Superoxide in the Vascular SystemWolin M.S. · Gupte S.A. · Oeckler R.A.
Department of Physiology, New York Medical College, Valhalla, N.Y., USA Corresponding Author
Oxidant production and regulation is becoming increasingly important in the study of vascular signaling mechanisms, and recent reviews have characterized some of the possible roles for known downstream products of superoxide formation. In this review, we will examine current research in the field, with a special emphasis on the role of the superoxide molecule itself and its place amongst the slightly better understood roles of peroxide and peroxynitrite. The regulatory roles of oxidant species are wide-ranging, and their involvement in processes ranging from intracellular and receptor signaling mechanisms that regulate endothelial mediator release and vascular contractile function to processes that control cellular growth and apoptosis has been implied. Cellular sources of superoxide production and metabolism and the chemical interaction of oxidant species with specific components of cellular signaling mechanisms are considered important factors which determine physiological responses that control vascular function.
Copyright © 2002 S. Karger AG, Basel
This review will focus on providing a concise overview of the field of vascular oxidant signaling, with an emphasis on examining novel aspects of how superoxide functions in some of the better understood mechanisms. While superoxide is the major product of most of the oxidases present in vascular tissue, its conversion to other oxidant species such as hydrogen peroxide (H2O2) and its reaction with nitric oxide (NO) [which generates peroxynitrite (ONOO–)] have important roles which have generally been emphasized in previous reviews [1, 2, 3] on vascular oxidant signaling mechanisms. As the rate of superoxide production increases, the interactions of superoxide itself with components of cellular regulatory processes appear to become very important processes in the oxidant signaling mechanisms that are expressed. Some of the key processes regulated by oxidant signaling are listed in table 1.
Table 1. Cellular signaling systems of vascular smooth muscle and/or endothelial cells that demonstrate evidence of oxidant regulation mechanisms
Properties of Superoxide
Superoxide is the molecule which results from a one-electron reduction of molecular oxygen by various oxidases and autooxidation processes that occur in biological systems [1, 3]. At physiological pH, superoxide is both a free radical (because it contains an unpaired electron) and a negatively charged species (as a result of it having a pKa of 4.8) . Its anionic and free radical properties, as well as its ability to participate in either electron-accepting (oxidation) or -donating (reduction) reactions, have a major influence on how superoxide interacts with vascular signaling systems. The anionic charge of superoxide results in it using anion channels as a primary mechanism for its transport across biological membranes , while its free radical character causes superoxide to react with other radicals with rate constants in the range of diffusion-controlled reactions. For example, under physiological conditions, superoxide has been reported to react with NO with a rate constant of 7 × 109 mol–1·l·s–1 . The extremely efficient reactions of superoxide with NO and the various forms of superoxide dismutase (SOD) are essential to the mode of interaction of superoxide with important physiological signaling systems.
Control of Superoxide Levels in Vascular Tissue: The Sources
The local level of superoxide is determined by the balance between its rate of formation by various oxidases and autooxidation processes and its rate of removal via SOD and reaction with various molecules. Vascular tissue contains multiple oxidases whose activity and expression appear to be highly regulated. Cell types found in the vessel wall generally have NAD(P)H oxidases which are active under basal physiological conditions, and these oxidases seem to serve varying roles in oxygen-sensing mechanisms that control both vascular function and adaptive gene expression responses [3, 6]. Some of these NAD(P)H oxidases appear to contain b558-type cytochromes and subunit structures which resemble the phagocytic cell NADPH oxidase system [1, 6, 7], while other proteins, including NO synthase (NOS), cytochrome P450, cyclooxygenase and xanthine dehydrogenase, have NAD(P)H oxidase activity under certain conditions. The mitochondrial electron transport chain is also thought to produce superoxide at two sites in the NADH dehydrogenase and coenzyme Q or Q-cycle region , and these mitochondrial sites may also be important participants in oxygen sensing-related vascular redox signaling [9, 10]. There are many stimuli-linked regulatory mechanisms which increase superoxide production, and thus it is important to consider how each oxidase present in the vessel wall potentially functions as a source of superoxide.
The predominant cell types of the vessel wall, including the endothelium [11, 12], vascular smooth muscle [13, 14] and fibroblasts , are thought to contain oxidases with subunits that resemble the well-studied phagocytic oxidase. The phagocytic cell oxidase appears to produce superoxide through a highly regulated transfer of electrons from NADPH to flavin sites, which then reduce a b558-type cytochrome. Subsequent reaction of the reduced form of this cytochrome with molecular oxygen is thought to generate superoxide . The phagocytic oxidase contains membrane-bound p22phox and gp91phox subunits whose superoxide-producing activity is thought to be controlled by cellular activation processes that promote the assembly and binding of cytosolic rac G protein, p40, p47 and p67 subunits to the membrane-bound subunits . While vascular tissue appears to contain most of these subunits, there is evidence for homologues of the gp91phox subunit, currently termed NOX-1 and NOX-4 . The vascular NAD(P)H oxidases seem to show both a receptor-stimulated activation mechanism similar to that of the phagocytic oxidase and regulation via modulation of NOX and p22phox subunit expression . In contrast to the phagocytic oxidase, which uses NADPH for a substrate, most studies of the vascular oxidase have detected significant levels of NADH oxidase activity. While vascular cells are likely to maintain high levels of NADPH through the activity of the pentose phosphate pathway of glucose metabolism, the cytosolic NAD(H) pool is thought to be maintained primarily in its oxidized, or NAD form, as a result of the way glycolysis and mitochondrial shuttles function [17, 18, 19]. Thus, the availability of NADH may be an additional factor in controlling vascular NAD(P)H oxidase activity, and hence increases in local lactate concentrations are potentially an important factor in increasing vascular NADH oxidase activity . Since the basal rate of superoxide production by NAD(P)H oxidases seems to be regulated by physiological levels of oxygen, there has recently been much interest and evidence that these oxidases may function as tissue oxygen sensors [19, 20]. The signaling mechanisms activated in response to various pathophysiological stimuli may involve oxidant overproduction with or without the overexpression of the enzymatic sources of reactive oxygen species (ROS), and this can occur either by neutrophilic invasion or alterations in cellular metabolic and regulatory pathways. Certain pathophysiological messenger responses of the vasculature, such as signaling through the angiotensin type 1 receptor, appear to function by stimulating or increasing the expression of NAD(P)H oxidase activity [1, 21, 22, 23].
NOS has the potential to be an important source of vascular superoxide in cell types in which the enzyme is highly expressed, such as endothelial cells [24, 25]. The concentration of L-arginine is a limiting factor for supporting NO production by NOS, and this has been thought to be a factor in increasing superoxide production by this system. However, recent evidence has emerged to suggest that a loss of the tetrahydrobiopterin cofactor for NOS through either its depletion or oxidation may be the key factor in increasing the oxidase activity of this enzyme at the expense of NO biosynthesis from L-arginine . In fact, in diabetes, endothelial NOS has an impaired ability to generate NO, and this has recently been demonstrated to be due to an altered expression of GTP-cyclohydrolase I, the first and rate-limiting enzyme in the de novo biosynthesis of tetrahydrobiopterin . Hence, the ability of NOS to generate radicals, coupled with the subsequent known involvement of superoxide, may be of importance in certain pathological states, including atherosclerosis, diabetes and chronic smoking, where deficiencies in tetrahydrobiopterin and L-arginine may lead to changes in vascular function [27, 28, 29].
While it has been known for some time that the oxygenated form of cytochrome P450 is capable of releasing superoxide, evidence that this is an important source of endothelium-derived superoxide has only recently been reported [30, 31]. An isozyme of P450 in porcine arteries has recently been identified as an endothelium-derived hyperpolarizing factor synthase, and it is now thought that this endothelium-derived hyperpolarizing factor synthase may have a physiological role in vascular homeostasis via production of both ROS and epoxyeicosatrienoic acids . The contribution of cytochrome P450 to cellular oxidant production is also accentuated by disease or autoinduced states such as those seen in alcohol- or drug-induced upregulation of these enzyme systems, and thus ROS production via P450 may be of greater importance in certain pathologic states, including alcoholism and cases of altered fatty acid oxidation .
The peroxidase component of the cyclooxygenase reaction is able to oxidize NAD(P)H and other substances to species such as NAD(P)·, which autooxidize (i.e. react with molecular oxygen) to generate superoxide . It has been demonstrated that the cyclooxygenase reaction can act as a significant source of vascular superoxide while simultaneously generating prostaglandins (PGs) as a result of endothelial cell activation by stimuli such as bradykinin [34, 35].
Most sources of vascular endothelium have xanthine dehydrogenase, which possesses low xanthine oxidase activity. The xanthine oxidase activity of xanthine dehydrogenase can be increased by thiol oxidation and limited proteolysis . The availability of the substrate hypoxanthine, derived from the degradation of ATP via adenosine, can be a limiting factor in the expression of xanthine oxidase activity. Exposure of endothelium to ischemia followed by reperfusion or hypoxia and reoxygenation appears to generally cause marked increases in endothelial xanthine oxidase activity [36, 37]. This activity results in a brief period of production of superoxide and peroxide, which may then react with iron to form hydroxyl radical-like species. These reactive species may in turn cause lipid peroxidation and the release of chemoattractants to facilitate cellular damage and an immune response in a dynamic positive feedback loop which would enhance existing vasculopathies .
Studies on isolated mitochondria have provided evidence that the electron transport chain produces superoxide at two sites in the NADH dehydrogenase and coenzyme Q or Q-cycle region . The actions of inhibitors at specific sites in the electron transport chain upon oxygen sensing-related vascular signaling have demonstrated that these mitochondrial sites are potentially important participants in the generation of oxidant species that mediate the redox signaling mechanisms involved [10, 38]. However, there are few actual data which demonstrate the significance of mitochondria as a source of oxidants in vascular tissue under normal physiologic conditions. Conversely, there is substantial evidence that apoptotic signaling activates a high rate of mitochondrial oxidant production [39, 40, 41, 42], and hence mitochondria are likely to be an important source of oxidants in vascular tissue when apoptotic mechanisms are activated.
Control of Superoxide Levels in Vascular Tissue: The Scavengers
Local mechanisms of superoxide consumption play a major role in controlling overall levels of superoxide. Since superoxide reacts with itself to form H2O2 and molecular oxygen at a rate of 8 × 104 mol–1·l·s–1 , substances that are significant scavengers of superoxide need to be present at concentrations which compete with the spontaneous dismutation reaction. Vascular tissue contains three forms of SOD which accelerate the dismutation of superoxide into H2O2 and molecular oxygen, with rate constants in the vicinity of 2.9 × 109 mol–1·l·s–1 . The first form is the cytosolic Cu,Zn-SOD, which is thought to lower superoxide levels from the nanomolar to picomolar concentration range. Mitochondria contain a second form, Mn-SOD, and lastly, arterial smooth muscle cells are the principal source of the extracellular Cu,Zn form of the SOD enzyme in the vascular wall . Since NO reacts with superoxide at a rate which is about 3 times faster than that of the SOD reaction (7 × 109 mol–1·l·s–1) , it can become an important scavenger of superoxide as NO levels approach local concentrations of SOD . Additionally, antioxidants such as ascorbate scavenge superoxide, although many of these substances may not be present at levels able to compete with the consumption of superoxide by SOD (and NO). Cellular uptake by cells such as erythrocytes in the blood is also a potentially significant mechanism for the removal of superoxide. Finally, the cell maintains the ability to control the expression of SOD isoforms, and alterations here may lead to changes in ROS levels and cellular scavenging ability. Recent evidence implicates the alteration of SOD levels due to either genetic mutation  or oxidant damage of metal ion binding sites  in the development of amyotrophic lateral sclerosis and other neurodegenerative disorders.
Interactions of Superoxide with Signaling Systems
The balance between the superoxide-producing oxidases and SOD activities appears to keep basal superoxide concentrations below the range in which this species could directly interact with vascular signaling. However, basal levels of H2O2, derived primarily from superoxide, appear to regulate some of the signaling systems thought to participate in oxygen and redox sensing [19, 38]. As NO levels increase into the range of the local concentrations of SOD, the scavenging of superoxide by reaction with NO has been hypothesized to be a source of activation of signaling mechanisms, these mechanisms being in addition able to direct NO effects, as a result of the formation of ONOO– [3, 5, 48]. Physiological or pathophysiological conditions that increase superoxide production or decrease tissue SOD levels activate additional mechanisms through superoxide or its derived oxidant species. Under these conditions, the mechanisms expressed are often dependent on the level of activation of signaling mechanisms controlled by other physiological processes and the function of cellular redox and antioxidant systems that control signaling processes . The following sections provide an overview of some of the better understood aspects of these processes, with an emphasis on the novel aspects in which superoxide may function.
H2O2 appears to be one of the most important oxidant species because it is known to interact with multiple signaling systems. The signaling mechanisms that appear to be the most sensitive to changes in peroxide are those interactions controlled by enzymes which metabolize peroxide . Its metabolism by the peroxidase reaction of cyclooxygenase results in the activation of PG production as a consequence of the oxidation of the ferric heme of this enzyme to a ferryl redox state that appears to be required for the catalytic process which converts arachidonic acid to PGG2 . PGG2 is a peroxide, and its metabolism to PGH2 by the peroxidase reaction of cyclooxygenase is thought to sustain PGG2 production. The dominant response of several microvascular preparations to peroxide appears to be either vasodilation or vasoconstriction, mediated through the production of PGH2 or PGs derived from the metabolism of PGH2 [50, 51]. Interestingly, there is minimal evidence consistent with a direct influence of endogenously generated superoxide on the production of PGs, whereas ONOO– seems to function as a peroxide which stimulates cyclooxygenase activity and PG production [52, 53, 54]. Thus, species derived from superoxide, including peroxide and ONOO–, seem to be more directly involved in the regulation of PG biosynthesis.
The metabolism of peroxide by catalase appears to be linked to a mechanism of stimulation of soluble guanylate cyclase (sGC) and cGMP-mediated relaxation . This vasodilator mechanism has been observed in several vascular preparations, and it seems to be closely linked to the activity of NADH oxidase in vascular smooth muscle  and endothelium-derived H2O2 . Superoxide and oxidation products of NO have been observed to be potent endogenous inhibitors of this H2O2-mediated relaxing mechanism through processes potentially involving the inhibition of catalase  and possibly direct interactions with sGC .
Peroxides may have additional interactions with signaling systems through direct reactions with protein functional groups, such as reactive thiols, or as a result of thiol redox changes via the metabolism of peroxide by glutathione peroxidase. Multiple signaling systems, listed in table 2, appear to be regulated through poorly understood mechanisms which may involve peroxide-elicited modifications of thiol redox. Although not yet completely clarified, it appears that both protein kinases and phosphatases may in some cases be influenced by oxidants. The reactive thiols at the active sites of tyrosine phosphatases or proteins containing bound zinc, or thiols which are readily converted to disulfides through enzymatic S-thiolation or oxidation activated by the presence of oxidized glutathione, are all potentially important factors which appear to be regulated by peroxide generation . For example, members of the MAP kinase pathways appear to be activated by superoxide-derived peroxide, but sometimes the results of two or more studies in this area appear at first glance to be in conflict. For instance, in Chinese hamster ovary cells , peroxide has been observed to activate jun N-terminal kinase (JNK) and p38, but not extracellular-regulated kinase (ERK) MAP kinases, yet only ERK seems to be activated in intact bovine coronary and pulmonary arteries . A possible explanation for these conflicting results, even within the same cell and species type, is the idea of cellular localization. Scaffolding proteins within the cell may be activated under certain conditions, causing compartmentalization of various systems . In the aforementioned example, this might entail a physical stimulus such as stretch, causing the individual units of the MAP kinase pathway , or other components of signaling pathways such as NADH oxidase, to interact with chaperones (e.g. heat shock proteins) and scaffolding proteins, aiding in the translocation and assembly of these proteins into a localized functional unit consisting of large immobilized complexes of signal-transducing molecules . A different stimulus or condition might set up a different scaffolding of signaling molecules, and thus a modularity of interchangeable signaling units may unfold. Hence, it is possible that peroxide may stimulate a pathway under one set of conditions, but not under another, when the necessary constituents are not properly localized to carry out their functions. Since the endpoint or output is dependent upon cellular conditions, this sets up the cell for a complex method of dynamically modulating signaling output based upon current cellular conditions and state.
Table 2. Specific targets for oxidant interactions within vascular smooth muscle and/or endothelial cells
Peroxide- and superoxide-derived ROS may account for mechanisms of signal transduction that are currently poorly understood, such as the mechanotransduction of a physical stimulus such as stretch or pressure into a biochemical signaling cascade and outcome. Current studies in vascular smooth muscle indicate that NADH oxidase-derived superoxide and peroxide are involved in alterations in contractility and responses to stretch on the vessel wall , and there is emerging evidence that ROS also mediate the myogenic response . Evidence for further involvement in transduction lies in the transition between receptor binding and second messenger activation, such as appears to occur at the serotonin receptor in vascular smooth muscle, where binding of serotonin to its receptor seems to activate ERK in a peroxide-dependent manner . A similar role has been noted in the case of norepinephrine-stimulated α1-adrenergic induction of cardiac myocyte hypertrophy . Some kinases may possibly be the source of ROS, as suggested by recent studies revealing superoxide generation by ras , a central signaling mediator, and by PI3 kinase in the presence of TNF-α . It seems more likely that these kinases are key regulators of downstream sources of superoxide rather than the actual sources themselves; the biochemical structures of the aforementioned kinases do not seem to have appreciable oxidase capacity, and any potential generation of radicals as a by-product would likely be at concentrations too low to be of consequence. The processes through which superoxide and superoxide-derived peroxide could potentially interact with these and other poorly understood mechanisms are described further in the sections on superoxide interaction with metal- and thiol-related redox-regulated systems and cellular proliferation and apoptosis.
Inhibition of NO signaling as a result of its inactivation by superoxide is one of the most important regulatory-like actions of superoxide. Based on observations made in many vascular preparations, it appears that a balance exists between NO and superoxide which controls the expression of important physiological and pathophysiological processes . Since NO is a dissolved hydrophobic gas, it readily diffuses throughout the cells in the vessel wall. Although there is a high level of compartmentalization of the sources of production and metabolism of superoxide, regions of high rates of superoxide production and/or low SOD activity become the means of NO removal from the area surrounding these sites. The reaction of NO with superoxide consumes NO and prevents its interaction with the signaling mechanisms normally regulated by NO. Thus, NO-elicited vasodilation through activation of sGC is inhibited by superoxide, and the reversible inhibition of mitochondrial respiration by NO is converted to an irreversible attenuation of respiration through inhibition of mitochondrial electron transport by ONOO– [3, 22, 69].
Elevated levels of NO are likely to cause the generation of ONOO– in amounts that potentially interact with signaling systems in the region where it is produced. Since NO often appears to interact with superoxide in various physiological and pathophysiological situations such as ischemia, reperfusion and inflammation, the interactions of ONOO– with signaling systems may be of substantial importance.
Acute exposure of vascular preparations to elevated levels of NO has been shown to activate several different signaling-like mechanisms. It can cause the trapping and regeneration of NO through ONOO–- and thiol-dependent mechanisms . The inactivation of mitochondrial electron transport and respiration by endogenously generated ONOO– appears to be readily observed when cardiac muscle is exposed to hypoxia and reoxygenation or ischemia and reperfusion, and endothelial sources of superoxide appear to be important contributors to the observed inhibition . While the inactivation of mitochondrial aconitase by superoxide is considered an indicator of the formation of superoxide [71, 72], the formation of ONOO– appears to markedly increase the potency of superoxide as an inhibitor of mitochondrial respiration . In addition to stimulating cyclooxygenase production of PGH2 in a fashion similar to peroxide, ONOO– also has an extremely potent inhibitory effect on PGI2 synthase. This enzyme has a tyrosine residue in its active site which is extremely sensitive to nitration by ONOO–. Once PGI2 synthase is nitrated, PGH2 can either be converted to other PGs, or the actions of PGH2 itself may be observed . For example, exposure of isolated coronary arteries to hypoxia-reoxygenation causes just this type of modification, suggesting that ONOO– generation has the potential to convert the production of endothelium-derived PGI2, an antithrombotic vasodilator, to PGH2, a prothrombotic vasoconstrictor . Thus, endogenously formed ONOO– can potentially activate multiple vascular signaling mechanisms which are of potential importance to the overall state of vascular function.
ONOO– can interact with and alter cellular signaling mechanisms by disrupting the balance maintained between cellular kinase and phosphatase activity. Phosphatase inhibition can potentially activate many pathways, which is generally accomplished by maintaining enzymes in their phosphorylated or active forms, and ONOO– has been demonstrated to irreversibly inhibit the protein tyrosine phosphatases PT1B, CD45 and LAR . Interestingly, superoxide may also reversibly inhibit PT1B via direct oxidation of its cysteine active site , and thus this site may prove to be involved in conversion from a physiological signaling mechanism to a pathological one at higher superoxide concentrations. In addition, there is evidence that NO may increase phosphatase activity, and thus superoxide scavenging of NO via ONOO– formation may in itself alter the kinase/phosphatase balance . Other influences of ROS on intracellular signaling cascades will be addressed in further detail according to cell type and pathway involvement.
The interactions of ONOO– with antioxidant enzymes can also have signaling-like effects, because there is evidence that ONOO– inhibits glutathione peroxidase , catalase  and mitochondrial or Mn-SOD . The cysteine residues in the zinc fingers of the regulatory domain make this an attractive target for redox regulation. Studies have shown that a positively charged zinc-thiolate structure is susceptible to oxidation by negatively charged oxidants such as ONOO– . It is also possible that ONOO– may affect redox control mechanisms, such as mitochondrial electron transport  and cytosolic NAD(H) , in a manner that promotes superoxide generation [82, 83]. In addition, Cu,Zn-SOD functions as a catalyst for tyrosine nitration in the presence of ONOO–, and this could also have an effect on signaling mechanisms . Thus, further elevation of ONOO– generation is likely to be associated with a deterioration of antioxidant defense mechanisms, and this is likely to promote activation of signaling mechanisms linked to apoptosis and cell death.
Superoxide has the potential to interact with metal and reactive thiol sites on proteins either directly or through the generation of ONOO–, and some of these interactions could be components of vascular signaling mechanisms. A brief overview of possible interactions appears in table 2. For example, ceruloplasmin and the enzymatic regulation of myeloperoxidase, which catalyzes the rate-limiting step in the oxidation of low-density lipoprotein (LDL) via reduction of the iron of heme, is thought to be controlled by the rate of superoxide production [85, 86]. Conversely, in cell-free systems, an iron-centered 15-lipooxygenase activity appears to be inhibited by superoxide ion . Additionally, superoxide could influence the activation of zinc finger proteins such as A20 and protein kinase C (PKC) , specifically suppressing the antiapoptotic gene via zinc protein A20  and possibly stimulating autonomous PKC activity viathiol oxidation and release of zinc from cysteine-rich regions of PKC . Interestingly, SOD significantly reduces the xanthine/xanthine oxidase-induced increase in autonomous PKC activity, while catalase has no effect, which suggests that superoxide, but not peroxide, is the molecule responsible for modification of the thiol-containing cysteine-rich groups of PKC . This is currently an area of active investigation, merging protein and structural chemistry with physiological understanding, and the list of mediators we have given is by no means exhaustive. It now appears that there may be many other regulatory proteins whose metal- or thiol-containing structural regions could lend themselves to redox modulation, some, such as PKC, affecting a broad range of cellular signaling mechanisms. Since many of these superoxide- and ONOO–-mediated interactions with signaling systems appear to involve direct chemical reactions with functional groups, there is likely to be selective activation of signaling mechanisms dependent on these processes in regions where these oxidases are localized.
Role of Superoxide and the Control of Vascular Regulatory Systems
The activation of oxidases in endothelium by physiological and pathophysiological processes is likely to be involved in the regulation of multiple signaling mechanisms that affect endothelial function. In addition, oxidase activation could result in the extracellular secretion by endothelium of oxidant species at concentrations which influence the function of vascular smooth muscle or other cell types present in blood or vascular tissue. It appears that endothelial cells contain a phagocytic cell-like NAD(P)H oxidase which generates superoxide under basal conditions [12, 91], and although these cells appear to contain all of the components of the phagocytic oxidase, the importance and role of gp91phox versus NOX-type subunits in the NAD(P)H oxidase(s) present are not yet resolved . Stimuli such as prolonged exposure to nitroglycerin , oxidized LDL  and angiotensin  appear to increase the expression and/or activity of endothelial NAD(P)H oxidase. Depletion or oxidation of the tetrahydrobiopterin cofactor for NOS appears to readily result in an uncoupling of NO production and an increase in NADPH oxidase activity [26, 27, 28, 29], to the extent that endothelium-dependent vasodilator responses normally mediated by NO have been observed to be converted to responses caused by NOS-derived peroxide . It has recently been observed that cytochrome P450 can be a source of endothelium-derived H2O2, as an isozyme homologous to cyp2C9 has been shown to produce physiologically relevant levels of ROS . Cyclooxygenase has been observed to generate vasodilator levels of superoxide-derived oxidant species in the cerebral microcirculation upon exposure to bradykinin or brain injury . Exposure of endothelium to hypoxia and reoxygenation increases endothelial superoxide generation by xanthine oxidase [36, 37, 93]. However, there is only limited information available on the impact of activation of xanthine oxidase on vascular function, with the exception of lung injury. The pulmonary microcirculation appears to bind and accumulate circulating forms of xanthine oxidase which are released by injury to various systemic organs, and this process is thought to contribute to the evolution of pulmonary endothelial dysfunction . Thus, a diversity of physiological and pathophysiological stimuli appear to be able to selectively activate the various oxidases present in endothelium, and the activation of individual oxidases may be linked to the control of specific vascular responses.
Increases in superoxide generation appear to have a profound effect on the type of mediators released from the endothelium. Attenuation of the release and actions of NO is one of the most readily observed effects of an increase in endothelial superoxide production. The major mechanism through which superoxide appears to influence endothelial PG generation is via the stimulation of cyclooxygenase due to the formation of H2O2 and perhaps ONOO–. Although the stimulation of vasoactive PG generation is readily observed when vascular preparations are exposed to peroxide, only a few studies have considered the role of peroxide (or ONOO–) in the control of PG production during vascular responses to physiological stimuli. Studies on the mechanism of the reactive hyperemia elicited by the release of a 15-second occlusion of arterioles in the rat skeletal muscle microcirculation have provided evidence consistent with a role for the stimulation of PGs by endogenously formed H2O2 in the transient vasodilator response that is observed . Other processes, such as the hypothesized role of NO in inhibition of the generation of cytochrome P450-derived vasoactive mediators  and the impact of the inactivation of PGI2 synthase by ONOO– , could also be important in contributing to alterations in endothelial function, as oxidase activation results in increases in superoxide production. Thus, oxidase activation with a subsequent increase in superoxide production has a major impact on the vasoactive mediators that are released from endothelium, a summary of which appears in figure 1.
Fig. 1. Superoxide interactions modulate vasoactive mediator release from endothelium, which affects neighboring smooth muscle cells. AA = Arachidonic acid; EET = epoxyeicosatrienoic acids; L-Arg = L-arginine; p450 = cytochrome P450-containing enzymes.
Oxidase activation also appears to have important roles in modulating signaling mechanisms that regulate other aspects of endothelial function, including the control of permeability, adhesion protein expression, growth and apoptosis. Superoxide  and peroxide have been implicated in the alteration of endothelial cell contractility leading to changes in vascular permeability in both the microvasculature  and in the lung during the acute respiratory distress sequelae of increased permeability and leukocytic migration . The actual targets have yet to be elucidated, although alterations leading to the actin and myosin interaction have been proposed. A similar oxidant role has been demonstrated for the expression of cellular adhesion proteins such as intercellular adhesion molecule-1 [98, 99, 100], yet as with the permeability story, the targets of ROS interaction remain elusive. Various mechanisms regulating Ca2+ homeostasis in the endothelial cell are activated by superoxide in a concentration-dependent manner. It has been observed that at low concentrations, superoxide has little effect on basal intracellular Ca2+ levels, but elevates the agonist-stimulated Ca2+ influx. With increasing levels of superoxide, a rapid, transient increase in intracellular Ca2+ is seen due to the release of intracellular Ca2+ stores, which is then followed by a sustained increase in intracellular Ca2+ via extracellular Ca2+ influx . Evidence indicates that superoxide elevates the generation of 1,4,5-triphosphate, which in turn depletes intracellular Ca2+ from the sarcoplasmic reticulum (SR) and enhances capacitative Ca2+ entry from the extracellular space . The mechanism by which superoxide elevates 1,4,5-triphosphate levels in endothelium, as well as the action of superoxide on the uptake of Ca2+ by the endoplasmic recticulum, is currently unknown. There are also observations that peroxide can attenuate agonist-stimulated Ca2+ responses . Selective interactions of oxidant species with systems that influence tyrosine phosphorylation, PKC, MAP kinases and nuclear factors are likely to participate in endothelial cell responses through mechanisms that are currently poorly understood, such as permeability, adhesion protein expression, growth and apoptosis.
Studies in vascular smooth muscle have provided evidence that NAD(P)H oxidases [19, 103] and mitochondrial systems [9, 10] have important roles in the generation of superoxide, and that oxidant species control several signaling mechanisms, including sGC, PG production, potassium and calcium ion transport and tyrosine kinase-regulated systems (protein kinase B, PKC and MAP kinases) [3, 7]. These oxidant-regulated signaling mechanisms appear to participate in the control of the generation of vascular force and adaptive growth and apoptotic signaling systems that are involved in remodeling. Since lactate appears to relax vascular smooth muscle through a mechanism that seems to involve the stimulation of sGC by H2O2, increases in cytosolic NADH could be potentially important in controlling basal levels of oxidant generation. Furthermore, the NADH oxidase involved seems to function as a PO2 sensor . Growth factors such as angiotensin II have been demonstrated to increase vascular smooth muscle NAD(P)H oxidase activity and expression of the p22phox and NOX-1 subunits of this oxidase associated with the activation of cell growth signaling [7, 14]. Thus, oxidant signaling systems appear to have fundamental roles in oxygen sensing and the control of multiple aspects of force generation and vascular remodeling.
The production of cGMP by sGC is one of the most important processes through which superoxide and its derived species control vascular function. Based on studies in bovine pulmonary and coronary arteries, it appears that superoxide originating from NADH oxidase can stimulate sGC during posthypoxic reoxygenation as a result of H2O2 formation and its metabolism by catalase . However, superoxide generated by this oxidase inhibits cGMP-mediated relaxation caused by NO or H2O2 when Cu,Zn-SOD activity is inhibited [18, 104]. Exposure of human placental vessels to hypoxia and reoxygenation appears to cause an endothelium-independent PG-mediated contractile response that seems to be mediated through increased peroxide generation during reoxygenation . Interestingly, this mechanism appears to be enhanced in placental vessels from patients with gestational diabetes as a result of the loss of what appears to be an endogenous peroxide-mediated vasodilator mechanism which is simultaneously activated during reoxygenation . The properties of this mechanism suggest that it originates from NADH oxidase-derived H2O2, producing cGMP-mediated relaxation. This response seems to be attenuated in vessels from patients with gestational diabetes by increased NO levels which inhibit catalase metabolism of peroxide and thus the linked impairment of the stimulation of sGC. Peroxide also appears to activate contractile responses in vascular smooth muscle that seem to originate from the activation of mechanisms potentially involving processes such as the activation of tyrosine kinases, PKC and p42/p44 MAP kinases (ERK 1/2) [59, 107]. Thus, superoxide-derived oxidant species appear to affect the function of multiple signaling mechanisms which are involved in the control of vascular force.
In smooth muscle cells, both superoxide and peroxide disrupt the SR Ca2+-ATPase and inactivate plasma membrane Ca2+ pumps, leading to both short- and long-term effects on smooth muscle Ca2+ handling. Recent studies suggest that there are inositol 1,4,5-triphosphate-sensitive and -insensitive regions in the SR and that superoxide selectively inhibits SR Ca2+-ATPase clustered in the inositol 1,4,5-triphosphate-sensitive pool. The mechanism of SR Ca2+-ATPase inhibition by superoxide is possibly through the irreversible oxidation of sulfhydryl groups or by direct attack on the ATP-binding site, because reducing agents and sulfhydryl-binding agents prevent inhibition of SR Ca2+-ATPase activity, and ATP promotes recovery of activity in isolated SR vesicles . Additionally, studies have suggested that the effect on vascular tone of superoxide anion originating in the adventitia is mediated by inactivation of endothelium-derived NO, which promotes smooth muscle calcium influx via L-type Ca2+ channels and spontaneous tone . Redox reagents potentially react with thiol molecules on the α1-subunit of human recombinant L-type Ca2+ channel protein and modulate the voltage gating and influx of Ca2+ currents [109, 110]. Moreover, since superoxide activates PKC, it is possible that activated PKC may regulate the L-type Ca2+ current. It has been shown that PKC activates L-type Ca2+ channels by phosphorylating a Ca2+ channel protein of vascular smooth muscle cells . However, this area has not been well studied. Besides, superoxide decreases voltage-dependent K+ current density by reducing a methionine residue on the K+ channel protein of coronary arterial myocytes, leading to impairment of vascular function . In contrast, superoxide also inhibits ATP-dependent K+ currents in cardiac myocytes . Therefore, it is clear that these channels contain thiol groups which may subject them to redox regulation, but the actual interactions remain to be elucidated.
Superoxide and peroxide can influence the growth, as well as death, of both vascular and nonvascular cells through the mechanisms shown in figure 2. Cellular release of superoxide and peroxide, either constitutively in the case of tumor cells or following cytokine stimulation, may serve as an intercellular messenger to stimulate proliferation via mechanisms common to natural growth factors . Superoxide seems to function as a signal transduction messenger, mediating the downstream effects of Ras and Rac in nonphagocytic cells. As such, superoxide contributes to the unchecked proliferation of Ras-transformed cells . Superoxide and peroxide are implicated in the activation of protein tyrosine kinases which subsequently stimulate downstream signaling pathways, such as MAP kinase and phospholipase C=γ, and ultimately reach the nucleus, where they result in altered gene expression mediated by the activation of several transcription factors . One such example is peroxide activation of apoptosis-stimulated kinase 1, which is required for sustained activation of JNK/p38 MAP kinases and apoptosis . Recent evidence also indicates a role for these oxidants in the control of vascular smooth muscle proliferation both in vitro and in vivo, and oxidative stress has been shown to mediate hormone-induced hypertrophy and under some circumstances to induce apoptosis .
Fig. 2. Targets for interactions between ROS and signaling systems that potentially influence the expression of vascular function and cellular growth and death processes in vascular smooth muscle and endothelial cells. MEK and MAPKK are MAP kinase kinases. Ras, Raf and Rac are G proteins. PKB = Protein kinase B; ASK = apoptosis-stimulated kinase.
Both physiological cell death (apoptosis) and, in some cases, accidental cell death (necrosis) involve a two-step process, each of which may be altered by ROS. At the first step, various physiological and pathological stimuli trigger an increase in mitochondrial membrane permeability, which releases apoptogenic factors through the outer membrane and dissipates the electrochemical gradient of the inner membrane. An increase in the level of superoxide induces the formation of mitochrondrial pores via interaction with membrane protein thiols, producing cross-linkage reactions that may open membrane pores upon the binding of Ca2+. Peroxide causes oxidation of mitochondrial pyridine nucleotides and thereby stimulates a specific Ca2+ release from intact mitochondria. Due to the binding of a unique cyclophilin (cyclophilin-D), the adenine nucleotide translocase is converted into a nonspecific translocase through a calcium-mediated conformational change. The binding of cyclophilin-D is increased in response to oxidative stress and some thiol reagents which sensitize the mitochondrial transition pores to intracellular calcium [119, 120, 121]. At the second step, the consequences of mitochondrial dysfunction due to loss of mitochondrial membrane potential, uncoupling of the respiratory chain, hyperproduction of superoxide, disruption of mitochondrial biogenesis, outflow of matrix calcium and glutathione and release of soluble intermembrane proteins result in the translocation of apoptosis-inducing factor to the nucleus, activation of nucleases that hydrolyze nuclear DNA and the activation of procaspase 9 and other procaspases leaked into the cytosol from the disrupted mitochondrial membrane and mitochondrial pores. Increased production of superoxide by mitochondrial dysfunction may lead to further inhibition of aconitase and destabilization of the mitochondrial membrane potential and entail a bioenergetic catastrophe giving rise to the disruption of plasma membrane integrity and necrosis [122, 123, 124, 125]. In addition, the activity of caspases, which are cysteine proteases, may be regulated by oxidative stress in a manner that controls the expression of apoptosis and necrosis. Besides attacking mitochondria, superoxide and ONOO– directly damage genomic DNA, which in turn activates poly (ADP-ribose) synthetase (PARS). In some cell types, PARS has been implicated in the process of apoptotic cell death. However, high levels of PARS deplete energy metabolites, and this is likely to promote necrotic cell death. While it is well established that superoxide-derived oxidant-activated mechanisms have a marked influence in controlling the expression of cellular growth and cell death by necrosis and apoptosis, the fundamental processes involved remain very poorly understood.
Oxidant species also appear to have important roles in modulating signaling mechanisms that regulate vascular smooth muscle growth and apoptosis. Growth factors which increase NAD(P)H oxidase activity associated with increased expression of the p22phox and NOX-1 subunits appear to activate protein kinase B through oxidant mechanisms . Studies have indicated that apoptosis contributes to postpartum arterial remodeling in the neonatal lamb . Furthermore, it has been reported that apoptosis of vascular smooth muscle cells is regulated by p53-dependent and -independent pathways. Apoptotic death induced by the transcription factors c-myc and E1A is mediated by and dependent upon p53, whereas the mechanisms involved in the free radical-induced apoptosis of vascular smooth muscle cells are still not clearly understood. It is quite possible that overproduction of superoxide radical may elicit one of the many pathways resulting in the triggering of apoptotic cell death.
Roles for Superoxide in Vascular Disease Processes
Oxidant signaling mechanisms appear to have important roles in the expression of many aspects of multiple vascular diseases. Physiological responses of the vasculature such as shear and pressure often begin the process of increased NO and superoxide production which is associated with the activation of adaptive physiological responses . Alterations related to vascular disease processes function to perturb the changes in the production and interactions of NO and superoxide in a manner that activates additional signaling responses associated with the progressive complex pathophysiology of each vascular disease. In hypertension, the prolonged exposure to increased pressure and circulating hormones like angiotensin II appear to have important roles in the development of a loss of endothelium-derived NO-mediated regulation and the occurrence of a prooxidant state that promotes processes such as growth or hypertrophy of vascular smooth muscle and increased expression of adhesion proteins which promote inflammatory responses and the development or progression of atherosclerosis [68, 127]. These changes have been observed to be associated with increases in the expression and activation of NAD(P)H oxidases [128, 129], elevated levels of superoxide generation in both inflammatory cells and vascular cells normally present in the vessel wall [130, 131] and increases in the expression of extracellular SOD which help preserve NO-mediated responses . This may explain the success of certain therapies for hypertension and coronary artery disease; in addition to their roles in increasing NO biosynthesis, statins [129, 133] and estradiol  have been shown to inhibit NADPH oxidase expression. Thus, the clinical success of such therapies, especially of the statin drugs, may be attributed as much to prevention or attenuation of the production of oxidant species as to any involvement in other aspects of NO-dependent mechanisms.
Elevated glucose and advanced glycosylation products seen in diabetes and elevated LDL associated with hypercholesteremia function to enhance oxidant production [87, 89, 129], and this further accelerates the progression of the vascular disease process [6, 25, 56]. For example, as other oxidases (e.g. NOS and xanthine oxidase) in the endothelium become activated and processes such as the promotion of oxidant-associated stimulation of cell growth signaling pathways (e.g. protein kinase B and ERK 1/2 ) by angiotensin II occur, these systems could function to promote proliferation and intimal thickening of the microvasculature associated with the progression of hypertension, as well as to potentiate immune involvement in the development of atherosclerotic sites in larger vessels such as coronary arteries. Under these conditions, the signaling mechanisms controlling vascular function generally shift to an increased sensitivity to vasoconstrictors, which enhances the potential for vasospasm. Vasospasm could also enhance the activation of oxidant mechanisms as a result of it causing repeated periods of tissue ischemia and reperfusion. Eventually, as the disease process progresses, localized sites of extreme dysfunction of homeostatic mechanisms and cell death are likely to occur, and this would be associated with processes such as thrombosis , plaque rupture  and arterial occlusion .
What role ROS may play in the phenomena of preconditioning, which is the ability of short periods of ischemia to lead to protection of the vascular endothelium and smooth muscle from the oxidant-mediated damage of a prolonged ischemia-reperfusion episode, is currently unclear but under investigation. Recent work indicates that preconditioning can prevent intercellular adhesion molecule-1 expression via a PKC- and NO-mediated pathway, resulting in a lesser adhesion of neutrophils to endothelial cells, and that this may contribute to its protective effect against reperfusion-induced endothelial injury . Additionally, the impairment of endothelium-dependent coronary vasodilation has been shown to be paralleled by an SOD-preventable disruption of the endothelial glycocalyx, suggesting a role for superoxide in the mechanism of endothelial dysfunction in postischemic guinea pig hearts . Thus, there seems to be both a quantitative and temporal aspect to ROS involvement, in that short episodic periods of ischemia appear to be protective, while larger, extended bursts associated with a prolonged ischemia-reperfusion event appear to cause ROS-mediated cellular damage. While oxidant signaling is only one component of the very complex set of signaling mechanisms that are associated with vascular disease progression, the extent of activation of oxidant production in the various cellular compartments present in the vessel wall has major roles in determining the progression of the pathophysiology that is observed at the cellular level.
Once thought to be destructive or by-products of cellular metabolism, it is now becoming clear that ROS play a vital role in both inter- and intracellular signaling pathways. Enzymatic and reactive formation of ROS now appear to play vital roles in maintaining cellular homeostasis, and the chemistry of these intermediates allows for ultrafast modulation and communication, giving these molecules the unique ability to respond rapidly via interaction, and then with a sustained response via changes in the signaling intermediates and in the longer-term alteration of gene transcription. With further elucidation, it may soon be clear which targets and delivery methods would be most effective in pathological states in which disruption of these processes occurs. One thing is becoming very clear: an understanding of superoxide and its metabolites is quintessential to a comprehensive knowledge of cellular function, from vascular reactivity and proliferation to apoptosis and cell death.
Recent studies from the authors’ laboratories discussed in this review have been funded by United States Public Health Service grants HL31069, HL43023 and HL66331.
Dr. Michael S. Wolin
Department of Physiology
New York Medical College
Valhalla, NY 10595 (USA)
Tel. +1 914 594 4093, Fax +1 914 594 4826, E-Mail firstname.lastname@example.org
Received: October 22, 2001
Accepted after revision: December 17, 2001
Published online: August 22, 2008
Number of Print Pages : 17
Number of Figures : 2, Number of Tables : 2, Number of References : 139
Journal of Vascular Research
Founded 1964 as Angiologica by M. Comèl and L. Laszt (1964–1973) continued as Blood Vessels by J.A. Bevan (1974–1991)
Vol. 39, No. 3, Year 2002 (Cover Date: May-June 2002)
Journal Editor: M.J. Mulvany, Aarhus
ISSN: 1018–1172 (Print), eISSN: 1423–0135 (Online)
For additional information: http://www.karger.com/journals/jvr