitrotyrosine formation in vivo, it is now recognized that 3-nitrotyrosine may be formed as a result of neutrophil myeloperoxidase (MPO)-mediated oxidation of NO2− to •NO2, peroxidase(s) or heme/hemoprotein-catalyzed reactions/Fenton chemistry, and/or MPO-generated hypochlorous acid (HOCl)-induced generation of nitryl chloride (ClNO2) [37–39].
Unsaturated fatty acids react with •NO-derived species via the addition of •NO2 across the double bond to generate carbon-centered radical species known as nitro-fatty acids [64, 79]. Although nitro-fatty acids have been detected in plasma, urine, and tissues from healthy and hypercholesterolemic individuals, the exact mechanism by which nitration of fatty acids occurs in vivo remains unknown [79–81]. It has been suggested that nitration may occur via •NO2 reacting with unsaturated lipids or LOO• to yield lipid derivatives that are of the nitro-allylic, nitroalkene, dinitro, or nitro-hydroxy forms [49, 82]; peroxidase-mediated or Fenton-induced oxidation of NO2− to •NO2 [48, 68, 83]; interaction with ONOO− or ONOOH via oxidation, nitrosation, or nitration reactions; and/or electrophilic addition reactions with NO2+ [84, 85]. Nitrated lipids, such as nitroalkenes, may undergo aqueous decay and release •NO (independent of thiols), isomerize to a nitrite ester with N-O bond cleavage, or undergo one-electron reduction to generate an enol group and •NO [86, 87]. Nitroalkenes may also participate in Michael addition reactions with cysteine and histidine residues in proteins and with the thiolate anion of glutathione (GSH) to initiate reversible post-translational modification(s) of proteins [88, 89].
In vivo, nitroalkene and nitrohydroxy nitro-fatty acids are the predominant species formed . In plasma, following the administration of the exogenous nitro-fatty acid, nitro-9-cis-octadecenoic acid, it was found that this nitro-fatty acid was adducted rapidly to GSH or thiol-containing proteins, underwent reversible release from GSH, and after 90 minutes was predominantly localized to the muscle and liver compartments . These nitro-fatty acids have been shown to mediate a number of inflammatory and vascular cell signaling processes [79, 92, 93]. Although nitro-fatty acids, such as nitrolinoleic acid (LNO2) and nitro-hydroxy arachidonic acid derivatives, induce •NO-dependent guanylyl cyclase-mediated vasodilation, other biological actions attributed to nitro-fatty acids do not result from activation of this signaling pathway and have been attributed, in part, to activation of PPAR-γ and inhibition of NF-kB [81, 86, 92–94]. Nitrolinoleic acid and nitrated oleic acid inhibit neutrophil activation, O2•− synthesis, and degranulation by limiting proinflammatory STAT signaling [95, 96]. Nitroarachidonic acid has also been shown to limit the deleterious effects of macrophage activation by decreasing transcription and expression of the inducible form of nitric oxide synthase (iNOS) in an •NO-independent manner . Leukocyte recruitment and adhesion to the vessel wall are also inhibited by LNO2, which decreases CD11b and vascular cell adhesion molecule-1 (VCAM-1) expression [92, 95]. Platelet function is modulated by nitro-fatty acids, and LNO2 has been shown to decrease thrombin-induced aggregation by increasing platelet cAMP levels . Interestingly, LNO2 has been shown to exert some of its vasoprotective effects via upregulation of the antioxidant enzyme heme oxygenase-1 (HO-1) mRNA and protein levels in both an •NO-dependent and -independent manner; the •NO-independent effects result from activation of the ERK pathway and eIF2α to increase transcription and translation [99, 100]. Nitrolinoleic acid also inhibits vascular smooth muscle cell proliferation by enhancing nuclear translocation of the transcription factor Nrf2 . Thus, nitro-fatty acids represent a novel class of endogenous mediators with vasoprotective effects that are generated by lipid nitration under pathophysiological hypoxic or ischemic conditions.
Thiyl radicals (RS•) may be formed by the •OH, ONOO−, and/or by Fe3+-mediated one-electron oxidation of thiols, such as GSH or cysteine residues in proteins [102–104]. Thiyl radicals may also be derived from other sulfur-containing moieties, including disulfide, thioester, or thioether functionalities, under conditions of oxidant stress [105–107]. Once formed, RS• react not only with themselves and oxygen but also oxidize biological electron donors, including ascorbic acid, NADH, and ferricytochrome c; react with tocopherol species (limiting their availability for reaction with unsaturated fatty acids); and react across carbon-carbon double bonds [108–113].
Evidence for the in vivo formation of RS• was demonstrated in rats administered the spin trap 5,5-dimethyl-1-pyrroline (DMPO) and given oxidized linoleic or linolenic acid via direct injection into the stomach. Analysis of bile samples revealed that a DMPO-RS• adduct, identified as DMPO-GS•, was formed presumably due to a reaction between the oxidized fatty acids with biliary GSH . In human plasma isolated from healthy adult volunteers, incubation of the samples with ONOO− (0.5 mM) and the spin trap α-phenyl-N-tert-butyl nitrone (PBN) (50 mM) resulted in the detection of an EPR spectrum that was nearly identical to that obtained for a PBN-albumin-RS• radical adduct following ONOO−-mediated oxidation of bovine serum albumin [115, 116].
The importance of RS• to atherogenesis results from their ability to participate in reversible addition reactions to double bonds and to abstract hydrogen atoms from bisallylic carbon-hydrogen bonds. Thiyl radicals induce cis-trans isomerization of double bonds in monounsaturated and polyunsaturated fatty acids through a radical intermediate where the carbon-carbon bond is capable of rotation [117–121]. The RS• adds to the double bond, induces rotation of a carbon-carbon bond, and subsequently ejects an RS• by β-scission to perpetuate the cycle . The RS•-mediated isomerization process is random with respect to the position of the double bond within the structure, and, as the trans-conformation is preferred energetically by 0.6–1 kcal/mol, at equilibrium the trans-olefin will predominate (80% vs. 20% cis isomer) [117, 118, 121–124]. The kinetics of cis-trans isomerization are determined by the rotation barrier, which itself is dependent upon the thiol structure but independent of the choice of RS• . In the case of polyunsaturated fatty acids, cis-trans isomerization proceeds via the formation of intermediate mono-trans isomers followed by RS• addition to yield di-trans isomers . When present, trans-fatty acids in membrane lipids alter the physical properties of the phospholipid bilayer resulting in decreased fluidity and permeability [112, 125].
Trans-isomers of unsaturated fatty acids are a component of the Western diet as they occur naturally in dairy products and beef fat, or are formed during the partial hydrogenation of vegetable oils. Intake of these food products results in the absorption of geometrical isomers of C18 unsaturated fatty acids that may be converted to long chain fatty acids or incorporated into lipids [126–128]. Trans-fatty acids may also be formed in vivo. In rats fed a diet devoid of trans-fatty acids, analysis of membrane lipids isolated from different tissues revealed the presence of trans-fatty acids indicating that endogenous formation had occurred. When rats were subjected to CCl4 poisoning to induce •CCl3 and CCl3O2• radical formation in vivo, there was a fourfold increase in plasma lipid trans-fatty acid content suggesting that free radicals participate in the endogenous formation of trans-fatty acids [129, 130].
Increased intake of trans-fatty acids has been associated with an elevated risk of cardiovascular disease; however, the contribution of endogenously formed trans-fatty acids to disease risk has not been evaluated to date . Mechanistically, the adverse effects of dietary trans-fatty acids have been attributed to increased total and LDL cholesterol with a concomitant decrease in HDL cholesterol [132, 133]; increased levels of lipoprotein(a) and triglycerides ; elevated plasma levels of the inflammatory markers, tumor necrosis factor-α, tumor necrosis factor receptor 1, interleukin-6, and C-reactive protein [135, 136]; increased soluble intercellular adhesion molecule-1 (ICAM-1), VCAM-1, and plasma E-selectin levels ; and, impaired endothelium-dependent vascular reactivity assessed by brachial artery flow-mediated dilation .
Thiyl radicals also have a propensity to initiate lipid peroxidation by abstracting hydrogen atoms from bis-allylic methylene groups of fatty acids to yield pentadienyl radicals. These radicals, in turn, may react with oxygen to generate peroxyl radicals. Early studies using pulse radiolysis to generate RS• revealed the absolute rate constants for the reactions between cysteinyl RS• and linoleic (18:2), linolenic (18:3), and arachidonic (20:4) acids in water/alcohol mixtures to be 1.3–16 × 106 M−1s−1, and these rates were increased with both the number of bis-allylic functions and the polarity of the solvent . In studies performed with other RS•, including RS• derived from GSH, the rate constant range for the reaction with fatty acids containing between 2 and 6 double bonds was determined to be 0.3–7.5 × 107 M−1s−1 . Further characterization of the hydrogen abstraction reaction from bis-allylic centers estimated that the carbon-hydrogen bond dissociation energy was 75–80 kcal/mol . Thiyl radical reactivity with polyunsaturated fatty acids was also found to be determined by the distance between the reactive –S• center and the ionic group in the attacking molecule, the number of ionic functions in RS•, and the lipophilicity of the attacking RS• .
Experimental evidence supports the role of RS• in lipid peroxidation in complex biological systems. For example, in monkey arterial smooth muscle cells, ROS generation and oxidation of LDL was found to be L-cystine-dependent and required micromolar concentrations of Cu(II) or Fe(III). This oxidatively modified LDL was taken up and degraded by mouse peritoneal macrophages to a greater extent than that observed for non-modified LDL. These findings demonstrated that sulfur-containing amino acids support ROS generation resulting in modification of lipoproteins and, therefore, may play a role in atherogenesis . This conclusion is further supported by the observation that trans-isomers of arachidonic, linoleic, or oleic acid residues were also found to be present in membrane phospholipids isolated from human monocytic leukemia cells following incubation with thiol compounds, including β-mercaptoethanol, GSH, or 3-(mercaptoethyl)quinazoline-2,4(1H,3H)dione .
Tyrosyl radicals are generated by the one-electron oxidation of L-tyrosine in reactions with •OH and transition metal ions, as well as by reaction with ONOO− derivatives, and by MPO and H2O2 [143–145]. Tyrosyl radicals readily combine to form o,o’-dityrosine and react with O2•− at a rate (1.5 × 109 M−1s−1) that is approximately 3-fold faster than that for dimerization to occur (0.45 × 109 M−1s−1) [146, 147]. Tyrosyl radicals generated by MPO are highly reactive and have been shown to damage proteins by forming o,o’-dityrosine crosslinks. Studies performed with human neutrophils demonstrated that the MPO-H2O2 system was able to generate protein-bound dityrosine with free L-tyrosine independent of free metal ions [143, 148]. In a chloride-free system, the major product of L-tyrosine oxidation by MPO was determined to be dityrosine with lesser amounts of pucherosine and trityrosine formed; however, at physiological concentrations of L-tyrosine and Cl−, the amphipathic aldehyde p-hydroxyphenylacetaldehyde was determined to be the major product of the neutrophil MPO-H2O2 system. This aldehyde was shown to modify proteins covalently via a reduced Schiff base mechanism involving the aldehyde and lysine residues’ Nε-amino moieties [149, 150]. Tyrosyl radicals generated by MPO have also been implicated in the initiation of lipid peroxidation. In vitro studies of LDL oxidation by activated human neutrophils, which contain abundant MPO and H2O2, revealed that lipid peroxidation occurred only in the presence of free L-tyrosine implying that tyrosyl radical formation by MPO was necessary for the initiation of lipid peroxidation . Tyrosyl radicals have also been shown to play a role in LDL oxidation in vivo and in atherogenesis. Analysis of LDL isolated from human vascular tissue demonstrated that o,o’-dityrosine levels were 100 times greater than that observed in circulating LDL . Similarly, o,o’-dityrosine formation was found to be increased 11-fold in atherosclerotic fatty streaks and 6-fold in advanced atheromas compared to normal aortic tissue, indicating that tyrosyl radical formation was capable of protein damage in vivo .
In vascular cells, circulating inflammatory cells, and platelets, free radicals and non-radical reactive species may be generated from reactions between O2•− and/or •NO and other free radicals, lipids, and proteins. In these cell types, O2•− is synthesized by the enzymes NADPH oxidase, MPO (which also synthesizes reactive halogenating species) xanthine oxidase, lipoxygenases, and uncoupled nitric oxide synthase(s), and as a byproduct of mitochondrial respiration (Fig. 1). Within the cell, some level of O2•− is required to maintain cellular homeostasis; however, when these O2•−-generating sources remain activated after a physiological stimulus has abated, the continued production of O2•− alters cellular redox homeostasis resulting in increased oxidant stress. As such, there is evidence to support a role for the activation of the aforementioned sources of O2•− in the pathogenesis of atherosclerosis.
The NADPH oxidases are a family of multicomponent enzymes that assemble upon stimulation in the cell membrane, and are a predominant source of vascular cell O2•− formation. These enzymes utilize NADPH as an electron donor to reduce molecular oxygen to O2•−, have low levels of ambient activity, and in vascular cells, are differentiated by their subunit composition. NADPH oxidase is comprised of up to three cytosolic subunits (p40phox, p47phox, and p67phox), a regulatory G-protein (Rac1 or Rac2), and a membrane-bound cytochrome b558 reductase domain, consisting of two proteins (p22phox and gp91phox); homologs of the gp91phox in vascular cells have been termed Nox [153–158]. Although there is no crystal structure of the Nox enzymes, they contain 6 transmembrane domains, and domains III and IV contain 2 histidine residues that span 2 asymmetrical heme groups. The −COOH terminus resides in the cytoplasm and contains the NADPH and flavin adenine dinucleotide binding domains. Thus, to generate O2•−, an electron is transferred from NADPH to flavin adenine dinucleotide, to the inner heme, to the outer heme, and then to oxygen [159–163]. Vascular endothelial cells have been shown to contain Nox1, Nox2, Nox4, and Nox 5 and predominantly generate O2•− that is released to the extracellular compartment, while vascular smooth muscle cells express Nox1, Nox4, and Nox5 and synthesize intracellular O2•− and H2O2 [164–167]. Regulation of NAD(P)H oxidase activation occurs via phosphorylation of p47phox. In p47phox, intermolecular interaction between two tandem SH3 domains limit p47phox binding to p22phox; however, once p47phox is phosphorylated, it may interact with the cytoplasmic tail of p22phox and facilitate the full assembly of NAD(P)H oxidase . Once the enzyme is assembled, it is active and O2•− is generated by transfer of electrons from cytosolic NADPH to oxygen in the luminal or extracellular space. The Km of the Nox2 subunit for NADPH has been reported to be 40–45 µM .
Vascular NAD(P)H oxidase-derived O2•− production is augmented by stimuli with pathophysiological relevance for atherosclerosis, including tumor necrosis factor-α and inflammatory cytokines, angiotensin II, aldosterone, bone morphogenetic proteins, and oscillatory shear stress [170-176]. Superoxide levels produced by NAD(P)H oxidase may also be influenced by heritable polymorphisms with functional effects on enzyme activity. To date, the association between the C242T polymorphism of the p22phox subunit and atherosclerotic risk has received the greatest attention. One analysis of 201 patients with coronary artery disease compared with 201 control subjects revealed that this variant was protective with respect to atherosclerosis (OR = 0.49, 95% CI: 0.28, 0.87, p=0.015) . Similarly, another study 3,085 subjects with established atherosclerosis compared with 2,163 control subjects confirmed that the C242T polymorphism was associated with a decreased risk of atherosclerosis . This protective effect may be attributed, in part, to the finding that basal and stimulated vascular NAD(P)H oxidase-mediated O2•− production is decreased ~30% in individuals with the C242T polymorphism compared to controls . In contrast, two polymorphisms in the promoter region of p22phox, −675 A/T and −930 G/A, were associated with elevated NAD(P)H oxidase expression and activity, and increased carotid intima-media thickness; the −930 G/A polymorphism was also found to have an interaction with tobacco use [180, 181].
In atherosclerosis, evidence demonstrates that NAD(P)H oxidase expression and activity are elevated. Increased expression of the NAD(P)H oxidase subunits p22phox, p47phox, and p67phox, calcium-dependent Nox5, and O2•− production have been found in coronary arteries isolated from patients with established atherosclerosis, and saphenous veins and internal mammary arteries obtained from diabetic patients [182–184]. Similarly, enhanced expression of the p22phox and gp91phox subunits has been shown to correlate with both the degree of coronary artery atherosclerotic lesion severity and plaque cross-sectional area [167, 185].
Despite these observations, data to support a causal role for NAD(P)H oxidase in atherosclerotic lesion formation remain controversial, and studies performed in mouse models with genetic deletion of NAD(P)H oxidase subunit(s) have yielded conflicting results. For example, gp91phox−/− mice develop the same degree of atherosclerosis as control mice, and when crossed with apolipoprotein E knockout (apoE−/−) mice, deletion of gp91phox had no influence on vascular atherosclerotic burden . In contrast, genetic deletion of p47phox, an activator of Nox2, appeared to influence atherogenesis: p47phox/apoE double knockout mice have a marked reduction in atherosclerosis compared to apoE−/− mice, in part, due to a decrease in monocyte/macrophage-mediated oxidation of LDL [187, 188].
Myeloperoxidase, which is expressed abundantly in neutrophils and monocytes, is a basic (pI > 10) heme protein that utilizes chloride as a substrate and H2O2 as a cosubstrate to generate reactive oxidants and radical species that oxidatively modify lipids and proteins [189, 190]. Immunohistochemical and biochemical analyses have demonstrated that MPO and its oxidation products are present within human atherosclerotic lesions [72, 191, 192]. Among the MPO-derived oxidants, HOCl is a potent chlorinating oxidant; importantly, MPO is the only known pathway for synthesizing reactive chlorinating species in humans [192–194]. Myeloperoxidase has also been shown to use other free radicals and low-molecular-weight compounds as substrates, including NO2−, tyrosine, ascorbate, urate, catecholamines, estrogens, and serotonin [148, 195–199]. In human plasma, it has also been shown that MPO catalyzes the oxidation of •NO to NO2− by a mechanism that involves the MPO-mediated formation of a substrate radical, which, in turn, reacts with •NO [200–202].
Regulation of MPO peroxidase activity is dependent, in part, upon •NO levels [200, 203]. Under anaerobic conditions, it was found that •NO binds to MPO-Fe(III), which is catalytically active, with kon = 1.07 µM−1s−1 and koff = 10.8 s−1, while •NO was found to bind to MPO-Fe(II) with kon = 0.1 µM−1s−1 and koff = 4.6 s−1, suggesting that following heme reduction, there is a conformational change that modifies the accessibility of •NO to the heme pocket . Furthermore, •NO was found to modulate MPO peroxidase activity by two mechanisms: low levels of •NO reduce Compound II of the enzyme to the Fe(III) form while high levels of •NO lead to inhibition of MPO by forming the nitrosyl complex, MPO-Fe(III)-NO [200, 201, 203]. As such, it has been suggested that MPO may serve as a catalytic sink for •NO and limit its bioavailability .
The MPO-derived oxidant, HOCl, mediates chlorination of amines and unsaturated lipids, oxidation of thiols and thiol esters, and oxidative bleaching of heme groups and iron sulfur centers [193, 204–206]. The formation of 3-chlorotyrosine, a stable chlorination product, has been detected in LDL isolated from human atherosclerotic aortas, and levels were found to be 30-fold higher than those measured in LDL from healthy donors . Nitrotyrosine may also be formed through MPO-dependent pathways, including the reaction of NO2− with HOCl to generate nitryl chloride (Cl-NO2) and •NO2 [67, 68, 195, 207]. Similarly, other oxidation products derived from MPO have been demonstrated in human atherosclerotic tissues. Acrolein, an aldehyde formed by threonine exposed to MPO-derived oxidants, was shown to modify lysine residues covalently in human atherosclerotic plaques . The MPO-mediated oxidation of serine results in formation of the glycoaldehyde, carboxymethyllysine, an intermediate in advanced glycosylation end-product generation, which has been detected in atherosclerotic lesions . In addition, Schiff base adducts between p-hydroxyphenylacetaldehyde, a tyrosine oxidation product, and the ε-amino moiety of protein lysine residues and amino lipids have been identified in human atherosclerotic lesions [150, 210].
Lipoproteins exposed to MPO-generated tyrosyl radicals demonstrate o,o’-dityrosine formation, protein cross linking, and lipid peroxidation, and these modifications have been found in lipoproteins isolated from human atheroma [72, 73, 143, 151]. HOCl oxidizes lysine residues in apolipoprotein B-100 (apoB-100) of LDL and converts LDL into an high-uptake form [211, 212]. Exposure of LDL to MPO-HOCl-NO2− also promotes apoB-100 protein nitration, and effectively initiates lipid peroxidation in serum [207, 213, 214]. These nitrated LDL species are converted into NO2-LDL that is recognized by the scavenger receptor CD36 and are avidly taken up and degraded by macrophages, resulting in foam cell formation and cholesterol deposition [214–216]. Other studies have identified oxidized phospholipids, characterized by an sn-2 acyl group with a terminal γ-hydroxy(or oxo)-α,β-unsaturated carbonyl, as high affinity ligands for the macrophage scavenger receptor CD36, on both NO2-LDL and other oxidized forms of LDL [217, 218].
MPO-catalyzed oxidation of thiocyanate (SCN−), which is present in human plasma and elevated in smokers, induces protein carbamylation at sites of inflammation and may have relevance to atherogenesis. The oxidation products of SCN− include hypothiocyanous acid (HOSCN), a weak oxidant, and trace amounts of cyanate (OCN−) [219, 220]. Cyanate may react with protein nucleophilic groups through carbamylation, possibly leading to functional impairment. MPO-carbamylated lipoproteins have atherogenic properties and have been associated with coronary artery disease; in a case-control (1:2) study performed in 450 subjects, individuals with plasma concentrations of homocitrulline (ε-carbamyllysine) in the highest quartile were found to have a 7–8–fold higher risk of having clinical or angiographic evidence of atherosclerotic vascular disease compared to those in the lowest quartile .
In the vasculature, MPO-mediated consumption of •NO may contribute to a pathophysiological phenotype that facilitates atherothrombogenesis. Levels of MPO-derived HOCl contribute to a prothrombotic phenotype and have been shown to induce endothelial cell apoptosis and detachment . MPO-derived lipid oxidation products increase expression of P-selectin to enhance platelet adhesion and increase tissue factor activity [222, 223]. MPO has also been implicated in the transition of a stable plaque to a vulnerable plaque, as HOCl activates matrix metalloproteinase-7 (MMP7) via oxygenation of a thiol residue in the enzyme’s cysteine switch; however, there is also dynamic homeostatic control of MMP7 activity as oxidative modification of tryptophane and glycine residues located in the catalytic domain of MMP7 decrease proteolytic activity .
Statins have been shown to downregulate MPO mRNA expression by 60-fold in human monocytes compared to levels in untreated cells by blocking geranylgeranylation of rhoA GTPase . In statin-treated human transgenic MPO mice, MPO mRNA levels were reduced significantly, and this reduction correlated with a reduction in MPO activity . Statins have also been shown to lower systemic levels of protein-bound nitrotyrosine: in hypercholesterolemic patients without evidence of atherosclerosis, atorvastatin decreased plasma levels of protein bound nitrotyrosine, chlorotyrosine, and dityrosine by 25%, 30%, and 32%, respectively [13, 226].
Epidemiological studies have found that individuals with total or near total deficiency of MPO, which occurs in the Caucasian population with a frequency of ~ 1 in every 2000–4000, are less likely to develop coronary artery disease . Individuals with a −463G/A promoter polymorphism that results in a 2-fold decrease in MPO expression have decreased angiographic evidence of coronary artery disease, nonfatal myocardial infarction, and cardiac death [228, 229]. In contrast, individuals with elevated levels of MPO have an increased risk of angiographically documented coronary artery disease. In one study, individuals with MPO levels in the highest quartile were 15–20-fold more likely to have coronary artery disease compared with those in the lowest quartile . Serum MPO levels have also been shown to predict independently endothelial dysfunction in humans (OR = 6.4, 95% CI: 2.6 – 16.0, p=0.001 for highest vs. lowest quartile), to predict future risk of coronary artery disease in healthy individuals with LDL cholesterol < 130 mg/dL (OR = 1.52, 95% CI: 1.21–1.91) or HDL > 50 mg/dL (OR = 1.59, 95% CI 1.24–2.05), and to be significantly associated with recurrent ischemic events in patients with acute coronary syndromes [231–233].
Xanthine oxidase catalyzes the oxidation of hypoxanthine to xanthine and uric acid in a reaction that involves a one-electron transfer to oxygen to yield O2•− [234–237]. Xanthine oxidase also catalyzes hydroxylation of N-heterocyclic and aldehyde substrates, as well as functions as an NADH oxidase [238–241]. The enzyme exists as a homodimer, and each subunit possesses four redox centers: a molybdopterin site, a single flavin adenine dinucleotide site, and two Fe2S2 sites [242, 243]. Xanthine oxidase circulates in plasma at levels reported to be 0–4200 mU/L and is expressed only by vascular endothelial cells in the vessel wall [244–246]. In the endothelium, xanthine oxidase exists as two isoforms, dehydrogenase and oxidase . While only the oxidase isoform generates O2•−, isoform switching is mediated irreversibly by proteolysis or by reversible oxidation of Cys535 and Cys992, indicating that isoform activity is redox-sensitive [248, 249]. Under physiological conditions, the enzyme exists mainly in the dehydrogenase form; however, in the setting of hypoxia, the oxidase isoform predominates [250, 251].
Although increased expression and activity of xanthine oxidase have been demonstrated in atherosclerosis and serum uric acid has been shown to be an independent predictor of cardiovascular mortality, evidence implicating xanthine oxidase in atherogenesis in humans has been largely indirect [184, 252]. Inhibition of xanthine oxidase activity with oxypurinol has been shown to improve impaired vasodilation and vascular reactivity in individuals with hypercholesterolemia and coronary artery disease . In these patients, increased circulating levels of xanthine oxidase may contribute significantly to vascular oxidant stress as studies performed in hypercholesterolemic rabbits revealed that hypercholesterolemia stimulates the release of xanthine oxidase from the liver to increase circulating xanthine oxidase activity two-fold . Similarly, patients who underwent coronary artery bypass grafting surgery and were treated with allopurinol demonstrated decreased histological evidence of myocardial ischemia-reperfusion injury compared to untreated patients; in addition, patients with an acute myocardial infarction who received allopurinol prior to primary angioplasty had improved left ventricular function post-procedure and at 6 months [255, 256]. Recently, investigators have provided more direct experimental evidence to link xanthine oxidase activation with the development of atherosclerosis. In atherosclerotic apoE−/− mice, inhibition of xanthine oxidase activity with tungsten attenuated vascular O2•− production, improved endothelial function and vascular reactivity, and prevented the development of atherosclerosis .
One plausible explanation for the difficulty in implicating xanthine oxidase with atherogenesis is that the enzyme may function as a NO3− and NO2− reductase at the molybdopterin site and generate atheroprotective •NO, especially under hypoxic conditions [258–262]. A second rationale has been exposed by population-based phenotyping studies that have identified significant interindividual and interethnic differences in xanthine oxidase activity, with Caucasians having approximately 20% lower activity than other groups . Furthermore, while several genetic polymorphisms have been identified, only the A69901C variant has been associated with carotid atherosclerosis in a study of 953 hypertensive subjects and 1,818 individuals from Japan . Interestingly, it has been shown that xanthine oxidase expression and O2•− production are upregulated by NAD(P)H oxidase, suggesting that factors that regulate activity of this oxidase may also influence xanthine oxidase .
The lipoxygenases are non-heme iron-containing dioxygenases that promote atherosclerosis by generating oxidatively modified, biologically active lipids. Lipoxygenases catalyze the stereospecific insertion of oxygen into polyunsaturated fatty acids, yielding unsaturated hydroperoxy fatty acid derivatives, including prostaglandins, thromboxanes, and leukotrienes [266, 267]
Expression of 12/15-lipoxygenase has been detected in human atherosclerotic plaques, and using chiral phase HPLC, lipoxygenase-mediated oxidation of linoleate and cholesteryl hydroperoxy-octadecadienoate isolated from human atherosclerotic plaques was confirmed by the high degree of observed stereospecificity of oxidation [268–270]. In apoE−/− and LDLR−/− mice, genetic deletion of 12/15 lipoxygenase was associated with a decrease in vascular oxidant stress, macrophage foam cell formation, monocyte adhesion to the endothelium, and atherosclerosis [271–275]. Interestingly, genetic studies in humans do not confirm this association between 12/15-lipoxygenase expression and atherosclerosis. In two large community-based populations (1,809 subjects, 1,734 controls in the Coronary Artery Risk Development in Young Adults Study; 12,974 participants in the Atherosclerosis Risk in Communities study), the T560M polymorphism, which results in a 20-fold reduction in the catalytic activity of 12/15-lipoxygenase, was associated with a 1.31–1.62-fold increased risk of atherosclerosis suggesting that near deletion of enzymatic activity did not protect against atherosclerotic vascular disease . In contrast, increased 12/15-lipoxygenase activity, as occurs with the −611A/G promoter polymorphism, was associated with carotid plaque (OR = 4.01, 95% CI: 1.39, 11.53, p=0.0005) when examined in 556 individuals with coronary heart disease; however, the association was not found in 1,111 community-based individuals .
The enzyme 5-lipoxygenase (5LO), which synthesizes the proinflammatory leukotrienes, leukotriene B4, leukotriene C4, leukotriene D4, and leukotriene E4, has been linked to atherosclerosis [278–287]. Early studies in the atherosclerosis resistant CAST mouse model identified 5-LO in the chromosome 6 region that was found to be a major locus for atherosclerosis susceptibility. When this locus was bred into atherosclerosis susceptible B6 mice (known as CON6 mice) and subsequently crossed with LDLR−/− mice, the mice demonstrated decreased atherosclerotic lesion formation on an atherogenic diet compared to LDLR−/− on a similar diet . Further studies performed in 5LO−/− mice bred onto an LDLR−/− background demonstrated a 26-fold decrease in atherosclerotic lesion formation compared to LDLR−/− mice, similar to what was observed in CON6 mice, and that 5LO was abundantly expressed in atherosclerotic plaques and appeared to colocalize with macrophages .
In human atherosclerosis, 5LO expression was confirmed in plaques isolated from the aorta, carotid, and coronary arteries at different stages of atherosclerosis. Here, 5LO was localized to macrophages, dendritic cells, foam cells, mast cells, and neutrophilic granulocytes, and the number of 5LO expressing cells in the plaque was increased in advanced lesions . Interestingly, 5-LO expression was found predominantly in activated macrophages, and cells expressing 5-LO were localized to the shoulder region of atherosclerotic plaques [290, 291]. Similarly, plaque analysis of endarterectomy specimens obtained from patients with carotid atherosclerosis revealed that 5-LO expression, leukotriene B4 production, and leukotriene A4 hydrolase expression were increased in unstable as compared to stable plaques, and this was associated with an increase in MMP2 and MMP9 expression and activity as well as decreased collagen content [286, 287, 291].
Further support for the role of 5LO in atherosclerosis was provided by several studies that related polymorphisms of the 5LO promoter or 5-lipoxygenase activating protein (FLAP) with an increased risk of atherosclerosis. In a study of 470 healthy individuals enrolled in the Los Angeles Athersoclerosis Study, 6% of the study cohort was identified as carriers of a variant genotype (increased or decreased number of GC-boxes in tandem in the promoter) with racial and ethnic differences in expression; the highest rates of variant genotypes were observed in Asian or Pacific Islanders and African-Americans as compared to Hispanics and non-Hispanic Caucasians. In this study population, presence of the 5LO polymorphism was associated with an increase in carotid intima-media thickness (80 ± 19 µm; 95% CI: 43 – 116, p<0.001). Furthermore, carriers of the 5LO polymorphisms were found to have increased intima-media thickness that was associated with increased intake of arachidonic acid and linoleic acid, a finding that was not observed in noncarriers .
Polymorphisms in the gene encoding FLAP have been associated with the presence of xanthomas in familial hypercholesterolemia as well as an increased risk of myocardial infarction and stroke. In 945 patients with heterozygous familial hypercholesterolemia, the rs9551963 polymorphism was associated with an increased risk of xanthomas (OR = 1.52, 95% CI: 1.11–2.07, p=0.01), while the rs17222842 polymorphism was found to be protective (OR = 0.62, 95% CI: 0.43–0.90, p=0.01) . In a genome-wide linkage analysis of 296 families in Iceland, including 713 individuals with myocardial infarction, a four SNP haplotyope in the FLAP gene, HapA, was found to be associated with myocardial infarction (RR = 1.8, adjusted p=0.005); HapA had a frequency of 15.8% in affected individuals compared to 9.5% in controls. In this population, HapA was also found to be associated with an increased risk of stroke (RR = 1.7, p=000095). Neutrophils isolated from individuals with a history of myocardial infarction demonstrated increased production of leukotriene B4 following stimulation with ionomycin compared to neutrophils from controls. To confirm that these findings were not unique to the Icelandic population, a similar genetic analysis was performed in 753 individuals from the United Kingdom with myocardial infarction and 730 controls. In this population, the HapA haplotype was not associated with myocardial infarction; however, the investigators identified a second 4 SNP haplotype, HapB, that was associated with an increased risk of myocardial infarction (RR = 1.95, p=0.00037) . These findings were also confirmed in a Scottish population of 450 patients with ischemic stroke .
A haplotype in the leukotriene A4 hydrolase gene, HapK, has also been described. In an Icelandic population, this haplotype was associated with an increased risk of myocardial infarction and additional cardiovascular disease (RR = 1.45, p = 0.0091). A replication study performed in Atlanta, Cleveland, and Philadelphia revealed that self-described European Americans, 27% of whom carried at least one copy of HapK, were at increased risk of myocardial infarction (RR=1.19, p=0.006); however, African-Americans, of whom only 6% were carriers, were found to be at significantly increased risk (RR = 3.57, p= 0.000022) .
Further support for the contribution of 5-LO and FLAP to atherosclerotic vascular disease was provided by a study of apoE/LDLR double knockout mice treated with MK-866, an inhibitor of FLAP. In these mice, MK-866 treatment decreased aorta atherosclerotic lesion size and lesion cross-sectional area, and increased plaque stability by increasing collagen and smooth muscle cell content while limiting macrophage infiltration . Clinically, a randomized, prospective, placebo-controlled crossover study of the FLAP inhibitor (DG-031) was carried out in patients with myocardial infarction who were found to carry at-risk polymorphisms in the FLAP or leukotriene A4 hydrolase genes. In this study, 191 patients were identified as carriers of polymorphisms in either gene; 87% with a FLAP polymorphism, 13% with a leukotriene A4 hydrolase polymorphism. Treatment with DG-031 (750 mg/d) resulted in a 26% decrease in ionomycin-stimulated neutrophil production of leukotriene B4 (95% CI: 10% – 39%, p=0.003) and a 12% reduction in neutrophil myeloperoxidase release (95 % CI 2% – 21%, p=0.02) .
The endothelial isoform of nitric oxide synthase (eNOS) catalyzes the oxidation of L-arginine to L-citrulline and •NO. The enzyme exists as a dimer with both reductase and oxygenase domains, and produces •NO via electron transfer from the cofactor NADPH through flavin adenine dinucleotide and flavin mononucleotide to heme where substrate L-arginine is oxidized to form •NO and L-citrulline. When L-arginine or the cofactors NADPH or 5,6,7,8-tetrahydrobiopterin (BH4) are not sufficient, eNOS “uncouples” and reduces oxygen to yield O2•− in preference to its oxidizing L-arginine to yield •NO . Similarly, release of zinc from the zinc-thiolate moiety of eNOS by ONOO− results in disulfide bond formation between eNOS monomers, and uncoupling of the enzyme to generate O2•− .
Although eNOS is atheroprotective, overexpression of eNOS in apoE−/− mice resulted in decreased •NO, elevated levels of vascular O2•−, and increased atherosclerotic lesion formation . This finding was attributed to a relative deficiency in BH4, leading to eNOS uncoupling; dietary supplementation with BH4 restored •NO levels and diminished vascular oxidant stress . In a GTP cyclohydrolase I transgenic mouse model, characterized by increased synthesis of BH4 and limited BH4 oxidation, eNOS uncoupling was prevented both in the absence of vascular disease and in experimental atherosclerosis [302, 303]. When this mouse model was crossed with an apoE−/− mouse that overexpressed eNOS, there was a reduction in vascular oxidant stress and atherosclerotic lesion formation, demonstrating the importance of cofactor supply to prevent eNOS uncoupling . Recently, a novel haplotype of GCH1, the gene encoding GTP cyclohydrolase I, was identified and found to be associated with reduced enzymatic activity [305, 306]. The relationship between this haplotype, defined by 3 single nucleotide polymorphisms (rs8007267G>A in the promoter region, rs3783641A>T in intron 1, and rs10483639C>G in the 3’ untranslated region) and vascular biopterin levels, oxidant stress, and vasodilation was examined in arterial and venous segments obtained from 347 patients with coronary artery disease at the time of elective coronary artery bypass grafting. Here, it was found that homozygous carriers of the haplotype had significantly lower plasma and vascular BH4 and total biopterin levels compared to heterozygous or noncarriers of the haplotype. Vascular O2•− production was increased significantly and endothelium-dependent vascular reactivity was decreased significantly in vascular segments from homozygous carriers compared to heterozygous or noncarriers of the haplotype. Thus, GCH1 haplotype appears to regulate vascular BH4 levels with consequences for eNOS coupling . Other investigators have identified a C243T polymorphism in the 3’-UTR that was associated with •NO production, baroreflex coupling, and systolic and diastolic blood pressure levels . C-reactive protein, which is associated with endothelial dysfunction and atherosclerosis, was shown to produce its adverse vascular effects, in part, by uncoupling eNOS through decreased GTP cyclohydrolase I expression resulting in BH4 deficiency . In addition, the compound AVE9488 was found to limit atherosclerosis in apoE−/− mice by stimulating eNOS promoter activity, and, therefore, eNOS expression, concomitant with increasing vascular content of BH4 to prevent eNOS uncoupling .
Clinically, eNOS uncoupling has been associated with hypertension, diabetes mellitus, hypercholesterolemia, and atherosclerosis, suggesting that the association between eNOS cofactor supply, eNOS uncoupling, and eNOS-mediated O2•− production defined in animal models may also be operative in human disease [311–314]. In fact, in explants of saphenous veins and internal mammary arteries obtained from patients undergoing coronary artery bypass grafting surgery, high vascular levels of BH4 were associated with decreased eNOS-derived O2•− production and improved endothelium-dependent vascular reactivity . When these explanted tissues were incubated with 5-methyltetrahydrofolate to scavenge ONOO− or 5-methyltetrahydrofolate was infused in vivo prior to vessel harvest, there was a further increase in vascular BH4 levels and the BH4/total biopterin ratio, and diminished eNOS-mediated O2•− production. The eNOS dimer/monomer ratio was also increased, resulting in enhanced eNOS activity and •NO levels .
In atherosclerosis, vascular smooth muscle cells, monocytes, macrophages, and dendritic cells all express iNOS. Induction of iNOS may occur following exposure to inflammatory cytokines, including interleukin-1β, interferon-γ, and tumor necrosis factor-α. In contrast to eNOS, iNOS binds Ca2+/calmodulin tightly and does not require an increase in intracellular Ca2+ for activation .
The presence of iNOS localized to macrophages and vascular smooth muscle cells has been documented in human atherosclerotic plaques [318–320]. In these vascular atheromas, iNOS expression was detected as early as the fatty streak stage and was present in advanced atheromas coinciding with evidence of protein nitration . Moreover, iNOS expression was found to colocalize with oxidized lipid and protein derivatives found in atherosclerotic plaques . Further evidence for the role of iNOS in atherosclerosis is derived from studies in animal models. In the apoE−/− mouse, iNOS expression was detected in macrophages and vascular smooth muscle cells of developing plaques . Although genetic deficiency of iNOS itself did not influence atherosclerotic lesion size in diet-induced atherosclerosis, several studies performed in iNOS/apoE double knockout mice show a reduction in atherosclerotic plaque formation [323–326].
To determine the association between iNOS-derived •NO and ONOO− formation in atherosclerosis-prone mouse models, 3-nitrotyrosine accumulation and BH4 oxidation were assessed in apoE−/− mice fed a Western diet. Here, 3-nitrotyrosine and dihydrobiopterin (BH2) levels were found to be low in 3-week old apoE−/− mice but increased significantly in 27-week old mice fed a Western diet. In contrast, apoE-iNOS double knockout mice fed a Western diet demonstrated a reduction in tissue 3-nitrotyrosine accumulation indicating that 3-nitrotyrosine expression in this model was, in part, associated with iNOS-derived •NO. Similarly, increased BH2 levels associated with Western diet feeding were decreased in double knockout mice, indicating iNOS likely played a role in oxidation of BH4 .
In a streptozotocin-induced diabetic rat model, electron paramagnetic resonance spectroscopy detected •OH formation and LOO• radicals generated by uncoupled iNOS. Treatment of the rats with L-arginine decreased O2•− generation and ONOO− production resulting in decreased LOO• formation. As diabetes progressed in this animal model, there was an associated rise in iNOS protein expression, and iNOS colocalized with 3-nitrotyrosine and 4-hydroxynonenal formation. Thus, iNOS-mediated •OH production likely also contributes to atherogenesis by initiating lipid peroxidation in experimental diabetes .
During oxidative phosphorylation, mitochondria generate O2•− at several sites within the electron transport chain. As electrons are transferred from NADH through Complexes I–IV to generate ATP, 0.2–2.0 % of the electrons leak to form O2•−, primarily via Complex I and Complex III . Within the mitochondria, O2•− may be produced within the mitochondrial matrix and the intermembrane space, and it has been suggested that mitochondrial membranes produce ~ 24 nmol/O2•−/min per gram of tissue [329, 330]. The majority of O2•− is converted to H2O2 by manganese superoxide dismutase (Mn-SOD), localized to the matrix, to maintain an estimated intramitochondrial steady-state concentration of O2•− of 8 × 10−12 M or greater . Mitochondrial dysfunction has been linked to atherosclerosis, and aortic specimens obtained from patients with atherosclerosis show a greater degree of mitochondrial DNA damage than that observed in the absence of atherosclerosis . Mitochondrial DNA damage has been shown to precede atherosclerosis and to correlate with the extent of disease in apoE−/− mice . The mechanistic role of O2•− in mitochondrial DNA damage was determined when Mn-SOD) activity was inhibited in a murine atherosclerosis model: when Mn-SOD activity was deficient, there was evidence of increased mitochondrial DNA damage and accelerated atherosclerosis . It has also been shown that oxidized lipids colocalize with mitochondria in vitro and that oxidized LDL (oxLDL) increases mitochondrial O2•− formation, suggesting another mechanism by which mitochondrial dysfunction may be associated with atherosclerotic lesion formation [333, 334]. Finally, deficiency of complex I, NADH:ubiquinone oxidoreductase, is the most common enzymopathy associated with oxidative phosphorylation and is associated with an increase in O2•− generation as a result of decreased complex I activity .
To limit the deleterious effects of reactive free radical and non-radical species and their pathophysiological sequelae, the intracellular redox state is balanced by small molecule antioxidants that include GSH, α-tocopherol, ubiquinones, and ascorbic acid, as well as antioxidant enzymes that modulate the intra- and extracellular reactions of oxidizing metabolites. The enzymes that comprise this antioxidant defense system are extensive and include the SODs, which dismutate O2•− to H2O2 and oxygen; catalase, which catalyzes the decomposition of H2O2 to water and oxygen; the glutathione peroxidases, which utilize GSH to reduce H2O2 and fatty acyl peroxides to water and lipid alcohols, respectively; glutathione reductase, which reduces oxidized glutathione or glutathione disulfide to GSH; glutathione-S-transferases, which glutathiolate oxidants to decrease their oxidizing potential; thiol-disulfide oxidoreductases, which preserve the thiol redox state of proteins; peroxiredoxins, which catalyze the reduction of H2O2 to water; HO-1, which degrades free heme to carbon monoxide (CO) and biliverdin; paraoxonases, which limit oxidation of LDL; and glucose-6-phosphate dehydrogenase, which is the principal intracellular source of NADPH, a key reducing equivalent and cofactor for other antioxidants (Table 1)(Fig. 1).
The importance of these antioxidants in modulating the oxidative risk for atherothrombotic vascular disease is revealed when the aforementioned antioxidant enzymes have diminished catalytic activity. This insufficiency may occur as a result of decreased enzyme expression or activity as a consequence of a heritable polymorphism, or in a significant oxidizing environment where oxidative modification of the enzyme alters its functional properties. As a result of decreased activity of these antioxidant enzymes, a pro-oxidative environment is created that favors the development of atherosclerotic lesions.
Once formed, O2•− is dismutated enzymatically to H2O2 and oxygen by the SOD family of antioxidant enzymes, which include intracellular SOD (Cu,Zn-SOD), extracellular SOD (EC-SOD), and mitochondrial Mn-SOD, also referred to as types I, III, and II, respectively [336–342]. These enzymes catalyze the dismutation of O2•− with a rate constant of 2.3 × 109 M−1s−1 that is 10,000 times faster than the spontaneous dismutation of O2•− [22, 343]. Cu,Zn SOD requires Cu2+ and Zn2+ for catalytic activity and resides in the cytoplasmic compartment of cells, although some enzymatic activity has been detected in the nucleus, lysosomes, and peroxisomes, as well as the intermembrane cisterns of the mitochondria [336, 337]. EC-SOD resides in the extracellular matrix, requires Cu2+ and Zn2+ for catalytic activity, and contains an intrasubunit disulfide bridge (Cys107-Cys189) that is necessary for enzymatic activity [338, 339, 344]. In the mitochondria, Mn-SOD dismutates O2•− released into the matrix, and is a pH sensitive enzyme with a loss of activity observed as pH rises [340–343].
Among these SODs, a common human gene variant in the heparin-binding domain of EC-SOD has been linked to cardiovascular disease risk. The R213G polymorphism results in a decreased affinity for heparin without a purported effect on specific enzymatic activity. This polymorphism has been linked to increased body weight and triglycerides in a Swedish population study of 4,925 randomly selected 74-year old subjects, increased mortality from ischemic heart disease and cerebrovascular events in patients with diabetes on renal replacement therapy, and a 2.3-fold increased risk of ischemic heart disease in a nested case-control study that included 1,912 individuals from a Danish population [345–347]. In experimental rat models, overexpression of R213G EC-SOD as compared to native EC-SOD failed to limit vascular O2•− and ONOO− levels, decrease blood pressure, or improve impaired endothelium-dependent vasodilation under basal conditions or in the setting of increased inflammation, suggesting a mechanistic basis for the observations in humans [348, 349]. Clinical studies have also shown a decrease in EC-SOD activity in aged individuals, African-Americans with hypertension, patients with vasospastic angina, thoracic aortic aneurysm patients, and in calcific aortic stenosis; however, genotyping was not performed in these studies, and, thus, it remains unknown if decreased SOD activity may be attributable to a heritable polymorphism [350–354].
A functional polymorphism has also been identified in the signal sequence of MnSOD (Ala16 to Val) that results in structural alterations to the mitochondrial targeting domain, implying that its decreased antioxidant potential may be a consequence of limited post-transcriptional transport to the mitochondria. A study of 989 individuals revealed that those who harbored this variant had an increased carotid intima-to-media thickness and were at increased risk for coronary artery disease and acute myocardial infarction [355, 356].
Increased vascular oxidant stress may also lead to a functional decrease in SOD activity through oxidative posttranslational modification of the enzyme. In an angiotensin II-infused rat model, Mn-SOD was subject to tyrosine nitration and accounted for ~20% of the total pool of nitrated proteins; this resulted in a 50% decrease in Mn-SOD activity in the absence of a decrease in protein expression . Similar findings were observed in the kidney of diabetic apoE−/− mice. In these studies, it was found that Tyr34 was exquisitely sensitive to nitration, and this posttranslational modification may result in a decrease in activity as this residue is located in the active site of the enzyme . Mass spectrometry revealed that Tyr45 and Tyr193 were also nitrated following exposure of the enzyme to ONOO− . Other studies have reported that when CuZn-SOD was exposed to ONOO−, nitration of Trp32 occurred, leading to a 6-nitrotryptophan modification with a corresponding 15% decrease in enzymatic activity [360, 361].
Catalase is located primarily in peroxisomes where it exists as a tetramer of four identical subunits that each contains a heme at the active site. Catalase reduces H2O2 to H2O in a reaction that occurs at the diffusion limit of H2O2 [362, 363]. During this reaction, the heme iron of catalase is oxidized by H2O2 resulting in the formation of an oxyferryl group with the porphyrin (Por) π radical and conformational conversion to Compound I (Por+•−FeIV=O). Compound I subsequently reacts with a second molecule of H2O2 to yield water and oxygen with return of the enzyme to the resting conformational state (Por−FeIII). A second state of catalase, Compound II (Por−FeIV=O), is formed by the reduction of Compound I; Compound II does not participate in the reduction of H2O2 but may revert to Compound I, albeit slowly. Catalase also binds NADPH to limit oxidative inactivation of the enzyme by H2O2 (i.e., formation of Compound II) and to facilitate tetramer formation [364–367].
Heritable catalase deficiency is an autosomal recessive disorder and reported solely in families of Japanese, Hungarian, or Swiss origin. These individuals have no overt phenotype with the exception of oral ulcers reported in Japanese individuals with acatalasemia. In a Hungarian cohort of 30 patients compared with 29 family members with normal catalase activity, several polymorphisms (G insertion at exon 2, GA insertion at exon 2, or T→G substation at intron 7) were associated with a 12.7% incidence of familial diabetes mellitus, elevated levels of plasma homocysteine, and lower levels of folate, suggesting an adverse risk profile for cardiovascular disease [368, 369]. Despite these findings, a recent analysis of 420 patients with type 1 and type 2 diabetes mellitus patients did not find an association between these polymorphisms and disease state, regardless of the degree of glycemic control .
Under conditions of oxidant stress, catalase has been shown to undergo tyrosyl nitration leading to a decrease in enzyme activity that results from oxidative modification of Cys377 to form a cysteic acid [371, 372]. Catalase is also subject to chlorotyrosyl modification in the presence of MPO and to protein carbonyl formation through the actions of oxidant species on lysine, arginine, proline, or threonine residues with a resulting decrease in intrinsic catalase activity [372, 373].
The glutathione peroxidases (GPx) are selenocysteine-containing enzymes that catalyze the reduction of H2O2 and lipid hydroperoxides to H2O and lipid alcohols, respectively, in a reaction that utilizes GSH as a reducing cosubstrate [374–377]. The glutathione peroxidases may also function as ONOO− reductases [378, 379]. There are 5 known forms of GPx: cellular GPx (GPx-1), gastrointestinal GPx (GPx-2), plasma GPx (GPx-3), phospholipid GPx (GPx-4), and sperm GPx (snGPx).
The importance of the GPx family of antioxidant enzymes in limiting the oxidative risk for atherothrombosis is increasingly recognized. Evidence from genetically modified mouse models has confirmed the role of GPx in vascular redox homeostasis. GPx-1- deficient mice demonstrate endothelial dysfunction and have spontaneous neointima formation, increased periadventitial inflammation, and collagen deposition surrounding the coronary arteries [380, 381]. When GPx-1−/− mice were crossed with apoE−/− mice to examine the influence of GPx-1 deficiency in atherosclerosis, decreased GPx-1 activity was associated with increased atherosclerotic lesion formation, increased vascular oxidant stress, and decreased •NO levels as compared to apoE−/− mice with normal GPx-1 activity . When these double knockout mice were made diabetic, increased atherosclerotic lesion formation was associated with elevated levels of adhesion molecules, receptors for advanced glycation end products, and proinflammatory and profibrotic markers . Conversely, overexpression of GPx-4, which reduces oxidized phospholipids, cholesterol hydroperoxides, as well as proinflammatory lipid peroxides generated by lipoxygenases and cyclooxygenases, has been shown to decrease vascular oxidant stress and the progression of atherosclerosis in apoE−/− mice . The central role of GPx-4 in modulating vascular redox tone was elucidated recently. Studies performed in a GPx-4 conditional knockout mouse model revealed that inhibition of GPx-4 resulted in activation of 12/15-lipoxygenase to initiate lipid peroxidation and stimulate apoptosis-inducing factor-mediated cell death [385, 386].
Clinical studies have shown that erythrocyte GPx-1 activity is an independent predictor of adverse cardiovascular events in individuals without traditional risk factors for atherosclerosis. In one study of 636 individuals with suspected coronary artery disease, individuals with the highest level of GPx-1 activity had the lowest risk of cardiovascular events a (HR = 0.29, 95% CI: 0.15, 0.58, p<0.001) compared to individuals with the lowest levels of enzyme activity . In a second study of 508 patients with documented coronary artery, carotid, and/or peripheral arterial disease followed for a median of 6.5 years, those individuals with the lowest levels of erythrocyte GPx-1 activity were more likely to experience a cardiovascular event than those with the highest levels (HR = 2.3; 95% CI: 1.4, 4.0 for lowest vs. highest GPx-1 activity, p=0.002 adjusted) .
Deficient GPx-1 activity may be acquired, as occurs in hyperhomocysteinemia, which has been shown to decrease expression of GPx-1 via an effect on translation, or associated with a heritable polymorphism [389, 390]. A Pro198Leu variant that decreases GPx-1 activity by 40% has been associated with increased carotid intima-media thickness, peripheral arterial disease, and the prevalence of cardiovascular disease . This polymorphism has also been associated with increased coronary artery calcification in diabetic patients, and an increased risk of coronary artery disease and thoracic aortic aneurysm [392–394].
As a large number of oxidant factors that mediate atherosclerotic risk are circulating in plasma, recent attention has focused on GPx-3, or plasma GPx, and its role in modulating oxidant stress. Deficient GPx-3 activity has been associated with increased platelet activation, decreased platelet •NO responsiveness, and cerebrovascular arterial thrombosis . Molecular examination of the GPx-3 promoter revealed seven polymorphisms within the promoter that are tightly linked and form two novel haplotypes. The H2 (less prevalent) haplotype was associated with decreased GPx-3 transcriptional activity, was more prevalent among patients with stroke and cerebral venous thrombosis, and conveyed an approximate two-fold increased, independent risk of thrombotic stroke in young patients [396, 397].
The glutathione-S-transferases (GSTs) are xenobiotic-metabolizing enzymes that detoxify reactive electrophiles, such as those contained in tobacco smoke, to limit their oxidizing capabilities. The GSTs are dimeric cytosolic enzymes that catalyze the conjugation of an active xenobiotic to GSH, an endogenous water-soluble substrate. Each dimer contains 2 active sites, which function independently, and each has a ligand binding site for GSH that is specific as well as an electrophilic binding site that is less specific . The N-terminus contains a catalytically essential cysteine, serine, or tyrosine residue that interacts with GSH to lower the pKa to 6–7 from its baseline value of 9 [399, 400]. These enzymes may catalyze Michael additions to α,β-unsaturated ketones as well as epoxide ring-opening reactions and nucleophilic aromatic substitutions to yield GSH conjugates, as well as reduce hydroperoxides to form GSSG [401, 402]. Once the GSH conjugate is formed, it is eliminated via several transporters, including the broad-specificity anion transporter of dinitrophenol S-GSH conjugates, ATP-dependent GS-X pump, P-glycoprotein, and the multidrug resistance-associated protein [403–408]. In addition to their catalytic role in detoxification, GSTs have also been found to possess selenium-independent peroxidase activity with hydroperoxides; steroid isomerization capacity; and bind and transport bilirubin, heme, bile salts, and steroids, in a process that is associated with a decrease in enzymatic activity [409–411].
Several studies have found an association between GST polymorphisms that decrease enzymatic activity and atherosclerosis. In fact, in 75 patients with established atherosclerosis, elevated levels of plaque DNA damage, as well as levels of the inflammatory markers C-reactive protein, fibrinogen, and adhesion molecules, were detected in individuals with the GSTM1*0 null allele [412, 413]. A meta-analysis revealed that smokers with the GSTM1*0 allele were at increased risk for coronary artery disease (OR 2.3; 95% CI: 1.4, 9.0), as were individuals with a GSTT1*1 null allele (OR 2.5; 95% CI: 1.3, 4.8) [414, 415]. Interestingly, smokers with angiographically documented coronary artery disease were more likely to harbor the GSTT1*1 allele than smokers who did not have significant disease . In addition to genetic polymorphisms influencing GST activity, GST is subject to tyrosine nitrosation and carbonylation (by reactive aldehydes) with a concomitant reduction in enzyme activity, although the precise functionalities subject to these posttranslational modifications have not yet been identified [417, 418].
Heme oxygenase catalyzes the oxygen-dependent regiospecific metabolism of heme to Fe(II), CO, and biliverdin, which, in turn, is converted to bilirubin via biliverdin reductase. Heme oxygenase exists as two isoenzymes: the inducible isoform HO-1 and the constitutive isoform HO-2. These isoenzymes similarly metabolize heme; however, HO-2 contains heme regulatory motifs that are absent in HO-1, and its primary function has been suggested to be the generation of CO to serve as a signaling molecule [419– 422]. The observation that HO-1 is induced by oxidant stress has led to consideration of this enzyme as an antioxidant. The antioxidant properties attributed to HO-1 result from its ability to degrade the heme moiety released by destabilized or denatured proteins bearing it and, thereby, limit LDL oxidation ; to generate vasoprotective levels of CO that limit cell proliferation and inflammation, and stimulate vasodilation, predominantly through cGMP-independent effects on Ca2+-dependent potassium channels [424–427]; and to increase levels of bilirubin, which itself is a free radical scavenger that inhibits NADPH oxidase enzyme assembly and activity to decrease O2•− production [428, 429]. HO-1 is induced by 1-palmitoyl-2-isoprostanoyl-sn-glycero-3-phsophorylcholine and linoleyl hydroperoxide contained within oxLDL, and its expression has been detected in early fatty streaks as well as advanced atherosclerotic lesions [430–433]. Induction of HO-1 through ROS-responsive transcription factor(s), such as Nrf-2, results in increased levels of bilirubin and CO to decrease oxidant stress, and has been shown to limit atherosclerotic lesion formation and vein graft failure in mouse models of hypertension and cardiac hypertrophy [434–436].
A number of studies have examined HO-1 polymorphisms and their association with atherothrombotic vascular disease. Recent studies have focused on a (GT)n repeat in the HO-1 promoter and found that as the repeat lengthened, ROS were less effective at inducing HO-1 expression . In fact, in 110 patients with established coronary artery disease, and presumably increased systemic oxidant stress, the ability to induce HO-1 in peripheral monocytes was negatively correlated with the degree of atherosclerosis . Studies have also shown that short repeats (<25 GT) were associated with a decreased risk of restenosis in 472 patients with peripheral arterial disease following peripheral intervention (HR = 0.46, 95% CI: 0.24, 0.87, p=0.016), while a study of 323 patients with successful coronary artery stenting procedures revealed that long repeats (>25 GT) were associated with an increased risk of restenosis (OR = 3.74, 95% CI: 1.61, 8.70, p=0.002) as well as major adverse coronary and cerebrovascular events [439, 440]. In 263 diabetic patients, those with long repeats (> 27 GT) had lower levels of serum bilirubin and an increased risk of cardiovascular disease after controlling for conventional coronary heart disease risk factors (OR 2.81; 95% CI 1.22, 6.47, p=0.015), compared to individuals with short repeats (< 27 GT) .
Heme oxygenase activity is also regulated at the posttranscriptional level. Heme oxygenase-1 may undergo tyrosine nitration or sulfhydryl oxidation in the presence of ONOO−, although neither modification has been proven conclusively . In contrast, HO-2 activity may be regulated by oxidative modification. HO-2 contains an intramolecular disulfide bond that links Cys265 of one heme regulatory motif with Cys282 of a second heme regulatory motif; this covalently linked conformation binds heme tightly. When this disulfide bond is reduced, elevated levels of free heme are detectable in cells. Therefore, reduction of the disulfide bond would likely increase cellular heme, decrease enzyme activity, and decrease atheroprotective levels of CO and bilirubin .
Glucose-6-phosphate dehydrogenase (G6PD) is the first and rate-limiting enzyme in the pentose phosphate pathway and the principal source of intracellular NADPH. NADPH is utilized as a reducing equivalent to maintain thiol redox balance and as a cofactor by other antioxidant enzymes. The G6PD gene is highly polymorphic with a large number of recognized functional variants ; G6PD deficiency is the most common enzymopathy worldwide with more than 440 biochemical variations owing to ~120 different mutations. The majority of described mutations are single or double missense mutations that have arisen owing to selective pressure for decreased G6PD activity in the malaria belt to render erythrocytes more susceptible to oxidative damage . As a result, erythrocytes parasitized with plasmodium species are cleared by the reticuloendothelial system at a faster rate than those with normal G6PD activity .
The central role of G6PD in modulating the cardiovascular redox milieu has been demonstrated by our group. G6PD limits oxidative injury in vascular endothelial cells, smooth muscle cells, and cardiomyocytes and has been associated with a decrease in angiogenesis and worsening indices of left ventricular function following ischemia-reperfusion injury in a G6PD-deficient mouse model [447–450]. In addition to genetic G6PD deficiency, G6PD deficiency may also be acquired. For example, G6PD activity was decreased in the kidneys of diabetic rats with a normal G6PD genotype and reduced by 80% in mice with a normal genotype that were fed a high fat diet rich in polyunsaturated fats [451, 452]. In the setting of hyperaldosteronism, G6PD expression and activity were decreased leading to increased vascular oxidative stress, decreased •NO, and impaired vascular reactivity in mice with a normal G6PD genotype . Clinically, G6PD deficiency has been linked to an increased risk for hypertension, gestational hypertension, preeclampsia, and diabetes, all of which are associated with vascular endothelial dysfunction [454–457].
Although the aforementioned studies implicate increased activation or dysfunction of individual ROS-generating or antioxidant enzymes, respectively, in the pathogenesis of atherosclerosis, it is more likely that functional alterations occur in several enzymatic systems concurrently to render a pro-oxidant vascular redox milieu. The accumulation of biologically active free radicals and their ability to oxidize a range of target molecules provide support for the role of oxidant stress in the pathogenesis of atherothrombosis. Prevailing cholesterol theory dictates that oxidant stress is a mechanistic necessity for atherogenesis; the oxidative modification hypothesis contends that only oxLDL are taken up by macrophages to form foam cells and serve as the nidus for the developing atherosclerotic plaque [458, 459]. For this reason, we now turn to a discussion of the role of ROS in lipoprotein oxidation as a precursor to atherothrombotic vascular disease.
During the process of LDL oxidation, ROS target polyunsaturated fatty acids that contain bis-allyllic hydrogens to oxidize them (Fig. 2). These oxidized fatty acids degenerate to form a number of reactive low-molecular-weight aldehyde species, which, in turn, react with the ε-amino groups of lysine residues of apoB-100 to modify the surface charge of LDL particles [212, 460]. Proteomic analysis of apoB-100 and oxLDL has also identified histidine and tryptophan residues as targets for oxidative modification . In the presence of MPO-derived reactive nitrogen species, human LDL particles are also subject to tyrosine nitration of apoB-100, rendering these nitrated LDL species proatherogenic and capable of initiating lipid peroxidation . Once peroxidation of phospholipids present in the outer layer of LDL occurs, the particle promotes a conformational change in apoB-100 such that it is displaced from the LDL particle’s hydrophobic zone into the aqueous phase, leading to increased nonreceptor-mediated capture of the oxLDL particle by vascular cells .
LDL may undergo oxidation of particle components other than apoB-100; for example, minimally oxidized forms of LDL were found to contain polyoxygenated cholesteryl ester hydroperoxides. Moreover, evidence exists that minimally oxLDL is formed in vivo and has been identified in extracts of atherosclerotic lesions isolated from apoE−/− mice . Minimally oxLDL has been shown to activate macrophages and induce cell membrane ruffling and spreading via CD14. Toll-like receptor 4 also binds to minimally oxLDL leading to the secretion of the proinflammatory cytokines, including monocyte chemotactic protein-1, tumor necrosis factor-α, and interleukin-6, to enhance monocyte binding to endothelial cells and induce platelet shape change and aggregation [463–465].
Following oxidative or nitrosative modification, LDL is taken up by macrophages via scavenger receptor pathways resulting in the formation of cholesteryl ester-rich foam cells. Similarly, endothelial cells also take up oxLDL via the lectin-like oxidized LDL receptor-1 . Intracellular endothelial accumulation of oxLDL results in diminished •NO production, increased expression of leukocyte adhesion molecules, promotion of a prothrombotic surface, and synthesis of smooth muscle cell mitogenic factors [466–468]. This results, in part, from oxLDL-mediated NADPH oxidase activation to increase endothelial cell oxidant stress; carbonylation and nitration of proteins, including the epidermal growth factor receptor and eNOS; and increased proteasomal degradation of eNOS . These functionally-deficient endothelial cells, together with atherogenic macrophage-derived foam cells, facilitate atherosclerotic lesion formation as confirmed by the presence of oxLDL in human atherosclerosis. Immunohistochemical analysis of atherosclerotic plaques demonstrates abundant staining for apoB-100, which is absent in human arteries without evidence of atherosclerosis . Furthermore, therapies aimed at reducing oxLDL in preference to LDL have been shown to decrease experimental atherosclerosis: gene transfer of LOX-1 to the livers of apoE−/− mice decreased plasma oxLDL levels and inhibited atherosclerosis progression without influencing total cholesterol or LDL levels . Interestingly, there is also evidence to suggest that macrophages may take up nonoxidized LDL, which subsequently undergoes oxidative modification to form oxLDL within the cell. In vitro studies have demonstrated that macrophages take up acetylated LDL, which is shuttled to lysosomes that serve as the intracellular site for lipid particle oxidation .
Oxidized LDL may exit the subendothelial space and has been measured in blood, where it has been reported to account for 0.001% of the total LDL in normal individuals and to increase as high as 5% of total LDL in patients with an acute myocardial infarction [473–476]. The presence of circulating oxLDL has been supported by the detection of plasma autoantibodies to oxLDL, and the observation that human plasma possesses immunoreactivity to oxLDL epitopes [477–479]. Elevated plasma levels of oxLDL, which presumably reflect the vascular burden of oxLDL, correlate with incident metabolic syndrome, peripheral arterial disease, acute coronary syndromes, angiographically documented coronary artery disease, and stroke [480–483].
High-density lipoproteins (HDL) derive their atheroprotective functions from their ability to reverse transport cholesterol from the vasculature to the liver; however, components of HDL are subject to oxidative modification to generate oxidized lipoprotein particles that are dysfunctional, proinflammatory, and paradoxically promote atherosclerotic lesion formation [484–487]. In vitro studies have shown that metal ions, ROO• and •OH, aldehydes, tyrosyl radicals, lipoxygenase, and HOCl oxidize HDL, while in vivo it has been suggested that HDL oxidation results from exposure to MPO, O2•−, H2O2, ONOO−, and lipoxygenases [485, 486, 488–497]. Once oxidized, HDL exhibits an increase in fluidity that is associated with impaired reverse cholesterol transport and anti-inflammatory properties as well as cytotoxicity towards cells [487, 489, 498–503].
High-density lipoprotein particles are heterogeneous and composed of free cholesterol, phospholipids, and apolipoproteins that surround a triglyceride- and cholesteryl ester-rich core. High-density lipoprotein particles also carry enzymes with antioxidant properties, including paraoxonase-1 and lecithin:cholesterol acyltransferase (LCAT) [504–506]. Within the HDL particle, these antioxidant enzymes serve to limit free radical oxidation of circulating LDL and HDL particles as well as those particles resident in the vessel wall. In fact, plasma HDL is the predominant carrier of lipid hydroperoxides that are reduced to their corresponding hydroxides when apolipoprotein A-1 (apoA-1) methionyl residues are modified to methionine sulfoxide [493–495].
Under basal conditions, HDL maintains reverse cholesterol transport by enhancing apoA-1-mediated cholesterol efflux from macrophages via the ATP-binding cassette (ABC)A1 and esterification of cholesterol by LCAT. HDL also prohibits oxidation of phospholipids in LDL through apoA-1 and paraoxonase 1, and inhibits inflammatory cytokine and adhesion molecule expression. In this manner, both oxidation of LDL and uptake by macrophages are limited as is oxLDL-mediated O2•− generation [493–495, 507– 514]. The importance of HDL in modulating inflammation and atherosclerosis may be ascertained by studies of patients treated with recombinant HDL. A single infusion of recombinant HDL in 7 hypercholesterolemic men improved endothelial function and brachial artery flow-mediated dilation . In 36 patients with acute coronary syndromes and established atherosclerosis, treatment with apoA-1Milano, an anti-atherogenic apoA-1 variant, weekly for 5 weeks resulted in a significant decrease in total coronary artery atheroma volume compared to that observed in 11 placebo-treated individuals .
In contrast, in a pro-atherogenic milieu, HDL transition to a proinflammatory phenotype is characterized by several structural changes: there is a decrease in apoA-1 (or it may form dimers, trimers, or aggregates with other apolipoproteins), paraoxonase-1, and LCAT content, and the particle becomes enriched in apolipoprotein J and serum amyloid A [488, 491, 493, 494, 497, 498, 517–520].
Notably, the pro-atherogenic environment is characterized by an influx of inflammatory macrophages that generate O2•−, H2O2, ONOO−, and, via MPO, HOCl and NO2− [68, 521]. Both MPO and lipoproteins modified by HOCl have been detected in atherosclerotic lesions [498, 522]; oxidative modification of protein tyrosyl residues to form 3-chlorotyrosine and 3-nitrotyrosine has been detected in HDL isolated from atherosclerotic plaques . In fact, HDL particles isolated from patients with established cardiovascular disease demonstrated 13-fold higher levels of protein-bound 3-chlorotyrosine as well as higher levels of 3-nitrotyrosine compared to control subjects, indicating that MPO-derived species oxidized HDL in vivo [76, 523].
The functional consequences of MPO-mediated oxidatively modified HDL were determined by examining the effect of oxidation of apoA-1 on cholesterol efflux from macrophages. Here, chlorination of apoA-1 resulted in a marked decrease in cholesterol efflux, even greater than that observed with nitration of apoA-1 [523, 524]. Mass spectrometry revealed that Tyr192 of apoA-1 was the predominant chlorination site in apoA-1, and that chlorination of this residue correlated with loss of (ABC)A1 transport activity [523–525]. As apoA-1 Tyr192 is located near a lysine residue and HOCl reacts with lysine ε--amino groups to form chloramines, the mechanism by which HOCl chlorinates Tyr192 likely requires participation of the lysine residue, which was confirmed in site-directed mutagenesis studies [524, 526]. Interestingly, methionine residues inhibit tyrosine chlorination by scavenging chlorinating intermediates [526, 527]; however, methionine oxidation in the absence of tyrosine chlorination does not inhibit apoA-1 cholesterol efflux activity . Furthermore, site-specific substitution of methionine to valine increased the sensitivity of apoA-1 to MPO-generated oxidants confirming the role of methionine as a local residue that limits tyrosine chlorination . In addition to Tyr192 chlorination, apoA-1 cholesterol efflux is also impaired by acrolein that is generated by MPO and modifies Lys226 . Other studies using hydrogen-deuterium exchange mass spectrometry analysis of apoA-1 have implicated Tyr166 as a residue that preferentially undergoes oxidative modification . This residue interacts with and activates LCAT. Thus, oxidative modification of apoA-1 Tyr 166 may influence cholesterol esterification by modulating LCAT function.
It has been suggested that polymorphisms in the paraoxonase 1 promoter or gene may facilitate oxidative modification of HDL and predispose individuals to atherosclerosis; however, the results from clinical studies have been largely contradictory. Paraoxonase 1 limits oxidation of LDL and HDL and mice deficient in this enzyme have been shown to have increased susceptibility to atherosclerosis [531–533]. To date, two coding variants, Q192R and L55M, and five promoter polymorphisms have been identified and a functional decrease in paraoxonase 1 activity has been reported in patients with types 1 and 2 diabetes mellitus, familial hypercholesterolemia, renal disease, and established coronary heart disease [534–539]. Despite these observations, the ability to link this decrease in HDL antioxidant capacity with a particular polymorphism has been challenging and has been attributed, in part, to different polymorphism distributions between races and within populations . Moreover, a recent meta-analysis of 43 studies that included 11,212 cases of coronary heart disease and 12,786 control subjects revealed that only the Q192R polymorphism was associated with a slight increase in the risk for coronary heart disease . Furthermore, many of the early genetic studies failed to correlate the polymorphisms with activity. Therefore, clarity about the relationship between paraoxonase 1 polymorphisms, HDL redox status, and their influence on atherogenesis awaits further study .
Clinical studies do, however, confirm that HDL, possibly through the aforementioned post-translational modifications, can also mediate atherosclerosis. Patients with established coronary heart disease despite very high levels of HDL cholesterol (≥84 mg/dl) have impaired functional HDL responses. Compared to healthy control subjects with similar HDL levels, HDL isolated from these patients exhibits a proinflammatory response in monocyte chemotactic assays and increases LDL oxidation . Similar findings have been observed in patients with stable coronary heart disease or poorly controlled diabetes mellitus, in whom HDL particles exhibited decreased antioxidant enzyme activity and increased lipid hydroperoxide content [542, 543]. Patients with systemic inflammation attributable to non-coronary heart disease, including chronic kidney disease, rheumatoid arthritis, systemic lupus erythematosus, and solid organ transplant recipients, also demonstrate proinflammatory HDL. Interestingly, these patients are also at higher risk for developing atherosclerosis [544–548].
In addition to the important role that oxidative modification of lipids and lipid-associated proteins plays in atherogenesis, there is compelling evidence to demonstrate that neutrophils participate in atherosclerosis. Epidemiological studies have demonstrated that the neutrophil-lymphocyte ratio on hospital admission for an acute coronary syndrome has prognostic value. In one study of 422 patients with ischemia and angiographically documented coronary artery disease, the neutrophil-lymphocyte ratio was found to be an independent predictor of cardiac death (HR = 8.13, p=0.02), and those patients with the highest ratio had a 10.2% decrease in event-free survival . Similarly, a study of 2,833 patients with myocardial infarction revealed that patients with the highest neutrophil-lymphocyte ratio had higher in-hospital (8.5% vs. 1.8%) and 6-month (11.5% vs. 2.5%) mortality as compared to patients with the lowest ratios. When adjusted for patient clinical characteristics, individuals with the highest ratios continued to have an increased risk for in-hospital (OR = 2.04, p=0.013) and 6-month mortality (OR = 3.88, p < 0.001) . Furthermore, the admission neutrophil-lymphocyte ratio on admission for acute myocardial infarction was found to predict independently an increased risk of mortality over a 2–3 year follow-up period [551, 552].
Animal models have demonstrated a predilection of neutrophils for atherosclerotic plaques. In apoE−/− mice crossed with mice homozygous for a knock-in mutation of green fluorescent protein in the lysozyme M locus, fluorescent neutrophils and monocytes were detected at the periphery of atherosclerotic plaques after 26 weeks . In LDLR−/− mice, neutrophils were found attached to the plaque cap in intermediate and advanced atherosclerotic lesions after 6 weeks . In human atherosclerotic plaques isolated by atherectomy or at autopsy neutrophil infiltration was also observed, mainly at the sites of plaque erosion or rupture .
Neutrophil recruitment to atherosclerotic plaques has been shown to be regulated by the CXC receptor 4/CXC ligand 12 axis; disruption of this axis induces a release of hematopoietic stem cells and leukocytes, including neutrophils from the bone marrow [556, 557]. CXCL 12 has been detected in atherosclerotic plaques and an inverse relationship exists between levels of CXCL 12 and unstable coronary artery disease [558, 559]. In apoE−/− or LDLR−/− mice, blockade of CXC receptor 4 resulted in increased numbers of neutrophils as well as elevated neutrophil content in atherosclerotic plaques .
The contribution of neutrophils to atherothrombotic vascular disease is, in part, related to production of reactive free radical species via NAD(P)H oxidase, MPO, and iNOS that increase oxidant stress and promote lipid peroxidation; these mechanisms link inflammation with oxidant stress. Fully activated neutrophils have been shown to increase O2•− production and contribute to systemic oxidant stress in hypertension, diabetes, tobacco use, and hyperlipidemia [561–564]. Studies performed with neutrophils isolated from hypercholesterolemic patients demonstrated increased phorbol myristate acetate-stimulated O2•− production and adherence to endothelial cells as compared to those neutrophils isolated from control subjects . Ex vivo studies performed with human neutrophils also revealed that they oxidize LDL thiols by MPO-derived HOCl and N-chloramine-dependent mechanisms .
Circulating monocytes undergo phenotype transition to macrophages that accumulate lipid and serve as the nidus for atherosclerotic lesion formation (reviewed in ). Early studies that implicated monocytes in atherogenesis were performed in cholesterol-fed rabbits where examination of the aorta revealed accumulations of lipid-filled macrophages. Prior to the appearance of visible atherosclerotic plaques, macrophages were adherent to the endothelial surface and had migrated to the subendothelial space. In advanced atherosclerotic lesions, macrophage foam cells resided in the superficial layers of the plaque and within areas of necrosis. After cessation of the high cholesterol diet, the number of macrophages in the vessel wall decreased significantly, implying that monocytes participated in the initial phase of atherogenesis . Monocytes and macrophages were also identified in fibro-fatty lesions and to a lesser extent in fibrous plaques and advanced lesions . Immunohistochemical studies using FOH1a/DLH3, a monoclonal antibody that recognizes oxidized phosphatidylcholine in oxLDL, revealed accumulation of oxLDL in macrophages in human atherosclerotic plaques [570, 571]. Other studies that examined the progression of atherosclerosis in mice deficient in chemokines and their receptors that promote monocyte recruitment, including CCL2, CCR2, CCL5, and CCR5, revealed that genetic deletion of these chemokines or receptors increased atherosclerosis. Similarly, animal models in which monocytes or macrophages were selectively depleted reveal that atherogenesis, but not established atherosclerosis, is decreased implying that monocytes and macrophages play a critical role in early atherogenesis [572, 573]. Furthermore, the degree of monocyte accumulation in the vessel wall was found to correlate with plaque size .
Monocytes contribute to atherogenesis by altering the local redox state, thereby providing another link between inflammation and oxidant stress. Monocytes possess an NAD(P)H oxidase that is differentially regulated compared to neutrophil NAD(P)H oxidase, generates O2•− with peak production at 1 hour, and are capable of lipid peroxidation via NAD(P)H oxidase activation in a Ca2+- and phospholipase A2-dependent manner [575–579]. In human atheroma isolated by directional atherectomy, macrophage infiltration and O2•− production were increased in plaques isolated from patients with acute coronary syndromes compared to those obtained from patients with stable angina. In these atherosclerotic plaques, O2•− colocalized with NAD(P)H oxidase subunits and oxLDL deposition . The relative contribution of monocyte/macrophage-derived O2•− to atherosclerosis was determined through bone marrow transplantation studies performed with p47phox−/− mice. In these studies, deficiency of NAD(P)H oxidase activity decreased monocyte/macrophage infiltration of the vessel wall, plasma oxLDL levels, aortic O2•− production, and atherosclerotic lesion formation . Monocytes also generate reactive nitrogen species, and studies performed with isolated human monocytes in the presence of NO2− were found to nitrate tyrosine residues of apoB-100 and initiate LDL lipid peroxidation. Under these conditions, both LDL nitration and lipid peroxidation were found to require monocyte activation and NO2− but were inhibited in the presence of catalase and MPO inhibitors. Further study revealed that at low rates of •NO flux, monocyte-mediated LDL oxidation occurred via a MPO-H2O2 pathway, while at higher levels of •NO flux, this process was more dependent upon ONOO− formation . Immunohistochemical studies to identify HOCl-modified epitopes in human atherosclerotic plaque using a monoclonal antibody directed against human HOCl-modified LDL demonstrated abundant staining in fibroatheroma and complex lesions that localized to monocytes/macrophages .
Platelets, which are essential for primary hemostasis, promote atherothrombotic vascular disease by contributing to plaque genesis and expansion under conditions that favor increased oxidative or nitrosative stress [582–587]. Much experimental evidence supports the role of platelets in atherogenesis. Studies performed in apoE−/− mice demonstrate that platelets are adherent to the endothelium prior to development of visible atherosclerotic plaques . Similarly, mice deficient in von Willebrand factor, which recruits platelets to sites of vascular injury, have less atherosclerosis than wild-type mice . Furthermore, the ablation of platelet-derived thromboxane delays atherogenesis in atherosclerosis-prone LDLR−/− mice [583, 584]. Platelets also facilitate the progression of atherosclerosis through the surface expression and release of P-selectin to stimulate monocytes and macrophages to synthesize and release chemoattractrants or growth factors; increased P-selectin levels have been shown to modulate atherosclerosis in apoE−/− and LDLR−/− mice [586, 587, 590, 591].
Platelet reactivity is regulated by vascular endothelium-derived prostacyclin, which inhibits platelets via increased intracellular cAMP levels; ecto-ADPase, which decreases plasma levels of ADP and ATP to limit platelet recruitment; and •NO, which diffuses into platelets and modulates platelet function through cGMP-dependent and –independent processes. Platelets themselves express NOSs and synthesize •NO, albeit at lesser amounts than endothelial cells ; however, platelet-derived •NO has been shown to modulate primary platelet aggregation, contribute to platelet disaggregation, and inhibit platelet recruitment to the growing thrombus [592–595]. Increased levels of platelet •NO stimulate soluble guanylyl cyclase to elevate cGMP levels, resulting in decreased intracellular Ca2+ stores, phosphorylation and inhibition of the thromboxane A2 receptor, and inhibition of phosphodiesterase type 3 to elevate cAMP levels and limit platelet aggregation [596–598]. Elevated cGMP levels also inhibit phosphatidylinositol 3-kinase to prevent glycoprotein IIb/IIIa activation . Interestingly, •NO inhibits platelet reactivity in a number of cGMP-independent mechanisms, including decreased ATP-dependent Ca2+ uptake by platelets, and by limiting exocytosis of platelet granules via S-nitrosation of N-ethylmaleimide-sensitive factor .
Reactive oxidant species released by vascular cells or generated by platelet themselves directly influence platelet responses, in part, by reacting with •NO and inhibiting the redox-sensitive ecto-ADPases [601–605]. Ex vivo studies of platelet reactivity have shown that platelets exposed to O2•−-generating systems have a decreased threshold for activation by collagen, thrombin, and ADP [604, 606]. Platelets themselves also produce free radical oxidants as evidenced by a burst of O2 consumption, O2•− production, and GSSG formation during platelet aggregation [607, 608]. There appears to be several platelet sources of ROS as thrombin-stimulated aggregation is accompanied by mitochondrial generation of H2O2, which is believed to mediate thrombin-induced apoptosis . Platelets also express NAD(P)H oxidase, and activation of this enzyme to produce O2•− occurs concomitant with platelet activation [610, 611]. Platelet NAD(P)H oxidase activation stimulates glycoprotein IIb/IIIa and αIIbβ3 integrin activation, but does not influence α-granule or dense granule release [612, 613]. Increased platelet NAD(P)H oxidase-derived O2•− has also been shown to induce oxidative modifications of LDL as demonstrated by increased conjugated diene and lysophosphatidylcholine formation in collagen-stimulated platelets . Enhanced lipid peroxidation, in turn, increases platelet activation and adhesion as observed in platelets exposed to F2-isoprostanes . Activation of platelet NAD(P)H oxidase has also been shown to regulate platelet release of CD40 ligand (CD40L), which initiates an inflammatory response by increasing adhesion molecule expression and chemokine secretion by endothelial cells [615, 616]. Redox sensitive interactions between CD40-CD40L increase the adherence of platelets to the vessel wall . Ex vivo studies performed with washed platelets revealed that uncoupled eNOS may also contribute to platelet O2•− generation, which has pathophysiological implications for atherothrombosis; however, it remains unknown if this observation resulted from depletion of cofactors that occurred during platelet processing . Furthermore, it is now recognized that platelet agonists target either intra- or extracellular O2•−-generating enzymes, and selective activation of these different enzymatic sources influences the downstream effect(s) on platelet signaling. For example, both intra- and extracellular O2•− may modulate CD40L expression and release while studies performed with extracellular antioxidants suggest that only extracellular O2•− stimulates P-selectin expression, and intracellular O2•− appears to regulate αIIbβ3 integrin activation .
Studies examining the direct effects of ONOO− on platelet reactivity have yielded conflicting results. Early studies suggested that ONOO− inhibited platelet aggregation in response to ADP, collagen, or thrombin when performed in the presence of plasma, while more recent studies performed by generating endogenous ONOO− at nanomolar concentrations revealed that ONOO− formation increased concomitant with prostaglandin endoperoxide H2 synthase-1 activity, thromboxane A2 release, and platelet aggregation [601–603, 619]. In these studies, inhibition of either •NO or O2•− generation limited platelet aggregation . A number of plausible mechanisms have been advanced to explain these seemingly discordant results, including the use of pharmacological concentrations of ONOO− in some studies, the formation of S-nitrosothiols that act as a reservoir of •NO, agonist-dependent differences in the relative amount of •NO and O2•− formed resulting in non-steady state levels of ONOO−, and •NO-mediated S-glutathiolation of SERCA to lower intracellular Ca2+ levels [61, 602, 620, 621].
In the setting of hyperlipidemia, biologically active oxidized lipids implicated in atherogenesis have also been shown to promote platelet aggregation. Exogenously oxidized lipoproteins or oxidized LDL and HDL isolated from hypertriglyceridemic subjects increased the ADP-stimulated maximal aggregation rate of platelets compared to nonoxidized lipoproteins . Oxidation of LDL yields a family of oxidized choline glycerophospholipids (oxPCCD36) that function as ligands for the macrophage scavenger receptor CD36, have been demonstrated in vivo, and have been implicated in atherosclerosis [216–218]. Platelets also express CD36 and studies performed in hyperlipidemic apoE−/− and LDLR−/− mice as well as with human plasma revealed that oxPCCD36 accumulated in the plasma, bound to platelet CD36, and induced platelet activation . Mechanistically, CD36-mediated platelet activation occurs through JNK phosphorylation .
Increased oxidant stress also favors a procoagulant state by influencing the function of key mediators of fibrinolysis. Fibrinogen is subject to a number of posttranslational modifications that result from oxidation, nitration, and homocysteinylation and influence its function. Oxidative modification of fibrinogen results in an increased carbonyl-to-protein molar ratio and decreased α-helical content of the protein. Oxidized fibrinogen forms fibrin at an accelerated rate, supports platelet aggregation, and is less efficient in stimulating plasminogen activation by tissue- plasminogen activator than non-oxidized fibrinogen . Nitration of fibrinogen β-chain at Tyr292 and Tyr422 significantly accelerated thrombus formation with clots demonstrating large bundles of twisted, thin fibrin fibers with distorted clot architecture, and increased fibrin stiffness [626, 627]. Fibrinogen also undergoes posttranslational modification by homocysteine thiolactone, a metabolite of homocysteine, to yield fibrin clots that are composed of thin, densely packed fibrin fibers with increased resistance to fibrinolysis owing to homocysteinylation of twelve lysine residues and possible steric interference with the tissue plasminogen activator and plasminogen binding sites . Oxidation of annexin II, an endothelial cell receptor for tissue plasminogen activator, by homocysteinylation of Cys9 leads to decreased tissue plasminogen activation, binding, and decreased activation of cell-surface plasminogen activation and fibrinolysis .
Oxidant stress also increases the transcription of plasminogen activator inhibitor-1 (PAI-1) that limits the effects of tissue plasminogen activator . While clinical studies often find that there are increased levels of tissue plasminogen activator in patients at risk for atherosclerosis, this increase is accompanied by elevated PAI-1 levels that lead to a net inhibition of tissue-plasminogen activator activity and, therefore, explain the prothrombotic phenotype in the setting of elevated levels of tissue-plasminogen activator . Although smokers, who have increased indices of oxidant stress, also have an elevated baseline release of tissue plasminogen activator and PAI-1, stimulated release of tissue plasminogen activator was significantly reduced compared to nonsmokers suggesting that a net imbalance that favored a prothrombotic state existed [9, 632]. Furthermore, in the presence of minimally oxLDL, a proatherogenic lipoprotein that contains oxidized phospholipids, endothelial cell PAI-1 protein levels were increased 2-fold and activity increased by 35% while tissue plasminogen activator protein levels decreased by 45% indicating a prothrombotic phenotype .
Coagulation is supported by tissue factor, which binds to and allosterically activates factor VIIa to generate thrombin for physiological secondary hemostasis. Vascular endothelial cells, smooth muscle cells, monocytes, and macrophages all express tissue factor in response to inflammatory mediators, such as tumor necrosis factor-α, interleukin-1β, CD40L, and oxLDL [634–646]. Platelets also express tissue factor, although this occurs via microparticle-bound tissue factor exchange . In addition, a soluble, alternatively spliced form of tissue factor with procoagulant activity has been detected in blood; however, its clot-forming potential remains controversial [648, 649].
Oxidized phospholipids present in minimally oxLDL have been shown to stimulate coagulation by increasing expression of tissue factor via two distinct signaling pathways: activation of PKC/ERK/EGR-1 and Ca2+/calcineurin/NFAT . Oxidized phospholipids have also been shown to decrease the expression of thrombomodulin protein levels by 40% resulting in a 30% reduction in activity. The oxidation products of 1- and/or 2-oleoyl phosphatidylcholine or phosphatidylethanolamine were also found to reduce tissue factor pathway inhibitor (TFPI) activity by interacting directly with the C-terminal basic region of TFPI . In fact, in patients with coronary artery disease, elevated plasma levels of TFPI were found to be higher than in control subjects and associated with higher levels of oxidized lipids .
Tissue factor coagulant activity is dependent upon an extracellular Cys186-Cys209 disulfide bond . The redox state of this bond is modulated by protein disulfide isomerase that targets the bond to suppress coagulation via thiol/disulfide exchange; this action is regulated by •NO . Thus, in the setting of increased oxidant stress resulting in limited •NO, this mechanism will favor the procoagulant effects of tissue factor.
Atherogenesis involves mediators of the immune system, and studies performed in immunodeficient mice or apoE−/− mice crossed with interferon-γ−/− mice reveal a significant delay in both the onset and progression of atherosclerotic plaque formation . Immunohistochemical analysis of human atherosclerotic plaques, which have been shown to contain polyclonal memory T lymphocytes that are reactive against oxLDL epitopes, further supports these observations. CD4+ lymphocytes and immune cell derived-cytokines have been shown to activate NAD(P)H oxidase, increase O2•− production, and impair vascular reactivity, early findings in atherosclerosis (Fig. 3) . Together, these studies implicate oxidant stress as an integral component of the cellular immune response associated with atherosclerosis.
In advanced human atheromas, T cells comprise approximately 10–20% of the total cell population and are of the T helper-1 subset . These T cells recognize antigens in a major histocompatibility complex class II manner, as well as secrete proinflammatory cytokines, including interferon-γ, interleukin-2, and tumor necrosis factor-α, that activate macrophages to increase O2•− production . In a mouse model, it was shown that mice that mount a predominantly T helper-1 subset immune response have increased levels of serum interleukin-6 and develop significantly more atherosclerosis than those that mount a T helper-2 cell response, independent of lipid levels .
The role of oxLDL as a candidate autoantigen has been confirmed experimentally and in clinical studies [660, 661]. In apoE−/− mice with atherosclerosis, CD4+ and CD8+ T cells have been identified in the plaques, and the mice have increased levels of circulating antibodies against oxLDL . Patients who present with acute coronary syndromes have been shown to have high serum levels of oxLDL antibodies that are decreased following treatment with a statin [663, 664]. Furthermore, scavenger receptors expressed by macrophages, including CD36, CD68, CXCL16, scavenger receptor-A and –B, and LOX-1, modulate the uptake of oxLDL for presentation to antigen-specific T cells [665– 671]. This effect, in turn, leads to upregulation of toll-like receptor-2 and toll-like receptor-4, and oxidized derivatives of LDL have been shown to bind toll-like receptor-4-CD14 complexes on monocytes/macrophages to augment the immune response . Antibodies to oxLDL also bind to other oxidatively modified targets, including oxidized phosphatidylcholine generated on the surface of bacteria or endothelial cells as well as oxidation-specific epitopes on apoptotic cells .
In addition to oxLDL, other ROS-related candidate antigens include lipoprotein(a), advanced glycation end products, heat shock proteins, and bacterial-derived antigens. Lipoprotein(a) has been shown to be the preferential carrier of oxidized phospholipids in human plasma, and in vitro transfer studies revealed that oxLDL donates these oxidatively modified phospholipids to lipoprotein(a) in a time- and concentration-dependent manner . Cross-reactivity between these candidate antigens with antibodies that recognize oxLDL epitopes augments immune activation . Advanced glycation end products facilitate antigenic stimulation in atherosclerosis by increasing ROS levels to induce maturation of dendritic cells that stimulate T cell proliferation and cytokine secretion . Advanced glycation end products stimulate the immune response further by signaling through toll-like receptors to propagate an immune-inflammatory response . Oxidant stress has also been shown to increase expression of lymphocyte Hsp60 and members of the heat shock protein (Hsp) family, including mycobacterial Hsp65 and chlamydial Hsp60 that resemble human Hsp60 and exist in atherosclerotic lesions in experimental animal models [676–680]. Furthermore, immunization with Hsp65 or Hsp60 augments atherosclerosis in both mice and rabbits [677, 681]. Interestingly, the antioxidant capacity of macrophages may regulate the chronicity of Chlamydophila pneumoniae infection. Macrophages with decreased antioxidant enzyme activity resulting in elevated levels of oxidant stress have a significant increase in the levels and duration of Hsp60 protein expression .
The relationship between bacterial-derived antigens and atherosclerosis is exceedingly complex. Although early studies focused on a select group of pathogens, namely Chlamydophila pneumoniae and Porphyromonas gingivalis, analysis of human coronary artery atherosclerotic plaque excised from patients by directional atherectomy revealed a high bacterial diversity with DNA from more than 50 species present that was absent from postmortem patients without heart disease . Collectively, components of these bacterial pathogens may induce an immune inflammatory response through molecular mimicry, serve as ligands for toll-like receptors, and upregulate the ROS-responsive transcription factor NF-KB to promote a cascade of inflammatory events [684, 685].
Cell-free hemoglobin has also been proposed as a candidate antigen. Intraplaque hemorrhage with accumulation of erythrocyte cholesterol-containing membranes and release of hemoglobin is a frequent finding in advanced atherosclerosis . Elevated levels of oxidant stress within the plaque lead to oxidative modification of cell-free hemoglobin generating targets that are identified as autoantigenic epitopes. In patients with carotid atheromas, these novel epitopes were shown to increase the population of T cells that secrete interferon-γ as well as promote maturation of human monocyte-derived dendritic cells. As such, ROS-mediated modification of cell-free hemoglobin likely contributes to the local immune-inflammatory response within the atheroma .
Free radical species may also provide a mechanistic link between the innate and adaptive immune responses to atherosclerosis. These radicals influence T cell expression of CD40L implying that oxidant stress may importantly modulate B cell antibody formation to putative target epitopes and antigens . In fact, it has been demonstrated that activated T cells that express CD40L bind to CD40 on B cells to yield circulating antibodies to oxLDL [689, 690]. Increased oxidant stress also perpetuates the immune response by rendering T cells refractory to deactivation . These observations suggest that immunomodulation of atherosclerosis as a targeted therapy may also benefit from a concomitant reduction in oxidant stress.
Local and systemic alterations of ambient biomechanical forces inherent to the flow of blood through the vascular system also promote free radical formation and increase the predilection for atherosclerotic lesion formation. In vivo, blood vessels are subject to shear stress, a longitudinal frictional force at the blood-endothelial surface that varies between 15–75 dyne/cm2 over the cardiac cycle, as well as mechanical strain, a tangential force on the vessel wall related to blood pressure [692–694]. Under basal conditions, these forces are atheroprotective and serve to limit deleterious free radical formation within physiological parameters; however, once perturbed, these protective effects are lost, flow becomes bidirectional with a time-average near 0, and free radicals may engage in local oxidation reactions more efficiently [692–694]. The relationship between these hemodynamic forces, oxidant stress, and atherosclerosis is best recognized at arterial bifurcations where alternating turbulent and stagnant blood flow creates a complex flow profile and increases the susceptibility to atherothrombogenesis [695–697].
Experimental evidence demonstrates the atheroprotective effects of laminar or unidirectional shear stress on the vascular endothelium. Laminar shear stress increases eNOS activity by elevating intracellular Ca2+, increases eNOS phosphorylation at Ser1177, and upregulates eNOS expression to increase •NO production, as well as stimulates prostacyclin synthesis, all limiting endothelial oxidant stress [698–706]. Interestingly, only steady-state pulsatile shear stress may maintain redox homeostasis. Using human coronary arteries, an abrupt increase in shear stress was shown to stimulate mitochondrial H2O2 production, and carotid arteries exposed to abnormally elevated laminar flow exhibit NAD(P)H oxidase activation and persistent O2•− production [707, 708]. Pulsatile shear stress modulates the redox environment further by increasing the antioxidant capacity of the vascular endothelium. Studies performed in vitro and using ex vivo arterial specimens have shown that pulsatile shear stress increases the expression of Cu,Zn-SOD, Mn-SOD, GPx-1, thioredoxin-1, peroxiredoxins, HO-1, NADPH:quinine oxidoreductase 1, ferritin, microsomal expoxide hydrolase, glutathione-S-transferase, gamma-glutamylcysteine synthase, and intracellular GSH stores [709–716].
By contrast, oscillatory shear stress and its characteristic pattern of turbulent blood flow renders segments of the vasculature prone to atherosclerosis (Fig. 4). In vivo, low shear stress (< 0.5 Pa) predicts the development of atherosclerotic plaques characterized by abundant lipid accumulation, inflammatory cells, and a thin fibrous cap with evidence of vascular remodeling . Furthermore, in a clinical study, alterations in shear stress patterns in human coronary arteries were found to correlate with areas of high strain, implying that these forces may act in concert to modulate net vascular oxidant stress .
Oscillatory shear stress increases oxidant stress by activating O2•−-generating enzymes and shifting mechanical-transcriptional coupling to downregulate the expression of atheroprotective mediators . In vitro and in vivo, oscillatory shear stress increases vascular endothelial oxidant stress owing to increased activation and expression of the O2•−-generating enzymes NAD(P)H oxidase and xanthine oxidase with a concomitant and prolonged increase in O2•−, H2O2, and ONOO− levels . Turbulent flow also diminishes antioxidant capacity by downregulating antioxidant enzymes, including Mn- SOD, and decreasing GSH levels and eNOS expression [719, 720]. In this redox active environment, O2•−-sensitive transcription factors such as NF-kB are activated to induce the expression of the endothelial adhesion molecules VCAM-1, ICAM-1, and E-selectin, as well as chemokines and inflammatory cytokines that are important for leukocyte recruitment and extravasation .
Loss of atheroprotective hemodynamic forces, as occurs at arterial bifurcations, has further implications for the retention and oxidation of LDL. Immunostaining for 3-nitrotyrosine of explanted human coronary arteries from patients with ischemic cardiomyopathy at transplant revealed increased staining at arterial branch points characterized by oscillatory shear stress . In vitro studies confirmed that under conditions of oscillatory shear stress, increased endothelial cell O2•− and ONOO− production allowed for tyrosyl nitration of apoB-100 . Using 2-photon microscopy of porcine or murine coronary arteries, it was determined that the vascular architecture, in fact, is permissive for atherosclerotic plaque formation at arterial branch points. Here, it was shown that the absence of the luminal surface elastin layer at arterial branch points facilitates extensive binding of LDL at these sites, while the presence of elastin throughout the remainder of the vessel lumen retarded LDL retention . Taken together, these studies provide evidence for the link among flow disturbances, oxidant stress, and the inciting events of atherosclerosis.
In the vasculature, an increase in oxidant species, including reactive oxygen, nitrogen, or halogenating species, and free radical species such as thiyl, tyrosyl, or protein radicals defines the pathobiological state of oxidant stress. Oxidant stress can arise from pathophysiological stimulation of enzymatic sources, decreased antioxidant capacity, or both increased production and decreased antioxidant capacity. This redox imbalance in the vessel wall mediates a transition to a vascular phenotype that facilitates oxidative modification of lipids and proteins, infiltration of inflammatory and immune cells, platelet activation, and altered local hemodynamic forces. In this perturbed redox environment, oxLDL particles are engulfed by macrophages and transited to the vessel wall to serve as the nidus for atherosclerotic plaque formation. Dysregulation of antiatherogenic HDL and immune mediators by increased oxidant stress augment this response. Moreover, aberrant lipid deposition and disruption of atheroprotective laminar flow results in a pattern of alternating turbulent flow and stasis. Taken together, these findings demonstrate that free radical and redox reactive non-radical species play an integral, mechanistic role in all aspects of atherogenesis, and that adverse modification of the factors that govern the vascular redox state increase the oxidative risk of atherothrombotic cardiovascular disease.
The authors wish to thank Ms. Stephanie Tribuna for expert secretarial assistance. This work was supported by grants HL81110, HL70819, HL61795, HL81857, and HV28178.
Disclaimer: The project described was supported by Grant Numbers HL81110, HL70819, HL61795, HL81857, and HV28178 and its contents are solely the responsibility of the authors and do not necessarily represent the official views of the NIH.