I. Introduction

Top Previous Next References

It was believed initially that the toxicity ascribed to the superoxide radical was caused by superoxide's direct interaction with biological targets. It is now clear that many tissue effects of O₂ result from the secondary formation of other oxygen radicals in addition to direct reactions of superoxide (or its conjugate acid) with biological targets, such as lipids (Aikens and Dix, 1991; Dix and Aikens, 1993), catecholamines (Misra and Fridovich, 1972; Heikkila and Cohen, 1973; Rao and Hayon, 1975; Macarthur et al., 2000), and DNA (Dix et al., 1996). Superoxide in aqueous media undergoes a spontaneous second-order reaction with itself, a dismutation reaction that yields one molecule each of H₂O₂ and oxygen (see reaction 1, Table 1) in a relatively slow reaction at pH 7.4 (the second-order rate constant is of the order of 10^4.5) when compared with the rate at which superoxide or HO₂ can abstract an H-atom from such key biological targets as catecholamines or the allylic CH in lipid where the second-order rate constant exceeds 10⁷.

Although the dismutation would be spontaneous at physiological pH at high superoxide concentrations, the concentration of superoxide approaches 10 µM (physiological) as the self-reaction slows down considerably and its lifetime becomes extended to many seconds. Consequently, nature has evolved a class of superoxide dismutase (SOD²) enzymes to remove this deleterious free radical byproduct of oxygen metabolism. These enzymes can react rapidly with superoxide (rates approaching or exceeding 10⁹) and dismutate the radical to the nonradical products, O₂ and H₂O₂, faster than superoxide can react with other potential biological targets. The short half-life should not be misinterpreted as mitigating the potential reactivity of O₂ because the half-life is actually quite long in relation to the phenomenal diffusion coefficient of the radical. Given that superoxide can interact with a variety of biological target molecules, the reaction with the enzyme literally can shunt the superoxide production into H₂O₂ and oxygen. Thus, it is conceivable that, in vivo, the presence of the highly active SOD enzymes will lead to an increase in the local concentration of H₂O₂.

The most reactive oxy radical is HO₂. It was proposed many years ago that it could be produced from the interaction of O₂ and H₂O₂ by a chemical process known as the Haber-Weiss reaction (reaction 2, Table 1). However, detailed studies of the rate of this reaction have shown that it could not take place under physiological conditions. An alternative explanation, which is now widely accepted, is that trace amounts of metal ions, primarily ferrous ion, react with H₂O₂ in what is known as the iron-catalyzed Fenton reaction to produce the hydroxyl radical. Normally, ferrous ion is not present in vivo, but it is produced by the action of superoxide on ferric ion (reactions 3 and 4, Table 1) present in iron storage proteins; thus, liberating soluble ferrous ion. There is considerable debate as to whether protein-bound metal ions (e.g., lactoferrin, hemoglobin, etc.) catalyze this reaction to any great degree. Sensitive measurements of the free (unbound) iron concentration in tissues, such as synovial fluid, have been reported to show a sufficient concentration to catalyze reactions (3) and (4) in an inflamed joint. Finally, recent attention has been drawn to what is called "site-specific" hydroxyl radical formation (Czapski et al., 1983), wherein an iron ion bound to a macromolecule catalyzes HO^· generation at the actual site on the substrate where cleavage eventually ensues.

In summary, tissue toxicity from extracellular superoxide generation seems to be based on its direct reactivity with numerous types of biological molecules (lipid, DNA, RNA, catecholamines, steroids, etc.) and from its dismutation to form H₂O₂ and the concomitant reduction of ferric ion to ferrous ion; reaction of these two products yields the highly toxic hydroxyl radical that may cleave covalent bonds in proteins and carbohydrates, cause lipid peroxidation, and destroy cell membranes. There are three strategies available to "detoxify" or prevent formation of locally produced oxygen radicals: 1) deliver SOD or an SODm to the area; 2) deliver catalase or a related peroxide scavenger, or 3) chelate (and thereby inactivate) the trace iron that catalyzes the reaction.

NO^· is synthesized from the guanidino group of L-arginine by a family of enzymes termed NO^· synthases (NOSs). Three isoforms have been described and cloned: endothelial cell NOS (ecNOS or type 3), brain NOS (bNOS, nNOS, or type 1), and inducible macrophage-type NOS (iNOS or type 2). All of the NOS isoforms can be inhibited to varying degrees with N-substituted L-arginine analogs (e.g., N-methyl-L-arginine). The formation of NO^· is linked to the incorporation of O₂ into the molecule. All NOS isoforms are dependent on NADPH and calmodulin. In iNOS, calmodulin is present in a tightly bound form; thus iNOS produces NO^· in a sustained manner in the presence of adequate substrate (Geller and Billiar, 1998; Marletta, 1993; Stuehr, 1997). Many of the biological actions of NO^· are mediated through the guanylyl cyclase/cyclic GMP (cGMP) system. NO^·, a lipophilic small molecule, diffuses to adjacent cells and readily enters the cytosol, where it activates soluble guanylyl cyclase by binding to the iron on its heme component, thereby moving the iron out of the plane of the porphyrin ring. Increased levels of cGMP trigger a reduction of calcium concentration by enhancing extrusion of calcium and its sequestration into intracellular stores. The decrease in intracellular calcium concentration is responsible for the NO^·-mediated relaxation of vascular and nonvascular smooth muscle, inhibition of platelet adherence and aggregation, inhibition of neutrophil chemotaxis, and signal transduction in the central and peripheral nervous systems (Ignarro, 1991; Moncada et al., 1991; Dusting, 1995). It is now well established that NO^· also has several cGMP-independent actions. The cytotoxic effects of NO^· (in high local concentrations) involves the inhibition of key mitochondrial iron-sulfur enzymes, including NADH:ubiquinone oxidoreductase, NADH:succinate oxidoreductase, and aconitase (Nathan, 1992). cGMP-independent activation by NO of other enzymes, such as cyclooxygenase, has also been described. This action may be related to the reaction of NO^· with the iron-heme center at the active site of the enzyme (Salvemini and Masferrer, 1996). NO^· inhibits the activity of cytochrome P-450 enzymes (Khatsenko et al., 1993). NO^· may modulate gene transcription and translation: in endothelial cells, it activates c-fos (Felley-Bosco et al., 1994), whereas in neurons it potentiates the effect of calcium on gene expression linked to the c-fos promoter (Peunova and Enikolopov, 1993). Many inflammatory conditions are associated with production of comparatively large amounts of NO^·, produced by iNOS, with consequent cytotoxic effects. iNOS, first identified in macrophages, can be expressed in essentially any cell type. Although constitutive expression of iNOS has been localized to the kidney, the intestine, and the bronchial epithelia, iNOS is expressed typically in response to immunological stimuli and produces nanomoles, rather than picomoles, of NO^·. Once produced in high local concentrations, NO^· may act as cytostatic and cytotoxic molecules for fungal, bacterial, helminthic, and protozoal organisms, as well as tumor cells. Bacterial lipopolysaccharide and a variety of proinflammatory cytokines also induce the expression of iNOS in a number of nonhematopoietic cells, including fibroblasts, glial cells, and cardiac myocytes, as well as vascular and nonvascular smooth muscle cells (Nathan, 1992). iNOS produces large amounts of NO^· for prolonged periods. The expression of iNOS is regulated both at the level of transcription and at the level of iNOS mRNA stability. The mechanism of iNOS induction involves de novo transcription and the biosynthesis of new protein. Induction of iNOS can be inhibited by numerous agents, including glucocorticoids, thrombin, macrophage deactivation factor, tumor growth factor-, platelet-derived growth factor, interleukin (IL)-4, IL-8, IL-10, and IL-13. Induction of iNOS may have either toxic or protective effects. Factors that seem to dictate the consequences of iNOS expression include the type of insult, the tissue type, the level and duration of iNOS expression, and probably the redox status of the tissue. Much attention has focused on the toxicity of iNOS. For example, induction of iNOS in endothelial cells produces endothelial injury (Palmer et al., 1992). Induction of iNOS has been shown to inhibit cellular respiration in macrophages and vascular smooth muscle cells; these processes can lead to cell dysfunction and cell death. Such processes, when occurring within vascular smooth muscle cells, play a key role in the pathogenesis of the vascular hyporeactivity and progressive vascular decompensation associated with various forms of circulatory shock (Szabó, 1995). In clear contrast, expression of iNOS in liver cells suppressed endotoxin and tumor growth factor--induced toxicity (Kim et al., 1997; Ou et al., 1997). Overexpression of iNOS by gene transfer also limits lipopolysaccharide (LPS)-induced toxicity in endothelial cells (Tzeng et al., 1997).

Simultaneous generation of NO^· and O₂ favors the production of a toxic reaction product, peroxynitrite anion (ONOO) (Beckman et al., 1990), and this product may account for some of the deleterious effects associated with NO^· production. This peroxynitrite-forming reaction has since been shown to be diffusion controlled (k_obs, = 6.7 × 10⁹ M¹ s¹), indicating that competition of NO^· with SOD for superoxide is feasible (Huie and Padmaja, 1993).

Beckman noted that peroxynitrite production increases as the square of the fluxes of these precursors. Moreover, certain forms of SOD enzymes are inactivated by reaction with peroxynitrite, and this can create positive feedback for ONOO formation (Ischiropoulos et al., 1992a; Beckman et al., 1994a). Hence, it is reasonable to conclude that peroxynitrite overproduction may occur readily in vivo. Once near or inside a cell, ONOO can damage or deplete a number of vital components [e.g., DNA by strand scission (King et al., 1992; Groves and Marla, 1995; Groves et al., 1996), lipids by peroxidation (Radi et al., 1991a; Rubbo et al., 1994), aconitase (Castro et al., 1994; Hausladen and Fridovich, 1994), and antioxidant availability (Van der Vliet et al., 1994; Vasquez-Vivar et al., 1996)].

A considerable portion of the toxic effects previously attributed to NO^· or O₂ alone may in fact be modulated by peroxynitrite (Table 2). The resulting oxidative stress may cause cell death and tissue damage that characterize a number of human disease states, among them neurological disorders and stroke, inflammatory bowel disease, arthritis, toxic shock, and acute reperfusion injuries. In fact, recent studies suggest that peroxynitrite, and not NO^·, may be the ultimate cytotoxic species in many conditions (Castro et al., 1994; Hausladen and Fridovich, 1994; Szabó et al., 1996a).

In cells exposed to exogenous peroxynitrite or to compounds that simultaneously generate NO^· and superoxide, marked changes in the level of cellular energetics and DNA integrity occur. For instance, in pulmonary type II cells, inhibition by peroxynitrite of sodium uptake has been reported (Hu et al., 1994). Mitochondrial respiration is profoundly inhibited by peroxynitrite in neurons and glial cells (Bolanos et al., 1995), cultured monocytic macrophages (Szabó and Salzman, 1995; Szabó et al., 1996b), and cultured rat aortic smooth muscle cells (Szabó et al., 1996b). Although a decrease in the activity of succinate-cytochrome c reductase and cytochrome c oxidase was found in neurons exposed to peroxynitrite, only cytochrome c oxidase was affected in isolated mitochondria exposed to peroxynitrite. These findings suggest the contribution of secondary cellular pathways to the toxicity of peroxynitrite (Bolanos et al., 1995). Inactivation of mitochondrial enzymes increases the amounts of H₂O₂ generated by the mitochondria (Radi et al., 1994), which may further contribute to cellular injury, in an additive or synergistic fashion.

Similarly, in macrophages (Szabó and Salzman, 1995), smooth muscle cells (Szabó et al., 1996b), and neurons (Heales et al., 1994), immunostimulation leads to the inhibition of mitochondrial respiration. This inhibition is due to peroxynitrite, rather than "pure" NO^· formation, because the suppression of cell respiration can be restored by both NOS inhibitors and by superoxide or peroxynitrite scavengers.

Although exposure to high concentrations of peroxynitrite leads to rapid cell death associated with rapid energetic derangements, lower concentrations of peroxynitrite can, after several hours, lead to apoptotic cell death (Bonfoco et al., 1995; Estevez et al., 1995; Salgo et al., 1995). In isolated tissues and organs, peroxynitrite elicits a variety of alterations. Peroxynitrite infusion causes a reduction in myocardial contractility in isolated perfused hearts (Schulz et al., 1995) and induces an impairment of the endothelium-dependent relaxant ability (Villa et al., 1994). The finding that the development of this endothelial dysfunction can be prevented by NO^· donors (Villa et al., 1994) supports the notion that toxic acute effects are due to OONO formation (Moro et al., 1994, 1995).

	II. DNA Damage

Top Previous Next References

Free radical-mediated reactions can cause structural alterations in DNA (e.g., nicking, base-pair mutations, rearrangements, deletions, insertions, and sequence amplification). The endogenous reactions that are likely to contribute to ongoing DNA damage are oxidation, methylation, depurination, and deamination (Totter, 1980; Ames, 1989). NO or, more likely, reactive products derived from it, such as NO₂, ONOO, N₂O₃ and HNO₂, are mutagenic agents, with the potential to produce nitration, nitrosation, and deamination reactions on DNA bases (Routledge et al., 1994). Methylation of cytosines in DNA is important for the regulation of gene expression, and normal methylation patterns can be altered during carcinogenesis (Weitzman et al., 1994). Conversion of guanine to 8-hydroxyguanine, a frequent result of reactive oxygen species (ROS) attack (Halliwell and Aruoma, 1991; Dizdaroglu, 1993; Box et al., 1995), has been found to alter the enzyme-catalyzed methylation of adjacent cytosines (Weitzman et al., 1994)thus providing a link between oxidative DNA damage and altered methylation patterns.

The chemistry of DNA damage by several ROS has been well characterized in vitro (Steenken, 1989; Dizdaroglu, 1993; Epe, 1993; Box et al., 1995), although specific information is needed about the changes produced by peroxyl (RO₂), alkoxyl (RO^·), ozone (O₃), and several of the reactive nitrogen species (RNS) (e.g., ONOO) is lacking. Different ROS affect DNA in different ways [e.g., H₂O₂ does not react with DNA bases at all (Halliwell and Aruoma, 1991; Dizdaroglu, 1993)], whereas HO^· generates a multiplicity of products from all four DNA bases, and this pattern seems to be a diagnostic "fingerprint" of HO^· attack (Halliwell and Aruoma, 1991). By contrast O₂ selectively attacks guanine (Epe, 1993; Van den Akker et al., 1994). The most commonly produced base lesion, and the one most often measured as an index of oxidative DNA damage, is 8-hydroxyguanine. It is sometimes measured as the nucleoside, 8-hydroxydeoxyguanosine (Floyd et al., 1986; Ames, 1989). These assay methods have been reviewed in detail (Floyd et al., 1986; Halliwell and Aruoma, 1991; Halliwell and Dizdaroglu, 1992; Dizdaroglu, 1993).

Damage to DNA by ROS/RNS seems to occur naturally, in that low steady-state levels of base damage products have been detected in nuclear DNA from human cells and tissues (Floyd et al., 1986; Ames, 1989; Halliwell and Dizdaroglu, 1992; Richter, 1992; Musarrat and Wani, 1994). The pattern of damage to the purine and pyrimidine bases suggests that at least some of the damage occurs by HO^· attack, suggesting that HO^· is formed in the nucleus in vitro (Halliwell and Dizdaroglu, 1992). ROS/RNS can also damage mitochondrial DNA, and such damage has been suggested to be important in several human diseases and in the aging process (Harman, 1992; Shigenaga et al., 1994). Mitochondria are often said to be the most important intracellular source of ROS, but it is difficult to unambiguously confirm this postulate (Halliwell and Gutteridge, 1985). However, it seems very likely that the mitochondrial electron transport chain generates ROS in vivo (Ambrosio et al., 1993; Guidot et al., 1993) and that mitochondrial DNA is damaged by them. The roles that ROS or RNS play in the DNA damage have not yet been completely elucidated. This apparent increased net oxidative damage in mitochondrial DNA compared with nuclear DNA could be because of the proximity of mitochondrial DNA to ROS generated during electron transport, the lack of histone proteins to protect the DNA against attack, or inefficient repair, so that base damage accumulates to higher levels.

DNA damage can be repaired by the action of a series of enzymes (Demple and Harrison, 1994). However, DNA from human cells and tissues contains low levels of DNA base damage products (Ames, 1989; Malins and Haimanot, 1991; Halliwell and Dizdaroglu, 1992; Bashir et al., 1993; Jaruga et al., 1994; Adachi et al., 1995), suggesting that these enzymes do not achieve complete removal of modified bases, perhaps because they operate at close to maximum capacity in vivo. DNA damage by ROS/RNS can cause multiple lesions, including single and double strand breaks, apurinic/apyrimidinic sites and modified pyrimidines and purines. Repair of these lesions occurs primarily by base excision repair, although nucleotide excision repair may also be involved. A repair system for the abasic apurinic/apyrimidinic sites produced by spontaneous depurination also exists. Areas of current interest include the role of poly(ADP-ribose) polymerase (PARP) in the rejoining of DNA strand breaks, including those induced by ROS (Satoh et al., 1993; Satoh and Lindahl, 1994).

	III. Poly(ADP-Ribose) Synthetase

Top Previous Next References

Poly(ADP) synthetase (PARS) [also known as PARP or poly(ADP-ribose) transferase] is a protein modifying and nucleotide-polymerizing enzyme that is present abundantly in the nucleus (Althaus and Richter, 1987; De Murcia and Menissier-De Murcia, 1994). The obligatory trigger of PARS activation is the nicks and breaks in the DNA strand, which can be induced by a variety of environmental stimuli and free radical (or oxidant) attacks; these include the oxidants HO₂, HO^·, and ONOO, ionizing radiation, and genotoxic agents, such as N-methyl-N'-nitro-N-nitrosoguanidine. The physiological function of PARS and poly(ADP-ribosylation) is still under much debate. From studies using pharmacological inhibitors of PARS, poly(ADP-ribosylation) has been suggested to regulate gene expression and gene amplification, cellular differentiation and malignant transformation, cellular division, and DNA replication, as well as apoptotic cell death (Althaus and Richter, 1987; Lautier et al., 1993; De Murcia and Menissier-De Murcia, 1994; Lindahl et al., 1995; Wang et al., 1995; Simbulan-Rosenthal et al., 1996). However, recent studies using cells from PARS(-/-) mice have failed to demonstrate a role for PARS in the process of apoptosis induced by various apoptotic signals, such as the Fas ligand or dexamethasone (De Murcia et al., 1997; Morrison et al., 1997; Wang et al., 1995, 1997).

In the 1980s, Berger and Okamoto have observed rapid depletion of NAD+ due to PARS activation, leading to cellular ATP depletion, and functional alterations of the cell, with eventual necrotic-type cell death. The main cytotoxic triggers used in these studies in vitro were alkylating agents, radiation, and H₂O₂, whereas the most frequently used PARS inhibitors were nicotinamide, 3-aminobenzamide, and benzamide.

Research into the "suicidal" role of PARS gained new momentum in the mid-1990s because of the observations in vitro that NO^· or peroxynitrite can trigger DNA single-strand breakage and PARS activation (Radons et al., 1994; Eliasson et al., 1997; Szabó et al., 1996b). NO^· and peroxynitrite can also inhibit mitochondrial respiration and exert other cytotoxic effects on their own. Thus, it is likely that a synergistic relationship exists between the PARS-mediated pathways and PARS-independent pathways of cellular metabolic suppression (Fig. 1). Furthermore, the observations that NO^· and peroxynitrite are important mediators of the cellular damage in various forms of inflammation and reperfusion injury suggest that the PARS-related suicide pathway might play a role in various pathophysiological conditions in vivo (Fig. 2).

View larger version (46K): [in this window] [in a new window]   Fig. 1.   Suggested mode of activation of PARS in inflammation and shock.

View larger version (186K): [in this window] [in a new window]   Fig. 2.   PARS activity: Staining was observed at 4 h after carrageenan-induced pleurisy (A); at 5 days after gentamicin-induced renal injury (B); at 60 min after splanchnic artery occlusion shock (C); and at 35 days following collagen-induced arthritis (D). Original magnification, 125×. Figure is representative of at least three experiments performed on different experimental days.

	IV. Relative Importance of Reactions of Glutathione with Nitric Oxide, Oxyradicals, and Peroxynitrite in Endotoxic Shock and Inflammation

Top Previous Next References

Glutathione is a known oxyradical scavenger (Darley-Usmar and Halliwell, 1996). Moreover, glutathione can react with NO^· to form S-nitrosoglutathione, a vasodilator compound (Simon et al., 1993). Thus, theoretically, the mechanism of the observed vascular alterations in the BSO-pretreated cells, and tissues may be related to peroxynitrite, oxyradicals, NO^·, or the combination of these. From data in the literature, it seems unlikely that a glutathione-NO^· reaction plays a major role in the observed changes. This conclusion was based on the following considerations: 1) in accordance with previous studies (Jia and Furchgott, 1993; Stamler, 1995), we found no differences in the endothelium-independent relaxant effect of S-nitroso-N-acetyl-DL-penicillamine in control and BSO-treated animals, suggesting that the reduction in endogenous glutathione does not affect NO-induced relaxant responses; and 2) in vitro studies in macrophages and other cell types have established that endogenous glutathione only protects against very high (pharmacologically relevant) fluxes of NO^·, but not against lower levels of NO^· production, such as the ones that are relevant to the in vitro or in vivo conditions in our experiments (Sakanashi et al., 1991; Walker et al., 1995). On the other hand, the possibility that the enhancement of cytotoxicity in BSO-treated cells or tissues after LPS treatment is related, in part, to increased oxyradical-induced cytotoxic effects is a good possibility. Data supports that depletion of endogenous glutathione enhances the cytotoxic effects of H₂O₂ and oxyradicals, and we have also observed an enhancement of H₂O₂ toxicity in endothelial cells and smooth muscle cells (Cuzzocrea et al., 1998c). In the experiments that involve LPS stimulation, it is conceivable that a more pronounced inhibition of mitochondrial respiration by oxygen-derived free radicals and oxidants can lead to a dysfunctional electron transfer, with more superoxide production from the mitochondria. This positive feedback cycle would also lead to an enhancement of peroxynitrite production, with subsequent increased cytotoxicity. It is noteworthy in this context that it is the production of superoxide, not the production of NO^·, that represents the rate-limiting factor in the production of peroxynitrite during endotoxemia (Szabó and Salzman, 1995).

It must be kept in mind that, in immunostimulated cells, the production of various oxygen- and nitrogen-derived free radicals and oxidants occurs in a simultaneous fashion. Therefore, it is conceivable that important interactions exist between these various species in terms of oxidative potential and cytotoxicity. For instance, in respect to peroxynitrite-induced oxidative injury, it is well established that the ratio of NO^· and superoxide determines the oxidant capacity, and excess NO^· reduces peroxynitrite-induced oxidative processes (Rubbo et al., 1994; Szabó and Salzman, 1995; Petit et al., 1996). H₂O₂, on the other hand, prolongs the half-life of peroxynitrite (Miles et al., 1996), and synergizes with peroxynitrite in terms of cytotoxicity. Thus, it is possible that the cytotoxic effects we observed in response to immunostimulation represent the sum of a complex interaction between various oxygen- and nitrogen-derived radicals and oxidants. Nevertheless, based on the similarities between the effects of exogenously added peroxynitrite and LPS treatment, and considering the simultaneous protective effects of N-methyl-L-arginine and tetrakis-(4-benzoic acid) porphyrin (MnTBAP) against the vascular failure in response to LPS challenge (see above), we propose that peroxynitrite, or a peroxynitrite-derived oxidant, contributes to protein oxidation in response to immunostimulation.

Recent studies demonstrate that endogenous glutathione plays an important role in reducing vascular hyporeactivity and endothelial dysfunction in response to peroxynitrite and endotoxic shock, as well as in acute inflammation. In fact we have shown that BSO-treated rats developed a significant inflammatory response, as compared with the animals that have a normal glutathione system. These findings are in agreement with previous suggestions that glutathione plays an important role in blocking the oxidant-induced injury and, specifically, in blocking the peroxynitrite-induced injury (Karoui et al., 1996; Cuzzocrea et al., 1998c). A variety of additive or synergistic cytotoxic processes triggered by peroxynitrite may contribute to acute and delayed cytotoxicity, and depletion of glutathione may also interfere with these pathways. This points out the importance of intact glutathione pools, as protective mechanisms against the vascular failure under conditions of oxidant stress, shock, and inflammation. There are several ways to improve glutathione status and/or replenish cellular glutathione stores. For instance, cell-permeable glutathione analogs have been described (Morris et al., 1995). These strategies may represent alternative or additional approaches to other approaches directed toward the prevention of the loss of vascular patency in shock and inflammation.

	V. Superoxide Dismutase

Top Previous Next References

Under normal circumstances, formation of O₂ (the one-electron reduction product of oxygen) is kept under tight control by SOD enzymes. These include the Mn enzyme in mitochondria (SOD2) and Cu/Zn enzyme present in the cytosol (SOD1) or extracellular surfaces (SOD3). The importance of SOD2 is highlighted by the findings that, in contrast to SOD1 (Reaume et al., 1996) and SOD3 (Carlsson et al., 1995), SOD2 knockout is lethal to mice (Lebovitz et al., 1996; Melov et al., 1999). In acute and chronic inflammation, the production of O₂ is increased at a rate that overwhelms the capacity of the endogenous SOD enzyme defense system to remove them. The result of such imbalance results in O₂-mediated damage. A proposal that O₂ was intimately involved with the inflammatory response was raised as early as the 1970s through the pioneering work of McCord and Fridovich (McCord and Fridovich, 1969). Some important proinflammatory roles for O₂ (Fig. 3) include endothelial cell damage and increased microvascular permeability (Droy-Lefaix et al., 1991; Haglind et al., 1994), formation of chemotactic factors such as leukotriene B₄ (Fantone and Ward, 1982; Deitch et al., 1990), recruitment of neutrophils at sites of inflammation (Boughton-Smith et al., 1993; Salvemini et al., 1996a, 1999), lipid peroxidation and oxidation, DNA single-strand damage (Dix et al., 1996), and formation of ONOO (Beckman et al., 1990; Ischropoulos et al., 1992a; Crow and Beckman, 1995; Salvemini et al., 1998, 1999). Most of the knowledge gathered about the roles of superoxide in disease has been collected by the use of the native SOD enzyme and, more recently, by data generated in transgenic animals that overexpress the human enzyme (Huber et al., 1980; Uematsu et al., 1994; Fridovich, 1995).

View larger version (23K): [in this window] [in a new window]   Fig. 3.   Biochemical impact of superoxide generation.

	VI. Radical Generation

Top Previous Next References

Following ischemia, superoxide is produced during the reperfusion phase, and it rapidly reacts with NO^· and forms ONOO. This has been demonstrated in the heart (Matheis et al., 1992; Naseem et al., 1995; Schulz and Warnbolt, 1995), liver (Ma et al., 1995), kidney (Yu et al., 1994), intestine (Szabó et al., 1995a), brain (Cazevielle et al., 1993; Fagni et al., 1994; Gunasekar et al., 1995), and lung (Ischiropoulos et al., 1995; Kooy et al., 1995). Under these conditions, prevention of ONOO generation by inhibition of NO^· biosynthesis markedly reduces reperfusion injury, as shown by reduced pulmonary lipid peroxidation (Ischiropoulos et al., 1995) or improved myocardial mechanical performance (Schulz and Warnbolt, 1995).

A growing body of evidence supports a role for ONOO and other reactive species in neuronal injury associated with ischemia/reperfusion injury in the central nervous system. The original proposition (Beckman, 1991), that ONOO (and not NO^· or O₂ independently) is a major cytotoxic mediator in the neuronal injury during stroke and N-methyl-D-aspartic acid (NMDA) receptor activation, was based on theoretical considerations and previous evidence showing that reperfusion injury in the central nervous system is associated with activation of NMDA receptors, which then triggers the production of O₂ and NO^·. There is now indirect evidence to show that NMDA receptor activation is associated with a marked increase in HO^·-like activity in the brain (blocked by inhibition of NOS), which is presumably due to ONOO generation (Hammer et al., 1993). The involvement of O₂ and the protective effect of O₂ neutralizing strategies (Cazevielle et al., 1993; Dawson et al., 1993; Lafon-Cazal et al., 1993; Fagni et al., 1994; Beal et al., 1995; Crow and Beckman, 1995; Dawson, 1995; Gunasekar et al., 1995), as well as the involvement of NO^· and the protective effect of NOS inhibition (Huang et al., 1994; Smith et al., 1994; Schulz et al., 1995; Zielasek et al., 1995), has been well established in various forms of central nervous system injury.

Similar to inflammation and shock, the mechanism of ONOO-induced cellular damage in the ischemia/reperfusion remains a subject for future investigations, but presumably involves multiple mechanisms. Both in vivo and in vitro evidence clearly suggest the involvement of PARS in the neuronal damage associated with NO^· (or ONOO) production in response to NMDA receptor activation (Wallis et al., 1993; Cosi et al., 1994; Zhang et al., 1994, 1995).

Endothelial cells appear to be major regulators of neutrophil traffic, regulating the process of neutrophil chemoattraction, adhesion, and migration from the vasculature to the tissue. During the early phase of reperfusion, P-selectin is rapidly released to the cell surface from preformed storage pools after exposure to certain stimulisuch as H₂O₂, histamine, or complementand allows the leukocytes to roll along the endothelium (Geng et al., 1990). ICAM-1, constitutively expressed on the surface of endothelial cells, is then involved in neutrophil adhesion (Geng et al., 1990; Farhood et al., 1995). Hypoxic endothelial cells synthesize proinflammatory cytokines, which can up-regulate endothelial expression of the constitutive adhesion molecule ICAM-1 in autocrine fashion (Shreeniwas et al., 1992). The expression of P-selectin and ICAM-1 corresponds with the induction of neutrophil recruitment, which is maximal within the first hour of reperfusion, and persists at a lower rate in the late phase of reperfusion (Clark et al., 1995). In accordance with these findings, it has been demonstrated that ischemia and reperfusion induced the appearance of P-selectin on the endothelial vascular wall and up-regulated the surface expression of ICAM-1 on endothelial cells. Thus, it has been hypothesized that oxidative and nitrosative damage in ischemia/reperfusion and shock requires the presence of ROS, which are mainly produced from the massive neutrophil infiltration (Cuzzocrea et al., 1998c).

Important cardiovascular consequences of circulatory shock include reduced responsiveness of arteries and veins to exogenous or endogenous vasoconstrictor agents (vascular hyporeactivity), myocardial dysfunction, and disrupted intracellular energetic processes. Recent studies prompted these conclusions based mainly on results obtained with the use of NOS inhibitors, but they did not or could not distinguish between the effects of NO^· versus ONOO. Recent data demonstrate that authentic ONOO is capable of mimicking many of the cardiovascular alterations associated with shock (endothelial dysfunction, vascular hyporeactivity, myocardial failure, and cellular energetic failure) (see above). In circulatory shock, proinflammatory cytokines invoke a pleiotropic cellular response, including the stimulation of oxygen-centered free radicals, such as O₂. The majority of NO^· produced by macrophages is converted to ONOO (Ischiropoulos et al., 1992b). The production of ONOO (evidenced by increased nitrotyrosine immunoreactivity or increased oxidation of the fluorescent probe dihydrorhodamine 123 to rhodamine 123) has recently been demonstrated in endotoxic shock and in hemorrhagic shock (Wizemann et al., 1994; Szabó et al., 1995b).

A large number of studies demonstrate the protective effect of SOD in various models of endotoxic and hemorrhagic shock and splanchnic artery occlusion/reperfusion injury (McKechnie et al., 1986; Wang et al., 1990; Rhee et al., 1991; Youn et al., 1991; McCord, 1993; Kapoor and Prasad, 1995; Salvemini et al., 1999). Furthermore, there is a large amount of evidence to show that the production of reactive species such as O₂, H₂O₂, and HO^· occurs at the site of inflammation and contributes to tissue damage (Salvemini et al., 1996a; Cuzzocrea et al., 1997). Inhibitors of NOS activity reduce the severity of inflammation and support a role for NO^· in the pathophysiology associated with various models of inflammation (Tracey et al., 1995; Wei et al., 1995; Salvemini et al., 1996b; Cuzzocrea et al., 1997). In addition to NO^·, ONOO is also generated during inflammation damage (Salvemini et al., 1996b; Cuzzocrea et al., 1997). The involvement of ONOO in these conditions is strongly supported by direct measurements. For example, in arthritis, nitrotyrosine levels increase in plasma and synovial fluid (Kaur and Halliwell, 1994). In ileitis (Miller et al., 1995) and in endotoxin-induced intestinal inflammation (Chamulitat et al., 1996), there is immunocytochemical documentation (increased nitrotyrosine immunoreactivity in the inflamed tissues) of augmented ONOO production (Fig. 4).

View larger version (172K): [in this window] [in a new window]   Fig. 4.   Nitrotyrosine formation. Staining was present at 4 h after carrageenan-induced pleurisy (A); at 5 days after gentamicin-induced renal injury (B); at 60 min after splanchnic artery occlusion shock (C); and at 35 days following collagen-induced arthritis (D). Original magnification, 125×. Figure is representative of at least three experiments performed on different experimental days.

The pathophysiological role of NO^· and ONOO in the gastrointestinal damage elicited by endotoxin or chronic inflammation has been the subject of a variety of detailed investigations. The ability of authentic ONOO to cause severe colonic inflammation has been documented (Rachmilewitz et al., 1993). The production of ONOO in colitis may be even more pronounced because of the parallel down-regulation of SOD (Seo et al., 1995), which makes the O₂ available for coupling with NO^·. Desferrioxamine, a putative peroxynitrite scavenger (Denicola et al., 1995), or SOD protects against the gastric damage elicited by NO^· donors, supporting the view that peroxynitrite (and not NO^· per se) is the cytotoxic species in these models (Lamarque and Whittle, 1995). Recent investigations have also concluded that inhibition of PARS exerts beneficial effects in shock (Szabó et al., 1997), reperfusion injury (Zhang et al., 1994; Zingarelli et al., 1996; Cuzzocrea et al., 1997; Thiemermann et al., 1997), and inflammation (Szabó et al., 1997, 1998; Cuzzocrea et al., 1998a,b).

	VII. Pharmacological Intervention to Reduce Reactive Oxygen Species Generation in Shock, Inflammation, and Ischemia/Reperfusion

Top Previous Next References

Interventions, which reduce the generation or the effects of ROS exert beneficial effects in a variety of models of inflammation and shock. These therapeutic interventions include a vitamin E-like antioxidant (Cuzzocrea et al., 1999a), an SODm (Wang et al., 1990; Cuzzocrea et al., 1999b), and a ONOO decomposition catalyst (Salvemini et al., 1998). The therapeutic efficacy of SOD itself in animals with systemic inflammation, hemorrhage, or shock is controversial. The following reasons may explain the lack of effect of SOD against the tissue injury associated with local or systemic inflammation: 1) SOD metabolized O₂ to H₂O₂. Without efficient removal of H₂O₂, however, H₂O₂ is converted to the highly toxic HO^· (Goode and Webster, 1993). Indeed, SOD may function as a pro-oxidant by catalyzing the conversion of H₂O₂ to HO^· (Yim et al., 1990), such as is believed to be the case in Down's syndrome. 2) Neither SOD nor O₂ easily cross biological membranes. Thus, an increase in the amounts of extracellular SOD does not attenuate the effects of the O₂ generated by intracellular sources (Meister, 1992). In contrast to SOD, spin-trapping nitrones, such as phenyl-N-tert-butyl nitrone, consistently improve outcome in rat models of endotoxic (McKechnie et al., 1986; Hamburger and McCay, 1989) and traumatic shock (Novelli et al., 1986; Novelli, 1992). The early phase of the inflammatory process is related to the production of histamine, leukotrienes, platelet-activating factor, and possibly cyclo-oxygenase products, whereas the delayed phase of the inflammatory response has been linked to neutrophil infiltration and the production of neutrophil-derived free radicals and oxidants, such as H₂O₂, O₂, and HO^·, as well as the release of other neutrophil-derived mediators (Ohishi et al., 1989; Salvemini et al., 1996b).

The peroxynitrite anion is formed and can be prepared via by a number of pathways (Fig. 5), particularly through various oxidations of nitrogen oxides and photolysis and radiolysis of solid nitrate salts. As previously noted, formation of peroxynitrite by the combination of NO^· and O₂ radicals is quite favorable (Huie and Padmaja, 1993), as is the combination of NO₂ and HO^· (k_obs = 4.5 × 10⁹ M¹ s¹) (Logager and Sehested, 1993). Four practical laboratory syntheses of peroxynitrite are known, including: 1) the original route, in which intermediate NO⁺ from dehydration of nitrous acid is trapped with peroxide (Beckman et al., 1994a); 2) nucleophilic attack of peroxide on alkyl nitrites (Moncada and Higgs, 1993); 3) ozonolysis of aqueous azide solutions (Gleu and Hubold, 1935; Pryor et al., 1995); and 4) photolysis and radiolysis of nitrate salts (King et al., 1992). Yields are readily observed and quantified spectrophotometrically by the characteristic yellow color of the anion (_max = 302 nm, _max = 1670 ± 50 M¹ cm¹) (Beckman et al., 1994b).

View larger version (16K): [in this window] [in a new window]   Fig. 5.   Known and possible reactions of peroxynitrite.

Peroxynitrite is stable in alkaline solution, but the conjugate acid (pK_a = 6.80) (Logager and Sehested, 1993) is colorless and unstable, and isomerizes rapidly to nitrate, which is considerably more stable (H° = 40 kcal/mol) (Squadrito et al., 1995). An additional acid-catalyzed pathway is observed at extreme acidity, which apparently corresponds to decomposition of [H₂OONO]⁺ (Benton and Moore, 1970).

The rearrangement of peroxynitrite to nitrate is coupled intimately with the oxidation chemistry of this species, and both reactions have been the subject of recent investigations and intense debate. Peroxynitrite and its conjugate acid are strong oxidants, capable of effecting one- and two-electron reactions akin to those of HO^·, nitrogen dioxide (NO₂), and nitrosoniun cation. Oxidations of thiols (Radi et al., 1991b; Van der Vliet et al., 1994), sulfides (Pryor et al.; 1994; Padmaja et al., 1996), transition metal complexes (Goldstein and Czapski, 1995; Groves and Marla, 1995), deoxyribose (Beckman et al., 1990), halide ions (Hughes et al., 1971; Goldstein and Czapski, 1995), ascorbate (Barlett et al., 1995; Squadrito et al., 1995), olefins (Halfpenny and Robinson, 1992), benzenes (Halfpenny and Robinson, 1992), phenols (Halfpenny and Robinson, 1992, 1996; Beckman et al., 1992; Ischiropoulos et al., 1992a; Van der Vliet et al., 1994a,b; Groves and Marla, 1995; Ischiropoulos et al., 1996; Ramezanian et al., 1996), and other aromatics (Halfpenny and Robinson, 1996; Van der Vliet et al., 1994b; Alvarez et al., 1996) by peroxynitrite have been described.

Peroxynitrite is a particularly effective oxidant of aromatic molecules and organosulfur compounds that include free amino acids and peptide residues. Cysteine and glutathione, which are significant components of antioxidant reservoirs, are converted to disulfides (Radi et al., 1991b; Van der Vliet et al., 1994b). Methionine is converted to sulfoxide or is fragmented to ethylene and dimethyldisulfide (Pryor et al., 1994; Padmaja et al., 1996). Dimethyl sulfoxide is oxidized to formaldehyde (Beckman et al., 1990). Methyl acrylate is polymerized (Halfpenny and Robinson, 1992). Tyrosine and tryptophan undergo one-electron oxidations to radical cations, which are competitively hydroxylated, nitrated, and dimerized (Ischiropoulos et al., 1992b; Van der Vliet et al., 1994b; Alvarez et al., 1996; Ramezanian et al., 1996). The formation of nitrotyrosine is particularly favorable, and the appearance of this product in biological samples is taken as diagnostic of exposure to peroxynitrite (Salvemini et al., 1996b). Purine nucleotides are vulnerable to oxidation (Douki and Cadet, 1996; Uppu et al., 1996) and to adduct formation (Douki et al., 1996).

Peroxynitrite reacts with a number of metal-containing enzymes. It can inactivate aconitase as can superoxide (Keyer and Imlay, 1996) by oxidative cleavage of the labile iron site from the parent cluster (Hausladen and Fridovich, 1996), and certain SOD enzymes by nitration of critical residues (Ischiropoulos et al., 1992b). Peroxynitrite is proposed to down-regulate neuronal NOS by oxidation of a requisite cofactor (Huhmer et al., 1996). Heme prosthetic groups of cytochromes (Thomson et al., 1995) and peroxidases (Floris et al., 1993) are oxidized reversibly by peroxynitrite. Furthermore, peroxynitrite can release metal ions from the active sites of other enzymes (Swain et al., 1994; Crow et al., 1995).

The kinetics of substrate oxidation by peroxynitrite is extremely complex (Pryor and Squadrito, 1995). When the solution pH is fixed near physiologically relevant values, so that peroxynitrite and its conjugate acid are present in fixed ratios, observed rates of substrate oxidation and peroxynitrite loss are first-order in peroxynitrite. These rates typically increase from the peroxynitrite isomerization limit in proportion to increasing substrate concentrations; however, a significant zero-order limit is frequently encountered. Fairly high substrate concentrations, for example, several millimolar ascorbate (Barlett et al., 1995), are often required to attain first-order behavior. This behavior has been interpreted to involve competition between direct, first-order reaction of peroxynitrite with substrate and quenching by substrate of any activated form of peroxynitrite complex (Pryor and Squadrito, 1995).

The reaction of peroxynitrite with methionine at pH 7.4, 25°C, is the prototypical example (Pryor et al., 1994), although similar kinetics is observed with ascorbate (Squadrito et al., 1995). At high methionine concentrations, the sulfoxide is formed from a two-electron oxidation by peroxynitrite; observed peroxynitrite decay rates are first-order in methionine, but the extrapolated intercept with no added methionine is several times faster than the intrinsic isomerization rate of peroxynitrite. At lower methionine concentrations, the observed rates drop below the first-order limit, and an increasing mole fraction of the observed products is ethylene produced by decomposition of a sulfur-centered radical cation that results from a one-electron oxidation. The one-electron and two-electron oxidation products are proposed to arise from discrete reactions with activated and ground-state peroxynitrite, respectively.

The approach to a zero-order limit with increasing substrate indicates that the activated form is formed in a rapidly reversible equilibrium between peroxynitrite/peroxynitrous acid, which is a stronger oxidant. Furthermore, the rates of isomerization and zero-order oxidation are typically similar, which suggests that these processes follow coincident paths through the intermediate.

The nature of the activated state is the subject of considerable debate. Some researchers favor the formulation as a pair of radicals, HO^· and NO₂, which results from reversible peroxide bond homolysis and can proceed either to oxidations or to recombination as nitrate (Halfpenny and Robinson, 1996; Beckman et al., 1990; King et al., 1992; Van der Vliet et al., 1994a; Alvarez et al., 1996). It has also been suggested that the intermediate is a trans-isomer that is formed via isomerization of the thermodynamically favored cis geometry by hindered rotation about the nitrogen peroxide bond (Hughes and Nicklin, 1968; Koppenol et al., 1992; Goldstein and Czapski, 1995; Pryor et al., 1994, 1996; Squadrito et al., 1995; Tsai et al., 1996a,b). The discrepancy in viewpoints is of more than academic interest, due to the extremely rapid and general oxidative reactivity of the product: HO^·.

Interest in peroxynitrite chemistry recently has grown in significance, but experimental evidence on this point is limited and not completely definitive. Certain results seem to favor the radical hypothesis. For example, HO^· and NO₂ radicals recombine to form peroxynitrite as a major product (Logager and Sehested, 1993), which provides reversibility. The bond dissociation enthalpies can be calculated with some assumptions and coincide closely with experimental activation enthalpies for isomerization (e.g., 17 versus 18 ± 1 kcal/mol) (Mahoney, 1970; Koppenol et al., 1992). However, other possible processes are calculated to have similar activation enthalpies, and it has been argued that the experimental entropy is too small to accommodate homolysis (Koppenol et al., 1992). Radical trapping products, particularly hydroxylated aromatics and oxidized spin traps, are reported to form (Halfpenny and Robinson, 1996; Beckman et al., 1990; Van der Vliet et al., 1994b), but yields relative to charged peroxynitrite are invariably very low (Shi et al., 1994; Richeson and Ingold, 1996). Finally, the rate of isomerization to nitrate shows none of the dependence on solution viscosity that is expected for diffusive radical recombination (Pryor et al., 1996). A detour around such objections is an efficient cage recombination process, but the distinction between tightly caged radicals and a geometric isomerization becomes an issue more of semantics than of physics.

Formulation of the activated state as the trans-isomer seems to be reasonable from a structural and theoretical standpoint. High-level calculations find that the cis-isomer is more stable (McGrath et al., 1988; Tsai et al., 1996a,b), and predictions for this isomer match experimental spectra (Tsai et al., 1996a,b). Isomerization of the trans-isomer to nitrate is predicted to be more facile than that of the cis-isomer, the terminal peroxide oxygen atom being more favorably disposed geometrically and electronically for intramolecular addition to nitrogen in the former (Tsai et al., 1996a). Furthermore, interconversion of the isomers by hindered rotation about the partial bond between nitrogen and the peroxide is predicted to be markedly more facile for the protonated acid than for the anion (McGrath and Rowland, 1994; Tsai et al., 1996a) that neatly explains the acid requirement of the nitrate isomerization reaction. However, reports of direct experimental observation of the trans-isomer are notably absent from the literature. In addition, yields of oxidized product versus peroxynitrite loading from some substrates that react only with the activated state (McGrath et al., 1988; Yang et al., 1992; Crow et al., 1994) and ferricyanide ([Fe(CN)₆]⁴) (Goldstein and Czapski, 1995) are reported to be substoichiometric and independent of substrate concentration. This result implies that substrate quenching of the intermediate is in competition with nitrate formation or some other decay reaction (e.g., to nitrite), directly from the ground state (Goldstein and Czapski, 1995).

Other activation mechanisms, such as heterolysis to HO^· and NO₂⁺, are ruled out by the experimental observation that double labeling of peroxynitrite with ¹⁸O across the peroxide bond produces double-labeled nitrate upon isomerization (Anbar and Taube, 1954). Wider use of isotope labels as tracers (¹⁸O) or as magnetic probes (¹⁵N) may assist future clarification of the nature of the activated state in the oxidation and isomerization reactions.

It should be noted that a few examp